STATISTICAL PROCESS CONTROL

BASIC
D ate:
⌧
⌧-R Ch ar t
⌧
⌧
S E CTION: P RO DUCT: CHA RA CTE RIS TICS :

P E RSO N IN-CHA RGE :

RT

Eve nts:
X- C
x-C HA
art
h

σ
R-Cart
h

Σ
µ
SAMPLE # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
DATE
TI ME
X1
X2
X3
X4
X5
x
R

Presented by:

1.2 1.3 1.4 1.5 1.6 1.7 1.8

Surendra P Tiwari,
Executive Director
Quality Growth Services Pvt. Ltd.
H-13, Kirti Nagar, New Delhi – 110 015
Phone/ Fax: +91-
+91-11
11--25431737 / 25438598
E-mail: [email protected]
URL: www.qgspl.com

AGENDA-- In scope
AGENDA
 Module-1: SPC Introduction and Concept of Control chart:
 Voice of Process & SPC
 Variation: Common and Special Cause
 Basic Statistics: Location, Spread, Shape
 Normal Distribution
 Control chart fundamentals
 Type of data and control charts
 Selection of Control charts
 Control chart preparatory steps
 Module 2: Variable Common Control charts
 X bar & R Chart
 Data Collection
 Limit calculation and validation
 Process capability- Method of calculation
 Cp, Cpk
 Pp, Ppk
 Control chart interpretation rules
 I-MR Chart
 Module-3:Attribute Control Charts Introduction :
 p/ np Chart
 C/U Chart (Appendix dfor further study)

QGS\SPC 2

1
Out of Scope
 Probability based charts and Short Run Charts
 Stoplight Control Chart
 Pre-Control Chart
 DNOM Chart
 Standardized Xbar-R Chart
 Standardized Attribute chart
 I-MR-R Chart
 CUSUM Chart
 EWMA Chart
 Zone Chart
 Process capability: Advanced concept
 Dealing with Non-normal data
 Non-normal Control charts
 Central Limit Theorem
 Data Transformation
 Box-Cox Transformation
 Johnson Transformation
 Non-normal Process Capability

QGS\SPC 3

Section-1

SPC-
SPC- BASICS

QGS\SPC 5

2
 TRADITIONAL PROCESS CONTROL

PRODUCT
THE
or
PROCESS SERVICE

• BASED ON DEFECT DETECTION

• LITTLE OR NO REFERANCE TO THE Inspect
PROCESS
• A GOAL POST MENTALITY Correct Detect
Reject

QGS\SPC 8

Traditional Philosophy
– Goal Post Mentality

LSL USL

2 1 4
5
µ
3
6

Is there a meaningful
difference between
kicks 2 and 6?

QGS\SPC 9

3
 Traditional Philosophy

LSL USL

Anything outside
the specification limits
represents quality losses
Goalpost Mentality
QGS\SPC 10

PROCESS CONTROL: A BETTER APPROACH

M ethod
E nvironment PRODUCT
THE
P eople or
E quipment PROCESS
SERVICE
M aterial

•Listen Voice of Process

Collect

Act Record
Analyze

VOICE OF THE PROCESS

QGS\SPC 11

4
VOICE OF THE
PROCESS
THROUGH SPC

SPC HISTORY
Developed By
Dr. Walter A. Shewhart
During 1920’s in Bell Lab

Since then SPC has evolved to cover

different processes

5
 VARIATION:
The basic principle of SPC

QGS\SPC 14

One Problem With Mother Nature

Everything is Different
VARIABILITY
• No two things are exactly alike…
• No two people are same…
• Temperature changes continuously…
• The products we produce change
continuously…

6
 Common Causes
• Common to all individual readings in
time periods.

A process operating under common

cause is called
under statistical control

Special/ Assignable Causes

• Sudden in nature
• Usually attract the attention of local
people associated with the process.

They are not common to all time periods,

but they can cause process fluctuations
which are large in magnitude

7
 Process Behavior

QGS\SPC 18

How to estimate process behavior?

Shape

Location
Spread
QGS\SPC 19

8
 Processes can differ in;

Location Spread

Shape
Or any combination
of these….
QGS\SPC 20

If only common cause of variation are present. The voice of the

process is stable & predictable and is said to be under statistical
control

Prediction

Prediction?

