NPTEL - Acceptance Sampling

Acceptance Sampling
Accepting or rejecting a lot (of

parts, components, etc.) based
on the inspection of a sample
drawn from it
Copyright Tapan P Bagchi 1

Acceptance Sampling
● Accept/reject entire lot based on sample results

● Created by Dodge and Romig during WWII
● Not consistent with TQM of Zero Defects
● Does not estimate the quality of the lot
A “Lot” of
goods

What is acceptance
sampling?
Lot Acceptance Sampling

¾ A SQC technique, where a random sample is
taken from a lot, and upon the results of
appraising the sample, the lot will either be
rejected or accepted
¾ A procedure for sentencing incoming batches
or lots of items without doing 100% inspection
¾ The most widely used sampling plans are given
by Military Standard (MIL-STD-105E)

What is acceptance
sampling?
z Purposes
¾ Determine the quality level of an incoming shipment
or at the end of production
¾ Judge whether quality level is within the level that has
been predetermined
z But! Acceptance sampling gives you no

idea about the process that is producing
those items!

Acceptance Sampling
z Another area of quality control and improvement

z Closely connected with inspection and testing of product
z Inspection can occur at many points in a process
Acceptance
Acceptance Sampling:
Sampling: the the inspection
inspection and
and classification
classification ofof aa
sample
sample of
of nits
nits selected
selected at
at random
random from
from aa larger
larger batch
batch or
or lot
lot
and
and ultimate
ultimate decision
decision about
about disposition
disposition of
of the
the lot
lot ––
Lot
Lot Disposition
Disposition or
or Lot
Lot Sentencing
Sentencing
Whenparts
When partsare
arereceived
received
Two common points of inspection
Afterproduction
After production
Sampling Plans
z Acceptance sampling: Form of inspection applied
to lots or batches of items before or after a
process, to judge conformance with
predetermined standards
z Sampling plans: Plans that specify lot size,
sample size, number of samples, and
acceptance/rejection criteria
¾ Single-sampling
¾ Double-sampling
¾ Multiple-sampling and Sequential sampling

Types of sampling plans
z Sampling by attributes vs. sampling by

variables
z Incoming vs. outgoing inspection
z Rectifying vs. non-rectifying inspection
¾ What is done with nonconforming items found during
inspection
¾ Defectives may be replaced by good items
z Single, double, multiple and sequential
plans

How acceptance sampling
works
z Attributes (“go no-go” inspection)

¾ Defectives-product acceptability across range
¾ Defects-number of defects per unit
z Variable (continuous measurement)
¾ Usually measured by mean and standard
deviation

Why use acceptance
sampling?
z Can do either 100% inspection, or inspect a sample

of a few items taken from the lot
z Complete inspection
¾ Inspecting each item produced to see if each item
meets the level desired
¾ Used when defective items would be very
detrimental in some way

Sampling Terms
z Acceptance quality level (AQL): the
percentage of defects at which consumers are
willing to accept lots as “good”
z Lot tolerance percent defective (LTPD or
RQL): the upper limit on the percentage of
defects that a consumer is willing to accept
z Consumer’s risk: the probability that a lot
contained defectives exceeding the LTPD will be
accepted
z Producer’s risk: the probability that a lot
containing the acceptable quality level will be
rejected
Why not 100% inspection?
Problems with 100% inspection

¾ Generally very expensive
¾ Can’t use when product must be destroyed to test it
¾ Handling by inspectors can induce defects
¾ Inspection may be very tedious so defective items
may slip through even 100% inspection

A Lot-by-Lot Sampling Plan
Count
N Accept or
n Number
(Lot) Conforming Reject Lot
z Specify the plan (n, c) given N

z For a lot size N, determine
¾ the sample size n, and
¾ the acceptance number c.
z Reject lot if number of defects > c
z Specify course of action if lot is rejected
Quality Definitions
z Acceptance quality level (AQL)

