Sequential sampling plans for inspection by attributes

The specified plans apply to: end items, components and raw materials, operations, materials in process, supplies in storage, maintenance operations, data or records, administrative procedures. Annexes A, B and C form an integral part of this standard.

Plans d'échantillonnage progressif pour le contrôle par attributs

1.1.1 La présente Norme internationale présente des plans et des règles d'échantillonnage progressif pour le contrôle par attributs d'unités discrètes. Les plans dans le corps principal de la norme sont indexés en termes de point du risque fournisseur et de point du risque client. L'annexe A donne des plans et des règles d'échantillonnage progressifs indexés en termes de niveau de qualité acceptable (NQA) en complément du système des plans d'échantillonnage de l'ISO 2859-1. Le but de la présente Norme internationale est de fournir des règles basées sur la détermination progressive des résultats de contrôle, afin de persuader le fournisseur, par des pressions économiques et psychologiques, de la non-acceptation de lots de qualité inférieure, pour qu'il fournisse des lots de qualité ayant une forte probabilité d'acceptation. En même temps le client est protégé par une limite supérieure imposée de la probabilité d'accepter des lots de faible qualité. 1.1.2 Les plans d'échantillonnage contenus dans la présente Norme internationale sont applicables, notamment mais d'une manière non limitative, aux contrôles ci-après: -- produits finis; -- composants ou matières premières; 1286opérations; -- matériaux en cours de fabrication; -- fournitures en stock; -- opérations d'entretien; -- informations ou enregistrements; -- procédures administratives. La présente Norme internationale contient des plans d'échantillonnage pour le contrôle par attributs d'unités discrètes. Les plans d'échantillonnage sont utilisables lorsque l'évaluation de la non-conformité s'exprime soit en termes de proportion (ou pourcentage) d'unités non conformes, soit en termes de non-conformités par unité (par 100 unités). Les plans d'échantillonnage sont fondés sur l'hypothèse que les non-conformités surviennent de façon aléatoire et sont statistiquement indépendantes. Il peut exister des raisons de supposer qu'une non-conformité dans une unité pui

Sequential sampling plans for inspection by attributes

General Information

Status
Withdrawn
Publication Date
06-Nov-1991
Withdrawal Date
06-Nov-1991
Current Stage
9599 - Withdrawal of International Standard
Completion Date
17-Oct-2006

Relations

Buy Standard

Standard
ISO 8422:1991 - Sequential sampling plans for inspection by attributes
English language
45 pages
sale 15% off
Preview
sale 15% off
Preview
Standard
ISO 8422:1996
English language
45 pages
sale 10% off
Preview
sale 10% off
Preview
e-Library read for
1 day
Standard
ISO 8422:1991 - Plans d'échantillonnage progressif pour le contrôle par attributs
French language
47 pages
sale 15% off
Preview
sale 15% off
Preview
Standard
ISO 8422:1991 - Plans d'échantillonnage progressif pour le contrôle par attributs
French language
47 pages
sale 15% off
Preview
sale 15% off
Preview

Standards Content (Sample)

INTERNATIONAL
8422
STANDARD
First edition
1991-11-15
Sequential sampling plans for inspection by
attributes
Plans d’&hntillonnage progressifs pour le contr6le par attributs
----
____________-_-.___--_-----------
I_
----- ----
-- ----
Reference number
-.-_________- - IS0 8422:1991(E)
---_- -- _ _ __ .__

---------------------- Page: 1 ----------------------
IS0 8422:1991 (E)
,
Contents
Page
Section 1 General ,,.,.,,.,~.,.,.,,.,.,,.,,.,.,.~.~. 1
1.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .-.,. 1
1.2 Normative references . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I
2
1.3 Definitions and symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . 3
1.4 Principle of a sequential sampling plan by attributes
5
Section 2 Choice of sampling plan .,.,.,.,.,.,.
2.1 Choice between sequential, single, double and multiple sampling
plans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
. . . . . . . 5
2.2 Particular reservations on the inspection of small lots
2.3 Selection of a sampling plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 Pre-operation preparations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
,.,,.,. 11
Section 3 Operation of a sequential sampling plan
............... ......................................... 11
3.1 Specification of the plan
............................. .............................. 11
3.2 Drawing of the sample
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3 Cumulative count .,.
3.4 Determination of acceptability .,,.,.,. 11
13
3.5 Operating characteristic curves and average sample size
,.,. 13
3.6 Validity of approximations
Annexes
A Sequential sampling plans corresponding to IS0 2859-l sampling
,. 18
plans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .~.*.
B Determination of the parameters of a sequential sampling
40
plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C Calculation of the operating characteristic curve and average
41
sample size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
D Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0 IS0 1991
All rights reserved. No part of this publication may be reproduced or utilized in any form
or by any means, electronic or mechanical, Including photocopying and microfilm, without
permission in writing from the publisher.
International Organization for Standardization
Case Postale 56 l CH-1211 Get&e 20 l Switzerland
Printed in Switzerland
ii

---------------------- Page: 2 ----------------------
IS0 84223 991 (E)
Foreword
IS0 (the International Organization for Standardization) is a worldwide
federation of national standards bodies (IS0 member bodies). The work
of preparing International Standards is normally carried out through IS0
technical committees. Each member body interested in a subject for
which a technical committee has been established has the right to be
represented on that committee. International organizations, govern-
mental and non-governmental, in liaison with ISO, also take part in the
work. IS0 collaborates closely with the International Electrotechnical
Commission (IEC) on all matters of electrotechnical standardization.
Draft International Standards adopted by the technical committees are
circulated to the member bodies for voting. Publication as an Interna-
tional Standard requires approval by at least 75 % of the member bodies
casting a vote.
International Standard IS0 8422 was prepared by Technical Committee
ISO/TC 69, Applications of slafistical mefhods, Sub-Committee SC 5,
Acceptance sampling.
Annexes A, B and C form an integral part of this International Standard.
Annex D is for information only.
. . .
III

---------------------- Page: 3 ----------------------
This page intentionally left blank

---------------------- Page: 4 ----------------------
Sequential sampling plans for inspection by attributes
Section 1: General
-
administrative procedures.
1.1 Scope
It contains sampling plans for inspection by attrib-
utes of discrete items. The sampling plans may be
1 .l .l This International Standard specifies se-
used when the extent of nonconformity is expressed
quential sampling plans and procedures for in-
either in terms of proportion (or percent) noncon-
spection by attributes of discrete items.
forming items or in terms of nonconformities per
The plans in the main body of the standard are
item (per 100 items).
indexed in terms of the producer’s risk point and the
The sampling plans are based on the assumption
consumer’s risk point.
that nonconformities occur randomly and with sta-
Annex A specifies sequential sampling plans and
tistical independence. There may be good reasons
procedures indexed in terms of the acceptable
to suspect that one nonconformity in an item could
quality level (AQL) to supplement the system of
be caused by a condition also likely to cause others.
sampling plans in IS0 2859-l.
If so, it may be better to consider the items just as
conforming or not, and ignore multiple nonconform-
The purpose of this International Standard is to pro-
ities.
vide procedures for sequential assessment of in-
spection results that may be used to induce the
The plans in annex A are primarily intended to be
supplier through the economic and psychological
used for inspection of a continuing series of lots
pressure of non-acceptance of lots of inferior quality
from the same production run. The plans in the main
to supply lots of a quality having a high probability
body of this International Standard may also be used
of acceptance. At the same time, the consumer is
for inspection of lots in isolation.
protected by a prescribed upper limit to the proba-
bility of accepting lots of poor quality.
1.2 Normative references
1.1.2 The sampling plans designated in this lnter-
national Standard are applicable, but not limited, to
The following standards contain provisions which,
inspection in different fields, such as
through reference in this text, constitute provisions
-
of this International Standard. At the time of publi-
end items;
cation, the editions indicated were valid. All stan-
-
dards are subject to revision, and parties to
components and raw materials;
agreements based on this International Standard
-
are encouraged to investigate the possibility of ap-
operations;
plying the most recent editions of the standards in-
-
dicated below. Members of IEC and IS0 maintain
materials in process;
registers of currently valid International Standards.
-
supplies in storage;
IS0 2859-l :I 989, Sampling procedures for inspection
-
by attributes -- Part 1: Sampling plans indexed by
maintenance operations;
acceptable quality level (AQL) for lot-by-lot in-
spection.
- data or records;

---------------------- Page: 5 ----------------------
IS0 8422:1991(E)
IS0 3534-l:+ Statistics - Vocabulary and symbols A Acceptance number for sequential sam-
- Part 1: Probability and general statistical terms. pling.
A Acceptance number corresponding to the
IS0 3534-2:-l), Statistics - Vocabulary and symbols t
curtailed value of the cumulative sample
- Part 2: Statistical quality control.
size.
1.3 Definitions and symbols
Consumer’s risk quality level (in percent
CR4
nonconforming).
I .3.1 Definitions
n Cumulative count.
For the purposes of this International Standard, the
Multiplier of the cumulative sample size
8
definitions given in IS0 3534-1, IS0 3534-2 and
that is used to determine the acceptance
IS0 2859-1, together with the following definitions,
and the rejection numbers (slope of the
apply.
acceptance and rejection lines).
3.3.1.1 cumulative count (I>): When sampling in-
h Constant that is used to determine the
A
spection of items from a lot is performed sequen-
acceptance numbers (intercept of the
tially, the total number of nonconforming items
acceptance line).
(nonconformities) found during inspection, counting
h Constant that is used to determine the
from the start of inspection up to, and including, the
R
rejection numbers (intercept of the re-
* item last inspected.
jection line).
1.3.1.2 cumulative sample size (n,,,,,): When sam-
Sample size for a corresponding single
n,
pling inspection of items from a tot is performed se-
g plan.
samplin
quentially, the total of inspected items, counting
from the start of the inspection up to, and including,
n Average sample size.
av
the item last inspected.
n Cumulative sample size.
cum
1.3.1.3 acceptance value for sequential sampling
(A): A value derived from the specified parameters Curtailment value of cumulative
nt
of the sampling plan and the cumulative sample sample size.
size. Whether the lot may yet be accepted is deter-
Lot or process quality level (in proportion
P
mined by comparing the cumulative count with the
nonconforming or nonconformities per
acceptance number.
item).
NOTE 1 In sampling by attributes, acceptance values
NOTE 3 To convert p to percent noncon-
are integers and are referred to as acceptance numbers.
forming or nonconformities per 100 items,
This latter term is used throughout this International
multiply by 100.
Standard.
Producer’s risk quality level. I’a = 1 - cx
1.3.1.4 rejection value for sequential sampling (R): PA
when p = Pi.
A value derived from the specified parameters of the
sampling plan and the cumulative sample size.
Consumer’s risk quality level. Pa = /I
PR
Whether the lot shall yet be considered not accept-
when p = /)R.
able is determined by comparing the cumulative
count with the rejection number.
The probability of acceptance.
P
a
NOTE 2 In sampling by attributes, rejection values are
Producer’s risk quality level (in percent
PRQ
integers and are referred to as rejection numbers. This
nonconforming).
latter term is used throughout this International Standard.
Rejection number for sequential sam-
JR
pling.
1.3.2 Symbols
R Rejection number corresponding to the
The symbols used in this International Standard are t
curtailed value of the cumulative sample
as follows:
size.
A Acceptance number for a corresponding
0
single sampling plan.
1) To be published.
2

---------------------- Page: 6 ----------------------
IS0 8422:1991(E)
If, at a given stage, the cumulative count is such that
a The producer’s risk?
the risk of accepting a lot of an unsatisfactory quality
The consumer’s risk.*)
P level (the consumer’s risk) is sufficiently low, the lot
is considered acceptable and sampling of that lot is
R Index parameter that is used to deter-
terminated.
mine approximations to the OC curve at
general quality levels. (See C.2.2.)
If, on the other hand, the cumulative count is such
that the risk of non-acceptance for a lot of a satis-
1.4 Principle of a sequential sampling factory quality level (the producer’s risk) is suffi-
ciently low, the lot shall be considered not
plan by attributes
acceptable, and sampling of that lot is terminated.
Under a sequential sampling plan by attributes,
If the cumulative count does not allow either of the
items are selected at random and subjected to in-
above decisions to be taken, then an additional item
spection one by one, and a cumulative count is kept
is inspected. The process is continued until sufficient
of the number of nonconforming items (or of the
sample information has been accumulated to war-
number of nonconformities). After the inspection of
rant a decision that the lot is acceptable or not ac-
each item, the cumulative count is used to assess
ceptable.
whether there is sufficient information to sentence
the lot at that stage of the inspection.
2) a and /? may be considered to be the type I and type II risks, respectively, when testing the null hypothesis
against the alternative hypothesis

---------------------- Page: 7 ----------------------
IS0 8422:1991(E)