If special cause of variation are present. The voice of the process is

not stable & predictable and is said to be out of control
QGS\SPC 21

9
Location: MEAN- Centre of Gravity
The Mean of ‘n’ values is the total of the values
divided by ’n’
x1 + x2 + x3.....xn
x=
n
In Standard Mathematical Notation it is
i −n i−N
x=∑ xi
n estimates
xi
N µ =∑
i −1 i −1
The symbol Σ” means ‘sum of’

MEAN: Example
Calculate the mean of the following data set

Trial Result
1 6
2 3
n

∑ (x )
3 7
4 6
i
5 10
101
6
7
7
4 X = i =1
= = 6 .733
8
9
8
9
n 15
10 4
11 8
12 8
13 6
14 7
15 8
sum 101
mean 6.733

10
 Spread-Range
The difference between the largest and the smallest of a
set of numbers. It is designated by a capital “R”
R = X Max − X Min

R = X Hi − X Low
Examples
Data Set: 45 47 49 51 Range is: 54-45=9
46 47 49 52
47 47 50 53
47 48 51 54
Range is not a very powerful statistic to measure dispersion

Spread- Standard Deviation

The average distance between the individual numbers

and the mean.It is designated by “σ”

( x 1 − x ) + ( x 2 − x ) ....( x N − x )
2 2 2
∧
σ =
N

s =
∑( x − x )
i
2
estimates σ = ∑ (x − x)
i
2

n −1 N

11
 Standard Deviation: Example

Determine the standard deviation for the following

sample data set: 1 2 3

( x 1 − x ) + ( x 2 − x ) ....( x n − x )
2 2 2
∧
s =
n-1

∧ ( 1 − 2)2 + ( 2 − 2)2 +( 3-2 ) 2

s =
3-1

Shape: Histogram
Histograms give a graphical view of the
distribution of the values
It reveals the amount of variation that any
process has within it.
5
4
FREQUENCY 3
2
1
64 65 66 67 68 69 70 71 72
HEIGHT(Inches)

12
Normal Distribution Curve-Relation between spread & sigma

σ
-3σ σ -1σ
-2σ σ x σ
+σ σ +3σ
+2σ σ
68.26
95.46
99.73

By collecting sample data from the process

and computing their
• Mean
• Standard deviation and
• Shape
Prediction can be made about the process

QGS\SPC 29

13
For routine process control, we need
• Simple computation
• Easy to use by operators for ongoing process control
• Help the process perform consistently , predictably for quality and
cost
• Achieve
• Less variation in output
• Lower unit cost
• Increase effective capacity
• Provide a common language for discussing process performance
• Distinguish special from common causes of variation , as a guide to
local or management action.
QGS\SPC 30

CONTROL CHARTS
• Transformation of a normal distribution curve in the form of 3 parallel
lines, where
• The middle line indicates mean and called central line (C.L.)
• The upper line indicates Mean +3 Sigma and called upper control limit (UCL)
• The lower line indicates Mean -3 Sigma and called lower control limit (LCL)

U.C.L.
σ
+3σ
C.L.
Mean

σ
-3σ
L.C.L.
QGS\SPC 31

14
 Types of Data
• Attribute • Variable
 Anything that can be  Anything that can be
classified measured
 Either/Or  Height
 Pass/ fail  Dollars
 Good/ Bad  Distance
 Go/No Go  Speed
 Count  Continuous, infinite
 Discrete  Stimulates Prevention
 Detection Oriented

Common Control Charts

Variable
 Average and Range ( X-R )
 Individual and Moving Range (X – MR/I-MR)

Attribute
 p/ np Chart : Unit Nonconforming (Defectives)
 c/u Chart : Number of Nonconformities (Defects)

QGS\SPC 33

15
 Other Control Charts
 Average and Standard Deviation ( X-S )
 Individual and Moving Range (X – MR/I-MR)
 Average-Range-Moving Range (I-R-MR)
 Stoplight chart
 DNOM (Deviation from Nominal) chart
 Z-Chart
 Zone Chart
 Cusum (Cumulative Sum) Chart
 EWMA (Exponentially Weighted Moving Average )Chart
 Multivariate Chart
 Regression chart etc…

QGS\SPC 34

Steps for Control Charts

1. Complete preparatory steps
2. Data Collection
3. Making Trial Control Limits
4. Validation of Control limits
5. Process capability Study
6. On going control
7. Improvement

QGS\SPC 35

16
 Control Chart: Preparatory Steps
 Create a suitable (conducive) environment
 Select characteristics
 Verify Measurement System capability
 Select suitable control chart

QGS\SPC 36

SPC – Charts For All Occasions

%
Variable No Defective or Defects
Data? Defects?

Yes % Defective

No Constant Constant
Rational
I & MR chart Sample Sample
Subgroups?
Size? Size?

No No
Yes Yes Yes

p-chart u-chart
Yes Easy to
Subgroup
Compute
Size >8?
sigma?
np- or p-chart c- or u-chart
No Yes
No

X-bar & R chart X-bar & S chart

QGS\SPC 37

17
 Section-2

VARIABLE CONTROL CHARTS

QGS\SPC 38

CONTROL CHARTS (Variable)

Average--Range ( X-R )Chart
Average
 When ?
Measurement must be variable
Situation must be practically feasible to have at
least 2 measurements in short span.
Mass production
Suitable for Product (Output) characteristics
Suitable for both normal and non-normal data

QGS\SPC 39

18
 Average--Range (
Average X-R )Chart
 Data Collection
Decide the Subgroup Size
 Rational subgroup: Variability within subgroup should be small

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

QGS\SPC 40

CONTROL CHARTS (Variable)

Average--Range ( X-R )Chart
Average
 Data Collection
Decide Subgroup Frequency
 Detect change in the Process over span of time.
 For initial study, may be consecutive or a very short
interval.