The smallest percentage of defectives that will
make the lot definitely acceptable. A quality level
that is the base line requirement of the customer
z RQL or Lot tolerance percent defective (LTPD)
Quality level that is unacceptable to the customer

How acceptance sampling
works
z Remember
¾ You are not measuring the quality of the lot,
but, you are to sentence the lot to either reject
or accept it
z Sampling involves risks:
¾ Good product may be rejected
¾ Bad product may be accepted
Æ Because we inspect only a sample, not the

whole lot!
Acceptance Sampling

Three approaches for sentencing a lot
1. Single
1. Single sampling
sampling
• For low cost or low impact material on the subsequent process
• From trusted supplier
3. Double/multiple
3. Double/multiple sampling
sampling
• when 100% inspection is too costly

• Trusted supplier with potential risk
2. Rectified
2. Rectified sampling
sampling
• Inspect every item in the lot, then remove the defective units by
discarding, reworking, or returning to suppliers
• For critical process. Defective input may result high failure cost
• From new supplier
Types of Sampling Plan
SamplingPlan
Sampling Planfor
forAttributes
Attributes SamplingPlan
Sampling Planfor
forVariables
Variables
1. Single-sampling plan Standard sample size

2. Double-sampling plan
3. Multiple-sampling plan Smaller sample size on average
4. Sequential sampling plan

Advantages and Disadvantages of
Sampling
SSP’s Advantages Disadvantages
z Less expensive z Risk of accepting a lot of
because of less poor quality
inspection
z Risk of rejecting a lot of
¾ Works with single sample
¾ Protects both producer
acceptable quality
and consumer z Requires planning and
z Rejection on entire lot documentation
motivates quality z Require extensive study
improvement for on customer’s
suppliers
requirement
Lot Formation
1. Lots should be such that …

¾ produced on the same machines, by same operators, from
common raw materials, at approximately the same time
period
2. Larger lots are better than smaller lots
¾ These are more representative of overall quality
3. Lots should be conformable to the material
handling systems and personnel

Random Sampling
z Inspected units should be
selected at random
z Inspected units should
represent all items in the lot
z Potential bad stories:
¾ Sampled from the front, or
top of pile
¾ Did not randomize
¾ Did not stratify the lot and
did not sample from each
stratum

The Single Sampling Plan
z The most common and easiest plan to use but not most
efficient in terms of average number of samples needed
z One sample drawn from the lot and 100% inspected
z Single sampling plan
N = lot size
n = sample size (randomized)
c = acceptance number
d = number of defective items in sample
z Rule: If d ≤ c, accept lot; else reject the lot

Take a randomized The Single
sample of size n Sampling
from the lot N procedure
Inspect all items in the

sample
Defectives found = d
Yes
d≤c? Accept lot
No
Reject lot
Return lot Do 100%

to supplier inspection
Single-Sampling Plans for
Attributes
z A lot of size N has been submitted for

inspection Example:
Example:
z Sample size n NN==10,000
10,000
n=
n=8989
z Acceptance number of defective c cc==22
z Lot sentencing is based on one sample of
size n
Sample
exactly n parts
N = 10,000
n = 89 and inspect
every part
Producer’s & Consumer’s Risks
due to mistaken sentencing of lot
z TYPE I ERROR = P(reject good lot)

α or Producer’s risk
5% is common
z TYPE II ERROR = P(accept bad lot)

β or Consumer’s risk
10% is typical value

Acceptance sampling contd.
z Producer’s risk
¾ Risk associated with a lot of acceptable quality rejected
z Alpha α
= Prob (committing Type I error)
= P(rejecting lot at AQL quality level)
= producers risk

Acceptance sampling contd.
z Consumer’s risk
¾ Receive shipment, assume good quality, actually bad
quality
z Beta β
= Prob (committing Type II error)
= Prob (accepting a lot at RQL quality level)
= consumers risk
The Operating Characteristic (OC) curve for a
sampling plan quantifies these risks

