2.

Statistical Process Control

Introduction

 Statistical quality control (SQC) is collection, analysis and interpretation of

data for use in quality control.

 Major branches of SQC is statistical process control (SPC). From the

definition, ‘process’ is specific sets of functions or mechanism established by
management and engineers to produce goods or deliver service for meeting the
customer requirements.

 SPC is used to detect the presence of assignable variations in the process.

 The power of Shewhart’s control chart, however, lies in its ability to separate
out the assignable causes of variations.
 It is utilized for identifying problems in the process, stabilizing it,
monitoring it, controlling it and even improving it.

 SPC is also used to monitor process capability whether the process is

capable of producing or providing goods/service on the basis of
specifications

1
 CAUSES OF VARIATIONS IN QUALITY-:

1. ASSIGNABLE CAUSES-: It refers to those changes in the

quality of the products which can be assigned or attributed to
any particular causes like defective materials, unskilled
worker, machine wear, etc. and which can be corrected.

2. CHANCE CAUSES-: These causes take place as per chance

or in a random fashion as a result of the cumulative effect of
a multiplicity of several minor causes which cannot be
identified.
 These causes are inherent in every type of production.
 Seven quality control tools
 Ishikawa suggested that 90 to 95% of quality problems
can be solved by the use of following seven quality
control tools.
 Tools for generating ideas
 Check sheet
 Scatter diagram
 Cause and effect diagram
 Tools to organize data
 Pareto charts
 Process charts (Flow diagrams)
 Tools for identifying problems
 Histograms
 Statistical process control chart
3
 1. Check Sheet
Data collection tool generally used in the preliminary stages of a study of a
quality problem
 Data often entered by worker as check marks in a given category
Examples:
 Process distribution check sheet - data on process variability
 Defective item check sheet – types and frequencies of defects on the
product
 Defect location check sheet - where defects occur on the product
 Simple data check-off sheet designed to identify type of quality problems
at each work station; per shift, per machine, per operator
 2. Scatter Diagrams
An x-y plot of data collected on two variables, where a
correlation between the variables is suspected
 The data are plotted as pairs; for each xi value, there is a
corresponding yi value
 The shape of the collection of data points often reveals a
pattern or relationship between the two variables
 Scatter Diagram
Effect of cobalt binder
content on wear
resistance of a
cemented carbide
cutting tool insert
 3. Pareto Analysis

 Named after the 19th century Italian economist Vilfredo Pareto

 Often called the 80-20 Rule

 Principle is that quality problems are the result of only a few

problems

 e.g. 80% of the problems are caused by 20% of the causes.

7
 3. Pareto Analysis of Wine Glass Defects (Total Defects = 77)

70 93% 97% 100% 100%

88%
60 54
72% 80%
Frequency (Number)

Cumulative Percent
50
40 60%
30 40%
20 12
20%
10 5 4 2
70.1% 16% 6.5% 5.2% 2.6%
0 0%
Scratches Porosity Nicks Contamination Misc.

Causes, by percent total defects

6-8
 4. Process flow Chart
 Shows sequence of events in process
 Depicts activity relationships
 Has many uses
 Identify data collection points
 Find problem sources
 Identify places for improvement
 Identify where travel distances can be reduced

6-9
 5. Cause and Effect Diagram
 Used to find problem sources/solutions
 Other names
 Fish-bone diagram or Ishikawa diagram
 Steps
 Identify problem to correct
 Draw main causes for problem as ‘bones’
 Ask ‘What could have caused problems in these areas?’
Repeat for each sub-area.

6-10
 Cause and Effect Diagram Example

Problem

Too many
defects

6-11
 Cause and Effect Diagram Example
Method Manpower

Main Cause

Too many
defects

Material Machinery

Main Cause
6-12
 Cause and Effect Diagram Example
Method Manpower

Drill
Overtime
Too many
defects
Wood
Steel Lathe
Material Machinery
Sub-Cause

6-13
 Cause and Effect Diagram Example
Method Manpower

Drill Tired
Overtime
Slow Too many
defects
Wood Old
Steel Lathe
Material Machinery

6-14
 Fishbone Chart - Problems with Airline Customer Service

6-15
 6. Histograms
 A chart that shows the frequency distribution of observed values
of a variable like service time at a bank drive-up window

 Displays whether the distribution is symmetrical (normal) or

skewed

16
 7. Statistical Process Control Charts
 Industrial data can be divided into two types; variable and
attribute form of data.
 The data which can assume all possible values of variables and
which can be obtained by actual measurement is known as variable
form of data (also called continuous data).
Examples are weight, height, resistance, temperature etc.

