CHAPTERONE

I
NTRODUCTI
ON

1.
1Def
ini
ti
onsofOper
ati
onsr
esear
ch
Becauseoft
hewi
descopeofappl
i
cat
ionofoper
ati
onsr
esear
chgi
vi
ngapr
eci
sedef
ini
ti
oni
s
di
ff
icul
t.Howev
er,
afewdef
ini
ti
onsofORar
egi
venbel
ow:

Oper
ati
onsi
stheact
ivi
ti
escar
ri
edouti
nanor
gani
zat
ion,
and
Resear
chi
sthepr
ocessofobser
vat
ionandt
est
ingchar
act
eri
zedbyt
hesci
ent
if
icmet
hod,
Si
tuat
ion,pr
obl
em st
atement
,modelconst
ruct
ion,v
ali
dat
ion,exper
iment
ati
on,candi
dat
e
sol
uti
ons.
Oper
ati
onsResear
chi
sasy
stemat
icanal
ysi
sofapr
obl
em t
hroughsci
ent
if
icmet
hod,car
ri
ed
outbyappr
opr
iat
especi
ali
stt
osol
vet
hepr
obl
em.

Oper
ati
onsr
esear
chi
sthesy
stemat
icappl
i
cat
ionofquant
it
ati
vemet
hods,
techni
quesandt
ool
s
t
otheanal
ysi
sofpr
obl
emsi
nvol
vi
ngt
heoper
ati
onofsy
stems.Dael
l
enbachandGeor
ge,
1978

Oper
ati
onsr
esear
chi
sessent
ial
l
yacol
l
ect
ionofmat
hemat
icalt
echni
quesandt
ool
swhi
chi
n
conj
unct
ionwi
thasy
stemsappr
oach,ar
eappl
i
edt
osol
vepr
act
icaldeci
sionpr
obl
emsofan
economi
corengi
neer
ingnat
ure.Dael
l
enbachandGeor
ge,
1978

Theset
wodef
ini
ti
onsi
mpl
yanot
herv
iewofORasbei
ngt
hecol
l
ect
ionofmodel
sandmet
hods,
whi
chhav
edev
elopedl
argel
yindependentofoneanot
her
.

Oper
ati
ons r
esear
ch ut
il
izes t
he pl
anned appr
oach (
updat
ed sci
ent
if
ic met
hod)and an
i
nter
disci
pli
nar
yteam i
nor
dert
orepr
esentcompl
exf
unct
ionalr
elat
ionshi
psasmat
hemat
ical
model
sfort
hepur
poseofpr
ovi
dingaquant
it
ati
vebasi
sfordeci
sion-
maki
nganduncov
eri
ng
newpr
obl
emsf
orquant
it
ati
veanal
ysi
s.Thi
eraufandKl
ekamp,
1975

Oper
ati
onsr
esear
ch i
sadeci
sion-
maki
ng t
oolt
hathasbeen char
act
eri
zed byt
heuseof
sci
ent
if
icknowl
edget
hroughi
nter
disci
pli
nar
yteam ef
for
tfort
hepur
poseofdet
ermi
ningt
he
bestut
il
izat
ionofl
i
mit
edr
esour
ces.H.A.Taha,
1976

Theset
wodef
ini
ti
onsr
efert
othei
nter
disci
pli
nar
ynat
ureofOR.Howev
er,
ther
eisnot
hingwhi
ch
canst
oponeper
sonf
rom consi
der
ingsev
eralaspect
soft
hepr
obl
em underconsi
der
ati
on.One
oft
hebestdef
ini
ti
onsi
sasf
oll
ows.

Oper
ati
onsr
esear
chdef
inedast
heappl
i
cat
ionofsci
ent
if
icmet
hods,t
echni
quesandt
ool
s,t
o
pr
obl
ems i
nvol
vi
ng t
he oper
ati
ons ofa sy
stem so as t
o pr
ovi
de t
hose i
n cont
roloft
he
oper
ati
onswi
thopt
imum sol
uti
onst
othepr
obl
ems.Cur
chman,
Ackof
fandAr
nof
f,1957.

Oper
ati
onsr
esear
chi
sthear
tofwi
nni
ngwar
swi
thoutact
ual
l
yfi
ght
ingt
hem.Aut
herCl
arke.Thi
s
def
ini
ti
onr
efer
stot
hemi
l
itar
yor
igi
noft
hesubj
ectwher
eteam ofexper
tswer
enotact
ual
l
y
par
ti
cipat
ingi
nmi
l
itar
yoper
ati
onsf
orwi
nni
ngt
hewarbutpr
ovi
dingadv
isor
yandi
ntel
l
ect
ual
suppor
tfori
mit
ati
ngst
rat
egi
cmi
l
itar
yact
ions.

Af
ewot
herdef
ini
ti
onsofORar
easf
oll
ows:

Oper
ati
onsr
esear
chhasbeendescr
ibedasamet
hod,
anappr
oach,
asetoft
echni
ques,
ateam
act
ivi
ty,acombi
nat
ionofmanydi
sci
pli
nes,anext
ensi
onofpar
ti
cul
ardi
sci
pli
nes(
mat
hemat
ics,
engi
neer
ing,
andeconomi
cs)
.Anewdi
sci
pli
ne,
avocat
ion,
evenar
eli
gion.I
tisper
hapssomeof
al
ltheset
hings.SLCook,
1977

Oper
ati
onsr
esear
chmaybedescr
ibedasasci
ent
if
icappr
oacht
odeci
sion-
maki
ngt
hati
nvol
ves
t
heoper
ati
onsofor
gani
zat
ional
syst
em.F.S.Hi
l
lerandGJLi
eber
man,
1980

Oper
ati
ons r
esear
ch i
s a sci
ent
if
ic met
hod ofpr
ovi
ding execut
ive depar
tment
s wi
tha
quant
it
ati
vebasi
sfordeci
sionsundert
hei
rcont
rol
.P.M.Mor
eandGEKi
mbal
l

Oper
ati
onsr
esear
chi
sappl
i
eddeci
siont
heor
y.I
tusessci
ent
if
ic,
mat
hemat
ical
,orl
ogi
calmeans
t
oat
temptt
ocopewi
tht
hepr
obl
emst
hatconf
rontt
heexecut
ive,
whens/
het
ri
est
oachi
evea
t
hrough-
goi
ngr
ati
onal
i
tyi
ndeal
i
ngwi
thher
/hi
sdeci
sionpr
obl
ems.D.W.Mi
l
lerandM KSt
an

Oper
ati
onsr
esear
chi
sasci
ent
if
icappr
oacht
opr
obl
em-
sol
vi
ngf
orexecut
ivemanagement
.H.
M Wagner

Ast
hedi
sci
pli
neofoper
ati
onsr
esear
chgr
ew,numer
ousnameswer
egi
vent
oit
,suchas
Oper
ati
onsAnal
ysi
s,Sy
stemsAnal
ysi
s,Deci
sionAnal
ysi
s,Quant
it
ati
veAnal
ysi
s,Management
Sci
ence,Deci
sion Sci
ence and et
c.Howev
er,each oft
hese emphasi
zes t
he quant
it
ati
ve
appr
oacht
otheanal
ysi
sandsol
uti
onofmanagementpr
obl
ems.

1.
3.Hi
stor
yofOper
ati
onsResear
ch(
OR)
Ev
ent
hough,
thepar
ti
cul
armodelandt
echni
queofORcanbet
racedbackt
omuchear
li
ert
imes,
i
tisgener
all
yagr
eedt
hatoper
ati
onsr
esear
chcamei
ntoexi
stenceasadi
sci
pli
nedur
ingWor
ld
WarI
Iwhent
her
ewasacr
it
icalneedt
o managescar
cer
esour
ces.Thet
erm ‘
oper
ati
ons
r
esear
ch’wasconf
inedasar
esul
tofr
esear
chonmi
l
itar
yoper
ati
onsdur
ingt
hiswar
.Si
ncet
he
wari
nvol
ved st
rat
egi
c and t
act
icalpr
obl
ems whi
ch wer
e gr
eat
ly compl
i
cat
ed,t
o expect
adequat
e sol
uti
ons f
rom i
ndi
vi
dual
s orspeci
ali
stsi
n a si
ngl
e di
sci
pli
ne was unr
eal
i
sti
c.
Ther
efor
e,gr
oupsofi
ndi
vi
dualwhocol
l
ect
ivel
ywer
econsi
der
edspeci
ali
stsi
nmat
hemat
ics,
economi
cs,st
ati
sti
csandpr
obabi
l
ityt
heor
y,engi
neer
ing,behav
ior
al,andphy
sicalsci
encewer
e
f
ormedasspeci
aluni
tswi
thi
nthear
medf
orcest
odealwi
thst
rat
egi
candt
act
icalpr
obl
emsof
v
ari
ousmi
l
itar
yoper
ati
ons.
I
mpor
tanceandappl
i
cat
ionofoper
ati
onsr
esear
ch

