www.jntuworld.

com

STATISTICAL QUALITY CONTROL

Introduction: Quality is the determining factor the success of any product or service
large resource are committed in every organization to ensure quality
Definition: It is defined as customer satisfaction in general and fitness for use in
particular. Both the external consumer who buy the product and services and the
internal consumers that is, all divisions or departments of the business organization
are equally interested in the quality.
Statistical quality control: The process of applying statistical principles to solve the
problem of controlling the quality control of a product or service is called statistical
quality control.
Quality elements: a) Quality design b) Quality conformance
a) Quality design: Quality of design refers to product feature such as performance,
reliability durability, ease of use, serviceability
b) Quality conformance: Quality conformance means whether the product meets the
given quality specification or not
Inspection: The process of measuring the out put and comparing it to check whether
it meets the given specified requirements or not, is called inspection.
Inspection Methods: The following are the methods of inspection based on merits
1) Incoming inspection: In this method, the quality of the goods and services arriving
into the organization is inspected. This ensures that the material suppliers adhere to
the given specifications with this defective material cannot enter into the production
process. This focuses on the vendor’s quality and ability to supply acceptable raw
materials.
2) Critical point inspection: Inspecting at the critical points of a product manufacture
gives valuable insight into the completely functional process. At the points of
manufacture that involve high costs or which offer no possibility for repair or rework,
inspection is crucial further operation depend on these results critical point inspection
helps to drop the defective production, and thereby, facilitate avoiding unnecessary
further expenditure on them.
3) Process inspection: This is also called patrolling inspection or floor inspection or
roving inspection. Here the inspector goes around the manufacturing points in the
shop floor to inspect the goods produced on random sample basis from time to time.

www.jntuworld.com
www.jntuworld.com

4) Fixed inspection: It provides for a centralized and independent where work is

brought for inspection from time to time. This method is followed where the inspection
equipment cannot be moved to the points of productions.
5) Final inspection: This is centralized inspection making use of special equipment.
This certifies the quality of the goods before they are shipped.
Elements of statistical Quality Control: The technique under SQC can be divided
in to two parts a) Process control b) Acceptance sampling
a) Process control: Process control is a technique of ensuring the quality of the
products during the manufacturing process itself. If a process consistently produces
items with acceptable or tolerable range of specification. It is said to be statically
under control. Process control is achieved through control charts. Process control
aims to control and maintain the quality of the products in the manufacturing process.
Statistical control charts: A control chart compares graphically the process
performance data to computed statistical control limits. These control limits act as
limit lines on the chart control chats are the tools to determine whether the process is
under control or not.
The quality of the production process may be affected by chance cause or
assignable cause.
Chance cause: such causes, which may or may not affect the manufacturing process
are called chance cause, chance cause cannot even be identified. It is not possible to
always maintain the given specification.
Assignable Cause: Assignable causes affect the quality of the production process.
These causes can be identified and specified. Causes such as change in the labour
shift, power fluctuations, or excessive tool wear are said to be assignable causes as
they affect the quality of manufacturing process in different ways.
Process capability: Process capability refers to the ability to achieve measurable
results from a combination of machines, tools, methods, materials and people
engaged in production.

www.jntuworld.com
www.jntuworld.com

Confidence limits and control limit:

Confidence limit: It indicate the range of confidence level. A confidence level refers to
the probability that the value of measurement or parameter, such as length of screw,
is correct.
Ex: If a component is required with measurement of 50 mm. across, then the buy
accept all components measuring between 48 mm and 52 mm across, considering a
five percent confidence level.
Control limit: Control limits are found in the control charts. There are two control
limits 1) Upper control limit (UCL) and 2) Lower control limit (LCL). These are
determined based on the principles of normal distribution
Ex: In a pilot investigation of the length of the nails produced in the shop floor, it is

found that the mean length X is cm, the S.D 3σ, the measure of variability of the
nails produced 0.2 cm. How do you construct the control chart for this data.

+3σ Upper control limit

4.6

Length of nail 4 x

3.4
-3σ Lower control limit

Sample number

Control charts for variables: A variable is one whose quality measurement changes
from unit to unit. The quality of these variables is measured in terms of hardness,
thickness, length, and so on. The control charts for variables are drawn using the

principles of normal distribution. There are two types of control charts for variables x
and R chart.

