Control Charts, Theory of control charts, measurement range, construction and analysis of R

charts, process capability study, use of control charts. Attributes of Control Chart, Defects,
construction and analysis of charts, improvement by control chart, variable sample size,
construction and analysis of C charts
CONTROL CHARTS
Definition. A control chart is defined as a statistical tool used to detect the presence of
assignable causes in any manufacturing systems and it will be influenced by the pure
system of chance causes only. Control charts are of two types : Variable control charts and
Attribute control charts.
Variable Control Charts
A variable control chart is one by which it is possible to measure the quality
charactertistics of a product. The variable control charts are
(i) X — chart
(ii) R — chart
(iii) σ — chart
Attribute Control Chart
An attribute control chart is one in which it is not possible to measure the quality
characteristics of a product i.e., it is based on visual inspection only like good or bad,
success or failure, accepted or rejected. The attribute control charts are !
(i) p — chart
(ii) np — chart
(iii) c — chart
(iv) u — chart
Objectives of Control Charts
1. Control charts are used as one source of information to help whether an item or items
should be released to the customer.
2. Control charts are used to decide when a normal pattern of variation occurs, the process
should be left alone when an unstable pattern of variable occurs which indicates the
presence of assignable causes it requires an action to eliminate it.
3. Control charts can be used to establish the product specification.
4. To provide a method of instructing to the operating and supervisory personnel
(employees) in the technic of quality control.
Symbols or Notations
X : Mean of the sample
X σ : Standard deviation of the sample
X 1 : Mean of the population or universe
σ1 : Standard deviation of the population.
Central Limit Theorem
Irrespective of the shape of the distribution of the universe, the average value X of a sample size ‘n’
( X 1, X 2. X 3, ....... X n) drawn from the population will tend towards a normal distribution as n
tends to infinity Relationship between R and σ1
Interpretation of Control Charts
After plotting the points on the X – R charts, it shows two possible states of control. They are 1.
State of statistical control and 2. State of lack of control.
State of Satistical Control
A manufacturing process is said to be in a state of statistical control whenever it is operated upon
by a pure system of chance causes. The display of points in the X chart and R chart will be
distributed evenly and randomly around the central line and all the points should fall between the
UCL and LCL which is shown below.
State of Lack of Control
A process is said to be in a state of lack of control when the state of statistical control does not hold
good. In such a state we interpret the presence of assignable causes. The reasons for lack of
control are
(i) points violating the control limits
(ii) run
(iii) Trend
(iv) Clustering
(v) Cycle pattern.
Violation of Control Limits
A chart is said to be a violation of control limits, whenever the points cross the UCL, or LCL or
both, reasons for this are
(a) wrong setting
(b) Physical or environmental conditions
(c) Damage or mishandling of parts

Run (upward or downwards)

A run is said to occur whenever seven or more consecutive points occur on one side of the central
line, run can be either above the central line or below the central line, which is shown below.
Reasons for run are
(a) Different kinds of raw material used
(b) New worker
(c) Carelessness of operator
(d) Minor failure of machine parts

Trend
This refers to the gradual movement of points from one side of the central line to the other side of
the central line. A trend is said to occur whenever a correlation exists between the points with
reference to the sub-group number. Trends can be upward or downward.
The reasons for this are
(a) Tool wear
(b) Worker fatigue
(c) Effect of temperature and humidity
(d) Accumulation of waste products.

Clustering
This occurs whenever the points do not spread over the entire region between UCL and LCL.
Instead the points exists very close to the central or on the central line, itself. Clustering indicates
systematic differences within the subgroup.

Cycle Pattern
This pattern shows systematic differences between the sub-groups. Such things can occur
whenver the assignable causes are present.

