Download as pdf or txt
Download as pdf or txt
You are on page 1of 82

Ghislain Tremblay

Southern Illinois University Edwardsville


Fall 2023
Define Measure Analyze Improve Control
Six Sigma
Analysis Phase FMEA Risks Analysis
Perform FMEA
O1 O2 O3 O4
Perform Multi-vari Analysis I1
I2 FMEA
Identity Potential Critical Inputs I3
I4

Process Door Cause & Effect


VA NVA

Data Door
22 O O

21 O

20 O
n O n
O O
19 n n

18 X O X n X

17 n O n
X

16
. n X O

15 X n X

14 X n

13 X

12

Hypothesis-Testing 1 2 3 4 5 6 7 8 9 10

Regression Analysis
Chi-Square t-test Design of Experiments
c² ANOVA

Regression
X1
.
.
SIUE IE 462 Fall 2023 SIUE IE 462 Fall 2023
Improve 2
Objectives
Six Sigma

 At the end of this module,


you will be able to:

• Identify opportunities where DOE can be used to improve processes


• Plan and generate a basic DOE with Minitab using the appropriate
experimentation strategy
• Analyze a DOE both graphically and statistically
• Determine process changes needed to meet requirements

SIUE IE 462 Fall 2023


Design of Experiments Improve 3
Agenda
Six Sigma

Design of Experiments 400 minutes

DOE Process 35 minutes

Full Factorial Designs 160 minutes

Fractional Factorial Designs 125 minutes

Modeling Analysis 40 minutes

Blocking Designs 40 minutes

SIUE IE 462 Fall 2023


Design of Experiments DOE Process Improve 4
Tool Links
Six Sigma

You may need... To do...


Brainstorming Fishbone Diagram FMEA

PROBLEM
STATEMENT

FMEA
1 2 3
Minitab ANOVA Pareto
Pareto Chart of the Effects
DOE
(response is Distance, Alpha = .10)

A: StopAngl
C B: ArmLength
B C: PostHeig
D: StartAng
A E: BallType
D
CD
AC

Minitab 13 BC
DE
BD
AE
AB
BE
E
CE
AD

0 10 20 30

4 5 6

7 8 9
SIUE IE 462 Fall 2023
Design of Experiments Improve 5
Value Of DOE?
Six Sigma

Y = f ( x1, x2, …, xn)

$$$$$$$$$$$$

SIUE IE 462 Fall 2023


Design of Experiments DOE Process Improve 6
What Is a DOE?
Six Sigma

A systematic method:
- where a number of factors (Xs) are changed
What simultaneously, following a predetermined pattern,
to investigate their effect on the response or output
(Y)
- experimenting with complex processes with the
objective of optimizing the process.
DOE seeks to:
• Determine which variables affect the system.
• Determine how the magnitude of the variables
affects the system.
• Determine the optimum levels for the variables.
• Determine how to manipulate the variables to
control the response.

SIUE IE 462 Fall 2023


Design of Experiments DOE Process Improve 7
What is a Designed Experiment? Six Sigma

 A method to change all the factors at once in a structured


pattern to determine their effects on the output(s)

 The structured pattern is known as an orthogonal array

A B AXB
1 -1 -1 1
2 1 -1 -1
3 -1 1 -1
4 1 1 1

0 0 0

SIUE IE 462 Fall 2023 SIUE IE 462 Fall 2023 8


Improve 8
Why Use a DOE?
Six Sigma

 Do experiments with
more than one factor in
an efficient way

 Improve performance,
quality, and cost
Why
 Reduce variation

 Provide a framework for


choosing the
combination of factors
that optimize the outputs A key
for
success

SIUE IE 462 Fall 2023


Design of Experiments DOE Process Improve 9
Strategy of Experimentation
Six Sigma

1. Define the problem


2. Establish the objective
3. Select the response variable(s) (Ys)
Plan
4. Select the factors (Xs)
How 5. Choose the factor levels
6. Select the experiment design
7. Collect the data
8. Analyze the data Execute
9. Draw conclusions
10. Control the process
SIUE IE 462 Fall 2023
Design of Experiments DOE Process Improve 10
Statistical Thought Process VS Strategy of experimentation. Six Sigma

Correlation -Basically, it is integrated!

