Topic : Control Charts for Attributes

Subject Code : HMMS18005

Subject Name : Statistical Quality Control and
Reliability
Delivered by : Dr. A. Rathi
Introduction
• Many quality characteristics cannot be conveniently represented numerically.
• In such cases, we usually classify each item inspected as either conforming or
nonconforming to the specifications on that quality characteristic.
• The terminology defective or non-defective is often used to identify these two
classifications of product.
• More recently, the terminology conforming and nonconforming has become
popular. Quality characteristics of this type are called attributes.
Examples:
• The proportion of warped automobile engine connecting rods in a day’s
production.
• The number of non-functional semiconductor chips on a wafer.
• The number of errors or mistakes made in completing a loan application.
• The number of medical errors made in a hospital.
Attributes:
The term attributes in quality control refers to those quality characteristics,
which classify the items / units into one of the two classes:
• Conforming / Defective / Good
• Non – conforming / Non – Defective / Bad
There are two types of Attributes, They are:
(i) Where numerical measurements of the quality characteristics are not
possible. for example, colour, scratches, damages, missing pats etc.,
(ii) Where numerical measurements of the quality characteristics are possible
and items are classified as defective or non defective on the basis of the
inspection. For example, the diameter of a cricket ball can be measured by
the micrometre. but sometimes it may be more convenient to classify the
balls as defective and non – defective using a Go – No Go gauge.
• In inspection of attributes, actual measurements are not done, but
the number of defective items or number of defects in the item is
counted. The size of defect and its location is not important.
• Items are inspected and either accepted or rejected.
• There are different types of control charts for attributes for different
situations. We classify the control charts for attributes into two
groups as follows:
(i) Control Chart for Defectives.
(ii) Control Chart for defects.
Control chart for defectives are mainly of two types as given below:
(i) Control Chart for fraction defective (p – Chart)
(ii) Control Chart for number of defectives (np – Chart / d - Chart)
Both control charts for defectives are based on the Binomial
distribution.

Control chart for defects are also of two types as given below:
(i) Control chart for number of defects ( c – Chart)
(ii) Control chart for number of defects per unit (U – Chart)
Both control charts for defects are based on the Poisson distribution.
Control Charts for Fraction Defective (p – Chart):
• The most widely used control chart for attributes is the fraction (proportion)
defective chart, that is p- chart.
• The p – chart may be applied to quality characteristics, which cannot be
measured or impracticable and uneconomical to measure it.
Fraction Defective:
Fraction Defective (p) is defined as the ratio of the number of defective items/
units/ articles found in any inspection to the total number of items / units / articles
inspected. Symbolically we write,
Fraction Defective (p) =
For example, if 500 cricket balls are inspected and 20 balls are found to be defective, the
fraction defective of the balls is given as,
Fraction Defective (p) = 20 / 500 = 0.04
Control Limits for p – chart when
p is known:
The Mean and Variance of the sampling distribution of the fraction
defective is given by,
E(p) = P and Var (p) =
Here, P is the probability or fraction or proportion of defectives in the
process and Q = 1-P.
We also know the standard deviation of a random variable X is,
S.D(X) =
Therefore, the centre line and control limits for the p – chart are given
as follows:
CL = E(p) = P
UCL = E(p) + 3SD(p)
=p+3
UCL= p + 3
LCL = E(p) - 3SD(p)
=p-3
LCL= p - 3
 Control Limits for p – chart when p
is unknown
If are the fraction defectives of the 1st , 2nd ,….., sample respectively, we
calculate the average fraction defective as follows:

If the number of defective items for each sample is given, it is convenient to

calculate the average fraction defective as follows:
In this case, the centre Line and control limits of the p – chart is given
as:
CL = E(p) =
UCL = E(p) + 3SD(p) = + 3
LCL = E(p) - 3SD(p) = - 3
Revised Control Limits:
To bring the process under statistical control it is necessary to
investigate the assignable causes are eliminate, we delete the out of
control points and calculate the revised centre line and control limits
for the p – chart by using the remaining samples. These limits are
known as the revised control limits.
To calculate the revised limits of the p – chart, we first calculate the
new as follows:
= or =
Where d = the number of discarded samples,
= the sum of fraction defectives in the discarded samples
= the sum of number of defectives in the discarded samples
After finding , we reconstruct the centre line and control limits of the p
– chart by replacing by