QGS\SPC 41

19
 CONTROL CHARTS (Variable)
Average--Range ( X-R )Chart
Average
 Data Collection
Decide No. of Subgroups
(for initial study: to define the control limits)
 To incorporate Major source of variation (Generally 25
subgroups or more containing about 100 individual
measurements)

QGS\SPC 42

Average--Range (
Average X-R )Chart
 Data Collection
 On a data collection sheet, called control chart sheet

QGS\SPC 43

20
AVERAGE RANGE CHART
CUSTOMER PART NAME PART NO.
PARAMETER SPECS.:- NOMINAL MACHINE
TOLERANCE ( + )
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20
01 0.65 0.75 0.75 0.60 0.70 0.60 0.75 0.60 0.65 0.60 0.80 0.85 0.70 0.65 0.90 0.75 0.75 0.75 0.65 0.60
02 0.70 0.85 0.80 0.70 0.75 0.75 0.70 0.70 0.80 0.80 0.70 0.70 0.65 0.60 0.55 0.80 0.65 0.60 0.70 0.85
03 0.65 0.75 0.80 0.70 0.65 0.75 0.65 0.80 0.85 0.60 0.90 0.85 0.75 0.85 0.80 0.75 0.85 0.60 0.85 0.65
04 0.65 0.85 0.70 0.75 0.85 0.85 0.65 0.65 0.75 0.65 0.70 0.60 0.60 0.65 0.65 0.80 0.60 0.65 0.70 0.70
05 0.85 0.65 0.75 0.65 0.80 0.70 0.80 0.75 0.75 0.75 0.65 0.70 0.70 0.60 0.85 0.65 0.80 0.60 0.70 0.65
X-Bar
R
QGS\SPC 44

AVERAGE--RANGE CONTROL CHARTS X-R

AVERAGE

 Calculate Average of each Subgroup

X = ( X1 + X2 + … + Xn )/ n

 Calculate Range of each Subgroup

R= Xmax. - Xmin.

 X1, X2,…., Xn are individual values within the subgroup

 n is the Subgroup Sample Size.

QGS\SPC 45

21
 AVERAGE--RANGE CONTROL CHARTS
AVERAGE X-R

CUSTOMER PART NAME PART NO.

PARAMETER SPECS.:- NOMINAL MACHINE
TOLERANCE ( + )
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19
0.65 0.75 0.75 0.60 0.70 0.60 0.75 0.60 0.65 0.60 0.80 0.85 0.70 0.65 0.90 0.75 0.75 0.75 0.65
0.70 0.85 0.80 0.70 0.75 0.75 0.70 0.70 0.80 0.80 0.70 0.70 0.65 0.60 0.55 0.80 0.65 0.60 0.70
0.65 0.75 0.80 0.70 0.65 0.75 0.65 0.80 0.85 0.60 0.90 0.85 0.75 0.85 0.80 0.75 0.85 0.60 0.85
0.65 0.85 0.70 0.75 0.85 0.85 0.65 0.65 0.75 0.65 0.70 0.60 0.60 0.65 0.65 0.80 0.60 0.65 0.70
0.85 0.65 0.75 0.65 0.80 0.70 0.80 0.75 0.75 0.75 0.65 0.70 0.70 0.60 0.85 0.65 0.80 0.60 0.70
0.70 0.77 0.76 0.68 0.75 0.73 0.71 0.70 0.76 0.68 0.75 0.74 0.68 0.67 0.75 0.75 0.73 0.64 0.72
0.20 0.20 0.10 0.15 0.20 0.25 0.15 0.20 0.20 0.20 0.25 0.25 0.15 0.25 0.35 0.15 0.25 0.15 0.20

QGS\SPC 46

AVERAGE--RANGE CONTROL CHARTS X-R

AVERAGE

 Calculate Process average (Overall average)

X =(X1 + X2 + … + Xk)/ k

 Calculate Average Range

R = (R1 + R2 + … + Rk )/ k

 k = No. of Subgroups

QGS\SPC 47

22
AVERAGE--RANGE CONTROL CHARTS
AVERAGE

 Calculate Trial Control Limits for Range Chart

UCLR = D4 R
LCLR = D3 R.
 Calculate Trial Control Limits for Average Chart
UCLX = X + A2 R
LCLX = X - A2 R

D4, D3 and A2 are constant varying as per sample size (n).

