Sampling

Procedures
Inspection by Attributes

SAFM Consulting Services

Learning Objectives
Ø Explain the purpose of acceptance
sampling
Ø Contrast acceptance sampling and
process control
Ø Compare and contrast single and
multiple sampling plans
Ø Determine the average outgoing quality
of inspected lots

2
Sampling Plans
Ø Acceptance sampling: Form of
inspection applied to lots or
batches of items before or after
a process, to judge conformance
with predetermined standards
Ø Sampling plans: Plans that
specify lot size, sample size,
number of samples, and
acceptance/rejection criteria
Ø Single-sampling
Ø Double-sampling
Ø Multiple-sampling

3
Acceptance Sampling
Ø Acceptance Sampling most useful when
ØA large number of items must be
processed in a short time
ØThe cost consequences of passing
defects are low
ØDestructive testing is required
ØFatigue or boredom leads to inspection
errors

4
Introduction
Ø BS 6001-1 1999 (ISO 2859-1:1999)
Sampling
 procedures for inspection by attributes.
Part 1:
 Sampling schemes indexed by
acceptance quality
 limit (AQL) for lot-by-lot inspection.
Ø ISO 2859-2:1985 Sampling procedures for
 inspection by attributes. Sampling plans
indexed
 by limiting quality (LQ) for isolated lot
inspection.
5
Ø ISO 2859-0:1995 - Part 0: Introduction to
ISO 2859 Part 1

Ø Acceptance sampling system for inspection by

attributes.
Ø It is indexed in terms of acceptance quality
level (AQL)
Ø Aim - to induce a supplier to maintain a
process average at least as good as the
specified acceptance AQL, whilst at the
same time providing an upper limit for the
risk to the consumer of accepting the
occasional poor lot.
Ø Applicable to end products, raw materials,
operations, maintenance operations, 6
Usage of the Part 1 scheme.
Ø Intended to be used for a continuing series
of lots, which will allow the application of
switching
 rules. The rules provide:
 - protection to the consumer should a
deterioration in quality be detected.
 - an incentive to the supplier to
reduce
 inspection costs should consistently
good
 quality be achieved.
Ø Can also be used for inspection of lots in
7
 isolation
Terminology
Ø Inspection by attributes: inspection
where an
 item is classified as conforming or
nonconforming with respect to a specified
requirement or set of requirements.
Ø Nonconformity: Non fulfillment of a
specified
 requirement. Usually classified
according to the
 degree of seriousness. More serious
 nonconformities will usually be assigned
a very
 small AQL, whilst less serious8
Terminology (cont)
Ø Normal inspection: Use of a sampling plan
with
 acceptance criteria devised to secure the
producer a high probability of acceptance when
the process average of the lot is better than the
AQL. Used when there is no reason to suspect
the process average differs from an acceptable
level.
Ø Tightened inspection: Use of a sampling plan
 with an acceptance criteria that is tighter
than that for the corresponding plan for normal
inspection. Invoked when the inspection results
of consecutive lots indicate that the process 9
average might be poorer than the AQL.
Terminology (cont)
Ø Acceptable Quality Limit (AQL): Quality
level
 that is the worst tolerable process
average when a continuing series of lots is
submitted for acceptance sampling. AQL
does not mean ’desirable level’. ISO 2859
is designed to encourage suppliers to have
process averages consistently better than
the AQL, otherwise there is a risk of
switching to tighter inspection.
Ø The designation of an AQL does not imply
that the supplier has the right knowingly
to supply any
nonconforming items.
10

Terminology (cont)
Ø AQL values shall not exceed 10%
nonconforming.
Ø When the quality level is expressed as number
of
 nonconformities per 100 items, AQL values
up to
 1000 nonconformities per 100 items may be
used.
 Sampling:
Ø Sample selection must be drawn from the lot by
simple random sampling.
Ø When double or random sampling is to be used,
each subsequent sample shall be selected 11
Inspection Levels
Ø 4 special inspection levels - S1, S2, S3, S4.
Ø 3 general inspection levels - I, II, III.
Ø Special inspection levels used when
sample size
 must be kept small and larger sampling
risks can
 be tolerated.
Ø Level II will be used unless another
inspection level is specified.
Ø Level I is used when less discrimination is
required, Level III when greater
discrimination is
12
 required.
steps in the use of the
standard
be summarized as follows:


Ø Decide on the AQL.