CL = E(p) =
UCL = E(p) + 3SE(p) = + 3
LCL = E(p) - 3SE(p) = - 3
Example1:
A factory manufacturing small bolts. To check the quality of the bolts, the
manufacturer selected 20 samples of sample size 100 each from the manufacturing
process time to time. He / She visually inspected each selected bolt for criterion
defects. After the inspection, he/ she obtained the following data:
Sample 1 2 3 4 5 6 7 8 9 10
Number
Proportion 0.10 0.04 0.08 0.15 0.08 0 0.01 0.05 0.05 0.08
Defective
Sample 11 12 13 14 15 16 17 18 19 20
Number
Proportion 0.10 0 0.06 0.05 0.03 0.20 0.05 0.07 0.01 0.08
Defective

Estimate the proportion defective of the process. Does the process appear to be
under control with respect to the proportion of defective bolts.
Solution:
= X 1.29 = 0.065
CL = 0.065
UCL = + 3 = 0.065 + 3
= 0.065 + (3X 0.025) = 0.140
UCL = 0.140
LCL = - 3 = 0.065 - 3
= 0.065 - (3X 0.025) = -0.010
LCL = - 0.010 = 0
Interpretation of the Result:
From the P – control chart it is observed that the points corresponding
to the sample numbers 4 and 16 lie outside the upper control limits.
Therefore, the process is out of control. It means that some assignable
causes are present in the process.
Example 2:
The following data are found during the inspection of the first 15
samples of size 100 each from a lot of two – wheelers manufactured by
an automobile company:
Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Number
Number 3 4 6 2 12 5 3 6 3 5 4 15 5 2 3
Of
Defects

Draw the control chart for fraction defective (p) and comment on the
state of control. If the process is out of control, calculate the revised
centre line and control limits by assuming assignable causes for any out
of control point.
To calculate the control limits, we first calculate the fraction defective for each sample and then
Sample Number Sample Size (n) Number of Defectives (d) Proportion Defective ( p = )

1 100 3 0.03
2 100 4 0.04
3 100 6 0.06
4 100 2 0.02
5 100 12 0.12
6 100 5 0.05
7 100 3 0.03
8 100 6 0.06
9 100 3 0.03
10 100 5 0.05
11 100 4 0.04
12 100 15 0.15
13 100 5 0.05
14 100 2 0.02
15 100 3 0.03
Total 78 0.78
 = = = 0.052
CL = = 0.052
UCL = + 3 = 0.052 + 3
= 0.052 + 3 X 0.022 = 0.118
UCL = 0.118
LCL = - 3 = 0.052 - 3
= 0.052 - 3 X 0.022 = -0.014
LCL = - 0.014 (= 0)
Interpretation of the Result:
From the P – control chart it is observed that the points corresponding
to sample numbers 5 and 12 lie outside the upper control limits.
Therefore, the process is out of control. It means that some assignable
causes are present in the process.
For calculating the revised limits for the p – chart, we first delete the
out of control points ( 5 and 12) and calculate the new using remaining
samples.
In our case, n = 100, k = 15, d = 2,
= 12 + 15 = 27
= = = = 0.039
After finding the we reconstruct the centre line and control limits of the chart using
equations as follows:
CL = E(p) = = 0.039
UCL = E(p) + 3SE(p) = + 3
= 0.039 + 3
= 0.039 + 3 X 0.019
UCL = 0.096
LCL = E(p) - 3SE(p) = - 3
= 0.039 - 3
= 0.039 - 3 X 0.019 = negative value
LCL = 0
Control Chart for Number of Defectives
(np – Chart / d – chart)
• For p – chart, the fraction defective (proportion) for each sample is
calculated and plotted against the sample number.
• Alternatively, instead of calculating the proportion defective, we can
count the number of defective items in the samples and plot them
against the sample number. The chart thus obtained is known as the
np – chart.
• The np – chart is an adaptation of the basic p – chart.
• The purpose of this chart is to provide a better understanding to the
operator who may understand actual number of defective items easily
rather than the proportion defective.
• However there is one drawback of the np chart. It is used only for
equal sample size because the number of defectives cannot be
compared for variable sample size.
• The proportion defective p was obtained by dividing the actual
number of defectives (d) by the sample size (n), i.e. p = , so d = np.
• Therefore, the actual number of defectives may be represented by np.
• Suppose the items or units of a process (population) are classified into
two mutually exclusive groups as defective (Success) and non – defective
(Failure) and a random sample of size n is drawn from it.
• Then the number of defectives follows a Binomial distribution with mean
np and Variance npq where p is the fraction or proportion of defectives
in the process and q = 1-P