QGS\SPC 48

TABLE OF CONSTANTS FOR CONTROL

CHARTS

Subgroup
A2 d2 D3 D4
Size (n)
2 1.880 1.128 - 3.267
3 1.023 1.693 - 2.574
4 0.729 2.059 - 2.282
5 0.577 2.326 - 2.114
6 0.483 2.534 - 2.004
7 0.419 2.704 0.076 1.924
8 0.373 2.847 0.136 1.864
9 0.337 2.970 0.184 1.816

QGS\SPC 49

23
AVERAGE RANGE CHART
X- DoubleBar 0.718
R-Bar 0.21
FOR R-Chart
UCL D4×R-Bar 0.433
LCL D3×R-Bar 0.000
FOR X-Bar Chart
UCL X-DoubleBar+A2×R-Bar 0.837
LCL X-DoubleBar-A2×R-Bar 0.5991
QGS\SPC 50

DRAW AVERAGE-
AVERAGE- RANGE CHART
X-Bar Chart

0.80

0.70

0.60

0.50

R-Chart

0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

UCL LCL

QGS\SPC 51

24
AVERAGE-- RANGE CHART : Minitab output
AVERAGE
Xba r-R C hart of X-1 , ..., X-5
U C L= 0.8 342
0 .80
Sample M ean

0 .75
_
_
X= 0.7 18
0 .70

0 .65

0 .60 LC L= 0 .601 8
1 3 5 7 9 11 13 15 17 19
S a m p le

U C L= 0.4 258
0.4
Sample Range

0.3
_
0.2 R = 0. 201 4

0.1

0.0 LC L= 0
1 3 5 7 9 11 13 15 17 19
S a m p le

QGS\SPC 52

Exercise: DRAW AVERAGE-

AVERAGE- RANGE CHART

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

1 27 30 21 40 51 33 12 35 42 22 34 32 34 28 44 26 38 26 42 30 23 28 25 30 38

2 23 17 44 21 34 30 18 48 15 50 22 48 32 30 32 42 40 28 38 32 44 34 29 38 27

3 24 27 22 29 17 28 22 20 34 45 36 32 28 17 22 35 51 34 52 39 48 39 40 44 39

4 36 32 28 24 10 22 30 47 20 41 44 33 38 23 41 28 32 39 36 45 33 44 33 32 22

QGS\SPC 53

25
 QGS\SPC 54

Xbar-R Chart of data

50
U C L=45.97
Sample Mean

40

_
_
X=32.44
30

20 LC L=18.91
M ean 26.5 28.5 28.25 37.5 39.5 36.25 24.5 32.75 31.75 36.5 36.25 36
Range 15 19 11 28 28 16 13 16 13 15 16 14

U C L=42.36
40
Sample Range

30

_
20 R=18.57

10

0 LC L=0
M ean 26.5 28.5 28.25 37.5 39.5 36.25 24.5 32.75 31.75 36.5 36.25 36
Range 15 19 11 28 28 16 13 16 13 15 16 14

QGS\SPC 55

26
CONTROL CHART: Validation of Control Limit
Control limits should indicate the variation which comes due
to common causes only. So that, any assignable cause
variation is reflected.

Hence, it should be based on data when there is no assignable

cause.

Any control limit based on assignable cause data can not be

considered as Reliable

What to do ?
QGS\SPC 56

Validation of Control Limit

Identify any out of control or special cause situation
 Any point above UCL or below LCL:

R CHART

60
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

QGS\SPC 57

27
Validation of Control Limits
 For initial control chart,
 Discard all the subgroup showing out of control situation (starting from R-Chart)
 Re-Calculate Control Limits, plot and analyse for any out of control situation.
 Re discard if any out of control situation again found.
 Continue the above cycle until all plot indicate a control situation.
 Repeat same exercise with Average chart.
 If more than 50% data are required to be discarded, reject all data and re-
collect.
 Once initial control chart indicates control situation,

 Calculate Initial Capability

 Extend control limits for ongoing control.

QGS\SPC 58

Process Capability Study

 Compare Voice of Process with Voice of Customer
(Specification)
Voice of the Process

-3 σ +3 σ

Process Width

Design Width
LSL T Customer
Voice of the USL
QGS\SPC 59

28
PROCESS CAPABILITY
Subgroup
d2
Size (n)
2 1.128
3 1.693
 Calculate Process Standard Deviation
4 2.059
σ = R/d2 5 2.326
6 2.534
d2 is a constant varying as per sample size (n)
7 2.704
8 2.847
 Calculate Process Capability (Cp) 9 2.970

Cp = (USL - LSL) / 6σ USL = Upper Specification Limit

= Tolerance/ 6σ LSL = Lower Specification Limit

QGS\SPC 60

A Problem With Cp

-5 -4 -3 -2 -1 0 1 2 3 4 5 -5 -4 -3 -2 -1 0 1 2 3 4 5

 Cp considers only spread, not the location.