Ø Decide on the inspection level.
Ø Determine the lot size.
Ø Enter the table to find sample size code
letter.
Ø Decide on type of sampling to be used.
Ø Enter proper table to find the plan to be
used.
Ø Begin with normal inspection, follow the
switching rules and the rule for stopping
the inspection (if needed). 13
Sample size code letters
Lot Size Special Inspection Levels General Insp. Levels
S1 S2 S3 S4 I II III
2-8 A A A A A A B
9-15 A A A A A B C
16-25 A A B B B C D
26-50 A B B C C D E
51-90 B B C C C E F
91-150 B B C D D F G
151-280 B C D E E G H
281-500 B C D E F H J
501-1200 C C E F G J K

14
Sample size code letters
(cont.)
Lot Size Special Inspection Levels General Insp. Levels
S1 S2 S3 S4 I II III
1201-3200 C D E G H K L
3201-10000 C D F G J L M
10001-35000 C D F H K M N
35001-150000 D E G J L N P
150000- D E G J M P Q
500000
>=500000 D E H K N Q R

15
 Single Sampling plans
selected segment of normal inspection table
Code Letter Sample Size AQL
0.010 0.015 0.025 0.040 0.065 0.10 0.15
Ac Re AC Re AC Re AC Re AC Re AC Re AC Re

G 32 ↓ ↓ ↓ ↓ ↓ ↓ ↓

H 50 ↓ ↓ ↓ ↓ ↓ ↓ ↓

J 80 ↓ ↓ ↓ ↓ ↓ ↓ 01

K 125 ↓ ↓ ↓ ↓ ↓ 01 ↑

L 200 ↓ ↓ ↓ ↓ 01 ↑ ↓

M 315 ↓ ↓ ↓ 01 ↑ ↓ 12

N 500 ↓ ↓ 01 ↑ ↓ 12 23

P 800 ↓ 01 ↑ ↓ 12 23 34

Q 1250 01 ↑ ↓ 12 23 34 56

R 2000 ↑ ↑ 12 23 34 56 78
16
Double Sampling Plans
normal Inspection
Code Sample Sample Size Cumulative AQL
Letter Samples
1.5 2.5 4.5 6.5 10

Ac Re AC Re AC Re AC Re AC Re

G FIRST 20 20 02 03 13 25 36

G SECOND 20 40 12 34 45 67 910

H FIRST 32 32 03 13 25 36 59

H SECOND 32 64 34 45 67 910 1213

J FIRST 50 50 13 25 36 59 711

J SECOND 50 100 45 67 910 1213 1819

17
OC Curves
Operating Characteristic Curves


Ø OC curve is a graph showing what any

particular sampling plan can be expected to
do in terms of accepting and rejecting
batches.
Ø An understanding of the implications of an OC
curve helps understand the risks to the
manufacturer, consumer, and in deciding
inspection levels and batch sizes.
Ø Each possible plan has its own OC curve.
Ø Horizontal scale - shows the percentage
defective.
Ø Vertical scale - shows the percentage of batches
that may be expected to be accepted if18
OC Curve
Ø Operating Characteristic (OC) Curve:
Probability curve that shows the
probabilities of accepting lots with various
fractions defective.

19
Typical OC Curve
1
Probability of accepting lot
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
3%
0.1
0
0 .05 .10 .15 .20 .25
Lot quality (fraction defective)

20
Decision Criteria

1.00
Probability of accepting lot
Ideal

Not very
discriminating

“Good” “Bad”

Lot quality (fraction defective)

21
Sampling Terms
Ø Acceptance quality level (AQL): the
percentage of defects at which
consumers are willing to accept lots as
“good”
Ø Lot tolerance percent defective
(LTPD): the upper limit on the
percentage of defects that a consumer
is willing to accept
Ø Consumer’s risk: the probability that a
lot contained defectives exceeding the
LTPD will be accepted
Ø Producer’s risk: the probability that a
lot containing the acceptable quality 22
level will be rejected
Consumer’s and Producer’s
Risk

1
α = .10
Probability of accepting lot
0.9
0.8
0.7
0.6
0.5
LTPD
0.4
0.3
0.2 “Good” Indifferent “Bad”

0.1 β = .10
0
0 .05 .10 .15 .20 .25
AQL Lot quality (fraction defective)

23
 OC Curve
e.g. (n = 10, c = 1)
1
.9139
0.9
Probability of acceptance

0.8 .7361
0.7
0.6 .5443
0.5
0.4 .3758
0.3 .2440
0.2 .1493
0.1 .0860
0
0 .10 .20 .30 .40 .50
Fraction defective in lot

24
Average Quality

Ø Average outgoing quality (AOQ):

Average of inspected lots (100%) and
uninspected lots
 N − n
AOQ = Pac × p 
 N 

Pac = Probability of accepting lot

p = Fraction defective
N = Lot size
n = Sample size

25
 Example : AOQ

0 0
0 .0 5 0 .0 4 6
AOQ ( Fraction defective out )

