F.

Control Chart

Dr Arunangshu Mukhopadhyay
Professor
Dr B R Ambedkar NIT Jalandhar
1
2
 Control Charts
■ Is it possible to make a fabric same as before 1 year?
■ If we practically see this question we find that it is not possible because
there can be some variation related to men or machine or variation due to
variation in raw material( cotton fibre).
■ If we closely observe we can understand that variation is a day to day
process. But these variation affect quality and hence we make a chart
where we decide to keep a control of variation in process called Control
Chart.
■ The central idea in Control Chart is that, if the production process is
under control, then the values of the statistic will lie in between “mean ±
3σ.”
3
4
 Control Charts
Sl. No. Type of data R P nP C
5. 1Control Charts
Fibre properties such as strength, length and fineness ◘
2 The strength of yarn, cord, and fabric ◘
3 The amount of variation in strength within the breaks or number from a ◘
bobbin, cone, or spool of yarn
4 Yarn, rove and sliver linear density, lap weight ◘
5 Variation of linear density of yarn within a bobbin ◘
6 Percent bad quality ◘
7 Number of bad pieces or units ◘
8 Imperfections in yarn, number of particular defects in fabric ◘

9 Number of bad units with a reasonably constant sample size ◘

10 Number of bad units with a sample size that varies considerably from lot to ◘
lot, day to day, or week to week
5
Control Charts

UCL (mean + 3σ)

Mean

LCL (mean - 3σ)

A typical control chart

7
Specification Limit
Sometimes, the customer provides the control limits for the production
process, such control limits are called the upper specification limit (USL)
and lower specification limit (LSL) and the decision regarding the control
of the production process is taken by taking care of these specification
limits.
Control Charts
Upper Specification
limit

Control Specification
limit limit

Lower Specification
limit
 Control Charts
Type of Control Chart
Attributes Control Charts
Variables Control Chart
“Attributes” counts the number of defective items in a
“Variable” means measuring characteristics sample or the number of defects associated with a
such as time, weight, volume, length, particular type of item
pressure drop, concentration, etc

1. nP chart (control chart for number of

1. X bar chart (control chart for defectives).
mean). 2. P chart (control chart for proportion of
2. R chart (control chart for range). defectives).
3. Standard Deviation S control 3. C chart (control chart for number of
chart. defects).
4. U chart (control chart for number of
defects).
 UCL

Statistic
CL

LCL

Sample number

11
 Control Charts
Type of Control Chart
Variables Control Chart Attributes Control Charts
“Variable” means measuring “Attributes” counts the number of defective
characteristics such as time, weight, items in a sample or the number of defects
volume, length, pressure drop, associated with a particular type of item
concentration, etc

1. nP chart (control chart for number of

1. X bar chart (control chart for
defectives).
mean).
2. P chart (control chart for proportion
2. R chart (control chart for
range). of defectives).
3. C chart (control chart for number of
defects).
For process having mean µ and standard deviation σ
FOR KNOWN mean µ and standard FOR UNKNOWN mean µ and standard
deviation σ deviation σ

n is number of sample
 R chart
To draw R-chart, k different samples, each of size n are selected and the
points Ri (range of the ith sample) are plotted and according to these points
the decision is made.
FOR KNOWN mean µ and standard deviation σ

d2, D1 and D2 are tabulated in the statistical table

for the different values of n.

Ri is the range of the ith sample and k denotes total

samples of size n selected for study
 np chart
Variable X represents number of defective articles/products observed in the
sample of size n. Here X will be binomially distributed random variable with
parameters “n” and “p,” where p is the probability of getting defective
article produced by the process. NOTE THAT k is different samples, each
of size n.
“p” is known.

