TRAINING MATERIAL

ON

STATISTICAL PROCESS CONTROL

(SPC)

116-A, JWALAHERI MARKET, 2nd FLOOR,

PASCHIM VIHAR, NEW-DELHI-63
Ph: 011- 25275584 Mobile: 9810705999
E-mail : globalservices_delhi@yahoo.com
balakrishnan_babu@hotmail.com

TOOLS FOR PROCESS CONTROL

Detection:
A past oriented strategy that attempts to
identify unacceptable
output after it has been
produced and then separate it from the good output.
Defect Detection:
Is reactionary
Tolerates waste
Relies on inspection, audits, or checks of large samples
of
output
Reacts to all defects indiscriminately
Focuses on conformance to specifications
Involves action only on output
Relies on delayed feedback for defect detection
Is not cost effective

Prevention :A future oriented strategy that improves quality and

productivity by directing analysis and action toward correcting the
process itself so that unacceptable parts will not be produced.
Defect Prevention:
Is pro-active
Avoids waste
Uses small samples of product and process information
Is analytically based
Discriminates between potential defects based on causes
Involves action on the process or process parameters
Provides timely feedback
It is cost effective

TECHNIQUES FOR PROCESS CONTROL

1. Mistake Proofing :- In this technique 100% process
control is achieved by preventing all types of failures by
using modern techniques to get defect free product. Here
causes are prevented from making the effect.
2.

100% Inspection : In this technique 100% checking of all

the parameters of all products has been done to get defect
free product. Here only defects are detected.

3. Statistical Process Control : In this Statistical technique

such as Control Chart, Histogram etc. are used so as to
analyses the process and achieve and maintain state of
statistical control to get defect free product. Causes are
detected and prompting corrective action before defect
occurs.
WHY S.P.C. IS REQUIRED ?
Effectiveness of any activity in
an Organization is measured with respect to time and cost
involved in it.

Mistake Proofing 100% Inspection

In this method
more advanced
and modern
techniques are
used which
require substantial
investment during
its installation and
maintenance.

As it is detection type
of technique it cant
avoid failure but
rejects defective
products.
Requires more
inspectors, more
inspection times and
in turn more cost.

Statistical Process
Control
For this technique
investment is very less
and process is
controlled on each
workstation therefore
defective components is
not forwarded to next
operation. Predictability
reduces frequent
adjustments & in turn
increases productivity,
reduces inspection cost
at station & at final
inspection

From above we can observe that S.P.C. is the economical way of controlling
the process in comparison with Mistake Proofing and 100% inspection.

WHAT IS S.P.C. ?
1. Statistics :- A
sample data (e.g.
make inferences
output from which

value calculated from or based upon

a subgroup average or range) used to
about the process that produced the
the sample comes.

2. Statistical Control :- The condition describing a process

from which all special causes of variation have been
eliminated and only common causes remain.
3. Statistical Process Control :- The use of Statistical
techniques such as control charts to analyze a process or
its outputs so as to take appropriate actions to achieve and
maintain a state of statistical control and to improve the
process capability.

Population and Sample

Population:
A part of the universe, which is under study for gathering
information, is called population. The concept would be clearer from
the following examples:
1. The number of persons out of the total employees of a company, who
are engineers.
2. The total number of households in a city who have computers in
house.
3. The crop production of a particular variety in a district in the current
year.
4. The number of devotees visiting a shrine in a year.
Many events and activities are occurring in the world
(universe). By making the definition of population we clearly mark the
scope of the study/ information gathering process.
It is generally neither practical nor necessary to observe the
entire population. Rather, we observe only a small subset of it at a
time, often doing so periodically. When we do it this way, we are said
to be sampling the population.

Sample:
To find out how the population is behaving, we will observe only a
part of the population and gather data. We will then use this data to
infer something about the population. Sample is a sub-set of the
population.
For example
The number of engineers within a department, in a company.
The number of household in a block of a city, who have computers.
The crop production of a particular variety in a small village of the
district.
The number of devotees visiting a shrine per month.

UNDERSTANDING VARIATION
No two things are totally alike or identical. Hence variation is a
natural phenomenon and is universal.
No two products or characteristics are exactly alike, because any
process contains many sources of variability. The differences among
products may be large, or they may be almost extremely small, but they
are always present.
The diameter of a machined shaft, for instance, would be
affected by,
Machine (clearances, bearing wear)
Tool (strength, rate of wear)
Material (chemical composition, hardness)
Operator (part feed, accuracy of centering)
Maintenance (lubrication, replacement of worn parts)
Environment (temperature, constancy of power supply)

Some sources of variation in the process cause very short-run piece-to-piece

differences
e.g., backlash, clearances within a machine and its fixtures,
Other sources of variation tend to cause changes in the output only over a
longer period of time, either gradually as with tool or machine wear, step-wise as
with procedural changes, or irregularly, as with environmental changes such as
power surges .
Therefore, the time period and conditions over which measurements are
made will affect the amount of the total variation that will be present
From the standpoint of minimum requirements, the issue of variation is often
simplified - parts within specification tolerances are acceptable, parts beyond
specification tolerances are not acceptable; reports on time are acceptable, late
reports are not acceptable. However, to manage any process and reduce
variation, the variation must be traced back to its sources. The first step is to
make the distinction between causes of variation.