 Thef
iel
dofoper
ati
onsr
esear
chpr
ovi
desamor
epower
fulappr
oacht
odeci
sionmaki
ng
t
hanor
dinar
ysof
twar
eanddat
aanal
yti
cst
ool
s.
 Empl
oyi
ng oper
ati
ons r
esear
ch pr
ofessi
onal
s can hel
p compani
es achi
eve mor
e
compl
etedat
aset
s,consi
deral
lav
ail
abl
eopt
ions,pr
edi
ctal
lpossi
bleout
comesand
est
imat
eri
sk.
 Usecasest
odet
ermi
newhi
cht
echni
quesar
emostappr
opr
iat
etosol
vet
hepr
obl
em.
Oper
ati
onsr
esear
chcanbeappl
i
edt
oav
ari
etyofusecases,
incl
udi
ng:
 Schedul
i
ngandt
imemanagement
.
 Ur
banandagr
icul
tur
alpl
anni
ng.
 Ent
erpr
iser
esour
cepl
anni
ng(
ERP)andsuppl
ychai
nmanagement(
SCM)
.
 I
nvent
orymanagement
.
 Net
wor
kopt
imi
zat
ionandengi
neer
ing.
 Packetr
out
ingopt
imi
zat
ion.
 Ri
skmanagement
….et
c

pr
ocessofoper
ati
onsr
esear
ch
1.I
dent
if
yingapr
obl
em t
hatneedst
obesol
ved.
2.Const
ruct
ingamodel
aroundt
hepr
obl
em t
hatr
esembl
est
her
eal
wor
ldandv
ari
abl
es.
3.Usi
ngt
hemodel
toder
ivesol
uti
onst
othepr
obl
em.
4.Test
ingeachsol
uti
onont
hemodel
andanal
yzi
ngi
tssuccess.
5.I
mpl
ement
ingt
hesol
uti
ont
otheact
ual
probl
em.

Char
act
eri
sti
csofoper
ati
onsr
esear
ch

Ther
ear
ethr
eepr
imar
ychar
act
eri
sti
csofal
loper
ati
onsr
esear
chef
for
ts:
1.Opt
imi
zat
ion-Thepur
poseofoper
ati
onsr
esear
chi
stoachi
evet
hebestper
for
mance
undert
hegi
venci
rcumst
ances.Opt
imi
zat
ional
soi
nvol
vescompar
ingandnar
rowi
ngdown
pot
ent
ial
opt
ions.
 2.Si
mul
ati
on-Thi
sinv
olv
esbui
l
dingmodel
sorr
epl
i
cat
ionsi
nor
dert
otr
youtandt
est
sol
uti
onsbef
oreappl
yi
ngt
hem.
3.Pr
obabi
l
ityand st
ati
sti
cs-Thi
sincl
udesusi
ng mat
hemat
icalal
gor
it
hmsand dat
ato
uncov
erhel
p
f
uli
nsi
ght
sandr
isks,
maker
eli
abl
epr
edi
cti
onsandt
estpossi
blesol
uti
ons.

Adv
ant
agesofOper
ati
onsr
esear
ch
1.Enhancepr
oduct
ivi
ty
 Oper
ati
onResear
ch:
 hel
psi
nimpr
ovi
ngt
hepr
oduct
ivi
tyoft
heor
gani
zat
ions.
pr
ovi
de si
gni
fi
canti
nfor
mat
ion t
othe manager
s bef
ore maki
ng an i
mpor
tant
deci
si
on.
 hel
psi
nmaki
ngsmal
ldeci
sionsf
ori
mpor
tantdeci
sionsf
oranor
gani
zat
ion.
 al
sohel
psi
nper
for
mingday
-to-
dayt
asksl
i
kei
nvent
orycont
rol
,wor
kfor
cepl
anni
ng,
expansi
onoft
hebusi
ness,
inst
all
ati
on,
andup-
gradat
ionoft
echnol
ogy
.
 Ef
fect
iveandaccur
atedeci
sionmaki
nghel
psi
nimpr
ovi
ngt
hepr
oduct
ivi
tyoft
he
or
gani
zat
ion.
2.Li
nearpr
ogr
ammi
ng
ORpr
ovi
desmanyal
ter
nat
ivesf
oronepr
obl
em,
whi
chhel
pst
hemanagementt
ochooset
he
bestdeci
si
onandi
mpl
ementi
ttogetaposi
ti
veout
come.

Manager
scanev
aluat
ether
isksassoci
atedwi
theachsol
uti
onandcandeci
dewhet
hert
hey
wantt
ogowi
tht
hesol
uti
onornot
.
3.I
mpr
ovecoor
dinat
ion
 OR i
mpr
oves t
he coor
dinat
ion bet
ween di
ff
erentdepar
tment
s and empl
oyees of
anor
gani
zat
ion.Becauseoper
ati
onsr
esear
chf
ocusesont
hewhol
eor
gani
zat
ionand
doesnotf
ocusononedepar
tment
.
 Asar
esul
toft
hat
,manager
sofeachdepar
tmentknowwhatt
heyshoul
ddot
oachi
evea
commonobj
ect
iveoft
heor
gani
zat
ion.
 Ther
efor
e,manager
sofdi
ff
erentdepar
tment
scancoor
dinat
ewi
thoneanot
herbet
ter
 whensol
uti
onsar
eappl
i
edt
oal
lthedepar
tment
sal
toget
her
.

4.Lowerr
isksoff
ail
ure
 Oper
ati
onsr
esear
chl
ower
sthechancesoff
ail
ureaswi
tht
hehel
pofoper
ati
ons,and
r
esear
chmanager
sgett
oknowaboutal
ltheal
ter
nat
ivesol
uti
onsf
orasi
ngl
epr
obl
em.
 Al
lri
sksassoci
atedwi
thasol
uti
onar
eanal
yzedbef
orei
mpl
ement
ingi
t.Asar
esul
tof
whi
cht
her
iskoff
ail
urer
educesunl
esssomet
hingunexpect
edev
entt
akespl
ace.
5.Cont
rol
ont
hesy
stem
Oper
ati
onsr
esear
chhel
psi
nredef
ini
ngt
hecont
rol
ofasy
stem.
Oper
ati
onsr
esear
chpr
ovi
desi
n-dept
hknowl
edgeaboutapar
ti
cul
aract
ion,whi
chal
l
ows
manager
stot
akebet
tercont
rol
oft
hewor
k.
Theycancont
rolt
hei
rsubor
dinat
esi
nabet
terwayandcanmaket
hemostr
elev
antj
ob
doneonpr
ior
it
y.
I
n addi
ti
on t
othi
s,oper
ati
onsr
esear
ch al
so pr
ovi
desi
nfor
mat
ion aboutt
heexpect
ed
out
come.Hence,manager
sknow whatst
andar
dsofper
for
manceheshoul
dex
pectf
rom
hi
ssubor
dinat
es.
Usi
ngt
hisi
nfor
mat
ion,hecanmeasur
etheper
for
manceofempl
oyeesandcancompar
eit
wi
tht
hest
andar
dper
for
mance.
Li
mit
ati
ons/
Disadv
ant
agesofOper
ati
onsr
esear
ch
1.Cost
ly
• Thef
ir
standf
oremostdi
sadv
ant
ageofoper
ati
onsr
esear
chi
sit
shi
ghcost
.
• The oper
ati
ons r
esear
ch wor
ks on mat
hemat
icalequat
ions t
hatr
equi
re expensi
ve
t
echnol
ogyt
ocr
eat
ethem.
2.Technol
ogydependent
Anot
herl
i
mit
ati
onofoper
ati
onsr
esear
chi
sit
stechnol
ogydependence.
Themat
hemat
icalequat
ionscanonl
ybeanal
yzedoncomput
ers;
itsr
eli
anceont
echnol
ogy
makesi
tanon-
popul
archoi
ceofmanager
s.
3.Dependenceonexper
ts
• At
eam ofexper
tsi
srequi
redt
oper
for
m oper
ati
onsr
esear
ch.
• Ther
ear
evar
iousf
act
orsassoci
atedwi
tht
his,whi
chmakesoper
ati
onsr
esear
chan
unpopul
archoi
cef
ormanagement
.
• Forexampl
e,t
hesol
uti
onswi
l
lnotbeef
fect
iveandcan’
tbei
mpl
ement
edi
fther
eisa
 communi
cat
iongapbet
weenmanager
sandORexper
ts.
• Besi
dest
his,
allsol
uti
onswi
l
lbecomeusel
essandmi
ghtcausel
ossr
athert
hanbenef
iti
f
t
he3wr
ongi
nfor
mat
ioni
sshar
edwi
tht
heexper
ts.

