TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 13, No. 2, Februa
ry 20
15, pp. 215 ~ 222
DOI: 10.115
9
1
/telkomni
ka.
v
13i2.704
3
215
Re
cei
v
ed
No
vem
ber 1
8
, 2014; Re
vi
sed
De
cem
ber 2
6
,
2014; Accep
t
ed Jan
uary 1
0
, 2015
Optimal Warranty Policy Considering Two-dimensional
Imperfect Preventive Maintenance fo
r Repairable
Product
Xin-
y
u
e Lee*
1
, Yun-xian Jia
2
Dep
a
rtment of Mana
geme
n
t Engi
neer
in
g
Me
ch
an
i
c
al
Eng
i
ne
e
r
in
g
Co
l
l
eg
e
Hepi
ng str., Shijiaz
h
u
ang, C
h
i
n
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: oecl
x
y@
hot
mail.com
1
, y
unx
ia
n_
j
i
a
@
ho
tma
i
l
.
co
m
2
A
b
st
r
a
ct
For some kinds of products, such as car, ai
rcraft, governm
ent acquisition, the consumers hav
e
strict requir
e
ments to th
e re
li
abil
i
ty of th
ese
prod
ucts
. T
hen
the
manufact
u
rer is
incl
in
ed t
o
pr
ovid
e th
e t
w
o-
di
me
nsio
nal pr
eventiv
e mai
n
tena
nce
p
o
licy to
take
the us
age
degr
ee
of the prod
uct in
to accou
n
t. As
a
result, tw
o-di
me
nsi
ona
l pre
v
entive
mainte
nanc
e po
li
cy i
n
the w
a
rrant
y perio
d h
a
s
recently
obtai
n
e
d
incre
a
sin
g
atte
ntion fro
m
man
u
facturers a
nd
consu
m
er
s. In this pa
per, w
e
focused
on th
e
opti
m
i
z
at
ion
o
f
base
d
w
a
rra
nty cost a
n
d
p
r
opos
ed
a
ne
w
expecte
d
c
o
st
mod
e
l
of
the tw
o-d
i
me
nsio
nal
i
m
perf
e
ct
preve
n
tive
mai
n
tena
nce p
o
lic
y from the pe
rspective
of th
e ma
nufactur
e
. T
he optimal tw
o-dime
nsi
o
n
a
l
preve
n
tive
ma
i
n
tena
nce w
a
s
obtai
ne
d by mi
ni
mi
z
i
n
g
b
a
se
d
w
a
rranty cost.
And asy
m
metri
c
copul
a functi
on
w
a
s app
li
ed t
o
mod
e
li
ng
the
fail
ure f
unctio
n
of t
he
pro
d
u
c
t. At last, nu
mer
i
cal
ex
a
m
p
l
es
are
giv
e
n
to
illustrat
e
the pr
opos
ed
mo
dels
,
of
w
h
ich the re
sults prov
e the mo
de
l effective and va
li
date
.
Ke
y
w
ords
:
t
w
o-dimens
ion
a
l
w
a
rranty, tw
o-dimens
ion
a
l
i
m
p
e
rfect p
r
eventiv
e
mai
n
tena
nce, b
a
s
ed
w
a
rranty, extende
d w
a
rranty,
asymmetric co
pul
a functio
n
Copy
right
©
2015 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
In general, warranty has two sig
n
ifica
n
t roles
a
s
“p
ro
tector” and “p
romote
r” to consume
r
and m
anufa
c
turer when
the p
r
od
uct
s
fail to pe
rf
orm their p
r
e-specifie
d fun
c
t
i
ons du
ring
the
warra
n
ty peri
od [1-2]. Accordin
g to M
u
rthy and B
lischke [3], the
r
e
are
only two
kind
s of
wa
rranty
in the
pro
d
u
c
t life cycl
e in
cludi
ng b
a
se
d warranty a
nd exten
ded
warra
n
ty. Based
warranty
is
sold with p
r
o
duct. Its co
st is usually in
clud
ed
in the
sale pri
c
e. Ho
wever, to have extend
ed
warra
n
ty, the cu
stome
r
sho
u
ld pay extra
money to
get
the se
rvice
a
fter ba
sed
wa
rra
nty expire
s.
No
wad
a
ys, ei
ther for ba
se
d warra
n
ty or extended wa
rra
nty, one key issu
e of interest that ari
s
e
s
from wa
rrant
y analysis i
s
the modelin
g of wa
rrant
y cost. By optimizing
wa
rranty co
st, th
e
manufa
c
turer co
uld p
r
i
c
e
the
warrant
y prop
er
an
d ma
ke th
ei
r p
r
od
uct m
o
re
attra
c
tive to
c
o
ns
umers
.
