CHAPTER 08

Statistical Quality Control

1 INTRODUCTION
Theauality of a product is the'degree of perfection'. Every product has to be manufactured
according to
n specifications e.g., dimensions, strength,appearance, surface finish, hardness, etc. The quality of the
uct depends upon a number of factors such as its design and specifications, production process, quality of
a materials, quality of machines and
equipments, manpower expertise and skill and the inspection.
According to Alford and Beatly, "Quality control is defined as an industrial management technique by
=sof which products of uniform acceptable quality are manufactured. It is concerned mainly with making
ms right rather than discovering and rejecting those made wrong." It is preventive rather than a corrective
SUE. The corrective action rests with the personnel.
One simple way to control the quality is to conduct 100% inspection. However, it will be very costly and

suming. Nowadays statistics is used for quality control and this method is known as Statistical Quality
SOC)
Objectives of Quality Control
to achieve and maintain.
Se-up standards of quality acceptable to the customer and economical
and waste.
Ocate and identify the process faults in orderto control the defectives, scrap
3. of the products.
ake necessary corrective measures to maintain the quality
4. not reach the customers.
ure that sub-standards products do
3
Adv
avantages of Statistical Quality Control
. will spoil all the
When destructive testing, then 100% testing
quality of product is tested by destructed in testing.
few products will be
ucts. Under statistical qualitv control very
It ensur control, aintenance and improvement in the quality standards.

3 It cost.
arm

provides better quality

4. It red
assurance at lower inspective
to the minimum.
It reduces the inspection
and

he wastage of time and material

manufacturing cost and enhances profits.

Copyight MADE ERSY
Publlcations
www.madeeasypublications.org
 168 Mechanical Engineering Industrial Engineering
MAD
8.4 Causes of Variation in Quality ERSY
Pubic2.us
Variation in quality of the products can be due to chance causes or assignable cai.
Chance causes: These causes varyto random independent of each other and it ic
tural to
as uneconomical to detect and eliminate them. These causes are natur.
any ficult as wel
and are beyond human control. For instance, slight variation in temperature, manufactur
pressi ro rOCes
etc. interact randomly to produce slight variation in the quality of the product.
Assignable Causes: These are non-random causes that can be identified. The assinn
dhumidity
may creep in at any stage of the process, right from the arrival of the raw material to cause
the
the finished product. some of the assignable causes of variation are defective raw dispatd
materialeco
machine setups, worm equipments, unskilled and fatigued workers and so on.
The variaklOper
detected and eliminated and is, therefore, called controllable or preventable
variationvariation can be
8.5 Techniques of SQC
There are two broad ways ofcontrolling the quality of the product:
1. Process control: It is concerned with controlling the quality of the
product during the prodiuction
process. It ensures that a product of only requisite standard is produced and
makes use of tha
control charts'.
2. Product control: It is concerned with controlling the quality of the
product by critical examination
strategic points. It is concerned with inspection of goods already produced to ascertain whether at
are fit to be they
despatched or not. It makes use of sampling inspection plans to achieve the objective.
8.6 Control Charts
A control chart is an important statistical tool used for
the study and control of repetitive
The differences in dimensions or any other processes
quality the products are bound to happen if (i) different
of
machine tools are used (ii) different cutting tools are used (ii) materials of different properties are
used (iv) working conditions are different (v) workers of different skills manufacture the producls and
(vi) improper jigs, tools and fixtures are used.
However, variability in size is observed even if the
products are manufactured on the same macnine
tool, using the same cutting tool by the same worker on the
same material using proper jigs.
and fixtures.
A s already mentioned this
variability is due to chance causes as well as assignable cau t
chart accepts the normal variation due to
chance causes but eliminates entirely the
assignable causes. errd
Chance (usual) variations are riations,

normally of a lesser magnitude than asssignable (unusuaiits 8re

occur randomly and can be
described by the normal probability distribution curves.
defined within which variations are a
acceptable and beyond which they are unacceptaoE limits
probability distribution, 99.73 per cent of all chance variations are expected to
three standard deviations larger or smaller o n beyond
than the mean value of variable. Any
these limits, therefore, can be expected to be vae n e e d t o

caused by unusual assignable cause Whose

be investigated and then controlled. A items

process is said to be 'in control' if it proau

attributes or variables fall within the acceptable
range and is said to be 'out or coru
items whose attributes or variables are
beyond the acceptable range.
o Copyright