4.Unquant
if
iabl
efact
ors
• Theef
fect
ivenessof
• sol
uti
onsdev
elopedusi
ngoper
ati
onsr
esear
chl
argel
ydependsont
hev
ari
ousf
act
ors.
•I
tiseasyt
omeasur
equant
if
iabl
efact
orsa
duset
hem f
ort
heoper
ati
onsr
esear
ch,
butt
hepr
obl
em ar
iseswheni
mpor
tantf
act
orsar
ein
unquant
if
iabl
efor
m.
• Unquant
if
iabl
efact
orsr
esul
tini
naccur
atesol
uti
ons.
5.Di
ff
icul
ttoi
mpl
ement
Thesol
uti
onsobt
ainedf
rom oper
ati
onsr
esear
char
edi
ff
icul
ttoi
mpl
ement
,asmostof
t
hem ar
eusual
l
yunr
eal
i
sti
c.
Somemodi
fi
cat
ionsar
erequi
redt
omaket
oimpl
ementt
hesol
uti
onswhi
chhampert
he
ef
fect
ivenessoft
hesol
uti
on.
2.LI
NEARPROGRAMMI
NG

2.
1.Basi
cConcept
sofLi
nearPr
ogr
ammi
ng(
LP)
Al
argenumberofdeci
sionpr
obl
emsf
acedbyabusi
nessmanageri
nvol
veal
l
ocat
ionof
r
esour
cest
ovar
iousact
ivi
ti
es,wi
tht
heobj
ect
iveofi
ncr
easi
ngpr
ofi
tsordecr
easi
ngcost
s,
orbot
h.
 Whenr
esour
cesar
einexcess,nodi
ff
icul
tyi
sexper
ienced.Butsuchcasesar
ever
yrar
e.
Pr
act
ical
l
yinal
lsi
tuat
ions,t
hemanagement
sar
econf
ront
edwi
tht
hepr
obl
em ofscar
ce
r
esour
ces.

Nor
mal
l
y,t
her
ear
esev
eralact
ivi
ti
est
oper
for
m butl
i
mit
ati
onsofei
theroft
her
esour
cesor
t
hei
rusepr
eventeachact
ivi
tyf
rom bei
ngper
for
medt
othei
rbestl
evel
.Thus,t
hemanager
hast
otakeadeci
sionast
ohow bestt
her
esour
cesbeal
l
ocat
edamongt
hev
ari
ous
act
ivi
ti
es.

The deci
si
on pr
obl
em becomescompl
i
cat
ed when t
he numberofr
esour
cest
hatar
e
r
equi
redt
obeal
l
ocat
edandt
her
ear
esev
eralact
ivi
ti
est
oper
for
m.I
nsuchcases,ev
enan
exper
iencedmanager
,inal
lli
kel
i
hood,maynotpr
ovi
det
her
ightanswer
.Int
hiscase,
mat
hemat
ical
progr
ammi
ngmet
hodshel
ptoal
l
ocat
etheav
ail
abl
escar
cer
esour
cesamong
compet
ingends(
act
ivi
ti
es)t
oachi
evet
heobj
ect
ivesoft
heor
gani
zat
ion.

Mat
hemat
icalpr
ogr
ammi
ngi
nvol
vesopt
imi
zat
ionofacer
tai
nfunct
ioncal
l
edt
heobj
ect
ive
f
unct
ionsubj
ectt
ocer
tai
nconst
rai
nts.
 Forexampl
e,a managermay be f
aced wi
tht
he pr
obl
em ofdeci
ding t
he
 appr
opr
iat
einput
/out
putmi
xoft
hef
ourpr
oduct
s.Wi
tht
hepr
ofi
tabi
l
ityoft
hepr
oduct
s
al
ongwi
tht
hei
r
 r
equi
rement
sofr
awmat
eri
alsknown,t
hemanagerpr
obl
em canbef
ormul
atedas
amat
hemat
icalpr
ogr
ammi
ngpr
obl
em t
aki
ngt
heobj
ect
ivef
unct
ionast
hemaxi
mizat
ion
ofpr
ofi
tsobt
ainabl
efr
om t
hemi
x,keepi
ngi
nvi
ewt
hev
ari
ousconst
rai
nts(
theav
ail
abi
l
ity
andmi
x)ofr
awmat
eri
als,
laborsuppl
y,mar
ketandsoon.

Themet
hodsofmat
hemat
icalpr
ogr
ammi
ngcanbedi
vi
dedi
ntodi
ff
erentgr
oups:l
i
near
,
i
nteger
,non-
li
near
,goal
,dy
nami
cpr
ogr
ammi
ngandot
her
s.

Thi
schapt
eri
sdev
otedt
oli
nearpr
ogr
ammi
ngwhi
l
enon-
li
nearpr
ogr
ammi
ngwi
l
lal
sobe
cov
eredi
nthel
att
erchapt
er.Thei
ntegerandgoalpr
ogr
ammi
ngwi
l
lber
oughl
ydi
scussed
underLPPassumpt
ionv
iol
ati
onsect
ion.Ont
heot
herhand,dy
nami
cpr
ogr
ammi
ngwi
l
lnot
beconsi
der
edatt
hisl
evel
.

2.
2.Def
ini
ti
onofLi
nearPr
ogr
ammi
ng
Li
nearPr
ogr
ammi
ng (
LP)i
sa mat
hemat
icalmodel
i
ng t
echni
que usef
ulf
oreconomi
c
al
l
ocat
ion of“
scar
ce”or“
li
mit
ed”r
esour
ces,such as l
abor
,mat
eri
al,machi
ne,t
ime,
war
ehousespace,capi
tal
,et
c.t
osev
eralcompet
ingact
ivi
ti
essuchaspr
oduct
s,ser
vices,
j
obs,
newequi
pment
,pr
oject
sandet
c,ont
hebasi
sofagi
vencr
it
eri
onofopt
imal
i
ty.

 I
tisat
echni
quewhi
chi
nvol
vest
hepl
anni
ngofact
ivi
ti
est
oobt
ainanopt
imal
sol
uti
onofpr
obl
emsl
i
kedi
str
ibut
ionofcommodi
ti
es,
all
ocat
ionofmachi
nes,
schedul
i
ng
t
ranspor
tser
vicesandet
c…

LP i
sat
oolorat
echni
quebywhi
chonecansel
ectt
hebestal
ter
nat
ive(
theopt
imal
)
sol
uti
onf
rom asetoff
easi
bleal
ter
nat
ivesol
uti
ons,
insi
tuat
ionsi
nwhi
cht
heobj
ect
iveand
t
heconst
rai
ntf
unct
ionscanbeexpr
essedasl
i
nearmat
hemat
ical
funct
ions.
 Thewor
dli
nearr
efer
stol
i
nearr
elat
ionshi
psamongv
ari
abl
esi
nthemodel
.

Thi
stechni
quewasdev
eloped,byGeor
geB.Dant
zingwhi
l
ehewaswor
kingi
ntheUSAi
r
f
orcedur
ingWor
ldWarI
I,pr
imar
il
yforsol
vi
ngmi
l
itar
ylogi
sti
cspr
obl
ems.Butcur
rent
ly,
iti
s
bei
ngusedext
ensi
vel
yinal
lfunct
ionalar
easofmanagement
,ai
rl
ines,agr
icul
tur
e,mi
l
itar
y
oper
ati
ons, oi
lref
ini
ng, educat
ion, ener
gy, heal
thcar
e sy
stems, pol
l
uti
on cont
rol
,
t
ranspor
tat
ion,
planni
ngandschedul
i
ngr
esear
chanddev
elopment
,andet
c.

I
twasment
ionedt
hat
,al
thought
heseappl
i
cat
ionsar
edi
ver
se,al
lLPmodel
sconsi
stof
cer
tai
ncommonpr
oper
ti
esandassumpt
ions.Thesecommonpr
oper
ti
esandassumpt
ions
shoul
dbeconsi
der
edbyLPPanal
ystanddeci
sionmaker
sbef
oremodel
i
ngaLPt
echni
que
f
orsol
vi
ngar
eal
li
fedeci
sionpr
obl
em.