---------------------- Page: 8 ----------------------
IS0 8422:1991 (E)
Section 2: Choice of sampling plan
2.1.2 Caution
2.1 Choice between sequential, single,
double and multiple sampling plans
The choice between single, double, multiple and
sequential sampling plans shall be made before the
inspection of a lot is started. During the inspection
2.1 .I Advantages and disadvantages of
of a lot, it is not permitted to switch from one type
sequential plans
of sampling plan to another as the operating char-
acteristic of the plan may be drastically changed if
The average sample size is the average of the vari-
the actual inspection results influence the choice of
ous sample sizes which may occur under a sam-
acceptance criterion.
pling plan for a given lot or process quality level.
Like double and multiple sampling plans, the use of
2.1.3 Curtailment of the sample size
sequential sampling plans leads to a smaller aver-
age sample size than single sampling plans having
Although a sequential sampling plan is on average
the same operating characteristic. However, the av-
much more economical than the equivalent single
erage savings are even greater when using a se-
sampling plan, it may occur, during the inspection
quential sampling plan than when a double or
of a particular lot, that acceptance or non-
multiple sampling plan is used.
acceptance comes at a very late stage because the
cumulative count remains between the acceptance
For good quality lots, the savings may reach, or
number and the rejection number for a long time.
even exceed, 50 % compared with a maximum sav-
With the graphical method this corresponds to the
ing of 37 % with double sampling. Annex C gives a
random progress of the step curve remaining in the
method for determining approximate values of the
indecision zone. Such a situation is most likely to
average sample size.
occur when the quality level of the lot (in proportion
nonconforming or in nonconformities per item) is
On the other hand, the actual number of items in-
close to g, where g is the slope of the acceptance
spected for a particular lot when using a double,
and rejection lines.
multiple or sequential sampling plan may exceed
that of the corresponding single sampling plan. For
In order to alleviate this disadvantage, a maximum
double and multiple sampling plans there is an up-
cumulative sample size n, is set before sampling
per limit to the actual number of items to be in-
begins, and inspection is stopped if the cumulative
spected.
sample size reaches the curtailment value, q, with-
out a decision having been made. The acceptance
For sequential sampling plans, there is generally no
or non-acceptance of the lot is then determined in
such limit, and the number of inspected items may
accordance with a rule which is also agreed in ad-
considerably exceed the sample size of the corre-
vance of sampling. The curtailment rules of this
sponding single sampling plan and even exceed the
International Standard have been determined in
lot size. For the sequential sampling plans in this
such a way that the producer’s and consumer’s
International Standard, a curtailment rule (see
risks are hardly affected by this deviation from the
2.1.3) has been introduced in order to limit the po-
principles underlying the statistical theory of se-
tential number of inspected items.
quential sampling inspection. The curtailment rules
As the ultimate sample size from a particular lot is to be used are given in 2.4.2.
.
not known in advance, the selection of the sample
may present organizational difficulties when se-
2.2 Particular reservations on the
quential sampling plans are used. Moreover, the
inspection of small lots
sheduling of inspection operations may present dif-
ficulties when using a double, multiple or sequential
The statistical theory underlying the sequential
sampling plan. A further disadvantage is that the
sampling plans in this International Standard is
execution of a sequential sampling plan is more
based on the assumption that the samples taken
easily misunderstood by the inspectors than the
from the lot are “with replacement”, i.e. each sam-
simpler rules for single sampling.
pled item is replaced before the next item in the
sample is selected. When, as is usual, sampling is
The balance between the advantages of a smaller
without replacement, the theory remains valid for all
average sample size and the organizational dis-
advantages associated with a fluctuating inspection practical purposes if the cumulative sample size
load results in sequential sampling being suitable does not exceed one-tenth of iV, where N denotes
only when inspection of individual items is costly in the lot size; the theory remains approximately valid
even for cumulative sample sizes up to one-seventh
comparison with inspection overheads.
5

---------------------- Page: 9 ----------------------
IS0 8422:1991 (E)
of N. Unfortunately, in contrast to the situation for sumer’s risk points have to be determined from di-
single sampling plans, the actual cumulative sample rect considerations of the conditions under which
size that is necessary in a sequential sampling plan the sampling plan will operate.
will not be known in advance.
In the case of a small lot it is therefore advisable to
2.4 Pre-operation preparations
ensure that the size of the lot is sufficiently large to
allow a curtailed sequential sampling plan to oper-
ate under sampling without replacement, in accord-
2.4.1 Obtaining the parameters /I*, /IR and g
ance with the specified producer’s and consumer’s
risks. For the general sequential sampling plans
The criteria for acceptance or non-acceptance of a
described in 2.3.2 and 2.4.1, it is therefore recom-
lot that are invoked at each stage of the inspection
mended that the lot size exceed 7n,, where q is the
are determined from the parameters /zA, hR and I(.
curtailment value of the sequential sampling plan.
The values of these parameters corresponding to a
If the lot size is not sufficiently large to satisfy producer’s risk of r;r = 0.05, a consumer’s risk of
NOTE 4
the above requirement, both the consumer’s and the pro-
p -_ 0,lO and preferred values of the producer’s and
ducer’s risks will generally become less than their speci-
consumer’s risk quality levels are given in
fied values. If, however, the acceptance number of the
tables I-A and I-B.
corresponding single sampling plan is zero, then the pro-
ducer’s risk may slightly exceed the specified values.
Annex B gives general procedures for determining
/zA, h, and (4 for any combination of producer’s and
consumer’s risk points.
EXAMPLE
2.3 Selection of a sampling plan
The specification for a type of electrical insulator is
given as the nominal withstand value 1000 kV. An
2.3.1 Plans matching those of IS0 2859-l
inspection agency is using a single sampling plan
with sample size 65 and acceptance number 6 for
If it is required to find a sequential sampling plan
determination of the acceptability of production lots
matching a plan from IS0 2859-1:1989, then
of this insulator type. Since the build-up of the test
annex A may be used. Annex A contains sequential
voltage is consuming in both time and energy, it has
sampling plans indexed by acceptable quality level
been decided that a sequential sampling plan
(AQL) and sample size code letter. The operating
should be used in future to determine the accepta-
characteristic curves of these sequential sampling
bility of production lots of the insulator. The se-
plans match, as closely as practicable, those of the
quential sampling plan is to have an operating
corresponding plans in IS0 2859-l.
characteristic similar to that of the single sampling
plan being used.
2.3.2 General plans
proper-
The single sampling plan has the following
The general method described in 2.3.2 and in 2.4.1 ties:
is used when the requirements of the sequential
-
of the insulators from the production fail
if 5 9/b
sampling plan are specified in terms of two points
at the nominal voltage then the probability of ac-
on the operating characteristic curve of the plan.
ceptinq a lot is 095:
The point corresponding to the higher probability of
\
acceptance shall be designated the prodllcer’s risk
--
if 16 o/o of the insulators from the production fail
point; the other shall be designated the cor,stln,er’s
at the nominal voltage then the probability of ac-
risk point.
cepting the lot is 0.10.
The first step when designing a sequential sampling
These requirements correspond to fixing
plan is to choose these two points, if they have not
already been dictated by circumstances. For this
a) the producer’s risk quality (PRQ) at 5 % with
purpose, a producer’s risk of oc = 0,05 and a con-
95 % of lots expected to be accepted; i.e. such
sumer’s risk of p = 0,lO are often used. (See
that the producer’s risk is 5 %, or O! = 0,05.
figure 1.)
When the desired sequential sampling plan is re-
b) the consumer’s risk quality (CRQ) at 16 % non-
quired to have approximately the same operating
conforming with 10 % of lots expected to be ac-
characteristic as an existing single, double or multi-
cepted; i.e. such that the consumer’s risk is
ple sampling plan, the producer’s risk point and the
IO Oh, or /I = OJO.
consumer’s risk point may be read off from a graph
The requirements are indicated on the graph of the
or a table of the operating characteristic of that plan.
operating characteristic curve in figure 1.
When no such plan exists, the producer’s and con-
6

---------------------- Page: 10 ----------------------
IS0 8422:1991(E)
rounded up to the nearest integer and, in the
case of inspection for nonconformities per
100 items, 17~ is determined as
rounded up to the nearest integer.
2.4.2.2 Truncation for small lots
If the resulting value of q exceeds the lot size, then
the sequential sampling plan shall be used with the
curtailment value nt of the sample size equal to the
lot size.
15 loop
4 0 5 10
2.4.2.3 Example
Processquallty
I
I
CRQ level (In percent
PRQ
noncontormfng)
Consider the sequential sampling plan for inspection
for percent nonconforming with parameters
Figure 1 - Operating characteristic curve for a
= 1,750, ItR = 2,247 and g = 0,095 7 that were de-
/IA
sampling plan with producer’s risk TV = 0,05 and
termined in the example given in 2.4.1. The plan was
consumer’s risk /‘I = 0,lO
chosen to match the single sampling plan with
rb = 65 and A, = 6.
The curtailment value for the cumulative sample
From table I-A it is found that the parameters of the
size is therefore q = 98.
sequential sampling plan that satisfies these re-
quirements are Had the corresponding single sampling plan not
been known, the curtailment value would have been
h A = -1,750
determined by 2.4.2.1 b). Substituting the values of
/IA, h, and g in the formula in 2.4.2.1 b) leads to the
h, = 2,247
curtailed sample size n, = 91.
g = 0,0957
The s ame v alues could have been found by calcu-
2.4.3 Choosing the form of the sampling plan
lation using the procedu re given in annex B.
This International Standard gives two methods of
operating a sequential sampling plan: a numerical
method and a graphical method.
2.4.2 Determining the curtailment value of the
sample size
The numerical method has the advantage of being
accurate, thereby avoiding disputes about accept-
ance or non-acceptance.
2.4.2.1 Standard procedures
The graphical method is well suited to the inspection
a) If the sample size 4 of the single sampling plan
of series of lots, as the chart needs only to be drawn
that is equivalent to the sequential sampling plan
once, but the method is less accurate due to the in-
under consideration is known, the curtailment accuracy inherent in plotting points and in drawing
value for the cumulative sample size is deter-
straight lines. On the other hand, the method does
mined as nt = 1,5%, rounded up to the nearest
have the advantage of displaying the increase in the
integer.
information on the quality of the lot as additional
items are inspected, information being represented
b) If the sample size of the equivalent single sam-
by the progress of a broken line within the in-
pling plan is not known, the curtailment value
decision zone until the line reaches, or crosses, one
under inspection for percent nonconforming is
of the boundaries of that zone.
determined as
The numerical method is the standard method, so
2 hAhR
far as acceptance or non-acceptance of a lot is con-
I+= g(1 -g)
cerned. See the caution in 3.4.2.

---------------------- Page: 11 ----------------------
IS0 8422:1991(E)
The smallest cumulative sample size permitting ac-
2.4.3.1 Numerical method
ceptance of the lot is obtained by rounding /zA/g up
to the nearest integer.
For each value, %,,,,, of the cumulative sample size
that is less than the curtailment value of the sample
The smallest cumulative sample size permitting
size, the acceptance number A is found by rounding
non-acceptance of the lot under inspection for pro-
the quantity
portion nonconforming is obtained by rounding
. . . .
8%um - hA (2 1) h,/(l -g) up to the nearest integer.
The rejection number EXAMPLE
dow n to the nearest integer.
R is foun d by roundin g the qu antity
For the sequential sampling plan with parameters
. . . .
(2 2)
8Ylcum + 43 h = 1,750, h, = 2,247 and g = 0,095 7 that were de-
&mined in the example given in 2.4.1, the curtail-
up to the nearest integer.
ment value of the sample size was determined in the
example given in 2.4.2.3 to be q = 98. The corre-
acceptance number, A,, corresponding to the
The
sponding acceptance number is found by rounding
curt ailed sample size is determined as
gq = 9,38 down to the nearest integer, hence the
acceptance number /I, is 9 and the rejection number
A,=wt
R, is IO.
rounded down to the nearest integer.
The formula for the acceptance number /1 becomes
The corresponding rejection number is calculated
0,095 7&-.um - 1,750
as
rounded down to the nearest integer, and the for-
R, = A, + 1
mula for the rejection number R becomes
Whenever the value of equation (2.1) is negative, the
0,095 7n,,, + 2,247
cumulative sample size is too small to allow ac-
ceptance of the lot. Conversely, whenever the value
of equation (2.2) is larger than the cumulative sam-
ple size, the cumulative sample size is too small to rounded up to the nearest integer.
permit non-acceptance of the lot under inspection
The acceptance and rejection numbers correspond-
for proportion nonconforming.
ing to the cumulative sample sizes lzcun, = 1, 2, . . . . 97
are determined by successively inserting the values
The quantities given by equations (2.1) and (2.2)
in these formulae and rounding the result as
shall be determined to three decimal places before
Of ncum
described above. The result is shown in figure2.
rounding.

---------------------- Page: 12 ----------------------
IS0 8422:1991(E)
Cumulative sample
Acceptance number Rejection number
mum - h* !F cum + 17F?
size
n [equation (2.1)] A R
[equation (2.2)]
cum
1 - 1,654 * 2,343 *t
2 - 1,559 * 2,438 **
3 - 1,463 * 2,534 3
4 - 1,367 * 2,630 3
5 - 1,272 * 2,726 3
6 - 1,176 * 2,82 1 3
7 - 1,080 * 2,917 3
8 - 0,985 * 3,013 4
9 - 0,889 4 3,108 4
10 - 0,793 A 3,204 4
11 - 0,697 * 3,300 4
12 - 0,602 * 3,395 4
13 - 0,506 * 3,491 4
14 - 0,410 A 3,587 4
15 - 0,315 * 3,683 4
16 - 0,219 * 3,778 4
17 - 0,123 * 3,874 4
18 - 0,028 * 3,970 4
19 0,068 0 4,065 5
20 0,164 0 4,161 5
. .
. . .
. . . .
l
;
9; 7,;33 11,530 1;
98 - 9 - 10
*
indicates that the cumulative sample size is too small to permit acceptance.
*’ indicates that the cumulative sample size is too small to permit non-acceptance.
Figure 2 - Inspection record sheet for the sequential sampling plan considered in the example given in 2.4.3.1
The acceptance zone is the zone below (and in-
2.4.3.2 Graphical method
cluding) the acceptance line, together with that
Prepare a graph as shown in figure3, with the cu-
part of the curtailment line that is below (and in-
mulative sample size as the horizontal axis, the cu-
cludes) the point (q; A,).
mulative count as the vertical axis, and with the
quantities given by equations (2.1) and (2.2) repres- The rejection zone is the zone above (and in-
ented by two straight lines with the same slope, g. cluding) the rejection line together with that part
The lower line, with intercept -hA, is designated the of the curtailment line that is above (and in-
acceptance line, and the upper line, with intercept cludes) the point (I+; R,).
h,, is designat
...