www.madeeasypublications.org MADE ERSY

Pubtications
E EASY
Caro7S
Statisti ilQuality Cont
chart
c o n s i s t s
of 169
Control (CL) thatindicate the desired
centre line
control
or
control level of the
control limit (UCL) that indicates the process,
an upp
er upper tolerance limit,
control limit (LCL) hat indicates the lower tolerance imit.
0 a l o

art has ahorizontal

horizontal scl
scale that represents the
T h ec o n t r o l

cornsecutive sample number

the characteristic quality of each
sample. This is shown in Fig. 8.1.and a vertical
eallected on a series of samples taken during a
T h edata production
plotted on a graph. As long as the points fall within the process at different time
enals and arecontrol and we do not
control limits, the
tatistical question the quality of the product. However, if
process is
plotted point
under

a
control limits, this alerts the
falls utside the
production manager to the possibility that the
d tisis unacceptable and that the process is not under
the product quality of
statistical control. So he tries to
ermine whether variationis due to chance causes or some assignable causes.

ontrol chart indicates that the observed variation is due to chance causes alone, the
process
iseoid
saidtto be 'in control'.
If on the other hand, the control chart
indicates that the observed variation is
nat likely to be due to chance, it can be said that the manufacturing process is 'out of control". In that
event, the process is halted and effort is made to seek and correct the possible cause.

Out of control
Upper Control Limit (UCL)
----~------------------

Centre Line (CL) Acceptable

quality
(under control)

Lower Control Limit (LCL)

--------- ----------------------
Out of control

1 2 3 4 5 6 7 8 9 10 11 12 13
Sample number
Figure 8.1

wO
types of control charts are usually used level
charts for variables: Thev are used to achieve
and maintain an acceptable qualiy
ri measurements such as alame
can be subjected to
quantitative
f e n g twhose
process h of aproduct
bolt, thickness of a pipe, specific resistance of a wire and strength ofa

yarn, etc.
(i) Control char acceptable qualnyleve
and maintain an
Cnarts for attributes: Thev are used to achieve measurement but can
be
f subjected to quantitative surface finish of
a
c a n not be
w n o s e product acceptable'. For instance
or 'not
or "bad', as 'acceptable'
9Ood or not acceptable
either acceptable
oOur or brightness of an article is

ontrol Charts for Variables

hey are use characteristics. They aretwo Kinds:
measurable quality

.
Control harts for sample (T -charts) means

Control chartstfor sample ranges (R-charts) www.madeeasypublications.org

ytight MADE ERSY

Publications
 170 Mechanical Engineering Industrial Engineering9 MADE ERSy
These charts are constructed as follows:
-Publications
Step 1: A random sample of size n (n is usually 4 to 5 units) is taken during a manufacturins
g
Over a period of time and the quality measurements x. I2 X, are noted. process
Tep 2:1 he sample mean + and the sample range Rare calculated using

n 'i=1

and R max min

where xmax and xmin are the largest and smallest values respectively of measurements x,. x
Step 3: If the process is found to be stable, k successive samples (kusually varies between 20to?
to 30)
are selected and for each sample, mean and Range Rare calculated. I heni and R are computed iie.

=tt+_
K

and
K i=
Chart
(a) Centre line = , when specification limits are not given,

LSL+USL when
2 specification limits are given.
(6) The upper and lower control limits are set using

Here A, is found from table 8.1 and depends on the sample size n.

Remember: UCL = ++AR and LCL = -AR,

Chart
(a) Centre line = R
(b) The upper and lower control limits for R-chart are set using
Here D and D
L
are also found from table 8.1 and depend on the sample size n.

Remember UCL DAR and LCL =DR

Step 4: The upper and lower process tolerance limits
(also called natural tolerance limits) indiviaua
values of x are computed by using

3R
UTL + d2 and
LTL=d
=

Here d, is also found from table 8.1 and depends nits

upon the sample size. If these natural tolerance i
fall within the customer's specifications, the process is said to tomer
be capable of meeting the cu
specifications. The process capability

6o 6
d2

www.madeeasypublications.org MADE ERSY O Copyright

Publications
 EASY
Statistical Quality
VPublications Control 171
Table 8.1
Factors used in the ï and R
Quality Control Charts
Factors for R-chart
No. of units Factor for Lower control
in a sample T-chart limit (D) Upper control Factor for
limit (D)
(n) (A) estimating o
from R(d,)
2 1.88
3.27 1.13
3 1.02 0
2.57 1.69
4 0.73
2.28 2.06
5 0.58
2.11 2.33
6 0.48
2.00 2.53
7 0.42 0.08 1.92 2.70
8 0.37 0.14 1.86 2.85
9 0.34 0.18 1.82 2.97
10 0.31 0.22 1.78 3.08
11 0.29 0.26 1.74 3.17
12 0.27 0.28 1.72 3.26
13 0.25 0.31 1.69 3.34
14 0.24 0.33 1.67 3.41
15 0.22 0.35 1.65 3.47
16 0.21 0.36 1.64 3.53
7 0.20 0.38 1.62 3.59
18 0.19 0.39 1.61 3.64
19 0.19 0.40 1.61 3.69
20 0.18 0.41 1.59 3.74

