Design of Engineering Experiments

Part 5 – The 2k Factorial Design

• Text reference, Chapter 6
• Special case of the general factorial design; k factors, all at
two levels
• The two levels are usually called low and high (they could be
either quantitative or qualitative)
• Very
V widely
id l usedd in
i industrial
i d t i l experimentation
i t ti
• Form a basic “building block” for other very useful
experimental designs (DNA)
• Special (short-cut) methods for analysis
• We will make use of Design-Expert

Chapter 6 Design & Analysis of Experiments 1

7E 2009 Montgomery
 The Simplest
p Case: The 22

“-” and “+” denote the low and

high levels of a factor,
respectively
• Low and high are arbitrary
terms
• Geometrically, the four
runs form the corners of a
square
• Factors can be quantitative
or qualitative, although
their treatment in the final
model will be different
Chapter 6 Design & Analysis of Experiments 2
7E 2009 Montgomery
 Chemical Process Example
p

A = reactant concentration, B = catalyst amount,

y = recovery

Chapter 6 Design & Analysis of Experiments 3

7E 2009 Montgomery
 Analysis Procedure for a
Factorial Design
• Estimate factor effects
• Formulate model
– With replication, use full model
– With an unreplicated design, use normal probability
plots
• Statistical testing (ANOVA)
• Refine the model
• Analyze residuals (graphical)
• Interpret results

Chapter 6 Design & Analysis of Experiments 4

7E 2009 Montgomery
 Estimation of Factor Effects
A  y A  y A
See textbook, pg. 209-210 For
ab  a b  (1)
  manuall calculations
l l i
2n 2n
 21n [ab  a  b  (1)] The effect estimates are: A
= 8.33,
8 33 B = -5.00,
5 00 AB = 1.67
1 67
B  yB   yB 
Practical interpretation?
ab  b a  (1)
 
2n 2n Design-Expert analysis
 21n [ab  b  a  (1)]
abb  (1) a  b
AB  
2n 2n
 21n [ab  (1)  a  b]
Chapter 6 Design & Analysis of Experiments 5
7E 2009 Montgomery
 Estimation of Factor Effects
Form Tentative Model
Term Effect SumSqr % Contribution
Model Intercept
Model A 8.33333 208.333 64.4995
Model B -5 75 23 2198
23.2198
Model AB 1.66667 8.33333 2.57998
Error Lack Of Fit 0 0
Error P Error 31.3333 9.70072

Lenth's ME 6.15809
Lenth's
Lenth s SME 7.95671

Chapter 6 Design & Analysis of Experiments 6

7E 2009 Montgomery
 Statistical Testing - ANOVA

The F-test for the “model” source is testing the significance of the
overall model; that is, is either A, B, or AB or some combination of
these effects important?

Chapter 6 Design & Analysis of Experiments 7

7E 2009 Montgomery
 Design-Expert
g p output,
p , full model

Chapter 6 Design & Analysis of Experiments 8

7E 2009 Montgomery
 Design-Expert output, edited
or reduced model

Chapter 6 Design & Analysis of Experiments 9

7E 2009 Montgomery
 R id l and
Residuals d Di
Diagnostic
ti Ch
Checking
ki

Chapter 6 Design & Analysis of Experiments 10

7E 2009 Montgomery
 The Response Surface

Chapter 6 Design & Analysis of Experiments 11

7E 2009 Montgomery
 The 23 Factorial Design

Chapter 6 Design & Analysis of Experiments 12

7E 2009 Montgomery
 Effects in The 23 Factorial Design

A  y A  y A
B  yB   yB 
C  yC   yC 
etc, etc, ...

Analysis
done via
computer

Chapter 6 Design & Analysis of Experiments 13

7E 2009 Montgomery
 An Example of a 23 Factorial Design

A = gap, B = Flow, C = Power, y = Etch Rate

Chapter 6 Design & Analysis of Experiments 14

7E 2009 Montgomery
 Table of – and + Signs for the 23 Factorial Design (pg. 218)

Chapter 6 Design & Analysis of Experiments 15

7E 2009 Montgomery
 Properties of the Table
• Except for column I, every column has an equal number of + and –
signs
• The sum of the product of signs in any two columns is zero
• Multiplying any column by I leaves that column unchanged (identity
element))
• The product of any two columns yields a column in the table:

A  B  AB
AB  BC  AB 2C  AC

• Orthogonal design
• Orthogonality is an important property shared by all factorial
designs

