Vous êtes sur la page 1sur 5

TheSAVEITCOMPANYoperates areclamationcenterthatcollects fourtypesofsolid

waste materials and treats them so thatthey can be amalgamated into a salable product.
(Treatingandamalgamatingareseparateprocesses.)Threedifferentgradesofthisproduct
canbemade(seethefirstcolumnofTable3.16),dependinguponthemixofthematerials
used.Althoughthereissomeflexibilityinthemixforeachgrade,qualitystandardsmay
specifytheminimumormaximumamountallowedfortheproportionofamaterialinthe
productgrade.(Thisproportionistheweightofthematerialexpressedasapercentageofthe
totalweightfortheproductgrade.)Foreachofthetwohighergrades,afixedpercentageis
specifiedforoneofthematerials.ThesespecificationsaregiveninTable3.16alongwiththe
costofamalgamationandthesellingpriceforeachgrade.

Thereclamationcentercollectsitssolidwastematerialsfromregularsourcesandso
isnormallyabletomaintainasteadyratefortreatingthem.Table3.17givesthequanti
tiesavailableforcollectionandtreatmenteachweek,aswellasthecostoftreatment,for
eachtypeofmaterial.
TheSaveItCo.issolelyownedbyGreenEarth,anorganizationdevotedtodealing
withenvironmentalissues,soSaveItsprofitsareusedtohelpsupportGreenEarthsac
tivities.GreenEarthhasraisedcontributionsandgrants,amountingto$30,000perweek,
tobeusedexclusivelytocovertheentiretreatmentcostforthesolidwastematerials.The
boardofdirectorsofGreenEarthhasinstructedthemanagementofSaveIttodividethis
moneyamongthematerialsinsuchawaythat atleasthalfoftheamountavailableof
eachmaterialisactuallycollectedandtreated.Theseadditionalrestrictionsarelistedin
Table3.17.
WithintherestrictionsspecifiedinTables3.16and3.17,managementwantstode
terminetheamountofeachproductgradetoproduceandtheexactmixofmaterialstobe
usedforeachgrade.Theobjectiveistomaximizethenetweeklyprofit(totalsalesin
come minus totalamalgamationcost),exclusiveofthefixedtreatmentcostof$30,000
perweekthatisbeingcoveredbygiftsandgrants.
Formulation as a Linear Programming Problem. Beforeattemptingtoconstruct a
linearprogrammingmodel,wemustgivecarefulconsiderationtotheproperdefinitionof
thedecisionvariables.Althoughthisdefinitionisoftenobvious,itsometimesbecomes
TABLE 3.16 Product data for Save-It Co.
Amalgamation
Cost per Pound ($)

Selling Price
per Pound ($)

Material 1: Not more than 30% of total


Material 2: Not less than 40% of total
Material 3: Not more than 50% of total
Material 4: Exactly 20% of total

3.00

8.50

Material 1: Not more than 50% of total


Material 2: Not less than 10% of total
Material 4: Exactly 10% of total

2.50

7.00

Material 1: Not more than 70% of total

2.00

5.50

Grade

Specification

54

3 INTRODUCTION TO LINEAR PROGRAMMING

TABLE 3.17 Solid waste materials data for the Save-It Co.
Material

Pounds per
Week Available

Treatment Cost
per Pound ($)

1
2
3
4

3,000
2,000
4,000
1,000

3.00
6.00
4.00
5.00

Additional Restrictions
1. For each material, at least half of the
pounds per week available should be
collected and treated.
2. $30,000 per week should be used to
treat these materials.

thecruxoftheentireformulation.Afterclearlyidentifyingwhatinformationisreallyde
siredandthemostconvenientformforconveyingthisinformationbymeansofdecision
variables,wecandeveloptheobjectivefunctionandtheconstraintsonthevaluesof
thesedecisionvariables.
Inthisparticularproblem,thedecisionstobemadearewelldefined,buttheappro
priatemeansofconveyingthisinformationmayrequiresomethought.(Tryitandseeif
youfirstobtainthefollowinginappropriatechoiceofdecisionvariables.)
Becauseonesetofdecisionsisthe amount ofeachproductgradetoproduce,it
would seem natural to define one set of decision variables accordingly. Proceeding
tentativelyalongthisline,wedefine
yi_numberofpoundsofproductgradeiproducedperweek

(i_A,B,C).