Example 8.1 A company manufactures screws to a nominal diameter 0.500 t 0.030 cm.
amples were taken randomly from the manufactured lot and 3 measurements were taken on
ample at different lengths. The readings are shown in the table below.

Sample No. Measurements per sample (cm)x

1 2 3

0.488 0.489 0.505

0.494 0.495 0.499
0.498 0.515 0.487
0.492 0.509 0.514
0.490 0.508 0.499

Calculate the control limits on a n d R-charts and draw the charts.

Solution:
Calculation of mean
For sarmple1, 0.488+0.489 +0.505 -
0.494
3
For sample 2, 0.494+0.495+0.499
= 0.496

Srilarly 3 = 0.500, F4 0.505, T

=
= 0.499

t + g +*s =0.499
yiant MADE ERSY
www.madeeasypublications.orgg

Ditbhlications
172 Mechanical Engineering Industrial Engineering MADE ERSY
Calculation of range blications
For sample 1, maxmin) for sample 1
R =

= 0.505-0.488 . 0 1 7
Similarly. R2 0.499 0.494 0.005,
= =

Ra= 0.515-0.487 0.028,

R 0.508-0.490 0.018
= =

A A+R2+Ra+Ra +Rs = 0.018

5

F-chart
I0.52 - - - -

UCL 0.5174

0.50 CL = 0.4990

0.48
LCL 0.4807

1 4 5
Sample number
(a)

t0.05
0.05 - -R-chart
----- UCL = 0.0463

0.04

O.03

0.02
CL 0.0180
0.01

0.00 LCL 0.0000

4 5
Sample number
(b)
Trial control limits:
Control limits for -charts

CL =
= 0.499,
UCL
+ ApR =0.499+ 1.02x 0.018 0.5174,
=

LCL -
These control limits lie
AR =0.499 1.02 0.018 0.4807.
=
x

within the manufacturer's assigned limits of

Control limits for R-chart 0.50+0.03 0.53 and-0.03 =04 =

CL = R= 0.018,
UCL =
D4R =2.57 x 0.018 0.0463, (From table 8.1 tor ns
LCL =
DR
=0x0.018 O.00.
The 7-chart and R-chart
are shown plotted figure (a) and (b)
in
It can be seen from the figures that no value
of ï or Ais out of
respectively.
the control limits. Therefore, tne al
control limits are the actual control limits. The
process is in statistical control i.e., it is
only under the influence of chance causesoperating
from assignable causes of variations and is
of varianO
www.madeeasypublications.org MADE ERSY o Copyright
Publications
 P u bb l i c a l

al Quality Control

173
The
lowing are the mean
g a m p l e8 . 2

lengths
ranges of lengths of and
samples eachof size 5. The
ach of s
10 specification limits for length finished
a
om

Aoanils.
s and exanine whether the process is
are
under control and 200 t5
cm. Construct
1 2 3 state your
Samnple
no.

201 198
4 5 6 7 8
recommendations.
202 200 203 9 10
204 199 196
Mean
5 7 3 199 201
4 7
me for n= 5, Az =
Range
577, D = 0, D = 2.115 2 8 5 6
A S S U .

Solution:

mhe
specification
limits for length are given to be 200t5 cm. Hence,
mean is known whereas standard
viation is u n k n o w n .

Controlimits for F-chart

Central limit, CL = 200

UCL = H+ AR,
=2_47
where 10
UCL =
200+0.577 x4.7 202.71,
LCL =
u -

AR =
200- 0.577 x 4.7 197.29
Control limits for R-chart
CL R 4.7,
UCL = DR = 2.115 x 4.7 = 9.941,

LCL DR =0x 4.7 =0

The and R-charts are drawn in Figure (a) and (b) respectively.

