Anda di halaman 1dari 12

Lecture 1 Introduction &

Profit Upper Bound


(PUB)
Ch8L 4310
ur. Mauhew !. 8eal
Main Concepts I
rocess SynLhesls & ueslgn
ueLermlnlng Lhe process lnpuLs and
ouLpuLs
ueLermlnlng Lhe overall process
congurauon/ow
ueslgnlng Lhe process unlLs Lo achleve chemlcal
englneerlng funcuon
ueslgnlng Lhe process lnLeracuons Lo maxlmlze
ob[ecuves (e.g. recycle ow values, heaL
lnLegrauon)
Pow do we generaLe alLernauve deslgns?
Main Concepts II
ueclslons and ueclslon Maklng
ueLermlnlsuc Cpumlzauon
uncerLalnLy uullLy
leaslble
8eglon
Cpumal olnL
Pow do we accounL for
dlerenL declslon maker's rlsk
preferences?
Pow do we rapldly ellmlnaLe bad
alLernauves?
Pow do we choose beLween good
alLernauves?
Cb[ecuve luncuon
First Idea - Bounding
roblem solvlng heurlsuc: never do more work Lhan you have Lo."
!"#$ "&'()*+,- ./"& 0(12&**1( 3)44 54356* *&7 81(& 31(9 7"5: 61' "5;& +8& 71 <1=>
1) uo noL ellmlnaLe a posslbly opumal alLernauve
2) 1ry Lo ellmlnaLe alLernauves as qulckly as posslble
llnd a bound on Lhe quallLy of Lhe deslgn and ellmlnaLe deslgns LhaL have bounds LhaL
make Lhem clearly bad.
rocess
8aw MaLerlals roducL MaLerlals
roL upper 8ound" = value of roducL MaLerlals - value of 8aw MaLerlals
Profit Upper Bound Analysis
roL upper 8ound" = value of roducL MaLerlals - value of 8aw MaLerlals
1hls musL be CSl1lvL for Lhere Lo be any polnL ln furLher examlnlng an alLernauve
lL lCnC8LS cosLs of Lhe process lLself - Lherefore a lower bound on cosL
Cpumlsuc bounds overesumaLe value and underesumaLe cosL, we wlll always wanL
Lo use opumlsuc bounds Lo ellmlnaLe alLernauves.
We wlll someumes compare opumlsuc bounds Lo pesslmlsuc bounds and
ellmlnaLe alLernauves because Lhey cannoL be beuer Lhan an exlsung soluuon.
roL
AlLernauve A
AlLernauve 8
roL Lower 8ound of A > roL upper 8ound of 8
==> We can ellmlnaLe 8
Profit Upper Bound
Where do you geL chemlcal prlces?
hup://www.lcls.com/chemlcals/channel-lnfo-chemlcals-a-z/
1hese are lndlcauve" prlces and qulLe ouL of daLe.
CerLaln prlces can be obLalned easlly - such as for energy, www.ela.gov
Profit Upper Bound Example
Vinyl Chloride
C
2
P
4
+ Cl
2
! C
2
P
3
Cl + PCl
8ouLe 1 - Chlorlnauon of LLhylene
8ouLe 2 - AceLylene + Pydrogen Chlorlde
C
2
P
2
+ PCl

! C
2
P
3
Cl
8ouLe 3 - LLhylene Cxldauon + Pydrogen Chlorlde
2C
2
P
4
+ Cl
2
+ x C
2
! 2C
2
P
3
Cl + P
2
C
C
2
P
4
+ PCl

+ x C
2
! C
2
P
3
Cl + P
2
C
8ouLe 4 - LLhylene Cxldauon + Chlorlne
!"#$%&'( *+%&# !,(-
C
2
P
4
30
C
2
P
2
80
Cl
2
18
PCl

23
P
2
C 0
C
2
(alr) 0
C
2
P
3
Cl 33
lrom LexLbook
PUB Example Vinyl Chloride
Cont
C
2
P
4
+ Cl
2
! C
2
P
3
Cl + PCl
8ouLe 1 - Chlorlnauon of LLhylene - ?),9 5 @5*)* 12 A 4@ 12 B):64 !"41()<&.
( )
62.5
71x18 28x30
C/lb Cost Material Raw
+
=
( )
62.5
36.5x25 62.5x35
C/lb Cost Material Product
+
=
( ) ( )
lb C lb C / 16 / 712 . 15
5 . 62
18 71 30 28
62.5
25 36.5 35 62.5
! =
" + "
#
" + "
./01# *23 !,(- 4%56( &"(/+%7#
16
-16
7
11
C
2
P
4
+ Cl
2
! C
2
P
3
Cl + PCl
C
2
P
2
+ PCl

