STATISTICAL

PROCESS CONTROL.
Learning :-

KRISHNA TOMAR
 STATISTICAL PROCESS CONTROL

SPC :. STATISTICAL PROCESS CONTROL.

Requirements of SPC & Current edition :-

 Controlling the process variation by using statistics .
 2nd edition published in July 2005.

Introduction:-
 Statistical Process Control (SPC) is an industry-standard methodology for
measuring and controlling quality during the manufacturing process.
Quality data in the form of Product or Process measurements are obtained
in real-time during manufacturing. This data is then plotted on a graph with
pre-determined control limits.
History of SPC :-
 SPC is developed by Dr. Walter A. Shewhart during 1920 in Bell Lab.
Principal of SPC :-
 Basic principal of SPC is to reduce variation.

KRISHNA TOMAR
Krishna Tomar
Statistics :-
 The set of information derived from the sample data to estimate the
process are called statistics.
Process :-
 Converting an input into output by using Man, Machine, Material,
Method and environment.
Control :-
 Ensuring to make a variable to be with in the stated limit.
Variation:-
Difference between actual output and desired output is called variation.
One problem with mother nature is , Everything is different, No two things produces same in
the world. By all efforts we can only reduce the gap between the two. This gap is known as
variation.
Why do we want to reduce the variation ?
Because variation is the main source of wastage, undesired reworks, rejection, customer
dissatisfaction and many hidden, uncover able costs. Variation can not be eliminate, only we can
minimize the variation.

Effect Results
Less Higher
Variation Quality

High Poor
Variation Quality

KRISHNA TOMAR
Special Cause :-
 Anything that causes variations that are not part of the stable process is called a special
cause, assignable cause, or unnatural cause.

 Batch of defective raw material (दोषपूर्ण कच्चे माल का बैच)

 Faulty set-up (दोषपूर्ण सेटअप)
 Human error (मानव त्रु टट)
 Incorrect recipe (गलत नुस्खा)

Common Cause :-
The cause of variations in a stable process is called a Common Cause. A common cause
is a natural cause of variation in the system.

 Machine vibration (मशीन कंपन)

 Temperature fluctuations (तापमान में उतार-चढाव)
 Slight variation in raw materials (कच्चे माल में थोडा अंतर)
 Human variation in setting control dials (टनयंत्रर् डायल सेट करने में मानवीय टिन्नता)

Process Location :-
 Mean of the process shows the location of process. It is mean value or say average value.

Note :- How much our process location is shift from customer location is tell that how much worst is
our process running.

KRISHNA TOMAR
Spread :- Standard deviation ( Spread also called depression)
 The average distance between the individual numbers and the mean , it is denoted by sigma.
Note (1) :- To find total variation of process - Multiply
6 to Standard deviation or sigma.
Example :- Customer voice is 0.30
Sigma value is :- 0.04, Then 6 x 0.04 = 0.24 (Process
voice) Process is in good condition.
Note (2) :- If our process spread is less than customer
spread then our process is in good condition.
Note (3) :- It is thumb rule that if our process sigma
value is less, then process is running in good condition.
Because sigma is process variation

KRISHNA TOMAR
 Histogram
What is a Histogram ?

A histogram is used to summarize data of one continuous variable that is dimension, weight etc. It is the
most commonly used graph to show frequency distribution.

History of Histogram :-

invented by Karl Pearson follower of Sir Francis Galton. And histogram is one of the seven basic tools of
quality control.

Histogram :- It is used to check

centering, spread and shape of the
process. It is invented by Karl Pearson
in 1891 follower of Sir Francis Galton.

Notes
It looks very much like a bar chart. But there are important difference between them. It differs from a bar
graph, in the sense that a bar graph relates two variables, but a histogram relates only one. A histogram
groups numbers into ranges. It is recommend that bar charts have gaps between the rectangles to clarify
the difference.

Why use Histogram :-

Data have always variation, and variation has a pattern, and with the help of Histogram pattern can be easily
seen when summarized pictorially.
It helps in understand ,Process centering , Spread and shape of process. Helps answer the question, is
process capable to meeting customer requirement. It helps indicate if there has been any change in process .

KRISHNA TOMAR
Krishna Tomar
When to use Histogram :-
When the data is numerical. It is used to check past history, or lot which is manufactured in past .

Importance of Histogram :-
1. Simple to use.
2. Use on all variable data.
3. Histogram provides a visual representation of data distribution.