E(d) = np and
Var(d) = npq = np(1-p)
S.D(d) =
Control Limits for np chart / d –
chart when P is known:
The centre line and control limits for the np – chart are given as follows:

Centre Line (CL ) = E(d) = np

Upper Control Limit (UCL) = E(d) + 3SD(d) = np + 3
Lower Control Limit (LCL) = E(d) - 3SD(d) = np - 3
Control Limits for np chart / d – chart when p is unknown:

In practice, the fraction defective (p) of the process is not known. Therefore,
it is necessary to estimate it by the average sample fraction defective.
If we draw k samples of size n and are the number of defective items in 1st,
2nd, …, sample, respectively, we can calculate the average fraction defective of
the samples as follows:
= =
Hence in this case, the centre line and control limits of the np – chart are
obtained by replacing p by in the equations as follows:
Centre Line (CL ) = E(d) = n
Upper Control Limit (UCL) = E(d) + 3SD(d) = n + 3
Lower Control Limit (LCL) = E(d) - 3SD(d) = n - 3
Example: 1 To monitor the manufacturing process of laptops, a quality control engineer randomly selects 50
laptops from the production line each day over a period of 20 days. The laptops are inspected for certain
defects and the number of defective laptops found each day is recorded in the following table:
Day Number of Number of Day Number of Number of Defective
Laptops Defective Laptops Laptops Laptops
Inspected Inspected
1 50 4 11 50 6
2 50 8 12 50 1
3 50 6 13 50 5
4 50 10 14 50 3
5 50 4 15 50 2
6 50 3 16 50 3
7 50 4 17 50 7
8 50 7 18 50 9
9 50 8 19 50 2
10 50 4 20 50 4

Construct the appropriate control chart and state whether the process is in control.
The number of laptops inspected in each sample (n) = 50
The number of samples (k) = 20
The total number of defectives is = 4+8+ …+4 = 100
= =
= X 100 = 0.10

Centre Line = n = 50 x 0.1 = 5

Upper Control Limit (UCL) = n + 3 = 5+6.3640 = 11.3640
Lower Control Limit (LCL) = n - 3 = 5 – 6.3640 = 0
Result: The process is under control.
Control Charts for Number of Defects (c – Chart)

• The c - control chart is used for monitoring and controlling the number of
defects in an item / unit.
• Traditionally, the number of defects was denoted by c. Therefore, the
number of defects is known, we can use the c – chart. It is used when the
item / unit is largely assembled. Eg. TV, Computer, mobile, laptop, aircraft,
etc.,
• In such cases, there are many opportunities for a defect to occur and the
probability of the occurrence of a defect is very small.
• For example, in the case of a commercial airplane there is a large number
rivets, but a small chance of any rivet having defects.
• For obtaining the centre line and control limits of the c- chart, we
require the sampling distribution of the number of defects.
• In this case, when there are many opportunities for a defect to occur
and the probability of the occurrence of a defect is very small, the
Poisson distribution is applied.
• So the number of defects follows the Poisson distribution. If is the
average number of defects in the process, the mean and variance of
the Poisson distribution are given as follows:
 E(c) = and Var(c)
We know the standard error of a random variable x is,
S.D(c) =