1. Formulate the question


2. State Hypothesis Plan -DOE
3. Determine Risk Level
4. Collect Data
5. Verify Assumptions
Execute - DOE
6. Compute Test Statistic
7. Interpret Results

SIUE IE 462 Fall 2023


Design of Experiments DOE Process Improve 11
Effects Six Sigma

1. Factor A affects the mean

A1 A2

2. Factor B affects the


standard deviation
Manipulate the
B1
controllable factors
How (Xs) at different levels B2
Y
to determine their
effect on some 3. Factor C affects the
response (Y) mean and the standard
deviation
C1
C2
Y

4. Factor D has no effect

D1 D2
SIUE IE 462 Fall 2023 Y
Design of Experiments DOE Process Improve 12
Terminology of DOE – Classes
Six Sigma

Screening Characterization Optimization

A E B
C B D K K
Factors I K F F
F
H G J J

Time

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 13
Basic Experimental Terms
Six Sigma

Model: Yij =  +  i +  j + ij

factor block effect error term


strip

Blocks: Blocking is a technique used to increase the precision of the experiment


by organizing the data into blocks.
A block is a portion of the experimental units collected under a certain condition.
Although every measurement should be taken under identical experimental
conditions, but this is not always possible.
Factors that affect the experiment and can be measured, but not controlled, can
be accounted for using a blocked design.
For examples, an experiment may be carried out over several days with
variations in temperature and humidity, or data may be collected in different
plants, or by different technicians could be called a block.

SIUE IE 462 Fall 2023


Improve 14
Terminology of DOE – Terms
Six Sigma

Test Score (Y) = f (A, B, C)


Process
Factor Variable Setting Level
A Complexity Complex +1 Treatment or
Simple -1 Factors Combination
B Training Advanced +1
Basic -1 Sequence Process Variables Response
Order A B C (score)
C Method Method A +1
Method B -1 1 -1 -1 -1 4.2
2 +1 -1 -1 2.1
Order 3 -1 +1 -1 7.2
4 +1 +1 -1 5.2
Data 5 -1 -1 +1 9.3
6 +1 -1 +1 5.5
Levels 7 -1 +1 +1 8.3
8 +1 +1 +1 9.9
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 15
Basic Experimental Terms Randomization
Six Sigma

 Important concept in DOE

 Refers to the order in which


experiments are conducted

 Eliminates systematic effects that


could occur over the time period the
data are collected

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 16
Basic Experimental Terms– Replication
Six Sigma

 Replication means to duplicate the same


experimental condition or test more than one
time.

 Allows you to measure Variation

 To do Replications:
Run each setup experiment once and repeat the
whole sequence again. This means you have to redo
the setup at each run.

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 17
Basic Experimental Terms– Repetition
Six Sigma

 Repetition is taking several measurements of


an experimental condition or test without
changing the setup.

 To do Repetitions:
Run each experiment’s setup several times in a
row.

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 18
Basic Experimental Terms – Center points
Six Sigma

Center points: Design points with all factor levels


set halfway between the high and low settings.

Adding center points to a design allows you to


Centre point detect curvature in the response surface.

If there is curvature that involves the center of the


b ab design, the response at the center point will either
+1 be higher or lower that the average of the factorial
(corner) points.
B
You can also add center points to increase the
-1 (1) power of your design.
a
True center points can only be used with numeric
-1 +1 factors, Minitab creates pseudo-center points in
A
factorial designs. In this case, for Q text factors,
Minitab adds 2Q times as many to the design or if
the design is blocked, to each block.
SIUE IE 462 Fall 2023
Improve 19
Basic Experimental Terms -Confounding
Six Sigma

Confounding: Effects that cannot be estimated separately from one


another.

For example, if a factor A is confounded with the 3-way interaction BCD,


then the estimated effect for A also includes any effect due to BCD.

These effects are also said to be aliased.

Confounding occurs when you use a fractional factorial design, because


you don’t run all the factor level combinations.

Minitab displays an alias table that specifies the confounding patterns in your
design.

