STATISTICAL QUALITY CONTROL

CHAPTER FIVE:
ACCEPTANCE SAMPLING
 Acceptance Sampling

 Acceptance sampling is an important field of

statistical quality control that was popularized
by Dodge and Romig and originally applied by
the U.S. military to the testing of bullets
during World War II.
 The principle works in such a way that a

sample should be picked at random from the

lot, and on the basis of information that was
yielded by the sample, a decision should be
made either to accept or reject the lot.
 Acceptance Sampling (Cont’d)

 Acceptance sampling is "the middle of the

road" approach between no inspection and
100% inspection.
 Acceptance sampling is employed when

one or several of the following hold:

 Testing is destructive,

 The cost of 100% inspection is very

high,
 100% inspection takes too long.
 Lot Acceptance Sampling Plans

 A lot acceptance sampling plan (LASP) is a

sampling scheme and a set of rules for
making decisions.
 The decision, based on counting the
number of defectives in a sample, can be
to accept the lot, reject the lot, or even,
for multiple or sequential sampling
schemes.
LASPs fall into the following categories:

1. Single sampling plans

2. Double sampling plans
3. Multiple sampling plans
4. Sequential sampling plans
5. Skip lot sampling plans:
 single sampling plan

 In a single sampling plan, the information

obtained from one sample is used to make a
decision to accept or reject a lot.
 There are two parameters in this sampling

plan: the sample size n and the acceptance

number c.
 The plan operates as follows. A random
sample of size n is selected from the batch.
The number of nonconforming items or
nonconformities in the sample is found and
compared to the acceptance number c.
 If the observed
number is less than
or equal to the
acceptance number,
the lot is accepted.
 If more than c
nonconforming items
or nonconformities
are found in the
sample, the lot is
rejected.
 Double sampling plan

 A double sampling plan involves making a decision to

accept the lot, reject the lot, or take a second sample.
 If the inference from the first sample is that the lot
quality is quite good, the lot is accepted.
 If the inference is poor lot quality, the lot is rejected.
 If the first sample gives an inference of neither good
nor poor quality, a second sample is taken. Thereafter,
based on the combined number of nonconforming
items or nonconformities in both samples, a decision is
made to accept or reject the lot.
 Let's consider the following double
sampling plan where attribute inspection
is con­ducted to find the number of
nonconforming items:

N = 5000

n1 = 40 n2 = 60
c1 = 1 c2 = 5
r1 = 4 r2 = 6
 The working procedure for this plan is initially, to
select a random sample of 40 items from the lot of
size 5000.
 If 1 or fewer nonconforming items are found, the

lot is accepted, but if 4 or more nonconforming

items are found, the lot is rejected.
 If the observed number of nonconforming items is

2 or 3, a second sample of size 60 is selected.

 If the combined number of nonconforming items

from both samples is less than or equal to 5, the

lot is accepted; if it is 6 or more, the lot is rejected.
 iii. Multiple sampling plans:

 In this plan more than two samples are

needed to reach a conclusion.
 Three, four, five, or as many samples as

desired may be needed to make a decision

regarding the lot.
 Lot Acceptance Sampling Plans
(Cont’d)

iv. Sequential sampling plans: This is the

ultimate extension of multiple sampling where
items are selected from a lot one at a time and
after inspection of each item a decision is
made to accept or reject the lot or select
another unit.
v. Skip lot sampling plans: Skip lot sampling
means that only a fraction of the submitted
lots are inspected.
Definitions of basic Acceptance Sampling
terms:

 Acceptable Quality Level (AQL):

 The AQL is the notion of the quality level of

lots that defines acceptable level.

 It is a percent defective that is the base line

requirement for the quality of the producer's

product.
 The producer would like to design a sampling

plan such that there is a high probability of

accepting a lot that has a defect level less
than or equal to the AQL.
 Lot Tolerance Percent Defective (LTPD):
 The LTPD is a designated high defect

level that would be unacceptable to the

consumer.
 The consumer would like the sampling

plan to have a low probability of

accepting a lot with a defect level as high
as the LTPD.
 Cont’d

 Type I Error (Producer's Risk): This is the

probability, for a given (n, c) sampling plan,
of rejecting a lot that has a defect level
equal to the AQL.
 Type II Error (Consumer's Risk): This is the

probability, for a given (n, c) sampling plan,

of accepting a lot with a defect level equal
to the LTPD.
Acceptance Sampling Risks
 EVALUATING SAMPLING PLANS