Causes of Variation:
There are two causes of variations
Common Causes (also called as Chance Causes or Non-controllable
Causes)
Special Causes (also called as Assignable Causes or Controllable
Causes)
COMMON CAUSES
Arise from causes that are inherent in the process
Some degree affect all the output of the process
Exist even as the process is statistically stable and behave like a
constant system.
While individual measured values are different as a group they tend
to form a distribution pattern that can be described as predictable in
terms of Location (typical value), Spread (amount by which the
smaller values differ from the larger ones) and Shape (the pattern of
variation - whether it is symmetrical, peaked, etc.).

The extent of Common Causes of variation can be indicated by

simple statistical techniques, but the causes themselves need more
detailed analysis to isolate.
These common causes of variation are usually the responsibility of
management to correct, although other people directly connected with
the operation sometimes are in a better position to identify these
causes and pass them on to management for correction.
Overall, though, the resolution of common causes of variation usually
requires actions on the system.

SPECIAL CAUSES
Affect only some output of the process and are not inherent in the
process.
Special Causes refer to any factors causing variation that cannot be
adequately explained by any single distribution of the process output,
as would be the case if the process were in statistical control.
Unless all the special causes of variation are identified and corrected,
they will continue to affect the process output in unpredictable ways.
Examples of Special Causes of Variation include
People
Fatigue, illness, state of health.
Lack of training.
Lack of discipline.
Lack of knowledge of the requirements.
Lack of motivation, attitudes.
Uncontrolled experimentation

Plant/Machines
Rotation of machines
Differences in test or measuring devices
Worn out machinery or tools
Lack of scheduled maintenance
Badly designed equipment
Process/methods
Unsuited techniques of production
Untried /new production processes
Materials
Changes in inspection methods
Mixing of parts, batches
Accumulation of waste products or impurities
Lack of homogeneity
Changes of sources of supply
Environment
Gradual deterioration in conditions.
Seasonal, daily, weekly changes.
Variations in temperature & humidity
which affects the process
Variation of noise & dust which
affect the operator

Special causes of variation result in output behaviour that is usually erratic

and unpredictable.
The discovery of a special cause of variation, and its removal, are usually
the responsibility of someone who is directly connected with the operation,
although management sometimes is in a better position to correct.
The resolution of a special cause of variation, then, usually requires local
action. Charting highlights occurrence of special causes.
Special Causes of Variation
Have frequency distributions that are unstable over time
Result in unpredictable outcomes
Reappear local unless action is taken
Can only be eliminated by actions on the shop floor

Random Variation

Non-Random Variation

Only common cause are

present

Common & Assignable cause

are present

Common causes are more in

nos.

Assignable causes are very few

in nos.

Common causes are part of

process

Visitor to the process

Contributes to constant
variation

Highly fluctuating variation

Predictable

Unpredictable

Statistics Apply

Statistics shall not apply

Management controllable

Operating personnel
controllable

Only a relatively small proportion of all process troubles (industrial

experience suggests about 15%) is correctable locally by people directly
connected with the operation; the majority
the other 85% - is correctable only by management action on the system.
Confusion about the type of action to take is very costly to the organization,
in terms of wasted effort, delayed resolution of trouble, and aggravated
problems. It would be wrong; for example, to take local action (e.g.,
adjusting a machine) when management action on the system was required
(e.g., selecting suppliers that provide consistent input materials).
ACTIONS taken to combat the above types of variation include,
Local Actions are those that can be taken by people close to the process
and are usually required for eliminating special causes of variation. These
actions can only correct about 15% of the process problems.

Actions on the System are those that almost always require management
action for correction and are usually required for eliminating common
causes of variation. These actions can correct about 85% of the process
problems.

VARIABLE & ATTRIBUTE

Every process generates data that can be categorized in Variables

data & Attributes data.
Variable data:
Variables data relates to what can be measured and
expressed quantitatively in specific units of measurements as
dimensions, volume, temperature, pressure, time, strength, etc.
The examples include:
Diameter of a bearing
Volume of the cylinder
Temperature of an oven
Weld strength of a location
Capsule filling time
Hardness of a gear

Attributes data: Attributes data are qualitative data where the results
are recorded in terms good or bad, defective or defect free, pass or fail,
yes or no, acceptable or not acceptable, conforming or non-conforming,
presence or absence of a desired characteristic.
Examples include
sticker printing acceptable or not acceptable
an item meets the gauge or not
Dents on body pass or fail
Brazing joint good or bad

SELECTING SAMPLE SIZE

One of the first basic questions which comes in our mind before
carrying out a survey or statistical study is How many samples
should I select?
Generally we select the samples as per our convenience or
without any particular logic except cost & time.
sampling is time consuming and costly, our objective in selecting a
sample is to obtain a specified amount of information about a
population at a minimum cost.
Samples / subgroups for control charts
Rational subgroups or samples are collection of individual
measurements, whose variation is attributable only to a constant
system of common causes. In the development & continuing use of
control charts, subgroups or samples should be chosen in a way that
provides the maximum opportunity for measurements within each
subgroup to be alike (subject only to forces of common cause
variation) and maximum chance for subgroup to differ from one
another if special cause arise between subgroups.