Chapter 6 Design & Analysis of Experiments 16

7E 2009 Montgomery
 E ti ti off Factor
Estimation F t Effects
Eff t

Chapter 6 Design & Analysis of Experiments 17

7E 2009 Montgomery
 ANOVA Summary – Full Model

Chapter 6 Design & Analysis of Experiments 18

7E 2009 Montgomery
 Model Coefficients – Full Model

Chapter 6 Design & Analysis of Experiments 19

7E 2009 Montgomery
 Refine Model – Remove Nonsignificant Factors

Chapter 6 Design & Analysis of Experiments 20

7E 2009 Montgomery
 Model Coefficients – Reduced Model

Chapter 6 Design & Analysis of Experiments 21

7E 2009 Montgomery
 Model Summary Statistics for Reduced Model

• R2 and adjusted R2
5 106  10
SS Model 5.106 5
R  2
  0.9608
SST 5.314  10 5

SS E / df E 20857.75 /12
R 2
 1  1  0.9509
5 314  10 /15
Adj 5
SST / dfT 5.314
• R2 for p prediction (based
( on PRESS))
PRESS 37080.44
RPred  1 
2
 1  0.9302
SST 5.314  10 5

Chapter 6 Design & Analysis of Experiments 22

7E 2009 Montgomery
 Model Summary Statistics

• Standard error of model coefficients (full

model))
2 MS E 2252.56
se( ˆ )  V ( ˆ )  k
 k
  11.87
n2 n2 2(8)

• Confidence interval on model coefficients

ˆ  t / 2,df se( ˆ )    ˆ  t / 2,df se( ˆ )

E E

Chapter 6 Design & Analysis of Experiments 23

7E 2009 Montgomery
 The Regression Model

Chapter 6 Design & Analysis of Experiments 24

7E 2009 Montgomery
 Model Interpretation
p

Cube plots are

often useful visual
displays
p y of
experimental
results

Chapter 6 Design & Analysis of Experiments 25

7E 2009 Montgomery
 Cube Plot of Ranges

What do the
large ranges
when gap and
power are at the
high level tell
you?

Chapter 6 Design & Analysis of Experiments 26

7E 2009 Montgomery
Chapter 6 Design & Analysis of Experiments 27
7E 2009 Montgomery
 The General 2k Factorial Design
• Section 6-4,
6-4 pg.
pg 227
227, Table 6-9,
6-9 pg.
pg 228
• There will be k main effects, and
k 
  two-factor interactions
 2
k 
  three-factor interactions
 3

1 k  factor interaction
Chapter 6 Design & Analysis of Experiments 28
7E 2009 Montgomery
 6.5 Unreplicated
p 2k Factorial Designs
g
• These are 2k factorial designs with one
observation at each corner of the “cube”
• An unreplicated 2k factorial design is also
sometimes
ti called
ll d a “single
“ i l replicate” th 2k
li t ” off the
• These designs are very widely used
• Risks…if
Risks if there is only one observation at each
corner, is there a chance of unusual response
observations spoiling the results?
• Modeling “noise”?

Chapter 6 Design & Analysis of Experiments 29

7E 2009 Montgomery
 Spacing of Factor Levels in the
Unreplicated 2k Factorial Designs

If the factors are spaced too closely, it increases the chances

that the noise will overwhelm the signal in the data
M
More aggressive
i spacing
i isi usually
ll best
b t
Chapter 6 Design & Analysis of Experiments 30
7E 2009 Montgomery
 Unreplicated 2k Factorial Designs
• Lack of replication causes potential problems in
statistical testing
– Replication admits an estimate of “pure error” (a better
phrase is an internal estimate of error)
– With no replication, fitting the full model results in zero
degrees
g of freedom for error
• Potential solutions to this problem
– Pooling high-order interactions to estimate error
– Normal probability plotting of effects (Daniels, 1959)
– Other methods…see text

Chapter 6 Design & Analysis of Experiments 31

7E 2009 Montgomery
 Example of an Unreplicated 2k Design

• A 24 factorial was used to investigate the

effects of four factors on the filtration rate of a
resin
• The factors are A = temperature,
temperature B = pressure,
pressure
C = mole ratio, D= stirring rate
• Experiment was performed in a pilot plant

Chapter 6 Design & Analysis of Experiments 32

7E 2009 Montgomery
 The Resin Plant Experiment

Chapter 6 Design & Analysis of Experiments 33

7E 2009 Montgomery
 The Resin Plant Experiment

Chapter 6 Design & Analysis of Experiments 34

7E 2009 Montgomery
Chapter 6 Design & Analysis of Experiments 35
7E 2009 Montgomery
 Estimates of the Effects

Chapter 6 Design & Analysis of Experiments 36

7E 2009 Montgomery
Model Interpretation – Main Effects and Interactions

Chapter 6 Design & Analysis of Experiments 37

7E 2009 Montgomery
 Design Projection: ANOVA Summary for
th M
the d l as a 23 in
Model i Factors
F t A,
A C,
C and
dD

