Vous êtes sur la page 1sur 8

tractability toanalysis

(McMasterUniversity) (McMasterU
Sangat akurat
berbagai
kondisi lebih
kompleks Lagi
yang lebih
perhitungan

Itu keabsahan convenientanalysis-Bagaimana


Itu tractability
ofamodelisthedegreetowhichinferencesdrawnfrom
ofamodelisthedegreetowhichthismodeladmits
themodelholdfortherealsystem
TheImportanceofaGoodFormulation
TheStandardFormulation
GraphicSolutionandOptimizationOutcomes
1 2

Garis besar 1 2 3 Validityvs.Tractability Tradeo ff:optimizationmodelsandtheir


Decisionmakersalmostalwaysconfronta tradeo ff anta
BenoıtChachuat BenoıtChac

1/31 3/31
fi nditusefultoformulatetheo
tidak dari theproblem!

4G03 4G03

falseconclusions
FormulatinganOptimizationProblem Formulating

Theconclusionsaredrawnfrom
themodel
program),
(matematika
Sebuah
leadsto
inadequatemodel
Themodelmustbe
computationallytractable
analysismustbepractical!
khas :nya
1 2 3

penilaian

Pemodelan Matematika
McMasterUniversity

berbasis odel
program Solusi

DepartmentofChemicalEngineering (McMasterUniversity) (McMasterU

memecahkan

BenoıtChachuat < benoit@mcmaster.ca>


Kesimpulan

FormulatinganOptimizationProblem
ChE4G03: OptimizationinChemicalEngineering
Optimization

Masalahuntuk
Keputusan

TheImportanceofaGoodFormulation
M Prosedur: Analisis
Dofasco:
“Kamisolveit ...”
BenoıtChachuat BenoıtChac
masalah

(McMasterUniversity) (McMasterU

modifythebehaviorofthesystem
Variabel selectedfortheindependentvariables
independen
Dependentvariables
, whosevaluescanbechanged
, whosebehaviorisdeterminedbythevalues
secara bebas untuk
◮ ◮

Pada
kategori: Suchagroupinghelpsunderstanding!
involvingmanyvariables
formulationstage . decisionvariablescanbegroupedintotwo
Padamenjadi dibedakan :Desain:
thesolutionmethodjustseesanoptimization
Operasi:
reactorvolume,
Pengelolaan:
temperatur,
numberoftrays,
feedtype,
aliran, tekanan,
purchaseprice, nilai-nilai
heatexchangerarea,
valveopening,
SalesPrice,
Sebuah
dll.
dll.dll. (misalnya,
selang semua-atau-tidak, baik-ordecisions)
Sebuah
1 2 1 2

DecisionVariables, x ExamplesofDecisionVariables: Discretevs.ContinuousDecisionVariables:


DecisionVariables, x ( lanjutan) heuristik:
mixed-integerprograms
Whenthereisanoption,
Contoh:
havenopractica
modelingwith
todiscretevariables Variablema
(Fortractabi
Problemswithbothdiscreteandcontinuousvariablesarecalle
BenoıtChachuat BenoıtChac

5/31 7/31

masalah decisionvariables andseekvaluesthatmaximize (atau

4G03 4G03

FormulatinganOptimizationProblem Olahraga:
Mengenali
tergantung
independen
variabeldan Formulating

Fungsi objektif
Equalityconstraints
Inequalityconstraints
Variablebounds

←- ←- ←- ←-

kapak

(McMasterUniversity) (McMasterU
xm

(x) = 00 x≤
f ( x) ≤

max x st h g (x) ≤x m
objectivefunctions ofthedecisionvariablessubjectto

StandardModel:
Optimizationmodels
choicesas
memperkecil)
kendala
OptimizationModelFormulationpilihan
(disebut
onvariablevaluesexpressingthelimitsonpossibledecision
juga mathematicalprograms )mewakili DecisionVariables, x ( lanjutan)
BenoıtChachuat BenoıtChac
untuk beberapa saya ∈ { 1, ..., n}
1, ..., 100} ∀i=
.
kapak x saya
:i= 0,
= ≥
kapak, saya
{ x saya xm
. x saya
withtheindex saya xm xm
(McMasterUniversity) = (McMasterU