Within sample or subgroup only common cause variation should be

present.
Sample size considerations
The size of a rational sample is governed by the following
considerations.
Subgroups should be subject to common cause variation.
Subgroup should ensure the presence of normal distribution for the
sample means
Subgroups should ensure good sensitivity to the detection of special
causes.
Subgroups should be small enough to be economically appealing
from a collection & measurement standpoint
Generally 4 to 6 sample / subgroup is commonly used
considering the above factors.

Note :
Shewhart suggests 4 as the ideal subgroup size In
industrial use of control chart 5 is most common Essential idea
of control chart is to select sub groups in such away that it gives
minimum opportunity for variation within sub group, it is
desirable to be as small as possible (for economic purpose).
A size of 4 is better than 2 or 3 on statistical grounds;
the distribution of 8 is nearly normal for sub groups of four or
more even though the samples are taken from non normal
universe.
Sub groups of 2 or 3 are used ONLY when cost of
measurement is high

METHODOLOGY TO IDENTIFY ASSIGNABLE CAUSES

This write up is for assignment any one identify Assignable

causes. As you are aware the success of any SPC program is not in our
ability to collect data, draw charts etc., but in effectively identifying and
eliminating assignable causes. Assignable causes are those causes that
do not allow one to predict the behavior of processes. There is no
meaning in calculating Process Capability without having a predictable
process.
Many companies have initiated SPC charts. But the charts do not
benefit them. One of the main reason for this is that they have not
stopped the process when an assignable cause is indicated and
eliminated the cause. This is not done because no body is aware on how
to do it. Many experts only say that the cause is to be eliminated but no
one is able to assist a company in doing this. We are sharing with you our
approach for doing this cause elimination.

Before starting the SPC data collection, let us do the following steps:
1. Identify the characteristic for which SPC is to be done.
2. Have a brainstorming to list all the causes that may influence the
variation in this
characteristic
3. Prepare a Cause & Effect Diagram
4. Prepare a Master Cause Analysis Table (Annexure 1)
5. Prepare a Why-Why Analysis Table (Annexure 2)
6. Identify factors that may affect Average and those that may affect
Range
After completion of the above, plan for data collection, calculation
of preliminary limits, etc. Then use the chart for On -Line control.
When you are routinely using the chart, when ever a point goes
beyond the control limits, using the Master Cause Analysis Table, we
can narrow the assignable cause (Based on our preliminary listing as
mentioned in Sl. No.6 above) by verifying the condition of the cause
from the limits specified in the table.

ANNEXURE 1
MASTER CAUSE ANALYSIS TABLE
Sl.No

Cause

Is there
any
Spec.?

If so,
what is
the
Specn.?

Basis
For the
Specn?

Is it
checked
& how?

What is
the
actual?

Diff. in
Specfn.
Vs
Actual

Action
plan

ANNEXURE 2
WHY WHY ANALYSIS TABLE
Sl.No

Cause

Why

Why

Why

Why

Why

GUIDELINES FOR USING ANNEXURE 1

1.

Enter a serial number

2.

Enter the cause from the cause and effect diagram. All the
causes from the cause and effect diagram must be covered

3.

For each cause ask the Question: Is there a Specification? Please

note that the specification is for the cause. The answer can be
Yes or No.

4.

If there is a Specification, write the actual value of the

specification. If there is no specification, Enter in the Action Plan
column Specification is to be established.

5.

Give the basis for the specification mentioned in Column No.4.

Sometimes the Specification may be based on the drawing,
machine manufacturers catalogue, work instruction, Past
Experience, etc. Do not write your expectations. Only enter what
is actually existing. Remember that there has to be some basis.

6.

Is this specification being checked. If yes, write the actual method used
for checking. If it is not being checked, then enter No. It may be possible
that there are methods for checking but not done here, in which case the
answer is No. If the answer is No, then enter in the Action Plan Column
Method of checking is to be established.

7.

Enter here the actual value of the cause by using the method of
checking. Sometimes it may be the range of variation (Ex: Input material
condition) or it may be One Value (Ex: Taper in the fixture). This is the
actual value and not a guess. Time may be required to complete this
column.

8.

If there is a difference between the actual value and the specification,

then examine how important based on technical knowledge. If the
difference is not major, then mention No. Otherwise mention Yes. If the
answer is Yes, then enter in the Action Plan Column that further analysis
is needed like Why-Why Analysis or correction to eliminate the variation.

9.