X and R Chart: The X chart is used to show the process variations based on the
average measurement of samples collected. It shows more light on diagnosing quality
problem when read along with R chart. It shows the erratic or cyclic shifts in the
manufacturing process. It can also focus on when to take a remedial measure to set

www.jntuworld.com
www.jntuworld.com

right the quality problems. However, collecting data about all the variables involves a
large amount of time and resources.
The R chart is based on the range of the items in the given ample. It highlights
the changes in the process variability. It is a good measure of spread or range. It

shows better results when read along with the X chart.

For x charts: UCL = x + A2 R When x = Mean of Means

LCL = x - A2 R R = Mean of sample range

A2= Constant

For R chart: UCL = D4 R D4, D3 are constants

LCL = D3 R

R is the average of sample ranges (Ranges is the difference between the

maximum variable and minimum variable)

EX: Construct x and R charts from the following information and state whether the

process is in control for each of the following x has been computed from a sample of
5 units drawn at an interval of half an hour from an ongoing manufacturing process.
Samples 1 2 3 4 5 6 7 8 9 10

x 24 34 35 39 26 29 13 34 37 29

R 23 39 14 5 20 17 21 11 40 10

Solution: The mean of means x =

∑x =
300
= 30
n 10

R is calculated as R =
∑R = 200
= 20
n 10
x Chart: x hart
c UCL and LCL compute at sample size 5 A 2 table value is 0.58

UCL= x +A2 R = 30 + (0.58x20) =41.6

LCL = D3 R = 30 – (0.58x20) = 18.4

www.jntuworld.com
www.jntuworld.com

60

50
UCL = 41.6
40
Measurement
30 x = 30
of variables
20
LCL = 18.4
10

0
1 2 3 4 5 6 7 8 9 10
Number of samples

www.jntuworld.com
www.jntuworld.com

R Chart: R chart UCL and LCL compute at sample size 5, D4 table value is 2.11 and
D3 table value is 0

UCL = D4 R = 2.11 x 20 = 42.2

LCL = D3 R = 0 x 20 = 0

45
UCL = 42.6
25

20 R = 42.6
Measurement
of variables 15

10

0 LCL = 0
1 2 3 4 5 6 7 8 9 10
Number of samples

Therefore 3, 7 points the process is out of control.

Control charts for attributes: The quality of attributes can be determined on
the basis of ‘Yes’ or ‘No’, ‘Go’ or ‘No go’. In other words, in case of a mirror
glass, even if there is one scratch it is not considered to be a quality mirror, in
such a case quality is decided base on whether the mirror has any scratch or not.
The control charts for attributes are ‘C’ chart and ‘P’ charts
‘C’ Chart: ‘C’ chart is use where there a number defects per unit. This control
charts controls the number of defects per unit. Here the sample size should be
constant. This calculate as below.

UCL = c + 3 c and LCL = c - 3 c

Where the c = Total number of defects in all the samples

Total number of samples inspected

www.jntuworld.com
www.jntuworld.com

Ex:
Sample Number No. of defects Sample Number No. of defects
1 5 11 4
2 4 12 6
3 9 13 7
4 7 14 3
5 8 15 5
6 9 16 3
7 4 17 3
8 5 18 1
9 2 19 7
10 6 20 2
Total number of defects = 100

c = 100 = 5
20

UCL = c + 3 c = 5 + 3 5 = 11.69

LCL = c -3 c = 5-3 5 = 0
LCL = 0 means, LCL got negative value, take it as equal to zero

20

15
UCL = 11.69
10
No. defects
per each 8
sampled unit
6
5 c= 5
4
2
0 LCL = 0
2 4 6 8 10 12 14 15 18 20
Number of samples

www.jntuworld.com
www.jntuworld.com

‘P’ Chart: ‘P’ Chart is used where there is date about the number of defectives
per sample. It is also called fraction defective chart or percentage defectives
chart. Here each item is classified on ‘go or no go’ basis that is good or bad.
Hence if the sample size is larger, the results could be better.
UCL =
LCL =
Total no. of defective found
Where average defective ( p ) =
Total no. of pieces inspected
‘n’ = Number of pieces inspected per day
Ex: For each of the 14 days a number of magnets used in electric relays are
inspected and the number of defectives is recorded. The total number of
magnets tested is 14,000. The following are the particular of the number of
defectives found every day.
Number of Number of
Day number Day number
defective defective
1 100 8 120
2 50 9 60
3 150 10 140
4 200 11 50
5 150 12 70
6 50 13 40
7 80 14 40