 There are some measures to evaluate the

goodness (performance) of a sampling plan.
 These involve the operating characteristics

curve, average quality level of batches

leaving the inspection station, the average
number of items inspected before making a
decision on the lot, and the average
amount of inspection per lot if a rejected lot
goes through 100% inspection.
 Operating Characteristic (OC)
Curve:
 OC curve plots the
probability of accepting
the lot (Y-axis) versus the
lot fraction or percent
defectives (X-axis).
 It measures the
performance of a
sampling plan.
 It shows the
discriminatory power of
the sampling plan.
 Operating Characteristic (OC)
Curve:
 For all sampling plans, we want to accept
lots with a low proportion nonconforming
most of the time and we do not want to
accept batches with a high proportion
nonconforming very often.
 The OC curve indicates the degree to

which we achieve this objective.

 Average Outgoing Quality (AOQ):
 A common procedure, when sampling and
testing is non-destructive, is to 100% inspect
rejected lots and replace all defectives with
good units.
 In this case, all rejected lots are made perfect

and the only defects left are those in lots that

were accepted.
 The AOQ is the average quality level (defect

level) of a series of lots that leave the inspection

station, assuming rectifying inspection.
 Cont’d

 In AOQ , if all lots come in with a defect

level of exactly p, and the OC curve for the
chosen (n, c) LASP indicates a probability
Pa of accepting such a lot, over the long
run the AOQ can easily be shown to be:
Pa P ( N  n)
AOQ 
N

Where; N is the lot size

 Average Outgoing Quality Level
(AOQL):
 The average outgoing quality limit (AOQL)
is the maximum value, or peak, of the AOQ
curve.
 It represents the worst average quality that

would leave the inspection station,

assuming rectification.
 A plot of the AOQ (Y-axis) versus the incoming lot p
(X-axis) will start at 0 for p = 0, and return to 0 for p
= 1 (where every lot is 100% inspected and
rectified).
 In between, it will rise to a maximum. This
maximum, which is the worst possible long term
AOQ, is called the AOQL.
 Average Total Inspection (ATI):

 The ATI represents the average number of

items inspected per lot.
 If a lot has no nonconforming items, it will

obviously be accepted by the chosen sampling

plan, and only n items (the sample size) will be
inspected for a lot.
 At the other extreme, if the lot has 100%

nonconforming items, the number inspected

per lot will be N (the lot size) assuming that
rejected lots are screened.
 For a lot quality between these extremes,
the average amount inspected per lot will
vary between these two values.
 For single sampling plans, the average

total inspection per lot for lots with an

incoming quality level p is given by:
ATI n  (1  Pa )( N  n)
 Designing sampling plans

i. Single Sampling Plan:

 A single sampling plan, as previously
defined, is specified by the pair of
numbers (n, c).
 The sample size is n, and the lot is
rejected if there are more than c
defectives in the sample; otherwise the
lot is accepted.
 Sampling Plans (Cont’d)

 There are two widely used ways of picking

(n, c):
 Use tables (such as MIL STD 105D) that
focus on either the AQL or the LTPD
desired.
 Specify 2 desired points on the OC curve
and solve for the (n, c) that uniquely
determines an OC curve going through
these points.
 Sampling Plans (Cont’d)

a. Military Standard 105E sampling plan:

 These Standard military sampling
procedures for inspection by attributes were
developed during World War II.
 After then many universities and
organizations adopted it with a little
modification.
b. Military Standard 105D sampling plan:
 This document is essentially a set of
individual plans, organized in a system of
sampling schemes.
 Sampling Plans (Cont’d)

 In applying the Mil. Std. 105D it is

expected that there is perfect agreement
between Producer and Consumer regarding
what the AQL is for a given product
characteristic.
 Sampling Plans (Cont’d)