204 F-chart
- - UCL = 202.71
203
202

201

200
CL 200
199
198
--- LCL= 197.29

197 FE
196

195 8 9 10
4 5 6 7
1 2 3
Sample number
(a)
variability is, therefore,
tcan b that all points lie within the control limits of
R-chart. The process
outside the control limits
under contro.
Ontrot H to sample
no. 5, 6 an 8 lie be
corresponding should
three points therefore,
1 wever, The process,

hal-chart.
statistical control. should be
The PIOcess
te to check process is, therefore, not in
are found,
the
If they
process

t oche any assignable

causes.

eadk
sadjusted
e d to torem
remov
tnere
are
f l u c t u a t i o n s are
going to be
there.

remove them, erwise

w w w . m a d e e a s y p u b l i c a t i o n s . o r g

MADE ERSY
Publications
 -

Sample range R

TT

9
 E ERS
Statistical
Puolicatie
cations

CL
Quality Control 175
R-cha
for
hart =
R =
0.0191,
UCL
l i m i t s

rpntrol DAR 2.11 x0.0091 0.0192,

=
=

LCL DR =0x0.0091 0.00

=

A-chart. are shown plotted in Figure

and below (a) and
f-chart is coming under (b) respectively.
ne that the process
statistical control
arti
sn d i c a t e t

from and it is
unacceptable (samplebeginning
from earlier samples were
earlier to settle down.
an
lengths probably
no. 3 and
the process was under better contr 4) but
Them

towards
of the day
12.006 -chart
12.004 UCL 12.0047
12.002
12.000
CL 11.9994
11.998
11.996
11.994
LCL 11.994
11.992
11.990
12 3 4 5 6 7 8 9 10 11
12 13 14
Sample number
(a)
R-chart
0.020 UCL 0.0192
0.018
0.016
0.014
0.012
0.010
CL = 0.0091
0.008
0.006
0.004
0.002
0.000 LCL 0.00
12 3 45 6 7 8 9 10 11 12 13 14
Sample number
(b)

Example 8.4
's of fuses. Constructa control chart for mean and the range for the following data on
o magnitude). samples of 5 being taken every hour (each set of 5 has been arranged in ascending
42
42 19
85 36 42 51 60 18 15 69 64 61
45 24
75 54 51 74 60 20 30 109 90 78
78 68 80 93 4
89 57 75 72 27 39 113
87 72 81 77 59 42 62 118 109 109
90 78 95
81 112 153 136
mment on whether 132 84
138 78
60
GTner the production process seems to be under control.
84

heht
MADE ERSY www.madeeasypublications.org
Pubiications
 Industrial Engineering MADE ERSY
Mechanical Engineering
176 Publicatuons
Solution: calculated in table below
variance Rare
Ihe sample mean r and sample
Sample observation,x
Sample
mean Sample range
d5
Sample
no. (1) (2)
347 69.4 45
75 78 87
42 65 63.4
72 90 317 48
45 68 57.0
31 81 285 2
19 24 80
84 340 68.0 53
36 54 89
78 287 57.4 36
51 57 59
42
132 410 82.0 81
74 75 78
51
138 425 85.0 78
60 72 95
60
60 167 33.4 42
20 27 42
8 18 46.0
39 62 84 230 69
9 15 30
118 153 562 1124 84
10 69 109 113
112 468 93.6 48
11 54 90 93 109
109 136 478 95.6 75
2 51 78 94
Total: 863.2 721

721
0 0 12 =71.92, A==
12
60.08

2.11, Dg 0
From table 8.1, for n 5, A
= =
0.58, D = =

Control limits for 7-charts CL = 71.92,

i+AR =71.92 +0.58 60.08 x 106.77.

LCL =- AR = 71.92-0.58x 60.08 37.07

Control limits for R-chart CL = R= 60.08,

UCL DR =2.11x 60.08 126.77,
LCL = DR =0x 60.08 =0.00.

The -chart and R-chart are shown plotted in Figure below (a) and (b) respectively.
Since sample mean corresponding to sample number 8 and 10 is outside the control limits, the
-chart indicates the presence of some assignable causes of variations which should be detected
and corrected. The R-chart showsthatthe process variability is under control since all sample range
points are within the control limits.