For s
o
me k
i
nds
of
produc
ts
(s
uc
h as
airc
raft,
government ac
quis
ition), the
con
s
um
ers h
a
ve strict req
u
irem
ents to
the reliab
ility of these prod
ucts. So the
manufa
c
turer is
inclin
ed to provide the two
-
dime
nsi
onal
prevent
ive m
a
intena
nce policy to take the usage de
g
r
ee
of the produ
ct into accou
n
t, which ha
s two lim
its of coverage in t
e
rm
s of age and u
s
age. F
o
r
example, a sold ca
r
may
be cove
red
a
3
ye
ar and
3
0
,000 mile warranty,
and
t
he ca
r sho
u
ld
be
che
c
k u
p
i
n
4
s
(Sal
e, Spa
r
e
pa
rt, Service
and
Su
rvey)
sto
r
e at
p
e
r year and
15,
000 mile.
F
r
o
m
the perspe
c
ti
ve of the consum
ers, two-dim
e
n
s
iona
l preventiv
e maintena
nce
policy might be
better than
o
ne-di
men
s
ion
a
l wa
rranty for its o
ne m
o
re bo
und
ary t
hat co
uld ma
ke the
pro
d
u
c
ts
be mai
n
taine
d
in time.
Ho
wever, it
coul
d also pla
c
e
heavier bu
rd
en on
the m
anufa
c
ture
rs. So
the man
u
fact
ure
r
s sh
ould
price the
wa
rranty pro
perly
to make thei
r
prod
uct
s
mo
re attra
c
tive a
nd
guarantee th
eir p
r
ofit. As
a re
sult, ho
w to minimizi
n
g
the warrant
y cost in
the
two-di
men
s
io
nal
warra
n
ty be
comes a
pra
c
tical requi
re
ment. To
ou
r kno
w
led
ge,
few
analytical or num
eri
c
al
method
s h
a
ve be
en
rep
o
rt
ed in
the lite
r
ature,
wh
i
c
h offers
a de
cision
fra
m
e
w
ork
to optimi
z
e the
two-di
men
s
io
nal p
r
eve
n
tive mai
n
tena
nce pe
rio
d
in
th
e warranty
p
e
riod.
In thi
s
perspe
c
tive,
we
develop in thi
s
pap
er a
ne
w mathem
atical mo
del
to
find the opti
m
ized
preve
n
t
ive maintena
nce
policy for ba
sed wa
rranty.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 2, Februa
ry 2015 : 215 – 222
216
The re
st of this pa
per i
s
orga
nized a
s
follows: Section 2 introdu
ce
s the relati
ve work
existing in t
he literature
with p
r
od
u
c
t wa
rr
anty
polici
e
s. Se
ction 3 o
u
tlines the
mod
e
l
assumptio
n
s
and n
o
tation.
Section 4 i
s
d
edicated
to d
e
velopme
n
t o
f
the mathem
atical mo
del.
A
nume
r
ical example illu
stra
tes our a
ppro
a
ch to pr
ove the model validate and effective in Sect
ion
5.
2. Litera
ture
R
e
v
i
e
w
Different wa
rranty
polici
e
s has different warr
a
n
ty co
st. The charact
e
risti
c
s of wa
rra
nty
polici
e
s a
r
e
essential fa
ctors i
n
determining wa
rra
n
ty cost in
cluding
warra
n
t
y service o
b
j
ect,
warra
n
ty rene
wal me
cha
n
ism and wa
rra
nty dimensio
n etc. Blisch
ke and Murthy
[4]
formulate
d
a
cla
ssi
c taxon
o
my for wa
rranty policie
s.
But w
hen th
e prod
uct
s
are becoming
more
compli
cate,
the warranty polici
e
s have
some chang
es. So based
on the syste
m
of the warranty policie
s [5,
18, 24], we furthe
r devel
o
p
the taxon
o
m
y for warr
a
n
ty policie
s,
sho
w
n
as Fi
gure
1. At first,
warra
n
ty poli
c
ie
s
could
be
divided i
n
to
two type
s in
cluding
sin
g
le
-comp
one
nt warranty [9, 1
9
]
and multi-co
mpone
nt warranty. Multi-compon
ent wa
rranty often consi
d
e
r
s a
fleet of produ
cts.
Then th
e wa
rra
nty coul
d
be divide
d i
n
to two
sub
-
grou
ps as rene
wing
wa
rranty an
d n
o
n
-
renewing warranty. Rene
wing warranty policy means that the
warranty period
will restart at the
failure time if the failure occurs in the
warr
anty peri
o
d [1, 21]. Non
-re
ne
wing
wa
rra
nty policy
will
end at the wa
rra
nty period
W
no matter
how ma
ny failure
s hap
pen i
n
the warra
n
ty period [2, 27].