Under this column enter the specific Action Plan needed as already
mentioned above.

GUIDELINES FOR USING ANNEXURE 2

This table can be used for all the causes identified in the
Cause and Analysis Table with top priority for the Causes found to
have variation from the Master Cause Analysis Table.
1. Enter the running serial number
2. Enter the cause to be studied
3. Enter Why this cause should vary. There may be more than one

reason. Enter all the reasons one below the other.

4. For each of the Why identified in Column 3, write the possible

causes. Note that the cause is to be identified only for column 3

and not backward.
5. Proceed in the same manner as Column 4. Ensure that each time

the focus is only on the previous column.

6. Proceed in the same manner as Column 4. Ensure that each time

the focus is only on the previous column.

7. Proceed in the same manner as Column 4. Ensure that

each time the focus is only on the previous column.

Continue in this manner, till any of the following happen:
a. No further Why can be answered
b. The system cause has been identified (Ex: No system
for
checking, verification, control, etc.)
c.The reverse is the solution
8. Based on the listed why's, develop the action plan for

implementation.
Remember:
.

It is the system, which is at the Root Cause of all

problems and not individuals.

SPC PROCEDURE FLOW CHART

MSA

DATA COLLECTION

NO
Is Process
Predicatbl?
(I.e. In Control

Find out assignable

cause and eliminateit.

NO

Yes

Is process No
Capable?

Yes
Establish the Control
Limits
Prepare Reaction
Plan
On going
Process Control

Improve the
process

Measures of central tendency & dispersion

There are various measures of central tendency & dispersion, the most
commonly used are as follows
Mode: It is the most often occurring observation. It may be noted that
more than one mode may exist in a population.
Range: It is the midpoint between the highest & lowest observation.
Median: It is the middle observation when all observations are arranged
in the order of magnitude.
Mean: It is arithmetic average of observations
Example: Consider a sample of 10 observations as follows :
2.1, 2.5, 2.3, 2.2, 2.4, 2.4, 2.3, 2.3, 2.3, 2.1
Find out Mode, Range, Median, Mean

CONTROL CHARTS : TOOL FOR SPC

Dr. Walter Shewhart of the Bell Laboratories, in the 1920's, first made
the distinction between controlled (Common Causes) and uncontrolled
variation (Special Causes), and developed a simple but powerful tool to
dynamically separate the two - the control chart
Several types of control charts have been developed, to analyse both
variables and attributes. However, all control charts have the same two
basic uses, they are:
A judgment, to give evidence whether a process has been
operating in a state of statistical control, and to signal the
presence of special causes of variation so that corrective action
can be taken.
An operation, to maintain the state of statistical control, by
extending the control limits as a basis for real-time decisions .

GUIDELINES FOR SELECTION OF APPROPRIATE CHART

Various types of Control Charts are employed based on the type of data
generated. The chart below provides information about selection criteria of
control charts used in various situations.
T y p e o f D a ta

V a ria b le

A ttrib u te

Is n = 1 o r n > 1

Is n = 1 o r n > 1

n=1

n >1

n=1

n >1

X & MR

X & R or X & s

U n it o r P o r t io n

C o n s t a n t o r V a r y in g

U n it

P o r tio n

C o n s ta n t

V a ry in g

np

PROCESS CONTROL :A process is said to be operating instate of statistical control when the
only source of variation is common causes.
PROCESS STABILIITY: The process is said to be stable when the process is in control and
variation is constant with respect to time i.e. Being in statistical control.
PROCESS CAPABILITY:
The measure of inherent variation of the process ( i.e Six Sigma (+/- 3 )
when it is in stable condition is called as process capability.
OVER ADJUSTMENT:
It is the practice of adjusting each deviation from the target as if it were
due to a special cause of variation in the process. If stable process is adjusted
on the basis of each measurement made, then adjustment becomes an
additional source of variation and inturn it increases the total variation.

CHECKING FOR PROCESS PREDICTION :Process is said to be predictable when it is in control and stable i.e. when
all Special causes are removed from the process. The process can be
checked from Control Chart and Histogram.
Control Chart: - When all points are within control limits or there is no
obvious run or non-random pattern of points with in the control limits.
Histogram: - When bell shape is observed on Histogram.
REMOVING ASSIGNABLE CAUSES:
When process is fail to satisfy above requirements then existence of
special causes may be there. In this cause find special causes and
remove.

CALCULATING PROCESS CAPABILITY: After removing all special causes from the process calculate the capability
indices Cp and Cp k.
If Cp and Cp k is greater than 1.33 then process is said to be within acceptable
capability. Based on the priority make improvement plan for the process.
If Cp and Cp k is less than 1.33, then find out major common causes and remove
it.
ESTABLISHING CONTROL LIMITS: When Cp and Cp k is greater than 1.33 , Then , Establish UCL/LCL and CL
marked on control chart and Issued to Operators for ongoing control.
PREPARE REACTION PLAN: After deciding control limits, Corrective and disposition actions to be given to the
operators for any special causes expected to occur during the process. These
corrective and disposition actions can be documented in Reaction Plan.

