Statistical

Process
Control
Trainer: Amit Ballal
Reference: AIAG SPC 2 Edition manual
nd
 At the end of this training, you will learn:

• Understand basics of statistics

• Understand variation, state of statistical control Vs process control
• Identify correct control chart to be used
• Identify causes of variation and take action to reduce variation
• Apply SPC techniques in practical and real life

2
 Statistics

Estimation
about

Sample Population
3
 Symbols used
Symbols used for samples & Population
Parameter Sample Population

Sample size n N

Mean (Average) X μ
Standard Deviation
s σ
(Sigma)
Variance s² σ²

Range R R

Mode ------- -------

Median M or m or -------

4
 Process

Who is involved ?
With what What skills /
resources ? trainings are
required ?

Input Process Output

Method Measures (How effectiveness &

(Work Procedure / Efficiency of process will be
Instructions/ SOP, etc.) measured) 5
6
 Control

Variation is only due to

common causes
(No special causes)
 Dr. Walter A. Shewhart
Born: Died:
18 March 1891 11 March 1967

Summary:
American physicist, engineer and statistician who worked on statistical quality control.
Dr. Walter A. Shewhart
Achievements:
Awarded to Deming in 1955

“Shewhart medal” was named Awarded to Kaoru Ishikawa in 1982

after him which was started in
1948 by ASQ Awarded to Dorian Shainin in 1989

Invented control charts at Bell

telephones

8
Books by Dr. Walter A. Shewhart
A study of the accelerated motion of
small drops through a viscous
medium

Correction of Data for Errors of

Averages: A Discussion of a Method for
Correcting Data for Errors of Averages
Obtained from Small Samples
10
 Common Vs Special causes
Common cause Special cause

 Occurs regularly / part of the process  Doesn’t occur regularly / Not part of the process

 Difficult to identify  Easy to identify

au
 Can be reduced at some extent (that too
se
 Can be eliminated completely
with lots of effort / process re-design),
cannot be eliminated

11
 I-MR (X-MR) Chart

• I = Individual
• MR = Moving Range

Where to use ?
• Sample (Subgroup) size = 1

12
Sr. Reading MR (Moving Range)
No.
1 290

2 288 2

3 285 3

4 290 5

5 291 1

6 287 4

7 284 3

8 290 6

9 290 0

10 288 2

X = 288.3 MR = 2.89
13
 Setting Control Limits
I Chart
UCLx = x + E2MR LCLx = x - E2MR

MR Chart
UCLMR = D4.MR LCLMR = D3.MR

14
 Setting Control Limits
Process average x = 288.3 mm
Average range MR = 2.89 mm
Sample(subgroup) size n = 1

UCLx = x + E2.MR UCL = 295.99

= 288.3 + (2.66)(2.89)
= 288.3 + 7.69
= 295.99 mm
Mean = 288.3

LCLx = x - E2.MR LCL = 280.61

= 288.3 – 7.69
= 280.61 mm
X Chart
15
 Setting Control Limits
Process average x = 288.3 mm
Average range MR = 2.89 mm
Sample(subgroup) size n = 1

UCLMR = D4.MR UCL = 9.44

= (3.267)(2.89)
= 0.429 mm
Mean = 2.89

LCLMR = D3.MR LCL = 0

= (0)(2.89)
= 0 mm

MR Chart
16
 Change event log

OCS Subgroup No. Event log

1 New setup, insert IC27, material lot #1984

4 New setup, insert IC27, material lot #1991

6 New setup, insert IC27, material lot #1997

* 7 material lot #2002, bad insert replaced

8 material lot #2102

11 material lot #2112

14 material lot #2120

 Setting control limits

Control Chart
for sample of Variation due
Out of to assignable
9 boxes control causes

17 = UCL

Variation due to
16 = Mean natural causes

15 = LCL

Variation due
| | | | | | | | | | | |
to assignable
1 2 3 4 5 6 7 8 9 10 11 12 Out of causes
Sample number control