The steps in the use of the standard can be

summarized as follows:
 Decide on the AQL.
 Decide on the inspection level.
 Determine the lot size.
 Enter the table to find sample size code
letter.
 Decide on type of sampling to be used.
 Enter proper table to find the plan to be used.
 Begin with normal inspection; follow the
switching rules and the rule for stopping the
inspection (if needed).
Example
For a lot size of 2000, an AQL of 0.65% , and an
inspection level of II, determine the single
sampling plans for normal inspection.
Solution
Solution
given :
N = 2,000
AQL = 0.65%

using the lot size N=2,000 and normal

inspection, the sample size code letter K is
obtained from the above Table I.
Solution (cont’d)
 Solution (cont’d)
Again from table II (single sampling plans for
Normal inspection), the desired plan is
obtained for code letter K and AQL 0.65%. It is
n=125, c=2 and r=3.
Thus, from a lot of 2,000, a random sample of
125 is inspected. If 2 or fewer nonconforming
units are found, the lot is accepted; if 3 or more
nonconforming units are found, the lot is not
accepted.
 Sampling Plans (Cont’d)

 c) Choosing a Sampling Plan with a given

OC Curve

1.0
0.8 n = 52
c=3
0.6
Pa
0.4
0.2
0
1 3 5 7 9 11 13 15 17 19
p percent defective per lot
 Sampling Plans (Cont’d)

 How the points on this curve are

obtained?
We assume that:
 The lot size N is very large, as compared
to the sample size n, so that removing
the sample doesn't significantly change
the remainder of the lot.
 The number of defectives, d, in a random
sample of n items is approximately
binomial with parameters n and p.
 Sampling Plans (Cont’d)

 The probability of observing exactly d

defectives is given by the binomial
distribution n!
P(d )  f (d )  p d (1  p ) n  d
d !(n  d )!

 The pa is the probability that d, the

number of defectives, is less than or equal
to c, the accept number. c
This means
n!
that :- P apd  c   d!(n  d )!p (1  p)
d 0
d n d
 Sampling Plans (Cont’d)
 Sample table for Pa, Pd using the binomial
distribution using this formula with n = 52
and c=3 and p = 0.01, 0.02, ...,.012 we
find
Pa 0.998 0.980 0.930 0.845 0.739 0.620 0.502 0.394 0.300 0.223 0.162 0.115

Pd 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.11 0.12
 Sampling Plans (Cont’d)

 Solving for (n, c)

 In order to design a sampling plan with a

specified OC curve one needs two designated

points.
 Let us design a sampling plan such that the
1 
probability of acceptance is for lots with

fraction defective p1 and the probability of
acceptance is for lots with fraction defective p2.
 Typical choices for these points are: p1 is the

AQL, p2 is the LTPD and, are the Producer's Risk

and Consumer's Risk, respectively.
 Sampling Plans (Cont’d)
 If we are willing to assume that binomial
sampling is valid, then the sample size n,
and the acceptance number c are the
solution to c
n!
1    p1d (1  p1 ) n  d
d 0 d !(n  d )!
c
n!
  p2d (1  p2 ) n  d
d 0 d !( n  d )!
Example
 Suppose AQL = .02 with α = 5% and LTPD

= .08 with β = 10%. Use Larson’s

nomograph to determine n and c.
Larson’s nomograph can be used as follows:
 The vertical line on the left-hand side is for lot

percentage defectives such as AQL and LTPD.

 The vertical line on the right-hand side is for the

probability of lot acceptance such as (1 – α) and β.

 Connect AQL with (1 – α) and LTPD with β with

straight lines.
 The intersection of these two lines gives the sample

size n and acceptance number c. In this case, n =

90 and c = 3.
 Sampling Plans (Cont’d)
 We can also calculate the AOQ for a (n, c) .
 Assume all lots come in with exactly a

proportion of p defectives.
 After screening a rejected lot, the final

fraction defectives will be zero for that lot.

 However, accepted lots have fraction

defective p0. Therefore, the outgoing lots

are a mixture of lots with fractions
defective p0 and 0.
 Assuming the lot size
p a p (is
N N,
n) we have.
AOQ 
N
 Sampling Plans (Cont’d)

 For example, let N = 10000, n = 52, c = 3,

and p, the quality of incoming lots, = 0.03.
Now at p = 0.03, we glean from the OC
curve table that pa = 0.930 and

AOQ = (0.930)*(0.03)*(10000-52) / 10000

= 0.02775
 Sampling Plans (Cont’d)
Sample plot of AOQ versus p
AOQ p A plot of the AOQ versus p
.0010 .01
.0196 .02
.0278 .03
.0338 .04 0.04
.0369 .05 0.035
.0372 .06 0.03

.0351 .07 0.025

.0315 .08 0.02

AOQ
.0270 .09 0.015

.0223 .10 0.01

.0178 .11 0.005

0
.0138 .12 0 0.025 0.05 0.075 0.10 0.125
p = Incoming Quality Level
 Sampling Plans (Cont’d)
 From examining the curve shown in figure
above, we observe that:
 When the incoming quality is very good,

then the outgoing quality is also very

good.
 When the incoming lot quality is very bad,

most of the lots are rejected and then

inspected. Therefore, the AOQ, becomes
very good.
 Sampling Plans (Cont’d)