ON GOING PROCESS CONTROL :Continuous Periodical review of control chart and recorded process
events to identify the preventive action and revise the control limits.
OPERATOR'S ROLE IN SPC FOR INDIVIDUAL SUB-GROUP.
Data Collection

Plot On Chart

Is process in Control?

Yes

I.e. no special cause

No

Refer Reaction Plan

Take Corrective Action

Take Disposition
Action ( If Reqd.)

NOTE :
1.After Corrective action taken, the Immediate subgroup shall be measured and plotted.
2.Record in detail the causes, corrective action and disposition action taken for every out of
control condition.

FACTORS FOR COMPUTING LIMITS

n
2
3
4
5
6

d2
1.128
1.693
2.059
2.326
2.534

A2
1.880
1.023
0.729
0.577
0.483

D3
0
0
0
0
0

D4
3.268
2.574
2.282
2.114
2.004

E2
2.66
1.77
1.46
1.29
1.18

1.

DATA COLLECTION
x - R CHART:
x-R chart is developed from measurements of a particular
characteristic of a process output. x-R chart explains process data in
terms of both its spread (piece to piece variability) and its location
(process average).
DATA COLLECTION:
x
Measure of Location
R Measure of Spread
Scan the plot points, confirm that the calculations and plots are
correct. Make sure that the plot points for the corresponding x
and R is
vertically in line.
Initial study charts used for first time capability or for studies after
process improvements/changes should be the only process control
charts allowed on the production floor which do not have control limits
placed on them.

2. CALCULATE CONTROL LIMITS :

R = ( R1 + R2 + + RK ) / K
x= ( X1 + X2 + + XK ) / K
Where
K is the number of subgroups.
R1 is the range of the first subgroup.
x
is the average of the first subgroup.
Setup control charts :
x and R charts are normally drawn with the x chart above the R
chart.
Data block should include spare for each individual reading, average
(x ), Range ( R ) and the date/time or other identification of the
subgroup.

Characteristics to be plotted are the sample average (x ) and

the sample size ( R ) for each subgroup, collectively these reflect
the overall process average and its variability.
Average (x ) = ( X1 + X2 + . +Rn ) / n
Where n subgroup sample size.
Range ( R ) = Highest Lowest
Select the Scales for control charts :
Some general guidelines for determining the scales may be
helpful, although they may have to be modified in particular
circumstances.
For x Chart :The difference between the highest and the
lowest values on the scale should be at least two times the
difference between the highest and the lowest of the subgroup
averages (x ).

For R Chart :
Value extends from zero to an upper value about two times the largest range.
For R Chart :
UCLR = D4 RR
LCLR = D3 RR
For X Chart :
UCLx = XX + A2 RR
LCLx = XX - A2 RR
Where D4, D3, A2 are constants varying by sample size with values from sample
sizes from 2 to 10.
Draw the average (RR ) and process average (XX ) as solid horizontal lines.
Control limits ( UCLR , LCLR,UCLx, LCLx ) as dashed horizontal lines. Label the lines.

3. INTERPRETATION FOR PROCESS CONTROL

Since the ability to interpret either the subgroup ranges or
subgroup averages depends on the estimate of piece to piece
variability, the R chart is analyzed first. The data points are
compared with the control limits, for points out of control or for
unusual patterns or trends.
For Range Chart :
(a) Points beyond the control limits are primary evidence of noncontrol of that point. Any point beyond a control limit is the signal
for immediate analysis of the operation for the special cause.
A point above the control limit is generally due to
(1) Plot point may be miscalculated.
(2)Piece to piece variations has increased.
(3)Measurement system has changed.

A point below the control limit is generally due to

(b)

(1)

Plot point is in error.

(2)

Piece to piece variation has decreased.

(3)

Measurement system has changed.

Presence of unusual patterns or trends even when all ranges

are within control limits, can be evidence of change of process
spread, also indicates some special causes.

(c) Runs
(1)7 points in a row on one side of the average indicate that
the process is not normally distributed and there is shift in the
process average.
(2) 7 points in a row that are consistently increasing or
decreasing.
(d) Presence of cycles in the chart indicates that special causes
due to machine set up, non-uniformity in the material, wear of
machine etc.,

Find and Address Special Causes

For each indication of special cause in the range data, conduct an
analysis of the operation of the process to determine the cause and to
improve the process.
A process log may also be a helpful source of information in terms of
identifying special causes of variation. Single point out of control is reason to
begin an immediate analysis of the process.
Recalculate Control Limits
When conducting an initial process study or a reassessment of process
capability, the control limits should be recalculated to exclude the effects of
control periods for which process causes have been clearly identified and
removed.