18
 Range / variation chart

(b)
These
sampling (Sampling mean
distributions is constant but
result in the dispersion is
charts below increasing)

UCL
(x-chart does not
x-chart detect the
increase in
LCL dispersion)
UCL
(R-chart detects
R-chart increase in
dispersion)
LCL

19
 Average chart

(a)
These (Sampling mean is
sampling shifting upward but
distributions range is consistent)
result in the
charts below

UCL
(x-chart
x-chart detects shift in
central
LCL tendency)
UCL
(R-chart does not
R-chart detect change in
mean)
LCL

20
 Patterns in control charts

UCL (Upper control limit)

+3σ (UCL)
+2σ

+1σ
X
X

-1σ

-2σ

LCL (Lower control limit) -3σ (LCL)

Normal behavior. Process is
“in control.”

21
22
 Patterns in Control Charts

Upper control limit

Target
Rule-1

Lower control limit

Any point out of control limit

23
Reference: AIAG SPC 2nd edition manual
 Potential causes

Man Machine Method

Lack of Not following

Machine
Training SOP
lubrication
Process
Machine
Rule-1

Lack of parameters
Lack Of Bearing clearance
Competence not
Experience wear maintained

Material
properties Hardness of
Temperature material
Humidity
Thickness of
material

Environment Material 24
 Patterns in Control Charts

Upper control limit

Target
Rule-2

Lower control limit

7 continuous points on one

side of centre line

25
Reference: AIAG SPC 2nd edition manual
 Potential causes

Shifting
of
stopper

Change in
Rule-2

tool- Such
as

Change
in pin
location
 Patterns in Control Charts

Upper control limit

Target
Rule-3

Lower control limit

6 continuous points in
increasing order

27
Reference: AIAG SPC 2nd edition manual
 Patterns in Control Charts

Upper control limit

Rule-3 (Continued….)

Target

Lower control limit

6 continuous points in
decreasing order

28
Reference: AIAG SPC 2nd edition manual
 Potential causes

Tool wear
Rule-3
 Patterns in Control Charts

Upper control limit +3σ (UCL)

+2σ
Potential causes
+1σ

Target X
Rule-4

-1σ

-2σ

Lower control limit -3σ (LCL)

14 points in row alternating

up and down

30
Reference: AIAG SPC 2nd edition manual
 Patterns in Control Charts

Upper control limit +3σ (UCL)

+2σ
+1σ

Target X
Rule-5

-1σ

-2σ

Lower control limit -3σ (LCL)

2 out of 3 points >2 std dev

from centreline (Same side)

31
Reference: AIAG SPC 2nd edition manual
 Patterns in Control Charts

Upper control limit +3σ (UCL)

+2σ
Potential causes
+1σ

Target X
Rule-6

-1σ

-2σ

Lower control limit -3σ (LCL)

4 out of 5 points beyond 1σ

limit on same side of
centreline

32
Reference: AIAG SPC 2nd edition manual
 Patterns in Control Charts

Upper control limit +3σ (UCL)

+2σ
Potential causes
+1σ

Target X
Rule-7

-1σ

-2σ

Lower control limit -3σ (LCL)

15 points in row within 1

standard deviation of Centre
line

33
Reference: AIAG SPC 2nd edition manual
 Patterns in Control Charts

Upper control limit +3σ (UCL)

+2σ
Potential causes
+1σ

Target X
Rule-8

-1σ

-2σ

Lower control limit -3σ (LCL)

8 points in row > 1 std. dev

from centreline (either side)

34
Reference: AIAG SPC 2nd edition manual
 Summary of “Control chart” rules
1 1 point more than 3s limits

2 7 points in row in same side of centreline

3 6 points in row all increasing or decreasing

4 14 points in row alternating up and down

5 2 out of 3 points >2 std dev from centreline (Same side)

6 4 out of 5 points >1 std dev from centreline (Same side)

