Quality Assurance

MT-257

SPC

Statistical Applications
Statistical Thinking

Collection, organization, analysis, interpretation and

presentation of Data

Understanding Variation is key

Variation occurs in all processes
Variation is the root cause of all problems

Scooter Inc.
Consider the Steer Support
Extrusion followed by CNC Machine
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Statistical Applications

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Statistical Applications
Scooter Inc.
Variation Still Exits from:
Differences in Raw Materials
Placement on CNC machine
Temperature of room at time of processing
Mistake in CNC programming
According to Design
Ideally should be 79.95 mm
Heights must fall between 79.90 80.00
If less than 79.0 Loose fit will rattle
If greater than 80.0 may not fit in gap
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Variation Types
What to do?
Statistical

Process Control (SPC)

Take samples and measure accurately

On a daily basis

Define

performance measurements

Variation: Two Types

Common Causes
Common

Variation reflecting pure randomness in the

process
No two snowflakes are alike
No two flowers are exactly identical
The same with Scooter Inc.
Cannot Predict, but can Describe in form of Statistical
Distribution that applies to the larger population

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Variation Types
Variation: Two Types
Assignable Causes
Changes

in the Parameters of
Underlying Statistical Distribution
of Process
Mistake in Programming of CNC
Wear & Tear of Extrusion Machine

Not

common to all steer support

parts
Only affect a subset
Distribution

will look different

Higher variance
Different Mean

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Variation Types
Variation Types
Write

the letter R using your normal writing

hand

Switch

hands and write the R with the other

Use

the normal for first 4 and other for

remaining

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Variation Types

Variation Types
Common

Cause Variation
Not every letter exactly the same

Still

a Common Cause Variation but much

larger
No Pattern to predict shape of next letter

Assignable

Causes Switch in Hand after 4th R

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Common Cause Variability

Common Cause Variability
Common causes relate to the design of the

system
Expects a bell shaped or normal distribution
Further over time, this distribution should not
change
The process is said to be In control
Every Process shows this Common Variation

Time
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Assignable Cause Variability

Assignable Cause
Operator error, bad batch, environment change, etc.
Over time a process may change
Changes can occur in the mean, variance or shape
The process is said to be Out of Control if such

changes occur

Time
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

10

SPC Objectives

Objectives of SPC
Alert

management of Assignable Causes

Do not want to alert management of small random

variations
Do Not alert on letter to next in the first row
Measure the amount of Variation in the Process
Create an Objective measure of Consistency
By how much is the first row of Rs better than the
second Row
Assign cause to variation currently perceived as pure
randomness and control these causes, leading to reduced
11
variation and higher consistency

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Scooter Inc. Control Charts

Control Charts
Used to distinguish between
Assignable and Common Causes of
Variation
Track the Process Over-Time
Based on Samples
Sample Sizes between 2 10
Over several time periods; 20 to
50
Lets take 5 sample
sizes over a 25 day
period
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

12

Calculate x-bar and range for each

X-Bar and R Control Charts: Establishing the

X
&
Y
axis

Control Charts

Establish

X-axis:

Y-axis:
X-bar

the X & Y coordinates

Time Periods

Chart:
Mean of each Sample
R (Range) Chart
Range of each Sample

Mean
or
Range
of
each
Sampl
e

Calculate x-bar and range for each

sample
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

13

X-Bar and R Control Charts: Establishing the

X & Y axis
X-Bar Chart

Document Trends over time corresponding to Assignable Causes of

Variation
Identify unexpected drifts wear of tool
Jumps New person operating

is the Sample Size in each period

R Chart
Range:

Difference between highest and lowest in the

sample
A form of Variation in the Process

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

14

X-Bar and R Control Charts: Establishing

the X & X-bar
Y axis
Calculating
and Range for each Period
For Example: Day 14

X = (79.973 + 79.986 + 79.942 +

79.978 + 79.979) / 5

R = Max ( 79.973,79.986,
79.942,79.978,79.979)
Min (79.973,79.986,79.942, 79.978,
79.979)