Simple warra
n
ty policy h
a
s
only
one
warranty poli
c
y will be
ap
pli
ed in th
e warranty pe
riod
[22],
and different warra
n
ty policie
s will b
e
a
pplied in th
e perio
d of co
mbined
wa
rranty policy [1
6].
Acco
rdi
ng to
the wa
rranty
boun
dari
e
s, t
he warr
anty
polici
e
s
co
uld
be furth
e
r
di
vided into o
n
e
-
dimen
s
ion
a
l
warra
n
ty, two-dim
e
n
s
iona
l wa
rranty a
nd n
-
dim
e
n
s
i
onal
wa
rrant
y (
3
n
). One
-
dimen
s
ion
a
l
warra
n
ty is li
mited by time
[23], and
two-dim
e
n
s
iona
l wa
rranty is
usu
a
lly limite
d
by
age and u
s
a
ge (such as
miles an
d re
volutions) [8, 10, 26]. For
some p
r
od
ucts, which
req
u
ire
high reli
abilit
y, multi usage limits will be applie
d to rest
rain th
e warra
n
ty period. Th
en the
warra
n
ty cou
l
d be n
-
dim
ensi
onal
wa
rranty. Th
e
r
e
are fe
w lite
r
atures
abo
u
t
n-dime
nsi
o
nal
warra
n
ty. However, alo
n
g
with the industrial p
r
o
d
u
c
t become m
o
re co
mplex, the intere
sting
of
the n-dime
nsi
onal
wa
rranty wo
uld i
n
crea
se. O
b
viou
sly wa
rranty poli
c
y is the
re
su
lt of the g
a
m
e
betwe
en m
a
n
u
facturer an
d
co
nsume
r
. So the
co
st all
o
catio
n
me
ch
anism
will
greatly influen
ce
the warra
n
ty deci
s
io
n and
effect. And the warra
n
ty
policy coul
d be furthe
r divided
into free-rep
a
ir
warra
n
ty, pro-rate
warra
n
ty and
reb
a
te warr
anty
etc
by cost allocation
mech
ani
sm.
The
manufa
c
turer will take all
expense in
the free-re
p
a
ir wa
rranty perio
d. On the co
ntra
ry,
the
warra
n
ty cost
will be sh
are
by manufact
u
re
rs a
nd
co
nsum
ers in th
e perio
d of pro-rata wa
rran
ty
or re
bate warranty [7].
Figure 1. The
classification
of warranty policie
s
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Optim
a
l Warranty Policy Consi
deri
ng T
w
o-dim
ens
io
nal Im
perfect Preventive…
(Xin-yue L
e
e
)
217
Obliviou
s
ly, two-dime
nsio
nal
warra
n
ty is o
ne
of the
main
conte
n
ts of th
e
wa
rra
nty policy
For two
-
dim
e
nsio
nal warra
n
ty, there are
thr
ee meth
o
d
s that ha
s b
een devel
ope
d for analy
z
in
g
two-di
men
s
io
nal wa
rra
nty [29], ie univariate method
[14, 19, 26],
bivariate met
hod [8] and time
scale m
e
tho
d
[12]. Bivairate metho
d
i
s
simple
r an
d
mo
re strai
ghtforward. And
on
e wa
y
to
formulate
the
bivariate
di
stribution
is ma
king
fu
ll u
s
e
of Co
pula
fun
c
tion.
Cop
u
la
s a
r
e
a tool
for
con
s
tru
c
ting
multivariate
distrib
u
tion
s and de
scri
bing the d
e
pend
en
ce b
e
twee
n ra
nd
om
variable
s
Furthermo
re
sy
mmetric cop
u
la fun
c
tion
were al
rea
d
y appli
ed to
model th
e p
r
odu
ct
reliability. In this pape
r,
we propo
se the co
st model of th
e two-dimen
s
ion
a
l preve
n
tive
maintenance in the
warranty period by
utilizing
asy
mmetric copula function.
And the optimal
two-di
men
s
io
nal preventive mainten
a
n
c
e p
e
rio
d
is
obtaine
d by minimizi
ng th
e wa
rranty cost.
The advanta
g
e
s of the mod
e
l prop
osed a
r
e as follo
ws.
a)
It is able
to
handl
e the
tail dep
end
ence of a
g
e
and
usage
to cal
c
ul
ate
two-
dimen
s
ion
a
l
warra
n
ty co
st
whi
c
h
co
uld
be mo
re
accurate. T
he a
symmetric
co
p
u
la
function i
s
ap
plied rath
er th
an assumi
ng
the age an
d u
s
ag
e have lin
ear relation
sh
ip.
b)
The two-dim
ensi
onal
pre
v
entive ma
i
n
tenan
ce pe
riod co
uld be
obtai
ned
by
minimizi
ng b
a
se
d warrant
y cost. It is
more
re
alisti
c that the PM
action i
s
also
two-
dimen
s
ion
a
l to make the p
r
odu
ct more reliable.