x m, saya

(atau subskrip represents100similarvalueswiththesame


quantitieswithasinglesymbol,
indeks fi appropriateindexesforthedi
Itumisalnya
) permitrepresentingcollectionsofsimilar
mereka rststep
multipleindexes x ff
nama, membedakan
erentdimensionsoftheproblem;
informulatingalargeoptimizationmodelistochoose
areextremelycommon! Settingtheminandmaxvaluesequalsetsthevariabletoaconstan
nilai: Themostcommonvariableboundconstraintis
1 2 1 2

pengindeksan:
Inrealapplications,
sometimesmillions,
DecisionVariables, xkeeplargemodelsmanageable!
( lanjutan) ofvariables: indexednotationalschemes areneededto Variablebounds
VariableBounds, specifythedomainofde nitionfordecisionvariables fi: yang
setofvaluesforwhichthevariableshavemeaning
BenoıtChachuat BenoıtChac

9/31 11/31

tocustomersin
istocarryouttheseproduction (penggunaan

4G03
4G03

Letthethreemaindimensionsinthisproblem:

20facilitiesproducing
FormulatinganOptimizationProblem Formulating

largemanufacturerofcornseedoperates ℓ =
anddistributionoperationsatminimum.
25hybridcornvarietiesanddistributesthem
1, ..., ℓ)

1, ..., m)
1, ..., n)

(McMasterUniversity) (McMasterU

=
hybridvarietynumber
productionfacilitynumber ( f( =v =( r =
salesregionnumber

theoptimaltemperatureofachemicalprocess
thewarehouseslotassignedaparticularproduct
whetheracapitalprojectisselectedforinvestment
ballbearingsinaplantthatmanufactures10,000
thenumberofaircraftproducedonadefensecontract
+ / hari f= v= r=
30regions.Theoptimizationproblem Chooseappropriatedecis
pengindeksan).
Whatisthetota
1 2 3 4 5 1 2

DecisionVariables, xbestemployedtomodeleachofthefollowingquantities:
Olahraga: Decidewhetheradiscreteoracontinuousvariablewouldbe
( lanjutan) Olahraga: seedsof
Sebuah
DecisionVariables, nx =( mbiaya bedenotedas
lanjutan) pertanyaan:
BenoıtChachuat BenoıtChac
0and ≤
0 (x) = 0

0, g (x) ≤ 0, h

(McMasterUniversity) (McMasterU

indeksasi isusedforconstraintstoo!)
optimasi
Theseare (
Lihat kenapa? feasibleregion oran unboundedsolution
Inequalityconstraintsoftheform
Therecanbemanyoftheseinequalities,
Wemustbecarefultopreventde sothat g (x) isavector Konvensi
Thesedescribe ( indeksasi
Equalityconstraintsarewrittenwitha
interaksi isusedforequalityconstrain
decisionvariablesinthemode
Therecanbemanyoftheseinequalities,
betweenthevariablesinthemodel
Therecan tidakzeroright-handside
bemore (indepen
sotha
1 2 3 4 1 2 3 4

Mainconstraints
interaksi, g
Mainconstraints, InequalityConstraints:
ofoptimizationmodelsspecifytherestrictionsand
otherthanvariablebounds,
(x) ≤ thatlimitdecisionvariablevalues. EqualityConstraints:
Mainconstraints, g (x) ≤
BenoıtChachuat BenoıtChac

13/31 15/31

4G03 4G03

h (x) = 0 ( lanjutan)

hatlimitsthepossiblesolutionstotheproblem?

Olahraga:
Proposeboundsfor
variablesintheprocess FormulatinganOptimizationProblem Formulating

1, ..., N t
Aliran rateduetopumplimit
(lanjutan) fl ow rateontrays t =
kapak
0,
xm
.
xm
(McMasterUniversity) (McMasterU

Keamanan
Produkkualitas
Equipmentdamage
Equipmentoperation
(kontrak)
Hukum /(jangka
ethicalconsiderations
Max.investmentavailable
panjang)
Max. Min.liquid
Max.pressureinaclosedve
Max.regionwithinwh
diterima

