TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 13, No. 3, March 2
015,
pp. 546 ~ 55
4
DOI: 10.115
9
1
/telkomni
ka.
v
13i3.712
2
546
Re
cei
v
ed
De
cem
ber 1, 20
14; Re
vised Janua
ry 3
1
, 20
15; Accepted
Febrary 14, 2
014
Financial Feasibility Study of Waste Cooking Oil
Utilization for Biodiesel Production Using ANFIS
Imam Ahmad*
1
, Irman Hermadi
2
, Yandra Ark
e
man
3
1,2
Departement
of Computer S
c
ienc
e, F
a
cult
y of Mathematic
s and Natur
a
l
Scienc
es,
Bogor Agr
i
cult
ural U
n
ivers
i
t
y
,
1668
0 Bog
o
r,
Indon
esi
a
, Ph/F
ax: +
62-2
51-6
284
48/6
229
61
3
Departem
ent of Agroin
dustri
a
l T
e
chnolo
g
y
, F
a
cult
y
of Agri
cultura
l
Engi
ne
erin
g and T
e
ch
nol
og
y,
Bogor Agr
i
cult
ural U
n
ivers
i
t
y
,
1668
0 Bog
o
r,
Indon
esi
a
, Ph/F
ax: +
62-2
51-8
620
22
4/862
19
74
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: imamahm
ad
666
@gma
il.co
m
1
, Irmanherm
adi@
a
p
p
s.ip
b.ac.id
2
,
ya
nd
a_i
pb
@
y
a
hoo.com
3
A
b
st
r
a
ct
Cons
u
m
ptio
n o
f
fuel oil is
incr
eas
i
ng, but it is
not acco
mp
an
i
ed by
su
pp
ly. T
r
ansp
o
rtatio
n sector is
spent th
e
most existi
ng fu
el.
T
herefore,
it is
crucia
l to
lo
ok
for alt
e
rnativ
e
type of fu
el
su
ch as
bi
odi
ese
l
to
overco
me the
fuel sh
ortag
e
.
The purp
o
se
of this
study
is to inv
e
stig
ate the fe
asib
ility o
n
how
t
h
e
invest
me
nt of
w
a
ste cooki
n
g
oil
as
the
bi
o
d
ies
e
l
mate
ria
l
pro
ductio
n
. T
h
is stu
d
y co
nsi
s
ts of tw
o ph
a
s
es.
First is calcu
l
a
t
ing feas
ib
ility
fu
zz
y
mode
l w
i
th the i
n
p
u
t pr
ice of b
i
o
d
ies
e
l, w
a
ste cooki
ng o
il pr
ices
a
n
d
interest rates,
and
as its outp
u
t are N
e
t Pre
s
ent Val
ue (N
PV), Internal R
a
te of Retur
n
(
I
RR), Net Ben
e
fit
Cost Ratio (N
et B/C) and Payback Per
i
od
(PBP). Sec
ond is pred
icting
the feas
ibi
lity of using Ad
apt
i
v
e
Neur
o F
u
zz
y
Inferenc
e Syste
m
(ANF
IS). T
he resu
lts of
the
ana
lysis for e
a
ch type
of me
mb
ershi
p
funct
i
o
n
(mf) w
e
re obtai
ned tria
ng
ular
accuracy
77%,
accuracy g
aus
sian 5
3
% a
nd trape
z
o
id
al acc
u
racy 61
%.
Ke
y
w
ords
: AN
FIS, biodiese
l
, feasib
ility, w
a
ste cooki
ng o
i
l
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
Energy is a necessity that
cannot be
sep
a
rate
d from human li
fe, such a
s
fuel oil
nowaday
s co
nsum
ed in
cre
a
sin
g
ly. Agen
cy for the A
s
sessment a
nd
Applicatio
n of
Tech
nolo
g
y in
Indone
sia
rev
ealed
that th
e final
ene
rgy
co
nsumpt
io
n
ba
sed
on
typ
e
in
201
1
wa
s d
o
min
a
ted
by
the fuel, for e
x
ample die
s
el
fuel (46%
), g
a
soli
ne
(42%), jet fuel (6%), kerosene
(3
%) and fu
el
oil
(3%) [1
-2]. T
hehig
h
con
s
u
m
ption of di
e
s
el i
s
trig
ge
red by
con
s
u
m
ption of the
motor ve
hicl
e.