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Next need to establish the

upper and lower control
15
limits

X-Bar Control Charts: Upper and Lower Control

Limit
Upper and Lower Control LimitsX
for

Where A2 is based on
the number of
observations in the
subgroup (n)
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

16

R Control Charts: Upper and Lower Control

Limit
Upper and Lower Control Limits for R Chart

Where D3 & D4 are

based on the number
of observations in the
subgroup (n)

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

17

Scooter Inc. Control Charts

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

18

X-Bar and R Control Charts

X-Bar and R Charts

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

19

INTERPRETING CONTROL
CHARTS
NATURAL PATTERNS OF VARIATION

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

20

INTERPRETING CONTROL
CHARTS
X-Bar and R Charts

What is the Probability of 8 subsequent points above or below the center line?
(0.5)^

8 = 0.4% Highly Unlikely Corresponds to out of

control

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

21

Interpreting Control
Charts
Out of Control
What is the Probability?
A. 1 or more points fall outside control limits
B. 2 out of 3 consecutive points fall in the same region
C.
D.
E.
F.

A
4 out 5 consecutive points fall in the same region A
or B
9 consecutive points on the same side of
average
G
A B
D
F of decreasing
E
C points
6 consecutive
increasing
A
14 Bconsecutive points alternating up and down
C
B
A

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

22

Using X-Bar and Range Control Charts

Together
Use the X-Chart and R-Chart Together to determine if the process is

In Control
If.the process mean is Shifting Upwards
Sampling
Distribution

(process mean is
shifting upward)

UCL

Reveals shift

x-Chart

Then we will
see the
following:

LCL

UCL

R-chart
LCL

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Does not
reveal shift
23

Together
Use the X-Chart and R-Chart Together to determine if the process is

In Control
If.the process Variability is Increasing
Sampling
Distribution

(process variability is
increasing)

UCL

x-Chart

Then we will
see the
following:

LCL

Does not
reveal increase

UCL

R-chart

Reveals increase
LCL

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

24

X-Bar using Standard Deviation & S-bar

Datachart
Pool
Data

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

0.65

0.75

0.75

0.6

0.7

0.6

0.75

0.6

0.65

0.6

0.8

0.85

0.7

0.65

0.9

0.75

0.75

0.75

0.65

0.6

0.5

0.6

0.8

0.65

0.65

0.7

0.85

0.8

0.7

0.75

0.75

0.8

0.7

0.8

0.7

0.75

0.75

0.7

0.7

0.8

0.8

0.7

0.7

0.65

0.6

0.55

0.8

0.65

0.6

0.7

0.65

0.75

0.8

0.7

0.65

0.75

0.65

0.8

0.85

0.6

0.9

0.85

0.75

0.85

0.8

0.75

0.85

0.6

0.85

0.65

0.65

0.65

0.75

0.65

0.7

0.65

0.85

0.7

0.75

0.85

0.85

0.75

0.75

0.85

0.8

0.5

0.65

0.75

0.75

0.75

0.8

0.7

0.7

0.65

0.6

0.8

0.65

0.65

0.6

0.6

0.85

0.65

0.75

0.65

0.8

0.7

0.7

0.75

0.75

0.65

0.8

0.7

0.7

0.6

0.85

0.65

0.8

0.6

0.7

0.65

0.8

0.75

0.65

0.7

0.65

a) Calculate X-Bar for each Subgroup

b) Calculate Standard Deviation for each Subgroup
c) Determine the X-Double bar
d) Determine the S-Bar (average of standard
Deviations)