Chapter 6 Design & Analysis of Experiments 38

7E 2009 Montgomery
 The Regression Model

Chapter 6 Design & Analysis of Experiments 39

7E 2009 Montgomery
 Model Residuals are Satisfactoryy

Chapter 6 Design & Analysis of Experiments 40

7E 2009 Montgomery
 Model Interpretation – Response Surface Plots

With concentration at either the low or high g level,, high

g temperature
p and
high stirring rate results in high filtration rates
Chapter 6 Design & Analysis of Experiments 41
7E 2009 Montgomery
 The Half-Normal Probability Plot of Effects

Chapter 6 Design & Analysis of Experiments 42

7E 2009 Montgomery
 Other Analysis Methods for
Unreplicated 2k Designs
• Lenth’s method (see text, pg. 235)
– Analytical method for testing effects, uses an estimate
of error formed by pooling small contrasts
– Some adjustment to the critical values in the original
method can be helpful
– Probably most useful as a supplement to the normal
probability plot
• Conditional inference charts (pg. 236)

Chapter 6 Design & Analysis of Experiments 43

7E 2009 Montgomery
 Overview of Lenth’s method

For an individual contrast, compare to the margin of error

Chapter 6 Design & Analysis of Experiments 44

7E 2009 Montgomery
Chapter 6 Design & Analysis of Experiments 45
7E 2009 Montgomery
 Adjusted multipliers for Lenth
Lenth’ss method
Suggested because the original method makes too many
type
yp I errors,, especially
p y for small designs
g ((few contrasts))

Simulation was used to find these adjusted multipliers

Lenth’s method is a nice supplement
pp to the normal
probability plot of effects
JMP has an excellent implementation of Lenth’s method
in the screening platform
Chapter 6 Design & Analysis of Experiments 46
7E 2009 Montgomery
Chapter 6 Design & Analysis of Experiments 47
7E 2009 Montgomery
Outliers: suppose that cd = 375 (instead of 75)

Chapter 6 Design & Analysis of Experiments 48

7E 2009 Montgomery
 Dealing with Outliers
• Replace with an estimate
• Make the highest-order
highest order interaction zero
• In this case, estimate cd such that ABCD = 0
• A l
Analyze only
l the
h data
d you have
h
• Now the design isn’t orthogonal
• Consequences?

Chapter 6 Design & Analysis of Experiments 49

7E 2009 Montgomery
Chapter 6 Design & Analysis of Experiments 50
7E 2009 Montgomery
 The Drilling Experiment Example 6.3

A = drill
d ill load,
l d B = flow,
fl C = speed,
d D = type
t off mud,
d
y = advance rate of the drill

Chapter 6 Design & Analysis of Experiments 51

7E 2009 Montgomery
 Normal Probability Plot of Effects –
The Drilling Experiment

Chapter 6 Design & Analysis of Experiments 52

7E 2009 Montgomery
 Residual Plots

D ESI GN -EX PER T P lo t R esiduals vs. Pred icted

a d v. _ ra te
2.5 8625

1.4 4875

Re sid ual s
0.3 1125

-0.82625

-1.96375

1.69 4.70 7.70 10.71 13.7 1

P red icte d

Chapter 6 Design & Analysis of Experiments 53

7E 2009 Montgomery
 Residual Plots
• The residual plots indicate that there are problems
with the equality of variance assumption
• The usual approach to this problem is to employ a
transformation on the response
• Power family transformations are widely used

y y
*

• Transformations are typically performed to

– Stabilize variance
– Induce at least approximate normality
– Simplify
p y the model

Chapter 6 Design & Analysis of Experiments 54

7E 2009 Montgomery
 S l i a Transformation
Selecting T f i
• E
Empirical
i i l selection
l ti off lambda
l bd
• Prior (theoretical) knowledge or experience can
often suggest the form of a transformation
• Analytical selection of lambda…the Box-Cox
(1964) method (simultaneously estimates the
model parameters and the transformation
pparameter lambda))
• Box-Cox method implemented in Design-Expert

Chapter 6 Design & Analysis of Experiments 55

7E 2009 Montgomery
 (15.1)

Chapter 6 Design & Analysis of Experiments 56

7E 2009 Montgomery
 The Box-Cox Method
DE S IG N-E X P E RT P l o t B ox-C ox P lot for Power Transforms
a d v._ ra te

Lam bda
6.85
A log transformation is
Cu rre n t = 1
B e st = -0 .2 3 recommendedd d
L o w C.I. = -0 .7 9
Hi g h C.I. = 0 .3 2 5.40
The procedure provides
SS)

Re co m m e n d tra n sfo rm :
Log
a co
confidence
de ce interval
te va
L n(R es idualS

(L a m b d a = 0 )
on the transformation
3.95
parameter lambda
Iff unity
i isi included
i l d d in
i
2.50
the confidence interval,
no transformation
1.05 would be needed
-3 -2 -1 0 1 2 3

Lam bda
Chapter 6 Design & Analysis of Experiments 57
7E 2009 Montgomery
 Effect Estimates Following the
Log Transformation

Three main effects are

large
No indication of large
interaction effects
Whatt hhappened
Wh d tto th
the
interactions?