Ther
efor
e,i
nor
dert
oappl
yli
nearpr
ogr
ammi
ng,
ther
ear
ecer
tai
nrequi
rement
s(condi
ti
ons)
t
obemet
.Theser
equi
rement
sincl
ude:

a) Ther
eshoul
dbeanobj
ect
ivewhi
chshoul
dbecl
ear
lyi
dent
if
iabl
eandmeasur
abl
e
i
n quant
it
ati
vet
erms.I
tcoul
d be,f
orexampl
e,maxi
mizat
ion ofsal
es,ofpr
ofi
t,
mi
nimi
zat
ionofcost
,andsoon.
b) Theact
ivi
ti
est
obei
ncl
udedshoul
dbedi
sti
nct
lyi
dent
if
iabl
eandmeasur
abl
ein
quant
it
ati
vet
erms,
fori
nst
ance,
thepr
oduct
sincl
udedi
napr
oduct
ionpl
anni
ngpr
obl
em.
c) Resour
cesoft
hesy
stem whi
char
etobeal
l
ocat
ed,f
ort
heat
tai
nmentoft
hegoal
,
shoul
dal
sobei
dent
if
iabl
eandmeasur
abl
equant
it
ati
vel
y.Resour
cest
obeusedi
nthe
act
ivi
ti
esshoul
dal
sobel
i
mit
edi
nquant
it
y(r
esour
cesmustbei
nli
mit
edsuppl
y).Hence,
t
het
echni
quewoul
dinv
olv
eal
l
ocat
ionoft
heser
esour
cesi
nmannert
hatwoul
dtr
adeof
f
t
her
etur
nsont
hei
nvest
mentoft
her
esour
cesf
ort
heat
tai
nmentoft
heobj
ect
ive.
d) The r
elat
ionshi
ps r
epr
esent
ed by t
he obj
ect
ivef
unct
ion and t
he const
rai
nt
equat
ions/i
nequal
i
tiesmustbel
i
neari
nnat
ure.
e) Ther
eshoul
dbeaser
iesoff
easi
bleal
ter
nat
ivecour
seofact
ionsav
ail
abl
efort
he
deci
si
onmakert
hati
sdet
ermi
nedbyt
her
esour
ceconst
rai
nts.

2.
3.Basi
cassumpt
ionsofLi
nearPr
ogr
ammi
ng
Al
ltheassumpt
ionsofl
i
nearpr
ogr
ammi
ngpr
obl
em act
ual
l
yar
eimpl
i
citi
nthemodel
f
ormul
ati
on.Theseassumpt
ionsar
etheassumpt
ionofpr
opor
ti
onal
i
ty,
addi
ti
vi
ty,
cont
inui
ty,
f
ini
teness,
cer
tai
ntyandopt
imal
i
ty.Thesear
edi
scussedher
eunder
.
Pr
opor
ti
onal
i
tyassumpt
ioni
sanassumpt
ionaboutbot
htheobj
ect
ivef
unct
ionandt
he
f
unct
ionalconst
rai
ntswhi
chst
atest
hatt
hecont
ri
but
ionofeachact
ivi
ty(
Cj)t
othev
alueof
t
heobj
ect
ivef
unct
ion(
Z)i
spr
opor
ti
onalt
othel
eveloft
heact
ivi
ty(
Xj,
)asr
epr
esent
edby
t
heCjXjt
ermi
ntheobj
ect
ivef
unct
ion.Si
mil
arl
y,t
her
esour
ceconsumpt
ion(
aij)ofeach
act
ivi
ty(
Xj,
)ineachf
unct
ionalconst
rai
nti
spr
opor
ti
onalt
othel
eveloft
heact
ivi
ty(
Xj)
,as
r
epr
esent
edbyt
heaijxjt
ermi
ntheconst
rai
nt.

Forexampl
e,i
foneuni
tofapr
oduct(
Xj,
)isassumedt
ocont
ri
but
ebi
rr10t
owar
dpr
ofi
t,t
hen
t
het
otal
cont
ri
but
ionwoul
dbeequal
to10Xj:
wher
eXj,
ist
henumberofuni
tsoft
hepr
oduct
.
Thus,i
fthepr
oduct(
sal
es)i
sdoubl
ed,t
hent
hepr
ofi
twoul
dal
sobedoubl
ed.Si
mil
arl
y,i
f
oneuni
tof(
Xj,
)takes2hour
sofl
aborofacer
tai
nty
pe,
10uni
tswoul
drequi
re20hour
s...
andsoon.
 I
nef
fect
,then,pr
opor
ti
onal
i
tymeanst
hatt
her
ear
econst
antr
etur
nst
oscal
eand
t
her
ear
enoeconomi
esofscal
e.
 Whathappenswhent
hepr
opor
ti
onal
i
tyassumpt
iondoesnothol
dev
enasa
r
easonabl
eappr
oxi
mat
ion?I
nthi
scase,
wemustusenon-
li
nearpr
ogr
ammi
ngpr
obl
em.

Addi
ti
vi
tyassumpt
ionst
atest
hatev
eryf
unct
ionofaLP model(
whet
hert
heobj
ect
ive
f
unct
ionsort
hef
unct
ionont
hel
eft
-handsi
deofat
echni
calconst
rai
nt)i
sthesum oft
he
i
ndi
vi
dualcont
ri
but
ions/
resour
ceconsumpt
ionoft
her
espect
iveact
ivi
ti
es.Thus,t
het
otal
pr
ofi
tint
heobj
ect
ivef
unct
ioni
sdet
ermi
nedbyt
hesum oft
hepr
ofi
tcont
ri
but
edbyeachof
t
hepr
oduct
ssepar
atel
y.Si
mil
arl
y,t
het
otalamountofar
esour
ceusedi
sequalt
othesum
oft
her
esour
cev
aluesusedbyv
ari
ousact
ivi
ti
es.Thi
sassumpt
ioni
mpl
i
est
hatt
her
eisno
i
nter
act
ionamongt
hedeci
sionv
ari
abl
es(
int
eract
ioni
spossi
blewhen,f
orexampl
e,some
pr
oducti
saby
-pr
oductofanot
herone)
.
 Whathappenswhent
headdi
ti
vi
tyassumpt
iondoesnothol
dev
enasar
easonabl
e
appr
oxi
mat
ion?I
nthi
scase,
wemustusenon-
li
nearpr
ogr
ammi
ngpr
obl
em.
Di
vi
sibi
l
ity
/cont
inui
tyassumpt
ionst
atest
hatdeci
sionv
ari
abl
esi
nal
i
nearpr
ogr
ammi
ng
modelar
econt
inuous.i
.e.deci
sionv
ari
abl
esal
l
owedt
ohav
eanyv
alues,i
ncl
udi
ngnon-
i
ntegerv
alues,whi
chcansat
isf
ythef
unct
ionalandnon-
negat
ivi
tyconst
rai
nts.Thus,t
hese
v
ari
abl
esar
enotr
est
ri
ctedt
ojusti
ntegerv
alues.Si
nceeachdeci
sionv
ari
abl
erepr
esent
s
t
hel
evelofsomeact
ivi
ty,i
tisbei
ngassumedt
hatt
heact
ivi
ti
escanber
unatf
ract
ional
l
evel
s.
I
ncer
tai
nsi
tuat
ions,
thedi
vi
sibi
l
ityassumpt
ionmaynothol
dbecauseoft
hef
actt
hatsome
oral
lthedeci
si
onv
ari
abl
esmustber
est
ri
ctedt
oint
egerv
alues.I
nthi
scase,wemustuse
i
ntegerpr
ogr
ammi
ngmodel
s.

Cer
tai
ntyassumpt
ionconcer
nsaboutt
hepar
amet
ersoft
hemodel
,namel
y,t
hecoef
fi
cient
s
i
ntheobj
ect
ivef
unct
ionCj,t
hecoef
fi
cient
sint
het
echni
calconst
rai
ntsaij,andt
her
ight
-
handsi
desconst
ant
soft
het
echni
calconst
rai
ntsbi.I
tst
atest
hatt
hev
alueassi
gnedt
o
eachpar
amet
erofal
i
nearpr
ogr
ammi
ngmodeli
sassumedt
obeaknownconst
ant(
known
wi
thcer
tai
nty
)anddonotchangedur
ingt
heper
iodundert
hest
udy
.Hence,LPPobv
iousl
y
i
sdet
ermi
nist
ici
nnat
ure.Howev
er,
inr
ealappl
i
cat
ions,
thecer
tai
ntyassumpt
ioni
ssel
dom
sat
isf
ied pr
eci
sel
y.Ther
efor
e,t
he par
amet
erv
alues det
ermi
ned,whi
ch i
s based on
pr
edi
cti
onoff
utur
econdi
ti
ons,i
nev
itabl
yint
roducessomedegr
eeofuncer
tai
nty
.Fort
his
r
eason,
iti
susual
l
yimpor
tantt
oconductsensi
ti
vi
tyanal
ysi
saf
terasol
uti
oni
sfoundt
hati
s
opt
imal
undert
heassumedpar
amet
erv
alues.

Fi
nit
eness:Anopt
imum sol
uti
oncannotbecomput
edi
nthesi
tuat
ionswher
ether
ear
e
i
nfi
nit
enumberofal
ter
nat
iveact
ivi
ti
esandr
esour
cer
est
ri
cti
ons.
Opt
imal
i
ty:I
nli
nearpr
ogr
ammi
ng,
theopt
imalsol
uti
onal
way
soccur
satt
hecor
nerpoi
ntof
t
hesetoff
easi
blesol
uti
ons.