Therefore, th
e govern
m
ent
through the
dire
ctor'
s
de
cision a
bout Rene
wable En
ergy and En
e
r
gy
Con
s
e
r
vation
Numbe
r
: 72
3 K / 10 / DJE / 2013 enforce
d that 10
% blending in
to diesel fuel and
biodie
s
el
qua
lity standa
rd
s for di
strib
u
tion. Ma
ch
mu
d su
gge
sted
that by mixing 10%
(B10
) in
diesel fuel, it
woul
d gene
ra
te gas emi
ssi
ons (CO and
CO
2
) in the lo
we
st diesel vehicl
es [3-4].
Biodiesel is
a
biofuel, whi
c
h is ma
de fro
m
v
egetable
oils o
r
anim
a
l
fats [5]. The pro
c
e
s
s
of makin
g
bi
odie
s
el can
be don
e by
esterifi
cation
and tra
n
sest
erificatio
n [6-7]. Biodiesel
is a
rene
wa
ble li
mit fuel and
environm
ent
ally fri
endly, and also it produ
ce
e
m
issi
on which is
relatively cl
ea
ner th
at die
s
e
l
fuel [8]. In In
done
si
a, th
ere are 5
0
type
s of
ra
w m
a
te
rials which
ca
n
be used to produ
cebi
odie
s
el [9]. Howev
e
r, there a
r
e
only six raw
material type
s potentially
use
d
for produ
cin
g
biodie
s
el, n
a
m
ely palm oil
,
coconut, al
g
ae, ru
bbe
r, a
nd waste
co
o
k
ing
oil. Of the
six types, cooking
oil has t
he continuous availabilit
y.
Therefore, waste cook
ing
oils i
s
chosen as
the raw m
a
te
rial for this rese
arch si
nce frying
usin
g wa
ste coo
k
ing oil is still
a great co
n
c
ern
[10].
Addition, this
resea
r
ch is
carri
ed out in o
ne of
biodi
esel indu
strie
s
i
n
Bogor. A
c
cordin
g to
data from th
e
Nation
al Socioeconomi
c
S
u
rvey in
20
1
3
, the co
nsu
m
ption of ho
use
hold
co
oking
oil in Indon
esia is 8.9 lite
r
s / capita / year. Ac
cording
to the Cent
ral Statistics
Agency in
20
13,
the populatio
n of Bogor h
a
s the den
se
st popul
at
ion
in West Jav
a
, namely 4,989 million [
2
].
Furthe
rmo
r
e,
Environme
n
t
al Manag
em
ent Agen
cy Bogor
re
co
rd
ed in 20
11,
16.090 lite
r
s of
wa
ste coo
k
in
g oil per yea
r
can b
e
pro
c
e
s
sed in
to biodie
s
el a
s
12,050 liters.
With the large
numbe
r of p
opulatio
n an
d the larg
e
numbe
r of wast
e
coo
k
ing
oil, it is expected to a
c
ti
vely
contri
bute in
colle
cting
wa
ste co
oki
ng o
il. Produc
i
ng
biodie
s
el fro
m
wa
ste co
o
k
ing oil h
a
s t
w
o
purp
o
ses n
a
m
ely incresi
n
g food safet
y
of public
and en
ch
aci
ng the re
ne
wabl
e ene
rg
y to
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Financial Feasibility Study of
Waste
Cooking Oil
Utilization for Biodi
esel
… (Im
a
m Ahm
ad)
547
produce energy [11]. Thus
this study is
aimed to determine
the feasibility of the waste cooking
oil availability colle
cted from the com
m
unity
which
can be
re
cycled a
s
bio
d
iesel and gi
ves
recomme
ndat
ions a
bout fe
asibility of bio
d
iesel pro
d
u
c
tion eco
nomi
c
ally.