3.
Model As
su
mptions and
Nota
tion
This
se
ction provide
s
mo
d
e
l assumptio
n
s an
d notati
on.
(
T
0
,
U
0
)
The two
-
dime
nsio
nal preve
n
tive mainten
ance (PM) p
e
riod impl
eme
n
ted
(W
B
, U
B
)
The ba
sed
warranty time and usage limi
t
0
0
0
B
B
U
U
r
WT
(0
1
)
PM level. When
1
, PM is perfect maintena
nce. When
0
, PM is
minimal re
pai
r.
C
p
The co
st of PM
C
m
The co
st of CM
Assu
ming th
e
usa
ge of the
prod
uct i
s
rel
a
tive
ly steady
for individu
al
cu
stome
r
. It doe
sn’t
mean the u
s
age for indivi
dual cu
stom
e
r
doe
sn’t
ch
a
nge but ch
an
ge little. The based wa
rra
n
ty
coverage
a
r
e
a
could
be
di
vided into
two pa
rt: D
1
an
d D
2
. For
D
1
,
warranty is likely terminated at
usa
ge limit. On the cont
rary
, warranty is likely termin
ated at time limit for D
2
.
Figure 2. Two
dimensi
onal
warra
n
ty coverag
e
Preventive m
a
intena
nce is im
pe
rfect
re
pair
and
the
co
rrective
maintena
nce
(CM) i
s
minimal re
pa
ir. Compa
r
e
d
to the mean time be
tween failu
re
s, the maintenan
ce time
is
negligibl
e
. A failure
ca
n be
detecte
d im
mediately whi
c
h
re
sults i
n
an imme
diate
claim, the
n
the
manufa
c
turer will respon
d
all the clai
ms. Assu
min
g
a failure could be dete
c
ted imme
di
ately
whi
c
h re
sult
s in an immediate claim a
nd the manuf
acturer will resp
ond all th
e claim
s
. In this
study, the
warra
n
ty co
st is
comp
o
s
ed
of
prev
entive maint
enan
ce
co
st
and
corre
c
tive
maintena
nce co
st. The cu
mulative failure dist
ribut
io
n
function, failure de
nsity function a
nd fai
l
ure
rate of the produ
ct are [20]
:
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 2, Februa
ry 2015 : 215 – 222
218
12
01
2
1
1
2
1
2
2
2
1
1
2
3
(,
)
(
(
)
,
(
)
,
)
(
(
),
(
)
,
)
(
)
(
1
(
)
,
(
),
)
(
)
(
(
)
,
1
(
)
,
)
Ft
u
C
F
t
F
u
p
C
Ft
F
u
p
F
u
C
F
t
F
u
p
F
t
C
Ft
F
u
(1)
(,
)
(,
)
F
tu
ft
u
tu
,
(,
)
(,
)
1
(,
)
Ft
u
ht
u
F
t
u
tu
(2)
12
((
)
,
(
)
,
)
CF
t
F
u
is the a
s
ym
metric
co
pul
a functio
n
a
nd
2
0
1
i
i
p
.
12
1
((
)
,
(
)
,)
CF
t
F
u
is
symmetri
c
co
pula fu
nctio
n
, whi
c
h
sho
u
ld be
sele
cted
by
hi
stori
c
al
warra
n
ty data. The commo
n
simple
symm
etric copul
a functio
n
s in
clu
de Gau
s
sian
cop
u
la, t-co
p
u
la, Gumbel
cop
u
la, Clayt
o
n
cop
u
la an
d Frank
cop
u
la et
c. Gumb
el co
pula is:
1/
'
12
1
2
((
)
,
(
)
,
)
e
x
p
(
l
n
((
)
)
(
l
n
(
(
)
)
CF
t
F
u
F
t
F
u
(3)
For the two
-
dime
nsi
onal
imperfe
ct preventiv
e p
o
licy, the failure rate
ri
ght after
perfo
rming th
e jth PM acti
on is give
n b
y
Equation (4). Obviou
sly
,
the PM action effect is t
he
redu
ction of
failure inte
nsity [11].
10
B
nU
U
,
20
B
nW
T
(
*
: In
teger part of a
re
a
l
number)
.
