1 The 2 k-p Fractional Factorial Design Text reference, Chapter 8 Motivation for fractional...
-
Upload
cory-newton -
Category
Documents
-
view
214 -
download
1
Transcript of 1 The 2 k-p Fractional Factorial Design Text reference, Chapter 8 Motivation for fractional...
1
The 2k-p Fractional Factorial Design
• Text reference, Chapter 8• Motivation for fractional factorials is obvious; as the
number of factors becomes large enough to be “interesting”, the size of the designs grows very quickly
• There may be many variables (often because we don’t know much about the system)
• Emphasis is on factor screening; efficiently identify the factors with large effects
• Almost always run as unreplicated factorials, but often with center points
2
Why do Fractional Factorial Designs Work?
• The sparsity of effects principle– There may be lots of factors, but few are important– System is dominated by main effects, low-order
interactions
• The projection property– Every fractional factorial contains full factorials in
fewer factors
• Sequential experimentation– Can add runs to a fractional factorial to resolve
difficulties (or ambiguities) in interpretation
3
The One-Half Fraction of the 2k
• Section 8-2, page 283
• Notation: because the design has 2k/2 runs, it’s referred to as a 2k-1
• Consider a really simple case, the 23-1
• Choose {a, b, c, abc} as an one-half fraction
4
• The 23-1 design contains only those treatment combinations with a “+” in the ABC column (ABC is a “generator” of the fraction)
• Note that I =ABC (defining relation)
• Main effects of A, B, and C
lA = ½(a-b-c+abc) = lBC
lB = ½(-a+b-c+abc) = lAC
lC = ½(-a-b+c+abc) = lAB
• It is impossible to differentiate between A and BC, B and AC, and C and AB – This phenomena is called aliasing and it occurs in all fractional designs (confounding)
• Aliases can be found directly from the columns in the table of + and - signs
• Notation for aliased effects:
• A = BC, B = AC, C = AB
, , A B CA BC B AC C AB
5
• Aliases can be found from the defining relation I = ABC by multiplication:
AI = A(ABC) = A2BC = BC
BI =B(ABC) = AC
CI = C(ABC) = AB
• Using this fraction, instead of estimating A, we are estimating A+BC, etc.
• The two blocks/fractions can be determined by
I =+ABC
(principal fraction)
or
I =-ABC
(alternate/complementary fraction)
The One-Half Fraction of the 23
6
The Alternate Fraction of the 23-1
• I = -ABC is the defining relation• Implies slightly different aliases: A = -BC, B= -
AC, and C = -AB• Both designs belong to the same family, defined by
• Suppose that after running the principal fraction, the alternate fraction was also run
• The two groups of runs can be combined to form a full factorial – an example of sequential experimentation
I ABC
7
Running Both of the One-Half Fractions of the 23-1
• All the effects can be estimated by analyzing the full factorial 23 design, ordirectly from the two individual fractions. E.g.,
½(lA + l’A) = ½(A + BC + A – BC) -> A
½(lA - l’A) = ½(A + BC - A + BC) -> BC• Thus, all main effects and two factor interactions
will be estimated, but not three-factor interaction ABC. Why?
8
Design Resolution• Resolution III Designs:
– Main effects are aliased with two-factor interactions– example
• Resolution IV Designs:– Two-factor interactions are aliased with each other– example
• Resolution V Designs:– Two-factor interactions are aliased with three-factor interactions– Example
• The resolution of a two-level fractional factorial design = the smallest number of letters in any word in the defining relation
3 12III
4 12IV
5 12V