STA4C04 – STATISTICAL INFERENCE AND QUALITY CONTROL

Module 4

STATISTICAL QUALITY CONTROL

Quality is the attribute of a product that determines its

fitness for use.
Quality means a standard a consumer expects from a
product or service. It is a major decision factor for
selecting a product or service.
Quality control is a powerful productivity technique for
effective diagnosis of lack of quality in any of the
materials , processes, machines and end products.
It is essential that the end product possess the qualities
that consumers expect from them. Every company should
take measures to ensure that their product meets the
desired quality conformation. Quality control ensures this
by insisting on quality specifications all along the line
from the arrival of materials through each of their
processing to the final delivery of goods.
Quality is inversely proportional to the variations.
Chance causes and assignable causes
The variations in the quality of manufactured product can
be classified as being due to two causes namely
1) Chance causes
2) Assignable causes
Chance causes are the causes of variations in the quality
which are beyond human control and cannot be prevented
or eliminated under any circumstances.
Such variations are usually termed as allowable
variations. The range of such variations is known as
natural tolerance limits of the process.
Assignable causes are causes of variation which can be
prevented and controlled.
The variations due to assignable causes are termed as
preventable variations.
Some of the important factors of assignable causes are
substandard or defective raw material , negligence of the
operators , wrong or improper handling of machines ,
faulty equipment etc.

Statistical quality control (SQC)

Statistical quality control means planned collection and
effective use of data for studying causes of variation in
quality either as between processes ,procedures materials
machines etc. or over periods of time.
The main purpose of SQC is to device statistical
techniques which would help us in separating the
assignable causes of variation from the chance causes thus
enabling us to take immediate remedial action whenever
assignable causes are present.
Uses of SQC
1) SQC helps in the detection and correction of many
production troubles and brings about a substantial
improvement in the product quality and reduction of
spoilage and rework.
2) It provide better quality assurance at lower inspection
cost
3) SQC reduce waste of time and material to the
absolute minimum by giving an early warning about
the occurrence of defects.
4) It tells us when to leave the process alone and when
to take action to correct troubles.
5) The very presence of a quality control scheme in a
plant improves and alerts the personnel.
Process control and product control
The statistical quality control can be classified into two -
process control and product control.
Process control is a method of SQC in which the quality
of a product is controlled by the products in the process
of production. In process control, control charts are used
to detect the variations and control the quality of a
product.
Product control is a method of SQC in which the quality
of a product is controlled while the product is ready to
dispatch or sell to the customers. The product control
uses the technique of acceptance sampling to detect the
defects and control the quality of the product.

Control charts
Control charts are the graphic devices for detecting the
variations in the production process and determining the
permissible limits of variation. The technique of control
charts are pioneered by W.A .Shewart.
It consists of three horizontal lines
1) Central line (CL) which indicates the desired
standard level of the process.
 2) Upper control limit (UCL) which indicates the upper
limit of tolerance.
3) Lower control limit (LCL) which indicates the lower
limit of tolerance.

Constructing a control chart

For constructing a control chart we will have to select
multiple samples of data values from the output of a
process and calculate an appropriate statistic t . This
statistic t maybe sample mean ̅x , sample range R , no. of
defectives in the sample d etc. A control chart is drawn
taking sample numbers along x axis and the statistic t
along the y axis . Three horizontal lines are drawn in the
control chart corresponding to the calculated values of the
central line (CL) ,lower control limit (LCL) and upper
control limit (UCL) . Then each value of the statistic t
plotted against the corresponding sample number. the
process is said to be under statistical control if all the
plotted points lies within the two controlled LCL and
UCL. If any one of the plotted points lies outside the
control limits then the process is said to be out of control.

Types of control charts

Two types of control charts
 1) Control charts for variables and
2) Control charts for attributes.

Control charts for variables

Control charts for variables can be applied to any quality
characteristic that is measurable like height ,weight
,length etc.
Two control charts for variables are X̅ chart (or mean
chart) and R chart ( or range chart)
X̅ chart or mean chart
It is employed to monitor the changes in the average of
the quality characteristic.
For constructing a mean chart we draw multiple samples (
say m samples) of size n from the output of the processes
and compute their means 𝑥₁ ̅ , 𝑥₂ ̅̅̅̅ and ranges R₁ ,
̅ ...... 𝑥ₘ
R₂.......Rₘ
̅̅̅+𝑥₂
𝑥₁ ̅̅̅+⋯+𝑥ₘ
̅̅̅̅
The mean of the means , 𝑥̿ = and 𝑅̅ =
𝑚
𝑅₁+𝑅₂+⋯+𝑅ₘ
𝑚

For a X̅ chart or mean chart

Central line , CL = 𝑥̿
Lower control limit , LCL = 𝑥̿ - A₂ ̅𝑅
Upper control limit , UCL = 𝑥̿ + A₂ ̅𝑅
Where A₂ is obtained by referring the SQC tables for
sample size n.
To construct a mean chart we take the sample numbers
along the X axis and sample means along the Y axis.
Horizontal lines corresponding to CL , LCL and UCL are
drawn on the graph . Then sample means are plotted
against the corresponding sample numbers . The process
is said to be under statistical control if all the plotted
points lies between the control limits LCL and UCL. If
any of the plotted points lies outside the control limits
then the process is said to be out of control.