00
0
0
0
00
0
0
0
P
M
a
t
(,
)
(
1
)
(,
)
PM
a
t
(
t
,
)
(
1
)
(
/
,
)
jT
h
j
T
u
h
j
T
j
T
r
jU
h
j
U
h
j
U
r
j
U
j=
1,2,…
(4)
4. Analy
t
ical
Warr
anty
Cost An
aly
s
is
For t
w
o
-
dime
nsio
nal
wa
rranty, there
a
r
e t
w
o
po
ssi
ble
situation
s
that ba
si
c
warra
n
ty
expire
s. One
is the u
s
a
g
e
of the pro
d
u
c
t exce
ed
s th
e mileag
e lim
itation. And the othe
r is th
e
age of the produ
ct excee
d
s
the ti
me limitation. So th
e warra
n
ty co
st inclu
d
e
s
two parts:
D
1
and
D
2
. For the
warranty cove
rage of D
1
, let
N
t
be the
nu
mber
of PM obse
r
ved u
p
to
time t. Then the
failure rate be
tween two PM action is:
t
N0
,
(,
)
ht
u
;
t
N1
,
00
0
(,
)
(
,
)
ht
u
h
U
r
U
;
…
t
N
j
,
1
0
0
0
0
()
(,
)
(
1
)
,
(
)
j
i
i
ji
U
ht
u
h
j
i
U
r
The cost of the wa
rranty inclu
d
ing two-di
men
s
ion
a
l imperfe
ct pre
v
entive maintenan
ce
is:
00
1
1
0
0
1
10
(1
)
1
1
0
1
0
00
0
0
1
10
10
0
0
0
()
(,
)
(
1
)
,
(
)
()
(,
)
(
1
)
,
(
)
B
jU
u
r
n
j
i
Dp
m
ji
jU
ur
U
n
i
m
i
nU
ji
U
C
C
n
C
h
t
u
h
j
i
U
d
td
tu
r
ni
U
Ch
t
u
h
n
i
U
d
t
d
t
u
r
(5)
At the same princi
ple, the cost of the
warranty including two
-
di
mensi
onal im
perfe
ct
preventive m
a
intena
nce for
the wa
rra
nty covera
ge of D
2
is
:
00
2
2
0
0
2
20
(1
)
1
1
2
00
0
00
0
1
20
20
0
0
0
(,
)
(
1
)
(
)
,
(
)
(,
)
(
1
)
(
)
,
(
)
B
jT
r
t
n
j
i
Dp
m
ji
jT
rt
W
n
i
m
i
nT
CC
n
C
h
t
u
h
j
i
T
j
i
T
r
d
u
d
t
C
h
t
u
h
n
i
T
n
i
T
r
du
dt
(6)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Optim
a
l Warranty Policy Consi
deri
ng T
w
o-dim
ens
io
nal Im
perfect Preventive…
(Xin-yue L
e
e
)
219
So the total expected b
a
se
d warra
n
ty co
st is:
1
2
1
2
12
12
(,
)
(
,
)
()
(,
)
(
,
)
(,
)
(
,
)
DD
BD
D
DD
DD
f
t
u
d
t
d
u
f
t
u
dt
du
EC
C
C
f
t
u
dt
d
u
f
t
u
d
t
d
u
f
t
u
dt
du
f
t
u
d
t
d
u
(7)
5.
Numerical E
xample
We use
the
a
s
ymmetri
c
co
pula pro
p
o
s
e
d
by
Wu(201
4). Th
e
symmetric copul
a
functio
n
whi
c
h is a
ppli
ed to con
s
tru
c
tion the a
s
ymmetric
cop
u
l
a is:
12
1
2
0
1
2
(
(
)
,
()
)
(
)
(
)
1
(
1
(
)
,
1
()
)
CF
t
F
u
F
t
F
u
C
F
t
F
u
1/
0
(,
)
e
x
p
l
n
(
)
l
n
(
)
Cx
y
x
y
is the
G
u
m
bel
cop
u
la.
Suppo
se
p
C
150,
m
C
820 a
nd
0.36
. The ba
se
d
warra
n
ty and
extended
wa
rra
nty cove
ra
ge a
r
e (5, 10
0000
)
and
(10, 2
0
0000
). The l
i
fe of
the produ
ct is(30,
6000
00).
0
0.9
p
,
1
0.1
p
,
1
3.76
and
2
0.
48
,
1
0.
7
8
,
2
0.77
,
1
2.6
,
2
4
840
0
. The marginal c
u
mulative failure
distrib
u
tion
s are:
1
1
1
()
1
e
x
p
t
Ft
,
2
2
2
()
1
e
x
p
u
Fu
In the ba
se
d
warra
n
ty, the optimal p
r
eve
n
ti
ve mainten
ance pe
riod
could b
e
obtai
ned by
min
i
mize
th
e
b
a
s
e
d
warra
n
t
y shown as
Figure 3. A
nd the partial results sho
w
ed
as Table 1. The
optimal warra
nty cost is 51
3.1398 when
the pr
eventive maintenance period
is (1.2, 24000). From
results b
e
low, we co
uld find the wa
rra
nty cost
will
decrea
s
e
s
at
first then in
crea
ses with
th
e
increa
sing of
T
0
or
U
0
.