T-chart
110 ----- UCL= 106.77

100

90
80
70 CL 71.92

40 - - LCL = 37.07

30
1 2 3 4 5 6 7 8 9 10 11 12
Sample number
(a)

www.madeeasypublications.org MADE EASY Copyright

-Publications
 ERSY Statistical Quality
Publications trol 177
R-chart
130
120
- -

UCL =
126.77

110
100
I 90

80
70
60 CL 60.08
50
40
30
- LCL= 0.00
1 2 3 4 5 6 7 8 9 10 11 12
Sample number
(b)

Example 8.5
A drilling machine bores holes with a mean diameter of 0.5230 cm and a

d
deviation of 0.0032 cm. Calculated the 2-sigma and 3-sigma upper and lower control limits for
of 4.
safsample
Solution:
= 0.5230 cm, o= 0.0032 cm and n= 4
Here

(x, o) chart.
We have to construct

2sigma limits for means

CL 0.5230 cm,

UCL +2 =0.5230+2x0.0032
0.5262 cm,
n 4

LCL F-2 0.5230-2x =

0.5198 cm.
4
3-sigma limits for means
CL = 0.5230 cm,

UCL = +3=0.5230+3x 0.5278 cm,

f-3=0.5230-3x =0.5182 cm
=
4

Control Charts for Variables sometimes

ninspectiont done. However, this is
the dimensions is
aWell as vaables, actual measurement of

rspectionuneconomica.
by attributes is just th the other way of inspection.
In this method
actual
The
measurements are not

slead the numt defect and its location is not important.

size of
ds are of defectives are counted. The accepted or rejected.
s0a and are either
Same way as by GO' and 'NOT GO gauges

yight MADE ERSY

www.madeeasypublications.org

Publications
 178 Mechanical Engineering Industrial Engineering MADE ERSY
Publications
Some examples of control by atributes are
() Number of defectives in a lot.
(i) Number of mistakes on the part of a typist
(im) Number of spots in a distempered wall
(v) Number of absentees in a department
etc.
(v) Number of complaints on a product per month,

for attributes are

NOTE The three most commonly used control charts
1. Control chart for fraction defective (p-chart).
2. Control chart for number of defectives (np-chart).

3. Control chart for number of defects (C-chart).

Control Chart for Fraction Defective (p-Chart)
Let samples (10 to 20) ofsize n be taken from the production process randomly at different-intervals

If d is the number of defectives in a sample, then the fraction defective in the example, p =
n
No. of defectives in a sample
or actual number of defectives, d= np.
Total no. ofitems inthe samole
Then if p represents the proportion of defectives produced by the entire process, it will be given by

Total no. of defective items inallthe samplesinspected

Total no. of items in all the samples inspected
The statistical theory of binomial distribution is used to construct the p-Chart. According -- this distribution,

standard deviation o, is given by

E1-
n
Thus the control limits for the p-chart are
CL = p.

Since number of defectives cannot be negative, if LCL sometimes comes out to be negative, it is taken
as zero. p-chart is normally used to plot and control fraction defectives when the sample size does not remain
uniform or it varies. However, it may also be used when the sample size remains constant or uniform.

Remember
UCL +3o, B+3,l=p =

LCL= -3a, = p-3,P1-D

n

Control Chart for Number of Defectives (np-Chart)

Let samples of size n be taken from the production process randomly at different intervals or fdis

the number of defectives in the sample and p is the fraction defective in the
sample, then d =
np
Now if np represents the number of defectives per sample of a constant size, then

n otal no.of defectives of all samples

Number of samples inspected

www.madeeasypublications.org MADE ERSY Copyrighe

-Publications
 ERSY

Publhcations
Statistical Quality Control
ation a of the np-chart is given by
ndar
dard 179

the
Control
limits for the np-chart are 1-P)-Jnp1-P)
n

CL np
us

mes Out to be negative, it should be

taken as zero.
CL
defectives
tives whe when the samplesize remains
constant. np-chart is used to plot and
control thee

UCL np+30np np +3/np(1-).

LCL np-30np np-3/np(1-.

art for Number of Defects (C-Chart)

arol
nther method of plotting attribute characteristic. When the
f large number of components, such as in the manufacture ofproduction
erblyo a
process involves complex
an aircraft, an
cloth, etc. aa knowledge
knowledge of the number of defects automobile, a T.V., a
rouler,c
per unit is useful in
maintaining satisfactory level of
a
Insuch cases we can base control limits on the assumption that the Poisson distribution is applicable.
h mean and variance of a Poisson distribution are equal the standard
deviation is given by

where C denotes the average number of defects in a sample and is given by

Total no. of defects in all the samples

Total no. of samples inspected
Thus the control limits for the C-chart are
CL = C.
The C-chart relates to samples of constant size. In case the samples size varies, the control chart tor
2C
Er OT detects per unit, u = i s used. The average number of defects per unit, u and the contro

nts are Pubiications

given by
CL = U,

UCL k+3
LCL -3

Margin of safety =n -1

Remember UCL C+3o, =Õ+3T,

LCL -30 = -3NT www.madeeasypublicationsog

MADE ERSY

Publications
 188 Mechanical Engineering Industrial Engineering MADE EASY
-Pubiications
Example 8.14 The number of defects inspected in rotational subgroups of size 5 andtfor
a
sample of size 10 are as follows:

Sample No. Number of defects

1 6/5
2 0 3 3/5
3 2 1 7 7/5
4 2/5
2 1 8/5
6 0 8 8/5
7 0 0 3 3/5
8 2 1 1 8/5
9 2 5 1
10 3 2 3 9 9/5

Set up appropriate control chart and give your comments.