7 15 points in row within 1 std. dev of centreline (Either side)

8 8 points in row > 1 std. dev from centreline (Either side)

36
 Which control chart shows special cause? (1/3)
+3σ (UCL)
+2σ
Rule 2 +1σ

-1σ
7 continuous points on one

-2σ

-3σ (LCL)
side of center line

+3σ (UCL)

+2σ

+1σ

-1σ

-2σ

-3σ (LCL)
37
 Which control chart shows special cause? (2/3)
+3σ (UCL)
+2σ
Rule 3
+1σ

-1σ

-2σ

-3σ (LCL)

+3σ (UCL)

+2σ

+1σ

-1σ

-2σ

-3σ (LCL)
38
 Which control chart shows special cause? (3/3)
+3σ (UCL)
+2σ

+1σ

-1σ

-2σ

-3σ (LCL)
Rule 5
+3σ (UCL)
+2σ

+1σ

-1σ

-2σ

-3σ (LCL)
39
Tampering (Over adjustment)

40
Data has no meaning
apart from its context

- Dr.Walter Shewhart
 Statistical control Vs Process Control
UCL=76.2 UCL=62.2

Target Target

LCL=74.5 LCL=57.3

Weight of A Weight of B

42
43
 Control Chart Selection
Quality Characteristic
Variable Attribute
defective defect
no
n>1? x and MR
yes constant
yes constant
sample p or sampling
size? np unit?
n>=10 or no
x and R
computer? yes no
no
yes
p-chart with c u
x and s variable sample
size
44
 Control chart differences & Application
Type of Control chart
Attribute
Variable
Defectives Defects

Median Chart (With

Parameter Xbar-R chart Xbar- s chart I-MR (X-MR) charts P np chart C chart U chart
R chart)

Sample size is Sample size is Sample size is Sample size is

Use Subgroup size < 9 Subgroup size > 9 Any subgroup size Subgroup size = 1
varying constant constant varying

Xbar chart & R Xbar chart & s Median chart & R chart Individual chart & Moving
Only p chart is Only np chart is Only c chart is Only u chart is
Charting chart both to be chart both to be both to be used range chart to be used
used used used used
used together used together together together

1. Individual readings
Charting s (Of subgroup) is are plotted in place of
Median is plotted in
(Comparison with ------- plotted in place of Xbar ------- ------- ------- -------
place of Xbar
Xbar-R chart) R 2. MR is plotted in place
of R

1. X dbar (For 1. X dbar (For 1. Median of medians

1. Xbar (For individual Or
Charting Xbar chart) Xbar chart) (For median chart)
X chart) p bar np bar c bar u bar
(Centre line) 2. Rbar (For R 2. s bar (For s 2. Median of R (For R
2. MR bar (For MR chart)
chart) chart) chart)

45
46
 Process / Machine
Part No: 1234 Part name XYZ ABC Characteristics 8± 2 mm M1TR-01
Operation no:

Subgroup # 1 2 3 4 5 6 7 8 9
Date, 12-8-
Time 13-8-22 14-8-22 16-8-22 17-8-22 18-8-22 19-8-22 20-8-22 21-8-22
122

1 9 11 10 10 8 10 8 8 10

2 8 10 9 9 10 9 7 6 8

3 8 11 8 8 11 8 8 9 9

4 9 8 8 10 8 10 10 10 8

5 10 10 9 8 9 8 9 9 9

X bar 8.8 10 8.8 9 9.2 9 8.4 8.4 8.8

47
R 2 3 2 2 3 2 3 4 2
 Xbar-R chart
Where,
X Chart R =average range of the samples

UCL: X + A2.R A2 =control chart factor found in Table 1.1

LCL: X - A2.R X =mean of the sample means

Where,
Range Chart R =average range of the samples

D3 = Control chart factor found in Table 1.1

D4 = Control chart factor found in Table 1.1

48
 Setting Control Limits
Process average x = 8.92 mm
Average range R = 2.56 mm
Sample(subgroup) size n = 5

UCLx = x + A2R UCL = 10.4

= 8.92 + (0.577)(2.56)
= 8.92 + 1.48
= 10.4 mm
Mean = 8.92

LCLx = x - A2R LCL = 7.44

= 8.92 - (0.577)(2.56)
= 8.92 – 1.48
= 7.44 mm X Chart
49
 Setting Control Limits
Process average x = 8.92 mm
Average range R = 2.56 mm
Sample(subgroup) size n = 5