Figure 3. Warranty cost curve of different
two-dimensio
nal preventive maintenance period
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 2, Februa
ry 2015 : 215 – 222
220
Table 1. Warranty cost of different
two-
dimensional preventive
maintenance perio
d
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
1000
0
1000.
7
725.8
701.2
708.8
715.2
720.4
749.8
772.5
790.9
2000
0
835.2
560.3
535.6
543.3
549.6
554.8
584.3
607.0
625.4
3000
0
820.4
545.5
520.8
528.5
534.8
540.0
569.4
592.2
610.5
4000
0
825.0
550.1
525.5
533.2
539.5
544.7
574.1
596.9
615.2
5000
0
828.8
554.0
529.3
537.0
543.3
548.5
577.9
600.7
619.0
6000
0
832.0
557.1
532.4
540.1
546.4
551.6
581.1
603.8
622.2
7000
0
849.7
574.8
550.2
557.9
564.2
569.4
598.8
621.6
639.9
8000
0
863.4
588.6
563.9
571.6
577.9
583.1
612.5
635.3
653.6
9000
0
874.5
599.6
574.9
582.6
589.0
594.2
623.6
646.3
664.7
Then we anal
yze the se
nsi
t
ive of the based
warranty cost mod
e
l b
y
changing
shown
as Figure 4. We could noti
c
e some interesting pheno
mena. Warra
nty cost of different PM inte
rval
change more
by increasing
1
and
2
. And
the optimal warranty cost is increa
sing when
1
is
increa
sing. O
n
the co
ntrary, it is decrea
s
ing when
2
is
increa
sing.
W
e
come
t
o
t
h
e con
c
lusion
that
1
and
2
have opposite influence on the based wa
rran
t
y cost.
Figute 4.
Sensitive analyze
of based warr
anty cost model considerin
g different
Then we a
nalyze the sensi
t
ive of
t
he based wa
rranty
cost model b
y
changing
shown a
s
Figure 5. And the optimal warra
nty cost is decreasin
g when
1
and
2
is increa
sing.
We com
e
to the conclusion that
1
and
2
have the uniform influence on the based warranty cost.
T
0
Co
st
U
0
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Optim
a
l Warranty Policy Consi
deri
ng T
w
o-dim
ens
io
nal Im
perfect Preventive…
(Xin-yue L
e
e
)
221
Figure 5. Sen
s
itive analyze
of based warr
anty cost model considerin
g different
6. Conclu
sion
This a
r
ticl
e propo
se
s a met
hod to find o
p
ti
mal two
-
di
mensi
onal
preventive mai
n
tenan
ce
perio
d by mi
nimizin
g
the
warra
n
ty cost
. The method
prop
osed co
uld help th
e manufa
c
turer to
price the ba
sed
warranty
prope
rl
y. Many extensio
ns to this work m
a
y be con
s
id
ere
d
. The
method of
fa
ilure red
u
ctio
n
mod
e
ls ha
ve
to
be fu
rther i
n
vestig
a
t
ed. And the
paramete
r
s
of
cop
u
la fun
c
tion have to
be theo
retica
lly studied.
In ord
e
r to check mo
del
s validity, some
good
ne
ss-of-
fi
t tests
sho
u
l
d
be devel
op
ed. At last, imperfe
ct corrective mainte
nan
ce shoul
d
be
inclu
ded in th
ese mo
del
s.
Referen
ces
[1]
Bai J, Pham H. Cost anal
ysis on rene
w
a
ble full-s
e
rvice
w
a
rr
anties for
multi-compo
n
ent s
y
stems
.
Europ
e
a
n
Jour
nal of Operati
o
nal
R
e
searc
h
2
006; 16
8(2): 49
2-50
8.
[2]
Bai J, Pham
H. Repair-lim
it
risk-free
w
a
rr
ant
y
policies w
i
t
h
imperfect repair
.
S
y
stems, Man and
C
y
ber
netics, Part
A: Sy
stems
and H
u
mans.
IEEE T
r
ansactions on.
2
005;
35(6): 76
5-7
7
2
.
[3]
Blischk
e
W
R
,
Murth
y
D
N
P. Product
w
a
rr
a
n
t
y
man
a
g
e
m
ent I: A
taxo
n
o
m
y
for
w
a
rr
a
n
t
y
p
o
lici
e
s.