Some research h
a
ve be
en
done
rel
a
ted
to feasibility
and bi
ode
sel
prod
uctio
n
.Amalia et
al ha
s b
een i
n
vestigate
d
the colle
cting
wa
ste coo
k
in
g oil in Bo
go
r. In ord
e
r to
see whethe
r o
r
not
biodie
s
el ind
u
stry is fea
s
i
b
ly establish
ed, it
require
s feasi
b
ility study
to determine wh
ether the
industry
can
provide future benefit [12]
. Morove
r, bi
odiesel feasi
b
ility st
udies
have been carried
out by
Widodo. He investi
gated th
e feasibility of biodiesel indust
ry
in producing biodiesel from
wa
ste co
oki
n
g oil usin
g co
nventional m
e
thod
s [13
]. Another
stud
y was al
so do
ne by Martini
who
investigate
d
f
u
zzy inve
stm
ent to
analyze the
finan
ci
a
l
feasi
b
ility of
desertificatio
n
ind
u
st
ry ba
sed
on su
garca
n
e
[14]. Furtherm
o
re, Sh
amshi
r
b
and
et alstudie
d
about ada
p
t
ive neuro f
u
zz
y
optimation of
wind farm
net pr
ofit to assess the f
easi
b
ility
of
wind power-based electri
c
ity
indu
stry [15]. Zlende
r et al have also
invest
igated
a feasibility analysi
s
of u
nderground
gas
stora
ge[16].
Ho
wever,
pre
s
ent
re
sea
r
ch propo
se
s t
o
dev
el
op a financi
a
l
fea
s
ibility
study model of
ANFIS in utilizing
waste co
oking oil as
material for bi
odiesel
production. Financi
al
fasibilty study
is importantly investigated as in
its process, there are many va
ri
ables affected feasibility of a
investment
comparing to
other aspect
s of feasibilit
y. Further, ANFIS, whic
h i
s
a com
b
ination of
two meth
ods
of Neu
r
al
Net
w
ork
(NN) an
d fuzzy
logi
c, has
advanta
ges
of lea
r
ni
ng (NN), pa
rt
ial
truth, an
d b
e
able to
explai
n the
process of re
a
s
oni
ng
(fu
zzy) [17].
Therefore,
ANFIS is
cho
s
en
as it is able to
solve un
cert
ainty
that can
predi
ct an in
put adaptivel
y.
2. Rese
arch
Metho
d
2.1. Rese
arh
Data
Data is
colle
cted from E
n
vironm
ental
Manage
men
t
Agency (B
PLH) Bog
o
r
in 2008
-
2013
and PT.
Meka
nika Ele
k
tri
c
a Egra (MEE). This
study is inve
stigated u
s
ing l
aptop comp
uter
with Processor Intel
Penti
u
m(R) CP
U
P6100
@2
.0
0GHz,
ran
d
o
m
a
c
cess m
e
mory
3 GB.
The
softwa
r
e
s
u
s
ed are: Wind
ows 7 a
s
op
eration
syste
m
, Micro
s
oft
Excel for dat
a analysi
s
a
nd
Matlab for progra
mming.
2.2. Determining Feasibilit
y
Criteria
Feasi
b
ility study, basi
c
all
y
, is used
to
determine business f
easi
b
ility based on
investment criteria. Fea
s
i
b
ility is an instrum
ent
in makin
g
de
cision to finance an investm
ent
proje
c
t b
a
se
d on
a te
ch
n
i
cal, e
c
o
nomi
c
, an
d fi
na
ncial [18-19]. P
r
oje
c
t fea
s
ibil
ity assessme
nt
sho
u
ld meet
some
crit
e
r
ia
[
13]
,
such a
s
,
(1)
Net Pres
ent Value (NPV), (2) Internal Rate of
Return (I
RR),
(3) Net Benefit Cos
t
Ratio
(Net
B/C) and (4) Payback
Pe
riod (PBP)
(s
ee Table
1).
The
crite
r
ia a
r
e
cho
s
e
n
a
s
they are
pa
rt of in
vestme
nt crite
r
ia
whi
c
h the
analy
s
is is ba
sed
o
n
ca
sh f
l
ow.