Where , q = 1 − p di be the number of defective articles/products

observed in the ith sample.
 p chart
Variable X represents number of defective articles/products observed in the
sample of size n. Here X will be binomially distributed random variable with
parameters “n” and “p,” where p is the probability of getting defective
article produced by the process. k different samples, each of size n
‘p’ is known. ‘p’ is unknown.

di = (number of defective articles)/(products observed in the ith sample)

pi = di /n = proportion of (defective articles)/(products produced by the process). These points pi are
plotted on the p chart and according to these plotted points the decision is made
 C Chart
To draw c-chart, k different sample units are selected from the production
process and are inspected. Let ci be the number of defects observed in the ith
sample unit produced by the process. These points cj are plotted on the
c-chart and according to these plotted points the decision is made.
Variable X = C represents number of defects observed per article/product.
Here X will follow Poisson probability distribution with parameter “λ” where
λ is the average number of defects per article produced by the process.
“λ” is known.
Control Charts
19
20
21
Control Chart

22
Pattern in the Control Chart

23
24
25
26
27
28
 Control Chart
■

29
 Control Chart
■

30
 Control Chart
■

31
 Sample no. 1 2 3 4 5 6 7 8
Avg. linear density (tex) 20 20 21 19 22 22 20 20
Range 1.2 2.1 1.6 1.8 2 1.7 2 1.8

32
UCL = 21.4992

CL = 20.475

LCL = 19.4508

33
 Control Chart
Range ‘R’ control chart:
■ This type of chart demonstrates the variability within a process. It is
suited to processes where the sample sizes are relatively small, for
example <10.
■ Sets of sample data are recorded from a process for the particular quality
characteristic being monitored.
■ For each set of date the difference between the smallest and largest
readings are recorded.
■ This is the range ‘R’ of the set of data. The ranges are now recorded onto
a control chart.
■ The center line is the averages of all the ranges.
34
 Control Chart
Q. A sample of size 4 was taken on each of 10 days from a machine
making sweater blanks, whose lengths should be 57cm. The following
length measurements were obtained. Use these data to calculate control
limits for a control chart for averages and ranges of samples of size 4.
DAY 1 2 3 4 5 6 7 8 9 10
57.7 57.2 57.7 57.9 56.1 56.4 56.1 55.4 56.9 58.2
LENGTHS OF
57.4 57.9 55.8 56.1 55.6 57.7 57.4 56.6 55.6 58.4
BLANKS
56.9 56.1 55.6 55.4 56.6 58.2 58.2 57.7 57.7 56.1
(CM)
56.1 55.9 56.6 55.4 57.2 56.1 58.4 56.9 56.4 57.2

SAMPLE AVERAGE
56.95 56.775 56.425 56.2 56.3 57.1 57.525 56.65 56.65 57.475
X

SAMPLE RANGE
1.6 1.8 2.1 2.5 1.6 2.1 2.3 2.3 2.1 2.3
R
35
 Control Chart
■

36
 Control Chart
■

37
 Control Chart
Standard Deviation ‘S’ control chart:
■ The ‘S’ chart can be applied when monitoring variable data. It is suited to
situations where there are large numbers of samples being recorded.
■ The ‘S’ relates to the standard deviation within the sample sets and is a
better indication of variation within a large set versus the range
calculation.
■ An advantage of using the standard deviation is that all data within a set
are utilized to determine the variation, rather than just the minimum and
maximum values.

38
 Control Chart
■

39
 Figure 6.5 - Control charts for Example 6-2.
Similarly, the Z-value at the lower control limit is - 3.00.
For these Z-values each tail area is found to be 0.0013.
The probability of a type I error, as shown by the shaded tail areas in Figure 6-5a, is
therefore 0.0026.
(c) What is the probability of detecting the shift by the second sample plotted after the
shift if the samples are chosen independently?

Solution –
Probability of detecting the shift by the second sample is -
P(detecting shift on sample 1) + P(not detecting shift in sample 1 and detecting shift in
sample 2)

First probability was found in part (b) to be 0.0548.

Second probability = (1 - 0.0548)(0.0548) = 0.0518

Total probability is 0.0548 + 0.0518 = 0.1066.

Thus, there is a 10.66% chance of detecting a shift in the process by the second sample.
 Control Chart
Attribute Control Charts:
■ Attribute control charts are utilized when monitoring count data.
■ There are two categories of count data, namely data which arises from
“pass/fail” type measurements, and data which arises where a count in the
form of 1,2,3,4,…. arises.
■ Depending on which form of data is being recorded, differing forms of
control charts should be applied.
1. ‘p’ and ‘np’ control charts.
2. ‘u’ and ‘c’ control charts.