X-bar Control
Limits

S-Bar Control
Limits

e) Find the Control Limits for both Charts

f) Plot X-bar and S-Values for each chart

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

25

X-Bar using Standard Deviation & S-bar

Datachart
Pool
Data

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

0.65

0.75

0.75

0.6

0.7

0.6

0.75

0.6

0.65

0.6

0.8

0.85

0.7

0.65

0.9

0.75

0.75

0.75

0.65

0.6

0.5

0.6

0.8

0.65

0.65

0.7

0.85

0.8

0.7

0.75

0.75

0.8

0.7

0.8

0.7

0.75

0.75

0.7

0.7

0.8

0.8

0.7

0.7

0.65

0.6

0.55

0.8

0.65

0.6

0.7

0.65

0.75

0.8

0.7

0.65

0.75

0.65

0.8

0.85

0.6

0.9

0.85

0.75

0.85

0.8

0.75

0.85

0.6

0.85

0.65

0.65

0.65

0.75

0.65

0.7

0.65

0.85

0.7

0.75

0.85

0.85

0.75

0.75

0.85

0.8

0.5

0.65

0.75

0.75

0.75

0.8

0.7

0.7

0.65

0.6

0.8

0.65

0.65

0.6

0.6

0.85

0.65

0.75

0.65

0.8

0.7

0.7

0.75

0.75

0.65

0.8

0.7

0.7

0.6

0.85

0.65

0.8

0.6

0.7

0.65

0.8

0.75

0.65

0.7

0.65

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

26

Attribute Control Charts

What if we have a Go, No-Go Situation

Binary:

Defective or Not Defective?

Does not have to based on a single dimension
Airline tracking Lost Luggage
Pharmacy providing wrong drugs to patient
We will use a P-Chart
Where P is the Fraction Defective
Sample

sizes are larger; 50 to 200

Especially if defects are rare

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

27

Constructing a P-Chart
1. Calculate P, Fraction of defects found for Sample

Size n

2. Determine the Average Fraction Defective

=

n = 100,
total number of defects in all groups = 194, Number of Parts
100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
10
7= 5
13
9
7
8
9
12
6
15
8
5
8
8
10
8
4
6
6
9
Inspected
2400

n 100 100
np
6
7
p 0.06 0.07

0.1 0.07 0.05 0.13 0.09 0.07 0.08 0.09 0.12 0.06 0.15 0.08 0.05 0.08 0.08

100
8
0.1 0.08 0.04 0.06 0.06 0.09 0.08

P-average 0.081
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

28

Constructing a P-Chart
3. Calculate the Upper and Lower Control Limits
a) If n is Constant

b) If n is Not Constant
Same equation but n = Average Sample Size

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

29

Constructing a P-Chart
n 100 100
np
6
7
p 0.06 0.07

100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
10
7
5
13
9
7
8
9
12
6
15
8
5
8
8
0.1 0.07 0.05 0.13 0.09 0.07 0.08 0.09 0.12 0.06 0.15 0.08 0.05 0.08 0.08

100 100 100 100 100 100 100

10
8
4
6
6
9
8
0.1 0.08 0.04 0.06 0.06 0.09 0.08

P-average 0.081

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

30

P-Chart Example

Five Production Runs are completed. Each had

some defectives. Compute the Control Limits for
P Chart based on the information given

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

31

P-Chart Example

Quantity is not the same

n = 1,550 / 5 = 310

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

32

P-Chart
Use the data to create a P-Chart

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

P-Chart
Calculate the Proportions for each Sub Group

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Process Capability
Difference between Capability and In Control
Recall:

Both Rows are In Control

But does the second row meet Design Specifications for R? No
2nd

Row Process is Not Capable

Control

Charts only measure to what extent the process is operating

based on its Historical Behaviour
We Set Limits of 79.928 & 79.974 based on how the process
performed in past 25 days

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

35

Process Capability
Conditions

Three Primary Conditions for a Capable

Process:
1. Process is Statistically in Control
2. Test for Normality and Central
Tendency

Do not use averages, rather Individual

Measures

Process
Spread is
too wide

Process
Center
needs
adjustment

3. The Control Limits should be well

Inside the Tolerance Limits

Measured by Cp Process Capability

Index Cp = Minimum Value of 1.33

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Capable Process
Spread centered and
within 36
Tolerance

Process Capability
Conditions

Three Primary Conditions for a Capable

Process:
Test for Normality and Central Tendency

Do not use averages, rather Individual

Measures

Draw the Nominal Dimension and

Tolerance Lines on the same distribution

and Calculate the Standard Deviation
If the Process is Centered
Use Cp
If the Process is Not Centered
Use Cpk

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

37

Process Capability
Conditions

Minimum Value Cp = 1.33

= 0.504 0.496

This is an Estimated Standard

Deviation Using Range

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

38

Process Capability
Conditions
What to do with a Process that is Not Capable?
Repair Existing Equipment
Modify Existing Process
Select a Better Process
Use 100% Inspection
Revise (Loosen) Drawing Tolerance where
possible
Purchase New Equipment
Etc.
39

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Process Capability Scooter

Inc.