Europ
e
a
n
Jour
nal of Operati
o
nal R
e
searc
h
. 199
2; 62(2): 12
7-14
8.
[4]
Blischke
WR, Murth
y
DNP.
W
a
rrant
y
C
o
st Anal
ys
is. CRC
Press, 1993.
[5]
Blischk
e
W
R
,
Murth
y
D
N
P., Product W
a
rra
nt
y
H
and
bo
ok
. CRC Press, 19
95.
[6]
Boug
uerra
S, Chel
bi
A, R
e
zg
N.
A
decisi
on
mode
l for a
d
o
p
ting
an
e
x
ten
ded
w
a
rra
nt
y u
nder
dif
f
ere
n
t
mainte
nanc
e p
o
lici
e
s.
Internat
ion
a
l Jour
na
l of Productio
n
Econo
mics.
201
2; 135(2): 84
0-8
49.
[7]
Chie
n
YH.
T
he ef
fect of a pr
o
-rata reb
a
te
w
a
rrant
y o
n
the
age r
epl
acem
e
n
t pol
ic
y
w
i
th
s
a
lva
ge va
lu
e
consi
derati
on. Reli
ab
ilit
y
,
IEE
E
T
r
ansactions
on.
201
0; 59(2
)
: 383-39
2.
[8]
Che
n
T
,
Popova E. Maintenance po
lici
e
s w
i
t
h
t
w
o-
dim
e
nsio
nal
w
a
rr
an
t
y
.
Relia
bil
i
ty
Engi
neer
in
g &
System
Safety
200
2; 77(1): 61
-69.
[9]
Chuk
ova S, Arnol
d R, W
ang
DQ. W
a
rrant
y anal
ys
is: An
appr
oach to m
ode
lin
g imp
e
rfect repairs
.
Internatio
na
l Journ
a
l of Prod
uction Eco
n
o
m
ics
. 2004 8
9
(1)
:
57-68.
[10]
Chuk
ova S, Johnston MR.
T
w
o-dime
nsi
ona
l w
a
rrant
y
r
e
p
a
ir
strateg
y
base
d
on minima
l and compl
e
t
e
repa
irs.
Mathe
m
atic
al a
nd co
mp
uter mod
e
li
ng
. 200
6; 44(1
1
): 1
133-1
1
4
3
.
[11]
Do
ye
n L, G
a
u
doi
n O. Cl
ass
e
s
of
imperfec
t
repa
ir mo
del
s bas
ed
on
re
ductio
n
of
fail
u
r
e i
n
tensit
y
o
r
virtual a
ge.
R
e
li
a
b
i
l
i
t
y En
gi
ne
eri
n
g
& System
Sa
fe
ty.
2004; 8
4
(1): 45-5
6
.
[12]
Gertsbakh IB, Kordo
n
sk
y KB. Parall
el time
scales and
t
w
o-dime
nsi
ona
l manufactur
e
r and ind
i
vid
u
a
l
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 2, Februa
ry 2015 : 215 – 222
222
customer w
a
rr
anties.
IIE transactions.
19
98;
30(12): 11
81-
118
9.
[13]
Hua
ng YS, Y
e
n C. A stu
d
y
of t
w
o
-
d
i
mens
i
ona
l w
a
rrant
y
polic
ies
w
i
th
p
r
eventiv
e
ma
in
tenanc
e.
IIE
T
r
ansactio
n
s
. 200
9; 41(4): 29
9-30
8.
[14]
Iskandar
BP,
Murth
y
DNP,
Jack N. A
ne
w
rep
a
ir
–rep
la
ce strateg
y
f
o
r items so
ld
w
i
t
h
a t
w
o-
dime
nsio
nal w
a
rrant
y.
Co
mp
uters & Operati
ons Res
earch.
200
5; 32(3): 66
9-68
2.
[15]
Iskandar BP
, Murth
y
DNP
. Repair-re
pla
c
e stra
tegies
for
t
w
o-
dim
ensi
ona
l
w
a
rr
ant
y
p
o
lic
ies
.
Mathe
m
atic
al a
nd Co
mputer
Mode
lli
ng.
20
0
3
; 38(1
1
): 12
33
-124
1.
[16]
Jack N, Iskandar BP
, Murthy
DNP
.
A
repair–
repl
ac
e strateg
y
bas
ed o
n
usage rate for ite
m
s sold
w
i
t
h
a
t
w
o-
dime
nsi
o
n
a
l w
a
rra
nt
y
.
R
e
liab
ility En
gin
e
e
rin
g
& System Safety
.
2009; 94(2): 61
1-61
7.
[17]
Mosko
w
i
tz H,
Chun
YH.