SIUE IE 462 Fall 2023


Improve 20
Confounding Six Sigma

A B C AB AC BC ABC
2 (a) 1 -1 -1 -1 -1 1 1
3 (b) -1 1 -1 -1 1 -1 1
5 (c) -1 -1 1 1 -1 -1 1
8 (abc) 1 1 1 1 1 1 1

Same Signs

Was “Y” affected by A or by the interaction of B and C?

SIUE IE 462 Fall 2023


Improve 21
Analysis of Variance, (ANOVA)
Six Sigma

 This is the heart of the matter of DOE


 Separates variation to distinguish sources
 Uses that data and regression data to form
predictive models to mimic responses
 Based off the fit of those models to the
experimental data, you either have something you
can use or not
 All DOE are based on the same statistical
principles and method of analysis - ANOVA and
regression analysis.

SIUE IE 462 Fall 2023


Improve 22
Variance
Six Sigma

 6Ms – these are the general sources of variation


present in all processes
• Machines
• Method
• Man (people)
• Materials
• Mother Nature (environment)
• Measurement
 In DOE, you are causing variation by changing
factors that affect one or more of the 6Ms
• Then we statistically break out those effects to
determine what variation was caused by which
source
SIUE IE 462 Fall 2023
Improve 23
What are the Types Design? Six Sigma

 Methods of Experimentation
◦ Trial and Error (Best Guess)
◦ Single Factor Experiment (OFAT)
 one change at a time
◦ Fractional Factorial Experiment
 change many things at a time
◦ Full Factorial Experiment
 change many things at a time
◦ Others (Box-Jenkins, Taguchi, etc.)
SIUE IE 462 Fall 2023
Improve 24
Types of Experimental Factors
Six Sigma

2
HELD CONSTANT FACTORS
Inputs that you can control and
elect to hold constant for the
experiment

1
CONTROLLABLE
FACTORS PROCESS OUTPUT (Y)
Inputs that you UNDER Response variable
wish to vary in a
DOE
STUDY

3
UNCONTROLLABLE (NOISE)
FACTORS
Inputs that you cannot control
• Environmental
• Unknown
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 25
Why Not One-Factor-at-a-Time?
Six Sigma

1 TEMPERATURE 3 TEMPERATURE A

Optimum
Maximum

PRESSURE
PRESSURE
Response
Response

TEMPERATURE A TEMPERATURE A
2 4
Actual
Optimum Test

PRESSURE
PRESSURE

Conclusion

Maximum B
Response

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 26
Benefits of Statistical Experiments
Six Sigma

Expert Change every- One-factor- Statistical


Evaluation criteria Observation Guessing judgement thing at once at-a-time experiments
(uncontrolled) experiments (Incl. DOE)

Time to get results Very long Long Medium Long Long overall Short

Cost Very high High Medium High High overall Relatively


low

Chance of detecting Very low Low Moderate Moderate Suboptimal Excellent


optimal levels likely

Confidence Very low Low Moderate Low High Very high


in results

Ability to identify
independent main None None None None None Yes
and interaction
effects

Conclusion: Statistical experiments are rich with information


SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 27
Example DOE
Six Sigma

 Problem: Gas mileage for car is 20 mpg.


Would like to get >30 mpg.

 Factors:
Measurement Method Machine

One factor at a time.


New tires
Gage Error
Tire pressure
Speed

MPG

Driving habits
Type of Gas

Temperature
New driver

Materials Man Mother Nature

SIUE IE 462 Fall 2023 SIUE IE 462 Fall 2023 28


Improve 28
OFAT “Reasonable Approach” Six Sigma

SIUE IE 462 Fall 2023


Improve 29
OFAT (One-Factor-At-A-Time) Six Sigma

SIUE IE 462 Fall 2023


Improve 30
Refresher: What Is an Interaction?
Six Sigma

A B AXB
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 31
Interactions?
Six Sigma

 Interactions occur when the effect of one factor depends upon the
level of another factor.
 Example,
• Drug “A” reduces blood pressure when used by itself,
• Drug “B” reduces anxiety when used by itself.
• If Drug A and B are used together then it may lead to a heart attack or
stroke.

SIUE IE 462 Fall 2023


Improve 32
Interactions
Six Sigma
 Factors may influence each other. E.g, the
optimal setting of a factor may depend on the
settings of the other factors.