! C
2
P
3
Cl
2C
2
P
4
+ Cl
2
+ x C
2
! 2C
2
P
3
Cl + P
2
C
C
2
P
4
+ PCl

+ x C
2
! C
2
P
3
Cl + P
2
C
LllmlnaLe by
boundlng
The General Formulation
Clven seLs of producLs and raw maLerlals
WrlLe down Lhe general sLochlomeLrlc balance, lnclude anyLhlng LhaL could be boLh a
producL and a reacLanL on boLh sldes of Lhe equauon, use one sLochlomeLrlc
coemclenL of producL as 1 (plcklng basls as 1 mole of LhaL producL.) 1here wlll be
M
S1CCP
-1 coemclenLs.
WrlLe down as many A1CM 8ALAnCLS as you have aLom Lypes (L.g. P, Cl, C, C),
n
A1CM
use Lhe molecular formula of basls producL as consLanL added Lo equauon.
Solve for (M
S1CCP
- n
A1CM
) coemclenLs ln Lerms of Lhe oLhers.
HCl O H Cl H C O HCl Cl H C ! " # $ % & + + ' + + +
2 3 2 2 2 4 2
M
S1CCP
= 6 n
A1CM
= 4 (C, P,Cl,C)
All coemclenLs posluve, usual convenuon on slgns ln balances (+ve producL, -ve reacLanL)
The General Formulation
0
0
1
3
2
0 1 2 0 0 0
1 0 0 1 2 0
1 2 0 1 0 4
0 0 0 0 0 2
=
!
!
!
!
"
#
$
$
$
$
%
&
+
!
!
!
!
!
!
!
!
"
#
$
$
$
$
$
$
$
$
%
&
!
!
!
!
"
#
$
$
$
$
%
&
'
' '
' '
'
(
)
*
+
,
-
4x6 maLrlx ! 2 uegrees of lreedom
Solve for 2 coemclenLs ln Lerms of Lhe oLher 4.
other. get the one fix
d, constraine be will - given, are , then specify
2
1
) 2 1 ( ) 1 ( 0 2 2 2
2 from 3 subtract
2 0 2
0 1 2
0 1 2 0 3 2 4
1
! " # $ %
% % # $ # $
% # # %
" ! $
" # ! " # ! &
&
' ( ) = ) = ( = ) +
= ( = + )
= + + ) )
= ) + + ) ( = + + + ) )
=
HCl O H Cl H C O HCl Cl H C ! " # $ % & + + ' + + +
2 3 2 2 2 4 2
O H Cl H C O HCl H C
2 3 2 2 4 2
2
1
: Gives
1 0
1 1 , 0
2
1
+ ! + +
= " =
# = # " = = " =
$ %
$ % & ' (
Lxample:
Degrees of Freedom
1hls ls whaL we x ln deslgn!
WlLhouL degrees of freedom - Lhere would be nC uLSlCn 8C8LLM
Pow do we nd Lhe values of Lhe degrees of freedom?
1hls ls where opumlzauon and more generally search comes ln.
Maxlmlze Lhe proL upper bound
u8 acLs as our ob[ecuve funcuon
Sub[ecL Lo:
ALom 8alances
SLochlomeLrlc CoemclenLs >= 0
8alances acL as consLralnLs
Summary
1here are many posslble deslgns Lo screen and Lo selecL from - deslgn ls abouL
maklng declslons eecuvely.
1ry Lo nd boundlng schemes Lo ellmlnaLe deslgns as qulckly as posslble
roL upper 8ound ls a very slmple scheme Lo underesumaLe cosLs.
uegrees of lreedom are whaL we x durlng deslgn.

ueslgns can add degrees of freedom ln dlerenL ways (e.g. by addlng more
posslble producLs and raw maLerlals, dlerenL Lypes of unlLs, dlerenL
reacuon schemes, eLc)
Cpumlzauon problems x Lhe degrees of freedom Lhrough an ob[ecuve
funcuon whlch ls sub[ecL Lo consLralnLs of dlerenL Lypes.
We need ways Lo search over Lhe uegrees of lreedom, Lhe problem can be
asLronomlcally blg (so boundlng.)

Anda mungkin juga menyukai