SLOVENSKI STANDARD
SIST ISO 8422:1996
01-september-1996
Sequential sampling plans for inspection by attributes
Sequential sampling plans for inspection by attributes
Plans d'échantillonnage progressif pour le contrôle par attributs
Ta slovenski standard je istoveten z: ISO 8422:1991
ICS:
03.120.30 8SRUDEDVWDWLVWLþQLKPHWRG Application of statistical
methods
SIST ISO 8422:1996 en
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.

---------------------- Page: 1 ----------------------

SIST ISO 8422:1996

---------------------- Page: 2 ----------------------

SIST ISO 8422:1996
INTERNATIONAL
8422
STANDARD
First edition
1991-11-15
Sequential sampling plans for inspection by
attributes
Plans d’&hntillonnage progressifs pour le contr6le par attributs
----
____________-_-.___--_-----------
I_
----- ----
-- ----
Reference number
-.-_________- - IS0 8422:1991(E)
---_- -- _ _ __ .__

---------------------- Page: 3 ----------------------

SIST ISO 8422:1996
IS0 8422:1991 (E)
,
Contents
Page
Section 1 General ,,.,.,,.,~.,.,.,,.,.,,.,,.,.,.~.~. 1
1.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .-.,. 1
1.2 Normative references . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I
2
1.3 Definitions and symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . 3
1.4 Principle of a sequential sampling plan by attributes
5
Section 2 Choice of sampling plan .,.,.,.,.,.,.
2.1 Choice between sequential, single, double and multiple sampling
plans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
. . . . . . . 5
2.2 Particular reservations on the inspection of small lots
2.3 Selection of a sampling plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 Pre-operation preparations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
,.,,.,. 11
Section 3 Operation of a sequential sampling plan
............... ......................................... 11
3.1 Specification of the plan
............................. .............................. 11
3.2 Drawing of the sample
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3 Cumulative count .,.
3.4 Determination of acceptability .,,.,.,. 11
13
3.5 Operating characteristic curves and average sample size
,.,. 13
3.6 Validity of approximations
Annexes
A Sequential sampling plans corresponding to IS0 2859-l sampling
,. 18
plans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .~.*.
B Determination of the parameters of a sequential sampling
40
plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C Calculation of the operating characteristic curve and average
41
sample size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
D Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0 IS0 1991
All rights reserved. No part of this publication may be reproduced or utilized in any form
or by any means, electronic or mechanical, Including photocopying and microfilm, without
permission in writing from the publisher.
International Organization for Standardization
Case Postale 56 l CH-1211 Get&e 20 l Switzerland
Printed in Switzerland
ii

---------------------- Page: 4 ----------------------

SIST ISO 8422:1996
IS0 84223 991 (E)
Foreword
IS0 (the International Organization for Standardization) is a worldwide
federation of national standards bodies (IS0 member bodies). The work
of preparing International Standards is normally carried out through IS0
technical committees. Each member body interested in a subject for
which a technical committee has been established has the right to be
represented on that committee. International organizations, govern-
mental and non-governmental, in liaison with ISO, also take part in the
work. IS0 collaborates closely with the International Electrotechnical
Commission (IEC) on all matters of electrotechnical standardization.
Draft International Standards adopted by the technical committees are
circulated to the member bodies for voting. Publication as an Interna-
tional Standard requires approval by at least 75 % of the member bodies
casting a vote.
International Standard IS0 8422 was prepared by Technical Committee
ISO/TC 69, Applications of slafistical mefhods, Sub-Committee SC 5,
Acceptance sampling.
Annexes A, B and C form an integral part of this International Standard.
Annex D is for information only.
. . .
III

---------------------- Page: 5 ----------------------

SIST ISO 8422:1996
This page intentionally left blank

---------------------- Page: 6 ----------------------

SIST ISO 8422:1996
Sequential sampling plans for inspection by attributes
Section 1: General
-
administrative procedures.
1.1 Scope
It contains sampling plans for inspection by attrib-
utes of discrete items. The sampling plans may be
1 .l .l This International Standard specifies se-
used when the extent of nonconformity is expressed
quential sampling plans and procedures for in-
either in terms of proportion (or percent) noncon-
spection by attributes of discrete items.
forming items or in terms of nonconformities per
The plans in the main body of the standard are
item (per 100 items).
indexed in terms of the producer’s risk point and the
The sampling plans are based on the assumption
consumer’s risk point.
that nonconformities occur randomly and with sta-
Annex A specifies sequential sampling plans and
tistical independence. There may be good reasons
procedures indexed in terms of the acceptable
to suspect that one nonconformity in an item could
quality level (AQL) to supplement the system of
be caused by a condition also likely to cause others.
sampling plans in IS0 2859-l.
If so, it may be better to consider the items just as
conforming or not, and ignore multiple nonconform-
The purpose of this International Standard is to pro-
ities.
vide procedures for sequential assessment of in-
spection results that may be used to induce the
The plans in annex A are primarily intended to be
supplier through the economic and psychological
used for inspection of a continuing series of lots
pressure of non-acceptance of lots of inferior quality
from the same production run. The plans in the main
to supply lots of a quality having a high probability
body of this International Standard may also be used
of acceptance. At the same time, the consumer is
for inspection of lots in isolation.
protected by a prescribed upper limit to the proba-
bility of accepting lots of poor quality.
1.2 Normative references
1.1.2 The sampling plans designated in this lnter-
national Standard are applicable, but not limited, to
The following standards contain provisions which,
inspection in different fields, such as
through reference in this text, constitute provisions
-
of this International Standard. At the time of publi-
end items;
cation, the editions indicated were valid. All stan-
-
dards are subject to revision, and parties to
components and raw materials;
agreements based on this International Standard
-
are encouraged to investigate the possibility of ap-
operations;
plying the most recent editions of the standards in-
-
dicated below. Members of IEC and IS0 maintain
materials in process;
registers of currently valid International Standards.
-
supplies in storage;
IS0 2859-l :I 989, Sampling procedures for inspection
-
by attributes -- Part 1: Sampling plans indexed by
maintenance operations;
acceptable quality level (AQL) for lot-by-lot in-
spection.
- data or records;

---------------------- Page: 7 ----------------------

SIST ISO 8422:1996
IS0 8422:1991(E)
IS0 3534-l:+ Statistics - Vocabulary and symbols A Acceptance number for sequential sam-
- Part 1: Probability and general statistical terms. pling.
A Acceptance number corresponding to the
IS0 3534-2:-l), Statistics - Vocabulary and symbols t
curtailed value of the cumulative sample
- Part 2: Statistical quality control.
size.
1.3 Definitions and symbols
Consumer’s risk quality level (in percent
CR4
nonconforming).
I .3.1 Definitions
n Cumulative count.
For the purposes of this International Standard, the
Multiplier of the cumulative sample size
8
definitions given in IS0 3534-1, IS0 3534-2 and
that is used to determine the acceptance
IS0 2859-1, together with the following definitions,
and the rejection numbers (slope of the
apply.
acceptance and rejection lines).
3.3.1.1 cumulative count (I>): When sampling in-
h Constant that is used to determine the
A
spection of items from a lot is performed sequen-
acceptance numbers (intercept of the
tially, the total number of nonconforming items
acceptance line).
(nonconformities) found during inspection, counting
h Constant that is used to determine the
from the start of inspection up to, and including, the
R
rejection numbers (intercept of the re-
* item last inspected.
jection line).
1.3.1.2 cumulative sample size (n,,,,,): When sam-
Sample size for a corresponding single
n,
pling inspection of items from a tot is performed se-
g plan.
samplin
quentially, the total of inspected items, counting
from the start of the inspection up to, and including,
n Average sample size.
av
the item last inspected.
n Cumulative sample size.
cum
1.3.1.3 acceptance value for sequential sampling
(A): A value derived from the specified parameters Curtailment value of cumulative
nt
of the sampling plan and the cumulative sample sample size.
size. Whether the lot may yet be accepted is deter-
Lot or process quality level (in proportion
P
mined by comparing the cumulative count with the
nonconforming or nonconformities per
acceptance number.
item).
NOTE 1 In sampling by attributes, acceptance values
NOTE 3 To convert p to percent noncon-
are integers and are referred to as acceptance numbers.
forming or nonconformities per 100 items,
This latter term is used throughout this International
multiply by 100.
Standard.
Producer’s risk quality level. I’a = 1 - cx
1.3.1.4 rejection value for sequential sampling (R): PA
when p = Pi.
A value derived from the specified parameters of the
sampling plan and the cumulative sample size.
Consumer’s risk quality level. Pa = /I
PR
Whether the lot shall yet be considered not accept-
when p = /)R.
able is determined by comparing the cumulative
count with the rejection number.
The probability of acceptance.
P
a
NOTE 2 In sampling by attributes, rejection values are
Producer’s risk quality level (in percent
PRQ
integers and are referred to as rejection numbers. This
nonconforming).
latter term is used throughout this International Standard.
Rejection number for sequential sam-
JR
pling.
1.3.2 Symbols
R Rejection number corresponding to the
The symbols used in this International Standard are t
curtailed value of the cumulative sample
as follows:
size.
A Acceptance number for a corresponding
0
single sampling plan.
1) To be published.
2

---------------------- Page: 8 ----------------------

SIST ISO 8422:1996
IS0 8422:1991(E)
If, at a given stage, the cumulative count is such that
a The producer’s risk?
the risk of accepting a lot of an unsatisfactory quality
The consumer’s risk.*)
P level (the consumer’s risk) is sufficiently low, the lot
is considered acceptable and sampling of that lot is
R Index parameter that is used to deter-
terminated.
mine approximations to the OC curve at
general quality levels. (See C.2.2.)
If, on the other hand, the cumulative count is such
that the risk of non-acceptance for a lot of a satis-
1.4 Principle of a sequential sampling factory quality level (the producer’s risk) is suffi-
ciently low, the lot shall be considered not
plan by attributes
acceptable, and sampling of that lot is terminated.
Under a sequential sampling plan by attributes,
If the cumulative count does not allow either of the
items are selected at random and subjected to in-
above decisions to be taken, then an additional item
spection one by one, and a cumulative count is kept
is inspected. The process is continued until sufficient
of the number of nonconforming items (or of the
sample information has been accumulated to war-
number of nonconformities). After the inspection of
rant a decision that the lot is acceptable or not ac-
each item, the cumulative count is used to assess
ceptable.
whether there is sufficient information to sentence
the lot at that stage of the inspection.
2) a and /? may be considered to be the type I and type II risks, respectively, when testing the null hypothesis
against the alternative hypothesis

---------------------- Page: 9 ----------------------

SIST ISO 8422:1996
IS0 8422:1991(E)

---------------------- Page: 10 ----------------------

SIST ISO 8422:1996
IS0 8422:1991 (E)
Section 2: Choice of sampling plan
2.1.2 Caution
2.1 Choice between sequential, single,
double and multiple sampling plans
The choice between single, double, multiple and
sequential sampling plans shall be made before the
inspection of a lot is started. During the inspection
2.1 .I Advantages and disadvantages of
of a lot, it is not permitted to switch from one type
sequential plans
of sampling plan to another as the operating char-
acteristic of the plan may be drastically changed if
The average sample size is the average of the vari-
the actual inspection results influence the choice of
ous sample sizes which may occur under a sam-
acceptance criterion.
pling plan for a given lot or process quality level.
Like double and multiple sampling plans, the use of
2.1.3 Curtailment of the sample size
sequential sampling plans leads to a smaller aver-
age sample size than single sampling plans having
Although a sequential sampling plan is on average
the same operating characteristic. However, the av-
much more economical than the equivalent single
erage savings are even greater when using a se-
sampling plan, it may occur, during the inspection
quential sampling plan than when a double or
of a particular lot, that acceptance or non-
multiple sampling plan is used.
acceptance comes at a very late stage because the
cumulative count remains between the acceptance
For good quality lots, the savings may reach, or
number and the rejection number for a long time.
even exceed, 50 % compared with a maximum sav-
With the graphical method this corresponds to the
ing of 37 % with double sampling. Annex C gives a
random progress of the step curve remaining in the
method for determining approximate values of the
indecision zone. Such a situation is most likely to
average sample size.
occur when the quality level of the lot (in proportion
nonconforming or in nonconformities per item) is
On the other hand, the actual number of items in-
close to g, where g is the slope of the acceptance
spected for a particular lot when using a double,
and rejection lines.
multiple or sequential sampling plan may exceed
that of the corresponding single sampling plan. For
In order to alleviate this disadvantage, a maximum
double and multiple sampling plans there is an up-
cumulative sample size n, is set before sampling
per limit to the actual number of items to be in-
begins, and inspection is stopped if the cumulative
spected.
sample size reaches the curtailment value, q, with-
out a decision having been made. The acceptance
For sequential sampling plans, there is generally no
or non-acceptance of the lot is then determined in
such limit, and the number of inspected items may
accordance with a rule which is also agreed in ad-
considerably exceed the sample size of the corre-
vance of sampling. The curtailment rules of this
sponding single sampling plan and even exceed the
International Standard have been determined in
lot size. For the sequential sampling plans in this
such a way that the producer’s and consumer’s
International Standard, a curtailment rule (see
risks are hardly affected by this deviation from the
2.1.3) has been introduced in order to limit the po-
principles underlying the statistical theory of se-
tential number of inspected items.
quential sampling inspection. The curtailment rules
As the ultimate sample size from a particular lot is to be used are given in 2.4.2.
.
not known in advance, the selection of the sample
may present organizational difficulties when se-
2.2 Particular reservations on the
quential sampling plans are used. Moreover, the
inspection of small lots
sheduling of inspection operations may present dif-
ficulties when using a double, multiple or sequential
The statistical theory underlying the sequential
sampling plan. A further disadvantage is that the
sampling plans in this International Standard is
execution of a sequential sampling plan is more
based on the assumption that the samples taken
easily misunderstood by the inspectors than the
from the lot are “with replacement”, i.e. each sam-
simpler rules for single sampling.
pled item is replaced before the next item in the
sample is selected. When, as is usual, sampling is
The balance between the advantages of a smaller
without replacement, the theory remains valid for all
average sample size and the organizational dis-
advantages associated with a fluctuating inspection practical purposes if the cumulative sample size
load results in sequential sampling being suitable does not exceed one-tenth of iV, where N denotes
only when inspection of individual items is costly in the lot size; the theory remains approximately valid
even for cumulative sample sizes up to one-seventh
comparison with inspection overheads.
5