 The data which can assume limited numerical values and which can
be obtained by counting is known as attribute form of data (also
called discrete data).
For example the number of students in a class, the number of cars sold
in a month etc are discrete data.

17
  The attribute data can be obtained by dividing the data into two groups,
one group possessing the attribute and another not possessing it.

 For example, the Ethiopian population can be divided into employed

and unemployed depending upon whether they posses employment or
not.

 Similarly the products can be divided into two groups, one confirming
to the specification (known as good or quality products) and another
not confirming to the specification (known as defectives or bad
products).

 Any product which do not confirm or meet the specifications is known

as defective.

18
 Control Chart
A control chart is a time plot of a statistic, such as a sample mean, range,
standard deviation, or proportion, with a center line and upper and lower
control limits. The limits give the desired range of values for the statistic.
When the statistic is outside the bounds, or when its time plot reveals certain
patterns, the process may be out of control.
Value This point is out of the control limits
UCL
3 Center
Line
3
LCL
Time

A process is considered in statistical control if it has no assignable causes,

only natural variation.
Control Chart for Samples of 9 Boxes
Variation due to assignable
causes

17=UCL

Variation due to natural

16=Mean causes

15=LCL

Variation due to assignable

causes
1 2 3 4 5 6 7 8 9 10 11 12
Sample Number
Out of control
 PURPOSE & USES OF CONTROL CHARTS
1. Helps in determining the quality standard of the
products.
2. Helps in detecting the chance & assignable variations
in the quality standards by setting two control limits.
3. Reveals variations in the quality standards of the
products from the desired level.
4. Indicates whether the production process is in control
or not.
5. Ensures less inspection cost & time in the process
control.
  Based on the type of data used control charts are broadly divided into two
types; control charts for variables and control charts for attributes.
 Control charts for variables makes use of variable form of data where as
Control charts for attributes makes use of attribute form of data.
The following are different types of control charts used.

I) Control charts for variables

 Control charts for Ranges ( R – Charts)

 Control charts for Standard deviation ( Sigma- chart)

II) Control charts for Attributes

 Control chart for fraction defective (p -chart)
 Control chart for number of defectives (np -chart)
 Control chart for fraction defects per sample (C -chart)
 Control chart for fraction defects per item (U -chart)

22
 EXAMPLE

Sample Quality Characteristic value Sample Sample Sample

Number Average Range standard
R deviation

X1 X2 X3 X4 X5 R1

1 ----- ----- ----- ----- ----- R2

2 ----- ----- ----- ----- ----- R3

3 ----- ----- ----- ----- ----- R4

- ----- ----- ----- ----- -----

- ----- ----- ----- ----- -----

- ----- ----- ----- ----- -----

k ----- ----- ----- ----- ----- Rk

𝜎
𝜎=
𝑘
23
  Schewhart deduced the following four empirical formulas

24
Figure 1- shows a normal probability distribution as a model for a quality
characteristic with the specification limits at three standard deviations on either
side of the mean.

25
 I. Construction of Variable Control Charts
 Variable data are plotted on a combination of two charts. Using X bar chart and
range (R) chart.
 However, S chart (standard deviation) chart should be used in place of a
range chart for larger sample size (n>10).
 The range method loses efficiency relative to S as sample size increases.
 X-bar charts measure shift in the central tendency of the process. It is computed
from the average of each subgroup data. The R Chart is calculated from the ranges
of each subgroup data, which is computed by subtracting the maximum and the
minimum value in each subgroup.
 Range charts monitor the dispersion or variability of the process.

Calculation of ‘X bar’ Control Limits

The center line for the X-bar is the mean of the sample means (X- double bar):
n
i.e.
X i
X  i 1
 (1)
n
n

X i
X  i 1
 ( 2)
26 n
 
where : R  Average Range of the samples
A2  Factor obtained fromtablebelow
n  Number of observation per sample
  Pr ocess s tan dard deviation
 x
 S tan dard deviationof the distribution of samplemeans

27
  If all the sample averages are lying within the control limits, on
either side of the central line in a zigzag manner without exhibiting
a systematic trend, then the process is under statistical control.
 If the sample averages leis either above UCL or below LCL, then
the process is out of control indicating the presence of assignable
variations.
 Then the analyst has to identify the causes or reasons for sample
averages lying outside the control limits, and take corrective action
to eliminate these reasons.
UCL

CL

LCL

Sample number

28
 Calculation of ‘R’ Control Limits

 The method for developing and using R-charts is the same as that for x-bar
charts. The center line of the R chart is the average range, and the upper
and lower control limits are computed as follows:

where D4 and D3 are constants and they depend only on sample size.
For a given sample size, their values can obtained from the standard table
below.