Theothersetofdecisionsisthe mix ofmaterialsforeachproductgrade.Thismixis


identifiedbytheproportionofeachmaterialintheproductgrade,whichwouldsuggest
definingtheothersetofdecisionvariablesas
zij_proportionofmaterialjinproductgradei

(i_A,B,C;j_1,2,3,4).

However,Table3.17givesboththetreatmentcostandtheavailabilityofthematerialsby
quantity(pounds)ratherthanproportion,soitisthisquantityinformationthatneedsto
berecordedinsomeoftheconstraints.Formaterialj(j_1,2,3,4),
Numberofpoundsofmaterialjusedperweek_zAjyA_zBjyB_zCjyC.
Forexample,sinceTable3.17indicatesthat3,000poundsofmaterial1isavailableper
week,oneconstraintinthemodelwouldbe
zA1yA_zB1yB_zC1yC_3,000.
Unfortunately,thisisnotalegitimatelinearprogrammingconstraint.Theexpressionon
thelefthandsideisnotalinearfunctionbecauseitinvolvesproductsofvariables.There
fore,alinearprogrammingmodelcannotbeconstructedwiththesedecisionvariables.
Fortunately,thereisanotherwayofdefiningthedecisionvariablesthatwillfitthe
linearprogrammingformat.(Doyouseehowtodoit?)Itisaccomplishedbymerelyre
placingeachproductoftheolddecisionvariablesbyasinglevariable!Inotherwords,
define
xij_zijyi
(fori_A,B,C;j_1,2,3,4)
_numberofpoundsofmaterialjallocatedtoproductgradeiperweek,

3.4 ADDITIONAL EXAMPLES

55

andthenweletthexijbethedecisionvariables.Combiningthexijindifferentwaysyieldsthefollowingquantitiesneeded
inthemodel(fori_A,B,C;j_1,2,3,4).
xi1_xi2_xi3_xi4_numberofpoundsofproductgradeiproducedperweek.xAj_xBj_xCj_numberofpounds
ofmaterialjusedperweek.
x

ij

x _x _x _x

____proportionofmaterialjinproductgradei.

i1 i2 i3 i4

Thefactthatthislastexpressionisanonlinearfunctiondoesnotcauseacomplication.Forexample,considerthe
firstspecificationforproductgrade A inTable3.16(theproportionofmaterial1shouldnotexceed30percent).This
restrictiongivesthenonlinearconstraint
x
A1

x _x _x _x

___ _0.3. A1 A2 A3 A4
However,multiplyingthroughbothsidesofthisinequalitybythedenominatoryieldsanequivalentconstraint
xA1_0.3(xA1_xA2_xA3_xA4),
so
0.7xA1_0.3xA2_0.3xA3_0.3xA4_0,
whichisalegitimatelinearprogrammingconstraint.
Withthisadjustment,thethreequantitiesgivenaboveleaddirectlytoallthefunctionalconstraintsofthemodel.The
objective function is based on managements objective of maximizing net weekly profit (total sales income minus total
amalgamationcost)fromthethreeproductgrades.Thus,foreachproductgrade,theprofitperpoundisobtainedbysubtracting
theamalgamationcostgiveninthethirdcolumnofTable3.16fromthesellingpriceinthefourthcolumn.These differences
providethecoefficientsfortheobjectivefunction.

Therefore,thecompletelinearprogrammingmodelis
MaximizeZ_5.5(xA1_xA2_xA3_xA4)_4.5(xB1_xB2_xB3_xB4)_3.5(xC1_xC2_xC3_xC4),
subjecttothefollowingconstraints:
1.Mixturespecifications(secondcolumnofTable3.16):
xA1 _ xA2 _ 0.3(xA1 _ xA2 _ xA3 _ xA4) 0.4(xA1 _ xA2 _ xA3 _ xA4)
xA3_xA4_ 0.5(xA1_xA2_xA3_xA4)0.2(xA1_xA2_xA3_xA4)

(gradeA,material1)

xB1 _ xB2 _ 0.5(xB1 _ xB2 _ xB3 _ xB4) 0.1(xB1 _ xB2 _ xB3 _ xB4)
x _
0.1(xB1_xB2_xB3_xB4)
B4

(gradeA,material3)

x _

(gradeB,material1)

C1

0.7(xC1_xC2_xC3_xC4)

(gradeA,material2)
(gradeA,material4).
(gradeB,material2)
(gradeB,material4).
(gradeC,material1).

