ISO 7870-3-2012 (E) - Acceptance Control Charts

Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

INTERNATIONAL ISO

STANDARD 7870-3

First edition
2012-03-01

Control charts —
Part 3:
Acceptance control charts
Cartes de contrôle —
Partie 3: Cartes de contrôle pour acceptation

Reference number
ISO 7870-3:2012(E)

© ISO 2012
ISO 7870-3:2012(E)

COPYRIGHT PROTECTED DOCUMENT


©  ISO 2012
All rights reserved. Unless otherwise specified, 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 either ISO at the address below or ISO’s
member body in the country of the requester.
ISO copyright office
Case postale 56 • CH-1211 Geneva 20
Tel. + 41 22 749 01 11
Fax + 41 22 749 09 47
E-mail [email protected]
Web www.iso.org
Published in Switzerland

ii  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

Contents Page

Foreword............................................................................................................................................................................. iv
Introduction......................................................................................................................................................................... v
1 Scope....................................................................................................................................................................... 1
2 Normative references.......................................................................................................................................... 1
3 Terms and definitions. ........................................................................................................................................ 1
4 Symbols and abbreviated terms...................................................................................................................... 2
4.1 Symbols.................................................................................................................................................................. 2
4.2 Abbreviated terms. .............................................................................................................................................. 3
5 Description of acceptance control chart practice....................................................................................... 3
6 Acceptance control of a process..................................................................................................................... 5
6.1 Plotting the chart.................................................................................................................................................. 5
6.2 Interpreting the chart. ......................................................................................................................................... 5
7 Specifications. ...................................................................................................................................................... 5
8 Calculation procedures...................................................................................................................................... 6
8.1 Selection of pairs of elements.......................................................................................................................... 6
8.2 Frequency of sampling....................................................................................................................................... 8
9 Examples................................................................................................................................................................ 9
9.1 Example 1 ............................................................................................................................................................. 9
9.2 Example 2 ............................................................................................................................................................10
10 Factors for acceptance control limits........................................................................................................... 11
11 Modified acceptance control charts. ............................................................................................................ 12
Annex A (normative) Nomographs for acceptance control chart design.......................................................... 14
Bibliography...................................................................................................................................................................... 20

© ISO 2012 – All rights reserved  iii


ISO 7870-3:2012(E)

Foreword

ISO (the International Organization for Standardization) is a worldwide federation of national standards bodies
(ISO member bodies). The work of preparing International Standards is normally carried out through ISO
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, governmental and
non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely with the International
Electrotechnical Commission (IEC) on all matters of electrotechnical standardization.

International Standards are drafted in accordance with the rules given in the ISO/IEC Directives, Part 2.

The main task of technical committees is to prepare International Standards. Draft International Standards
adopted by the technical committees are circulated to the member bodies for voting. Publication as an
International Standard requires approval by at least 75 % of the member bodies casting a vote.

Attention is drawn to the possibility that some of the elements of this document may be the subject of patent
rights. ISO shall not be held responsible for identifying any or all such patent rights.

ISO 7870‑3 was prepared by Technical Committee ISO/TC 69, Applications of statistical methods, Subcommittee
SC 4, Applications of statistical methods in process management.

This first edition of ISO 7870-3 cancels and replaces ISO 7966:1993.

ISO 7870 consists of the following parts, under the general title Control charts:

— Part 1: General guidelines

— Part 2: Shewhart control charts

— Part 3: Acceptance control charts

— Part 4: Cumulative sum charts

Additional parts on specialized control charts and on the application of statistical process control (SPC)
charts are planned.

iv  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

Introduction

An acceptance control chart combines consideration of control implications with elements of acceptance
sampling. It is an appropriate tool for helping to make decisions with respect to process acceptance. The bases
for the decisions may be defined in terms of

a) whether or not a d esignated percentage of units of a p roduct or service derived from that process will
satisfy specification requirements;

b) whether or not a process has shifted beyond some allowable zone of process level locations.

A difference from most acceptance sampling approaches is the emphasis on process acceptability rather than
on product disposition decisions.

A difference from usual control chart approaches is that the concept of process acceptance is introduced in
the process control. The process usually does not need to be in control about a single standard process level;
as long as the within-subgroup variability remains in control and is much smaller than the tolerance spread, it
can (for the purpose of acceptance) run at any level or levels within a zone of process levels which would be
acceptable in terms of tolerance requirements. Thus, it is assumed that some assignable causes will create
shifts in the process levels which are small enough in relation to requirements that it would be uneconomical
to attempt to control them too tightly for the purpose of mere acceptance.