---------------------- Page: 11 ----------------------

SIST ISO 8422:1996
IS0 8422:1991 (E)
of N. Unfortunately, in contrast to the situation for sumer’s risk points have to be determined from di-
single sampling plans, the actual cumulative sample rect considerations of the conditions under which
size that is necessary in a sequential sampling plan the sampling plan will operate.
will not be known in advance.
In the case of a small lot it is therefore advisable to
2.4 Pre-operation preparations
ensure that the size of the lot is sufficiently large to
allow a curtailed sequential sampling plan to oper-
ate under sampling without replacement, in accord-
2.4.1 Obtaining the parameters /I*, /IR and g
ance with the specified producer’s and consumer’s
risks. For the general sequential sampling plans
The criteria for acceptance or non-acceptance of a
described in 2.3.2 and 2.4.1, it is therefore recom-
lot that are invoked at each stage of the inspection
mended that the lot size exceed 7n,, where q is the
are determined from the parameters /zA, hR and I(.
curtailment value of the sequential sampling plan.
The values of these parameters corresponding to a
If the lot size is not sufficiently large to satisfy producer’s risk of r;r = 0.05, a consumer’s risk of
NOTE 4
the above requirement, both the consumer’s and the pro-
p -_ 0,lO and preferred values of the producer’s and
ducer’s risks will generally become less than their speci-
consumer’s risk quality levels are given in
fied values. If, however, the acceptance number of the
tables I-A and I-B.
corresponding single sampling plan is zero, then the pro-
ducer’s risk may slightly exceed the specified values.
Annex B gives general procedures for determining
/zA, h, and (4 for any combination of producer’s and
consumer’s risk points.
EXAMPLE
2.3 Selection of a sampling plan
The specification for a type of electrical insulator is
given as the nominal withstand value 1000 kV. An
2.3.1 Plans matching those of IS0 2859-l
inspection agency is using a single sampling plan
with sample size 65 and acceptance number 6 for
If it is required to find a sequential sampling plan
determination of the acceptability of production lots
matching a plan from IS0 2859-1:1989, then
of this insulator type. Since the build-up of the test
annex A may be used. Annex A contains sequential
voltage is consuming in both time and energy, it has
sampling plans indexed by acceptable quality level
been decided that a sequential sampling plan
(AQL) and sample size code letter. The operating
should be used in future to determine the accepta-
characteristic curves of these sequential sampling
bility of production lots of the insulator. The se-
plans match, as closely as practicable, those of the
quential sampling plan is to have an operating
corresponding plans in IS0 2859-l.
characteristic similar to that of the single sampling
plan being used.
2.3.2 General plans
proper-
The single sampling plan has the following
The general method described in 2.3.2 and in 2.4.1 ties:
is used when the requirements of the sequential
-
of the insulators from the production fail
if 5 9/b
sampling plan are specified in terms of two points
at the nominal voltage then the probability of ac-
on the operating characteristic curve of the plan.
ceptinq a lot is 095:
The point corresponding to the higher probability of
\
acceptance shall be designated the prodllcer’s risk
--
if 16 o/o of the insulators from the production fail
point; the other shall be designated the cor,stln,er’s
at the nominal voltage then the probability of ac-
risk point.
cepting the lot is 0.10.
The first step when designing a sequential sampling
These requirements correspond to fixing
plan is to choose these two points, if they have not
already been dictated by circumstances. For this
a) the producer’s risk quality (PRQ) at 5 % with
purpose, a producer’s risk of oc = 0,05 and a con-
95 % of lots expected to be accepted; i.e. such
sumer’s risk of p = 0,lO are often used. (See
that the producer’s risk is 5 %, or O! = 0,05.
figure 1.)
When the desired sequential sampling plan is re-
b) the consumer’s risk quality (CRQ) at 16 % non-
quired to have approximately the same operating
conforming with 10 % of lots expected to be ac-
characteristic as an existing single, double or multi-
cepted; i.e. such that the consumer’s risk is
ple sampling plan, the producer’s risk point and the
IO Oh, or /I = OJO.
consumer’s risk point may be read off from a graph
The requirements are indicated on the graph of the
or a table of the operating characteristic of that plan.
operating characteristic curve in figure 1.
When no such plan exists, the producer’s and con-
6

---------------------- Page: 12 ----------------------

SIST ISO 8422:1996
IS0 8422:1991(E)
rounded up to the nearest integer and, in the
case of inspection for nonconformities per
100 items, 17~ is determined as
rounded up to the nearest integer.
2.4.2.2 Truncation for small lots
If the resulting value of q exceeds the lot size, then
the sequential sampling plan shall be used with the
curtailment value nt of the sample size equal to the
lot size.
15 loop
4 0 5 10
2.4.2.3 Example
Processquallty
I
I
CRQ level (In percent
PRQ
noncontormfng)
Consider the sequential sampling plan for inspection
for percent nonconforming with parameters
Figure 1 - Operating characteristic curve for a
= 1,750, ItR = 2,247 and g = 0,095 7 that were de-
/IA
sampling plan with producer’s risk TV = 0,05 and
termined in the example given in 2.4.1. The plan was
consumer’s risk /‘I = 0,lO
chosen to match the single sampling plan with
rb = 65 and A, = 6.
The curtailment value for the cumulative sample
From table I-A it is found that the parameters of the
size is therefore q = 98.
sequential sampling plan that satisfies these re-
quirements are Had the corresponding single sampling plan not
been known, the curtailment value would have been
h A = -1,750
determined by 2.4.2.1 b). Substituting the values of
/IA, h, and g in the formula in 2.4.2.1 b) leads to the
h, = 2,247
curtailed sample size n, = 91.
g = 0,0957
The s ame v alues could have been found by calcu-
2.4.3 Choosing the form of the sampling plan
lation using the procedu re given in annex B.
This International Standard gives two methods of
operating a sequential sampling plan: a numerical
method and a graphical method.
2.4.2 Determining the curtailment value of the
sample size
The numerical method has the advantage of being
accurate, thereby avoiding disputes about accept-
ance or non-acceptance.
2.4.2.1 Standard procedures
The graphical method is well suited to the inspection
a) If the sample size 4 of the single sampling plan
of series of lots, as the chart needs only to be drawn
that is equivalent to the sequential sampling plan
once, but the method is less accurate due to the in-
under consideration is known, the curtailment accuracy inherent in plotting points and in drawing
value for the cumulative sample size is deter-
straight lines. On the other hand, the method does
mined as nt = 1,5%, rounded up to the nearest
have the advantage of displaying the increase in the
integer.
information on the quality of the lot as additional
items are inspected, information being represented
b) If the sample size of the equivalent single sam-
by the progress of a broken line within the in-
pling plan is not known, the curtailment value
decision zone until the line reaches, or crosses, one
under inspection for percent nonconforming is
of the boundaries of that zone.
determined as
The numerical method is the standard method, so
2 hAhR
far as acceptance or non-acceptance of a lot is con-
I+= g(1 -g)
cerned. See the caution in 3.4.2.

---------------------- Page: 13 ----------------------

SIST ISO 8422:1996
IS0 8422:1991(E)
The smallest cumulative sample size permitting ac-
2.4.3.1 Numerical method
ceptance of the lot is obtained by rounding /zA/g up
to the nearest integer.
For each value, %,,,,, of the cumulative sample size
that is less than the curtailment value of the sample
The smallest cumulative sample size permitting
size, the acceptance number A is found by rounding
non-acceptance of the lot under inspection for pro-
the quantity
portion nonconforming is obtained by rounding
. . . .
8%um - hA (2 1) h,/(l -g) up to the nearest integer.
The rejection number EXAMPLE
dow n to the nearest integer.
R is foun d by roundin g the qu antity
For the sequential sampling plan with parameters
. . . .
(2 2)
8Ylcum + 43 h = 1,750, h, = 2,247 and g = 0,095 7 that were de-
&mined in the example given in 2.4.1, the curtail-
up to the nearest integer.
ment value of the sample size was determined in the
example given in 2.4.2.3 to be q = 98. The corre-
acceptance number, A,, corresponding to the
The
sponding acceptance number is found by rounding
curt ailed sample size is determined as
gq = 9,38 down to the nearest integer, hence the
acceptance number /I, is 9 and the rejection number
A,=wt
R, is IO.
rounded down to the nearest integer.
The formula for the acceptance number /1 becomes
The corresponding rejection number is calculated
0,095 7&-.um - 1,750
as
rounded down to the nearest integer, and the for-
R, = A, + 1
mula for the rejection number R becomes
Whenever the value of equation (2.1) is negative, the
0,095 7n,,, + 2,247
cumulative sample size is too small to allow ac-
ceptance of the lot. Conversely, whenever the value
of equation (2.2) is larger than the cumulative sam-
ple size, the cumulative sample size is too small to rounded up to the nearest integer.
permit non-acceptance of the lot under inspection
The acceptance and rejection numbers correspond-
for proportion nonconforming.
ing to the cumulative sample sizes lzcun, = 1, 2, . . . . 97
are determined by successively inserting the values
The quantities given by equations (2.1) and (2.2)
in these formulae and rounding the result as
shall be determined to three decimal places before
Of ncum
described above. The result is shown in figure2.
rounding.

---------------------- Page: 14 ----------------------

SIST ISO 8422:1996
IS0 8422:1991(E)
Cumulative sample
Acceptance number Rejection number
mum - h* !F cum + 17F?
size
n [equation (2.1)] A R
[equation (2.2)]
cum
1 - 1,654 * 2,343 *t
2 - 1,559 * 2,438 **
3 - 1,463 * 2,534 3
4 - 1,367 * 2,630 3
5 - 1,272 * 2,726 3
6 - 1,176 * 2,82 1 3
7 - 1,080 * 2,917 3
8 - 0,985 * 3,013 4
9 - 0,889 4 3,108 4
10 - 0,793 A 3,204 4
11 - 0,697 * 3,300 4
12 - 0,602 * 3,395 4
13 - 0,506 * 3,491 4
14 - 0,410 A 3,587 4
15 - 0,315 * 3,683 4
16 - 0,219 * 3,778 4
17 - 0,123 * 3,874 4
18 - 0,028 * 3,970 4
19 0,068 0 4,065 5
20 0,164 0 4,161 5
. .
. . .
. . . .
l
;
9; 7,;33 11,530 1;
98 - 9 - 10
*
indicates that the cumulative sample size is too small to permit acceptance.
*’ indicates that the cumulative sample size is too small
...

NORME
INTERNATIONALE
Première édition
19914145
Plans d’échantillonnage progressif pour le
contrôle par attributs
Sequential sampling plans for inspection by attributes
Numéro de référence
ISO 8422:1991(F)

---------------------- Page: 1 ----------------------
ISO 8422:1991 (F)
Sommaire
Page
Généralités . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Section 1
1.1 Domaine d’application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Références normatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Définitions et symboles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Principe du plan d’échantillonnage progressif par attributs 3
Section 2 Choix d’un plan d’échantillonnage . . . .1.1.11.1. 5
Choix entre des plans d’échantillonnage progressif, simple,
2.1
double et multiple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
5
2.2 Réserves particulières sur le contrôle de petits lots . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Sélection d’un plan d’échantillonnage
6
2.4 Opérations préliminaires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mise en œuvre d’un plan d’échantillonnage progressif 11
Section 3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1 Spécification du plan
3.2 Prélèvement de l’échantillon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
11
3.3 Résultat du cumul . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4 Détermination de I’acceptabilité . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.5 Courbes d’efficacité et effectif moyen d’échantillon . . . . . . . . . . . 12
3.6 Validité des approximations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Annexes
A Plans d’échantillonnage progressif correspondant aux plans
18
d’échantillonnage de I’ISO 2859-l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
B Détermination des paramètres pour un plan d’échantillonnage
progressif . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
C Calcul de la courbe d’efficacité et de l’effectif moyen
d’échantillon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
......................................................... .................. 47
D Bibliographie
0 ISO 1991
Droits de reproduction réservés. Aucune partie de cette publication ne peut être reproduite
ni utilisee sous quelque forme que ce soit et par aucun procédé, électronique ou mécanique,
y compris la photocopie et les microfilms, sans l’accord écrit de l’éditeur.
Organisation internationale de normalisation
Case Postale 56 l CH-l 211 Genéve 20 l Suisse
Version française tiree en 1993
Imprimé en Suisse
ii

---------------------- Page: 2 ----------------------
ISO 8422:1991 (F)
Avant-propos
L’ISO (Organisation internationale de normalisation) est une fédération
mondiale d’organismes nationaux de normalisation (comités membres
de I’ISO). L’elaboration des Normes internationales est en général
confiée aux comités techniques de I’ISO. Chaque comité membre inté-
ressé par une etude a le droit de faire partie du comité technique créé
à cet effet. Les organisations internationales, gouvernementales et non
gouvernementales, en liaison avec I’ISO participent également aux tra-
vaux. L’ISO collabore étroitement avec la Commission électrotechnique
internationale (CEI) en ce qui concerne la normalisation électrotech-
nique.
Les projets de Normes internationales adoptes par les comités techni-
ques sont soumis aux comités membres pour vote. Leur publication
comme Normes internationales requiert l’approbation de 75 % au moins
des comités membres votants.
La Norme internationale ISO 8422 a eté elaborée par le comité techni-
que ISO/TC 69, Application des méthodes statistiques, sous-comité SC
5, Échantillonnage en vue d’acceptation.
Les annexes A, 6 et C font partie intégrante de la présente Norme
internationale. L’annexe D est donnée uniquement à titre d’information.