Point to be noted :-
1. Only 1 variable histogram is drawn in one time or one graph.
2. For draw histogram minimum 50 to 100 psc data should available.

Parts of a Histogram :-

1. The title:. The title describes the information included in the histogram.
2. X-axis:. The X-axis are intervals that show the scale of values which the measurements fall under
3. Y-axis:. The number of times that the values occurred within the intervals set by the X-axis
4. The bars:. The height of the bar shows the number of times that the values occurred within the interval, while the
width of the bar shows the interval that is covered. For a histogram with equal bins, the width should be the same across
all bars.

Distributions of a Histogram :-
A normal distribution:. When our data is as per control limits it is lie in normal distribution.

A normal distribution have 3 types :-

1. Bell Shaped
2. Right skewed distribution
3. Left skewed distribution.

We have to decided first, which type of normal distribution graph is suitable for our process.

Golden Point :- If your histogram have tail effect then histogram is ok.

KRISHNA TOMAR
Krishna Tomar
1. Bell shaped distribution or Symmetric distribution :-
it is suitable where the tolerance is given Bilateral ( both ± side ). Also if tolerance is unilateral, we have to mean the
tolerance and after use it.

Note :- Automobile industry followed Bell shaped distribution.

2. A Right - Skewed distribution :- 3. A Left - Skewed distribution :-

A right-skewed distribution is also called a A Left-skewed distribution is also called a Negatively skewed
positively skewed distribution. As per required distribution. As per required condition we have to use it,
condition we have to use it, For example :- In case of school result how much students gets
For example :- In case of doctor treatment how higher marks.
much patient cured in 1 day or 2 days etc.

KRISHNA TOMAR
Krishna Tomar
 7. Control Chart

What is a Control Chart ?

Control charts monitor how a process change over a time
Control charts, also known as Shewhart charts (after Walter A. Shewhart) or process-behavior charts, are a
statistical process control tool used to determine if a manufacturing or business process is in a state of control
or not. It shows stability or variability in process The control chart is one of the seven basic tools of quality
control
Inventor & History of Control Chart :-
The control chart was invented by Walter A. Shewhart working for Bell Labs in the 1920s.
Control chart is used to know weather the process in control or not. Invented by Dr. Walter Andrew Shewhart in 1924.
He also invented PDCA cycle.

Dr. Walter Andrew Shewhart

Benefits :-
Predict process out of control and out of specifications limit. Distinguish between specific, identifiable cause of
variations. Can be used for SPC

Types of control Chart :-

KRISHNA TOMAR
Krishna Tomar
Average Chart :-
X bar charts show the centering (mean) of the process, with the help of X bar chart we control the mean of the
process.
Range Chart :-
R charts shows the spread of the process with the help of R bar chart we control process spread.

What is the requirement to make X Bar R Chart :-

1. Data is variable
2. Mass production :- Production rate should moderate, not too much and not less. Generally automobile
sector have moderate production rate
3. Data should of single characteristics. That is only one dimension.
4. Generally dimension should CTQ (critical to customer) is used.
5. Data should taken in rotational sub-group.
6. e.g. If we check 5 psc in hour then continuous take data every hour of 5 psc.
7. Data should taken regular for sub-group, e.g. if our sub-group size is 5 psc, then we have to check regular 5
psc of production.
8. Rotational subgroup should 20 to 25 or 100 observation.
9. Control chart is not preferred for 1 shift production, Production should running minimum 2 to 3 days and
data
How to should be of 2 Chart
made control to 3 days.
:-

(1) Collect the data.

(a) Minimum 100 observation should taken.
(b) Number of samples should take 4 , 5 , 6 and should be taken continuously.

KRISHNA TOMAR
Krishna Tomar
 (a)

Points outside of the limit

Control limits are calculated to measure the natural
variability of a process. Any point or on outside the limit is
considered abnormal and requires investigation

(b)
8. List of coefficients of Average or Range Chart

Shift (Run):
A shift is indicated when points lie continually on one side
of the center line, A shift of seven consecutive points is
considered abnormal.

(c)

Trends :
Seven consecutive points in a continuous upward or
downward direction.

KRISHNA TOMAR
Krishna Tomar
 Why we not use line chart and run chart instead of control
charts:-
(d)
1. Firstly we used the line charts for understanding a
problem. But there is no center line, to show which is the
center of process
This is a simple chart , we can not get any result to see this
chart.