44
 Control Chart
‘P’ and ‘nP’ control charts:
■ P charts are utilized where there is a pass / fail determination on a unit
inspected.
■ The p chart will show if the proportion defective within a process
changes over the sampling period (the p indicates the portion of
successes).
■ In the p chart the sample size can vary over time. A similar chart to the p
chart is the np chart.
■ However, with the np chart the sample size needs to stay constant over
the sampling period.

45
 Control Chart
■

46
 Control Chart
■

47
Control Chart

48
 Control Chart
■

Sample No. No. of non-conforming Fractional non-conforming

1 12 0.24
2 15 0.30
………. ………. ……….
30 6 0.12
Total = 347

49
 Control Chart
■

50
 Control Chart
■
Sample No. Size Non-conforming Fractional non-conforming
1 100 12 0.12
2 …… 15 ………
……… …… ……… ………
25 90 6 0.16
∑=2450 ∑=234

51
 Procedures with variable sample size
■

52
Control limits based on Average Sample Size

53
■

Sample
no. (i)
1 100 12 0.120 0.029 0.83
2 90 9 0.100 0.031 0.13

Known through experiment

54
The Standardized Control Chart:
The third approach to dealing with variable size is to use a Standardized control
charts, where the points plotted in standard deviation units. Such a control chart
has the centre line at zero, and upper and lower control limits of +3 and -3,
respectively.

55
 Control Chart
‘U’ and ‘C’ control charts:
■ The ‘U’ and ‘C’ control charts are applied when monitoring and
controlling count data in the form of 1,2,3, …. i.e. specific numbers.
■ An example of such data is the number of defects in a batch of raw
material, or the number of defects identified within a finished product.
■ The C chart is used where there can be a number of defects per sample
unit and the number of samples per sampling period remains constant.
■ In the U chart, again similar to the c chart, the number of defects per
sample unit can be recorded, however, with the u chart, the number of
samples per sampling period may vary.

56
■

57
■

58
 Control Chart
■ Sample No. Non-conforming
1 21
2 24
……… ………
26 15
∑=516

59
 Control Chart
■
Sample No. Size Non-conforming Average no. (Ui)
1 5 10 2.0
2 5 12 2.4
……… …… ……… ………
20 5 5 1.0
∑=193

60
Q. The number of pills formed for 100 m length of cloth was examined. A sample
of the results is given below. Calculate control limits and draw the control chart.

Fabric Samples Pills Formed

1 16
2 12
3 12.5
4 13
5 14
6 13.5
7 15
8 12.8
9 13.5
10 15.2
61
Q. The number of thin places was counted over a 100 mt length of yarn. 10
samples each of 100 mt were taken for the test. Results of the sample are given
below. Calculate control limit.

1 16
2 12
3 12
4 13
5 14
6 13
7 15
8 12
9 13
10 15

62
QUES1. The no. of defects (oil stains, streaky lines and holes) in 50m length of a woven cloth was seen
and the results are given below. Draw control charts for each of these properties and check whether there
is an indication of any lack of control. Revise the limits if necessary.
SAMPLE NO. OIL STAINS STREAKY LINES HOLES
1 8 4 1
2 6 3 2
3 4 2 1
4 3 6 3
5 9 5 1
6 10 8 4
7 2 3 2
8 8 4 3
9 4 9 5
10 9 8 2
11 6 6 3
12 10 3 1
13 13 4 5
14 8 2 4
15 4 1 3
16 12 2 6
17 2 1 4
18 1 3 3
19 12 4 1
20 1 5 2
TOTAL 132 83 56
MEANS (μC) 6.6 4.15 2.8 63
1. CONTROL LIMITS FOR OIL STAINS

√ C = √6.6 = 2.56

Action Limits
C ± 3.09 √ C

6.6 ± 3.09 √6.6 = 6.6 ± 3.09 *2.56

6.6 ± 7.9104
(14.51, 0)