Table1. Fuzzy Criteria in
Feasibility Assessment
Cr
iter
ia Not
F
easible
Fairl
y
F
easible
Feasible
Strongl
y
Feasible
NPV
1
< 0
0 < NPV
≤
10% *
Inv
8% * Inv < NPV
≤
17% * Inv
NPV > 15% * Inv
IRR
2
<
r
r
≤
IRR
≤
6% +
r
4% + r < IRR
≤
1
5
% + r
IRR
≥
r + 1
2
%
B/C R
3
< 1
1 < B/C
≤
1,3
1,25 < B/C
≤
1,75
B/C > 1,6
PBP
4
>10
10 > PBP
≥
5
5,5 < PBP
≤
1,25
PBP
≤
1,5
Source: Martini (
2010)
2.3. Pre-proc
ess
It needs to identify and calcul
ate the feasi
b
ility study from eco
nomic a
s
p
e
ct
s before
investing,
su
ch a
s
, identif
ying the assumption
s,
investment
co
st, operational
co
st com
p
ri
si
ng
fixed and
variable
co
sts. T
he first p
r
o
c
e
s
s is to
creat
e fuzzificatio
n value
in i
n
put varia
b
le
s for
biodie
s
el
pri
c
e, wa
ste
coo
k
ing
oil p
r
i
c
e
and
interest
rate
by u
s
in
g
Tri
ang
ular Fuzzy
Num
ber
(TF
N
) [20]. Feasi
b
ility of th
is study is as
sessed based on
four
criteria as follows.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 3, March 2
015 : 546 – 5
5
4
548
2.3.1. NPV Fuzz
y
Acco
rdi
ng to Chiu d
an Park [21], when
ca
sh flow i
s
influen
ced by
intere
st rate
s by using
the fuzzy investment, fu
zzy NPV can
b
e
cal
c
ul
at
ed i
f
there a
r
e fu
zzy p
a
ra
mete
rs. Th
e eq
uat
ion
(1) i
s
:
∑
∏
Whe
r
e:
Ft = Ca
sh Flo
w
i = Fuzzy rep
r
ese
n
tation (T
FN)
PVi =
(
ai, b
i
, c
i
)
Each PV (TF
N
) ha
s three
values, they are:
PV
1
= (
a
1
, b
1
, c
1
)
PV
2
= (
a
2
, b
2
, c
2
)
PV
3
= (
a
3
, b
3
, c
3
)
,0
∐
1
,0
∏
1
∏
1
,
,0
∐
1
,0
∏
1
2.3.2. Bene
fit Cos
t
R
a
tio
(B/C
Ra
tio) F
u
z
z
y
B/C Ratio
is
value
comp
aring bet
wee
n
proje
c
t b
enefi
t
and
co
st of
a proje
c
t spe
n
t. This
study ado
pte
d
Kahram
an [22] method u
s
ed Equ
a
tion
(3) a
s
follo
ws.
/
∑
∑
,
∑
∑
(
3
)
2.3.3. Interna
l
Rate o
f
Re
t
u
rn (IR
R
) Fu
zzy
Similar to NP
V value, IRR is also un
ce
rtain.
IRR i
s
the retu
rn of i
n
vestment b
a
s
ed o
n
intere
st rate p
e
r year. Thi
s
study is u
s
ed
Equation (4)
as follo
ws.
∑
0
(
4
)
2.3.4. Pa
y
b
ack Period (P
BP) Fu
zzy
PBP refers to the estimation in
what year the investm
ent va
lue
will
return. I is the value
of investment
, Ab is net benefit obtained
. Th
is study is used Equ
a
tio
n
(5) a
s
follo
ws.
PBP
(
5
)
2.4. Rule Ba
se IF
-
TH
EN
Value in
put i
n
this
study i
s
fu
zzy va
ria
b
les
wh
ere fo
r the
biodi
ese
l
pri
c
e,
wa
ste
co
okin
g
price a
nd int
e
re
st rates
a
s
sumed
by t
he lo
w,
mo
d
e
rate
and
hi
gh
crite
r
ia. T
hese three i
nput
variable
s
an
d
three criteria
are obtai
ned
3
3
= 27 as a f
easi
b
ility rule base.