Hugging the center line :

When most points lie within the center line and 1.5 sigma,
the process is not in control state. This usually indicates
the mixing of data from different populations.

(e)
2. Secondly we upgrade to Run chart, there we have
available center Line, but this not show how much
fluctuation of line is ok for process, means there is no
upper limit and lower limit of process.

Cycling (periodicity) :
Any repeated, consistent up and down cycle is abnormal
and requires investigation.

(f)

3. Then we come to control chart for perfectly understand

a problem. It shows center, upper limit and lower limit of
process.

Approaching control limits :

2 of 3 points lying outside the 2 sigma line is considered
abnormal.

KRISHNA TOMAR Krishna Tomar

 Difference in Run Chart & Control charts :-

Run Chart :- Control Chart :-

1. A run chart is a line graph of data plotted over 1. A control chart is also a line graph of data plotted
time. over time with control limits.
2. A run chart does not show statistical control 2. Control chart have UCL and LCL, If there is any point
limits. going outside of the control limits it clearly shows
process is out of limits
3. A control chart also help in identification of special
3. A run chart can help in identification of special
causes in process and also help in achieve Statistical
causes in process.
process control.

Calculate process standard deviation or sigma for CP & CPK :-

σ = Ṝ/d₂
There are two types of method to find sigma (σ)
d₂ is constant varying as per sample size (n)
Calculate R (Range) 1. Root mean square method :- RMS method is used to
R = (Maximum value in a subgroup) - find sigma in Histogram , control charts, (Pp) Process
(Minimum Value in a subgroup) performance & (Ppk) process performance index.
Calculate Ṝ(Range bar)
Ṝ = R1 + R2 +……+ Rn
N k = Number of subgroups
2. σ = Ṝ/d₂, method of find sigma is used for Cp &Cpk.

KRISHNA TOMAR
Krishna Tomar
 Process Capability :-
Cp shows only spread of the process, It doesn’t show the location of process ,
Cp: This is the capability index which is defined as the tolerance width divided by the process
capability, irrespective of process centering. Typically, this is expressed as
Cp = (USL-LSL) / 6σ
Cp = Customer specification (Voice of customer )
Process specification (Voice of process )
Note :- 6σ is process specification
Example :- lets assume a dimension 20±0.20 then USL is 20.20 and LSL is 19.80
Customer specification :- USL - LSL = 20.20 - 19.80 = 0.40
Process specification :- 6σ (lets assume σ is 0.04) 6σ = 6*0.04 = 0.24
Cp = USL-LSL / 6σ = 0.40/0.24 = 1.67
KRISHNA TOMAR
 Compare voice of Process with Voice of customer :-

What is the meaning of Cp = 1 ? What is the meaning of Cp = 2 ?

Lets assume our customer tolerance is 1mm and
our process tolerance observed 1mm. Lets assume our customer tolerance is 1mm and our
Then cp = Customer tolerance \ process tolerance process tolerance observed 0.50 mm.
= 1 \ 1 =1 or 100% Then cp = Customer tolerance \ process tolerance
Means we have used 100 % customer tolerance. = 1 \ 0.50 = 2
Means we have used 50 % of customer tolerance.

What is the meaning of Cp = 1.33? What is the meaning of Cp = 0.50 ?

Lets assume our customer tolerance is 1mm and
Lets assume our customer tolerance is 1mm and our
our process tolerance observed 0.75 mm.
process tolerance observed 2 mm.
Then cp = Customer tolerance \ process
Then cp = Customer tolerance \ process tolerance
tolerance
= 1 \ 2 = 0.50
= 1 \ 0.75 =1.33
Means we have used 200 % of customer tolerance.
Means we have used 75 % of customer
tolerance.

What is the meaning of Cp = 1.67 ?