Take a randomized
sample of size n from The Single
the lot of Sampling
unknown quality p procedure

sample
Yes
d≤c? Accept lot
No
Reject lot
Return lot Do 100%

The Operating
Characteristic or OC
Curve
The OC curve indicates a
sampling plan’s ability to
discriminate between good and
bad lots

What is the Operating
Characteristic (OC) Curve?
z It is a graph of the % defective (p) in a lot or batch vs. the

probability that the sampling plan will accept the lot
z Shows probability of lot acceptance Pa as function of lot quality
level (p)
z It is based on the sampling plan
z Curve indicates discriminating power of the plan
z Aids in selection of plans that are effective in reducing risk
z Helps to keep the high cost of inspection down

OC Definitions on the Curve
100%
α = 0.10
90%
Probability of Accepting Lot
75%
50%
25%
RQL
AQL
Go
β = 0.10 od Indifferent Bad
.03 .06 .09
Lot Quality (Fraction Defective)
OC Curve Terms
z Acceptable Quality Level (AQL)
¾ Percentage of defective items a customer is willing
to accept from you (a property of mfg. process)
z Lot Tolerance Percent Defective (LTPD)
¾ Upper limit on the percentage of defects a customer
is willing to accept ( a property of the consumer)
z Average Outgoing Quality (AOQ)
¾ Average of rejected lots and accepted lots
z Average Outgoing Quality Limit (AOQL)
¾ Maximum AOQ for a range of fractions defective

The Perfect OC Curve
100%
75% This curve distinguishes

perfectly between good
and bad lots.
50%
What would allow you

to achieve a curve like
25%
this?
.03 .06 .09

OC Curves
100%
OC Curves come in
various shapes
75% depending on the
sample size and risk of
α and β errors
50%
This curve is more
discriminating
25% This curve is less

discriminating
.03 .06 .09

OC Curve and its interpretation
The oc-curve enables us to evaluate the probability of acceptance (Pa)

for any true lot quality level-on a what-if basis. This way, one can design
sampling plans that perform the way one wants.
The OC curve is Interpreted as follows:
y If the lot quality is 0.093 fraction defective, then Pa, is 0.05.
y If the lot quality is 0.018 fraction defective, then Pa, is 0.95.
Operating Characteristic Curve
1.00
α = 0.05
Probability of acceptance, Pa
{
0.80
OC curve for n and c

0.60
0.40
0.20
β = 0.10 {
0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20
Proportion defective p
AQL Copyright Tapan P Bagchi
LTPD 35
OC Curve helps visualize
producer’s and consumer’s points
z Type I and Type II decision errors correspond logically to
the two decision points on the oc-curve.
z Type I error -- Wrongful Rejection
A type I error is associated with the producer's point -- to
reject when the true value of the quality characteristic is
AQL. The risk of rejecting an AQL lot is the producer's
risk (α = alpha risk)
z Type II error -- Wrongful acceptance
A Type II error is to accept when the true value of the
quality characteristic is RQL -- at the consumer's point.
The risk of accepting a lot, if it is an RQL lot, is the
consumer's risk (ß = beta risk).
Decision Criteria
1.00
Probability of accepting lot
Ideal
Not very
discriminating
“Good” “Bad”
Lot quality (fraction defective)

Consumer’s and Producer’s Risk
1
α = .10
Probability of accepting lot
0.9
0.8
0.7
0.6
0.5
0.4 LTPD
0.3
0.2 “Good” Indifferent “Bad”
0.1 β = .10
0
0 .05 .10 .15 .20 .25
AQL Lot quality (fraction defective)
Example: QC Curve for 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
Types of OC Curves
z Type A
¾ Gives the probability of acceptance for an individual lot
coming from finite production
z Type B
¾ Give the probability of acceptance for lots coming from a
continuous process or infinite size lot

OC Curve Calculation
The Ways of Calculating OC Curves

¾ Binomial distribution (Type B)
¾ Hypergeometric distribution (Type A)
Pa = P(r defectives found in a sample of n)
¾ Poisson formula
P(r) = ( (np)r e-np)/ r!
¾ Larson nomogram