SUMMARY OF TYPICAL SPECIAL CAUSE CRITERIA

SL.NO

SPECIAL CAUSE CRITERIA

1 Point more than 3 standard deviations from centerline

7 Points in a row on same side of centerline

6 points in a row, all increasing or all decreasing

14 points in a row, alternating up or down

2 out of 3 points >2 standard deviations from center line (same side)

4 out of 5 points>1 standard deviations from center line (Same Side)

15 points in a row within 1 standard deviations of centerline (either

side)

8points in a row >1 standard deviation from centerline (either side)

Analyze the data on the AVERAGE CHART

When the ranges are in statistical control, the process spread the
within subgroup variation is considered to be stable. The averages can
then be analyzed to see if the process location is changing over time.
x
xControl limits for x are based upon the variation in the ranges. Then if
the averages are in statistical control, their variation is related to the
amount of variation seen in the ranges (common cause variation of the
system). If the averages are not in control, some special causes of
variation are making the process location unstable.
Points beyond control limits indicates that there is
(1) shift in process
(2) Plot points are in error.
Find and address the special causes and then recalculate the control
limits after eliminating the special causes.

4. INTERPRET FOR PROCESS CAPABILITY

Interpretation process capability is to be carried out only under the
following assumptions:
(1) Process is statistically stable.
(2) Individual measurements from the process conform to normal
distribution.
(3)

Design target is in the center of the specification width.

(4)

Measurement variation is small.

Having determined that a process is in statistical control, the question

still remains whether the process is capable of meeting customer needs.
To understand and improve the capability of a process, one should
understand that capability reflects variation from common causes and
management action on the system is required for capability
improvement.

Calculate Process Standard Deviation:

Since within subgroup process variability is reflected in the
subgroup averages, the estimate of the process
standard
deviation can be based on the average range (R).

= RR / d2
Where
RR the average of the subgroup ranges.
d2 the constant varying by sample size.

Capability can be described in terms of the distance of the

process average from the specification limits in
standard
deviation units, Z.

For unilateral tolerance

Z = ( USL XX ) /

(or) Z =( xX -LSL ) /

For bilateral tolerance

ZUSL =( USL XX ) / (or) ZLSL =(XX - LSL) /
Where
USL Upper specification limit,

LSL Lower specification limit

Z Negative value of Z indicates the process average is out

specification.

of

ZMIN Minimum of ZUSL and ZLSL

Z values can be used with the table of the standard normal distribution
(appendix Fin SPC manual) to estimate the proportion of output that will be
beyond any specification.
The value of ZMIN can also be converted to a capability index, CPK
CPK = ZMIN / 3 = Min [ (USL XX / 3

or

(XX LSL) / 3 ]

A process of ZMIN = 3 would have a capability index of CPK = 1.00

If ZMIN = 4, the process would have a capability index of CPK = 1.33

Evaluate process capability

It is necessary to evaluate the process capability in terms of meeting customer
requirements.
Fundamental goal is never ending improvement in process performance.
Improve the process performance by reducing the variation that comes from
common causes, or shift the process average close to the target. This generally
means taking management action to improve the system.
Improve Process Capability
To improve process capability, there must be increased attention on reducing
the common causes. Accounts must be directed towards the system namely,
the underlying process factors which account for the process variability such as
(1) Machine performance.
(2) Consistency of input materials.
(3) Basic methods by which process operates.
(4) Training methods.
(5) Working environment.
As a general rule these system-related causes for unacceptable process
capability may be beyond the abilities of the operator or their local supervisor
to correct. Instead they may require management intervention to make basic
changes, allocate resources and provide the co ordination needed to improve
the overall process performance.

XR S (Average and Standard Deviation charts)

Range charts were developed as measures of process variation
because the range is easy to calculate and is relatively efficient for
small subgroup sample sizes. Sample standard deviation S is
more efficient indicator of process variability especially with larger
sample sizes. It is sensitive in detecting special causes of variation.
Gather Data
(1) If raw data are voluminous, they are often recorded on a
separate data sheet.
(2) calculate subgroup sample standard deviation
S={

(XI XR )2 / ( n 1) }1/2

Where XI Individual values

XR Average
n Sample size.

Calculate Control Limits

UCLS = B4 s

LCLS = B3 s

UCLX = xX + A3 s

LCLX = xX A3 s

Where s Average of individual subgroup samples standard

deviation.
B4, B3, A3 Constants varying by sample sizes.
Interpret for Process Control
Refer XR R charts Interpretations for process control.
Interpret for Process Capability
Refer XR R charts Interpretations for process capability.
Process Standard deviation,
O = s / C4

where

s Average of sample standard deviation.

C4 Constant varying by sample size.

MEDIAN CHARTS
Median charts are alternatives to XRX and R charts for control of

processes with measured data.

(1) easy to use
(2)Since a single chart shows both the median and
spread, it can be used to compare the output of several
processes.
Gather data
(1) Median charts are used with subgroup sample size of
10 or less. Odd sample sizes are more convenient.
(2) If using even size subgroups, the median is the
average of middle two units.
(3) Enter the subgroups median (XRX ) and range ( R ) in
the table. It is recommended to also plot the range chart
to observe trends or runs in range.