The use of an acceptance control chart does not, however, rule out the possibility of identifying and removing
assignable causes for the purpose of continuing process improvement.

A check on the inherent stability of the process is required. Therefore, variables are monitored using Shewhart-
type range or sample standard deviation control charts to confirm that the variability inherent within rational
subgroups remains in a steady state. Supplementary examinations of the distribution of the encountered
process levels form an additional source of control information. A preliminary Shewhart control chart study
should be conducted to verify the validity of using an acceptance control chart.

© ISO 2012 – All rights reserved  v


INTERNATIONAL STANDARD ISO 7870-3:2012(E)

Control charts —

Part 3:
Acceptance control charts

1 Scope
This part of ISO 7870 gives guidance on the uses of acceptance control charts and establishes general
procedures for determining sample sizes, action limits and decision criteria. An acceptance control chart
should be used only when:

a) the within subgroup variation is in-control and the variation is estimated efficiently;

b) a high level of process capability has been achieved.

An acceptance control chart is typically used when the process variable under study is normally distributed;
however, it can be applied to a non-normal distribution. The examples provided in this part of ISO 7870 illustrate
a variety of circumstances in which this technique has advantages; these examples provide details of the
determination of the sample size, the action limits and the decision criteria.

2 Normative references
The following standards, in whole or in part, are normatively referenced in this document and are indispensable
for its application. For dated references, only the edition cited applies. For undated references, the latest edition
of the refferenced document (including any amendments) applies.

ISO 3534-1, Statistics — Vocabulary and symbols — Part 1: General statistical terms and terms used in probability

ISO 3534-2, Statistics — Vocabulary and symbols — Part 2: Applied statistics

3 Terms and definitions


For the purposes of document, the terms and definitions given in ISO 3534-1 and ISO 3534-2 apply.

3.1
acceptable process
process which is represented by a Shewhart control chart with a central line within the acceptable process zone

NOTE 1 Ideally, the average value X of such a control chart would be at the target value.

NOTE 2 The acceptable process zone is shown in Figure 1. Information on the Stewhart control chart can be found
in ISO 7870-2.

© ISO 2012 – All rights reserved  1


ISO 7870-3:2012(E)

Figure 1 — Two-sided specification limits: Upper and lower APL and RPL lines in relation to
processes of acceptable, rejectable, and indifference (borderline) quality

4 Symbols and abbreviated terms


NOTE The ISO/IEC Directives makes it necessary to depart from common SPC usage in respect to the differentiation
between abbreviated terms and symbols. An abbreviated term and its symbol can differ in appearance in two ways: by
font and by layout. To distinguish between abbreviated terms and symbols, abbreviated terms are given in Arial upright
and symbols in Times New Roman or Greek italics, as applicable. Whereas abbreviated terms can contain multiple letters,
symbols consist only of a single letter. For example, the conventional abbreviation of acceptable process limit, APL, is
valid but its symbol in equations becomes APL. The reason for this is to avoid misinterpretation of compound letters as an
indication of multiplication.

4.1 Symbols
ACL acceptance control limits

APL acceptable process level

L lower specification limit

n subgroup sample size

p0 acceptable proportion nonconforming items

p1 rejectable proportion nonconforming items

Pa probability of acceptance

RPL rejectable process level or non-acceptable process zone

T target value, i.e. the optimum value of the characteristic

U upper specification limit

X average value of the variable X plotted on a control chart

z variable that has a normal distribution with zero mean and unit standard deviation

zp′ normal deviate that is exceeded by 100p′ % of the deviate in a specified direction (similarly for zα , zβ,
etc.)

a risk of not accepting a process centred at the APL

β risk of not rejecting a process centred at the RPL

2  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

µ process mean

σw within-subgroup standard deviation corresponding to the inherent process variability

σX standard deviation of the subgroup average corresponding to the inherent process variability:
σX =σw / n

4.2 Abbreviated terms


ACL acceptance control limits

APL acceptable process level

L lower specification limit (used as a subscript)

OC operating characteristic

RPL rejectable process level or non-acceptable process zone

U upper specification limit (used as a subscript)

5 Description of acceptance control chart practice


In the pursuit of an acceptable product or service, there often is room for some latitude in the ability to centre
a process around its target level. The contribution to overall variation of such location factors is additional to
the inherent random variability of individual elements around a given process level. In most cases, some shifts
in process level must be expected and can be tolerated. These shifts usually result from an assignable cause
that cannot be eliminated because of engineering or economic considerations. They often enter the system at
infrequent or irregular intervals, but can rarely be treated as random components of variance.