 When factors are optimised separately, the


overall result (as function of all factors) may be
suboptimal ...
 Understanding interactions is important to
improve output
Yield of a chemical process is impacted by
operating temperature and reaction time.
The impact of changing temperature on
yield depends upon the reaction time.

SIUE IE 462 Fall 2023


Improve 33
Interactions
Six Sigma

30 The real maximum

40
50
60

The apparent maximum


factor B has
been optimised

factor A has been optimised


SIUE IE 462 Fall 2023
Improve 34
Interaction Strength
Six Sigma

No interaction Weak interaction Strong interaction


AxB between A x B between A x B

Y Y Y
50 50 50 B: -1
B: -1
40 40 40

30 30 30
B: -1
20 20 20

10 B: +1 10 10
B: +1
B: +1

-1 +1 A -1 +1 A -1 +1 A
1 2 3

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 35
Full Factorial Designs
Six Sigma

 Full Factorial: Examines all factor effects and interaction effects.


These become large very quickly.

• 22 Full Factorial = 2 factors, 2 levels = 4 runs


• 23 Full Factorial = 3 factors, 2 levels = 8 runs
• 24 Full Factorial = 4 factors, 2 levels = 16 runs
• 25 Full Factorial = 5 factors, 2 levels = 32 runs

 Used after initial screening experiments or where the process is


simple or well known. The experiment is run to optimize the
process using a vital few factors.

SIUE IE 462 Fall 2023


Improve 36
Power of 22 Factorial Design
Six Sigma

Speed (A) Tire Pressure (B) MPG


55 30
55 35
65 30
65 35

Factorial Design – each level of one factor is found in


combination with each level of the other factors.

Allowing both Main Effects and Interactions to be estimated


SIUE IE 462 Fall 2023
Improve 37
Two-Level Full Factorial Design - Keeping it Simple
Six Sigma

Run all high/low combinations of 2 (or more) factors

Use statistics to identify the critical factors

22 Full Factorial

What could be simpler?

SIUE IE 462 Fall 2023


Improve 38
22 Factorial Example – Main Effect
Six Sigma

Lubrication (B)

Hardness (A) “low” “high”


(i.e., dirty) (i.e., clean)
“low” 10 20

“high” 30 40

Objective: two factor experiment focusing on the impact of hardness and


lubrication on process yield (%).
Main Effects:
Effect of Hardness (A): average change in response (yield) as hardness goes from
a low to high level.
Effect of Lubrication (B): average change in response (yield) as lubrication goes
from a low to high level.

SIUE IE 462 Fall 2023


Improve 39
22 Factorial Example
Six Sigma

Lubrication (B)
Hardness (A) “low” “high”
(i.e., dirty) (i.e., clean)
“low” 10 20
“high” 30 40

Main Effects:
Main Effect of A: A = (30+40)/2 – (10+20)/2 = 20
Main Effect of B: B = (20+40)/2 – (10+30)/2 = 10
Interaction AB: AB = (40+10)/2 – (30+20)/2 = 0

SIUE IE 462 Fall 2023


Improve 40
Plot of Effects – Main Effect
Six Sigma

50
Lub “high”
40
Yield
30 Lub “low”

20

10

“low” “High”
Hardness Hardness

Parallel lines indicate the absence of an interaction effect


Effect of hardness on yield is the same regardless of whether clean or dirty lub is
used.

SIUE IE 462 Fall 2023


Improve 41
22 DOE Example - Interaction
Six Sigma

Lubrication (B)
Hardness (A) “low” “high”
(i.e., dirty) (i.e., clean)
“low” 10 20
“high” 30 0
Changed response
Main Effects: from 40, notice
Main Effect of A: A = (30+0)/2 – (10+20)/2 = 0 impact on effects
Main Effect of B: B = (20+0)/2 – (10+30)/2 = -10 calculation and
Interaction AB: AB = (0+10)/2 – (30+20)/2 = -20 effects plot.

SIUE IE 462 Fall 2023


Improve 42
Plot of Effects - Interaction
Six Sigma

50

40
Yield
30 Lub “low”

20

10
Lub “high”

“low” “High”
Hardness Hardness

Intersecting lines indicate an interactive effect.