 For a truly capable process
 Process spread must be smaller to specification and
 It should be located in a manner that its spread on both the sides
falls well within the specification.
 Capability index that considers both location and spread is called Cpk

QGS\SPC 61

29
PROCESS CAPABILITY

Calculate Process Capability (Cpk)

CpU = (USL -X) / 3σ =ZUSL/3
Subgroup
or CpL = (X - LSL) / 3σ =ZLSL/3 Size (n)
d2

2 1.128
Whichever is minimum will be Cpk 3 1.693
4 2.059
5 2.326
6 2.534
7 2.704
8 2.847
9 2.970

QGS\SPC 62

PROCESS CAPABILITY
Standard Deviation (σ) = R/d2
= 0.169/2.33 USL = 0.900
= 0.0725 LSL = 0.500
Process Capability
Cp = (USL - LSL) / 6σ = (0.900 - 0.500) / 6 x 0.0725
= 0.920
CpU = (USL - X) / 3σ = (0.900 - 0.738) / 3 x 0.0725
= 0.743
CpL = (X - LSL) / 3σ = (0.738 - 0.900) / 3 x 0.0725
= 1.093
Cpk = 0.743
LSL USL

X

1.093 0.743
QGS\SPC 63

30
Process Capability:

Cp Cpk Remarks

 Process Capable
 Continue Charting
 Bring Cpk closer to Cp
 Process has potential Capability
X  Improve Cpk by Local action

 Process lacks basic Capability

X X  Improve process by Management
action

QGS\SPC 64

What about PROCESS PERFORMANCE?

• Process Capability (Cp, Cpk) indicates the ability of the
process to meet the specification (Voice of Customer) when
Process operates under the Common Causes.

• In practical situation, a process shows variation due to both

common as well as assignable causes.

• One must analyse process behaviour due to combined effect

of Common and Assignable causes. The index is known as
Process Performance Index (Pp, Ppk)

QGS\SPC 65

31
What about PROCESS PERFORMANCE?
T

LSL USL
- 3s + 3s

Process Width

Design Width

QGS\SPC 66

PROCESS PERFORMANCE
Standard Deviation σs = ∑ (xi-X)2
i=1
for n=80

n-1 USL = 0.900

LSL = 0.500
Process Performance
Pp = (USL - LSL) / 6σs = (0.900 - 0.500) / 6 x 0.0759
= 0.880
PpkU = (USL - X) / 3σs = (0.900 - 0.738) / 3 x 0.0759
= 0.710
PpkL = (X - LSL) / 3σs = (0.738 - 0.900) / 3 x 0.0759
= 1.045
Ppk = 0.710

QGS\SPC 67

32
 Control Charts
Ongoing Process Control
• Collect the data at the frequency as established
• Plot on control chart
• Perform instant analysis and interpretation
• Give immediate feedback to the process for
action if any indication of process behaviour
change
•Record significant process events (Tool change, Operator change, Shift change, Breakdown etc…
•This helps identifying assignable causes
QGS\SPC 68

Interpreting Control Charts-Ongoing

 The Interpretation of control charts is based on the statistical probability

of a particular pattern occurring by complete chance (or being caused by
random variation).
 All of the tests identify events that have a less than 0.3 % chance of
occurring by random chance (outside of 3s probability of being caused by
random variation).
 Control charts are divided into sigma zones above and below the average
line.
 Zone C is < 1σ from the mean.
 Zone B is between 1σ and 2σ.
 Zone A is between 2σ and 3σ.
 Beyond Zone A is > 3σ.
 The following rules are from Journal of Quality Technology, October, 1984
and have been included in AIAG SPC manual-edition2

Control charts are instituted to be interpreted, not just for wallpaper.

QGS\SPC 69

33
Tests 1 & 2
Test 2. Seven Points in a Row
Test 1. One Point Beyond Zone A
on One Side of the Center Line
x
UCL
UCL A
A
B B
C C
C C
B B
A A
LCL
LCL
x

 Test 1 – the basic test.  Test 2

 Caused by a large change in  Caused by a process mean
the process shift
 Requires immediate action

QGS\SPC 70

Tests 3 & 4
Test 3. Six Points in a Row Steadily Test 4. Fourteen Points in a Row
Increasing or Decreasing Alternating Up and Down

UCL
A  Test 4 – Alternating Patterns
B x  Over adjustment
C  Shift-to-Shift variation
C  Machine-to-machine variation
B
A x
LCL

UCL
 Test 3 – Trends up or down A
 Caused by B
C x
 Mechanical wear
 Chemical depletion C
B
 Increasing contamination A
 Etc. LCL

QGS\SPC 71

34
Tests 5 & 6
Test 5. Two Out of Three Points in a
Row in the same Zone A or Beyond
x x
 Test 6
UCL
A  Another test for shifts
B  Test 1, 5, 6 are related and
C
show conditions of high
C
special cause variability.
B
A Test 6. Four Out of Five Points in a
LCL row in the same Zone B and Beyond
x

UCL
A x
 Test 5 B
 The second basic test C
C
 High variation w/o exceeding B x
the 3σ limit A
LCL
 Major special-cause variation
Why not include all of the tests? QGS\SPC 72