OC Curve Calculation by
Poisson distribution
z A Poisson formula can be used
¾ P(r) = ((np)r e-np) /r! = Prob(exactly r defectives in n)
z Poisson is a limit
¾ Limitations of using Poisson
n ≤ N/10 total batch

Little faith in Poisson probability calculation when n is
quite small and p quite large.
z For Poisson, Pa = P(r ≤ c)

For us, Pa = P(r ≤ c)

p
OC Curve Calculation by Binomial
Distribution
Note that we cannot always use the binomial distribution

because
Binomials are based on constant probabilities
N may not be infinite
p changes as items are drawn from the lot

OC Curve by Binomial Formula
Pa p (%)
.998 .01
.980 .02
Using this formula with n = 52 and .930 .03
c=3 and p = .01, .02, ...,.12 we find .845 .04
data values as shown on the right. .739 .05
This givens the plot shown below. .620 .06
.502 .07
.394 .08
.300 .09
.223 .10
.162 .11
.115 .12
The Ideal OC Curve
● Ideal curve would be Pa

perfectly perpendicular from
0 to 100% for a fraction 1.0
defective = AQL
● It will accept every lot with p
≤ AQL and reject every lot
with p > AQL
0.0
AQL pÆ

Guidelines for choosing
Producer’s and Consumer’s
decision points (AQL and RQL)
To choose the Producer's Point (AQL) in practice
z Lots at the producer's point quality level (AQL) should be
accepted most of the time. Define AQL accordingly. Take into
account historical quality levels and the consequence of returning
a lot to the producer at AQL quality level.
z Choose the producers risk of rejecting a lot that is of AQL quality.
Typical: α = 0.05.
To choose the Consumer's Point (RQL) in practice

z Lots at the consumer's point quality level (RQL) should be
rejected most of the time. Take into account the economic
consequence of the consumer’s accepting a lot at RQL quality
level and define RQL accordingly.
z Choose the consumers risk of accepting a lot that is of RQL
quality. Typical: β = 0.05.
The OC Curve by binomial formula
z Operating characteristic (OC)

curve
Curve plots the probability of accepting

the lot (Pa) versus the lot fraction
defective (p)

The OC Curve
for Single SP

Ideal OC curve
z Theoretically , it can be
achieved by 100%
inspection of the lot
z Due to sampling, ideal
OC curve cannot be
perfectly achieved.
z If sample size (n) is large,
the OC curve shape will
approach the ideal OC
curve

Properties of OC Curves
z The acceptance number c and sample size n are most

important factors in defining the OC curve
z Decreasing the acceptance number (c) is preferred over
increasing sample size (n)
z The larger the sample size n the steeper is the OC curve
(i.e., it becomes more discriminating between good and
bad lots)

Effects of n on OC curves
z As sample size (n)

increases, the OC
movement of OC
curve shape will
approach the ideal
OC curve ~ more
discriminating
z Note that c is kept
in proportional to n

Effects of c on OC curves
z Increasing
acceptance number
movement of OC (c) does not
significantly change
the shape of OC
z But Producer’s risk
(β) increases

OC curve
Acceptable quality level (AQL)
z The poorest quality level for the supplier’s process that a
consumer would consider to be acceptable
z A property of the supplier’s manufacturing process, not a property
of the sampling plan
Lot tolerance percent defective (LTPD)
z The protection obtained for individual lots of poor quality
z Also called rejectable quality level (RQL)
z LTPD is a level of lot quality specified by consumer, not a
characteristic of the sampling plan
z Sampling plans can be designed to have specified performance
at the AQL and the LTPD points

Type-A and Type-B OC Curves

Poisson distribution for Defects
z Poisson parameter: λ = np
z P(r) = (np)r e-np/r! = Prob(exactly r defectives in n)
z This formula may be used to formulate equations involving
AQL,RQL, α and β to given (n, c).
We can use Poisson tables to approximately solve these
equations. Poisson can approximate binomial probabilities if
n is large and p small.
Q. If we sample 50 items from a large lot, what is the probability
that 2 are defective if the defect rate (p) = .02? What is the
probability that no more than 3 defects are found out of the
50?