2.5. Dev
e
lop
m
ent of
ANF
IS
Figure 1. (a)
Sugeno M
o
d
e
l Orde 1; (b) Archite
c
ture
ANFIS (Ja
ng
1993
)
(2)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Financial Feasibility Study of
Waste
Cooking Oil
Utilization for Biodi
esel
… (Im
a
m Ahm
ad)
549
ANFIS (Adap
tive Neuro F
u
zzy Inferen
c
e System) i
s
a com
b
inat
ion of two method
s
namely n
eural network a
nd fu
zzy infe
ren
c
e
syste
m
(FIS). FIS u
s
ed
in thi
s
study is Su
ge
no
Model O
r
de
r 1. The com
b
i
nation of bot
h re
sults
ad
a
p
tive models
to the data input [15]. ANFIS
architectu
re
showed in Fig
u
re 1.
ANFIS architec
ture c
o
ns
ists
of five laye
r [23], as
follows
:
1.
Each
no
de i
in th
e first
layer i
s
ada
ptive towa
rd
s p
a
ramete
rs of a
n
a
c
tivation
function. The
output of each nod
e in the fo
rm of membershi
p
deg
ree
s
given by the
input memb
e
r
shi
p
fun
c
tion
s. This
study
is use
d
a tri
angul
ar mem
bership fun
c
ti
on
(Figu
r
e 2
)
wit
h
the Equatio
n (6) i
s
:
O
,
0
;
;
;
(
6
)
Figure 2. Tria
ngula
r
Fu
zzy
Numb
er (TFN)
Whe
r
e
{a,
b,
c}
are the
pa
rameters, b
=
1 a
s
an
ab
sol
u
te value.
If the valu
e of
th
ese
para
m
eters
are
cha
nge
d
,
then the shape
of
the
curve will also ch
ang
e.
The
para
m
eters in
this layer is u
s
ually called t
he pre
m
ise p
a
ram
e
ters.
2.
Each
nod
e in
the second l
a
yer i
s
fixed
node
s whi
c
h its
output
s re
sulted
f
r
om e
n
tire
input sign
al. In gene
ral, AND op
erat
or i
s
use
d
. Each
node represe
n
ts
α
pre
d
icate o
f
the i-th rule (Equation 7
)
.
O
,
w
i
µ
µ
(
7
)
3.
Each no
de in the third layer is fix
ed nodes
whi
c
h are the re
sult of the ratio
cal
c
ulatio
n of
α
predi
cate
(wi
)
, from th
e rule
s of i
-
th on the ov
erall a
m
ount
of
α
predi
cate. Eq
uation (8
) a
s
follow:
O
,
(
8
)
4.
Each no
de i
n
the fourth
layer is ad
aptive node
to an output
. With
wi
i
s
t
he
norm
a
lized firing st
rengt
h o
n
the third lay
e
r a
nd {
p
i
, q
i
, r
i
} a
r
e the
pa
rameters o
n
the
node. The p
a
r
amete
r
s in th
is layer a
r
e
c
a
lled co
nsequ
ent para
m
ete
r
s Equ
a
tion (9).
O
,
(
9
)
5.
Every neuro
n
in the fifth layer is a fi
xed node th
at is the su
m of all inpu
ts, as
Equation (10) below.
O
,
∑
∑
∑
(
1
0
)
ANFIS development do
ne
using Matla
b
,
the fi
rst step input value
obtained fro
m
the 27
con
d
ition
s
of feasibility. Total data obtained from
th
e conditio
n
s
of feasibility are 81
0 pairs o
f
data. Ho
wev
e
r, validation
of the total data wa
s
o
b
tained 7
08
valid data. The re
sult of this
validation is i
nput into ANF
I
S with 85% trainin
g
data a
nd 15% testin
g data.
(7)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 3, March 2
015 : 546 – 5
5
4
550
2.6. Analy
s
is
and Ev
aluation
A
naly
s
is
wa
s
car
r
ied out
by
compa
r
i
ng t
he re
sul
t
s f
r
om bot
h
conv
ent
io
na
l f
u
zzy
feasibility mo
del and ANFI
S model pre
d
iction devel
oped. The lat
t
er model re
sults small e
r
ror,
accuracy, a
nd co
nsi
s
te
ncy. Evaluation don
e is chan
ging
mf triangula
r
, gaussian
and
trape
zoid
al. Testing
wa
s carrie
d out with a pair o
f
input data and tho
s
e three mem
b
e
r
ship
function
s.