Warning Limits
C ± 1.96 √ C

6.6 ± 1.96 √6.6 = 6.6 ± 5.01

(11.61, 1.59)
64
65
2. CONTROL LIMITS FOR STREAKY LINES

Action Limits
C ± 3.09 √ C

4.15 ± 3.09 √4.15 = 4.15 ± 6.29

(0 , 10.44)

Warning Limits
C ± 1.96 √ C

4.15 ± 1.96 √4.15 = 4.15 ± 3.992

(0.15, 8.142)
66
67
3 CONTROL LIMITS FOR HOLES

Action Limits
C ± 3.09 √ C

2.8 ± 3.09 √2.8 = 2.8 ± 5.17

(0 , 7.97)

Warning Limits
C ± 1.96 √ C

2.8 ± 1.96 √2.8 = 2.8 ± 3.27

(0, 6.07)
68
Conclusion
The three control charts are shown in the figure and it is found that there are no
indications of any lack of control during the process
69
 Procedures with variable sample size
■
No. of
Roll Number of Total no. of No. of inspection
nonconformities per
No. square meters non-conformities units in Roll, n
inspection unit
1 500 14 10.0 1.40
2 400 12 8.00 1.50
… …… ……… ……… ………
10 625 23 12.5 1.84
∑=153 ∑= 107.50

70
71
 Procedures with variable sample size
■

Based on earlier sample,

with LCL = -3 and UCL = +3

72
Process Capability

73
74
 Control Chart
Process Capability Analysis:
■ A process capability study uses data from an initial run of parts to
predict whether a manufacturing process can repeatably produce parts
that meet specifications.
■ Process Capability measures how consistently a manufacturing process
can produce parts within specifications.
■ The basic idea is very simple. You want your manufacturing process to:
(1) be centered over the Nominal desired by the design engineer, and
(2) with a spread narrower than the specification width.
75
 Control Chart
Cp measures whether the process spread is narrower than the specification
width (Potential Process Capability)
Cpk measures both the centering of the process as well as the spread of the
process relative to the specification width (Actual Process Capability)
Cp = (Specification Width)/(Process Width)
Cpk = (Distance from Mean to nearest spec Limit)/(Distance from Mean
to Process Edge)

76
• Definition: “ Process Capability is the measured, inherent variation of the product
turned out by a process”

• What is Process: To some unique combination of machines, tools, methods,

materials & people engaged in production.
• Capability: An ability, based on tested performance, to achieve measurable results
• Process Capability = +3σ or -3σ ( a total of 6σ) Where, σ shows the standard
deviation of Process under a state of statistical control.
• Cp: process capability index
• Cpk: minimum process capability index
• Pp: process performance index
• Ppk: minimum process performance index

Cont..
 Calculations

USL: upper specification limit;

LSL: lower specification limit;
Mean: grand average of all the data
Sigma hat: estimated inherent variability (noise) of a stable process
SD: overall variability
79
80
 Interpretations
(1) When Cpk=Cp then the process is centered at the midpoint of the specifications.
(2) When Cpk < Cp then the process is running off center.
(3) When Cpk=0, the process mean is exactly equal to one of the specification limits.
(4) When Cpk<0 then the process mean lies outside the specification limit.
(5) When Cpk <-1 then the entire process lies outside the specification limits.

81
Cpk is:
• Process Capability measure
• Simple statistical measure estimate🡪 level 🡪 process output which will within specified limit
• Provides comparison between output of process Vs Process Specification
• Process Improvement:
Statistical Process Control tool

monitors process

tells whether capable or not meeting desired level of performance

action to be taken

To investigate concerns

Helps in process improvement and to achieve desired capability levels

Cont..
 Cp, Cpk Vs Pp, Ppk

Cp Cpk Pp Ppk
Process Capability Process Performance
Aims
Process Verification

Cp- Potential Capability -For Process Performance

What process can do under certain -Process in too new (At development
condition stage)
i.e. variation in short run for process - No historical Data
in state of statistical control - Sample size is larger from process
- Usage sample sigma for calculation
Cpk- Actual Capability - Cpk > Ppk
Estimate of capability what process is - Anomalies in case of: sample size is
doing over extended period of time small or data represents short
-Usage process Sigma for calculation amount of time only in that cases
-K stands for centralizing factor Cpk>Ppk is not true Cont..
 Relationship between Process
variability and Specification width
• Three possible ranges for Cp
– Cp = 1, as in Fig. (a), process
variability just meets
specifications