---------------------- Page: 3 ----------------------
Page blanche

---------------------- Page: 4 ----------------------
-_
~~ -~-
NORME INTERNATIONALE ISO 8422:1991 (F)
Plans d’échantillonnage progressif pour le contrôle par
attributs
Section 1: Généralités
-
opérations d’entretien;
1 A Domaine d’application
-
informations ou enregistrements;
1 .l .l La présente Norme internationale présente
des plans et des règles d’échantillonnage progressif
- procédures administratives.
pour le contrôle par attributs d’unités discrètes.
La présente Norme internationale contient des plans
Les plans dans le corps principal de la norme sont
d’échantillonnage pour le contrôle par attributs
indexés en termes de point du risque fournisseur et
d’unités discrétes. Les plans d’échantillonnage sont
de point du risque client.
utilisables lorsque l’évaluation de la non-conformité
s’exprime soit en termes de proportion (ou pour-
L’annexe A donne des plans et des régles
centage) d’unités non conformes, soit en termes de
d’échantillonnage progressifs indexés en termes de
non-conformités par unité (par 100 unités).
niveau de qualité acceptable (NQA) en complément
du système des plans d’échantillonnage de
Les plans d’échantillonnage sont fondés sur I’hypo-
I’ISO 2859-l.
thèse que les non-conformités surviennent de facon
aléatoire et sont statistiquement indépendantes. *
Le but de la présente Norme internationale est de
fournir des règles basées sur la détermination pro-
Il peut exister des raisons de supposer qu’une non-
gressive des résultats de contrôle, afin de persua-
conformité dans une unité puisse être provoquée
der le fournisseur, par des pressions économiques
par une condition et également en provoquer pro-
et psychologiques, de la non-acceptation de lots de
bablement d’autres. Dans ce cas, mieux vaut consi-
qualité inférieure, pour qu’il fournisse des lots de
dérer les unités comme conformes ou non et ignorer
qualité ayant une forte probabilité d’acceptation. En
les non-conformités multiples.
même temps le client est protégé par une limite
supérieure imposée de la probabilité d’accepter des
Les plans de l’annexe A sont destinés princi-
lots de faible qualité.
palement au contrôle de séries continues de lots
provenant du même cycle de production. Les plans
1.1.2 Les plans d’échantillonnage contenus dans
dans le corps principal de la présente Norme inter-
la présente Norme internationale sont applicables,
nationale peuvent également être utilisés pour le
notamment mais d’une manière non limitative, aux
contrôle de lots isolés.
contrôles ci-après:
produits finis;
1.2 Références normatives -
composants ou matières premières;
Les normes suivantes contiennent des dispositions
qui, par suite de la référence qui en est faite,
opérations; ’
constituent des dispositions valables pour la pré-
matériaux en cours de fabrication; sente Norme internationale. Au moment de la pu-
blication, les editions indiquées étaient en vigueur.
fournitures en stock; Toute norme est sujette à révision et les parties
1

---------------------- Page: 5 ----------------------
ISO 8422:1991 (F)
prenantes des accords fondés sur la présente NOTE 2 En échantillonnage par attributs, les valeurs de
rejet sont des entiers et sont appelées critères de rejet.
Norme internationale sont invitées à rechercher la
C’est ce dernier terme qui est utilisé dans le corps de la
possibilité d’appliquer les éditions les plus récentes
présente Norme internationale.
des normes indiquées ci-après. Les membres de la
CE1 et de I’ISO possèdent le registre des Normes
internationales en vigueur à un moment donné.
1.3.2 Symboles
ISO 2859-l :1989, Règles d’échantillonnage pour les
contrôles par attributs - Partie 1: Plans d’échan-
Les symboles utilisés dans la présente Norme
tillonnage pour /es confrôles lot par lot, indexés
inter nationale s Iont les s uivants:
d’après le niveau de qualité acceptable (NQA).
A Critère d’acceptation pour le plan
0
ISO 3534-l : -‘1, Statistique - d’échantillonnage simple correspondant
Vocabulaire et symbo-
- Partie 1: Probabilité et termes statistiques gé-
les
A Critère d’acceptation pour I’échantillon-
néraux.
nage progressif
ISO 3534-2:--l), Statistique - Vocabulaire et symbo-
A Critére d’acceptation correspondant à la
t
/es - Partie 2: Maîtrise statistique de la qualité,
valeur de troncage de l’effectif cumulé
d’échantillon
1.3 Définitions et symboles
D Résultat du cumul
Facteur de l’effectif cumulé d’échantillon,
g
1.3.1 Définitions
utilisé pour déterminer les critères de
rejet et d’acceptation (pente de la ligne
Pour les besoins de la présente Norme internatio-
d’acceptation et de la ligne de rejet)
nale, les définitions données dans I’ISO 3534-1,
I’ISO 3534-2 et I’ISO 2859-1, et les définitions sui-
h Constante utilisée pour déterminer les
A
vantes s’appliquent.
critéres d’acceptation (ordonnée à I’ori-
gine de la ligne d’acceptation)
1.3.1.1 résultat du cumul (D): Lorsque le contrôle
par échantillonnage d’unités d’un lot est effectué
Constante utilisée pour déterminer les
43
progressivement, nombre total d’unités non confor-
critères de rejet (ordonnée à l’origine de
mes (de non-conformités) trouvées pendant ce
la ligne de rejet)
contrôle, comptées à partir du début du contrôle
Effectif d’échantillon pour un plan
jusqu’à la dernière unité contrôlée incluse. no
d’échantillonnage simple correspondant
1.3.1.2 effectif cumulé d’échantillon (4&: Lorsque
Effectif moyen d’échantillon
SI
le contrôle par échantillonnage d’unités d’un lot est
effectué progressivement, total des unités contrô-
Effectif cumulé d’échantillon
Ilcum
lées, comptées à partir du début du contrôle jusqu’à
la dernière unité contrôlée incluse. Valeur de troncage de l’effectif cumulé
nt
d’échantillon
1.3.1.3 valeur d’acceptation pour l’échantillonnage
Niveau de qualité du lot ou processus (en
P
progressif (A): Valeur dépendant des paramètres
proportion de non conformes ou de non-
spécifiés du plan d’échantillonnage et de l’effectif
conformités par unité)
cumulé d’échantillon. L’acceptation du lot est alors
déterminée par la comparaison du résultat du cumul
NOTE 3 Pour convertir p en pourcentage
avec le critère d’acceptation.
de non conformes ou de non-conformités par
100 unités, multiplier par 100.
En kchantillonnage par attributs, les valeurs
NOTE 1
d’acceptation sont des entiers et sont appelées critères
Niveau de qualité du risque fournisseur.
d’acceptation. C’est ce dernier terme qui est utilise dans PA
P a=l-tXCJUaIldjJ=pA
le corps de la présente Norme internationale.
Niveau de qualité du risque client.
PR
1.3.1.4 valeur de rejet pour l’échantillonnage pro-
P,=/Iquandp=p,
gressif (R): Nombre dépendant des paramétres
spécifiés du plan d’échantillonnage et de l’effectif
P Probabilite d’acceptation
a
cumulé d’échantillon. L’acceptation du lot est alors
déterminée par la comparaison du résultat du cumul
QRC Niveau de qualité du risque client (en
avec le critère de rejet.
pourcentage de non conformes)
1) A publier.
2

---------------------- Page: 6 ----------------------
ISO 8422:1991 (F)
Niveau de qualité du risque fournisseur partir du nombre d’unités non conformes (ou du
QRF
(en pourcentage de non conformes) nombre de non-conformités). Après le contrôle de
chaque unité, le resultat du cumul est utilisé pour
Critère de rejet pour l’échantillonnage
décider si l’information est suffisante pour juger le
progressif
lot à ce stade du contrôle.
Critère de rejet correspondant à la valeur
Si, a un stade donne, le résultat du cumul est tel que
de troncage de l’effectif cumule d’échan-
le risque d’accepter un lot d’un niveau de qualité
tillon
insuffisant (le risque client) est suffisamment bas, le
lot est consideré acceptable et l’échantillonnage de
a Risque fournisseur*)
ce lot est terminé.
Risque client*)
P
Si, d’autre part, le résultat du cumul est tel que le
risque de non-acceptation pour un lot de niveau de
Paramètre utilise pour déterminer les
qualité satisfaisant (le risque fournisseur) est suffi-
approximations de la courbe d’efficacité
samment bas, le lot doit être considéré non-
pour des niveaux de qualité généraux
acceptable et l’échantillonnage de ce lot est
(voir C.2.2)
terminé.
1.4 Principe du plan d’échantillonnage
Si le resultat du cumul ne permet de prendre au-
progressif par attributs cune des deux décisions ci-dessus, alors une unité
supplementaire est contrôlée. Le processus est
Dans un plan d’échantillonnage progressif par attri- continué jusqu’à ce que l’information nécessaire ait
buts, les unités sont prélevées au hasard et soumi- ete accumulée afin de pouvoir prendre une décision
ses au contrôle une par une; un cumul est fait à sur I’acceptabilité ou la non-acceptabilité du lot.
f--, : p = PR

---------------------- Page: 7 ----------------------
ISO 8422:î 991 (F)

---------------------- Page: 8 ----------------------
ISO 8422:i 991 (F)
Section 2: Choix d’un plan d’échantillonnage
organisationnelles associées à des contrôles fluc-
2.1
Choix entre des plans
tuant est que l’échantillonnage progressif n’est ap-
d’échantillonnage progressif, simple,
proprie que lorsque le contrôle des unités discrètes
double et multiple
est couteux par rapport aux frais généraux de
contrôle.
2.1.1 Avantages et inconvénients des plans
d’échantillonnage progressif
2.1.2 Avertissement
L’effectif moyen d’échantillon est la moyenne des
Le choix entre des plans d’échantillonnage simple,
différents effectifs d’échantillon possibles qui peu-
double, multiple et progressif doit être fait avant le
vent être obtenus pour un plan d’échantillonnage
début du contrôle. II n’est pas possible pendant le
pour une qualité d’un lot ou processus donné.
contrôle d’un lot de changer un type de plan
Comme pour des plans d’échantillonnage double ou
d’échantillonnage contre un autre, car l’efficacité du
multiple, l’utilisation des plans d’échantillonnage
plan peut être radicalement modifiée si les résultats
progressif conduit a un effectif moyen d’échantillon
du controle réel influencent le choix du critère d’ac-
inférieur a celui qu’il faut pour un plan d’échan-
ceptation.
tillonnage simple ayant la même efficacité. Cepen-
dant, les économies moyennes sont également plus
2.1.3 Troncage de l’effectif d’échantillon
importantes lors de l’utilisation d’un plan progressif
plutôt que d’un plan double ou multiple.
Bien qu’un plan d’échantillonnage progressif soit en
moyenne plus économique que le plan d’échan-
Pour des lots de bonne qualité, les économies peu-
tillonnage simple équivalent, il peut arriver pendant
vent atteindre ou même excéder 50 %, comparées
le contrôle d’un lot particulier que l’acceptation ou
à une économie maximum de 37 % avec un échan-
à un stade très tardif, le
la non-acceptation arrive
tillonnage double. L’annexe C donne une méthode
résultat du cumul restant compris très longtemps
pour déterminer les valeurs approximatives de I’ef-
entre les criteres d’acceptation et de rejet. Pour la
fectif moyen d’echantillon.
méthode graphique cela se traduit par la progres-
sion aléatoire de la courbe dans la zone d’indéci-
D’autre part, le nombre réel d’unités contrôlées
sion du lot. Une telle situation est d’autant plus
pour un lot particulier en plan double, multiple ou
probable que le niveau de qualité du lot (en propor-
progressif peut dépasser celui correspondant à un
tion de non conformes ou en non-conformités par
plan d’échantillonnage simple. Pour des plans
unité) est proche de g, où g représente la pente des
d’échantillonnage double et multiple, il existe une
droites d’acceptation et de rejet.
limite supérieure au nombre réel d’unités à contro-
Ier .
Afin de pallier cet inconvénient, un effectif cumulé
d’échantillon maximal 4 est établi avant le début de
Pour les plans d’échantillonnage progressif, il
l’échantillonnage et le contrôle est arrêté si l’effectif
n’existe généralement pas une te.lle limite et le
cumulé d’échantillon atteint. la valeur de troncage,
nombre d’unités contrôlées peut considérablement
I+, sans qu’une décision ait été prise. L’acceptation
excéder l’effectif d’échantillon du plan d’échan-
ou la non-acceptation du lot est alors déterminée
tillonnage simple correspondant et même dépasser
selon une règle décidée également avant I’échan-
l’effectif du lot. Pour les plans d’échantillonnage
tillonnage. Les règles du troncage dans la présente
progressif de la présente Norme internationale, une
Norme internationale ont été déte,rminé.es de telle
règle de troncage (voir 2.1.3) a été introduite afin de
facon que les risques client et fournisseur soient à
limiter le nombre potentiel d’unités contrôlées.
peine affectés pa.r cet écart aux principes sous-
Puisque l’effectif final d’échantillon n’est pas connu
jacents de la théorie statistique sur les plans
à l’avance, la sélection de l’échantillon peut donc
d’échantillonnage progressif. Les règles de
présenter des difficultés organisationnelles. De plus,
troncage à utiliser sont données en 2.4.2.
l’utilisation d’un plan d’échantillonnage double,
multiple ou progressif peut entraîner des difficultés
2.2 Réserves particulières sur le contrôle
de planification des opérations de contrôle. Un autre
de petits lots
inconvénient est que la mise en œuvre d’un plan
d’échantillonnage progressif est plus facilement mal
La théorie statistique sous-jacente aux plans
interprétée par les contrôleurs que les règles plus
d’échantillonnage progressif de la présente Norme
faciles de l’échantillonnage simple.
internationale est fondée sur l’hypothèse que les
Le résultat de la balance entre l’intérêt d’un effectif
khantillons sont tirés du lot <, c’est-
moyen d’echantillon plus petit et les difficultés a-dire que chaque unité échantillonnée est remise