There are several seemingly different approaches to treating these location factors contributing variation beyond
that of inherent variability. At one extreme is the approach in which all variability that results in deviations from
the target value must be minimized. Supporters of such an approach seek to improve the capability to maintain a
process within tighter tolerance limits so that there is greater potential for process or product quality improvement.

At the other extreme is the approach that if a high level of process capability has been achieved, it is not only
uneconomic and wasteful of resources, but it can also be counterproductive to try to improve the capability
of the process. This often is the result of the introduction of pressures which encourage “tampering” with
the process (over-control) by people qualified to work on control aspects but not product or process quality
improvement programmes.

The acceptance control chart is a useful tool for covering this wide range of approaches in a logical and
simple manner. It distinguishes between the inherent variability components randomly occurring throughout
the process and the additional location factors which contribute at less frequent intervals.

When shifts appear, the process may then stabilize at a new level until the next such event occurs. Between
such disturbances, the process runs in control with respect to inherent variability.

An illustration of this situation is a process using large uniform batches of raw material. The within-batch
variability could be considered to be the inherent variability. When a new batch of material is introduced, its
deviation from the target may differ from that of the previous batch. The between-batch variation component
enters the system at discrete intervals.

An example of this within- and between-batch variation might very well occur in a situation where a blanking
die is blanking a machine part. The purpose of the chart is to determine when the die has worn to a point where
it must be repaired or reworked. The rate of wear is dependent upon the hardness of the successive batches
of material and is therefore not readily predictable. It will be seen that the use of an acceptance control chart
makes it possible to judge the appropriate time to service the blanking die.

© ISO 2012 – All rights reserved  3


ISO 7870-3:2012(E)

The acceptance control chart is based on the Shewhart control chart (i.e. X – R chart or X – s chart) but is
set up so that the process mean can shift outside of control limits of the Shewhart control chart if the
specifications are sufficiently wide, or be confined to narrower limits if the inherent variability of the process is
comparatively large or a large fraction of the total tolerance spread.

What is required is protection against a process that has shifted so far from the target value that it will yield
some predetermined undesirable percentage of items falling outside the specification limits, or exhibits an
excessive degree of process level shift.

When a chart of the average value of data sets from a process is plotted, in sequence of the production, one
notices a continual variation in average values. In a central zone (acceptable process, Figure 1), there is a
product that is indisputably acceptable. Data in the outer zones (Figure 1) represent a process that is producing
product that is indisputably not acceptable.

Between the inner and the outer zones are zones where the product is acceptable but there is an indication
that the process should be watched and, as the outer zone is approached, corrective action may be taken.
These criteria are the basic concepts for the acceptance control chart. The description in this part of ISO 7870
is designed to provide practices for the establishment of appropriate action lines for one- and two-sided
specification situations.

Since it is impossible to have a single dividing line that can sharply distinguish a good from an unsatisfactory
quality level, one must define a process level that represents a process that should be accepted almost always
(1 −  a). This is called the acceptable process level (APL), and it marks the outer boundary of the acceptable
process zone located about the target value (see Figure 1).

Any process centred closer to the target value than the APL will have a risk smaller than α of not being
accepted. So the closer the process is to the target, the smaller the likelihood that a satisfactory process will
not be accepted.

It is also necessary to define the process level that represents processes that should almost never be accepted
(1  −  β). This undesirable process level is labelled the rejectable process level (RPL). Any process located
further away from the target value than the RPL will have a risk of acceptance smaller than β.

The process levels lying between the APL and RPL would yield a product of borderline quality. That is, process
levels falling between the APL and RPL would represent quality which is not so good that it would be a waste
of time, or represent over-control, if the process were adjusted, and not so bad that the product could not be
used if no shift in level were made. This region is often called the “indifference zone”. The width of this zone is
a function of the requirements for a particular process and the risks one is willing to take in connection with it.
The narrower the zone, i.e. the closer the APL and RPL are to each other, the larger the sample size will have
to be. This approach will permit a realistic appraisal of the effectiveness of any acceptance control system, and
will provide a descriptive method for showing just what any given control system is intended to do.