Lets assume our customer tolerance is 1mm and our
process tolerance observed 0.60 mm.
Then cp = Customer tolerance \ process tolerance
= 1 \ 0.60 = 1.67
Means we have used 60 % of customer tolerance. KRISHNA TOMAR
 Compare voice of Process with Voice of customer :-
Cp = 1 & CPk = 1 Cp = 2 & CPk = 2

The Car is moving

The Car width is Same as smoothly in the highway
The Process is Running neck The Process is Running
highway width & The
to neck with Cust. Spec. & A with around 25% safe smoothly well with in the
driver has to drive very distance from both side of
slight process variation can customer specification limits.
cautiously otherwise the highway .
lead to defects.
Can be in trouble.
Cp = 1 & CPk = 0
Cp = 0.6 & CPk = 0.6

The car is around 50% out The Process is marginally

It is difficult for the car to The Process is not capable of the track & is not in a capable,50% Defects ,
run on the highway as the to produce as per cust. condition to continue is this investigation the Causes &
highway is to narrow for the Spec. Process Variation manner . Implements CA to reduce &
movement . need to reduced eliminate them
Cp = 1 & CPk = 0.5
Cp = 1 & CPk = 0.6

Process is completely off

This is an alarming
Center , may be the process
The Process is Off – situation you cant drive
The Car is Moving slightly itself is not stable & Need to
Center , move the Process your car in such a off track
beyond one edge of the check for any special cause
average closed to target to condition
highway variation .
have a defect free
operation KRISHNA TOMAR
 Process Capability Index :-
Cpk = This is the capability index which accounts for process centering and is defined as the minimum of
CPU or CPL ,
To find Cpk we have required process sigma and location (mean) of process.

Conclusion :-
If our Cpk value lies between 1.33 to 2 then our process mean is good.
If our Cpk is lies between 1 to 1.33 then immediate action is required. If our Cpk is less then 1
then our process is in worst condition, stop the process and take high priority action.
 How to calculate Sigma Level or Z score or Zigma ? :-
Method 1 :- Multiply Cpk value to 3, so we get sigma level
Example
If Cpk is 1.33
Then sigma level = 1.33 x 3 = 3.99 or say 4 sigma 4 sigma means process PPM is 6210

Method 2 :- Multiply Cpk value to 3, so we get sigma level

Difference in process sigma and sigma level ?? :-

KRISHNA TOMAR
 Simple Comparison of Process Capability & Performance using indices :-

Cp , CPk Pp , PPk

Cp , CPk are termed as Process Capability & Pp , PPk are termed as Process Performance
Process Capability index . & Process Performance index .
Sampling / Subgroups wise is data collected . Whole Population data is collected .

It captures the within subgroup variation . It captures the within & Between subgroup
variation ( Overall ).
Calculation :- Calculation :-
Cp = (USL-LSL) / 6σ Pp = (USL-LSL) / 6σ
CPk = Minim. (CPu = (USL - ) / 3σ Cpl = ( - PPk = Minim. (PPu = (USL - ) / 3σ PPL = ( - LSL
LSL ) / 3σ ) / 3σ

This sigma is calculated Rbar/d2 . This sigma is calculated By RMS Method( .

Process Capability is determined primarily Process Performance included both Common

by Common Cause of Variation . Cause & Special causes of Variation .

Typically studies are done once of the data / Typically done for the new process or during
Process is normal & Stable . PPAPs ( Also based on Cust. Requirement ).

These are used for process Improvement & These are used for assessing the overall
Capability studies . process performance .
In Cp and Cpk there is need to find Control In Pp and Ppk there is no need to find Control
Charts. Charts.

KRISHNA TOMAR
SPC Example with sample Data :-

KRISHNA TOMAR
 Benefits of SPC :-
SPC my have many benefits for manufacturing companies, including:
 Improved product quality
 SPC helps to ensure that products are more consistent with design spec. , which can reduce
defects, waste, and rework.
Increased productivity
SPC can help to identify and eliminate causes of inefficiency, which can lead to more efficient
operations.
Reduced costs
SPC can help to control costs by reducing the need for manual inspections and warranty
claims.
Enhanced customer satisfaction
SPC can help to identify imperfections before products are delivered, which can boost
customer satisfaction.
Better analytics and reporting
SPC charts can provide a visual representation of performance over time, with upper and
lower limits that indicate levels of variation. This can help to differentiate between common and
special cause variation, and to assess whether a process is stable or unstable.

Drawback to SPC :-
SPC can have several disadvantages, including:
Initial costs: SPC can be expensive to set up.
Complexity: SPC works best in stable, predictable, and repeatable processes, but it may not
able to account for all sources of variation in more complex processes.
Resistance to change: Staff may be resistant to changing their processes.
Data misinterpretation: There's a risk of misinterpreting data.
Requires skilled personnel: SPC may require skilled personnel.

KRISHNA TOMAR