29
 If all the sample ranges are lying within the control limits, on either
side of the central line in a zigzag manner without exhibiting a
systematic trend, then the process is under statistical control.
 If the sample ranges leis either above UCL or below LCL, then the
process is out of control indicating the presence of assignable
variations.
 Then the analyst has to identify the causes or reasons for sample
averages lying outside the control limits, and take corrective action
to eliminate these reasons.

UCL

CL

LCL

Sample number
  In the following control charts, though all the samples are within the
control limits, there exists a systematic trend and hence in the near
future the process may go out of control.
 The sample averages and the ranges should lie on both sides of the
central line in a zig-zag manner without exhibiting a systematic trend

31
 Factors for three-sigma control
limits of X-bar and R-charts

32
  If the process is under statistical control, the following
relations can be used to compute the mean and standard
deviation

33
 Steps in Creating Control Charts
1. Take representative sample from output of a process over a long
period of time, e.g. 10 units every hour for 24 hours.
2. Compute means and ranges for the variables and calculate the
control limits
3. Draw control limits on the control chart
4. Plot a chart for the means and another for the mean of ranges on
the control chart
5. Determine state of process (in or out of control)
6. Investigate possible reasons for out of control events and take
corrective action
7. Continue sampling of process output and reset the control limits
when necessary
  Problem 1; A manufacturer has decided to use control charts namely X bar

and R chart for shafts which he has to supply to an automobile

manufacturer. He has drawn samples of sample size five each. After 13
samples, he computed the sample averages and ranges and recorded as
below. Compute 3 sigma control limits for control charts for averages and
ranges. Is process under control?

Sample No 1 2 3 4 5 6 7 8 9 10 11 12 13

Averages 40.0 40.2 40.2 40.3 39.9 39.8 40.1 39.8 40.3 40.1 39.9 40.1 40.2

Ranges 0.3 0.4 0.6 0.5 0.2 0.1 0.2 0.5 0.3 0.4 0.3 0.2 0.4

35
 

36
 40.4

40.3

40.2
X

40.1

x-mean

40

LCL

39.9

UCL
39.8

39.7

39.6

39.5
1 2 3 4 5 6 7 8 9 10 11 12 13

37
  Inference: Since four sample averages are lying outside the

control limits, the process is out of statistical control.

 This indicates that there are assignable variations and hence the

quality control engineer should identify the assignable causes,

analyze them and take corrective action to eliminate them.

38
  R-Chart:

0.8

0.7
R
0.6

0.5 R-mean
0.4
LCL
0.3

0.2 UCL

0.1

0
1 2 3 4 5 6 7 8 9 10 11 12 13

 Inference: Since the sample ranges of sample are lying within the control
limits, on either side of the central line in a zigzag manner without exhibiting
a systematic trend, then the process is under statistical control.
 This indicates that there are no assignable variations and the process is
operating under chance variations only.
39
 Why to use both X bar and R -Charts together?
 The X - Chart makes use of sample averages. On the basis of X bar -
Chart, if one concludes that the process is out of control, it means
that there is variation from one sample to another sample.

 On the other hand R - Chart makes use of sample ranges. On the basis
of R - Chart, if one concludes that the process is out of control, it
means that there in a sample one quality characteristic value is too
high and another quality characteristic value too low. It means that
there is variation within the sample.
 In general, X bar- Chart shows variation between the samples and
R – Chart gives variation within the sample.
 To know variation between and within samples, X bar - Chart should
be used along with either R– Chart or sigma chart

40
  Problem 2: A textile manufacturer has decided to initiate the use control
charts for the strength of one of the textile yarn. He has drawn ten samples of
sample size five each. The average strength of each sample is determined. For
these ten samples the summation of the sample averages and the ranges are
found to be 800 and 48 units respectively. Compute 3 sigma control limits
for X bar - Chart and R– Chart and find the population average and
population standard deviation assuming the process under control.

Solution:

For sample size n = 5, from statistical tables, the various constants are obtained
as below.

A2 = 0.577, D3 = 0, D4 = 2.114 d2 = 2.326

41
  X-bar chart
= 80+(0.577)*4.8=82.7696
= 80- (0.577) *4 .8= 77.2304

 R-Chart

Assuming the process under statistical control,



42
 Moving Average (MA)/Range Chart
 Moving Average / Range Charts are a set of control charts for variables
data.
 The Moving Average chart monitors the process location over time,
based on the average of the current subgroup and one or more prior
subgroups.
 The Range chart monitors the process variation over time.