As with any acceptance sampling system, the four elements required for the definition of an acceptance
control chart are:

a) an acceptable process level (APL) associated with a one-sided a-risk;

b) a rejectable process level (RPL) associated with a one-sided β -risk;

c) an action criterion or acceptance control limit (ACL);

d) the sample size (n).

NOTE Generally, the defined risks are one-sided in this part of ISO 7870. In the case of two-sided specifications, the risks
are either a 5 % risk to go above an upper limit or a 5 % risk to go below a lower limit. This results in a 5 % (not 10 %) total risk.

Simplicity of operation is of critical importance to the use of a procedure such as an acceptance control chart.
Only the acceptance control limits and the sampling instructions (such as sample size, frequency, or method
of selection) need to be known to the operator who uses the chart, although training him to understand the
derivation is not difficult and can be helpful. It is thus no more complicated to use than the Shewhart chart.
The supervisor, quality expert, or trained operator will derive these limits without much effort from the above

4  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

considerations and will obtain a more meaningful insight into the process acceptance procedure, and a better
understanding of the control implications.

6 Acceptance control of a process

6.1 Plotting the chart


The sample average value of the quality characteristic is plotted on acceptance control charts in the following
way. A point is plotted on the chart for each sample with an identification number (numerical order, time order,
etc.) on the horizontal scale, and the corresponding sample average on the vertical scale.

6.2 Interpreting the chart


When the plotted point falls above the upper acceptance control limit ACLU or below the lower acceptance
control limit ACLL, the process shall be considered non-acceptable.

If a plotted point is close to the control line, the numerical values shall be used to make the decision.

7 Specifications
Theoretically, the specification of the values of any two of the defining elements APL (with a−risk), RPL (with
β-risk), acceptance control limit (ACL) or sample size (n) of an acceptance control chart system determines the
remaining two values; however, in practice, it is essential that APL (with a-risk) be defined first. In addition, the
within-rational subgroup value of σw must be known or have been estimated by the usual control chart
techniques such as using σˆ w = R / d 2 or s / c 4 . It is essential that the inherent random variability be in a state
of statistical control in order for the risk computations to be meaningful. This can be monitored through the use
of a Shewhart-type control chart for ranges or standard deviations. (See ISO 7870-2.)

Two selections of pairs of defining elements may be chosen.

a) Definition of the APL and RPL along with their respective a- and β -risks, and determination of the sample
size (n) and the acceptance control limit (ACL).

Often, a = 0,05 is chosen in acceptance control chart applications since there are few instances where
a process continuously runs at the APL. This means that the risk of rejection on each side of the target
value, T, should always be smaller than a.

This option is generally used when

1) acceptable processes are defined either for economic or other practical reasons in terms of process
capabilities that include allowance for small discrete shifts in process level in addition to inherent
random variation, or in terms of an acceptable quality level described by the percentage of items
exceeding specification limits, and

2) when rejectable processes are defined either for practical reasons in terms of unnecessarily large
shifts in process level, or in terms of a process level yielding an unsatisfactory percentage of items
exceeding specification limits.

b) Definition of the APL (with a) and the sample size n, and determination of the RPL for a given β -risk and
the ACL.

This option is used when acceptable processes are defined as in 1) above, and when there is a restriction
determining the allowable sample size. The option a) is preferable in most cases.

The examples in this part of ISO 7870 deal with variables data and are described in terms of two-sided
specifications with limits and levels defined both above and below the target value. However, the method is
equally valid for one-sided specification limits. In addition, there is no requirement that the values selected
above and below the target value be symmetrical should more latitude be desired on either side. If different

© ISO 2012 – All rights reserved  5


ISO 7870-3:2012(E)

values are selected above and below the target, the sample size required for the more stringent situation (i.e.
smaller distance between the APL and RPL) shall be used (see 8.1.1).