---------------------- Page: 9 ----------------------
ISO 8422:1991(F)
dans le lot avant la sélection de l’unité suivante. les circonstances. Dans cette perspective, le risque
Lorsque, comme souvent, l’échantillonnage est sans fournisseur, a = 0,05, et le risque client, /II = O,lO,
remise, la théorie reste valable dans tous les cas sont souvent utilisés (voir figure 1).
pratiques, si l’effectif cumulé d’échantillon n’excède
Lorsque le plan d’échantillonnage doit avoir ap-
pas un dixième de N, où N est la taille du lot; la
proximativement la même efficacité qu’un plan
théorie reste approximativement valable même pour
d’échantillonnage simple, double ou multiple exis-
des effectifs cumulés d’échantillon atteignant un
tant, le point du risque fournisseur et le point du
septième de N. Malheureusement, au contraire de
risque client sont choisis à l’aide d’un graphique ou
la situation d’un plan d’échantillonnage simple, I’ef-
d’une table de l’efficacité de ce plan. Quand un tel
fectif cumulé d’échantillon réel qui est nécessaire
plan n’existe pas, les points des risques client et
dans un plan d’échantillonnage progressif ne sera
fournisseur doivent être déterminés à partir des
pas connu à l’avance.
considérations directes des conditions dans les-
Par conséquent, dans le cas d’un petit lot, il est re-
quelles l’échantillonnage sera effectue.
commandé de s’assurer que l’effectif du lot est suf-
fisamment grand pour permettre une mise en œuvre
2.4 Opérations préliiminaires
d’un plan d’échantillonnage progressif tronqué avec
échantillonnage sans remise selon les risques four-
nisseur et client spécifiés. Pour les plans d’échan-
2.4.1 Obtention des paramètres hA, h, et g
tillonnage progressifs généraux décrits en 2.3.2 et
2.4.1, il est alors recommandé que l’effectif du lot
Les critères d’acceptation ou de non-acceptation
dépasse 7q, où q est la valeur du troncage du plan
d’un lot utilisés a chaque stade du controle sont
d’échantillonnage progressif.
déterminés a partir des paramètres hA, k et g.
NOTE 4 Si l’effectif du lot n’est pas suffisamment grand Les valeurs de ces paramétres correspondant à un
pour satisfaire à la condition ci-dessus, les risques client
risque fournisseur, 6c = 0,05, un risque client,
et fournisseur deviendront généralement inférieurs à
/? = 0,lO et des valeurs recommandées des niveaux
leurs valeurs spécifiées. Cependant, si le critère d’accep-
de qualité du risque client et du risque fournisseur
tation du plan d’échantillonnage simple correspondant est
sont données dans les tables 1-A et I-B.
zéro, alors le risque fournisseur peut excéder faiblement
les valeurs spécifiées.
L’annexe B donne des méthodes générales pour la
détermination de hA, h, et g pour toutes combinai-
2.3 Sélection d’un plan d’échantillonnage sons des points des risques fournisseur et client.
EXEMPLE
2.3.1 Plans équivalents aux plans de
I’ISO 2859-l
La spécifrcation d’un type d’isolateur électrique est
donnée par la valeur de résistance nominale de
S’il est nécessaire d’obtenir un plan d’échantillon-
1 000 kV. Une agence de contrôle utilise un plan
nage progressif équivalent à un plan de I’ISO 2859-1,
d’échantillonnage simple avec un effectif d’échan-
l’annexe A peut être utilisée. L’annexe A contient
tillon de 65 et un critère d’acceptation égal à 6 pour
des plans d’échantillonnage progressif indexés par
déterminer I’acceptabilité de la production des lots
le niveau de qualité acceptable (NQA) et la Iettre-
de ce type d’isolateur. Puisque la construction des
code de l’effectif d’échantillon. Les courbes d’effi-
essais de voltage consomme a la fois temps et
cacité de ces plans d’échantillonnage progressif
énergie, il a été décidé d’utiliser dans le futur des
sont aussi proches que possible de celles des plans
plans d’échantillonnage progressif pour déterminer
correspondants de I’ISO 2859-l.
I’acceptabilité de la production des lots d’isolateurs.
Le plan d’échantillonnage progressif doit avoir une
efficacité similaire au plan d’échantillonnage simple
2.3.2 Plans généraux
utilisé.
La méthode générale décrite en 2.3.2 et 2.4.1 est
Le plan d ‘echanti Ilonnage simple pos sède les pro-
utilisée lorsque les données d’un plan progressif
.
étés s U ivantes:
Pr1
sont spécifiées par deux points de la courbe d’effi-
cacité du plan. Le point correspondant à la plus
- si 5 % des isolateurs produits n’atteignent pas
haute probabilité d’acceptation doit être appelé le
la tension nominale, alors la probabilité d’ac-
point du risque fournisseur, l’autre doit être appelé
cepter le lot est de 0,95;
le point du risque client.
La première étape dans la détermination d’un plan - si 16 % des isolateurs produits n’atteignent pas
d’échantillonnage progressif est le choix de ces la tension nominale, alors la probabilité d’ac-
deux points si ceux-ci n’ont pas déjà été dictés par cepter le lot est de 0,lO.

---------------------- Page: 10 ----------------------
ISO 8422:1991(F)
I
#
80 -
60 -
PI
I I
1 0 5 10 15
WOP
Nfveau de qualttt! du processus
I
QRF QRC (en pourcentage de non conformes)
Figure 1 - Courbe d’efficacité pour un plan d’échantillonnage avec un risque fournisseur a = 0,05 et un risque
client p = 0,lO
Ces conditions correspondent à fixer
calculée par r+ = 1,5 4, arrondi au plus proche
entier supérieur.
a) la qualité du risque fournisseur (QRF) a 5 %
avec 95 % attendu de lots acceptés; c’est-à-dire
b) Si l’effectif d’echantillon du plan d’échantillon-
telle que le risque fournisseur est 5 %, ou
nage simple correspondant n’est pas connu, la
a = 0,05; valeur de troncage pour le contrôle de la pro-
portion de non conformes est déterminée par:
b) la qualité du risque client (QRF) à 16 % de non
2hAhR
conformes avec 10 % attendu de lots acceptes;
c’est-à-dire telle que le risque client est 10 %, q= gu -g>
ou a = 0,lO.
arrondi au plus proche entier supérieur, et dans
Ces conditions sont indiquées sur le graphique de le cas d’un contrôle de non-conformités par
la courbe d’efficacité en figure 1. unité, q est déterminé par
2hAhR
Les paramètres du plan d’échantillonnage progres-
=-
nt
sif, trouvés dans la table I-A, qui satisfont à ces g
conditions, sont:
arrondi au plus proche entier supérieur.
h A = 1,750
2.4.2.2 Troncage pour de petits lots
& = 2,247
Si la valeur calculée de I+ est supérieure à l’effectif
g = 0,095 7
du lot, alors la valeur de troncage I+ de l’effectif
d’echantillon égale a l’effectif du lot doit être utilisée
Les mêmes valeurs peuvent être calculées à l’aide
de la méthode donnée dans l’annexe B. pour le plan d’échantillonnage progressif.
2.4.2.3 Exemple
2.4.2 Détermination de la valeur de troncage
de l’effectif d’échantillon Considerons le plan d’échantillonnage progressif
pour le contrôle de pourcentage de non conformes
avec pour paramétres hA= 1,750, h, = 2,247 et
2.4.2.1 Procédures normales
g= 0,095 7, détermines dans l’exemple donné en
2.4.1. Le plan a été choisi pour correspondre au plan
a) Si l’effectif d’échantillon h du plan d’échan-
d’échantillonnage simple avec 4 = 65 et A, = 6.
tillonnage simple correspondant au plan
d’échantillonnage progressif est connu, la valeur La valeur de troncage pour l’effectif cumulé
de troncage de l’effectif cumulé d’échantillon est d’échantillon est alors 4 = 98.

---------------------- Page: 11 ----------------------
ISO 8422:1991(F)
Le critère de rejet correspondant est calculé par
Si le plan d’échantillonnage simple n’avait pas été
connu, la valeur de troncage aurait été déterminée
x, = A, + 1
selon 2.4.2.1 b). La substitution des valeurs de hA,
h, et g dans la formule de 2.4.2.1 b) donne l’effectif
Partout où la valeur de l’expression (2.1) est néga-
d’échantillon tronqué I+ = 91.
tive, l’effectif cumulé d’échantillon est trop petit pour
permettre l’acceptation du lot. Inversement, partout
où la valeur de l’expression (2.2) est supérieure à
2.4.3 Choix de la forme du plan
l’effectif cumule d’echantillon, celui-ci est trop petit
d’échantillonnage
pour permettre une non-acceptation du lot contrôlé
pour la proportion de non conformes.
La présente Norme internationale donne deux mé-
thodes de mise en œuvre d’un plan d’échantillon-
Les grandeurs données par les expressions (2.1) et
nage progressif: une méthode numérique et une
(2.2) doivent être déterminées avec trois décimales
méthode graphique.
avant d’être arrondies.
La méthode numérique a l’avantage d’être précise
L’effectif minimal cumulé d’échantillon permettant
et d’éviter ainsi des discussions sur l’acceptation
l’acceptation du lot est obtenu en arrondissant hA/g
la non-acceptation.
ou
au plus proche entier supérieur.
La méthode graphique est bien adaptée au contrôle
L’effectif minimal cumulé d’echantillon permettant
séries de lots, car la carte n’a besoin d’être
de
la non-acceptation du lot contrôlé pour la proportion
dessinée qu’une fois, mais elle est moins précise à
de non conformes est obtenu en arrondissant
cause de l’imprécision inhérente au marquage des
I&/(l -g) au plus proche entier supérieur.
points et dessin de droites. D’autre part, la méthode
a l’avantage de visualiser l’évolution de I’infor- EXEMPLE
mation sur la qualité du lot lorsque des unités sup-
Pour le plan d’échantillonnage progressif ayant pour
plémentaires sont contrôlées, l’information étant
paramètres hA = 1,750, & = 2,247 et g = 0,095 7 dé-
représentée par la progression d’une ligne brisée à
termines dans l’exemple donné en 2.4.1, la valeur
l’intérieur de la zone d’indécision, jusqu’à ce que
de troncage de l’effectif d’échantillon a éte détermi-
cette ligne atteigne ou dépasse une des limites de
née dans l’exemple donne en 2.4.2.3 comme étant
cette zone.
I+ = 98. Le critére d’acceptation correspondant est
La méthode numérique est la méthode de référence
calcule en arrondissant gn, = 9,38 au plus proche
dans la mesure où cela concerne l’acceptation ou entier inferieur; le critére d’acceptation A, est donc
la non-acceptation d’un lot. Voir l’avertissement en 9 et le critère de rejet Rt est 10.
3.4.2.
La formule pour le critère d’acceptation A devient
0,095 7 IZcgjm - 1,750
2.4.3.1 Méthode numérique
arrondi au plus proche entier inférieur, et la formule
Pour chaque valeur, hum, de l’effectif cumule
du critère de rejet R devient
d’échantillon, qui est inférieure à la valeur de
troncage de l’effectif d’échantillon, le critère d’ac-
0,095 7 %um + 2,247
ceptation A est trouvé en arrondissant la grandeur
arrondi au plus proche entier supérieur.
h . . . .
8n cum - A (2 1)
Les critères d’acceptation et de rejet correspondant
au plus proche entier inferie ur. Le critére de rejet
aux effectifs cumulés d’échantillon %um = 1, 2, . . . . 97
R est trouvé en arrond issant randeur
la cl
sont successivement déterminés en remplaçant
. . . . qum par sa valeur dans ces formules et en arron-
Pcum + k (2 2)
d
...