VariableBounds, W
Mainconstraints, g (x) ≤ contoh:
BenoıtChachuat BenoıtChac

dua sumber, SaudiArabiaandVenezuela,
ofeachbarrelislosttore
fi nerydistillscrudepetroleum dari fi ning):

f ( x)

max x
(McMasterUniversity) (McMasterU

re

scalarobjective functionispreferredforsolving, walaupun


Kita butuh ataumultipleobjectivesaretypicalinreallife!
“Baik”SEBUAH
sebuah“buruk”
kuantitatif
Thereisnofundamentaleorpracticaldi
adalah
minproblems:
mengukur;
tidak memadai!
aqualitativemeasuresuchas
ff erencebetweenmaxand and0.2barreloflubricants; barel; Upto6,000barrelsperdayofSaudicru
bahan bakar, Upto9,000barrelsperdayofSaudicrudeareavaila
and0.3barreloflubricants.
EachbarrelofSaudicrudeyields0.3barrelofgasoline,
EachbarrelofVenezuelancrudeyields0.4barrelofgasoline,
biaya0.4barrelofjetfuel,
$ 15perbarrel. 0.2barre

1 2 3

Fungsi objektif, Objectivefunctionsinoptimizationmodelstellus


f ( x) BenoıtChachuat
Casestudy:
Sebuah ModelFormulation
inchemicalcompositionandthusyielddi
ke tiga 10%
mainproducts: ff erentproductmixes
bensin, jetfuelandlubricants.Thetwocrudesdi
Thecrudesalsodi (sisa masa
ff er
ff erincostandavailability:
dayoflubrica
Contractswithindepend
barrelsperdayofga
BenoıtChac

17/31 19/31

4G03 4G03

Peringkat dari q
fi neascalarthatrepresentsperformance?
h (x) = 0 ( lanjutan)

} + { menghasilkan

) 0
q di luar FormulatinganOptimizationProblem Formulating

E RT =
{ Peringkat dari
2 FB
}- exp ( - -
dari
Ha ΔkT0

0, q di = =
Q r SEBUAH wede

} = { Peringkat (McMasterUniversity) 1, ..., n, andthecostassociatedwith


(McMasterU

i=
hubungan, egVLE
.
231 K
dari q

=
{ akumulasi
material, energi, saatpokok
ini, dll.hubungan,
saldo , egforconservedquantity
misalnya
keseimbangan
keputusan Tbytheengineer,
behaviorenforcedby
q:
s controller
misalnya ,FEgtemperatureset-point
SEBUAH Maximizepro
Maximizeproductquality
Mininizeenergyuse
fi t (ormininizecost)
uents Maximizesafety
FFL Minimizepollutinge
dan -Economics
lain-lain . Expresstheanswerasas
Expresstheanswe
1 2

ExamplesofEqualityConstraints:
Mainconstraints, g (x) ≤ Bagaimanaf ( x)
Fungsi objektif, Olahraga: Writetheexpressionforanobje
di semua
aku svariabel
c saya x saya
BenoıtChachuat BenoıtChac
0, x m

{ x: f ( x) =
0, h (x) = Δ=

Cz

(McMasterUniversity) (McMasterU

(Orfeasibleregion)
{ x: gofanoptimizationmodelisthe
(x) ≤
Δ= ofanobjectivefunction (inthedecisionvariablespa
S
,

modelThe fi rstissueingraphicsolutionistheThesetofallpointssatisfyinganequalityconstraintplotsasalineor
Verysimplegraphicsolutionhaveenoughpowertodealwith
feasibleset
acurveThesetofallpointssatisfyinganinequalityconstraintplotsasa
mungil
(in2-d)
: boundarylineorcurve,
withallpointsonwhicheversideoftheboundarysatisfythe
constraintasaninequality
wheretheconstraintholdswithequality,
Untuk
bersama
introducedintoaplot
fi ndoutthebestfeasiblepoint,
Objectivefunctionsarenormallyplottedinthesamecoordinate
sistem theobjectivefunctionmustbe
1 2 1 2

GraphingConstraints Thefeasibleset
collectionofchoicesfordecisionvariablessatisfying
S GraphingObjectiveFunctions Thecontour Cz
thelineorcurvepassingthroughpointshaving
BenoıtChachuat BenoıtChac

21/31 23/31

4 0 2

soastoful fi llthe ≤
4G03
2
)≤ 1
4G03

( x2| - x2
min. 2000 bl / hari min. 1500 bl / hari min. 500 bl / hari
3 2 1

2+ | x1
) 1

( x1
bahan bakar jet 2
bensin
pelumas

40% 20% FormulatinganOptimizationProblem Formulating

2 3 0 3
KILANG MINYAK

40% 20%
≤ = ≥
2
30% 30%
x2 x2 x2
.
+ + x1 1
x1
3 x1

3 2 1

(McMasterUniversity) (McMasterU

$ 20 / bl $ 15 / bl 2 3 0 3

max. 9000 bl / hari max. 6000 bl / hari


≤ ≥ ≥
2
x2 x2 x2
.
+ + x1 1
Arab
x1
3 x1
Saudi
Venezuela
3 2 1
Pertanyaan:
Casestudy: Formulateanoptimizationmodelinstandardform
requirementsinthemoste
ModelFormulation FFI cientmanner.
(Lanjutan) ExamplesofFeasibleRegions Olahraga: Plotthef
BenoıtChachuat BenoıtChac
f ( x),