Effect of hardness on yield depends on whether you are using clean or dirty lub.
SIUE IE 462 Fall 2023
Improve 43
Design Construction - Understanding Interactions
Six Sigma
Std A B C AB AC BC ABC

1 – – – + + + – y1 With eight, purpose-picked runs, we


7 8
2 + – – – – + + y2
can evaluate:
• three main effects (MEs)
3 4 3 – + – – + – + y3
• three 2-factor interactions (2FI)
4 + + – + – – – y4 • one 3-factor interaction (3FI)
5 – – + + – – + y5 • as well as the overall average
5 6 6 + – + – + – – y6

7 – + + – – + – y7
1 2
8 + + + + + + + y8
B
C
A
Note the pattern in each column:
• All of the +/- patterns are unique.
• None of the patterns can be obtained by adding or subtracting any combination
of the other columns
• This results in independent estimates of all the effects.

SIUE IE 462 Fall 2023


Improve 44
Multi-layered experiments
Six Sigma

Experiments are not one-shot adventures. Ideally one performs:

 an initial experiment

• check-out experimental equipment


• get initial values for quantities of interest

main experiment

• obtain results that support the goal of the experiment

confirmation experiment

• verify results from main experiment


• use information from main experiment to conduct more focussed
experiment (e.g., near computed optimum)

SIUE IE 462 Fall 2023


Improve 45
Characteristics of Good DOE Six Sigma

◦ The experiment should provide unbiased estimates of process


variable and treatment effects (factors at different levels).
◦ The experiment should provide the precision necessary to enable the
experimenter to detect important differences.
◦ The experiment should plan for the analysis of the results.
◦ The experiment should generate results that are free from ambiguity
of interpretation.
◦ The experiment should point the experimenter in the direction of
improvement.
◦ The experiment should be as simple as possible.
 Easy to set up and carry out
 Simple to analyze and interpret
 Simple to communicate or explain to others

SIUE IE 462 Fall 2023


Improve 46
Be Proactive!
Six Sigma

 DOE is a proactive tool.

 If DOE output is inconclusive:

• You may be working with the wrong variables


• Your measurement system may not be capable
• The range between high and low levels may be
insufficient

 There is no such thing as a failed experiment

• Something is always learned


• New data prompts asking new questions and

SIUE IE 462 Fall 2023


generates follow-on studies
Improve 47
Minitab Design Resolution Six Sigma

The resolution number tells you what factor and


interactions will be confounded with one another.
SIUE IE 462 Fall 2023
Improve 48
Intentionally left blank

SIUE IE 462 Fall 2023 49


Agenda
Six Sigma

Design of Experiments 400 minutes

DOE Process

Full Factorial Designs 75 minutes

Fractional Factorial Designs 125 minutes

Modeling Analysis 40 minutes

Blocking Designs 40 minutes

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 50
Factorial Designs
Six Sigma

What are they?

• Experiments involving two or more factors where


each factor may have two or more levels

Why are they important?

• Used to study the effects and interactions between


the factors

How are they analyzed?

• Graphical and statistical analysis

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 51
Full Factorial Designs
Six Sigma

◦ A full-factorial design consists of all possible combinations


of all selected levels of the factors to be investigated.
 Examines every possible combination of factors at all levels.
◦ A full-factorial design allows the most complete analysis
 Can determine main effects of the factors manipulated on
response variables
 Can determine effects of factor interactions on response
variables
 Can estimate levels at which to set factors for best result
◦ Is Time consuming…

SIUE IE 462 Fall 2023


Improve 52
Refresher…… Six Sigma

Statistical Thought Process


1. Formulate Question
2. State Hypothesis
3. Determine Risk Level

How 4. Collect Data


5. Verify Assumptions
6. Compute Test Statistic
7. Interpret Results

SIUE IE 462 Fall 2023


Multiple Sample Analysis Improve 53
Strategy of Experimentation - Plan
Six Sigma

1. Define the problem


2. Establish the objective
3. Select the response variable(s) (Ys)
Plan
4. Select the factor(s) (Xs)
How 5. Choose the factors levels Notice that the Statistical
Thought Process is
6. Select the experiment design integrated in the Strategy of
Experimentation
7. Collect the data
8. Analyze the data Execute
9. Draw conclusions
10. Control the process
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 54
Catapult
Six Sigma

 The Marketing department has seized the opportunity


to meet a customer need. Even though you have never
set-up or operated a catapult, your salesperson has
great confidence in your ability.