NORME
INTERNATIONALE
Première édition
19914145
Plans d’échantillonnage progressif pour le
contrôle par attributs
Sequential sampling plans for inspection by attributes
Numéro de référence
ISO 8422:1991(F)

---------------------- Page: 1 ----------------------
ISO 8422:1991 (F)
Sommaire
Page
Généralités . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Section 1
1.1 Domaine d’application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Références normatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Définitions et symboles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Principe du plan d’échantillonnage progressif par attributs 3
Section 2 Choix d’un plan d’échantillonnage . . . .1.1.11.1. 5
Choix entre des plans d’échantillonnage progressif, simple,
2.1
double et multiple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
5
2.2 Réserves particulières sur le contrôle de petits lots . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Sélection d’un plan d’échantillonnage
6
2.4 Opérations préliminaires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Mise en œuvre d’un plan d’échantillonnage progressif 11
Section 3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1 Spécification du plan
3.2 Prélèvement de l’échantillon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
11
3.3 Résultat du cumul . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4 Détermination de I’acceptabilité . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.5 Courbes d’efficacité et effectif moyen d’échantillon . . . . . . . . . . . 12
3.6 Validité des approximations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Annexes
A Plans d’échantillonnage progressif correspondant aux plans
18
d’échantillonnage de I’ISO 2859-l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
B Détermination des paramètres pour un plan d’échantillonnage
progressif . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
C Calcul de la courbe d’efficacité et de l’effectif moyen
d’échantillon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
......................................................... .................. 47
D Bibliographie
0 ISO 1991
Droits de reproduction réservés. Aucune partie de cette publication ne peut être reproduite
ni utilisee sous quelque forme que ce soit et par aucun procédé, électronique ou mécanique,
y compris la photocopie et les microfilms, sans l’accord écrit de l’éditeur.
Organisation internationale de normalisation
Case Postale 56 l CH-l 211 Genéve 20 l Suisse
Version française tiree en 1993
Imprimé en Suisse
ii

---------------------- Page: 2 ----------------------
ISO 8422:1991 (F)
Avant-propos
L’ISO (Organisation internationale de normalisation) est une fédération
mondiale d’organismes nationaux de normalisation (comités membres
de I’ISO). L’elaboration des Normes internationales est en général
confiée aux comités techniques de I’ISO. Chaque comité membre inté-
ressé par une etude a le droit de faire partie du comité technique créé
à cet effet. Les organisations internationales, gouvernementales et non
gouvernementales, en liaison avec I’ISO participent également aux tra-
vaux. L’ISO collabore étroitement avec la Commission électrotechnique
internationale (CEI) en ce qui concerne la normalisation électrotech-
nique.
Les projets de Normes internationales adoptes par les comités techni-
ques sont soumis aux comités membres pour vote. Leur publication
comme Normes internationales requiert l’approbation de 75 % au moins
des comités membres votants.
La Norme internationale ISO 8422 a eté elaborée par le comité techni-
que ISO/TC 69, Application des méthodes statistiques, sous-comité SC
5, Échantillonnage en vue d’acceptation.
Les annexes A, 6 et C font partie intégrante de la présente Norme
internationale. L’annexe D est donnée uniquement à titre d’information.

---------------------- Page: 3 ----------------------
Page blanche

---------------------- Page: 4 ----------------------
-_
~~ -~-
NORME INTERNATIONALE ISO 8422:1991 (F)
Plans d’échantillonnage progressif pour le contrôle par
attributs
Section 1: Généralités
-
opérations d’entretien;
1 A Domaine d’application
-
informations ou enregistrements;
1 .l .l La présente Norme internationale présente
des plans et des règles d’échantillonnage progressif
- procédures administratives.
pour le contrôle par attributs d’unités discrètes.
La présente Norme internationale contient des plans
Les plans dans le corps principal de la norme sont
d’échantillonnage pour le contrôle par attributs
indexés en termes de point du risque fournisseur et
d’unités discrétes. Les plans d’échantillonnage sont
de point du risque client.
utilisables lorsque l’évaluation de la non-conformité
s’exprime soit en termes de proportion (ou pour-
L’annexe A donne des plans et des régles
centage) d’unités non conformes, soit en termes de
d’échantillonnage progressifs indexés en termes de
non-conformités par unité (par 100 unités).
niveau de qualité acceptable (NQA) en complément
du système des plans d’échantillonnage de
Les plans d’échantillonnage sont fondés sur I’hypo-
I’ISO 2859-l.
thèse que les non-conformités surviennent de facon
aléatoire et sont statistiquement indépendantes. *
Le but de la présente Norme internationale est de
fournir des règles basées sur la détermination pro-
Il peut exister des raisons de supposer qu’une non-
gressive des résultats de contrôle, afin de persua-
conformité dans une unité puisse être provoquée
der le fournisseur, par des pressions économiques
par une condition et également en provoquer pro-
et psychologiques, de la non-acceptation de lots de
bablement d’autres. Dans ce cas, mieux vaut consi-
qualité inférieure, pour qu’il fournisse des lots de
dérer les unités comme conformes ou non et ignorer
qualité ayant une forte probabilité d’acceptation. En
les non-conformités multiples.
même temps le client est protégé par une limite
supérieure imposée de la probabilité d’accepter des
Les plans de l’annexe A sont destinés princi-
lots de faible qualité.
palement au contrôle de séries continues de lots
provenant du même cycle de production. Les plans
1.1.2 Les plans d’échantillonnage contenus dans
dans le corps principal de la présente Norme inter-
la présente Norme internationale sont applicables,
nationale peuvent également être utilisés pour le
notamment mais d’une manière non limitative, aux
contrôle de lots isolés.
contrôles ci-après:
produits finis;
1.2 Références normatives -
composants ou matières premières;
Les normes suivantes contiennent des dispositions
qui, par suite de la référence qui en est faite,
opérations; ’
constituent des dispositions valables pour la pré-
matériaux en cours de fabrication; sente Norme internationale. Au moment de la pu-
blication, les editions indiquées étaient en vigueur.
fournitures en stock; Toute norme est sujette à révision et les parties
1

---------------------- Page: 5 ----------------------
ISO 8422:1991 (F)
prenantes des accords fondés sur la présente NOTE 2 En échantillonnage par attributs, les valeurs de
rejet sont des entiers et sont appelées critères de rejet.
Norme internationale sont invitées à rechercher la
C’est ce dernier terme qui est utilisé dans le corps de la
possibilité d’appliquer les éditions les plus récentes
présente Norme internationale.
des normes indiquées ci-après. Les membres de la
CE1 et de I’ISO possèdent le registre des Normes
internationales en vigueur à un moment donné.
1.3.2 Symboles
ISO 2859-l :1989, Règles d’échantillonnage pour les
contrôles par attributs - Partie 1: Plans d’échan-
Les symboles utilisés dans la présente Norme
tillonnage pour /es confrôles lot par lot, indexés
inter nationale s Iont les s uivants:
d’après le niveau de qualité acceptable (NQA).
A Critère d’acceptation pour le plan
0
ISO 3534-l : -‘1, Statistique - d’échantillonnage simple correspondant
Vocabulaire et symbo-
- Partie 1: Probabilité et termes statistiques gé-
les
A Critère d’acceptation pour I’échantillon-
néraux.
nage progressif
ISO 3534-2:--l), Statistique - Vocabulaire et symbo-
A Critére d’acceptation correspondant à la
t
/es - Partie 2: Maîtrise statistique de la qualité,
valeur de troncage de l’effectif cumulé
d’échantillon
1.3 Définitions et symboles
D Résultat du cumul
Facteur de l’effectif cumulé d’échantillon,
g
1.3.1 Définitions
utilisé pour déterminer les critères de
rejet et d’acceptation (pente de la ligne
Pour les besoins de la présente Norme internatio-
d’acceptation et de la ligne de rejet)
nale, les définitions données dans I’ISO 3534-1,
I’ISO 3534-2 et I’ISO 2859-1, et les définitions sui-
h Constante utilisée pour déterminer les
A
vantes s’appliquent.
critéres d’acceptation (ordonnée à I’ori-
gine de la ligne d’acceptation)
1.3.1.1 résultat du cumul (D): Lorsque le contrôle
par échantillonnage d’unités d’un lot est effectué
Constante utilisée pour déterminer les
43
progressivement, nombre total d’unités non confor-
critères de rejet (ordonnée à l’origine de
mes (de non-conformités) trouvées pendant ce
la ligne de rejet)
contrôle, comptées à partir du début du contrôle
Effectif d’échantillon pour un plan
jusqu’à la dernière unité contrôlée incluse. no
d’échantillonnage simple correspondant
1.3.1.2 effectif cumulé d’échantillon (4&: Lorsque
Effectif moyen d’échantillon
SI
le contrôle par échantillonnage d’unités d’un lot est
effectué progressivement, total des unités contrô-
Effectif cumulé d’échantillon
Ilcum
lées, comptées à partir du début du contrôle jusqu’à
la dernière unité contrôlée incluse. Valeur de troncage de l’effectif cumulé
nt
d’échantillon
1.3.1.3 valeur d’acceptation pour l’échantillonnage
Niveau de qualité du lot ou processus (en
P
progressif (A): Valeur dépendant des paramètres
proportion de non conformes ou de non-
spécifiés du plan d’échantillonnage et de l’effectif
conformités par unité)
cumulé d’échantillon. L’acceptation du lot est alors
déterminée par la comparaison du résultat du cumul
NOTE 3 Pour convertir p en pourcentage
avec le critère d’acceptation.
de non conformes ou de non-conformités par
100 unités, multiplier par 100.
En kchantillonnage par attributs, les valeurs
NOTE 1
d’acceptation sont des entiers et sont appelées critères
Niveau de qualité du risque fournisseur.
d’acceptation. C’est ce dernier terme qui est utilise dans PA
P a=l-tXCJUaIldjJ=pA
le corps de la présente Norme internationale.
Niveau de qualité du risque client.
PR
1.3.1.4 valeur de rejet pour l’échantillonnage pro-
P,=/Iquandp=p,
gressif (R): Nombre dépendant des paramétres
spécifiés du plan d’échantillonnage et de l’effectif
P Probabilite d’acceptation
a
cumulé d’échantillon. L’acceptation du lot est alors
déterminée par la comparaison du résultat du cumul
QRC Niveau de qualité du risque client (en
avec le critère de rejet.
pourcentage de non conformes)
1) A publier.
2

---------------------- Page: 6 ----------------------
ISO 8422:1991 (F)
Niveau de qualité du risque fournisseur partir du nombre d’unités non conformes (ou du
QRF
(en pourcentage de non conformes) nombre de non-conformités). Après le contrôle de
chaque unité, le resultat du cumul est utilisé pour
Critère de rejet pour l’échantillonnage
décider si l’information est suffisante pour juger le
progressif
lot à ce stade du contrôle.
Critère de rejet correspondant à la valeur
Si, a un stade donne, le résultat du cumul est tel que
de troncage de l’effectif cumule d’échan-
le risque d’accepter un lot d’un niveau de qualité
tillon
insuffisant (le risque client) est suffisamment bas, le
lot est consideré acceptable et l’échantillonnage de
a Risque fournisseur*)
ce lot est terminé.
Risque client*)
P
Si, d’autre part, le résultat du cumul est tel que le
risque de non-acceptation pour un lot de niveau de
Paramètre utilise pour déterminer les
qualité satisfaisant (le risque fournisseur) est suffi-
approximations de la courbe d’efficacité
samment bas, le lot doit être considéré non-
pour des niveaux de qualité généraux
acceptable et l’échantillonnage de ce lot est
(voir C.2.2)
terminé.
1.4 Principe du plan d’échantillonnage
Si le resultat du cumul ne permet de prendre au-
progressif par attributs cune des deux décisions ci-dessus, alors une unité
supplementaire est contrôlée. Le processus est
Dans un plan d’échantillonnage progressif par attri- continué jusqu’à ce que l’information nécessaire ait
buts, les unités sont prélevées au hasard et soumi- ete accumulée afin de pouvoir prendre une décision
ses au contrôle une par une; un cumul est fait à sur I’acceptabilité ou la non-acceptabilité du lot.
f--, : p = PR

---------------------- Page: 7 ----------------------
ISO 8422:î 991 (F)

---------------------- Page: 8 ----------------------
ISO 8422:i 991 (F)
Section 2: Choix d’un plan d’échantillonnage
organisationnelles associées à des contrôles fluc-
2.1
Choix entre des plans
tuant est que l’échantillonnage progressif n’est ap-
d’échantillonnage progressif, simple,
proprie que lorsque le contrôle des unités discrètes
double et multiple
est couteux par rapport aux frais généraux de
contrôle.
2.1.1 Avantages et inconvénients des plans
d’échantillonnage progressif
2.1.2 Avertissement
L’effectif moyen d’échantillon est la moyenne des
Le choix entre des plans d’échantillonnage simple,
différents effectifs d’échantillon possibles qui peu-
double, multiple et progressif doit être fait avant le
vent être obtenus pour un plan d’échantillonnage
début du contrôle. II n’est pas possible pendant le
pour une qualité d’un lot ou processus donné.
contrôle d’un lot de changer un type de plan
Comme pour des plans d’échantillonnage double ou
d’échantillonnage contre un autre, car l’efficacité du
multiple, l’utilisation des plans d’échantillonnage
plan peut être radicalement modifiée si les résultats
progressif conduit a un effectif moyen d’échantillon
du controle réel influencent le choix du critère d’ac-
inférieur a celui qu’il faut pour un plan d’échan-
ceptation.
tillonnage simple ayant la même efficacité. Cepen-
dant, les économies moyennes sont également plus
2.1.3 Troncage de l’effectif d’échantillon
importantes lors de l’utilisation d’un plan progressif
plutôt que d’un plan double ou multiple.
Bien qu’un plan d’échantillonnage progressif soit en
moyenne plus économique que le plan d’échan-
Pour des lots de bonne qualité, les économies peu-
tillonnage simple équivalent, il peut arriver pendant
vent atteindre ou même excéder 50 %, comparées
le contrôle d’un lot particulier que l’acceptation ou
à une économie maximum de 37 % avec un échan-
à un stade très tardif, le
la non-acceptation arrive
tillonnage double. L’annexe C donne une méthode
résultat du cumul restant compris très longtemps
pour déterminer les valeurs approximatives de I’ef-
entre les criteres d’acceptation et de rejet. Pour la
fectif moyen d’echantillon.
méthode graphique cela se traduit par la progres-
sion aléatoire de la courbe dans la zone d’indéci-
D’autre part, le nombre réel d’unités contrôlées
sion du lot. Une telle situation est d’autant plus
pour un lot particulier en plan double, multiple ou
probable que le niveau de qualité du lot (en propor-
progressif peut dépasser celui correspondant à un
tion de non conformes ou en non-conformités par
plan d’échantillonnage simple. Pour des plans
unité) est proche de g, où g représente la pente des
d’échantillonnage double et multiple, il existe une
droites d’acceptation et de rejet.
limite supérieure au nombre réel d’unités à contro-
Ier .
Afin de pallier cet inconvénient, un effectif cumulé
d’échantillon maximal 4 est établi avant le début de
Pour les plans d’échantillonnage progressif, il
l’échantillonnage et le contrôle est arrêté si l’effectif
n’existe généralement pas une te.lle limite et le
cumulé d’échantillon atteint. la valeur de troncage,
nombre d’unités contrôlées peut considérablement
I+, sans qu’une décision ait été prise. L’acceptation
excéder l’effectif d’échantillon du plan d’échan-
ou la non-acceptation du lot est alors déterminée
tillonnage simple correspondant et même dépasser
selon une règle décidée également avant I’échan-
l’effectif du lot. Pour les plans d’échantillonnage
tillonnage. Les règles du troncage dans la présente
progressif de la présente Norme internationale, une
Norme internationale ont été déte,rminé.es de telle
règle de troncage (voir 2.1.3) a été introduite afin de
facon que les risques client et fournisseur soient à
limiter le nombre potentiel d’unités contrôlées.
peine affectés pa.r cet écart aux principes sous-
Puisque l’effectif final d’échantillon n’est pas connu
jacents de la théorie statistique sur les plans
à l’avance, la sélection de l’échantillon peut donc
d’échantillonnage progressif. Les règles de
présenter des difficultés organisationnelles. De plus,
troncage à utiliser sont données en 2.4.2.
l’utilisation d’un plan d’échantillonnage double,
multiple ou progressif peut entraîner des difficultés
2.2 Réserves particulières sur le contrôle
de planification des opérations de contrôle. Un autre
de petits lots
inconvénient est que la mise en œuvre d’un plan
d’échantillonnage progressif est plus facilement mal
La théorie statistique sous-jacente aux plans
interprétée par les contrôleurs que les règles plus
d’échantillonnage progressif de la présente Norme
faciles de l’échantillonnage simple.
internationale est fondée sur l’hypothèse que les
Le résultat de la balance entre l’intérêt d’un effectif
khantillons sont tirés du lot <, c’est-
moyen d’echantillon plus petit et les difficultés a-dire que chaque unité échantillonnée est remise