)
*

f(x
inanoptimizationmodelistheobjective 3 x2 7,5
*
isafeasiblechoicefordecisionvariableswith (McMasterUniversity) 2
6
(McMasterU
* +
≤ 0 4,5

3 x1 x2
≥ 3

+ x2
.x 2 .
1,5

0
x1
max x1
Sebuah
beberapa
unik
tidakoptimalsolution
optimalsolutions
alternatif optimalsolutions
(unboundedorinfeasiblemodels) 1
st x 1
◮ ◮ ◮

functionvalueofanyoptimalsolutions:
graphicallyaspointslyingonthebest f
objectivefunctioncontourthatintersectsthefeasibleregion
Itu optimalvalue
Anoptimizationmodelcanhave
Anoptimizationmodelmayhave:
f hanya satu optimalvalue
1 2 3 4

optimum
Sebuah optimalsolution
objectivefunctionvalueatleastequaltothatofanyothersolution
satisfyingallconstraints.Foraminimizationproblem:
OptimizationOutcomes x Apakah yang optimalsolution? Apakah yang optimalvalue?
ExamplesofGraphicSolutions Olahraga: Deter
BenoıtChachuat BenoıtChac

25/31 27/31

2 2
) 4G03 ) 4G03

2 8 2 8
2+ ( x2 0 2+ ( x2 0
) ≤ ) ≤
4,5 4,5

( x1 x2 ≥ 2 ( x1 x2 ≥ 2

x. 2 + x2 x. 2 + x2
. 0,5 . 0,5
x1
max x1 x1
max x1
2 1 FormulatinganOptimizationProblem 2 1 Formulating

st x 1 st x 1

x2 x2
2 2
+ 0 6
(McMasterUniversity)
+ 0 6
(McMasterU

≤ 2 ≤ 2

3 x1 x2
≥ 4,5
3 x1 x2
≥ 4,5

+ x2 1
3
+ x2 1
3
x
. 2 . x. 2 .
min
x1 x1 1,5 minx1 x1 1,5

0 0
st x 1 st x 1
2 1 2 1

ExamplesofGraphicSolutions Olahraga: Plottheobjectivefunctioncontoursinthetwo-crudecasestudy.


Apakah yang optimalsolution? Apakah yang optimalvalue?
ExamplesofGraphicSolutions
BenoıtChachuat BenoıtChac
ThisisanisothermalCFSTRwith
theseriesreaction:

cC
.
cB
.

T, c SEBUAH
(McMasterUniversity)

keluar
c
T

Fc
.
di SEBUAH
di c
T
di, c

ClassExercise:
RecognizingOptimizationOpportunity
F, T
BenoıtChachuat

29/31 31/31

di di
fl ow rateoffeed F fl ow rateoffeed F
Whatisoptimum forthefollowingprocesssystem? Whatisoptimum forthefollowingprocesssystem?
4G03
C 4G03

B→
BA →
SEBUAH →

FormulatinganOptimizationProblem SEBUAH → Formulating

Thegoalisto
engkauOnlythe
memaksimalkan
FFL
canbeadjusted
uentatsteadystate
CB Thegoalisto
engkauOnlythe
memaksimalkan
FFL
canbeadjusted
uentatsteadystate
Whatothervariablewouldy
liketoadjust?
CB

ThisisanisothermalCFSTRwith
theseriesreaction: ThisisanisothermalCFSTRwith
theparallelreaction:

cC cC
. .
cB cB
. .

T, c SEBUAH T, c SEBUAH
(McMasterUniversity) (McMasterU

keluar
c keluar
c
T T

Fc Fc
. .
di SEBUAH di SEBUAH
di c di c
T T
di, c di, c

ClassExercise:
RecognizingOptimizationOpportunity
F, T ClassExercise:
RecognizingOptimizationOpportunity
F, T
BenoıtChachuat BenoıtChac

Vous aimerez peut-être aussi