– Cp ≤ 1, as in Fig. (b), process not

capable of producing within
specifications

– Cp ≥ 1, as in Fig. (c), process

exceeds minimal specifications

• One shortcoming, Cp assumes that the

process is centered on the
specification range

• Cp=Cpk when process is centered

Process Capability analysis Process:
- Take representative sample of process output
- Statistical analysis of samples (Mathematical tools, Scattered Plot, Pareto Chart,
Histogram, (Minitab/Excel/QI Macros for exel)
- Calculate mean and Standard Deviation form the sample size
- Calculate Cp Value as well as Cpkl and Cpku
- Minimum value from Cpkl and Cpku is the value of Cpk
- By observing results of statistical analysis one can be explain or determine future
expected process capabilities (ex. In response to Productivity or Quality Attribute)
- Process Capability provides a single number which has ability to provide details of
process consistency output.
Requirement:
- Stable Process
 Interpretation of Values

Sigma Defect Rate Yield Cpk Sigma Defect Rate Yield Cpk
Level (DPMO) %Goods Level (DPMO) %Goods
1σ 691462 30.9 0.33 4σ 6210 99.40% 1.33
2σ 308770 69.10% 0.67 5σ 233 99.98% 1.67
3σ 66811 93.33% 1 6σ 3.44 99.99967% 2.0
Cpk Value Ranges
Process Capability Ratio for an Off-center Process
LSL USL
σ=2

Cp = 2.0
A Cpk = 2.0

38 44 50 56 62

LSL USL
σ=2

Cp = 2.0
B Cpk = 1.5

38 44 50 53 56 62
 LSL σ=2 USL
Cp = 2.0
C Cpk = 1.0

38 44 50 56 62

LSL USL
σ=2
Cp = 2.0
D
Cpk = 0

38 44 50 56 62

LSL USL
σ=2
Cp = 2.0
E Cpk = - 0.5

38 44 50 56 62 65
90
91
92
93
94
95
96
Recommended Minimum Values for the Process Capability Indices
Now suppose only LSL is given then,

Here also CPL < 1, which is undesirable

 What if process not capable
-Initial action - increase the inspection level and ensure that confidence with respect to
the quality of output product is increased.
-Clearly, quality cannot be inspected into a product or process, therefore, the net steps
will be to look at how to improve the capability of the process.
-Reviewing the product specifications, as by widening the specifications, the capability
can be increased. (This can only be performed, if any proposed specification changes are
acceptable per customer needs.)
-Then looking at the process/Actual Operations itself, there will be a need to identify
the sources of variation (Ex. Fish Bone i.e. 6M_Man, Material, Machine, Method,
Measurement, Mother Nature)
What if process not capable
What if process not capable
 Advantages and
1.) Product/Process Design and application
Understanding Stage:
-According to experimental design change input material attributes and process
parameter
-Allows to Identify special causes and develop control strategy to eliminate or reduce
variability
2. Process Scale up and Qualification Stage:
-To establish scientific evidence that the process is reproducible at commercial scale
and the process will consistently deliver a product that meets the quality standard
established in development stage.
Challenges:
-Limited commercial scale batches
-Higher level of sampling and testing to demonstrate product quality
Once sufficient data points are collected,
– Evaluate if the process reaches a stable state
– Establish control limits for control chart
 Advantages and
application
3. Routine Commercial Manufacturing Stage –
-Continual assurance that the process remains in a state of control (the validated
state) during commercial manufacture
-Traditional sampling and testing to confirm product quality
-Statistical process control tools to monitor the process
-Cpk/Ppk monitoring and trending data of process batches
105
106
Measure of Process Capability: Cpm

107
Measure of Process Capability: Cpmk

108
Comparison of Capability Indices
While Cp measures process potential, Cpk measures the actual process yield.
When a process is exactly centered between the specification limits, Cp = Cpk.