Control Limits

UCLR = D4 RR
UCLX = XRX + A2 RR

LCLR = D3 RR
LCLX = XRX - A2 RR

where , D4 ,D3 ,A2 are constants varying by sample sizes.

Interpret for process control and process capability
XRX R charts (refer back ).

is same as that of

CHARTS FOR INDIVIDUAL AND MOVING RANGE ( X - MR )

It is necessary for process control to be based on individual readings
rather than subgroups.
Control chart for individuals can be constructed as described below.
In such cases the within subgroup variation is effectively zero.
(1) Charts for individuals are not sensitive in detecting process
changes as XR and R charts.
(2) Since there is only one individual item per subgroup, values of X
and can have substantial variability (even though the process is
stable) until the number of subgroups is 100 or more.

Gather Data
(1) Individual readings (X) are recorded.
(2) Calculate the moving range (MR) between the individuals. It is
generally best to record the difference between each successive pair
of readings (eg: difference between first and second reading , the
second and third etc.).
(3) There will be one less such moving range than there are
individual readings.
Control Limits
UCLMR = D4 RR LCLMR = D3 RR
UCLX = X + E2 RR

LCLX = X - E2 RR

where RR is the average moving range.

XR is the process average.
D4 , D3 and E2 are constants that vary according to the sample
size n.
Interpret for process control and process capability is same as that of
XR R chart
(refer back).

Evaluation of process capability

There are two indices for calculating process capability.
These are as follows
Cp and Cpk.
The Cp index determines the spread or the distribution of the
process. It is given by
Upper spec. limit - Lower spec. limit
Cp
6 Standard Deviation
Or

USL - LSL
Cp
6
The Cpk index determines the central tendency of the process.
Not only the process has to be capable , but it has to meet the
specifications. It is given by the following formula
Cpk Minimum of

USL - Mean
Mean - LSL
and
3
3

The process must first be brought into statistical control by detecting and
eliminating special causes of variation. Then its performance is predictable, and its
capability to meet customer expectations can be assessed. This is the basis for
continuing improvement.
The following picture illustrates the factors on which Process Capability depends
LSL

USL

Specification
Limits

LSL

USL

Process Location

LSL

USL

Process Width or
Spread

LSL

USL

Process Width or
Spread

Two important Indices associated with Process Capability Analysis are Cp (Process
Potential Index & Cpk (Process Capability Index). Cp assesses the ability of a process
to meet Engineering Specifications and Cpk assesses the actual conformance of a
product or a process to the Engineering Specifications.

Process Potential Index Cp :

Cp is a measure of the ability of a Process to consistently generate a given
characteristic within the Specification Limits. Process Potential,
1. Measures what best the process can do
2. Does not always measure what the process is doing
3. Depends upon Process width (spread)
4. Depends upon the Engineering Tolerances
5. Is a comparison of the Process Variation with respect to the Engineering
Tolerance
It is obvious that Cp can only provide information about what the Process can do
and nothing about what the process is doing

See the following illustrations,

LSL

USL

(A)

LSL

USL

(B)

Engineering Tolerance
Process Width

LSL

USL

(C)

Engineering Tolerance
Process Width

Engineering Tolerance
Process Width

In all cases (A), (B), & (C) the there is no change in the values of Engineering
Tolerance and the Process Width. Thus, the Cp value will be constant and
same for (A), (B) as well as (C).
However, it is obvious that,
Case (A) : The process has wandered towards the Lower Specification Limit
(LSL), as such a proportion of the output will be below the LSL, which will not
satisfy the Engineering Specifications.
Case (B) : The process is centered and the Process Width is equal to the
Engineering Specifications, as such almost the entire population (99.73%) of the
output will satisfy the Engineering Specifications.
Case (C) : The process has wandered towards the Upper Specification Limit
(USL), as such a proportion of the output will be above the USL, which will not
satisfy the Engineering Specifications.

Interpretation of Process Capability

Having determined that the process is in statistical control, the question still
remains whether the process is capable?
Does its output meet customer needs?
Capability reflects variation from common causes and management
action on the system is almost always required for capability improvement.
Assessment of process capability begins after special causes have
been eliminated and outgoing control charts reflect, a process that is in
statistical control. In general the distribution of process is output is compared
with specifications, so as to ensure that these are met.

PROCESS CAPABILITY -- ACCURACY AND PRECISION

We have seen the concept of variation and also the distribution. That
was in isolation. It is very important to study their relation.
The process is said to be accurate if the value of its mean /central tendency of its
output is close to the targeted value.
The process is said to be precise if spread of the output is small. Thus lower the
spread value, more precise is the process. In other words precision is related to the
standard deviation.

Example showing the difference between PRECISION & ACCURACY

Precise & Accurate

Imprecise but Accurate

Precise but Inaccurate

Imprecise & Inaccurate

From a statistics perspective, There are only two problems.