 The customer wishes to purchase millions of catapults.


His specifications call for the catapult to launch a
projectile 60 inches with an accuracy of + / - 2.0 inches.

 The competition has been unable to make the transition


from “Building Block” assemblies to the catapult
business. The business is yours if you can produce
catapults that will operate to meet customer
specifications.

 Your first priority will be to determine if you can set-up


and operate this equipment.

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 55
Steps 1& 2 – Define Problem and Establish the Objective
Six Sigma

Key tasks for the Experimental Team:

• Write a Problem Statement

• Define the Objective of the DOE Study

Team charter,
Design of experiment planning form.xls
DMAIC_Experiment planning sheets_Rev 2.doc

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 56
Catapult – Steps 1 & 2
Six Sigma

Define Problem and Establish the Objective

Problem: Can the catapult be set up and a projectile


launched 60 inches consistently?

Objective: Set up the catapult and identify settings of the


factors that will allow operation of the catapult
to meet customer specifications

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 57
Catapult – Step 3 – Response Variable (Y)
Six Sigma

Experiment Planning Sheet – Page 1


CT Characteristics Relationship to Objective Specifications

Distance object Feasibility of operation Nominal target:


launched 60 inches +/- 2 inches

Design of experiment planning form.xls


DMAIC_Experiment planning sheets_Rev 2.doc

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 58
Step 4 – Select the Factors (Xs)
Six Sigma

 Controllable factors
 Held constant factors 3 types of factors
(FMEA, Fishbone diagram, SIPOC,
 Uncontrollable (noise) factors Brainstorm… can be good source
of information)
Input Factors (Xs) Output (Ys)

People Service
PROCESS
Material CTC
Equipment A blending of
Policies Product CTQ
inputs to
Procedures achieve the
desired CTD
Methods Task
outputs
Environment

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 59
Catapult – Step 4 – Define the “Control Factors”
Six Sigma
Experimental Planning Sheet – Top of page 2
Control Normal Proposed How are they
Factors Settings Settings Set
Stop Angle 60 degrees 50 and 70 Visually
degrees adjusted by the
operator

Arm Length 2 inches 1 and 3 inches Visually


adjusted by the
operator

Design of experiment planning form.xls


DMAIC_Experiment planning sheets_Rev 2.doc

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 60
Catapult – Step 4 – Define “Held Constant Factors”
Six Sigma

Experimental Planning Sheet – Bottom of page 2


Constant Desired How to Control How are they Set
Factors Settings
Post Height Post height: 4 in Recheck after each Operator sets tension
Start Angle Start Angle: 0 deg. launch by measurement.
Arm length set by
scale on arm.

No. and Type One large rubber Separated from other Chosen by supervisor
of rubber band and white ball choices (For the experiment)
bands
Type of
projectile

Design of experiment planning form.xls


DMAIC_Experiment planning sheets_Rev 2.doc

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 61
Catapult – Step 4 – Define “Noise Factors”
Six Sigma

Experiment Planning Sheet – Top of page 3

Noise Factors Strategy to Control Noise Factors


Operator Select experienced operators for this trial
run

Vibration of Tape catapult to table


catapult

Design of experiment planning form.xls


DMAIC_Experiment planning sheets_Rev 2.doc

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 62
Catapult – Step 5 – Choosing Factor Levels
Six Sigma
80

Response Variable (Y)


70 Levels for each
60
B difference factor should be
50
A difference balanced in relation
40
to each other (i.e., no
30

B- high
one dominant)

A-high
B-low

A-low
20

10
0 100 200 300 400 500 600 Y
Factor Settings
True Effect

Select factor settings


to ensure you do not
Experimental Effect
exclude the optimum
Low Hi
(-) (+) Factor Settings
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 63
Step 6 – Selecting the Experimental Design Six Sigma

Current Knowledge Type of Experiments Factors

Little knowledge Screening Studies Many factors, 2 levels

Some Knowledge Characterization Fewer factors, 2-3 levels

Much Knowledge Optimization Few factors, many levels

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 64
Step 6 – Randomization
Six Sigma

 Important concept in DOE

 Refers to the order in which


experiments are conducted

 Eliminates systematic effects that


could occur over the time period the
data are collected

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 65
Step 6 – Replication
Six Sigma

 Replication means to duplicate the same


experimental condition or test more than
one time.