Hypergeometric Distribution
z Hypergeometric formula: ⎛ n − r ⎞⎛ r ⎞
⎜⎜ ⎟⎟⎜⎜ ⎟⎟
N − M ⎠⎝ M ⎠
P(r ) = ⎝
⎛n⎞
⎜⎜ ⎟⎟
⎝N⎠
r defectives in sample size n when M defectives are in N.

z This distribution is used when sampling from a small
population. It is used when the lot size is not significantly
greater than the sample size.
z (Can’t assume here each new part picked is unaffected by the
earlier samples drawn).
Q. A lot of 20 tires contains 5 defective ones (i.e., p = 0.25). If an
inspector randomly samples 4 items, what is the probability of
3 defective ones?

Properties of OC Curves for finite size lots

Properties of OC Curves
z If the acceptance level

c is changed, the
shape of the curve will
change. All curves
permit the same
fraction of sample to
be nonconforming.

“Rectifying” Inspection
When the lot is rejected, it requires corrective action:

100% inspection and replacement of defectives by
good parts
Rectified
Passed

Average Outgoing Quality (AOQ)
z Expected proportion of defective items passed to

customer
Pa p ( N − n)
AOQ with rectifying inspection =
N
z Average outgoing quality limit (AOQL) is
¾The “maximum” point on AOQ curve

Average outgoing quality -
AOQ
Given the lot size of N, sample
size of n, and fraction defective
of p
z n items in the sample that, after
inspection, contain no
defectives, because all
discovered defectives are If N >> n, then
replaced
AOQ = Pa p
z N – n items that, if the lot is
rejected, also contain no
defectives
z N – n items that, if the lot is
accepted p(N-n) defectives Copyright Tapan P Bagchi 62
Average Quality
z Average outgoing quality (AOQ):

Average of inspected lots (100%) and
uninspected lots
AOQ = Pa× p⎜⎜⎜ N − n ⎟⎟⎟

⎛ ⎞
⎝ N ⎠
Pa = Probability of accepting lot

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

AOQLimit (AOQL) Estimation
0 0
0.05 0.046
0.1 0.1 0.074
AOQ (Fraction defective out)
Approximate AOQL = .082

0.080.15 0.082
0.2 0.075
0.060.25 0.061
0.04 0.3 0.045
0.35 0.03
0.02
0.4 0.019
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Incoming fraction defective

AOQ Curve
0.015
AOQL
Average
Outgoing 0.010
Quality
0.005
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
AQL LTPD
(Incoming) Percent Defective

Average outgoing quality -
AOQ

Double Sampling Plans
z Take small initial sample

¾If # defectives < lower limit, accept
¾If # defectives > upper limit, reject
¾If
# defectives between limits, take second
sample
z Accept or reject lot based on 2 samples
z Less inspection than in single-sampling

Multiple Sampling Plans
z Advantage: Uses smaller sample sizes

z Take initial sample
¾If # defectives < lower limit, accept
¾If # defectives > upper limit, reject
¾If # defectives between limits, re-sample
z Continue sampling until accept or reject lot based on
all sample data

Sequential Sampling
z The ultimate extension of multiple sampling

z Items are selected from a lot one at a time
z After inspection of each sample a decision is
made to accept the lot, reject the lot, or to
select another item
In Skip Lot Sampling only a fraction of the lots

submitted are inspected

Sequential Sampling Plans
z Sequential Analysis is the technique by which we build up our
sample one item at a time, and after inspecting each item, ask
ourselves: "Can we be sure enough to accept or reject this batch
on the information so far collected?“
z Seq SP offers Its value is in enabling reliable conclusions to be
wrung from a minimum of data. This was deemed sufficient to
require that it be classified "Restricted " within the meaning of the
Espionage Act during the war of 1939-45.
z Used for lot-by-lot acceptance plans for applications needing the
minimum possible average sample numbers.
z The SPR type sequential sampling plan is also known as
Columbia sequential sampling because the methodology was
developed at Columbia University.