3. Results a
nd Discu
ssi
on
3.1. Data o
f
Res
earc
h
Data
were ob
tained fro
m
PT. MEE and BPLH Bogo
r
whi
c
h can b
e
see
n
in Tabl
e 2. The
f
u
zzy
v
a
r
iabl
e value
s
a
r
e
biodie
s
el
pri
c
e, wa
ste
coo
k
ing
oil p
r
ice,
and inte
re
st
rate. All thre
e
are
rep
r
e
s
ente
d
in TFN a
s
low, moderate a
nd high
con
d
i
t
ion (se
e
Figu
re 3).
Table 2. Biodi
esel Pri
c
e a
n
d
Wa
ste Co
o
k
ing Oil in the
Last Six Years
No Yea
r
Waste
Cooking Oil
(Liter/
y
ear
)
Waste Cooking
O
il Pr
ice
(Rp/Liter
)
Biodiesel
Production
(Liter/
y
ear
)
Biodiesel
Price
(Rp/Liter
)
1
2008
3,120
2,500
2,496
6,500
2
2009
5,984
2,500
4,787
6,500
3
2010
10,950
3,000
8,760
6,500
4
2011
16,090
3,000
12,050
6,500
5
2012
16,600
3,000
13,280
9,000
6
2013
91,565
3,000
68,961
9,000
Source: BPLH 2
012
(a)
(b
)
(c)
Figure 3. Rep
r
esentation of
TFN (a
) biodi
esel
p
r
ice, (b) WCO p
r
i
c
e, and (c) inte
re
st rate
Data
need
ed
for the
fea
s
ibility of fuzzy
model
wa
s
origin
ally collected
in the
form
of
a
percenta
ge
o
f
biodie
s
el
p
r
odu
ction i
n
t
he first, se
co
nd, third u
n
til the te
nth ye
ar. Fu
rthe
rm
ore,
the other
assumptio
n
s are: the
percenta
g
e
of biodie
s
el sal
e
s, i
n
vestment
co
sts
(Rp.6
67.12
8.000),
fixed
co
sts
(Rp.96
.000.000
) a
n
d
varia
b
le
costs (Rp.2
5
0
.
800.000
). T
he
sup
portin
g
a
s
sumptio
n
s a
r
e the
co
st of
the inve
stmen
t
whi
c
h i
s
10
0% own
capit
a
l an
d bi
odie
s
el
prod
uctio
n
as much a
s
3 times a day, 1
50 liters p
e
r p
r
odu
ction in 2
8
workin
g da
y in a month.
3.2 Rule Bas
e
IF-THE
N
Rule b
a
se used in this stu
d
y were o
b
ta
ined from the
conditio
n
s o
f
27 feasibilit
y data
(Tabl
e 3) in
whi
c
h the co
ndition
s are
whe
n
pri
c
e
s
and interest rate fluctuatio
ns chan
ging t
o
be
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Financial Feasibility Study of
Waste
Cooking Oil
Utilization for Biodi
esel
… (Im
a
m Ahm
ad)
551
low, m
ode
rat
e
an
d
high.
T
he p
r
o
c
e
s
s of
takin
g
cri
s
p
v
a
lue
of the
fo
ur i
ndi
cators;
NPV fuzzy, I
R
R
fuzzy, B/CR f
u
zzy and PB
P fuzzy is
by giving wei
g
h
t
to each in
di
cator.
Dete
rmination of
crisp
values in thi
s
fuzzy fea
s
ibi
lity study use
s
Ce
ntroi
d
m
e
thod by giving wei
ght for each i
ndicator
and
when it i
s
added together yielding
value as
1 [24]. Each indi
cator
has a
weight to PBP =
0,25, NPV = 0,25, IRR = 0
,
25 and B/
CR = 0,25. Furtherm
o
re, the f
easibility crit
eria are divid
ed
into fourn
a
m
e
ly not feasi
b
le (0
), fairly feasible
(1
), feasible
(2
) and st
ron
g
ly feasibl
e
(3
). The
determi
nation
of 27 feasibili
ty condition
s sho
w
n in p
a
rt
in Table 3.