56

3 INTRODUCTION TO LINEAR PROGRAMMING

2.Availabilityofmaterials(secondcolumnofTable3.17):

x _x _x

A1 B1 C1

x _x _x

A2 B2 C2

x _x _x

A3 B3 C3

x _x _x

_3,000

(material1)

_2,000
_4,000

(material2)
(material3)

_1,000
(material4).
3.Restrictionsonamountstreated(rightsideofTable3.17):
A4 B4 C4

x _x

_x

_x

_x

_x

_x

x _x

_x

A1

A2
A3

B1

B2
B3

C1

C2
C3

_1,500

(material1)

_1,000
_2,000

(material2)
(material3)

C4
A4 B4
_500 (material4).
4.Restrictionontreatmentcost(rightsideofTable3.17):

3(xA1_xB1_xC1)_6(xA2_xB2_xC2)_4(xA3_xB3_xC3)_
5(xA4_xB4_xC4)_30,000.
5.Nonnegativityconstraints:
xA1_0,

xA2_0,

...,

xC4_0.

Thisformulationcompletesthemodel,exceptthattheconstraintsforthemixture
specificationsneedtoberewrittenintheproperformforalinearprogrammingmodelby
bringingallvariablestothelefthandsideandcombiningterms,asfollows:
Mixturespecifications:
0.7xA1 _0.3xA2_0.3xA3

_0.3xA4_0
_0.4xA1_0.6xA2_0.4xA3_0.4xA4_0
_0.5xA1 _0.5xA2_0.5xA3 _0.5xA4 _0
_0.2xA1_0.2xA2_0.2xA3_0.8xA4_0

(gradeA,material1)
(gradeA,material2)
(gradeA,material3)
(gradeA,material4).

0.5xB1 _0.5xB2_0.5xB3 _0.5xB4 _0


_0.1xB1_0.9xB2_0.1xB3_0.1xB4_0
_0.1xB1_0.1xB2_0.1xB3_0.9xB4_0

(gradeB,material1)
(gradeB,material2)
(gradeB,material4).

0.3xC1_0.7xC2_0.7xC3_0.7xC4

_0

(gradeC,material1).

AnoptimalsolutionforthismodelisshowninTable3.18,andthenthesexijvalues
areusedtocalculatetheotherquantitiesofinterestgiveninthetable.Theresultingop
timal value of the objective function is Z _ 35,108.90 (a total weekly profit of
$35,108.90).
TheSaveItCo.problemisanexampleofablendingproblem.Theobjectivefora
blendingproblem istofindthebestblendofingredientsintofinal productstomeet
certainspecifications.Someoftheearliestapplicationsoflinearprogrammingwerefor
gasolineblending,wherepetroleumingredientswereblendedtoobtainvariousgradesof

gasoline.TheawardwinningORstudyatTexacodiscussedattheendof
3.4 ADDITIONAL EXAMPLES

57

TABLE 3.18 Optimal solution for the Save-It Co. problem


Pounds Used per Week
Material
Grade

412.3
(19.2%)
2587.7
(50%)
0

B
C
Total

3000

2
859.6
(40%)
517.5
(10%)
0
1377

3
447.4
(20.8%)
1552.6
(30%)
0
2000

4
429.8
(20%)
517.5
(10%)
0

Number of Pounds
Produced per Week
2149
5175
0

947

Sec.2.5dealtwithgasolineblending(althoughTexacousedanonlinearprogrammingmodel).Otherblendingproblems
involvesuchfinalproductsassteel,fertilizer,andanimalfeed.

Vous aimerez peut-être aussi