Chapter 6 Design & Analysis of Experiments 58

7E 2009 Montgomery
 ANOVA Following the Log Transformation

Chapter 6 Design & Analysis of Experiments 59

7E 2009 Montgomery
 Following the Log Transformation

Chapter 6 Design & Analysis of Experiments 60

7E 2009 Montgomery
 The Log Advance Rate Model

• Is the log model “better”?

• We would g generallyy prefer
p a simpler
p model
in a transformed scale to a more
complicated model in the original metric
• What happened to the interactions?
• Sometimes transformations provide insight
into the underlying mechanism

Chapter 6 Design & Analysis of Experiments 61

7E 2009 Montgomery
 Other Examples of
Unreplicated 2k Designs
• The sidewall panel experiment (Example 6.4, pg. 245)
– Two factors affect the mean number of defects
– A third factor affects variability
– Residual plots were useful in identifying the dispersion
effect
ff t

Chapter 6 Design & Analysis of Experiments 62

7E 2009 Montgomery
Chapter 6 Design & Analysis of Experiments 63
7E 2009 Montgomery
fig_06_27
fig_06_28
fig_06_29
fig_06_30
The oxidation furnace experiment (Example 6.5, pg. 245)
Replicates versus repeat (or duplicate) observations?
M d li within-run
Modeling ithi variability
i bilit

fig_06_31
fig_06_32a
fig_06_32b
fig_06_33
fig_06_34
fig_06_35
 Addition of Center Points
to a 2k Designs
• Based on the idea of replicating some of the
runs in a factorial design
• Runs at the center provide an estimate of
error and allow the experimenter to
distinguish between two possible models:
k k k
First-order model ((interaction)) y   0    i xi    ij xi x j  
i 1 i 1 j i
k k k k
Second-order model y   0    i xi    ij xi x j    ii xi2  
i 1 i 1 j i i 1

Chapter 6 Design & Analysis of Experiments 74

7E 2009 Montgomery
Chapter 6 Design & Analysis of Experiments 75
7E 2009 Montgomery
 yF  yC  no "curvature"
curvature
The hypotheses are:
k
H 0 :   ii  0
i 1
k
H1 :   ii  0
i 1

nF nC ( yF  yC ) 2
SS Pure Quad 
nF  nC

This sum of squares has a

single degree of freedom

Chapter 6 Design & Analysis of Experiments 76

7E 2009 Montgomery
 Example 6.6, Pg. 248
Refer to the original experiment
shown
h i Table
in T bl 6.10.
6 10 Suppose
S that
h nC  4
four center points are added to this
experiment,
p and at the ppoints x1=x2 Usually between 3
=x3=x4=0 the four observed and 6 center points
filtration rates were 73, 75, 66, and will work well
69 The average of these four center
69. Design-Expert
points is 70.75, and the average of provides the analysis,
the 16 factorial runs is 70.06. includingg the F-test
Si
Since are very similar,
i il we suspect for pure quadratic
that there is no strong curvature curvature
ppresent.

Chapter 6 Design & Analysis of Experiments 77

7E 2009 Montgomery
Chapter 6 Design & Analysis of Experiments 78
7E 2009 Montgomery
 ANOVA for Example 6.6 (A Portion of Table 6.22)

Chapter 6 Design & Analysis of Experiments 79

7E 2009 Montgomery
 If curvature is significant,
g , augment
g the design
g with axial runs to
create a central composite design. The CCD is a very effective design
for fitting a second-order response surface model

Chapter 6 Design & Analysis of Experiments 80

7E 2009 Montgomery
 Practical Use of Center Points (pg
(pg. 260))
• Use current operating conditions as the center
point
• Check for “abnormal” conditions during the
time the experiment was conducted
• Check for time trends
• Use center p points as the first few runs when there
is little or no information available about the
magnitude of error
• Center
C points
i andd qualitative
i i factors?
f ?

Chapter 6 Design & Analysis of Experiments 81

7E 2009 Montgomery
 Center Points and Qualitative Factors

Chapter 6 Design & Analysis of Experiments 82

7E 2009 Montgomery