The centre line and control limits for the c – chart can be obtained as
follows:
Centre Line (CL) = E(c) =
Upper Control Limit (UCL) = E( c) + 3 S.D(c) = + 3
Lower Control Limit (LCL) = E( c) – 3 S.D(c) = - 3
Control Limits when is unknown:
• Suppose we draw k samples of constant size or size 1 and are the number of
defects within 1st, 2nd, …, sample, respectively.
• The average number of defects in the process is estimated by the average
number of defects in the sample which is calculated by the formula given
below:
= =
Hence in this case, the centre line and control limits of the c – chart are
obtained by replacing by in the equations as follows:
Centre Line (CL ) = =
Upper Control Limit (UCL) = + 3
Lower Control Limit (LCL) = - 3
Revised Control Limits:
• If all sample points are lie on or between the upper and lower control
limits, the control chart indicates that the process is under statistical
control. Only chance causes are present in the process and no assignable
cause is present.
• However, if one or more sample points lie outside the upper or lower
control limit, the control chart alarms that the process is not under
statistical control and some assignable causes are present in the process.
• To bring the process under statistical control, it is necessary to investigate
the assignable causes and take corrective action to eliminate them.
• Once the assignable causes are eliminated, we delete the out of control
points (samples) and calculate the revised centre line and control limits
for the c – chart by using the remaining samples. These limits are known
as revised control limits.

• For the revised limits of the c – chart, we first calculate new as follows:
=
Where d – number of discarded samples
- the sum of the number of defects within the discarded samples.
After finding the we reconstruct the centre line and the control limits
for the c – chart by replacing by in the equations as follows:

Centre Line (CL) =

Upper Control Limit (UCL) = + 3
Lower Control Limit (LCL) = - 3
Example: 4 The number of scratch marks on a particular piece of furniture is
recorded. The data for 20 samples are given below.
Sample 1 2 3 4 5 6 7 8 9 10
Number
Scratch 6 3 14 7 2 5 12 4 7 3
Mark
Sample 11 12 13 14 15 16 17 18 19 20
Number
Scratch 2 7 6 8 4 10 5 4 13 9
Mark

Draw the appropriate control chart and write comments about the state of the
process when (i) The management set a goal of 5 scratch marks on an average
per piece. (ii) The management does not set the average number of marks per
piece
The management set a goal of 5 scratch marks on an average per piece.
It means that the average number of defects in the process is known.

The average number of defects per sample (𝛌) = 5

The number of items in the sample (n) = 1
Therefore,
CL = 𝛌 = 5
UCL = 𝛌 + 3= 5 + 3 x = 11.708
LCL = 𝛌 - 3= 5 – 3 x = 1.708
Interpretation of the Result:

From the control chart it is observed that the points corresponding to the
sample numbers 3, 7 and 19 lie outside the upper control limit. Therefore the
process is not under statistical control with respect to the average number of
defects per piece which is 5.
(ii) Here k =20
Total number of defects = 6+3+….+ 13+9 = 131
= = X 131 = 6.55
CL = = 6.55
UCL = + 3 = 6.550 + 3 X 2.559 = 14.227
LCL = - 3 = 6.550 – 3 X 2.559 = -1.1270 (Negative)
Interpretation:
• From the c – chart, we observe that no point lies outside the
control limits and there is no specific pattern of the sample
points on the chart.
• So it indicates that the process is under control with respect to
the average number of defects per piece is 6.55.
Example: 5 As part of an overall quality improvement programme, a textile
manufacturer decides to monitor the number of defects found in each inspected bolt
(large bundle) of cloth. The data from 20 inspections are recorded in the table given
below:
Bolt of 1 2 3 4 5 6 7 8 9 10
Cloth
Number of 10 19 5 9 2 8 7 13 3 2
Defects
Bolt of 11 12 13 14 15 16 17 18 19 20
Cloth
Number of 22 4 6 9 7 2 5 12 4 2
Defects

(i) Which control chart should be used in this case? Calculate the control limits for this
chart.
(ii) Do these data from a controlled process? If not, calculate the revised control limits.
Here the manufacturer monitors the number of defects found in each
inspected bolt of cloth and only one bolt is inspected in each sample. So we
can use the c – chart.
The average number of defects in the bolt of cloth is not given in this case.
k = 20, and n=1
The Total number of defects in bolt = 10 +19+ ….+ 2 = 151
= =
CL = = 7.55
UCL = + 3= 7.55 + 3 X = 15.794
LCL = - 3= 7.55 – 3 X = - 0.6940 = 0
Interpretation:
• It is observed that the points corresponding to the sample numbers
(bolt) 2 and 11 lie outside the upper control limit. Therefore, the
process is not under statistical control with respect to the number of
defects.
• To bring the process under statistical control , it is necessary to
investigate the assignable causes and take corrective action to
eliminate them.
• For calculating the revised control limits for the c – chart, we first
delete the out of control points ( 2 and 11) and then calculate the
new by using remaining samples.
K = 20, d = 2, = 19 + 22 = 41