Tests 7 & 8
Test 7. Fifteen Points in a Row in Zone C
(Above and Below Centerline)

UCL
A  Test 8 – Alternating Means
B x  Mixtures
C  Over control
C  Two different processes on the
B same chart
A Test 8. Eight Points in a Row on Both Sides
LCL
of Centerline with None in Zone C

UCL
 Test 7 – The Whitespace Test A
x
 Occurs when within subgroup B
variation is large compared to C
between group variation or, C
 Old or incorrectly calculated limits B
A
LCL

QGS\SPC 73

35
Interpretation Exercise – 1

X-bar Chart for Length

5 .5
3.0 SL=5.357
 Interpret the following
Sample Mean

2.0 SL=5.233
1.0 SL=5.109
control charts according the
5 .0 X=4.985
-1.0 SL=4.86 1
the previous eight rules.
-2.0 SL=4.73 8
-3.0 SL=4.61 4
4 .5
0 5 10 15 20 25 X-bar Chart for Length
Sample Number 5.3
3.0SL=5.232
5.2
2.0SL=5.149
5.1

Sample Mean
1.0SL=5.066

 Are there any out of control 5.0

4.9
X=4.983

-1.0SL=4.900

conditions? Which rule? 4.8 -2.0SL=4.818

-3.0SL=4.735
4.7

4.6
0 5 10 15 20 25
Sample Number

QGS\SPC 74

Individuals (I & MR) Charts

When to use:
 Measurement is variable
 There is no rational basis for sub grouping or
 The measurements are expensive and/or destructive or
 Production rate is slow or
 Population is homogeneous
 Suitable for both process and product parameters

QGS\SPC 75

36
MOVING RANGE CONTROL CHARTS VARIABLE CONTROL CHART Form # S 16005
X-MR Chart
SECTION: PRODUCT: CHARACTERISTICS: SPECIFICATION:

SQC IN-CHARGE: INSPECTOR/OPERATOR: M/C #: FREQUENCY:

X- CHART
X-Chart
MR-Chart

SAMPLE # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
DATE
TIME
X
MR

QGS\SPC 76

MOVING RANGE CONTROL CHARTS

DATE READING RANGE
(X) (R)
01. 04.2000 8.00
02. 04.2000 8.50 0.50
03. 04.2000 7.40 1.10
04. 04.2000 10. 50 3.10
05. 04.2000 9.30 1.20
06. 04.2000 11. 10 1.80
07. 04.2000 10. 40 0.70
08. 04.2000 10. 40 0.00
09. 04.2000 9.00 1.40
10. 04.2000 10. 00 1.00
11. 04.2000 11. 70 1.70
12. 04.2000 10. 30 1.40
13. 04.2000 16. 20 5.90
14. 04.2000 11. 60 4.60
15. 04.2000 11. 50 0.10
16. 04.2000 11. 00 0.50
17. 04.2000 12. 00 1.00
18. 04.2000 11. 00 1.00
19. 04.2000 10. 20 0.80
20. 04.2000 10. 10 0.10
21. 04.2000 10. 50 0.40
22. 04.2000 10. 30 0.20
23. 04.2000 11. 50 1.20
24. 04.2000 11. 10 0.40

Ave ra ge 10. 23 1.31

QGS\SPC 77

37
MOVING RANGE CONTROL CHARTS

 Upper Control Limits UCLX = X+E2R

 Lower Control Limits LCLX = X-E2R
 Upper Control Limits UCLR = D4 R
 Lower Control Limits LCLR = D3 R

Subgroup Size (n) d2 D3 D4 E2

2 1.128 - 3.267 2.660

QGS\SPC 78

TABLE OF CONSTANTS FOR (I-

(I-MR) CHARTS

Subgroup
d2 D3 D4 E2
Size (n)
2 1.128 - 3.267 2.660
3 1.693 - 2.574 1.772
4 2.059 - 2.282 1.457
5 2.326 - 2.114 1.290
6 2.534 - 2.004 1.184
7 2.704 0.076 1.924 1.109
8 2.847 0.136 1.864 1.054
9 2.970 0.184 1.816 1.010

QGS\SPC 79

38
 Section-3

ATTRIBUTE CONTROL CHART

QGS\SPC 80

Control Charts for Attribute Data

 Defective Data  Defect Data
 Package does/does not leak  Bubbles in a windshield
 Lamp does/does not light  Paint flaws on a casing
 Go/no-go gauge data  Errors on an invoice
 Correct/Incorrect Shipment  Bad die on a wafer
 p- and np-charts  c- & u-charts
 p-charts for proportion non-  u-chart for the number of
conforming units, sample size not defects per inspection unit,
necessarily constant sample size not necessarily
 np-charts for number of non- constant
conforming units, sample size is  c-chart for the number of
constant defects, sample size constant

QGS\SPC 81

39
p-Chart
 When to use a p-chart
 When it is difficult or uneconomical to make a numerical
measurement
 When it is desired to combine different types of defects
into an overall proportion
 When the available data are for attributes
 When the data come from a binomial process
 Management summaries
 Many management summaries are attribute forms and
could benefit from control chart analysis.
 Examples: scrap rates, quality audits, first-run yields, etc.