Whent
heabov
est
atedcondi
ti
onsandassumpt
ionsar
esat
isf
iedi
nagi
vensi
tuat
iont
he
pr
obl
em canbeexpr
essedi
nal
gebr
aicf
orm cal
l
edt
heLi
nearPr
ogr
ammi
ngPr
obl
em (
LPP)
,
andt
hensol
vedf
oropt
imaldeci
sion.Weshal
lfi
rsti
l
lust
rat
ethef
ormul
ati
onofaLPPand
t
henconsi
dert
hemet
hodoft
hei
rsol
uti
ons.

2.
4.For
mul
ati
onofal
i
nearpr
ogr
ammi
ngpr
obl
em
For
mul
ati
onofal
i
nearpr
ogr
ammi
ngpr
obl
em i
nvol
vest
hef
oll
owi
ngst
eps:
1. I
dent
if
ythedeci
sionv
ari
abl
es(
act
ivi
ti
es)
2. I
dent
if
ytheobj
ect
ivef
unct
ionasal
i
nearf
unct
ionofi
tsdeci
sionv
ari
abl
es
3. St
ateal
lresour
cel
i
mit
ati
onsasl
i
nearequat
ionsori
nequal
i
tiesofi
tsdeci
sion
v
ari
abl
es.
4. Add non-
negat
ivi
tyconst
rai
ntsf
rom t
heconsi
der
ati
ont
hatnegat
ivev
aluesof
deci
sionv
ari
abl
esdonothav
eanyv
ali
dphy
sical
int
erpr
etat
ion
5. Usemat
hemat
icalt
echni
quest
ofi
ndal
lpossi
bleset
sofv
aluesoft
hev
ari
abl
esor
unknowns,
sat
isf
yingt
heconst
rai
nts
6. Sel
ectt
hepar
ti
cul
arsetofv
aluesofv
ari
abl
esobt
ainedi
nst
epf
ivet
hatl
eadst
o
t
heachi
evementoft
heobj
ect
ivef
unct
ion
Ther
esul
toft
hef
ir
stf
ourst
epsi
scal
l
edl
i
nearpr
ogr
ammi
ngpr
obl
em.Thesetofsol
uti
ons
obt
ainedi
nst
epf
ivei
sknownast
hesetoff
easi
blesol
uti
onsandt
hesol
uti
onf
inal
l
y
sel
ect
edi
nst
epsi
xiscal
l
edopt
imum (
best
)sol
uti
onoft
hel
i
nearpr
ogr
ammi
ngpr
obl
em.

Fr
om t
hef
ir
stf
ourst
eps,wecanf
ormul
ateat
ypi
call
i
nearpr
ogr
ammi
ngpr
obl
em t
hathas
t
wopar
ts:
Thesear
e:
 Theobj
ect
ivef
unct
ionand
 Theconst
rai
ntsf
unct
ions:
• Techni
cal
/f
unct
ional
const
rai
ntand
• Thenon-
negat
ivi
tyconst
rai
nt.

The obj
ect
ivef
unct
ion:i
s a mat
hemat
icalr
epr
esent
ati
on oft
he ov
eral
lgoaloft
he
or
gani
zat
ionst
atedasal
i
nearf
unct
ionofi
tsdeci
sionv
ari
abl
es(
Xj)t
oopt
imi
zet
hecr
it
eri
on
ofopt
imal
i
ty.I
tisal
socal
l
edt
hemeasur
eofper
for
mancesuchaspr
ofi
t,cost
,rev
enue,
di
stance,
etc.

I
nit
sgener
alf
orm,
theobj
ect
ivef
unct
ioni
sexpr
essedas:
Opt
imi
ze(
Maxi
mizeorMi
ni ze)Z=∑CjXj
mi

Opt
imi
zeZ=C1X1+C2X2+.
..
.+CnXn
Wher
e:Zi
sthemeasur
eofper
for
mancev
ari
abl
e(pr
ofi
t/
cost
),whi
chi
sthef
unct
ionofX1,X
.
2,.
.Xn.
. (
quant
it
iesofact
ivi
ti
es)
.
C1,C2,
Cj( ..
.Cn)(
. par
amet
ersorcoef
fi
cient
s)r
epr
esentt
hecont
ri
but
ionofauni
tof
t
her
espect
ivev
ari esX1,X 2,
abl ..
.X n.t
. othemeasur
esofper
for
manceZ(
theobj
ect
ive
f
unct
i .Xjr
on) epr
esentt
hev
aluesofact
ivi
ti
est
hatt
ell
stheext
entt
owhi
cheachoft
hese
act
ivi
ti
esar
eper
for
med.
 I
naLPmodelal
ldeci
sionv
ari
abl
esar
econt
inuous,cont
rol
l
abl
eandnon-
negat
ive.
 i
.e.
X10,
X20.
..Xn0.Thus,
onl
ynon-
negat
ivel
evel
sofact
ivi
tyar
econsi
der
edf
easi
ble.For

exampl
e,i
nthepr
oduct
ionpr
obl
ems,
theout
putcannotobv
iousl
ybenegat
ive.

The const
rai
nts: al
way
s,t
her
e ar
e cer
tai
nli
mit
ati
ons(
orconst
rai
nts)on t
he use of
r
esour
ces,e.
g.l
abor
,machi
ne,r
aw mat
eri
al,space,money
,et
c.t
hatl
i
mitt
hedegr
eet
o
whi
chanobj
ect
ivecanbeachi
eved.
 Suchconst
rai
ntsmustbeexpr
essedasl
i
nearequal
i
tiesori
nequal
i
tiesi
nter
msof
deci
si
onv
ari
abl
es.

Thegener
alf
orm oft
heconst
rai
ntf
unct
ionsar
eexpr
essedas:
∑aijXj(
<,,
>, )bi
=,

a11X1+a12X2+−−−a1nXn(
<,,
=,)b1
>,
a21X1+a22X2+−−−a2nXn(
<,,
=,)b2
>,

M
am1X1+am2X2+−−−amnXn(
<,,
=,)bn
>,
 Thecoef
fi
cient
s,aij’
sar
ecal
l
edt
het
echni
calcoef
fi
cient
sandmeasur
etheperuni
t
consumpt
ionoft
her
esour
cesf
orexecut
ingoneuni
tofunknownv
ari
abl
e(act
ivi
ti
es)Xj.
 aij’
scan be posi
ti
ve,negat
ive orzer
oint
he gi
ven const
rai
nts.Whati
sthe
i
mpl
i
cat
ionoft
heposi
ti
ve,
negat
iveorzer
otechni
cal
coef
fi
cient
s?
t
h
 Thebir
epr
esent
sthet
otal
avai
l
abi
l
ityoft
hei r
esour
ce.
 I
ti hatbi0f
sassumedt oral
li.Howev
er fanybi<0,t
,i henbot
hsi
desoft
he
const
rai
nti
canbemul
ti
pli
edby-
1tor
ever
set
hei
nequal
i
tyoft
heconst
rai bi>0)
nt( .

2.
5.Met
hodsofsol
vi
ngaLi
nearPr
ogr
ammi
ngPr
obl
ems
ALPPcanbesol
vedusi
ngbygr
aphi
cmet
hodoral
gebr
aicmet
hod.Theal
gebr
aicmet
hodi
s
al
soknownbyt
henamesi
mpl
exmet
hod.