UCLR = D4R UCL = 5.41

= (2.114)(2.56)
= 5.41 mm
Mean = 2.56

LCLR = D3R LCL = 0

= (0)(2.56)
= 0 mm

R Chart
50
51
52
53
 Mean
1, 46, 11, 46, 8 10, 42, 16, 46, 42, 46

1+46+11+46+8=112 10+42+16+46+42+46 = 202

Sum / n = (112/5) = 22.4 Sum / n = (112/5) = 33.67

54
 Mean (For practice)

10, 9, 11, 12, 11 10, 9, 11, 13, 11,12

55
56
 Median
1, 46, 11, 46, 8 10, 42, 16, 46, 42, 46

1, 8, 11, 46, 46 10, 16, 42, 42, 46, 46

11 42+42 = 42
2 57
 Median (For practice)

10, 9, 11, 12, 11 10, 9, 11, 13, 11,12

58
59
 Mode
1, 82, 11, 46, 82 10, 42, 16, 46, 42, 46

1,11, 46, 82, 82 10, 16, 42, 42, 46, 46

82 42, 46
60
 Mode (For practice)

10, 9, 11, 13, 12 9, 9, 11, 12, 11,12

61
Standard Deviation

62
 Standard Deviation
Sr. No.
Readings X-X (X – X) ² Σ(X – X) ²
1 10 10-10 = 0 (0) ² = 0

2 12 12-10 = 2 (2) ² = 4
0+4+1+1+4
3 11 11-10 = 1 (1) ² = 1
= 10
4 9 9-10 = -1 (-1) ² = 1

5 8 8-10 = -2 (-2) ² = 4

Mean s ² = Σ(X – X) ²
(Average): = 10 / 4
10 n-1 = 2.5

n=5 n–1=4 s =
Σ(X – X) ² = √ 2.5
= 1.58 63
n-1
 Standard Deviation (For practice)
Sr. No.
Readings X-X (X – X) ² Σ(X – X) ²
1 21

2 18

3 17

4 23

5 26

Mean s ² = Σ(X – X) ²
(Average): =
n-1
n= n–1= s =
Σ(X – X) ²
= 64
n-1
Normal Distribution
Mean
Median
Mode

Normal Distribution 65
 Normal Distribution

68.26%
95%
99.74%
-3 -2 -1 =0 1 2 3
 Example

Player-A Player-B
Mean 106 62
Median 28 65
Mode 17 67 67
 0

21.92

22.03

22.14

22.24

22.35

22.46
Histogram

22.57

22.68

22.78

22.89

23.00
 Kurtosis

• Where y = the deviation, μ = the mean, and σ = the standard deviation.

69
 Skewness

• If the skew factor is zero, there is no skew.

• Along with letting you compare histograms, the skew factor can tell you if the process has a tendency to lean toward upper or
lower specification limits.

70
Anderson-Darling test
73
 Cp, Cpk
USL- LSL
Cp:
6s

USL- X X- LSL
Cpk: 3s
Or
3s

74
Cp (Process Capability) /
Pp(Process Performance?

Process Variation
75
 Cpk (Process Capability Index) /
Ppk (Process Performance Index)?
Target X
LSL USL

Position of Process (Shift) 76

Can Cpk be Greater than Cp ?