R chart or range chart

It is employed to monitor the changes in the dispersion of
the quality characteristic.
For a R chart
CL = ̅𝑅
LCL = D3 ̅𝑅
UCL = D4 ̅𝑅
D3 and D4 are obtained by referring the SQC tables for
sample size n.
A R chart is constructed by taking sample numbers along
X axis and Sample ranges along Y axis . Horizontal lines
corresponding to CL , LCL and UCL are drawn on the
graph . Then sample ranges are plotted against the
corresponding sample numbers . The process is said to be
under statistical control if all the plotted points lies
between the control limits LCL and UCL. If any of the
plotted points lies outside the control limits then the
process is said to be out of control.

Control chart for attributes

Control charts for variables can be applied to any quality
characteristic that cannot be directly measured but can be
classified as conforming to specifications or not like
defective or non- defective.
Some control charts for attributes are
1) np chart or control chart of no. of defectives
2) p chart or control chart of fraction defectives
3) c chart or control chart of no. of defects

np chart or control chart of no. of defectives (d chart)

For constructing a np chart we draw multiple samples
( say m samples) of size n from the output of the
processes and note the no. of defectives in each sample
say d1,d2,......dm
𝑑₁+𝑑₂…+𝑑ₘ
Now average of the no. of defectives , n𝑝̅ = 𝑚
𝑛𝑝̅
𝑝̅ = 𝑛
For a np chart
CL = n𝑝̅
LCL = n𝑝̅ − 3√𝑛𝑝̅ 𝑞̅
UCL = n𝑝̅ + 3√𝑛𝑝̅ 𝑞̅
Where 𝑞̅ = 1- 𝑝̅
A np chart is constructed by taking sample numbers along
X axis and no. of defectives s along Y axis . Horizontal
lines corresponding to CL , LCL and UCL are drawn on
the graph . Then the no. of defectives are plotted against
the corresponding sample numbers . The process is said to
be under statistical control if all the plotted points lies
between the control limits LCL and UCL. If any of the
plotted points lies outside the control limits then the
process is said to be out of control.

P- chart (Control chart for fraction defectives)

For constructing a p-chart we draw multiple samples
( say m samples) of size n from the output of the
processes and note the no. of defectives in each sample
say d1,d2,......dm
𝑑₁+𝑑₂…+𝑑ₘ
Now average of the no. of defectives , n𝑝̅ = 𝑚
𝑛𝑝̅
𝑝̅ = 𝑛
For a p-chart
CL = 𝑝̅
𝑝̅ 𝑞̅
LCL = 𝑝̅ − 3√ 𝑛

𝑝̅ 𝑞̅
UCL = n𝑝̅ + 3√
𝑛

Where 𝑞̅ = 1- 𝑝̅
A p-chart is constructed by taking sample numbers along
X axis and fraction. of defectives s along Y axis .
Horizontal lines corresponding to CL , LCL and UCL are
drawn on the graph . Then the fraction defectives are
plotted against the corresponding sample numbers . The
process is said to be under statistical control if all the
plotted points lies between the control limits LCL and
UCL. If any of the plotted points lies outside the control
limits then the process is said to be out of control.

c-chart (Control chart for number of defects)

One of the important control chart for attributes is the c

chart. It is designed to control the number of defects per
unit. There are many situations in industry where the data
are obtained by counting and are of discrete type. For
example, the number of idle machines, number of defects
in castings, number of air bubbles in glasses, number of
blemishes in a sheet of paper, number of blemishes in
galvanized surfaces, number of weak spots in a given
length of wire, and number of breakdowns at this weak
spots, number of defects in body alignments of aircrafts,
and buses, number of rust spots in steel sheet, number of
mould marks on fibre glass canoes, are the examples
where c-chart is usually of one unit. It may be of fixed
time, length, area, a single unit or a group consisting of
different units. In the case of textile yarn, a fixed length is
taken as a unit. In the case of glass, a given area is taken
as a unit. In a glass vessel, one vessel is taken as a unit. In
the vase of wool or textile cloth, a given area is taken as
one unit. In the case of surface blemishes, a given area of
the surface is the unit of sample.

The theoretical basis for the c-chart is derived from the

Poisson distribution. In case of defects, in units, the
opportunity for the occurrence is very large (large number
of trials) while the actual occurrence of defects tend to be
small (the probability of success p is very small). This
situation can be very well represented by a Poisson
Distribution.

Procedure for the construction of c-chart

Let c denote the number of defects in one unit of cloth of
paper or any material.
𝑐1 +𝑐2 +⋯𝑐𝑛
Find the mean , c̄ = where 𝑐1 , 𝑐2 , … 𝑐𝑛 are
𝑛
the defects counted in several such units. The expected
standard or central line is . In a Poisson distribution, the
variance is equal to mean ie., σ² = c̄ or σ = √c̄ . Based on
this, 3σ limits for the upper and lower control limits are
obtained.

UCL = c̄+3σ=c+3√ c̄ and

LCL =c̄-3σ= c̄ -3√ c̄

The use of c-chart is valid and made appropriate when the

opportunities for the occurrence of a defect in each
production unit are infinite but the probability of a defect
at any point is very small and is constant. While using c-
chart uniform sample size is taken or unit sizes are taken.