= = = 6.111

CL = = 6.111
UCL = + 3 = 6.111 + 3 X = 13.527
LCL = - 3 = 6.111 - 3 X = -1.3050 = 0
Control Charts for Number of Defects per
Unit ( u – Chart)
• The c – chart is used for monitoring and controlling the number of
defects. But it is applied only when the sample (subgroup) size is one
or constant. i.e we have inspected the same number of units or items
in each sample.
• To overcome this drawback we use the u – chart. When the sample
size varies due to some reason, such as machinery, raw material,
workers etc., the u – chart can also be used when the sample size is
constant.
• The primary difference between the c – chart and the u – chart is that
instead of plotting the number of defects per sample, we plot the
number of defects per item / unit and monitor them.
Generally two approaches are used for constructing variable control limits for the u –
chart.
First Approach:
In this approach, we calculate the control limits for each sample. Suppose represent the
number of defects in 1st, 2nd, …., Samples of size respectively, we first calculate the
number of defects per item for each sample as follows:
, , ………,
The average number of defects per item is given by
or
CL =
UCL = + 3
LCL = - 3
Second Approach:
In this approach, we calculate the control limits by using the
average sample size. This approach is used only when there is no large
variation in the sample sizes and it is expected that the future sample
sizes will not differ significantly from the average sample size. Using this
approach, we get constant control limits just as we get in the case of
constant sample size. We calculate the average sample size as follows:
=
CL =
UCL = + 3
LCL = - 3
Example: 6 Twenty samples of carpets are inspected for varying sample size and the number
of defects in each sample is noted in the following Table:
Sample Number of Number of Sample Number of Number of
Number carpets defects Number carpets defects
inspected inspected

1 25 12 11 20 9
2 20 5 12 20 12
3 25 7 13 15 14
4 15 7 14 25 6
5 25 10 15 20 7
6 15 4 16 25 12
7 20 6 17 15 5
8 15 2 18 25 6
9 15 4 19 15 8
10 25 10 20 25 4