?
QGS\SPC 82

p- CHARTS (Attribute)

 Select the Subgroup Size

For p-chart, sufficiently large so that np>5)

QGS\SPC 83

40
p- CHARTS (Attribute)

 Select Subgroup Frequency

Detect change in the Process over span of time
 Select No. of Subgroups
Major source of variation incorporate (25 or more)

QGS\SPC 84

P Control Charts
DATE SAMPLE NO. OF PROPORTION
SIZE DEFECTIVES DEFECTIVES
01.04.2000 62 2 0.03
02.04.2000 62 5 0.08
03.04.2000 62 4 0.06
04.04.2000 62 3 0.05
05.04.2000 62 3 0.05
06.04.2000 62 6 0.10
07.04.2000 62 5 0.08
08.04.2000 62 0 0.00
09.04.2000 62 7 0.11
10.04.2000 62 5 0.08
11.04.2000 62 4 0.06
12.04.2000 62 1 0.02
13.04.2000 62 2 0.03
14.04.2000 62 3 0.05
15.04.2000 62 6 0.10
16.04.2000 62 3 0.05
17.04.2000 62 8 0.13
18.04.2000 62 4 0.06
19.04.2000 62 4 0.06
20.04.2000 62 4 0.06
21.04.2000 62 6 0.10
22.04.2000 62 4 0.06
23.04.2000 62 2 0.03
24.04.2000 62 3 0.05
25.04.2000 62 7 0.11
TOTAL: 101 1.629

QGS\SPC 85

41
P Control Charts
ATTRIBUTE CONTROL CHART
P/np- Chart
SECTION: PRODUCT: CHARACTERISTICS: SPECIFICATION:

SQC IN-CHARGE: INSPECTOR/OPERATOR: M/C #: FREQUENCY:

X- CHART

SA MPLE # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
DATE
TIME
n
d= np
d/n= p

QGS\SPC 86

Calculating Control Limits For a p-Chart

 For each subgroup

 Record: n – the number of units inspected
np – the number of nonconforming units
 Calculate p – the proportion
non-conforming: np
p=
n
3 p (1 − p )
UCLp = p +
 Calculate the control limits: n
n 1 p 1 + n 2 p 2 + ... + n kp k
p=
n 1 + n 2 + ... + n k 3 p (1 − p )
LCLp = p −
n

QGS\SPC 87

42
Interpreting p-Charts
 Attribute charts have fewer tests for special causes due to
the lower information content of attribute data
 A point outside the 3σ limit is assumed to be from special
cause variation.
 When np<5, the
likelihood of runs
below p-bar
increases.
Take care when
interpreting runs
in these cases.

QGS\SPC 88

Section -4

APPENDIX

QGS\SPC 89

43
Process Capability for Attribute

Capability (Attribute) = 1-p bar

or Using Z -Score

QGS\SPC 90

Process Capability for Attribute

QGS\SPC 91

44
np-Charts
 When to use an np-chart?
 Use the same criteria as a p-chart:
 When it is difficult or uneconomical to make a numerical
measurement
 When it is desired to combine different types of defectives into
a single value
 When the available data are for attributes
 When the data come from a binomial process
And the subgroups are all the same size
 How is it different from a p-chart?
 The actual number of nonconforming parts are plotted,
rather than the proportion defective
QGS\SPC 92

Control Limits for np-Charts

 For each subgroup
 Record: np – the number of nonconforming units
 Calculate np-bar – the mean # nonconforming units:
np 1 + np 2 + ... + np k
np =
k

 Calculate the control limits:

np
UCLp = n p + 3 n p (1 − ) = n p + 3 n p (1 − p )
n
np
LCLp = n p − 3 n p (1 − ) = n p − 3 n p (1 − p )
n
QGS\SPC 93

45
np-
np- CHART : Minitab output

NP Chart of DEFECTIVES
10 UCL=9.87

8
Sample Count

__
4 NP=4.04

0 LCL=0

1 3 5 7 9 11 13 15 17 19 21 23 25
Sample

QGS\SPC 94

np-Charts - Exercise
 Construct an np-chart from the data set
given below
 Use subgroup size = 100
 Interpret the results of the chart

QGS\SPC 95

46
Charts For Defects – c-Charts and u-Charts

 When to use c- or u-charts

 When the data is attribute data of defects, not defectives
 When the data comes from a Poisson distribution
 c-charts are charts constructed for number of occurrences
for a constant exposure
 u-charts are constructed for rate of occurrence for either
constant or varying exposure

QGS\SPC 96

Charts For Defects – c-Charts and u-Charts

 Examples
 Each month, 100 invoices are audited and the total number of
mistakes is recorded.