---------------------- Page: 9 ----------------------
ISO 8422:1991(F)
dans le lot avant la sélection de l’unité suivante. les circonstances. Dans cette perspective, le risque
Lorsque, comme souvent, l’échantillonnage est sans fournisseur, a = 0,05, et le risque client, /II = O,lO,
remise, la théorie reste valable dans tous les cas sont souvent utilisés (voir figure 1).
pratiques, si l’effectif cumulé d’échantillon n’excède
Lorsque le plan d’échantillonnage doit avoir ap-
pas un dixième de N, où N est la taille du lot; la
proximativement la même efficacité qu’un plan
théorie reste approximativement valable même pour
d’échantillonnage simple, double ou multiple exis-
des effectifs cumulés d’échantillon atteignant un
tant, le point du risque fournisseur et le point du
septième de N. Malheureusement, au contraire de
risque client sont choisis à l’aide d’un graphique ou
la situation d’un plan d’échantillonnage simple, I’ef-
d’une table de l’efficacité de ce plan. Quand un tel
fectif cumulé d’échantillon réel qui est nécessaire
plan n’existe pas, les points des risques client et
dans un plan d’échantillonnage progressif ne sera
fournisseur doivent être déterminés à partir des
pas connu à l’avance.
considérations directes des conditions dans les-
Par conséquent, dans le cas d’un petit lot, il est re-
quelles l’échantillonnage sera effectue.
commandé de s’assurer que l’effectif du lot est suf-
fisamment grand pour permettre une mise en œuvre
2.4 Opérations préliiminaires
d’un plan d’échantillonnage progressif tronqué avec
échantillonnage sans remise selon les risques four-
nisseur et client spécifiés. Pour les plans d’échan-
2.4.1 Obtention des paramètres hA, h, et g
tillonnage progressifs généraux décrits en 2.3.2 et
2.4.1, il est alors recommandé que l’effectif du lot
Les critères d’acceptation ou de non-acceptation
dépasse 7q, où q est la valeur du troncage du plan
d’un lot utilisés a chaque stade du controle sont
d’échantillonnage progressif.
déterminés a partir des paramètres hA, k et g.
NOTE 4 Si l’effectif du lot n’est pas suffisamment grand Les valeurs de ces paramétres correspondant à un
pour satisfaire à la condition ci-dessus, les risques client
risque fournisseur, 6c = 0,05, un risque client,
et fournisseur deviendront généralement inférieurs à
/? = 0,lO et des valeurs recommandées des niveaux
leurs valeurs spécifiées. Cependant, si le critère d’accep-
de qualité du risque client et du risque fournisseur
tation du plan d’échantillonnage simple correspondant est
sont données dans les tables 1-A et I-B.
zéro, alors le risque fournisseur peut excéder faiblement
les valeurs spécifiées.
L’annexe B donne des méthodes générales pour la
détermination de hA, h, et g pour toutes combinai-
2.3 Sélection d’un plan d’échantillonnage sons des points des risques fournisseur et client.
EXEMPLE
2.3.1 Plans équivalents aux plans de
I’ISO 2859-l
La spécifrcation d’un type d’isolateur électrique est
donnée par la valeur de résistance nominale de
S’il est nécessaire d’obtenir un plan d’échantillon-
1 000 kV. Une agence de contrôle utilise un plan
nage progressif équivalent à un plan de I’ISO 2859-1,
d’échantillonnage simple avec un effectif d’échan-
l’annexe A peut être utilisée. L’annexe A contient
tillon de 65 et un critère d’acceptation égal à 6 pour
des plans d’échantillonnage progressif indexés par
déterminer I’acceptabilité de la production des lots
le niveau de qualité acceptable (NQA) et la Iettre-
de ce type d’isolateur. Puisque la construction des
code de l’effectif d’échantillon. Les courbes d’effi-
essais de voltage consomme a la fois temps et
cacité de ces plans d’échantillonnage progressif
énergie, il a été décidé d’utiliser dans le futur des
sont aussi proches que possible de celles des plans
plans d’échantillonnage progressif pour déterminer
correspondants de I’ISO 2859-l.
I’acceptabilité de la production des lots d’isolateurs.
Le plan d’échantillonnage progressif doit avoir une
efficacité similaire au plan d’échantillonnage simple
2.3.2 Plans généraux
utilisé.
La méthode générale décrite en 2.3.2 et 2.4.1 est
Le plan d ‘echanti Ilonnage simple pos sède les pro-
utilisée lorsque les données d’un plan progressif
.
étés s U ivantes:
Pr1
sont spécifiées par deux points de la courbe d’effi-
cacité du plan. Le point correspondant à la plus
- si 5 % des isolateurs produits n’atteignent pas
haute probabilité d’acceptation doit être appelé le
la tension nominale, alors la probabilité d’ac-
point du risque fournisseur, l’autre doit être appelé
cepter le lot est de 0,95;
le point du risque client.
La première étape dans la détermination d’un plan - si 16 % des isolateurs produits n’atteignent pas
d’échantillonnage progressif est le choix de ces la tension nominale, alors la probabilité d’ac-
deux points si ceux-ci n’ont pas déjà été dictés par cepter le lot est de 0,lO.

---------------------- Page: 10 ----------------------
ISO 8422:1991(F)
I
#
80 -
60 -
PI
I I
1 0 5 10 15
WOP
Nfveau de qualttt! du processus
I
QRF QRC (en pourcentage de non conformes)
Figure 1 - Courbe d’efficacité pour un plan d’échantillonnage avec un risque fournisseur a = 0,05 et un risque
client p = 0,lO
Ces conditions correspondent à fixer
calculée par r+ = 1,5 4, arrondi au plus proche
entier supérieur.
a) la qualité du risque fournisseur (QRF) a 5 %
avec 95 % attendu de lots acceptés; c’est-à-dire
b) Si l’effectif d’echantillon du plan d’échantillon-
telle que le risque fournisseur est 5 %, ou
nage simple correspondant n’est pas connu, la
a = 0,05; valeur de troncage pour le contrôle de la pro-
portion de non conformes est déterminée par:
b) la qualité du risque client (QRF) à 16 % de non
2hAhR
conformes avec 10 % attendu de lots acceptes;
c’est-à-dire telle que le risque client est 10 %, q= gu -g>
ou a = 0,lO.
arrondi au plus proche entier supérieur, et dans
Ces conditions sont indiquées sur le graphique de le cas d’un contrôle de non-conformités par
la courbe d’efficacité en figure 1. unité, q est déterminé par
2hAhR
Les paramètres du plan d’échantillonnage progres-
=-
nt
sif, trouvés dans la table I-A, qui satisfont à ces g
conditions, sont:
arrondi au plus proche entier supérieur.
h A = 1,750
2.4.2.2 Troncage pour de petits lots
& = 2,247
Si la valeur calculée de I+ est supérieure à l’effectif
g = 0,095 7
du lot, alors la valeur de troncage I+ de l’effectif
d’echantillon égale a l’effectif du lot doit être utilisée
Les mêmes valeurs peuvent être calculées à l’aide
de la méthode donnée dans l’annexe B. pour le plan d’échantillonnage progressif.
2.4.2.3 Exemple
2.4.2 Détermination de la valeur de troncage
de l’effectif d’échantillon Considerons le plan d’échantillonnage progressif
pour le contrôle de pourcentage de non conformes
avec pour paramétres hA= 1,750, h, = 2,247 et
2.4.2.1 Procédures normales
g= 0,095 7, détermines dans l’exemple donné en
2.4.1. Le plan a été choisi pour correspondre au plan
a) Si l’effectif d’échantillon h du plan d’échan-
d’échantillonnage simple avec 4 = 65 et A, = 6.
tillonnage simple correspondant au plan
d’échantillonnage progressif est connu, la valeur La valeur de troncage pour l’effectif cumulé
de troncage de l’effectif cumulé d’échantillon est d’échantillon est alors 4 = 98.

---------------------- Page: 11 ----------------------
ISO 8422:1991(F)
Le critère de rejet correspondant est calculé par
Si le plan d’échantillonnage simple n’avait pas été
connu, la valeur de troncage aurait été déterminée
x, = A, + 1
selon 2.4.2.1 b). La substitution des valeurs de hA,
h, et g dans la formule de 2.4.2.1 b) donne l’effectif
Partout où la valeur de l’expression (2.1) est néga-
d’échantillon tronqué I+ = 91.
tive, l’effectif cumulé d’échantillon est trop petit pour
permettre l’acceptation du lot. Inversement, partout
où la valeur de l’expression (2.2) est supérieure à
2.4.3 Choix de la forme du plan
l’effectif cumule d’echantillon, celui-ci est trop petit
d’échantillonnage
pour permettre une non-acceptation du lot contrôlé
pour la proportion de non conformes.
La présente Norme internationale donne deux mé-
thodes de mise en œuvre d’un plan d’échantillon-
Les grandeurs données par les expressions (2.1) et
nage progressif: une méthode numérique et une
(2.2) doivent être déterminées avec trois décimales
méthode graphique.
avant d’être arrondies.
La méthode numérique a l’avantage d’être précise
L’effectif minimal cumulé d’échantillon permettant
et d’éviter ainsi des discussions sur l’acceptation
l’acceptation du lot est obtenu en arrondissant hA/g
la non-acceptation.
ou
au plus proche entier supérieur.
La méthode graphique est bien adaptée au contrôle
L’effectif minimal cumulé d’echantillon permettant
séries de lots, car la carte n’a besoin d’être
de
la non-acceptation du lot contrôlé pour la proportion
dessinée qu’une fois, mais elle est moins précise à
de non conformes est obtenu en arrondissant
cause de l’imprécision inhérente au marquage des
I&/(l -g) au plus proche entier supérieur.
points et dessin de droites. D’autre part, la méthode
a l’avantage de visualiser l’évolution de I’infor- EXEMPLE
mation sur la qualité du lot lorsque des unités sup-
Pour le plan d’échantillonnage progressif ayant pour
plémentaires sont contrôlées, l’information étant
paramètres hA = 1,750, & = 2,247 et g = 0,095 7 dé-
représentée par la progression d’une ligne brisée à
termines dans l’exemple donné en 2.4.1, la valeur
l’intérieur de la zone d’indécision, jusqu’à ce que
de troncage de l’effectif d’échantillon a éte détermi-
cette ligne atteigne ou dépasse une des limites de
née dans l’exemple donne en 2.4.2.3 comme étant
cette zone.
I+ = 98. Le critére d’acceptation correspondant est
La méthode numérique est la méthode de référence
calcule en arrondissant gn, = 9,38 au plus proche
dans la mesure où cela concerne l’acceptation ou entier inferieur; le critére d’acceptation A, est donc
la non-acceptation d’un lot. Voir l’avertissement en 9 et le critère de rejet Rt est 10.
3.4.2.
La formule pour le critère d’acceptation A devient
0,095 7 IZcgjm - 1,750
2.4.3.1 Méthode numérique
arrondi au plus proche entier inférieur, et la formule
Pour chaque valeur, hum, de l’effectif cumule
du critère de rejet R devient
d’échantillon, qui est inférieure à la valeur de
troncage de l’effectif d’échantillon, le critère d’ac-
0,095 7 %um + 2,247
ceptation A est trouvé en arrondissant la grandeur
arrondi au plus proche entier supérieur.
h . . . .
8n cum - A (2 1)
Les critères d’acceptation et de rejet correspondant
au plus proche entier inferie ur. Le critére de rejet
aux effectifs cumulés d’échantillon %um = 1, 2, . . . . 97
R est trouvé en arrond issant randeur
la cl
sont successivement déterminés en remplaçant
. . . . qum par sa valeur dans ces formules et en arron-
Pcum + k (2 2)
d
...

Questions, Comments and Discussion

Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.