Construct the suitable control chart for the number of defects per carpet.
First approach:
Sample No of No of u = c/n CL = = UCL = + 3 LCL = - 3
Number Carpets defects
Inspected (n) (c)
1 25 12 0.480 0.370 0.736 0.004
2 20 5 0.250 0.370 0.778 -0.0380
3 25 7 0.280 0.370 0.736 0.004
4 15 7 0.467 0.370 0.841 -0.101
5 25 10 0.400 0.370 0.736 0.004
6 15 4 0.267 0.370 0.841 -0.101
7 20 6 0.300 0.370 0.778 -0.038
8 15 2 0.133 0.370 0.841 -0.101
9 15 4 0.267 0.370 0.841 -0.101
10 25 10 0.400 0.370 0.736 0.004
First approach:
Sample No of No of u = c/n CL = = UCL = + 3 LCL = + 3
Number Carpets defects
Inspected (c)
(n)
11 20 12 0.450 0.370 0.778 -0.0380
12 20 14 0.600 0.370 0.778 -0.0380
13 15 6 0.933 0.370 0.841 -0.101
14 25 7 0.240 0.370 0.736 0.004
15 20 12 0.350 0.370 0.778 -0.0380
16 25 5 0.480 0.370 0.736 0.004
17 15 6 0.333 0.370 0.841 -0.101
18 25 8 0.240 0.370 0.736 0.004
19 15 4 0.533 0.370 0.841 -0.101
20 25 10 0.160 0.370 0.736 0.004
Total 405 150
 = = = 0.370
CL = = 0.370
UCL = +3 = 0.370 + 3
= 0.370 + 3X 0.122 = 0.736
UCL = 0.736
LCL = +3 = 0.370 - 3
= 0.370 - 3X 0.122 = 0.004
UCL = 0.004
Interpretation:
• From the u – chart we observe that the point corresponding to sample 13
lies outside the upper control limit.
• Therefore, the process is not under statistical control with respect to the
number of defects per carpet.
Second Approach:
According to this approach, the control limits are calculated using the
average sample size and we get constant control limits. We first
calculate the average sample size.
= = 20.25
CL = = 0.370
UCL = = 0.370 + 3 X
= 0.370 + 3 X 0.135
UCL = 0.775
LCL = = 0.370 - 3 X
= 0.370 - 3 X 0.135
LCL = -0.0350
Interpretation:
• Form the u – chart we observe that the point
corresponding to sample 13 lies outside the upper control
limit.
• Therefore, the process is not under statistical control with
respect to the number of defects per carpet.
Comparison between control charts for variables
and attributes
S. No Control Charts for variables Control Charts for Attributes
1 These charts are used for measurable These charts are used for non – measurable
characteristics. characteristics and the situations where the
control charts for variables are not applied due
to time and cost factors.
2 The types of charts are the chart and R The types of charts are the p – chart, np –
chart chart, c – chart and u - chart
3 A separate control chart is needed for A single chart is enough for a number of
each quality characteristics. quality characteristics
4 Small samples serve the purpose very well Large samples are required for correct
conclusions
5 The cost of inspection of units is large The cost of inspection of units is small since no
measurements are made.
6 These charts provide the better quality These charts are not very effective in quality
control control.
7 These charts are not easily understood These charts are easily understood.
Examples for np chart:
Find the UCL and LCL for np chart for the following information and
comment on the result:
Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
No of
100
No of 19 9 8 7 11 10 2 6 8 11 12 10 8 7 6
defecti
ves

= = 0.089
Central Line = n = 100 X 0.089 = 8.9
UCL = n + 3
= 8.9 + 3
= 8.9 + 3
UCL = 17.45
LCL = n - 3
= 8.9 - 3
= 8.9 - 3
LCL = 0.35
Example : A nationwide courier service desires to check the accuracy of
its clerical work in completing invoices. A sample of 200 invoices is
taken each day for 30 consecutive days for inspection. No. of defectives
found in each day is as follows:
2 6 8 5 2 3 5 2 10 2 10 9 7 4 1

4 6 8 6 1 8 3 6 3 2 4 9 3 7 4

Determine the upper and lower control limits using np chart and also
construct the control chart.
 = = 0.025
Central Line = n = 200 X 0.025 = 5
UCL = n + 3
=5+3
=5+3
UCL = 11.63
LCL = n - 3
=5-3
=5-3
LCL = -1.63 or 0
• A process is said to be stable if, over time, the process has both
constant means and constant variances as well as a constant
distribution.
Defect vs Defective:
• Defect – a single non conforming quality characteristic
• A defect is any item or service that exhibits a departure from specifications.
• A defect does not necessarily mean that the Producer or service cannot be used.
• A defect indicates only that the product result is not entirely as intended.
• A defect is an imperfection – a flaw or departure from the ideal.
• A defect can be large or small, rendering a product unusable or only outside of
specifications.
Example: Suppose service in a restaurant is being evaluated. If a waiter greets his
table after 5 minutes, the customer can still order and enjoy a meal even though the
promptness of the greeting did not meet expectations. Therefore, this could be
considered a defect (“ Late greeting” ) in the service.
A customer order a burger with tomato, but the burger delivered to them did not
contain tomato.
Products with defects can still be used for their original purpose, although the
customer experience may be sub – optimal.
Defective – items having one or more defects
• A defective is an item or service that is considered completely unacceptable for
use.
• Each item or service experience is either considered defective or not – there are
only two choices.
• A defective item, on the other hand doesn’t function as required. For this to
happen, the item will have one or more defects.
Example: Before final shipment, a quality inspector evaluates auto supply parts and
rates each item as “ pass” or “fail” to ensure that the company does not ship any
parts that will be unusable.
A burger patty is undercooked and represents a health risk, the customer cannot
safely eat it.
Defective products cannot be used for their original purpose without danger or
severe inconvenience to the customer.
Thank You