chapter 10 optimization designs
DESCRIPTION
Chapter 10 Optimization Designs. C. S. R. O. R. Optimization Designs. Focus : A Few Continuous Factors Output : Best Settings Reference : Box, Hunter & Hunter Chapter 15. Optimization Designs. C. S. R. O. 2 k with Center Points Central Composite Des. Box-Behnken Des. R. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/1.jpg)
10.1
Chapter 10
Optimization Designs
![Page 2: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/2.jpg)
10.2
Optimization Designs
C S
R O
R
Focus: A Few Continuous Factors Output: Best SettingsReference: Box, Hunter & Hunter
Chapter 15
![Page 3: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/3.jpg)
10.3
Optimization Designs
C S
R O
R
2k with Center PointsCentral Composite Des.
Box-Behnken Des.
![Page 4: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/4.jpg)
10.4
Response Surface Methodology
A Strategy of Experimental Design for finding optimum setting for factors. (Box and Wilson,
1951)
![Page 5: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/5.jpg)
10.5
Response Surface Methodology
• When you are a long way from the top of the mountain, a slope may be a good approximation
• You can probably use first–order designs that fit a linear approximation
• When you are close to an optimum you need quadratic models and second–order designs to model curvature
![Page 6: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/6.jpg)
10.6
Response Surface
-20
2
-2
0
2
0
20
40
60
80
-20
2
Yield
![Page 7: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/7.jpg)
10.7
First Order Strategy
-20
2
-2
0
2
0
20
40
60
80
Yield
![Page 8: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/8.jpg)
10.8
Second Order Strategy
02
-2
0
2
0
20
40
60
80
-2
Yield
![Page 9: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/9.jpg)
10.9
Quadratic Models can only take certain forms
Maximum Minimum Saddle Point Stationary Ridge
![Page 10: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/10.jpg)
10.10
Sequential Assembly of Experimental Designs as Needed
Fractional Factorial•Linear model
Full Factorial w/Center points•Main effects•Interactions•Curvature check
Central Composite Design•Full quadratic model
![Page 11: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/11.jpg)
10.11
Example: Improving Yield of a Chemical Process
Factor – 0 +
Time (min) 70 75 80
Temperature (°C) 127.5 130 132.5
Levels
![Page 12: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/12.jpg)
10.12
Strategy
• Fit a first order model
• Do a curvature check to determine next step
![Page 13: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/13.jpg)
10.13
Use a 22 Factorial Design With Center Points
x1 x2
1 – –2 + –3 – +4 + +5 0 06 0 07 0 0
![Page 14: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/14.jpg)
10.14
Plot the Data
70 80
125
135
130
Tem
pera
ture
(C
)
Time (min)
64.6 68.0
60.354.3
60.364.362.3
![Page 15: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/15.jpg)
10.15
Scaling Equations
x i (Original Units) Average of Original Units
Range of Original Units
( )
( )12
x2 temperature 130 C
2.5 Cx1
time 75 minutes
5 minutes
![Page 16: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/16.jpg)
10.16
Run Factors in original units
Factors in coded units Response
Time(min.)
Temp. (°C) Yield (gms)
x1 x2 x1 x2 y
1 70 127.5 – – 54.32 80 127.5 + – 60.33 70 132.5 – + 64.64 80 132.5 + + 68.05 75 130.0 0 0 60.36 75 130.0 0 0 64.37 75 130.0 0 0 62.3
Results From First Factorial Design
![Page 17: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/17.jpg)
10.17
Curvature Check by Interactions
• Are there any large two factor interactions? If not, then there is probably not significant curvature.
• If there are many large interaction effects then we should not follow a path of steepest ascent because there is curvature.
Here, time=4.7, temp =9.0, and time*temp=-1.3, so the main effects are larger and thus there is little curvature.
![Page 18: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/18.jpg)
10.18
Curvature Check with Center Points• Compare the average of the factorial points, ,
with the average of the center points, . If they are close then there is probably no curvature?
Statistical Test:
y f
yc
n
n
s r
f
c
c
# of points in factorial
# of points in center
Standard deviation of points in cente
If zero is in the confidence interval then there is no evidence of curvature.
![Page 19: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/19.jpg)
10.19
Curvature Check with Center Points
No Evidence of Curvature!!
![Page 20: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/20.jpg)
10.20
Fit the First Order Model
. . .y x x 62 0 2 35 4 51 2
![Page 21: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/21.jpg)
10.21
Path of Steepest Ascent: Move 4.50 Units in x2 for
Every 2.35 Units in x1
Equivalently, for everyone unit in x1 we move 4.50/2.35=1.91 units in x2
2.35
4.50
x1
1.91
1.0
x2
![Page 22: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/22.jpg)
10.22
The Path in the Original Factors
60 70 80 90
125
145
140
135
130
Tem
pera
ture
(C
)
Time (min)
64.6 68.0
60.354.3
60.364.362.3
![Page 23: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/23.jpg)
10.23
Scaling Equations
x1 time 75 minutes
5 minutesx2
temperature 130 C
2.5 C
time 5x1 75 minutes
temperature 2.5x2 130 C
![Page 24: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/24.jpg)
10.24
Factors in coded units
Factors in original units Run Response
Time(min.)