 In between these extremes, the AOQ rises,

reaches a maximum, and then drops.
 The maximum ordinate on the AOQ curve

represents the worst possible quality. It is

called the average outgoing quality limit,
(AOQL).
 One final remark: if N >> n, then the AOQ

~ pa p .
 Average Total Inspection (ATI)

 Similarly the ATI for the above example can

be calculated as follows:
 If all lots contain zero defectives, no lot will

be rejected. If all items are defective, all

lots will be inspected, and the amount to
be inspected is N.
 Finally, if the lot quality is 0 < p < 1, the

average amount of inspection per lot will

vary between the sample size n, and the
lot size N.
 Average Total Inspection (ATI)
(Cont’d)
 Let the quality of the lot be p and the probability
of lot acceptance be pa, then the ATI per lot is
ATI = n + (1 - pa) (N - n)
 Example

 Let N = 10000, n = 52, c = 3, and p = .03

We know from the OC table that pa = 0.930.

Then,
ATI = 52 + (1-0.930) (10000 - 52) = 753
(Note that while 0.930 was rounded to three
decimal places, 753 was obtained using more
decimal places.)
 Average Total Inspection (ATI)
(Cont’d)
A plot of ATI versus p, the Incoming Lot
Quality
ATI (ILQ)
P
70 .01
253 .02
753 .03 10000

1584 .04 9000

2655 .05 8000

Average Total Inspection (ATI)

7000

3836 .06 6000

5007 .07 5000

4000
6083 .08 3000

7012 .09 2000

7779 .10 1000

8388 .11 0.02 0.04 0.06 0.08 0.1 0.12 0.14

p = Incoming Quality Level

8854 .12
9201 .13
9453 .14
Example
 Draw the OC curve for a situation in which a

sample of n = 10 items is drawn from a

lot containing N = 2,000 items, and the
lot is accepted if no more than c = 1
defect is found and rejected if 2 or more
defects are found in the sample.
 Because the sample size is small relative to
the lot size (10/2000 = .005 < .1), it is
reasonable to use the binomial distribution
to obtain the probabilities that a lot will be
accepted for various lot qualities.
 A portion of the cumulative binomial table
found at the end of this supplement is
reproduced here to facilitate the discussion.
Example
 Shipments of 300 boxes of glassware are

received at a warehouse. Random samples of

five boxes are checked, and the lot is rejected
if more than one box contains a breakage.
Construct the OC curve for this sampling
plan.
 Because n/N = 5/300 = .0167 < .1, binomial
distribution can be used to obtain the
probabilities of acceptance, Pac, for various lot
percentage defectives.
 A portion of the cumulative binomial table; n =
5 and c = 1:
OC
curve
Develop the AOQ
curve for the
previous problem
Example
 Construct the AOQ curve for N = 500, n = 10,

and c = 1:
 Let values of p vary from .05 to .40 in steps of
.05.
 You can read the probabilities of acceptance,
Pac, from the binomial table.
 Double
Sampling
 Application of double sampling requires that a
first sample of size n1 is taken at random from
the lot.
 The number of defectives is then counted and

compared to the first sample's acceptance

number a1 and rejection number r1.
 Denote the number of defectives in sample 1

by d1 and in sample 2 by d2, then:

 If d1 <= a1, the lot is accepted.
 If d1 >= r1 , the lot is rejected.
 If a1 < d1 < r1, a second sample is taken.
 Double Sampling
(Cont’d)

 If a second sample of size n2 is taken, the

number of defectives, d2, is counted.
 The total number of defectives is D 2 = d1 +

d2 and this is compared to the acceptance

number a2 and the rejection number r2 of
sample 2.
 In double sampling, r2 = a2 + a1 to ensure

a decision on the sample.