2.
5.1.Gr
aphi
cMet
hod
Thegr
aphi
cappr
oachi
srest
ri
ctedi
nit
sappl
i
cat
ion.I
tcanonl
ybeusedf
ort
hepr
obl
ems
whi
chi
nvol
veonl
ytwodeci
sionv
ari
abl
es.ForLPpr
obl
emst
hathav
eonl
ytwodeci
sion
v
ari
abl
es,
iti
spossi
blet
hatt
heent
ir
esetoff
easi
blesol
uti
onscanbedi
spl
ayedgr
aphi
cal
l
y
bypl
ott
ingal
i
nearconst
rai
ntsongr
aphpapert
olocat
ethebest(
opt
imal
)sol
uti
on.
 I
nthi
scase,anopt
ional/f
easi
blesol
uti
ont
oaLPPi
sobt
ainedbychoosi
ngf
rom
sev
eralv
alues ofdeci
sion v
ari
abl
es (
X1, X2,..Xn)t
hatsat
isf
ies t
he gi
ven setof
const
rai
ntssi
mul
taneousl
yandal
sopr
ovi
dest
heopt
imum (
maxi
mum ormi
nimum)v
alue
oft
heobj
ect
ivef
unct
ion.
 Howev
er,t
hismet
hodei
therf
ail
sorbecomest
oocompl
exf
orappl
i
cat
ioni
fthe
pr
obl
em hasmor
ethant
wov
ari
abl
es.
I
mpor
tantdef
ini
ti
ons
 Sol
uti
on:
-Thesetofv
aluesofdeci
si
onv
ari
abl
esXi(
i=1,
2...
,
n)whi
chsat
isf
ythe
const
rai
ntf
unct
ionsofaLPPi
ssai
dtoconst
it
utesol
uti
ont
othatLPP
 Feasi
blesol
uti
ons:Thesetofv
aluesofdeci
sionv
ari
abl
esXi(i=1,
2,...
,n)whi
ch
sat
isf
yal
ltheconst
rai
ntsandnon-
negat
ivi
tycondi
ti
onsofLPsi
mul
taneousl
y
 I
nfeasi
blesol
uti
on:Thesetofv
aluesofdeci
sionv
ari
abl
eswhi
chdonotsat
isf
yal
l
t
heconst
rai
ntsandnon-
negat
ivi
tycondi
ti
onsofLPpr
obl
em.I
not
herwor
ds,atl
eastone
const
rai
nti
svi
olat
ed.
 Opt
imum sol
uti
on:Af
easi
blesol
uti
onwhi
chopt
imi
zes(
maxi
mizesormi
nimi
zes)
t
heobj
ect
ivef
unct
ionv
alueoft
hegi
venLPpr
obl
em i
scal
l
edanopt
imum basi
cfeasi
ble
sol
uti
on.
 Unbounded sol
uti
on:A sol
uti
onwhi
chcani
ncr
easeordecr
easet
hev
alueof
obj
ect
ivef
unct
ionofLPpr
obl
em i
ndef
ini
tel
yiscal
l
edun-
boundedsol
uti
on.
Thegr
aphi
cal
sol
uti
onpr
ocedur
econsi
stsoft
hef
oll
owi
ngst
eps:
1. For
mul
ateaLPP(
ident
if
ythedeci
sionv
ari
abl
es,
obj
ect
ivef
unct
ion,
andconst
rai
nt
r
est
ri
cti
ons)
2. Gr
apht
heconst
rai
nts(
drawagr
apht
hati
ncl
udesal
ltheconst
rai
nts)
3. I
dent
if
ythesol
uti
onspace(
ident
if
ythef
easi
bler
egi
on)
;
4. Locat
ethesol
uti
onpoi
nts(
Ident
if
ythecor
nerpoi
ntsf
rom t
hesol
uti
onspace)
;
5. Det
ermi
net
heopt
imalsol
uti
on(
asol
uti
ont
hatopt
imi
zest
heobj
ect
ivef
unct
ion)
and
6.I
nter
prett
her
esul
ts
Thet
ermf
ormul
ati
oni
susedt
omeant
hepr
ocessofconv
ert
ingt
hev
erbaldescr
ipt
ionand
numer
icaldat
aint
omat
hemat
icalexpr
essi
onswhi
chr
epr
esent
sther
elev
antr
elat
ionshi
p
amongdeci
si
onf
act
ors,
obj
ect
ivesandr
est
ri
cti
onsont
heuseofr
esour
ces.

2.
5.1.
1.Formaxi
mizat
ionpr
obl
em
Thegr
aphi
calsol
uti
ont
otheLPPcanbeappl
i
edt
othepr
obl
emswhi
chhav
emaxi
mizat
ion
i
nnat
ure.

Exampl
e1.Af
ir
misengagedi
npr
oduci
ngt
wopr s,AandB.Eachuni
oduct tofpr
oductA
r
equi
res2kgofr
aw mat
eri
aland4l
abourhour
sforpr
ocessi
ng,wher
easeachuni
tof
pr
oductBr
equi
res3kgofr
awmat
eri
aland3hour
sofl
abour
,oft
hesamet
ype.Ev
eryweek,
t
hef
ir
m hasanav
ail
abi
l
ityof60kgofr
awmat
eri
aland96l
abourhour
s.Oneuni
tofpr
oduct
AandBsol
dyi
eldsBi
rr40andBi
rr35,r
espect
ivel
y,aspr
ofi
t.Det
ermi
neast
ohow many
uni
tsofeachoft
hepr
oduct
sshoul
dbepr
oducedperweeksot
hatt
hef
ir
m canear
nthe
maxi
mum pr
ofi
t.

Answer
For
mul
ateaLPP
Theobj
ect
ivef
unct
ion:Thef
ir
stmaj
orr
equi
rementofanLPPi
sthatweshoul
dbeabl
eto
i
dent
if
ythegoali
nter
msoft
heobj
ect
ivef
unct
ion.Thi
sfunct
ionr
elat
esmat
hemat
ical
l
ythe
v
ari
abl
eswi
thwhi
chwear
edeal
i
ngi
nthepr
obl
em.
 Forourpr
obl
em,
thegoali
sthemaxi
mizat
ionofpr
ofi
t,whi
chwoul
dbeobt
ainedby
pr
oduci
ng(
andsel
l
ing)pr sAandB.I
oduct fwel
etX1andX2r
epr
esentt
henumberofuni
ts
pr
oducedperweekoft
hepr sAandBr
oduct espect
ivel
y,t
het
otalpr
ofi
t,Z=40X1+35X2
becauset
heuni
tpr
ofi
tfr
om t
het
wopr
oduct
sis,
respect
ivel
y,Bi
rr40andBi
rr35.
 Not
icet
hatt
hef
unct
ioni
sal
i
nearone.

Theconst
rai
nts:Ashasbeensai
d,anot
herr
equi
rementofl
i
nearpr
ogr
ammi
ngi
sthatt
he
r
esour
cesmustbei
nli
mit
edsuppl
y.Themat
hemat
ical
rel
ati
onshi
pwhi
chi
susedt
oexpl
ain
l
i
mit
ati
onofr
esour
cei
sinequal
i
ty.Thel
i
mit
ati
oni
tsel
fisknownasaconst
rai
nt.

 Eachuni oductAr
tofpr equi
res2Kgofr
awmat
eri
alwhi
l
eeachuni oductB
tofpr
 r
equi
res3kg.Thet
otalconsumpt
ionwoul
dbe2X1 +3X2,whi
chcannotexceedt
het
otal
av
ail
abi
l
ityof60kgev
eryweek.Wecanexpr
esst
hisconst
rai
ntas2X1+3X2≤60.
 Si
mil
arl
y,i
tisgi
vent
hatauni
tofAr
equi
res4l
abourhour
sfori
tspr
oduct
ionand
oneuni
tofBr
equi
re3l
abourhour
s.Wi
thanav
ail
abi
l
ityof96hour
saweek,
wehav
e4X1+
3X2≤96ast
hel
abourhour
’sconst
rai
nt.

I
tisi
mpor
tantt
onot
ethatf
oreachoft
heconst
rai
nts,i
nequal
i
tyr
athert
hanequat
ionhas
beenused.Thi
sisbecauset
hepr
ofi
tmaxi
mizi
ngout
putmi
ghtnotuseal
lther
esour
cest
o
t
hef
ull
-l
eav
ingsomeunused.Hence,t
he≤si
gn.Howev
er,i
tmaybenot
icedt
hatal
lthe
const
rai
nts,
li
keobj
ect
ivef
unct
ion,
areal
sol
i
neari
nnat
ure.

Non-
negat
ivi
tycondi
ti
on:Qui
teobv
iousl
y,X1 andX2,bei
ngt
henumberofuni
tspr
oduced
cannothav
enegat
ivev
alues.Thus,bot
hoft
hem canassumev
aluesonl
ygr
eat
er-
thanor
equal
-t
ozer
o.Thi
sist
henon-
negat
ivi
tycondi
ti
on,
expr
essedsy
mbol
i
cal
l
yasX1≥0andX2≥
0.

Now,
wecanwr
it
ethepr
obl
em i
ncompl
etef
orm asf
oll
ows.

Maxi
mize Z=40x1+35x2 Pr
ofi
t
Subj
ectt
o

2x1+3x
2≤60 Rawmat
eri
al
4x1+3x2≤96 const
rai
nt
x1,
x2≥0 Laborhour
s
const
rai
nt
Non-
negat
ivi
ty
r
est
ri
cti
on
Gr
aphi
ngt
heconst
rai
nts:
Weshal
lfi
rstchar
tthegi
venr
est
ri
cti
onsont
hegr
aph.
 The const
rai
ntofr
aw mat
eri
alav
ail
abi
l
it,2X1 +3X2 ≤ 60,st
y ates t
hatany
 combi
nat
ionof
2X1 +3X2 t
hatdoesnotexceed60i
saccept
abl
e.Wi
thv
ary
ingv
aluesofX1andX2,t
he
combi
nat
ioncanassumeamaxi
mum v
alueequalt
o60.Thi
sconst
rai
ntcanbedepi
cted
gr
aphi
cal
l
ybypl
ott
ingt
hest
rai ne2X1+3X2=60,si
ghtl
i nceonl
ytwopoi
ntsar
eneededt
o
obt
ainast
rai
ghtl
i
newecangett
het
wopoi
ntsbyset
ti
ngoneoft
hesev
ari
abl
esi
ntur
n
equalt
o0,
andcal
cul
ati
ngt
hev
alueoft
heot
her
.Thus,
whenX1=0,
X2woul
dequal20,
and
whenX2i
ssetequalt
o0,
X1woul
dbe30.Joi
ningt
hepoi
nts(
0,20)and(
30,
0)wegett
he
r
equi
redst
rai
ghtl
i
newhi
chi
sshownasl
i
nePT.