Example of Sequential
Decision Diagram
Results of sequential sampling data is plotted on the diagram

sequentially as it is collected. The accept or reject decision is made when
sample data plots on or over an acceptance line or rejection line. The
stair step pattern is caused by the discrete nature of the sample size and
the number of defectives in the n-table upon which this diagram is based.
Sequential Sampling: Acceptance
and Rejection Limit lines

Equations for the Limit Lines
The equations for the two limit lines are functions of

the parameters p1, , p2, and .
where

OC-Curve for Count Data in Sequential
Sampling—similar to other sampling plans

Choosing A Sampling Method:
Single, Double or Sequential?
z It is an economic decision
z Single sampling plans
¾ high sampling costs
z Double/Multiple sampling plans
¾ low sampling costs
¾ More complex to implement

Designing a Sampling
Plan
The plan must protect both the
consumer (the buyer) and the
supplier (the seller)

Designing The Take a randomized
sample of size n from
Single Sampling the lot of
plan unknown quality p

sample
Yes
d≤c? Accept lot
No
Reject lot
Return lot Do 100%

Sampling Plan Design by Binomial Distribution
z Binomial distribution:
P(x defectives in n) = [n!/(x!(n-x))!]px(1- p)n-x
Recall n!/(x!(n-x))! = ways to choose x in n
Q. If 4 samples (items) are chosen from a population
with a defect rate = .1, what is the probability that
a) exactly 1 out of 4 is defective?
b) at most 1 out of 4 is defective?

Designing a single-sampling plan
with a specified OC curve
To construct sampling plan such that
Probability of acceptance for lots with fraction defective p1
Two
simultaneous
conditions that
the (n, c) plan
should meet
Probability of acceptance for lots with fraction defective p2

Solving for (n, c)
To design a single sampling plan we need two points.
Typically these are p1 = AQL, p2 = LTPD and , are the
Producer's Risk (Type I error) and Consumer's Risk (Type
II error), respectively. By binomial formulas, n and c are
the solution to
These two simultaneous equations are nonlinear so there

is no simple, direct solution. The Larson nomogram can
help us here.
SSP Design
by Larson
Nomogram
● Applies to single
sampling plan
● Based on binomial
distribution
● Uses
1-α = Pa at AQL
β = Pa at RQL
● Can produce OC
curve

Two simultaneous conditions
that the (n, c) plan will meet
Designing a
single sampling
plan by Larson’s
Binomial
Nomogram
Inputs are AQL,

α, RQL and β.
Outputs are n and
c

The MIL-STD-105E approach
A Query from a Practitioner: Selecting AQL (acceptable quality levels)
“I'd like some guidance on selecting an acceptable quality level and inspection
levels when using sampling procedures and tables. For example, when I use
MIL-STD-105E, how do I to decide when I should use GI, GII or S2, S4?”
-- Confused in Columbus, Ohio
W. Edwards Deming observed that the main purpose of MIL-STD-105 was to
beat the vendor over the head.
"You cannot improve the quality in the process stream using this approach,"
cautions Don Wheeler, author of Understanding Statistical Process Control
(SPC Press, 1992). "Neither can you successfully filter out the bad stuff.
About the only place that this procedure will help is in trying to determine
which batches have already been screened and which batches are raw,
unscreened, run-of-the-mill bad stuff from your supplier. I taught these
techniques for years but have repented of this error in judgment. The only
appropriate levels of inspection are all or none. Anything else is just playing
roulette with the product."
MIL-STD-105E
z Original version (MIL STD 105A) issued in 1950 as tables; Last

version (MIL STD 105E) in 1989; ISO adopted it as ISO 2859
z Plan covers sampling by attributes for given lot size (N) and
acceptable quality level (AQL).
z Prescribes sample size n, acceptance number c, and rejection
number r
z Standard included three types of inspection—normal, tightened
and reduced and gives switching rules
z Plans assure producer’s risk (α) of 0.01 – 0.1. The only way to
control the consumer’s risk (β) is to change inspection level