Solution:
Here, we shall construct a u-chart, If is the total number of defects in the th
c, rational group of 5, then
the number of defects per unit in the ith
sample is u, c/5.
The process average number of defects
per unit is given by
Zu 59
10 5 x 101.18
The control limits are CL = U= 1.18,

UCL
=+3 1.18+3 1.15 2.64

3.0 U - chart
2.5 UCL 2.64
2.0

81.5
1.0 CL 1.18
0.5
LCL 0.00
0 1 2 3 4 5 6 7 8 9 10
Sample number

LCL -3 1.18-3 1.18 -0.280

The control chart is shown in
Figure above.
As all the points lie within the control
limits, the process is under statistical controi.

www.madeeasypublications.org mADE ERSY O Copyright

Publications
VAE,EPE
Publications

The observations of inspection of television sets on different dates are given

Example 8.15

control limits
the 12
Calculate 3 5 6 7 8 9 10 11
Day 6
24 3 65 2 6 4 2 65 5
No. i n s p e c t e d

19 14 9 18 13 6 24 15 15
13 12 7
Defects persample

Solution:

size varies, the control chart for

number of defects per unit, u =cln is used.
Since the s a m p l e
are calculated. The calculations are
shown in table below.
arate limits for each subgroup
Separa

Number Total Defects UCL LCL

Day per unit u-3 uln
inspected defects
cin
k+3/uln
C
n
3.16 5.33/n 3.16-5.33/n
5.54 .78
15
15 5.54 .78
5
5.34 0.98
3 24
6.93 -0.61
4 2
3 5.83 0.50
12
3 5.34 98
18
4.5 6.93 -0.61
2
2.8 5.54 0.78
8 14
3.2 5.34 0.98
9 19
2.3 6.24 0.08
10 7

Total 44 139

= 3.16
Now En
these control limtis are the actual control limits.
Since the defects per unit lie within the control limits,

8.9 Product Control

semi-finished goods or finished goods
Product control is concerned with classification of raw materials,
to judge
or rejectable items. It is concerned with
the inspection of goods already produced
CCeptable is normally used for product control
which
these are fit to be used or dispatched. Acceptance sampling
ner or another sample be taken
for deciding
whether the lot should be accepted without further inspection
des
about the quality of the lot.

8.10 Acceptance Samplin9

detecive
inspection so that no
product is to conduct 100%
Simple way to control the quality of a
aue
the
uneconomical or even impoSsible
f ui pass unnoticed. However, 100% inspection may be
Tollowing reasons then
When the quality of a product is tested by destructive testing (e.g., testing of electricalTUS
100% testing will spoil all the products. inspectors
of the
100% inspection causes enormousfatigue which affects the accuracy and judgementou ment