 Si
mil
arl
y,t
heot
herconst
rai
nt4x1+3x2=96canbepl
ott
ed.

I
dent
if
ythesol
uti
onspace(
ident
if
ythef
easi
bler
egi
on)
;
 Fr
om t
heabov
egr
aph,
thesol
uti
onspace(
thef
easi
bler
egi
on)i
sOPQR

Locat
ethesol
uti
onpoi
nts(
ident
if
ythecor
nerpoi
ntsf
rom t
hesol
uti
onspace)
;
 I
nli
nearpr
ogr
ammi
ng,si
nce,t
heopt
imalsol
uti
onal
way
soccur
satt
hecor
ner
poi
ntoft
hesetoff
easi
blesol
uti
ons,
thecor
nerpoi
ntsf
rom t
hesol
uti
onspacear
efound
 atO,
P,Q,
andR.

Bi
ndi
ngandNon-
Bindi
ngconst
rai
nts(
resour
ces)
Oncet
heopt
imalsol
uti
ont
oanLPPi
sobt
ained;wemaycl
assi
fyt
heconst
rai
ntsasbei
ng
bi
ndi
ngornon-
bindi
ng.Aconst
rai
nti
ster
medasbi
ndi
ngi
fthel
efthandsi
deandr
ighthand
si
deofi
tar
eequalwhenopt
imalv
aluesoft
hedeci
sionv
ari
abl
esar
esubst
it
utedi
ntot
he
const
rai
nt.

I
nExampl
e1abov
e,t
heopt
imalv
aluesofdeci
sionv
ari
abl
esar
e:X1=18andX2 =8.
Subst
it
uti
ngt
hesev
aluesi
nthet
woconst
rai
ntsweget
:

2x1+3x
2 or2x18+3x8=60RHSand
4x1+3x
2 or4x18+3x8=96-
RHS

Thus,
bot
htheconst
rai
ntsar
ebi
ndi
ngi
nnat
ure.
Ont
heot
herhand,i
fthesubst
it
uti
onoft
hedeci
sionv
ari
abl
esdoesnotl
eadt
oequal
i
ty
bet
weent
hel
eftandt
her
ighthandsi
desoft
heconst
rai
nt,
iti
ssai
dtobenon-
bindi
ng.

RedundantConst
rai
nt(
s):pl
ott
ingeachoft
heconst
rai
ntsont
hegr
aphser
vest
odet
ermi
ne
t
hef
easi
bler
egi
onofagi
venpr
obl
em.Howev
er,
ifaconst
rai
nt,
whenpl
ott
ed,
doesnotf
orm
par
toft
heboundar
ymar
kingt
hef
easi
bler
egi
onoft
hepr
obl
em,i
tissai
dtober
edundant
.
Ther
efor
e,t
hei
ncl
usi
onorexcl
usi
onofar
edundantconst
rai
ntobv
iousl
ydoesnotaf
fectt
he
opt
imal
sol
uti
onoft
hepr
obl
em.
 Z = 40x1 + 35x2 Pr
ofi
t
Maxi
mize Subj
ectt
o Exampl
e 2:
2x1+3x
2≤60 Raw mat
eri
al Cont
inui
ng wi
th
4x1+3x
2≤96 const
rai
nt exampl
e 1,suppose
4x1+3.
5x2≤105 Labourhour
s t
hat each of t
he
x1,
x2≥0 const
rai
nt pr
oduct
sar
erequi
red
Packagi
ng st
hour o be packed.Ev
ery
const
rai
nt uni oduct A
t of pr
r
equi
res4hour
swhi
l
e
ev
eryuni
tofpr
oductBr
equi
res3.
5hour
sforpackagi
ng.Supposet
hati
nthepackagi
ng
depar
tment
,105hour
sar
eav
ail
abl
eev
eryweek.Undert
hesecondi
ti
ons,whatpr
oductmi
x
woul
dmaxi
mizet
hepr
ofi
t?

Thepr
obl
em canber
est
atedasf
oll
ows:

 Theopt
imal
sol
uti
oni
stopr
oduce18uni
tsofpr
oductAand8uni
tsofpr
oductBev
ery
week,t
ogett
hemaxi
mum pr
ofi
tofBi
rr1000.Si
nce,t
het
hir
dconst
rai
ntdonothav
e
anyi
mpactont
heopt
imal
sol
uti
on,
iti
sar
edundantconst
rai
nt.
Exer
cise.Sol
vet
hef
oll
owi
ngLPPusi
nggr
aphi
cmet
hod

MaxZ=2X1+3X2
Subj
ectt
o:
4X1+3X2<18
5X1+2X2<19
X1,
X2>0
2.
5.1.
2.Formi
nimi
zat
ionpr
obl
em
Thegr
aphi
calsol
uti
ont
othel
i
nearpr
ogr
ammi
ngpr
obl
emscanbeappl
i
edt
othepr
obl
ems
whi
chhav
emi
nimi
zat
ionnat
ure.
Exampl
e.Theagr
icul
tur
alr
esear
chi
nst
it
utesuggest
edt
oaf
armert
ospr
eadoutatl
east
4800kgofaspeci
alphosphat
efer
ti
li
zerandnotl
esst
han7200kgofaspeci
alni
tr
ogen
f
ert
il
izert
orai
sepr
oduct
ivi
tyofcr
opsi
nhi
s/herf
iel
ds.Ther
ear
etwosour
cesf
orobt
aini
ng
t
hese-
mixt
uresAandB.Bot
hoft
hesear
eav
ail
abl
einbagswei
ghi
ng100kgeachandt
hey
costBi
rr40andBi
rr24r
espect
ivel
y.Mi
xtur
eAcont
ainsphosphat
eandni
tr
ogenequi
val
ent
of20kgand80kgr
espect
ivel
y,whi
l
emi
xtur
eBcont
ainst
hesei
ngr
edi
ent
sequi
val
entof50
kgeach.Det
ermi
nehowmanybagsofeacht
ypet
hef
armershoul
dbuyi
nor
dert
oobt
ain
t
her
equi
redf
ert
il
izeratmi
nimum cost
.

 Theopt
imalsol
uti
ont
othepr
obl
em i
stobuy144bagsofmi
xtur
eBonl
yandnone
ofmi
xtur
eA,
whi
chmi
nimi
zet
het
otal
costt
oBi
rr3456.

 Theopt
imal
val
ues,
maybesubst
it
utedi
nthet
woconst
rai
ntst
oget
:

20x1+50x2 or 20*0+50*144=7200≠4800(
RHS)and
80x1+50x2 or 80*0+50*144=7200=RHS.

 Accor
dingl
y,t
hef
ir
stconst
rai
nti
snon-
bindi
ngwhi
l
ethesecondonei
sabi
ndi
ng.

Exer
cise.Sol
vet
hef
oll
owi
ngl
i
nearpr
ogr
ammi
ngpr
obl
em usi
ngt
hegr
aphi
cmet
hod
Mi
ni zeZ=X1+1.
mi 5X2
Subj
ectt
o
X1+X2≥8
X1+X2≥10
X2≥2
X1,
X2≥0

2.
5.1.
3.Speci
alcasesofl
i
nearpr
ogr
ammi
ng
I
neachoft
heabov
edi
scussedexampl
es,wehav
eobt
ainedauni
queopt
imalsol
uti
on.We
now consi
dert
hreet
ypesofl
i
nearpr
ogr
ammi
ngpr
obl
emswhi
chdo nothav
euni
que
opt
imalsol
uti
ons.Some oft
hese cases ar
eresul
ted f
rom v
iol
ati
ons oft
he basi
c
assumpt
ionofl
i
nearpr
ogr
ammi
ng.Theyi
ncl
udemul
ti
pleopt
imalsol
uti
ons,nof
easi
ble
sol
uti
onsandunboundedsol
uti
ons.

1.Mul
ti
pleOpt
imalSol
uti
ons:-Asment
ionedi
ntheabov
e,i
neachoft
heexampl
est
hatwe
hav
econsi
der
ed,
wehav
eobser
vedt
hatt
heopt
ima
2. sol
uti
oni
sgi
venbyanext
reme/
cor
nerpoi
ntsoft
hef
easi
bler
egi
onandt
hesol
uti
oni
s
uni
que.I
tisal
sopossi
blet
hatagi
venpr
obl
em mayhav
emor
ethanoneopt
imal
sol
uti
on.
Eachoft
heopt
imawoul
dnat
ural
l
yyi
eldt
hesameobj
ect
ivef
unct
ionv
alue.
 Thet
wocondi
ti
onst
hatshoul
dbesat
isf
iedi
nor
dert
osaymul
ti
pleopt
imalsol
uti
ons
exi
star
e:
(
a) Theobj
ect
ivef
unct
ionshoul
dbepar
all
elt
osomeedgeofconst
rai
nt,
whi
chi
sal
so
boundar
ytot
hef
easi
bler
egi
on;
and
(
b) Theconst
rai
ntmustbeabi
ndi
ngconst
rai
nt.