Solution:
Total number of defectives = 14000
The average sample size(n) per day= 14000/14 days = 1000
Total no. of defective found per day
Percentage of defective per day =
Total no. of pieces inspected per day

Day Percentage of Number of Percentage of Day Number of

number defectives defective defectives number defective
1 100/1000=0.10 100 120/1000=0.12 8 120
2 50/1000 =0.05 50 60/1000 =0.06 9 60
3 150/1000=0.15 150 140/1000=0.14 10 140
4 200/1000=0.20 200 50/1000 =0.05 11 50
5 150/1000=0.15 150 70/1000 =0.07 12 70
6 50/1000 =0.05 50 40/1000 =0.04 13 40
7 80/1000 =0.08 80 140/1000=0.14 14 40

www.jntuworld.com
www.jntuworld.com

Total no. of defective found

(p)=
Total no. of pieces inspected
1400/14000 = 0.1
0.1(1 − 0.1)
UCL = 0.1 + 3 = 0.4
1000

0.1(1 − 0.1)
LCL = 0.1 - 3 =0
1000

0.45
0.40 UCL = 0.4

0.30
Percentage
of defective 0.20
0.15
0.10 p = 0.1
0.05
LCL = 0

1 5 10 15

Day Number

Acceptance Sampling: Acceptance sampling is a technique of deciding

whether to accept the whole lot or not based on the number of defectives from a
random drawn sample.
It is widely use in buying food products, such as rice, wheat etc. Before
buying the random samples drawn from the bags of say rice are tested. If the
quality of sample drawn looks good or free from defects then according to the
requirement the entire bag or part of it can be brought
The process of acceptance sampling through operating characteristic
curve (OCC)

www.jntuworld.com
www.jntuworld.com

Operating characteristic curve (OCC): The graphical relationship between

percentage defective in the lots being submitted for inspection and the
probability acceptance is termed as “operating characteristic of a particular
sampling plan”

C = 1 (Acceptance number)
100
80 n = 50 (sample size)
Probability of
acceptance 60
percentage
40
20

1 2 3 4 5 6 7 8 9 10
Actual percentage of defectives

It gives a clear picture about the probability of acceptance of lot for

various values of percent defectives in the lot. The probability of acceptance
of a lot is high for low values of actual percentage decrease and it is low for
high values of actual percentage defectives.
Construction of OC curve: To develop a sampling plan for acceptance
sampling, an appropriate O.C curve must be selected to construct an OC
curve an agreement has to be reached between the producer and the
consumer on the following four point.
1) Acceptable quality level (AQL): This is the maximum proportion of
defectives that will make the lot definitely acceptable.
2) Lot tolerance percentage defective (LTPD): This is the maximum
proportion of defectives that will make the lot definitely unacceptable.
3) Producers risk (α): This is the risk, the producer is willing to take that lots
of the quality level AQL will be rejected, even though, they are acceptable
usually α = 5%

www.jntuworld.com
www.jntuworld.com

4) Consumer risk (β): This is the risk, the consumer is willing to take that lots
of the quality level LTPD will be accepted, event though, they are actually
unacceptable usually β = 10%.

Producers
risk α = 10%
α AQL = 20%
Probability of LTPD = 60%
acceptance

Consumers β
risk β = 10% 20% 60% 100%

Percentage of defectives (p)

Sampling plans: Based on the number of samples drawn for taking accept/
reject decisions, the sampling methods are used. There are four methods of
acceptance samplings.
1) Single sampling plan: A lot is accepted or rejected on the basis of a single
sample drawn from that cost
2) Double sampling plan: If it is not possible to decide the fate of the lot on
the basis of first sample, a second sample is drawn and the decision is taken
on the basis of the combined results of first and second sample.
3) Multiple sampling plan: A lot is accepted or rejected based upon the result
obtained from several samples (of parts) drawn from the lot.
4) Sequential sampling plan: (Item by item analysis)
Sequential sampling involves increasing the sample size by one part at
a time till the sample becomes large enough and contains sufficient number
of defectives to decide intelligently whether to accept or reject the lot.

www.jntuworld.com