Table 3. Determination of
Feasi
b
ility Conditions
No
Biodisel
Price
WCO
Price
inter
e
st
rate
Feasibility
Agregation
1 Lo
w
High
Lo
w
PBP
-68.94
Not
Feasible
0
Not
F
easible
0
NPV
-759,573,
220
Not
Feasible
0
IRR
-0.39
Not
Feasible
0
B/C R
0.76 Not
Feasible
0
2 Lo
w
Moderate
Lo
w
PBP
9.65 Fairl
y
F
easible
1
Fai
r
l
y
F
easible
1.25
NPV
274,129,279
Strongl
y
Feasible
3
IRR
0.08 Not
Feasible
0
B/C R
1.04 Fairl
y
F
easible
1
3 Lo
w
Lo
w
High
PBP
3.68 Feasible
2
Feasible 2.00
NPV
539,213,403 Strongl
y
Feasible
3
IRR
0.20 Feasible
2
B/C R
1.18 Fairl
y
F
easible
1
4 High
Lo
w
Lo
w
PBP
1,51 Feasible
2
Strongl
y
F
easible
2.75
NPV
4,162,433,24
8
Strongl
y
Feasible
3
IRR
1.04 Strongl
y
Feasible
3
B/C R
2.08 Strongl
y
Feasible
3
3.3. Dev
e
lop
m
ent of
ANFIS Model
In develop
me
nt of ANFIS, there
are
trai
ni
ng d
a
ta an
d testing
dat
a. Total data
obtaine
d
from the fuzzy feasibility
study is 810
data, but no
t all
of the data i
n
cluded in the ANFIS
model.
This is be
cau
s
e th
ere
a
r
e
some
data
wi
th larg
e e
r
ro
r
excee
d
ing
0.
5; therefore, i
t
is o
b
taine
d
708
valid data
after valid
ation
sho
w
n i
n
Ta
ble 4. Th
e d
a
ta are trai
ne
d by a
s
mu
ch as 10
0 ep
o
c
h
s
and erro
r tole
ran
c
e 0.01.
Table 4. The
Total Data Be
fore and After Validation
No F
easibility
Data
Invalid
Valid
1 Not
Feasible
120
108
2 Fairl
y
F
easible
150
145
3 Feasible
420
390
4 Strongl
y
Feasible
120
65
Total data
810
708
3.4. Analy
s
is
and Ev
aluation
The sen
s
itivity analysis i
s
perfo
rmed
by co
mp
arin
g the co
nventio
nal mod
e
l an
d ANFIS
develop
ed. Wido
do’
s stu
d
y, usi
ng
co
nventional m
odel, showed
t
hatbiodie
s
el
prices
de
cre
a
se
24%, resulting fairly feasi
b
le cr
iteria
wi
th cri
s
p valu
e as
(1.5)
an
d the pri
c
e
o
f
waste
co
oki
ng
increa
sed
by
57%
with fai
r
ly fe
asible criteria (1.5).
ANFIS mo
de
l is d
e
v
e
l
op
ed
w
i
th th
e same
input value fo
r 57% in
crea
sing
biodie
s
el
price resultin
g fairly feasi
b
le crite
r
ia
wit
h
crisp val
ue
as
0.7 and
24%
decre
asin
g
wa
ste coo
k
in
g oil p
r
ice re
sult
ing fai
r
ly feasi
b
le
crite
r
i
a
(0.9
). It means
that ANFIS m
odel d
e
velop
ed have
bee
n su
cce
ssful
l
y
approa
ched
the fea
s
ibility of conve
n
tiona
l
model. Re
du
ction exceed
ed the value
of the
investment crite
r
ia i
s
not feasi
b
le
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 3, March 2
015 : 546 – 5
5
4
552
In this sta
ge,
evaluation i
s
ca
rrie
d
out
by
testing u
s
i
ng pai
rs
of in
put data. Mo
reove
r
,
testing of d
a
ta is d
one
by usin
g testin
g
data fr
om
training
re
sult. Testin
g is
al
so d
one
by 1
00
epo
ch a
nd 0.
01 e
rro
r b
a
la
nce
and th
e result
s succe
s
sfully app
roa
c
h ta
rget valu
e. The
re
sult
of
testing is
sho
w
ed in Ta
ble
5.