Design Specifications set by downstream or end

customer
Target Value = 79.950 mm
Tolerance level = 79.90, 80.00

How do we determine if the Process is Capable?

Tightness of the Design Specification

Difference between USL and LSL = 80.00 - 79.90 = 0.1

Amount of Variation in the current process

We estimate the Standard Deviation using R (for small sample
sizes)

= Estimated Standard Deviation

D2 = from Table based on Sample Size

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

40

Process Capability Scooter

Inc.
Scooter Inc. Capability
Recall R-bar = 0.0402

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

41

Process Capability
Combine these Two Measures into a Single Score

Process Capability Index

Measures

Allowable Tolerance relative to Actual

Variance

How can we Increase the Capability of the Process

in meeting Design Specifications?

Increase the Tolerance
Or
Decrease the Variability in the Process
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

42

Process Capability Analysis

Lower
Specification

Upper
Specification

=1

Process variability
matches specifications
Lower
Specification

>1

Upper
Specification

Process variability well

Lower
Upper
within specifications
SpecificationSpecification

If incapable:
redesign process or reduce variability
<1
use alternative process
use 100-percent inspection
relax design specification
Process variability
exceeds specifications

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

43

Process Capability Scooter

Inc.
Process Capability Index
USL

= 80.00
LSL = 79.90

0.017283

Experts

Recommend a Cp of 1.33
Motorola goal was 2.0 at every individual step
Corresponds to USL being 6 standard deviations
above and below the mean
Six-Sigma
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

44

Capability Example
The design specification for the width of a part is
between 101 mm and 101.8 mm. Which machines
are capable?
Machine

Standard
Deviation

Machine
Variability
6

0.13

0.78

0.08

0.48

0.16

0.96

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Machine Capability
Cp = spec/6

45

Capability Example
The design specification for the width of a part is
between 101 mm and 101.8 mm. Which machines
are capable?
Machine

Standard
Deviation

Machine
Variability
6

Machine Capability
Cp = spec/6

0.13

0.78

0.80/0.78 = 1.03

0.08

0.48

0.80/0.48 = 1.67

0.16

0.96

0.80/0.96 = 0.83

Cp > 1.33 is desirable

Cp = 1.00 process is barely capable
Cp < 1.00 process is not capable
2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

46

Using Cpk when Process Not

Centered
Use Cpk when process is Not Centered (But Normal)

C pk

USL LSL
min
,

3
3

A company creates small packages of cookies in a 16 gram

package. Government standards state that weights must be

within 5 percent of the weight advertised on the package
Inspectors test 1,000 packages of cookies and find an

average weight of 15.875 grams with a standard deviation

of 0.529 grams

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

47

Using Cpk Example

A company creates small packages of cookies in a 16 gram

package. Government standards state that weights must be

within 5 percent of the weight advertised on the package
Inspectors test 1,000 packages of cookies and find an
average weight of 15.875 grams with a standard deviation
of 0.529 grams

UCL= 16 + .05(16) = 16.8 grams

Mean = 15.875

LCL= 16 .05(16) = 15.2 grams

Center = (16.8 + 15.20) / 2 =

16.00
Will use Cpk

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

48

Using Cpk Example

C pk
C pk

Specification Limi

USL LSL
min
,

3
3

Upper Spec = 16.8

Lower Spec = 15.2

15.875 15.2 16.8 15.875

Min
;

3(.529)
3(.529)

C pk Min.4253; .5829

C pk .4253

2011 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

Observed Weight
Mean = 15.875 g
Std. Dev = .529 g

< 1.33,
Process is
Not Capable
49