Military Standard 105E
z Sampling procedure for inspection by attributes developed during
World War II and is the most widely used acceptance-sampling
system for attributes in the world today
z A collection of sampling schemes; therefore an acceptance-
sampling system
z Provides for three types of sampling: single, double, and multiple
z Primary focal point is the acceptable quality level (AQL)
¾ Different AQLs may be designated for different types of defects:
critical, major, and minor
¾ Generally specified in contract or by authority responsible for
sampling
z Sample size is determined by lot size and by choice of inspection
level
AQL Acceptance Sampling by
Attributes by MILSTD 105E
z Determine lot size N and AQL for the task at hand

z Decide the type of sampling—single, double, etc.
z Decide the state of inspection (e.g. normal)
z Decide the type of inspection level (usually II)
z Look at Table K for sample sizes
z Look at the sampling plans tables (e.g. Table IIA)
z Read n, Ac and Re numbers

Procedure
1. Choose the AQL
2. Choose the inspection level
3. Determine the lot size
4. Fine the appropriate sample size code letter (from
the MIL-STD 105E Table)
5. Determine the appropriate type of sampling plan to
use (single)
6. Enter the appropriate table to find the type of plan to
be used
7. Determine the corresponding normal and reduced
inspection plans to be used when required
MIL STD 105E Inspection
levels and Sample Size codes

OC Curves under different Inspection
Conditions

MIL STD 105E “Normal” Sample
Size n and defective limit c

The OC curve of a
MIL STD 105E Sampling Plan

MIL-STD-105E Normal Inspection

MIL-STD-105E Reduced Inspection

MIL-STD-105E Tightened Inspection

Sampling rules for switching between normal, tightened and
reduced inspection

OC Curves under different MIL
Inspection Conditions

Discussion of Military
Standard
z Several points about MIL STD 105E should be emphasized:
z MIL STD 105E is AQL-oriented, no concern for β
z Not all possible sample sizes are possible (2,3,5,8,13,20,32,50,
etc.)
z Sample sizes are related to lot sizes
z Switching rules are subject to criticism for both misswitching
between inspection plans and discontinuation even though there
has been no actual quality deterioration
z MIL STD 105E has been replaced by MIL STD 1916 by the US
Department of Defense as of October 2008
http://assist.daps.dla.mil/

How/When would you use
Acceptance Sampling?
z Advantages of acceptance sampling

¾ Less handling damages
¾ Fewer inspectors to put on payroll
¾ 100% inspection costs are to high
¾ 100% testing would take to long
z Acceptance sampling has some disadvantages
¾ Risk included in chance of bad lot “acceptance” and good
lot “rejection”
¾ Sample taken provides less information than 100%
inspection

Summary
z Many basic terms you need to know to be

able to understand acceptance sampling
¾ SPC, Accept a lot, Reject a lot, Complete
Inspection, AQL, LTPD, Sampling Plans,
Producer’s Risk, Consumer’s Risk, Alpha, Beta,
Defect, Defectives, Attributes, Variables, ASN,
ATI.

Some Key Definitions and Terms
Reference: NIST Engineering Statistics Handbook
Acceptable Quality Level (AQL): A percent defective

that is the base line requirement for the quality of the
producer's product. The producer would like the buyer 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) also called
RQL (Rejection Quality Level): 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.
Type I Error (Producer's Risk): The probability for a
given (n, c) sampling plan of rejecting a lot that has a
defect level equal to AQL. The producer suffers when
this occurs, because α lot with acceptable quality was
rejected. The symbol a is commonly used for the Type I
error and typical values for α to range from 0.2 to 0.01.
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. The consumer suffers
when this occurs, because a lot with unacceptable
quality was accepted. The symbol β is commonly used
for the Type II error and typical values range from 0.2 to
0.01.
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.
AOQ's refer to the long term defect level for this
combined LASP and 100% inspection of rejected lots
process. 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:
where N is the lot size. Copyright Tapan P Bagchi 102