deterioration.
Handlings of the product may induce defects or result in its www.madeeasypublicationsorg
MADE ERSY
OCopyright Pubiications
 190 Mechanical Engineering Industrial Engineering MADE ERSY
4.
Publications
Cost of inspection may be very high due to unpacking or dismanting of the product if it sor
so requires
and due to the use of special machines.
Inspection may be hazardous or even dangerous (e.g. testing of pressure vessels).
For these reasons, some form of sampling inspection is often required. In acceptance samplina
decisions about the quality of batches are made on the inspection of only a portion of the total number of ite the
T the sample of items conforms to the desired quality level, then the whole batch is accepted, if it does the
's
whole batch is rejected or subjected to further inspection. Acceptance sampling is, therefore, a quick, easve
economical method to take decisions about the quality levels of the products. However, the method involveand
certain amount of risk since the decisions of acceptance or rejection are based only on the sample which mavna
conform to the quality of the remaining batch.
Two types of risks viz. producer's risk and consumer's risk are involved in acceptance samplina, Ont
the
basis of sampl inspection under a sampling plan, there remains a possibility of accepting a lot which wise
would have been rejected i.e., the whole lot in reality does not conform to specifications, but in spite of this
fact,
the lot is accepted. This is known as consumer's risk. Similarly, there may be a possibility of rejecting a lot. on t
n the
basis of sampling inspection, which otherwise should have been accpeted ie., the lot in reality has a lower
percentage of reject items than specified but still the same is rejected. This will result in a loss to the producer
and is known as producer's risk or seller's risk. Acceptance sampling is used to help both the buyer and the seller
by providing the requisite degrees of protection from the risks to which they are exposed.
The two terms AQL (Acceptable Quality Level) arnd RQL (Reject Quality Level), or LTPD Iot Tolerance Per
cent Defective) are also used while developing an acceptance sampling plan. AQL reters to the maximum
percentage of defectives in the sample so that the lot is acceptable whereas RQL refers to the percentage of
defectives in the sample beyond which the lot is rejected.
All these parameters viz. AQL, RQL, consumer's risk and producers risk should be settled at the time of
contract between the producer on one side and the consumer on the other side. Once values for these parameters
are specified, a sampling plan is designed by applying appropriate techniques. In its simplest form, acceptance
sampling requires selecting a random sample of size number from a lot of size 'N', determining the number of
defectives in the sample and comparing this number with a predetermined value 'c', the critical value or the
acceptance number. If the number of defectives is less than or equal to 'c, the lot is accepted, otherwise it is
rejected
Acceptance sampling by attributes.
1.
2.
Acceptance sampling by attributes
Acceptance sampling by variables.
PubicEiOns
In acceptance sampling by attributes, the products are inspected on the basis of GO' and 'NOT GO
gauges. So the actual measurement is not done. If the product conforms to the given specfication it is acceptea,
otherwise rejected, however small mey be the error. Thus dimension the error is not important. For instance, ine
pipes without cracks are accepted, those with cracks are rejected, and size and shape of the crack is

measured.
In acceptance sampling by variables, actual measurement of dimensions or physical and che
properties are done.
Sampling inspection procedure consists of the following steps:
() Make inspection batches of the total quantity.
(i) Draw random samples from each batch.
(i) Carry out proper inspection to ascertain the quality of the random sample.
(iv) Select a sampling plan and establish an acceptable per cent defective.

O Copyright
www.madeeasypublications.org MADE ERSY
Publications
 ADE EASY
Publications
Statistical Quality Contro 191
Three types of sampling plans commonly used are:
plan
0 Single sampling
() Double sampling plan
m)Mutiple (sequential) sampling plan
The followings symb0lS will be used in all the
sampling plans:
N nurmber of products in a given batch,
n number of random samples drawn from the batch of N,
Cacceptance number i.e., maximum number of defectives allowed in a sample of size n.

8.11 Single Sampling Plan

The procedure of single sampling plan consists of the following steps:
1. Draw n number of samples from a batch of Nand
inspect them.
2 If the number of defectives in the sample of size nare s c, accept the batch of N.
3. ifthe number of defectives inn> c, inspect all the remaining (N-n) products.
4. Correct or replace all the defective pieces found.
For instance, assume N= 400, n = 20 and c = 2. According to this plan, 20 no. of samples would be
chosen at random out of 400 and inspected. If number of defectives is 2 or less, the whole batch of 400 will be
accepted. If the number of defectives is more than 2, then all the remaining 380 products would be inspected and
allthe defectives in them would either be rectified or replaced by good ones and only then the batch of 400 would
be accepted.

8.12 Operating Characteristic Curve (0.Cc.c.)

The graphical relationship between percentage defectives in the lots being submitted for inspection and
the probability of acceptance is termed as operating characteristics of a particular sampling plan. An operating
characteristic curve for a particular combination of n and cshows how well the given sampling plan discriminates
between good and bad lots. Figure 8.2 shows a typical .c.C.for a sampling plan with n= 50 and c=1. It gives
a clear picture about the probability of acceptance of a lot tor various values of per cent defectives in the lot.