Temp. (°C) Yield (gms)
x1 x2 x1 x2 y
0 0 75 130.0 5,6,7 62.31 1.91 80 134.8 8 73.32 3.83 85 139.63 5.74 90 144.4 10 86.84 7.66 95 149.15 9.57 100 153.9 9 58.2
center conditions
path of steepestascent
Points on the Path of Steepest Ascent
![Page 25: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/25.jpg)
10.25
60 70 80 90 100 110
125
145
140
135
130
150
155T
empe
ratu
re (
C)
Time (min)
58.2
86.8
64.6 68.0
60.354.3
60.364.362.3
73.3
Exploring the Path of Steepest Ascent
![Page 26: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/26.jpg)
10.26
60 70 80 90 100 110
125
145
140
135
130
150
155T
empe
ratu
re (
C)
Time (min)
86.889.7
64.6 68.0
60.354.3
60.364.362.3
78.8 84.5
77.491.2
A Second Factorial
![Page 27: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/27.jpg)
10.27
Run Factors in original units
Factors in coded units
Response
Time(min.)
Temp.(°C)
Yield (gms)
x1 x2 x1 x2 y
11 80 140 – – 78.8
12 100 140 + – 84.5
13 80 150 – + 91.2
14 100 150 + + 77.4
15 90 145 0 0 89.7
16 90 145 0 0 86.8
Results of Second Factorial Design
![Page 28: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/28.jpg)
10.28
Curvature Check by Interactions
Here, time=-4.05, temp=2.65, and time*temp=-9.75, so the interaction term is the largest, so there appears to be curvature.
![Page 29: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/29.jpg)
10.29
New Scaling Equations
x1 time 90 minutes
10 minutes
x2 temperature 145 C
5 C
![Page 30: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/30.jpg)
10.30
60 70 80 90 100 110
125
145
140
135
130
150
155T
empe
ratu
re (
C)
Time (min)
87.086.0
64.6 68.0
60.354.3
60.364.362.3
78.8 84.5
77.491.2
A Central Composite Design
83.3 81.2
81.2
79.5
![Page 31: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/31.jpg)
10.31
Run Variables in original units
Variables in coded units
Response
Time(min.)
Temp.(°C)
Yield (gms)
x1 x2 x1 x2 y
second first-order design
11 80 140 – – 78.8
12 100 140 + – 84.5
13 80 150 – + 91.2
14 100 150 + + 77.4
15 90 145 0 0 89.7
16 90 145 0 0 86.8
runs added to form a composite design
17 76 145 – 2 0 83.3
18 104 145 + 2 0 81.2
19 90 138 0 – 2 81.2
20 90 152 0 + 2 79.5
21 90 145 0 0 87.0
22 90 145 0 0 86.0
The Central Composite Design and Results
![Page 32: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/32.jpg)
10.32
Must Use Regression to find the Predictive Equation
The Quadratic Model in Coded units
ˆ y 87.38 1.38x1 0.36x2
2.14x12 3.09x2
2 4.88x1x2
![Page 33: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/33.jpg)
10.33
Must Use Regression to find the Predictive Equation
The Quadratic Model in Original units
![Page 34: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/34.jpg)
10.34
70 80 90 100 110
145
140
135
150
155
Tem
pera
ture
(C
)
Time (min)
87.086.0
78.8 84.5
77.491.2
The Fitted Surface in the Region of Interest
83.3 81.2
81.2
79.5
![Page 35: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/35.jpg)
10.35
Quadratic Models can only take certain forms
Maximum Minimum Saddle Point Stationary Ridge
![Page 36: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/36.jpg)
10.36
Canonical Analysis
Enables us to analyze systems of maxima and minima in many
dimensions and, in particular to identify complicated ridge systems,
where direct geometric representation is not possible.
![Page 37: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/37.jpg)
10.37
Two Dimensional Example
x0
x1
x2x2
x1
x2
x1
ˆ y 1x12 2x2
2ˆ y 0 1x1 2 x2 12x1x2
11x12 22x2
2
is a constant
222
21122110ˆ xxxxy
![Page 38: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/38.jpg)
10.38
Quadratic Response Surfaces
• Any two quadratic surfaces with the same eigenvalues (’s) are just shifted and rotated versions of each other.
• The version which is located at the origin and oriented along the axes is easy to interpret without plots.
![Page 39: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/39.jpg)
10.39
The Importance of the Eigenvalues
• The shape of the surface is determined by the signs and magnitudes of the eigenvalues.
Type of Surface
Minimum
Saddle Point
Maximum
Ridge
Eigenvalues
All eigenvalues positive
Some eigenvalues positive and some negative
All eigenvalues negative
At least one eigenvalue zero
![Page 40: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/40.jpg)
10.40
Stationary Ridges in a Response Surface
• The existence of stationary ridges can often be exploited to maintain high quality while reducing cost or complexity.
![Page 41: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/41.jpg)
10.41
Response Surface Methodology
Eliminate Inactive Factors Fractional Factorials (Res III)
Goals Tools
Find Path of Steepest AscentFractional Factorials (Higher Resolution from projection or additional runs)
Follow Path Single Experiments until Improvement stops
Find CurvatureCenter Points (Curvature Check)Fractional Factorials (Res > III) (Interactions > Main Effects)
Model CurvatureCentral Composite DesignsFit Full Quadratic Model
Continued on Next Page
BrainstormingSelect Factors and Levels and Responses
![Page 42: Chapter 10 Optimization Designs](https://reader036.vdocuments.us/reader036/viewer/2022062422/56812e55550346895d93fb02/html5/thumbnails/42.jpg)
10.42
Understand the shape of the surface
Goals Tools
Canonical Analysis• A-Form if Stationary Point is outside Experimental Region• B-Form if Stationary Point is inside Experimental Region
Find out what the optimum looks likePoint, Line, Plane, etc.
Reduce the Canonical Form with the DLR Method
If there is a rising ridge, then follow it.
Translate the reduced model back and optimum formula back to original coordinates
Translate the reduced model back and optimum formula back to original coordinates
If there is a stationary ridge, then find the cheapest place that is optimal.