Average Outgoing Quality Level (AOQL): 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): When rejected lots are

100% inspected (rectifying inspection), it is easy to
calculate the ATI if lots come consistently with a defect
level of p. For a sampling plan (n, c) with a probability Pa
of accepting a lot with defect level p, we have
ATI = n + (1 - Pa) (N - n)
where N is the lot size.
Average Sample Number (ASN):
For a single sampling plan (n, c) we know each
and every lot has a sample of size n taken and
inspected or tested.
For double, multiple and sequential plans, the
amount of sampling varies depending on the
number of defects observed.
For any given double, multiple or sequential plan, a
long term ASN can be calculated assuming all lots
come in with a defect level of p.
A plot of the ASN, versus the incoming defect level
p, describes the sampling efficiency of a given lot
sampling scheme.
Useful links
http://www.bioss.sari.ac.uk/smart/unix/mseqacc/slides/frames.htm
Acceptance Sampling Overview Text and Audio
http://iew3.technion.ac.il/sqconline/milstd105.html
Online calculator for acceptance sampling plans
http://www.stats.uwo.ca/courses/ss316b/2002/accept_02red.pdf
Acceptance sampling mathematical background
http://assist.daps.dla.mil/
The new US military sampling procedure

NPTEL - Acceptance Sampling

Transféré par

Informations du document

Description originale:

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

NPTEL - Acceptance Sampling

Transféré par

Droits d'auteur :

Formats disponibles

Acceptance Sampling

Accepting or rejecting a lot (of

Copyright Tapan P Bagchi 1

● Accept/reject entire lot based on sample results

Copyright Tapan P Bagchi 2

Lot Acceptance Sampling

Copyright Tapan P Bagchi 3

z But! Acceptance sampling gives you no

Copyright Tapan P Bagchi 4

z Another area of quality control and improvement

Copyright Tapan P Bagchi 6

z Sampling by attributes vs. sampling by

Copyright Tapan P Bagchi 7

z Attributes (“go no-go” inspection)

Copyright Tapan P Bagchi 8

z Can do either 100% inspection, or inspect a sample

Copyright Tapan P Bagchi 9

Problems with 100% inspection

Copyright Tapan P Bagchi 11

z Specify the plan (n, c) given N

z Acceptance quality level (AQL)

Copyright Tapan P Bagchi 13

Æ Because we inspect only a sample, not the

Copyright Tapan P Bagchi 15

• when 100% inspection is too costly

1. Single-sampling plan Standard sample size

4. Sequential sampling plan

Copyright Tapan P Bagchi 17

1. Lots should be such that …

Copyright Tapan P Bagchi 19

Copyright Tapan P Bagchi 20

Copyright Tapan P Bagchi 21

Inspect all items in the

Return lot Do 100%

z A lot of size N has been submitted for

z TYPE I ERROR = P(reject good lot)

z TYPE II ERROR = P(accept bad lot)

Copyright Tapan P Bagchi 24

Copyright Tapan P Bagchi 25

Copyright Tapan P Bagchi 26

Inspect all items in the

Return lot Do 100%

Copyright Tapan P Bagchi 28

z It is a graph of the % defective (p) in a lot or batch vs. the

Copyright Tapan P Bagchi 29

Copyright Tapan P Bagchi 31

75% This curve distinguishes

What would allow you

.03 .06 .09

25% This curve is less

.03 .06 .09

The oc-curve enables us to evaluate the probability of acceptance (Pa)

OC curve for n and c

Lot quality (fraction defective)

Copyright Tapan P Bagchi 40

The Ways of Calculating OC Curves

Copyright Tapan P Bagchi 41

 n ≤ N/10 total batch

Copyright Tapan P Bagchi 42

Copyright Tapan P Bagchi 43

Note that we cannot always use the binomial distribution

Copyright Tapan P Bagchi 44

n ≤ N/10 total batch