100
Sampling plan
n= 50,
c =1
80

60
8

40

20

2 4 6
8 10 12
Actual per cent 14
defectives
Figure 8.2 0.C
curve for sampling plan if n 50, c 1

MADE ERSH
Public- wAAw madaa
Probability of acceptance, %
Probability of acceptance, %

(D
A D EE R S Y
Statistical Quality
Publications
Control| 193
1.
ACCADtable Quality Level (AQL). This is the maximum proportion of defectives that
will make the
lot definitely acceptabie.
Lot Tolerance Percentage Defective (LTPD): This is the maximum
will make the lot definitely unacceptable. proportion of defectives that

100
1-a

AQL LTPD
Actual % defectives in a lot -

Figure 8.5 0.C. Curve

Producer's Risk (a). This is the risk, the producer is willing to take that lots of the quality level AQL
3.
will be rejected, eventhough, they are acceptable. Usually a. 5%.
consumer is willing to take that lots of the quality
level
4. Consumer's Risk (B). This is the risk, the
10%.
LTPD will be accepted, eventhough, they are actually unacceptable. Usually. B
=

can be constructed, as shown in Figure

8.5.
With the above four items known, O.C.C. for a sampling plan for
the various values of probability of lots being accepted,
An O.C.C. can also be plotted. by determining
i.e., values of per cent defectives.
various Pvalues (by Poisson or binomial distribution).

It is also
load is observed sequential sampling plan.
Remember: 1. Maximum data handling analysis
known as Multiple sampling plan.
known rejectable quality level.
percent defective is also
as
2. Lot Tolerance

8.13 Double Sampling Plan

batch,
of products in a given
Let N no.
in the first random sample,
no. of products
n =

acceptable number for the first sample,

C, the second random sample
no. products in
of
n =

number for both the samples combined.

acceptance
C consists of the following steps:
=

double sampling
plan
The procedure for randomly from
Nand inspect them.
number of samples further inspection.
1. Choose n, wnole batch of Nwithout any
s accept tne
defectives is C,
2. If the no. of the whole batch of N.
defectives in n, > Cz, reject
then choose a sample
of n, and inspect n
3. If the no. of l e s s tnan
Dut C2.
defectives in n,
more
is
than
Cc o m b i n e dis S Ca. accept the batch of
N.
4 If the no. of and na
in n,
of d e f e c t i v e s
5 If the total no. MADE ERSY
www.madeeasypublications.org

Publications

oCopyright
 Industrial Engineering
MADE ERSY
194 Mechanical Engineering ublication
the whole batch of Nor iner
combined is > C2, reject inspect a
.lt the total no. of defectives in n, and n,
the Nitems and replace the defectives
correct or that less inspection is require
more efficient
in the sense as
Ihe double sampling plan is usually batch in case of double samplin

Compared to single sampling plan. The

number of products
drawn from a
are good, seldom is the ne"
g plan
Tor the first sample is comparatively quite
small. In general,
when the producis
nro need for
becomes more economical
than the single sampling plan. Also, the producer Also, the
s e c o n d sample and the plan chance before being rejected.
derives the satisfaction that each batch
is given a second to the double samplina n i a .
and C=3. Then according
20, c, =1, n, 30
=

Example: Let N= 500, n, =

procedure.
1. Each batch contains 500 products.
random from 500 and inspect them.
2. Pick up sample of 20 items at
a
without further inspection.
the batch of 500
3. If the no. of defectives is 1, accept
500.
the whole batch of
4 If the no. of defectives is> 3, reject
then take a second sample of 30 products fram
more than 1 but s 3,
5. If the no. of defectives in 20 is
480 products and inspect them.
the batch of remaining 500-20
=

in 20 30 50 products is s 3, then accept the batch of 5oo

500,
6 If the total number of defectives
otherwise reject it.

less inspection is required as compared to single sampling plan.

Remember: In double sampline plan

Plan
8.14 Multiple or Sequential Sampling decision --garding
samples may be taken to reach
a
In this sampling plan one, two, three or more
of this plan, let us a s s u m e - following data:
acceptance or rejection of a batch. To explain the procedure
Sample size Combined Acceptance Rejection
number c number
sample size

50 50 s0
First sample
Second sample 25 75
Third sample 25 100
25 125 3 5
Fourth sample
Fifth sample 25 150 4

Precedure
1. Pick up a sample of 50 items from the batch of Nand inspect them.
2. If there is no defective item, accept the whole batch of Nwithout further inspection.
3. If it contains 2 3 defectives, reject the batch of N.
4. If it contains 2 defectives, take a second sample of 25 at random and inspect them.
batch ol
5. If the total number of defective from the first and the second samples is 1, accept the
67. If the total defectives from first and second samples is 2 4, reject the batch of N.
If the total defectives is less than 4 and great than 1, take third sample of 25 and inspect u
8. If the total defectives in 50+25 +25 100 samples is s 2, accept the batch of N.

9 If the total defectives is 2 5, reject the batch of N.

10. If the total defectives is more than 2 but less than 5, take another sample of 25 and inspeut
this way the sampling is carried out.

O Copyright

www.madeeasypublications.org MADE EASY

Publications