8 Calculation procedures

8.1 Selection of pairs of elements

8.1.1 Defining elements APL and RPL

In the case of variables ( X ), the APL may be selected in several ways. If the specification limits are known, as
well as the underlying distribution of the individual population items, the APL may be defined in terms of an
acceptable proportion (or percentage) p0 of nonconforming items which would occur when the process is
centred at the APL. See Figure  2. If the underlying distribution is normal (Gaussian), a one-tailed table of
standard normal deviate z values can be used where,

∞ 1  x2 
p= ∫z p 2π
exp  −
 2 

 dx

Figure 2 — Limits and defining elements of acceptance control charts

For samples of four or more, the assumption of a normal distribution for control purposes is generally valid for
X charting. However, the interpretation of the proportion (percentage) of nonconforming items associated with
the APL and RPL levels is dependent on the underlying distribution. Thus, for other distributions, appropriate
tables should be followed and the standard normal deviate values z p t replaced accordingly. The advantage of
the z approach in this application is that the limits and defining elements fall above and below the centre, so

6  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

that it is convenient to have identical a and β values on both sides of the target rather than having to deal with
a and 1 −  a or β and 1 −  β, depending on which side of the centre is involved. This also aids in a geometric
interpretation such as

zα σ X + z β σ X = RPL − APL
Upper APL ( APL U ) = U − z p 0 σ w
Lower APL ( APL L ) = L + z p 0 σ w
See example  1 in 9.1 where X charts with the APL and RPL are defined in terms of the percentage of
nonconforming items. A flowchart for the calculation procedure is shown in Figure 3.

U and L given
σ w given by R chart

Determination of p0 and p 1

Calculation of APL and RPL

Determination of α and β

Calculation of ACL and n

Figure 3 — Flowchart for calculation procedure (defining elements APL and RPL)

In some cases, the selection of an APL value may not be directly related to the specification limits, but may be
chosen on an arbitrary basis. Experience may show that the “uneconomic” or “not readily adjustable” causes
for shifts in process level correspond to a narrow band. The edge of this band may be arbitrarily designated as
the APL (see example 2 in 9.2). In this case, the normal distribution assumption is not invoked since the APL
is not directly related to the specification limits.

In a similar fashion, the RPL may be selected in several ways. It can be related to the specification limits by
defining an unacceptable proportion (percentage) p1 of nonconforming items which would occur when the
process is centred at the RPL.

Upper RPL ( RPL U ) = U − z p1σ w

Lower RPL ( RPL L ) = L + z p1σ w

Once the APL and a, and RPL and β, values are defined, the upper acceptance control limit (ACL U) is located at

 z 
ACL U = APL U +  α
 zα + z β (
 RPL U − APL U
 )
 
where zα and z β are the cut-off points for a proportion a and β respectively.

© ISO 2012 – All rights reserved  7


ISO 7870-3:2012(E)

The lower limit is located at

 z 
ACL L = APL L +  α
 zα + z β (
 APL L − RPL L
 )
 
When the a- and β -risks are selected to be equal, the acceptance control limit lies halfway between the APL
and RPL.

The sample size can be calculated as

2
 ( zα + z β )σ w 
n= 
 RPL − APL 
For asymmetrical limits, as at the end of Clause 7:

 2
 zα ,L + z β ,L σ w  
2

n = max 
( 
  zα ,U + z β ,U σ w  )or 
( )
 
 R
  PL U − A PL U   R − A PL L  
   PL L  
A nomograph, which also provides an OC (operating characteristic) curve, can be used instead of these
calculations. Both the calculation and nomograph methods are easy to use (see Annex A).

8.1.2 Defining elements APL, a, β and n

The APL may be selected as specified in 8.1.1. The sample size may be specified as a matter of convenience
in the operation, or it may be entered as a trial proposal to discover what kind of RPL and β values will result.
If these are unsatisfactory, the process can be iterated or one of the other combinations used so that n is
calculated. Given the APL, a and n values:

ACL U = APL U+ z α σ w n

ACL L = APL L − zα σ w n

RPL U = ACL U + z βσ w n

RPL L = ACL L − z β σ w n

See example 2 in 9.2. A flowchart for calculation procedure is shown in Figure 4.

8.2 Frequency of sampling


The relationship between sample size and the a- and β -risks has been discussed above. The determination
of frequency of sampling will not be treated in this part of ISO 7870. If the history of a process is one of well-
behaved inherent variability and of level shifts usually within the zone of acceptable processes, the sampling
frequency may be relatively low when compared to that for processes exhibiting less stability. The costs of
erroneous decisions are to some extent considered in the selection of the a and β values, but are clearly
related to the frequency of sampling as well.