Exampl
e:Sol
vet
hef
oll
owi
ngLPPgr
aphi
cal
l
y:
Maxi
mizeZ=8X1+16X2
Subj
ectt
o
X1+X2=200
X2=125
3X1+6X2=900
X1,
X2≥0
Theconst
rai
ntsar
eshownpl
ott
edont
hegr
aphi
ntheFi
gur
eabov
e.Al
so,i
so-
prof
itl
i
nes
(
theobj
ect
ivef
unct
ion)hav
ebeengr
aphed.Weobser
vet
hatt
hei
so-
prof
itl
i
nei
spar
all
elt
o
t
heequat
ionf
ort
hir
dconst
rai
nt3X1+6X2=900.Aswemov
ethei
so-
prof
itl
i
nef
art
herf
rom
t
heor
igi
n,i
tcoi
nci
deswi
tht
hepor
tonBCoft
i heconst
rai
ntl
i
net
hatf
ormst
heboundar
yof
t
hef
easi
bler
egi
on.

 I
timpl
i
est
hatt
her
ear
eani
nfi
nit
enumberofopt
imalsol
uti
onsr
epr
esent
edbyal
l
poi
ntsl
yi
ngont
hel
i
nesegmentBC,i
ncl
udi
ngB andC.Howev
er,si
ncet
heopt
imal
sol
uti
onal
way
sli
eatanext
remepoi
ntoft
hef
easi
bler
egi
on,wemaydi
sregar
dal
lot
her
sol
uti
onsandconsi
deronl
ytheext
remepoi
ntsr
epr edbyB(
esent 50,125)andC(
100,
100)t
oest
abl
i
sht
hesol
uti
ont
othi
spar
ti
cul
arLPP.

 Theext
remepoi
ntsoft
hef
easi
bler
egi
onar
egi
venandev
aluat
edher
e.

Poi
nt X1 X2 Z=8X1+16X2
O 0 0 0
A 0 125 2000
B 50 125 2400
C 100 100 2400Maxi
mum
D 200 0 1600

TheBandCcl
ear
lyr
epr
esentt
heopt
ima.I
nthi
sexampl
e,asment
ionedi
ncondi
ti
on(
a),i
f
suchaconst
rai
nt(
towhi
cht
heobj
ect
ivef
unct
ioni
spar
all
el)doesnotf
orm anedgeor
boundar
yoft
hef
easi
bler
egi
on,
thenmul
ti
pleopt
imal
sol
uti
onswoul
dnotexi
st.

Exer
cise:
Sol
vet
hef
oll
owi
ngl
i
nearpr
ogr
ammi
ngpr
obl
em.
Maxi
mize Z=110X1+110X2
Subj
ectt
o
X1+X29
X12
20X1+50X2360
X23
 X1,
X20

2.I
nfeasi
bleSol
uti
on:
-I
thasal
readybeenst
atedt
hatasol
uti
oni
scal
l
edf
easi
blei
fit
sat
isf
iesal
ltheconst
rai
ntsandt
henon-
negat
ivi
tycondi
ti
ons.Somet
imesi
tispossi
blet
hat
t
heconst
rai
ntsmaybei
nconsi
stentsot
hatt
her
eisnof
easi
blesol
uti
ont
othepr
obl
em.
Suchsi
tuat
ionsar
ecal
l
edi
nfeasi
bil
i
ty.

Exampl
e:Sol
vet
hef
oll
owi
ngLLPusi
nggr
aphi
cmet
hod.

Aswecanseei
nthegr
aph,
thef
easi
bler
egi
oncor
respondi
ngt
otheconst
rai
ntsi
sindi
cat
ed
i
ntheshadedr
egi
ons.Wecaneasi
l
yobser
vet
hatt
her
eisnocommonpoi
nti
nthear
eas
shaded.Ther
efor
e,al
ltheconst
rai
ntscannotbesat
isf
iedand,assuch,t
her
eisnof
easi
ble
sol
uti
ont
othegi
venpr
obl
em.

3.Unbounded Sol
uti
on:-Fora maxi
mizat
ion t
ype ofl
i
nearpr
ogr
ammi
ng pr
obl
em,
unboundednessoccur
swhenoneormor
eoft
hedeci
sionv
ari
abl
escanbei
ncr
eased
i
ndef
ini
tel
ywi
thoutv
iol
ati
nganyoft
her
est
ri
cti
ons(
const
rai
nts)
.Thus,anunboundedLPP
occur
sifi
tispossi
blet
ofi
ndpoi
ntsi
nthef
easi
bler
egi
onwi
thar
bit
rar
il
ylar
ge(
ifa
maxi
mizat
ionpr
obl
em)orar
bit
rar
il
ysmal
l(i
fami
nimi
zat
ionpr
obl
em)obj
ect
ivev
alues.
 Suchsi
tuat
ionmost
lyoccur
swhent
heassumpt
ionoff
ini
tenessi
svi
olat
ed(
oneor
mor
eoft
hev
ari
abl
escanbei
ncr
easedi
ndef
ini
tel
ywi
thoutv
iol
ati
ngt
hef
easi
bil
i
ty)or
whent
hepr
obl
em hasnotbeencor
rect
lyf
ormul
ated.

Exampl
e:Consi
dert
hef
oll
owi
ngexampl
e:

Exer
cise1:
-Sol
vet
hef
oll
owi
ngl
i
nearpr
ogr
ammi
ngpr
obl
em usi
nggr
aphi
cmet
hod.
Maxi
mize Z=30X1+50X2
Subj
ectt
o
3X1+X29
X1+2X212
X1+2X2 9
X1,
X20
Exer
cise2:Sol
vet
hef
oll
owi
ngLPPsusi
ngGr
aphi
cmet
hodandobt
aini
fther
eisapr
obl
em
i
nobt
aini
ngt
heopt
imal
sol
uti
on.
5. Acompanypr
oducest
wopr
oduct
sP1 andP2.Thepr
oduct
sar
epr
oducedandsol
dona
weekl
ybasi
s.Becauseofl
i
mit
edav
ail
abl
eresour
ces,t
heweekl
ypr
oduct
ioncan’
texceed25
uni
tsf
orpr
oductP1and35uni
tsf
orpr
oductP2.Thecompanyempl
oysat
otalof60wor
ker
s.
Pr
oductP1 r
equi
res2man-
weeksofl
abor
,whi
l
eP2 r
equi
resoneman-
weekofl
abor
.Pr
ofi
t
mar
ginonP1i
s60bi
rrandonP2i
s40.
a) For
mul
atet
heabov
epr
obl
em asl
i
nearpr
ogr
ammi
ngpr
obl
em
b) Fi
ndt
hepr
oductmi
xsoast
omaxi
mizet
hepr
ofi
toft
hecompany

6. Af
urni
tur
emanuf
act
urerpr
oducest
wot
ypesofdesks:st
andar
dandexecut
ive.These
desksar
esol
dtoanof
fi
cef
urni
tur
ewhol
esal
er,andf
oral
lpr
act
icalpur
poses,t
her
eisan
unl
i
mit
edmar
ketf
oranymi
xoft
hesedesks,atl
eastwi
thi
nthemanuf
act
urer
’spr
oduct
ion
capaci
ty.Eachdeskhast
ogot
hroughf
ourbasi
coper
ati
ons:cut
ti
ngoft
hel
umber
,joi
ningof
t
hepi
eces,pr
e-f
ini
shi
ng,andf
inalf
ini
sh.Eachuni
toft
hest
andar
ddeskpr
oducedt
akes48
mi
nut
esofcut
ti
ngt
ime,2hour
sofj
oini
ng,40mi
nut
esofpr
e-f
ini
shi
ng,and5hour
sand20
mi
nut
esoff
inalf
ini
shi
ngt
ime.Eachuni
toft
heExecut
ivedeskr
equi
red72mi
nut
esofcut
ti
ng,
3hour
sofj
oini
ng,2hour
sofpr
e-f
ini
shi
ngand4hour
soff
inalf
ini
shi
ngt
ime.Thedai
l
y
capaci
tyf
oreachoper
ati
onamount
sto16hour
sofcut
ti
ng,30hour
sofj
oini
ng,16hour
sof
pr
e-f
ini
shi
ngand64hour
soff
inalf
ini
shi
ngt
ime.Thepr
ofi
tperuni
tpr
oducedi
sBi
rr40f
ort
he
St
andar
ddeskandBi
rr50f
ort
heExecut
ivedesk.Det
ermi
net
hepr
oductmi
xthatmaxi
mizes
t
otal
rev
enue.