 Moving Average Charts are generally used for detecting small shifts in
the process mean.
 They will detect shifts of 0.5 sigma to 2 sigma much faster.
 They are, however, slower in detecting large shifts in the process
mean.
  Always look at the Range chart first.
 The control limits on the Moving Average chart are derived from
the average range, so if the Range chart is out of control, then the
control limits on the Moving Average charts are meaningless

 Many organizations are involved in continuous processes, such as

manufacturing steel, aluminum, paint, oil or chemicals.
 In those processes it is impractical to establish sample consisting
of successive units produced.

 In such cases a moving average (X-bar) and a moving range (R)

are calculated from individual measurements.

44
  The interpretation of a point outside control limits on moving
averages and range chart is the same as a point outside control limits
on conventional X-bar and R charts.

 However, because successive points on moving average and range

charts are not independent of one another, the interpretation of
several points in a row outside control limits is obviously not the
same.

 Similarly, runs above or below the central line do not have the
same significance on moving average and range charts as on
conventional X-bar and R charts.

45
 Working Rules

Analysis of Patterns on Control Charts

Western Electric Handbook Rules

A process is considered out of control if any of the
following occur:
1) One point plots outside the 3-sigma control limits.
2) Two out of three consecutive points plot beyond the 2-
sigma warning limits.
3) Four out of five consecutive points plot at a distance of
1-sigma or beyond from the center line.
4) Eight consecutive points plot on one side of the center
line.
46
 Analysis of Patterns on Control Charts

47 Figure 15-7 The Western Electric zone rules.

 X-bar and Sigma (S or ) Control Charts

3-sigma control limits for S:

An unbiased estimator of :

48
 X-bar and S Control Charts

S Chart:

x Control Chart (from S ):

49
 II. Control charts for attributes
 Defective : Any product which do not confirms or meets the
specification limits is know as defective (or non-confirming item).

 Defect: The reason for which the product is treated as defective is

known as defect.
 For example a product may be defective because of oversize. Then
oversize size is defect.
 Similarly, undersize, more concentration, more dilution, overweight,
under weight may be the possible defects.

 A defective item may have one or more number of defects.

 P and np-charts are for defectives and C and U-charts are for defects.
50
Control Charts for Attribute Data
 Here, you are counting something
 The number of defective units in a batch of units
 - OR - A better word here is
“nonconforming”

 The number of defects on one unit

A better word here is

“nonconformities”

 For attribute data, you draw one chart or the other (not
both)
 P-Chart (control chart for fraction defective or non-conforming )
 P-charts are used to measure the proportion that is non-conforming in a
sample. They are appropriate when both the number of non-conformities
measured and the size of the total sample can be counted. Control limits for
the P chart is calculated as:
P (1  P )
UC LP  P  3 (4)
n

C LP  P (5)

P (1  P )
LC LP  P  3 (6)
n
_
where : P  The average proportiondefectivein the sample
m m

m i p i
m
P i 1
 i 1
, p  i
i  1,2,..., k
kn k i n
k  numberofsamples
n  sample size
Di  # of defectives
pi  fraction defective
52
53
Example 3: Repeated samples size of 100 coffee cans are inspected to
determine whether a can is out of round or whether it contains leaks due
to improper construction. Such a can is said to be nonconforming. The
Following table shows the data provided for it.

Sample 1 2 3 4 5 6 7 8 9 10

Nonconforming # 19 10 4 6 8 9 3 1 0 4

Fraction defective 0.19 0.10 0.04 0,06 0,08 0.09 0.03 0.01 0.0 0.04

 m  64
p
 m

64
 0.064
n*k 100*10
p (1  p )
UCL  p  3  0.1374
nj
CL  0.064
p (1  p )
LCL  max[ 0, p  3 ]0
nj
 Figure 1: P- chart

55
 C-charts
 The C and U-charts are used for defects.
 The c-chart is used to monitor count of nonconformities within the
product whereas the u-chart is applied to control count of nonconformities
per unit.
 Usually an entire unit is labeled as nonconformities whether it has one or
more nonconforming products.
 For both c and u, a given unit equals a sample size (for example, 100-
meter of cloth, gross of pencil).

 When the sample size is constant use: C –chart and

 when it varies use: U-chart.

 Unlike c and u charts, the p-chart is used to monitor fraction or proportion

of nonconformities in the sample or the entire product.
 Unlike c and u charts, the sample size of p-chart is considered on the
number of individual pieces not in a unit.