 Allows you to measure Variation

 To do Replications:
Run each setup experiment once and repeat
the whole sequence again. This means you
have to redo the setup at each run.

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 66
Step 6 – Repetition
Six Sigma

 Repetition is taking several measurements of


an experimental condition or test without
changing the setup.

 To do Repetitions:
Run each experiment’s setup several times in a
row.

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 67
Example – Catapult Experiment
Six Sigma

1 Problem: Set up and 5 Levels: 1 & 3 inches


operation of new 50 & 70 degrees
product

2 Objective: Demonstrate 6 Experimental


catapult operability Design: 2 Factors
2 Levels
3 Replications
3 Measure: Distance in inches
Due to the high risk
involved with product
launch, management
4 Factors: Arm Length has only budgeted money
Stop Angle
for only 10-12 tests.
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 68
Generating a DOE in Minitab
Six Sigma

2
1

3
4

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 69
Generating a DOE in Minitab (Cont.)
Six Sigma

9
10
8

StdOrder RunOrder CenterPt Blocks StopAngle ArmLength


1 1 1 1 50 1
2 2 1 1 70 1
Unramdomized 3 3 1 1 50 3
Replicate 1
for training 4 4 1 1 70 3
purpose only – 5 5 1 1 50 1
demonstrate the 6 6 1 1 70 1
replicates 7 7 1 1 50 3
Replicate 2
8 8 1 1 70 3
File: Catapult_DOE_Data.xls
9 9 1 1 50 1
10 10 1 1 70 1
11 11 1 1 50 3
Replicate 3
12 12 1 1 70 3
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 70
Six Sigma

Ramdomiziation
of the runs

SIUE IE 462 Fall 2023


Improve 71
Checking for Interaction
Six Sigma

Interaction plot

4
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 72
Interaction Analysis
Six Sigma

Minitab output
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 73
ANOVA Analysis
Six Sigma

4
5

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 74
ANOVA Analysis (Cont.)
Six Sigma

Minitab output
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 75
Residuals Graphs
Six Sigma

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 76
Residuals Graphs
Six Sigma

Randomness of residuals

Minitab output

Normality of residuals

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 77
Step 9 – Conclusions and Recommendations
Six Sigma

Confirmation Run
 Check all observed data for accuracy

 Explain results of the experiment

 State results in practical terms as well


as statistical terms using graphs

 Limit conclusions to summary


evidence taken directly from the
experiment

 Conduct a confirmation run

 Point out the way ahead if further


experiments are required
SIUE IE 462 Fall 2023
Design of Experiments Full Factorial Designs Improve 78
Catapult – Step 9 – Conclusions and Recommendations
Six Sigma

 The primary reason for the tests was to


determine if the catapult could be set-up and
operated successfully.

 Conclusions (Based on the data):

• The Catapult was set-up and operated properly.


• Launch range: 24.5 inches to 63.5 inches
• Increasing both factors increased distance.
• The factors did not strongly interact with each
other.

 Recommendations: Test the remaining process


parameters to determine their settings before
releasing the catapult for production

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 79
Step 10 – Control the Process
Six Sigma

 When the process has been


optimized, it is time to
implement the changes.

 Develop procedures to control


key characteristics using
methods outlined in the control
phase

 Keep the process operating at


the peak of its performance

SIUE IE 462 Fall 2023


Design of Experiments Full Factorial Designs Improve 80
Six Sigma

Project #2

SIUE IE 462 Fall 2023


Improve 81
Catapult – DOE project 23
Six Sigma

Teams of 3 to 4
Duration: due on week
16 (_12/06_)
Deliverables:
• 23 DOE
Material:
• Catapult
 Select 3 Factors • Experiment
planning sheets
 Select 2 Levels for each Factor and results

 Use 2 Replications
 Generate the experiment with Minitab

Prepare a 10-15 minutes presentation

SIUE IE 462 Fall 2023


Full Factorial Designs Improve 82

You might also like