In terms of Cp

Consider figure 9.7,

USL = 62 mm, LSL = 38 mm, and T = 50 mm;
Three processes, A, B, and C, with different means and standard deviations, are shown.
Effect of process parameters on capability indices
Process Capability Indices for Processes with Various Parameters

From Table we can see that for processes B and C, whose means deviate from the target by
different amounts, the values of Cpk are the same. This is because Cpk is influenced by both the
process mean and the process standard deviation.
 Control Chart
Modified Control Chart (based upon α-error):

Z δσ µL X µU Z δσ
σx = (σ/√ (n))

Zασ/√(n) Zασ/√(n)

112
 Control Chart
■

113
 Control Chart
Practice Example (Modified Control Chart):
Q. LSL = 8, USL = 32, process followed is 6σ. What will be the Control
limit, if n = 4 and 6σ process can differ by 1.5σ.
Zδ = 4.5

σ=2

4.5σ 1.5σ 1.5σ 4.5σ

X
µL = 17 µU = 23
114
 Control Chart
Sol. Total process width = 6σ + 6σ = 12σ
USL – LSL = 24 => σ = 24/12 = 2 & Zδ = 6 – 1.5 = 4.5
Therefore, LCL = 8 + (4.5 – 3/√4) x 2 = 14
UCL = 32 – (4.5 – 3/√4) x 2 = 26
σx = (σ/√ (n)) = 1

14 Zασ/√(n) = 3 20 Zασ/√(n) = 3 26
µL = 17 µU = 23
115
 Control Chart
Modified Control Chart (based upon β-error):

Z λσ µL X µU Z λσ
σx = (σ/√ (n))

β-error β-error

Zβσ/√(n) Zβσ/√(n)

116
 Control Chart
■

117
 CUSUM Chart
■

118
 CUSUM Chart
■

119
 CUSUM Chart
Advantages of CUSUM chart
■ CUSUM chart is the best way to detect the small shifts of process mean
especially 0.5 to 2 SD from the target mean
■ It is easy to identify visually the shifts in process mean

Disadvantages of CUSUM chart

■ Establishing and maintaining of CUSUM charts are more difficult
■ CUSUM charts are slower in detect large process mean shift
■ Since CUSUMs are correlated, it is tough to interpret the patterns.

120
 EWMA Chart
■ Exponentially Weighted Moving Average chart is used in statistical
process control to monitor variables (or attributes that act like variables)
that make use of the entire history of a given output. This is different
from other control charts that tend to treat each data point individually.
■ Each output (previous sample mean) is given a weighting by the user.
The most recent samples are weighted the highest. This means the oldest
data is given the least amount of weight. The chart displays the
data geometrically. This gives the advantage of the chart not being
greatly affected when a small or large value enters the calculation.
■ The EWMA chart will detect shifts of .5 sigma to 2 sigma much faster
than Shewhart charts with the same sample size. They are, however,
slower in detecting large shifts in the process mean.
121
 EWMA Chart
■

122
 EWMA Chart
Examples of Uses of EWMA Control Charts
■ Detecting drift that is caused by tool wear.
◻ Ex. Manufacturer produces a widget of a certain diameter. If the
diameters of the widget are off, there are consequences. Measuring
using an EWMA Chart helps understand the manufacturing machine
wear and it’s impact to the creation of the widgets.
■ Accounting processes
◻ Ex. Day-to-day fluctuations in accounting processes may be large but
may not necessarily mean the process is unstable. The choice of
lambda can be determined to make the chart more or less sensitive to
these daily fluctuations.
123
 EWMA Chart
■ Chemical processes
■ Website visitors that fluctuate depending on the day of the week.
◻ Ex. This website gets far more visitors when people are at work
Monday through Thursday compared to even Friday, the weekends or
during holidays.

Important Notes on EWMA Charts

■ Your data must be time-ordered.
■ Consecutive points have the highest chance of being alike – so default to
a range of 2 when possible.
■ Created for normal data but is robust enough for non-normal data sets.
124
Thank you

125