 If D2 <= a2, the lot is accepted.
 If D2 >= r2, the lot is rejected.
 Multiple Sampling
 It involves inspection of 1 to k successive samples
as required to reach an ultimate decision.
 Mil-Std 105D suggests k = 7 is a good number.
 Multiple sampling plans are usually presented in

tabular form:
 The procedure commences with taking a random

sample of size n1 from a large lot of size N and

counting the number of defectives, d1.
 if d1 <= a1 the lot is accepted.
 if d1 >= r1 the lot is rejected.
 if a1 < d1 < r1, another sample is taken
 Multiple Sampling (Cont’d)

 If subsequent samples are required, the first

sample procedure is repeated sample by
sample.
 For each sample, the total number of
defectives found iat any stage, say stage i, is
Di  d j
j1

ai
 This is compared ri with the acceptance
number and the rejection number for
that stage until a decision is made.
 Sequential Sampling
 Sequential sampling is different from
single, double or multiple sampling.
 Here one takes a sequence of samples

from a lot. How many total samples looked

at is a function of the results of the
sampling process.
 The sequence can be one sample at a time,

and then the sampling process is usually

called item-by-item sequential sampling.
 Sequential Sampling (Cont’d)
 The cumulative observed number of defectives
is plotted on the graph as shown in figure.
 If the plotted point falls within the parallel lines

the process continues by drawing another

sample.
 As soon as a point falls on or above the upper

line, the lot is rejected.

 And when a point falls on or below the lower

line, the lot is accepted.

 The process can theoretically last until the lot is

100% inspected.
 Sequential Sampling (Cont’d)

 Sequential sampling

x r = h 2+ sn
reject
Number of Defectives

continue sampling x a = - h1+ sn

h2 accept

h1
n, number of items
 Sequential Sampling (Cont’d)
 The equations for the two limit lines are
functions of the parameters p1, p2, and .
xa   h1  sn Acceptance line
xr  h2  sn Rejection line.
1   1  p1 
log s  (log 
Where  )/k
h1   1  p2 
k
1   p2 1  p1 
log k  (log  ( ) )
h2    p1 1  p2 
k
 Sequential Sampling (Cont’d)
 Example :
 As an example, let p1 = .01, p2 = .10, α = .05,
β = .10
 The resulting equations are
Both acceptance numbers and rejection numbers
must be integers.
x a   0.939  0.04n x r  1.205  0.04n
 Sequential Sampling (Cont’d)
 Thus for n = 1, the acceptance number = -1,
which is impossible, and the rejection number =
2, which is also impossible. For n = 24, the
n acceptance n n number n is 0 n and nthe rejection number

=
inspect
3. accept reject inspect accept reject
---------------------------------------------------------------------------------------------------
1 x x 14 x 2
2 x 2 15 x 2
3
4
x
x
2
2
16
17
x
x
3
3
So, for n = 24 the
5 x 2 18 x 3 acceptance
6 x 2 19 x 3 number is 0 and
7
8
x
x
2
2
20
21
x
x
3
3 the rejection
9 x 2 22 x 3 number is 3.
10 x 2 23 x 3
11 x 2 24 0 3
12 x 2 25 0 3
13 x 2 26 0 3
 Sequential Sampling (Cont’d)

Other sequential plans are given by;

n n n
inspect accept reject
----------------------------------------------
49 1 3
58 1 4
74 2 4
83 2 5
100 3 5
109 3 6
 Skip Lot Sampling

 Skip Lot sampling means that only a

fraction of the submitted lots are
inspected.
 This mode of sampling is of the cost-

saving variety in terms of time and effort.

 However skip-lot sampling should only be

used when it has been demonstrated that

the quality of the submitted product is
very good.
 Skip Lot Sampling (Cont’d)
 A skip-lot sampling plan is implemented as follows:
 Design a single sampling plan by specifying the alpha

and beta risks and the consumer/producer's risks.

This plan is called "the reference sampling plan".
 Start with normal lot-by-lot inspection, using the

reference plan.
 When a pre-specified number, i, of consecutive lots

are accepted, switch to inspecting only a fraction f of

the lots. The selection of the members of that
fraction is done at random.
 When a lot is rejected return to normal inspection.
 Skip Lot Sampling (Cont’d)
 The parameters f and i are essential to
calculating the probability of acceptance
for a skip-lot sampling plan.
 In this scheme, i, called the clearance

number, is a positive integer and the

sampling fraction f is such that 0 < f < 1.
 Hence, when f = 1 there is no longer skip-

lot sampling.
 Skip Lot Sampling (Cont’d)
 The calculation of the acceptance probability
for the skip-lot sampling plan is given by;
fp  (1  f ) p i
pa ( f , i) 
f  (1  f ) p i
Where: P is the probability of accepting a lot with a
given p, from the OC curve .
 The following relationships hold:
 for a given i, the smaller is f, the greater is Pa
 for a given f, the smaller is i, the greater is Pa