8  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

U and L given
σ w given by R chart

Determination of p0

Calculation of APL

n given

Determination of α and β

Calculation of ACL
Confirmation of RPL

Figure 4 — Flowchart for calculation procedure (defining elements APL, a , β and n)

9 Examples

9.1 Example 1 (see also Figures A.3 and A.4)


Operation: Filling bottles with 10,0 cm3 ± 0,5 cm3 of solution.

Measurement: Amount of solution; nominal value 10 cm3.

Variability: The inherent variability due to random causes is known to have a normal distribution. Past experience
shows σw = 0,1 cm3.

Objective: It is desired to accept the set-up by an operator if less than 0,1 % of the bottles filled are above
and/or below the range 10,0 cm3 ± 0,5 cm3. It is desired to reject the set-up by an operator if more than 2,5 %
of the bottles are above and/or below 10,0 cm3 ± 0,5 cm3.

The following data are used to calculate the APL and RPL:

Upper specification limit: U = 10,5 cm3

Lower specification limit: L = 9,5 cm3

Process standard deviation: σw = 0,1 cm3

The critical value of z of the normal distribution (cutting off a tail area equal to the specified fraction exceeding
specification limits):

z p 0 = 3, 090 for p 0 = 0, 001

© ISO 2012 – All rights reserved  9


ISO 7870-3:2012(E)

z p1 = 1, 960 for p1 = 0, 025

The evaluation yields:

APL { U − z 0,001σ w =10,5 − 3,090 × 0,1 = 10,191


L + z 0,001σ w = 9,5 + 3,090 × 0,1 = 9,809

RPL { U − z 0,025σ w =10,5 − 1,960 × 0,1 = 10,304


L + z 0,025σ w = 9,5 + 1960
, × 0,1 = 9,696

It has been decided to take an a-risk of 5 % and a β -risk of 5 % so that zα  =  z β  = 1,645. Therefore:

 zα 
ACL U = APL U+  (
 RPL U − APL
 U )
 α + zβ
z 
= 10,191 + 0, 5 (10, 304 − 10,191)
= 10, 245
and

 z 
ACL L = APL L −  α  A
 zα + z β  PL L (− RPL L )
 
= 9, 809 − 0, 5 ( 9, 809 − 9, 696 )
= 9, 755
The sample size is:

n=
α(
 z +z σ
β ) 

 RPL − APL 
 
2
 (1, 645 + 1, 645 ) × 0,1
= 
 0,113 
2
= ( 2, 912 ) = 8, 48
The sample size is rounded up to n = 9 to ensure that the risks do not exceed the specified values of a and β.

The interpretation of the results leads to the following conclusions.

a) Operator set-ups that deviate from the nominal level by ±0,191 cm3 or less (which would mean that fewer than
0,1 % of the individual bottles will exceed specifications) are reasonably sure (95 % or higher) of being accepted.

b) Operator set-ups that deviate from the nominal level by ±0,304 cm3 or more (which would mean that more
than 2,5 % of the individual bottles would exceed specifications) are reasonably sure (95 % or higher) of
being rejected.

c) Operator set-ups that deviate from the nominal level by more than ±0,191 cm3 but less than ±0,304 cm3
may or may not be rejected for readjustment. These are considered not bad enough to be sure of rejecting
but not good enough to be sure of accepting. They represent borderline or “indifferent” quality with respect
to the accuracy of their set-up.

9.2 Example 2 (see also Figure A.5)


Operation: Coating process.

Measurement: Thickness of the coating.

Variability: The inherent variability of narrow lengthwise strips measured across the coating can be characterized
by the standard deviation within strips along the coating; σw = 0,005.

10  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

Objective: Since uniformity from strip to strip is more important than actual level, it is decided that strips having
mean values deviating from the grand average of all strips by less than ±0,008 mm should be accepted with a
risk of rejection of less than a = 5 %. For operating convenience, a sample size of n = 4 has been established.
Thus, the given parameters are σw = 0,005 and

APL L = −0,008 associated with a = 0,05 and zα = 1,645

APL U = +0,008 associated with a = 0,05 and zα = 1,645

The lower acceptance control limit is

ACL L = APL L − zα σ X
0, 005
= −0, 008 − 1, 645 ×
4
= −0, 012
The lower rejectable process level associated with a β -risk of 5 % is