A
poisson re
gr
essi
o
n
mode
l
for t
w
o-
attrib
ute
w
a
rrant
y
polic
ies.
Nava
l
Research Logistics.
1994; 41(
3): 355-3
76.
[18]
Murth
y
D
N
P
,
Blischke W
R
. W
a
rrant
y
ma
na
ge
ment and pr
od
uct manufactur
e
.
Spring
er
.
20
06.
[19]
Hua
ng YS, Z
huo YF
. Estimation of future
br
eak
do
w
n
s to
determin
e
op
timal
w
a
rra
nt
y polic
ies for
prod
ucts w
i
t
h
d
e
terior
ation.
R
e
lia
bi
lity Engi
n
eeri
ng & System Safety.
200
4
;
84(2): 163-1
6
8
.
[20]
Murth
y
D
N
P
,
W
ilson RJ.
Mo
dell
i
n
g
tw
o-dimensi
ona
l fail
ur
e free w
a
rranti
e
s
. Procee
din
g
s of the F
i
fth
S
y
mp
osi
u
m on
Applie
d stocha
stic models a
n
d
dat
a a
nal
ys
is
. 1991. (W
orld
Scientific, Sin
g
apor
e).
[21]
Park M, Pham H.
Altered quas
i-ren
e
w
a
l
c
oncepts for
model
ing r
e
n
e
w
a
ble
w
a
rra
nt
y
costs
w
i
t
h
imperfect repairs.
Mathematic
al an
d Co
mp
ut
er Mode
lli
ng.
2
010; 52(
9): 143
5-14
50.
[22]
Park M, Pham H. W
a
rrant
y c
o
st an
al
ys
es usi
ng
qu
a
s
i-rene
w
a
l
pro
c
esses for m
u
lticom
pon
en
t
s
y
stems. S
y
ste
m
s, Man and
C
y
ber
neti
cs, Part
A: Sy
stems
and H
u
ma
ns.
IEEE T
r
ansactions on.
201
0;
40(6): 13
29-
13
40.
[23]
Park M, Pham
H.
A
ne
w
w
a
r
r
ant
y
po
lic
y
w
i
th failur
e
times
and
w
a
rra
nt
y
servicin
g times
.
Relia
bi
lity
,
IEEE T
r
ansactions on
. 2
012;
61(3): 82
2-8
3
1
.
[24]
Shafie
e M, C
h
ukova
S. Mai
n
tenanc
e m
ode
l
s
in
w
a
rr
ant
y
:
A liter
ature r
e
vie
w
.
E
u
rop
e
a
n
Jo
urna
l
of
Operatio
nal R
e
search
. 20
13;
229(
3): 561-
57
2.
[25]
Su C, She
n
J.
Anal
ys
is
of e
x
t
end
ed
w
a
rra
nty p
o
l
i
cies
w
i
th
dif
f
ere
n
t rep
a
ir
optio
ns.
Eng
i
n
eeri
ng F
a
i
l
ur
e
Analys
is
. 201
2; 25: 49-62.
[26]
V
a
rnos
afader
a
n
i S, Chuk
ova
S.
A
t
w
o-
dime
nsio
nal
w
a
rra
n
t
y
servici
ng str
a
teg
y
b
a
sed
o
n
reducti
on i
n
prod
uct failure
intens
it
y
.
Co
mp
uters & Mathe
m
atics w
i
th
Ap
plicati
ons.
2
0
1
2
; 63(1): 20
1-2
13.
[27]
W
u
S, Longh
u
r
st P
.
Optimising ag
e-rep
l
ac
em
ent an
d e
x
tend
ed no
n-re
n
e
w
i
n
g
w
a
rr
ant
y po
lici
e
s in
lifec
ycle c
o
stin
g.
Internatio
nal
Journa
l of Productio
n
Econ
o
m
ics
. 20
1
1
; 13
0: 262–
26
7.
[28]
W
u
S. Construction of as
ymm
e
tr
ic copu
las a
nd its ap
plic
ati
on in t
w
o-
dime
nsio
nal re
lia
bi
li
t
y
mod
e
ll
ing
.
Europ
e
a
n
Jour
nal of Operati
o
nal R
e
searc
h
.
201
4; 238(
2): 476-4
85.
[29]
Y
un W
Y
, Kang
KM. Imperfect repair p
o
lic
ie
s und
er t
w
o
-
di
mensi
ona
l
w
a
r
r
ant
y
.
Proc
ee
d
i
ngs of th
e
Institution of Mecha
n
ica
l
Eng
i
neers.
Part O:
Journ
a
l of Risk
and Re
lia
bil
i
ty
.
2007; 2
21(4):
239-
247.
Evaluation Warning : The document was created with Spire.PDF for Python.