Thanks to visit codestin.com
Credit goes to www.scribd.com

Scribd
Upload
380 views19 pages

Box and Draper: Empirical Model-Building And: Response Surfaces

This document describes the use of response surface methodology to optimize settings for time and temperature in a manufacturing process. Initial experiments used a first-order model to identify significant factors. Further experiments using a central composite design found curvature, requiring a second-order model. Additional "star" runs were performed to estimate quadratic effects. The final second-order model identified optimal settings as well as interactions between time and temperature that maximized yield.

Uploaded by

Quea Apurimac
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
380 views19 pages

Box and Draper: Empirical Model-Building And: Response Surfaces

This document describes the use of response surface methodology to optimize settings for time and temperature in a manufacturing process. Initial experiments used a first-order model to identify significant factors. Further experiments using a central composite design found curvature, requiring a second-order model. Additional "star" runs were performed to estimate quadratic effects. The final second-order model identified optimal settings as well as interactions between time and temperature that maximized yield.

Uploaded by

Quea Apurimac
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

Box and Draper: Empirical Model-Building and

Response Surfaces

Khuri and Cornell: Response Surfaces

Myers and Montgomery: Response Surface


Methodology

-1-
An example from Box, Hunter and Hunter

Objective: Find settings of time (>) and temperture


(X ) that produced maximum yield.

Current setting: > œ 75 minutes, X œ 130‰ C


Past experience suggested that 5 œ 1.5.

-2-
-3-
First-order model:
5
C œ "!  "= B =  %
=œ"

Second-order model:
5 5
C œ "!  "= B =  "== B#=  "=> B= B>  %
=œ" =œ" "Ÿ=>Ÿ5

-4-
First first-order design

Runs time (min) Temperature (‰ C) B" B# C (grams)

" (! "#(Þ& " " &%Þ$


# )! "#(Þ& " " '!Þ$
$ (! "$#Þ& " " '%Þ'
% )! "$#Þ& " " ')Þ!
& (& "$!Þ! ! ! '!Þ$
' (& "$!Þ! ! ! '%Þ$
( (& "$!Þ! ! ! '#Þ$

The random order in which runs were actually


performed was 5,4,2,6,1,7,3
-5-
sC œ '#Þ!"  #Þ$&B"  %Þ&B#
=Þ.Þ Ð „ !Þ&(Ñ Ð „ !Þ(&Ñ Ð „ !Þ(&Ñ

Interaction test:

,"# œ !Þ'&, =Þ.Þ œ 0.75

Curvature test:

,""  ,## œ "% Ð&%Þ$  '!Þ$  '%Þ'  ')Þ!Ñ  "$ Ð'!Þ$


 '%Þ$  '#Þ$Ñ œ !Þ&!, s.d. œ "Þ"&

-6-
Estimation of error (from the three observations at
the center point): =# œ %Þ!

= œ #.0 agrees well with the preliminary value of


5 œ 1.5

-7-
Contour lines:

'#Þ!"  sC  #Þ$&B"  %Þ&B# œ !

Direction of steepest ascent is perpendicular to the


contour lines. Starting at the center of the
experimental region, simultaneously move
%Þ&Î#Þ$& œ "Þ*" units in B# for every unit moved in
B" .

-8-
Points on the path of steepest ascent

B" B# > X run C

! ! (& "$!Þ! &ß 'ß ( '#Þ$ (mean)


" "Þ*" )! "$%Þ) ) ($Þ$
# $Þ)$ )& "$*Þ'
$ &Þ(% *! "%%Þ% "! )'Þ)
% (Þ'' *& "%*Þ"
& *Þ&( "!! "&$Þ* * &)Þ#

-9-
Second first-order design

Runs time (min) Temperature (‰ C) B" B# C (grams)

"" )! "%! " " ()Þ)


"# "!! "%! " " )%Þ&
"$ )! "&! " " *"Þ#
"% "!! "&! " " ((Þ%
"& *! "%& ! ! )*Þ(
"' *! "%& ! ! )'Þ)

The random order in which runs were actually


performed was 11,15,13,12,16,14

-10-
sC œ )%Þ($  #Þ!#&B"  "Þ$#&B#
=Þ.Þ Ð „ !Þ&(Ñ Ð „ !Þ(&Ñ Ð „ !Þ(&Ñ

Interaction test:

,"# œ 4.88, =Þ.Þ œ 0.75

Curvature test:

,""  ,## œ &Þ#), s.d. œ "Þ$!

-11-
A first-order model is not adequate to represent the
local response function.

Second-order design (Star (axial) points are added to


form a central composite design)

-12-
Runs time (min) Temperature (‰ C) B" B# C (grams)

"" )! "%! " " ()Þ)


"# "!! "%! " " )%Þ&
"$ )! "&! " " *"Þ#
"% "!! "&! " " ((Þ%
"& *! "%& ! ! )*Þ(

È #
"' *! "%& ! ! )'Þ)

 È#
"( (' "%& ! )$Þ$

È #
") "!% "%& ! )"Þ#

 È#
"* *! "$) ! )"Þ#
#! *! "&# ! (*Þ&
#" *! "%& ! ! )(Þ!
## *! "%& ! ! )'Þ!

-13-
The random order in which the last 6 runs were
actually performed was 1*,#!,"(,#",##,")

sC œ )(Þ$'  "Þ$*B"  !Þ$(B#  #Þ"&B#"


 $Þ"#B##  %Þ))B" B#

-14-
-15-
-16-
-17-
-18-
-19-

You might also like