RPL L = ACL L − z β σ X
0, 005
= −0, 012 − 1, 645 ×
4
= −0, 016
Similarly,

ACL U = APL U + zα σ X
= 0, 012
RPL U = ACL U + z β σ X
= 0, 016
Interpretation of the results leads to the following conclusions.

a) Strips across the coating which have an average thickness deviating from the average thickness of the
entire coating by ±0,008 mm or less will be reasonably sure (95 % or higher) of being accepted for uniformity.

b) Strips across the coating which have an average thickness deviating from the average thickness of the entire
coating by ±0,016 mm or more will be reasonably sure (95 % or higher) of being rejected for lack of uniformity.

c) Strips across the coating which have an average thickness deviating from the average thickness of the
entire coating by more than ±0,008 mm but less than ±0,016 mm may or may not be rejected for lack of
uniformity. These represent thickness deviations which are not so small that they should definitely be
accepted nor so large that they should definitely be rejected.

Note that if this “indifference zone” of 0,008 mm to 0,016 mm is considered too wide, it can be reduced by the
use of a larger sample size. If n = 16 instead of n = 4, the acceptance control limits become ±0,010 mm and the
RPL values become ±0,012 mm. Or, if instead of demanding a smaller indifference zone, it is decided to demand
that a better job be done in obtaining uniform coatings, the APL can be shifted closer to the nominal value. For
example, if it is decided that a deviation of ±0,004 mm is as far as the 95 % acceptance protection is to go, then
for a sample size of 4, the new acceptance control limits would be ±0,008 mm and the RPL values ±0,012 mm.

10 Factors for acceptance control limits


Acceptance control limit factors are based on one-tail normal distribution probabilities unless the APLs lie
within 0, 85σ w / n of the target level, for a = 0,05, or within 0, 67σ w / n for a = 0,01. These values are the
outer bound for situations representing “tight” specification requirements for which the a-risk must be divided
appropriately on either side of the target value. The columns in Table 1 give:

a) the multiple of σ w / n ; the APL distance from the target level;

© ISO 2012 – All rights reserved  11


ISO 7870-3:2012(E)

b) the multiple of σ w / n ; the ACL distance from the target level which is the sum of the APL distance and
the corresponding z component for varying degrees of two-sided a-risks;

c) the Pa value for the APL using nomographs similar to Figures A.1 to Figure A.5.

It should be noted that when the difference between APL and the target value is small in units of σw, that is, for
the “tight” specification situations, the acceptance control chart is not appropriate.

Table 1 — Acceptance control limit factors

a = 0,05 a = 0,01
Difference Difference Difference Difference
between APL z between ACL Pa between APL z between ACL Pa
and target and target and target and target
col. 3 col.7
col.1 col. 2 col. 4 col. 5 col. 6 col. 8
(col. 1 + col. 2) (col. 5 + col. 6)
≥0,85 1,65 ≥2,50 0,950 ≥0,67 2,33 ≥3,00 0,990
0,80 1,65 2,45 0,951 0,60 2,33 2,93 0,990
0,70 1,66 2,36 0,952 0,50 2,33 2,83 0,990
0,60 1,67 2,27 0,953 0,40 2,37 2,77 0,991
0,50 1,68 2,18 0,954 0,30 2,37 2,67 0,991
0,40 1,71 2,11 0,956 0,20 2,41 2,61 0,992
0,30 1,75 2,05 0,960 0,10 2,52 2,62 0,994
0,20 1,80 2,00 0,964 0,00 2,58 2,58 0,995
0,10 1,87 1,97 0,969
0,00 1,96 1,96 0,975

NOTE The control limit factors given in this table are for use in locating acceptance and control limit lines:

APL = target value ± (factora) (σw n)


ACL = target value ± (factorb) ( σ w n)

a Use the appropriate factor from column 1 or 5.


b Use the appropriate factor from column 3 or 7.

11 Modified acceptance control charts


A modified acceptance control chart is a special case of a process acceptance control chart in which its
acceptance control limits can be determined in terms of its specification limits as shown in the following equations:

APL U = U − z p0 σ w

APL L = L + z p0 σ w

ACL U = APL U + zα σ w / n

12  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

ACL L = APL L − zα σ w / n

The acceptance control limits determined above are located inside the specification limits. The determination
procedure is similar to Example 1 shown in 9.1; however, it does not define the β -risk for specified rejectable
process levels nor does it provide rules for determining the sample size.