56
 C-control chart (control chart for non-conformities)
 C-charts are used to monitor the number of defects per unit.
 The types of units of measurement are a period of time, a surface area, or a
volume of liquid etc.
 The c chart is based on the Poisson distribution
 Control limits for the C chart is calculated as:
n

C i
C i 1
n
UCL  C 3 C (1)

CL  C (2)

LCL  C 3 C (3)
_
where : C  The averagenumber of defects

57
Example 4:

An automobile assembly worker is interested in monitoring and controlling the #

of minor paint blemishes appearing on the outside door panel on the driver’s side
of a certain make of automobile. The following data were obtained, using a
sample number of 10 door panel.

Sample 1 2 3 4 5 6 7 8 9 10
# of Paint
Blemishes 19 10 4 6 8 9 3 3 9 7

𝐶𝑖 78
𝐶= = =7.8
𝑘 10

𝑈𝐶𝐿 = 𝐶 + 3 𝐶 = 16.17
𝑈𝐶𝐿 = 𝐶 = 7.8

𝑈𝐶𝐿 = 𝐶 − 3 𝐶 = 0
 Figure 1: C- chart

59
 U-Chart(Control Chart for Number of defects per item)
 It is used when the sample size is varying.
 The chart uses the Number of defects per item. It is obtained by
dividing the control limits of C chart by the sample size as below.

𝐶+3 𝐶 𝑈
𝑈𝐶𝐿 = =𝑈+3
𝑛 𝑛
𝐶𝐿 = 𝑈
𝐶−3 𝐶 𝑈
𝐿𝐶𝐿 = =𝑈−3
𝑛 𝑛

60
61
62
 Process Capability
• Evaluating the ability of a production process to meet or exceed
preset specifications. This is called process capability.

• Product specifications, often called tolerances, are preset ranges of

acceptable quality characteristics, such as product dimensions.

Two parts of process capability

1) Measure the variability of the output of a process, and

2) Compare that variability with a proposed specification or product
tolerance.
 Measuring Process Capability
• To produce an acceptable product, the process must be capable
and in control before production begins.
USL  LSL
Cp 
6
Example
• Let’s say that the specification for the acceptable
volume of liquid is preset at 16 ounces ±.2
ounces, which is 15.8 and 16.2 ounces.

General Rule
Cp Capability
≥ 1.33 Good
1 ≤ Cp < 1.33 Adequate
Cp < 1 Inadequate
 Figure (a)
• The process produces 99.74 percent (three sigma) of
the product with volumes between 15.8 and 16.2
ounces.

Cp  1

In this case the process uses up all

the tolerance band
 Figure (b)
• The process produces 99.74 percent (three sigma) of the
product with volumes between 15.7 and 16.3 ounces.

Cp  1
the process uses up more than 100% of the
tolerance band.
In this case the process is very yield-sensitive,
and a large number of nonconforming
units will be produced.
 Figure (c)
• the production process produces 99.74 percent (three sigma)
of the product with volumes between 15.9 and 16.1 ounces.

Cp  1
This means that the process uses up
much less than 100% of the tolerance
band.
Consequently, relatively few
non-conforming units will be
produced by this process.
Example of different Cp values
 Process capability ratio (off centering
process)
• There is a possibility that the process mean may shift over a period of
time, in either direction, i.e., towards the USL or the LSL.
• This may result in more defective items than the expected.
• This shift of the process mean is called the off-centering of the process.

 USL     LSL 
C p k  min , 
 3 3 
General rule: Cpk >= 1 - the process is capable
Cpk < 1 - the process is not capable
 • Process mean:   15.9
• Process standard deviation:   0.067
Example • LSL = 15.8
• USL = 16.2

0.4
Cp   1.99
6(0.067)

 USL     LSL 
C p k  min , 
 3 3 
 16.2  15.9 15.9  15.8 
C p k  min , 
 3(.067) 3(.067) 
C pk  min1.49,0.49
C pk  0.49
Since Cpk value is < 1, the process is not capable
of meeting the process specs
Average run length for chart performance
Average Run Length
 The average run length (ARL) is a very important way of determining the
appropriate sample size and sampling frequency.
 In general, the expected number of samples taken before the shift is
detected is simply the average run length,
 Let p = probability that any point exceeds the control limits. Then,
 Limitations of Control Chart-:
Does not serve as a ‘PANACEA’ for all quality
evils/problems.
It cannot be used to all production process.
It involves mathematical & statistical problems
in the process of analysis & interpretation of
variations in quality.
Provides only an information services.
 THE END!!!!

75