Problem with Centering

Problem with Spread

Desired

Desired

Current
situation

Current
situation

LSL

USL

LSL

USL

Shift
Accurate but not Precise

Precise but not Accurate

3 Sigma Vs 6 Sigma
6 Sigma curve

LSL

USL
3 Sigma curve

10

11 12 13

14

Ina3sigmaprocessthevaluesarewidelyspreadalongthecenterline,
showingthehighervariationoftheprocess.Whereasina6Sigma
process,thevaluesareclosertothecenterlineshowing
lessvariationintheprocess.

15

16

3 Sigma Vs 6 Sigma

The comparison on the previous slide also shows a process

in which the process spread of Six Sigmas (ie. 3 Sigmas on either side
just fits within the specifications). In this case one must be extremely
careful to ensure that the process average never slips off the target,
otherwise the curve will shift and non-conforming items will increase.
With Six Sigma requirement the process mean can shift by as much
as 1.5 sigma before the likelihood of non-conforming items is
increased. Even if the process mean does shift off center by as much
as 1.5 sigma, only 3.4 non-conforming items per million parts should
result.

Amount of process shift allowed

1.5 SD

1.5 SD
LSL

USL
SD = 1

10

11 12 13

14

15

16

Sigma levels and defect rate

*Parts Per Million

Sigma levels & Cost of quality

Sigma

Defect
rate(PPM)

Cost of quality

3.4

<10%

233

10-15%

6210

15-20%

66807

20-30%

308537

30-40%

690000

>40%

Competitive level
World
Class

Industry
Average
Non
Competitive

P - CHART
P chart measures the proportion of non conforming items in a group of
items being inspected.
eg: 7 pieces are defective out of 70 pieces.
Before P chart can be used several preparatory steps must be
taken :
(1) Establish an environment suitable for action.
(2)Process must be understood in terms of its relationship to other
operations/users and in terms of the process elements (people,
equipment , material, methods, environment). Technique such as
cause and effect diagram help make these relationships visible.
Subgroup Size :
Charts for attributes require large subgroup sizes to be able to
detect moderate shift in performance.
eg: np > 5

Subgroup Frequency :
Subgroup frequency should make sense in terms of production periods.
short time intervals allow faster feedback.
Subgroup Number :
It must be large enough to capture all the likely sources of variation
affecting the process. Generally 25 or more subgroups.
proportion Non Conforming :
Number of items inspected n
Number of non conforming items found np
Proportion Non Conforming, P = np / n
Process Average Proportion Non Conforming,
P = ( n1 p1 + n2 p2 + . + nK pK ) / ( n1 + n1 + .. + n1 )

Control Limits :
UCLP = p + 3 { p (1 p) / n}1/2
LCLP = p - 3 { p (1 p) / n}1/2
where n is the constant sample size.
Suppose if the sample size varies then take average of sample size ( n ).
Then
UCLP = p+ 3 { p (1 p) / n}1/2
LCLP = p - 3 { p (1 p) / n}1/2
Interpret for Process Control :
. Points above upper and lower control limit is generally sign of higher
proportion non conforming.
Average number of non conforming items per subgroup (np) is large ( 9 or
more), the distribution of the subgroup is nearly normal and trend
analysis can be used. When np becomes small, trend and run analysis is
not applicable
Interpret For Process Capability :
For a P chart, process capability is reflected by the process average non
conforming p. eg: If P = 0.0312
Process capability currently is 3.12% failures of the functional check
(96.88% ok).

Evaluate the Process Capability :

Process capability as just calculated reflects the ongoing level of performance
that the process of generating and can be expected to generate as long as it
remains in control.
np CHART
np chart measures the number of non conforming items in an inspection. It
proves the actual number of non conforming items rather than proportion of the
sample.
Gather Data :
Sample sizes must be equal.
Samples should large enough to allow several non conforming items to appear
in each subgroup.
Control Limits :
Process Average Non Conforming is nP
xp)} 1/2
UCLnp =- nxp + 3 {nxp (1xp)} 1/2
LCL =- nxp - 3 {nxp (1np

Process Capability :
Note that the process capability for an np chart is still P.

C CHART
C-Chart is used when number of defects are found in single unit or
product.
Data Collection :
Sample size must be constant. It is applied to
(1) Non conformities are scattered through a continous flow of
product.
eg: flaws in a bolt of vinyl, bubbles in glam.
Calculate Control Limits :
Process average number of non conformities
C = (C1 + C2 + CK ) / K
UCLC = C + 3 ( C )1/2
LCLC = C - 3 ( C )1/2
Process Capability :
Process Capability is C.

U- CHART
U Chart measures the number of non conformities per inspection reporting
unit in subgroups which can have varying sample sizes.
It is similar to C-Chart except that the number of non conformities is
expressed on a per unit basis.
Control Limits :
U=C/n
where C number of non conformities found.
n sample size of the subgroup.
Calculate process average non conformities ( U ).
UCLU = U + 3( U / n)1/2
LCLU = U - 3( U / n)1/2
Process capability is U, the average number of non conformities per
reporting unit.

THANK YOU