© ISO 2012 – All rights reserved  13


ISO 7870-3:2012(E)

Annex A
(normative)

Nomographs for acceptance control chart design

A.1 General
Instead of a computation procedure, a nomograph procedure can be used for designing an acceptance control chart.
This approach includes the advantage of gaining easy access to any information on the accompanying OC curve.

A.2 Acceptance control charts for process mean, µ


The nomograph paper used with (approximately) normally distributed processes is given in Figure A.1. Assigning
a linear scale to the horizontal axis, any OC curve (probability of acceptance Pa as a function of the process
mean µ) may be represented as a straight line by appropriate choice of probability scale for the vertical axis.

The principle of the one-sided procedure is presented in Figure A.2. The OC curve is represented by a straight
line. For a =  β, the acceptance control limit is equal to the value of µ that yields a probability of acceptance
Pa  =  0,5  or 50  %. The slope of the OC curve depends upon the scale chosen for the horizontal axis and
upon the process standard deviation s and is related to the sample size n. The interrelation between these
parameters is represented by the dashed line parallel to the OC curve. This dashed line is needed for the
control chart design. Besides the process standard deviation, s, there are four parameters in the design:

a) the acceptable process level at probability of acceptance Pa = 1 − a, µ APL = APL ;

b) the rejectable process level at probability of acceptance Pa = β ; µ RPL = RPL ;

c) the acceptance control limit, µ ACL = ACL ;

d) the sample size, n.

If any two of these four parameters are given, the remaining two parameters can be deduced. The following
examples illustrate the procedures in detail:

EXAMPLE 1 (see Figure A.3 and Figure A.4)

Given: APL with Pa = 1 − a

RPL with Pa = β

EXAMPLE 2 (see Figure A.5)

Given: APL with Pa = 1 − a

n; s

14  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

Key
Pa probability of acceptance
µ process mean
n sample size
σ standard deviation (inherent variability)
Pa = Pa (µ)

Figure A.1 — Nomograph paper for acceptance control chart design

© ISO 2012 – All rights reserved  15


ISO 7870-3:2012(E)

Key
Pa probability of acceptance
µ process mean
n sample size
σ standard deviation (inherent variability)
Pa = Pa (µ)

Figure A.2 — Acceptance control chart design — One-sided approach

16  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

Key
Pa probability of acceptance
µ process mean
n sample size
σ standard deviation (inherent variability)
Pa = Pa (µ)

Figure A.3 — Acceptance control chart design — Example 1

© ISO 2012 – All rights reserved  17


ISO 7870-3:2012(E)

Key
Pa probability of acceptance
µ process mean
n sample size
σ standard deviation (inherent variability)
Pa = Pa (µ)

Figure A.4 — Acceptance control chart design — Example 1

18  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

Key
Pa probability of acceptance
µ process mean
n sample size
σ standard deviation (inherent variability)
Pa = Pa (µ)

Figure A.5 — Acceptance control chart design — Example 2

© ISO 2012 – All rights reserved  19


ISO 7870-3:2012(E)

Bibliography

[1] BELZ, M.H. Statistical Methods for the Process Industries, John Wiley & Sons, New York, 1973

[2] DUNCAN,  A.J. Quality Control and Industrial Statistics, 5th Edition, Richard D. Irwin, Inc.,
Homewood, IL, 1986

[3] FREUND, R.A. Acceptance Control Charts. Industrial Quality Control, 14(4), October 1957

[4] FREUND, R.A. A Reconsideration of the Variables Control Chart. Industrial Quality Control, 16(11), May 1960

[5] RICKMERS,  A.D. and TODD, H.N. Statistics, An Introduction, McGraw-Hill Book Co., New York, 1967

[6] SHEWHART, W.A. Economic Control of Quality of Manufactured Product (originally D. Van Nostrand Co.,
Inc., New York, 1931), republished by American Society for Quality Control, Inc., Milwaukee, WI, 1980

[7] ISO 7870-1, Control charts — Part 1: General guidelines

[8] ISO 7870-21), Control charts — Part 2: Shewhart control charts

[9] ISO 7870-4, Control charts — Part 4: Cumulative sum charts

1) Under preparation.

20  © ISO 2012 – All rights reserved


ISO 7870-3:2012(E)

ICS  03.120.30
Price based on 20 pages

© ISO 2012 – All rights reserved

You might also like