No

77
 Acceptance Criteria of Cp/Pp & Cpk/Ppk

• Cp / Pp: 1.67 minimum

• Cpk / Ppk: 1.67 Minimum

78
 Six sigma converter
Without shift With 1.5σ shift
Sigma(σ) level to PPM/DPMO & Cp PPM/DPMO to Sigma(σ) level & Pp
Sigma(σ) level PPM/DPMO Cp Sigma(σ) level PPM/DPMO Pp
6 3 2.00 5.5 0.0190 1.83
PPM/DPMO to Sigma(σ) level & Cp PPM/DPMO to Sigma(σ) level & Pp
PPM Sigma(σ) level Cp PPM/DPMO Sigma level Pp
233 5.000 1.67 3.4 4.500 1.50
Cp to Sigma(σ) level & PPM/DPMO Pp to Sigma(σ) level & PPM/DPMO
Cp Sigma(σ) level PPM Pp Sigma(σ) level PPM/DPMO
1.6667 5.000 233 1.67 5.010 0.272
Enter in these cells only
 Cp is 2.00 & Cpk is 0.67
What does it indicate & what should be done?

80
 What does Cp (Process Capability) /
Pp(Process Performance) Represent ?
• Lots of Process Variation is present

How to improve Cp/Pp ?

• By reducing variation in the process

• Brainstorm what can be the probable

causes for variation
• Validate actual causes & eliminate the
cause
81
Relation between Process capability and Rejection rate

Cp Rejection rate

1.00 0.270%

1.33 0.007%

1.50 6.8 PPM

2.00 2.0 PPB

82
83
 Defect Vs Defective
Defective Defect
Use of Attribute Control Charts

85
For Defectives

86
 P Chart
(When subgroup size varies)

87
• Subgroup size > 50
• Most sensitive attribute control chart

88
 Formulae

• p = Defectives found in subgroup /

total nos. inspected in subgroup
• P bar = Total defectives found /
total nos inspected

89
 P Chart

Population will be a binomial distribution,

but applying the Central Limit Theorem
allows us to assume a normal distribution
for the sample statistics
90
 p Chart calculations
Sr. No. Inspected / lot Qty. Rejection Qty. Proportion

1 100 2 0.02

2 82 3 0.0365

3 56 5 0.089

4 75 2 0.027

5 87 1 0.011

n (Average) 80 p bar = 0.0325

91
92
93
 np Chart
(When subgroup size is same)

94
Formulae

95
 Setting Control Limits
np Chart

96
 np Chart calculations
Sr. No. Inspected / lot Rejection Qty. Proportion
Qty.

1 100 2 0.02

2 100 3 0.03

3 100 5 0.05

4 100 2 0.02

5 100 1 0.01

n 100 p bar = 0.026

(Constant)
n pbar = pbar * n 2.6

97
For Defects

98
 c Chart
(When subgroup size is same)

99
C chart

100
 c Chart calculations
Sr. No. Inspected / lot Total defects found c
Qty.

1 100 2 0.02

2 100 3 0.03

3 100 5 0.05

4 100 2 0.02

5 100 1 0.01

n 100 c bar = 0.026

(Constant)

101
u chart

102
 u Chart calculations
Sr. No. Inspected / lot Qty. No. of defects u

1 100 2 0.02

2 82 3 0.0365

3 56 5 0.089

4 75 2 0.027

5 87 1 0.011

n (Average) 80 u bar = 0.037

103
104
 Other type of Control Charts
Probability Based Charts
• Stoplight Control (Also known as traffic light control
• Pre-Control
Short Run Control Charts
• Deviation from Nominal
• Standardized X-bar and R Charts
•Standardized Attributes Control Charts
Charts for Detecting Small Changes
• CUSUM Chart (Cumulative Sum)
• EWMA Chart (Exponentially Weighted Moving
Average) 105
 Pre-Control
• Pre-Control is a technique that sets the limits of the red
zones at the upper and lower specifications.
Control Charts Formulas

109
SPC Formulae

110
111
112
113
114
115
116
117