0.0 9 0 .1 0 .0 7 4 Approximate AOQL = . 082

0.0 8
0 .1 5 0 .0 8 2
0.0 7
0.0 6 0 .2 0 .0 7 5
0.0 50 .2 5 0 .0 6 1
0.0 4 0 .3 0 .0 4 5
0.0 3
0 .3 5 0 .0 3
0.0 2
0.0 1 0 .4 0 .0 1 9
0
0 0.0 5 0.1 0.1 5 0.2 0.2 5 0.3 0.3 5 0.4

Incoming fraction
defective

26
OC Curve

27
OC Curve example
Ø Code Letter M
Ø AQL 1.5
Ø Sample size - 315
Ø Accept: 10, Reject: 11
Ø What happens if a batch with 3% defectives is
submitted ?
Ø Find 3% on the X axis, and follow a vertical line up
until it meets the 1.5% curve.
Ø Take a horizontal line across to the Y axis, and read
off the value of 65%.
Ø I.e., 65% of batches would be accepted and 35%
rejected.
28
More OC curve examples
Pa AQL for Sample Plan M AQL for Sample Plan J
1.0 1.5 2.5 4.0 1.0 1.5 2.5 4.0
99 0.929 1.53 2.40 4.05 0.55 1.04 2.28 3.73
95 1.27 1.97 2.96 4.78 1.03 1.73 3.32 5.07
90 1.48 2.24 3.29 5.20 1.39 2.20 3.99 5.91
75 1.89 2.74 3.90 5.95 2.16 3.18 5.30 7.50
50 2.43 3.38 4.65 6.87 3.33 4.57 7.06 9.55
25 3.06 4.11 5.49 7.87 4.84 6.30 9.14 11.9
10 3.71 4.85 6.33 8.84 6.52 8.16 11.3 14.3
5 4.13 5.33 6.86 9.46 7.66 9.41 12.7 15.8
1 5.01 6.29 7.93 10.7 10.1 12.0 15.6 18.9

29
More examples
Ø Pinholes
Ø General inspection level 1, AQL 1.5%.
Ø Minimum Sample size code lettre L (200
Sample)
Ø If a batch with 3% pinholes is submitted,
there is a 75% chance of them being
accepted.
Ø Container
Ø Agreed AQL of 0.65%, general Inspection
level I.
Ø Sample code L (200 Sample).
Ø If a batch with 5% leakers is submitted,
there is a 99% chance of them being 30
Setting an Inspection Level
Ø Select the required AQL (as an average).
Ø Decide what quality should have a high
chance of rejection.
Ø Choose the appropriate sampling plan
by inspecting the OC curves.

31
 Setting an Inspection Level
 Example
Ø An AQL of 1.5% defective items has been chosen.
Ø It is desired to have at least an 80% chance of
rejecting a 6%
 defective batch under normal inspection.
Ø Code letters A->J fail to meet the requirement.
Ø Code letter K almost meets it, codes L-P more than
meet it.
Ø Decide the batch size. If batch size is 1000, check
sample size code letters table - we can use
general inspection level III.
Ø The sampling plan would therefore be AQL=1.5%,
General
 Inspection Level=III.
32
Switching Rules
Ø Normal to tightened: Shall be implemented as soon as
two out
 of five (or fewer than five) consecutive lots have been
 non-acceptable on original inspection.
Ø Tightened to normal: Shall be reinstated when five
consecutive
 lots have been considered acceptable on original
inspection.
Ø Normal to reduced: implemented when the switching
score is at
 least 30, production is at a steady state, and reduced
inspection is
 considered desirable by the responsible authority.
Ø Discontinuation of inspection: is implemented when
the
 number of lots not accepted in a sequence of 33
consecutive lots on
Switching scores
Single Sampling Plans
Ø Used when deciding whether to switch to reduced
inspection.
Ø Set switching score to 0 at start of normal inspection.
Ø Score updated following the inspection of each lot.
Ø When the acceptance number is >= 2, add 3 to the
switching
 score if the lot would have been accepted if the AQL
had been
 one step tighter; otherwise reset score to 0.
Ø When the acceptance number is 0 or 1, add 2 to the
switching
 score if the lot is accepted; otherwise reset score to 0.
Ø As in the previous slide, when the switching score is at
least 30
 and production is at a steady state, switch to reduced 34
Important points
Ø Setting an AQL does not guarantee to the
customer that batches of a worse quality will
not be accepted.
Ø If the average quality of batches being
submitted are a little worse than the AQL, a
number of batches will probably be accepted
before a switch to tightened inspection is
called for.
Ø In general, the customer gets a quality which is,
on average, better than the AQL, since the
tables are arranged to provide an economic
incentive -the manufacturer cannot afford to
have more than a small proportion of batches
rejected, and so will improve the quality if this
proportion is exceeded. 35