After plotting the X – R charts, the important points to be noted are :

1. Accept the process as confirming to the standards if the plotted points are randomly and evenly
distributed around the central line and lie within the control limits of both the charts.
2. If any plotted points fall on or above the UCL on the range chart, it indicates an increase in the
process dispersion which must be corrected by appropriate technical action.
3. If any plotted points fall on or above the UCL (a) LCL in the X chart, it indicates a shift in the
process average. This should be corrected by taking suitable technical action.
PROCESS CAPABILITY
The first priority is to bring a process into a state of statistical control. That is to say, make the
process stable and therefore predictable.
But this is not enough ; there is more to be done. Once we have concluded that a process is in
statistical control (by the absence of special causes on the control charts) there could still be two
possible conditions prevailing, i.e.,
(a) the process is stable (i.e., in control) and capable. In other words it continually produces parts
which are acceptable.
(b) the process is stable and incapable, i.e., it continually and predictably produces parts which are
not acceptable.
Obviously, it is the first alternative we want. So there must be some examination to test for the
inherest capability of the process.
This is done by comparing the output which the process can be expected to produce under normal
and random conditions with that required by the customer.
The natural spread of a process is the distribution of parts produced when the process is operating
in statistical control and being affected only by random variations.
So if we ran a process and measured every individual part produced the distribution would more
than often follow the normal bell-shaped curve with a standard deviation of σ.
The natural spread across the basis is 6σ

So this tells us what the process is actually giving us. We still don’t know if this is good enough for
what we want.
To determine this, we must compare the process spread with the design requirements, i.e., the
tolerance. The tolerance is, of course, the difference between the upper specification limit (USL)
and the lower specification limit (LSL).
The following diagrams illustrate the output distribution from five different processes. Processes
1 and 2 are producing components beyond the specification limits (either too low or too high).
This is because the process is wrongly set. Adjustment of the setting will bring the distribution
within the specification limits.
On the other hand, process 3 is producing unacceptable parts which are both too high and too low.
This is because the process spread is too wide. No end of adjustment to the setting will cure this
problem. The piece-to-piece variability must be reduced to make the process capable of producing
acceptable parts.
Process 4 is just capable of producing acceptable parts.
process 5 is the best.
This is the objective : to reduce the spread about a centre line which is on target setting.
For a variable process where we draw average/range charts the measure of capability can be
indicated by capability indices.
There are two indices and we shall deal with each in turn.
The Cp index
The Cp index gives a measure of capability for a process which is centred on the required design
value, i.e., it is centred on nominal.
The Cp index is found by comparing the tolerance (T) with the natural spread (6σ).

Attribute Control Charts

When a chart shows the number of articles conforming and the number of articles failing to
conform to any specified requirements then it is said to be an attribute control chart.
The attribute control charts are classified as
(a) p-chart or Fraction defective chart
(b) np-chart or Number defective chart
(c) c-chart or Defects per unit chart
(d) U-chart
p-chart
If the attribute data is collected for the number of defective articles produced in a given period and
a control chart is used to present the attribute data is known as p-chart. The fraction defective of
each sample is denoted by p

If p represents the fraction defective of the process then the probability of occurrence of ‘r’
defectives in a sample is given by the Binomial Distribution.
Steps to be followed to construct p-chart.
1. Collect about 25 samples of size n. If ‘n’ is constant then it is called a constant pchart. If ‘n’ is
varying then it is called variable p-chart.
2. Calculate the fraction defective of each sample

3. Compute the standard fraction defective

4. Compute the control limits

If n is constant then only one UCL and LCL is computed. If ‘n’ is varying then compute UCL and LCL
for each sample.
5. Compare the fraction defective p with its control limits. If all the samples are falling within the
limits then homogenisation is not necessary.
6. Otherwise delete the corresponding subgroups which are crossing the control limits and
calculate p 1.
7. Again compute the control limits for p 1.
8. Prepare a chart with the fraction defective on the vertical axis and subgroup No. on the
horizontal axis.
9. Draw the UCL and LCL on the chart and locate the points of p with respect to subgroup no.
Purpose of setting the lower control limits on the p chart or attribute control chart. When the
points fall below the LCL, one of the six factors are usually responsible for extra good quality.
(a) Unexpected improvement in the process.
(b) Faulty inspection.
(c) Error in charting or plotting.
(d) Chance variation.
(e) Use of very good raw material.
(f) Good training to the operator.