 In a molding process, there is a problem with pinholes in plastic

bottles. Each day, a number of bottles are examined and the
number of pinholes are recorded.

 Each month, the number of accidents that occur in an

organization is recorded.

QGS\SPC 97

47
Gathering c- or u-Chart Data

 Conditions for a Poisson distribution

 The number of occurrences in one subgroup is independent of the
number of occurrences in other subgroups.
 The rate of occurrences remain constant for all subgroups
 It is unlikely to have two occurrences in a very small area of
exposure.
 The value recorded in any subgroup is the number of occurrences
in the given exposure.

 Planning for c- and u-charts

 The event being studied can usually occur over some continuous
time or space, called the area of opportunity or exposure
 The area of opportunity can consist of some unit of time (8 hours
running time), an amount of material (2 sq. ft) or a number of
units (100 bottles)

QGS\SPC 98

Calculate c-Chart Control Limits

 For each subgroup

 Record: c – the number of defects found

 Calculate c-bar – the mean number of defects:

c 1 + c 2 + ... + c k
c=
k

 Calculate the control limits:

UCLc = c + 3 c
LCLc = c − 3 c

QGS\SPC 99

48
C Control Charts
DATE SAMPLE NO. OF
DATE SI ZE DEFECTS
01.04.2000 8 8
02.04.2000 8 17
03.04.2000 8 18
04.04.2000 8 15
05.04.2000 8 17
06.04.2000 8 9
07.04.2000 8 19
08.04.2000 8 6
09.04.2000 8 14
10.04.2000 8 17
11.04.2000 8 13
12.04.2000 8 15
13.04.2000 8 16
14.04.2000 8 22
15.04.2000 8 13
16.04.2000 8 10
17.04.2000 8 14
18.04.2000 8 9
19.04.2000 8 19
20.04.2000 8 11
21.04.2000 8 21
22.04.2000 8 23
23.04.2000 8 3
24.04.2000 8 8
25.04.2000 8 12
TOTAL: 349
QGS\SPC 100

C Control Chart
FRACTION DEFECTIVES = 13.96

UCL = C + 3 SQRT(C)

= 25.169

LCL = C - 3 SQRT(C)

= 2. 751

C CONTROL CHART
3.50
FECTS

3.00
2.50
NDE

2.00
RTIO

1.50
PROPO

1.00
0.50
0.00
00

00

00

00

00

00

00

00

00

00

00

00

00
20

20

20

20

20

20

20

20

20

20

20

20

20
4.

4.

4.

4.

4.

4.

4.

4.

4.

4.

4.

4.

4.
.0

.0

.0

.0

.0

.0

.0

.0

.0

.0

.0

.0

.0
01

03

05

07

09

11

13

15

17

19

21

23

25

DATE

PROPORTION DEFECTS UCL LCL AVERAGE

QGS\SPC 101

49
C-Chart interpretation

C Chart for Blemish

8
3.0SL=7.677
7
6 2.0SL=6.027

Sample Count
5
1.0SL=4.376
4

3
C=2.725
2

1 -1.0SL=1.074

0 LB=0.00E+00

0 10 20 30 40
Sample Number

 Interpretation rules for c- & u-charts are the same as those for p-
and np-charts.
 Because all of the points fall randomly within 3σ process limits, you
can conclude the process is running normally.

QGS\SPC 102

u-Charts

 When
 When the data are from Poisson process
 When the data are rates of occurrence with different sample
sizes
 Calculate Control Limits c 1 + c 2 + ... + c k
u=
 Calculate the process n 1 + n 2 + ... + n k
average nonconformities:
u u
UCLu = u + 3 =u+3
n n
 Calculate the control limits: u u
LCLu = u − 3 =u−3
n n

QGS\SPC 103

50
 Charts for Defects – Summary

 Charting defects is one step closer to the ideal, but these

charts still focus on outputs, not inputs
 c- and u-charts are also charts of attribute data and
suffer the same shortcomings as p- and np-charts
 It is easier to calculate c-charts than u-charts due to the
constant sample size
 When the sample sizes are large enough and the number
of defects is large, attribute data behaves like variable
data from a normal distribution

QGS\SPC 104

Attribute Control Form

AT T RIBUT ES CONT ROL CHART
p-CHART np-CHART c-CH ART u-CHART
Proportion Nonconforming Number Nonconforming Count of Nonconformities Nonconformities per Unit
PROCESS: CENTRLN
DEPART MENT : UPPER CL
CHARACTERISTIC: PREP. BY: LOWER CL
1 2 3 4 5 6 7 8 9 10 # 12 13 14 15 16 17 18 19 20 21 22 23 24 25
DATE
TIME
SHIFT
1
2
3
4
5
TOTAL
SAMPLE SIZE
PROPORTION
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

QGS\SPC 105

51