Table 5. Sam
pel of Data T
e
sting in ANF
I
S
No Input
Output
Biodiesel
price (Rp)
Waste cooking
oil price (Rp)
Interest rate
(%)
Target
Model
Error
1 9000
3000
12
2
2.2379
-0.2379
2 10500
3000
12
3
3.3544
-0.3544
3 11000
5300
26
1
0.6728
0.3272
4 9000
5300
12
0
0.3804
-0.3804
In addition, testing of fea
s
ibility ANFI
S model u
s
i
ng vario
u
s t
y
pes of me
mbershi
p
function
su
ch
as t
r
iang
ula
r
, and trape
zo
idal ga
us
sia
n
. As sho
w
ed
in Figu
re
4(a), (b)
and
(c),
each type of
membe
r
ship functio
n
(mf)of
ANFIS
target model have su
cces
sfully approa
che
d
the
target valu
e
of pair input
data. The
re
sults of th
e
a
c
curacy
of ea
ch mf
differ d
ue to the
out
put
model is ve
ry influential on
the mf value. The be
tter m
f
value the better model re
sulted. In brie
f,
the result of testing o
n
the pairs data in the input ANFI
S appro
a
che
s
target value
.
Statistics
cal
c
ulation tool
u
s
ed
to test th
e
a
c
c
u
ra
cy
of
A
N
FI
S
a
r
e t
he a
c
cur
a
cy
t
e
st
ing
on the input d
a
ta and RMS
E
(Root Mea
n
Square Error). Th
e equa
tions are:
100%
Whe
r
e
co
rrect data
= ANFIS trainin
g
dat
a , tota
l d
a
ta
= n
u
mbe
r
of i
nput d
a
ta. Th
erefo
r
e,
data accu
ra
cy obtained for triangula
r
are 77 % , Gaussi
an 53 % a
nd trape
zoi
d
a
l
61 %.
∑
Whe
r
e yoi
= target t
r
aini
ng
data f
r
om
pe
riod i
until
n ,
ypi = targ
et
model
of ANFIS data
from pe
riod
i until n. Thus, by the
s
e equatio
ns,
RMSE valu
es
are o
b
tai
ned for tri
a
n
gular
0.0246
0, gaussian 0.02
0
05 and tra
p
e
zoi
dal 0.
02
862. The re
sults of ANFIS model have
approa
che
d
the trainin
g
error tole
ran
c
e
value
determi
ned previou
s
l
y
by 0.01 with 100 epo
ch
s.
(a)
(b)
(c
)
Figure 4. Plot of Training
Result Datamf (a
) T
r
ian
gula
r
, (b) Gau
s
sia
n
, and (c) Tra
pezoid
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Financial Feasibility Study of
Waste
Cooking Oil
Utilization for Biodi
esel
… (Im
a
m Ahm
ad)
553
4. Conclusi
on
This
study ha
s re
ached
so
me co
ncl
u
si
o
n
s. T
he
re
sul
t
of feasibility study an
alysis u
s
ing
Anfis sh
owe
d
that the biodiesel in
du
stry assume
d for
about 10 yea
r
old proje
c
t is feasi
b
le to be
develop
ed o
n
sen
s
itivity level of incre
a
sin
g
pr
i
c
e 5
7
% (crips
0,7) and
de
cre
a
sin
g
pri
c
e 2
4
%
(crip
s
0,9) which su
cse
ssf
ully
appr
oa
ch
the target 1.
Fuzzy is a
b
l
e
to pre
d
ict p
o
ssibility risk
in
investment a
s
fea
s
ibility value ba
sed
on re
asonin
g
. Model fea
s
ibility study
Anfis ha
s b
e
e
n
develop
ed by
pair d
a
ta in
put; therefo
r
e
,
the resu
lts
of accuracy
MF triang
ular are 7
7
%, MF
gau
ssi
an 53
%, and MF trape
zoid 6
1
%.
In addition,
sugge
stion for
further research is that it
is better to cal
c
ulate feasibilit
y stud
y
for more than 10 years to obtain
more
data. Therefore, it will
increase the accuracy. Moreover,
future resea
r
ch, ab
out fu
zzy
rep
r
e
s
en
tation lo
w, m
oderate, and
high, can
b
e
used G
e
n
e
tic
Algoritm to obtain optimal
para
m
eter val
ue.
Ackn
o
w
l
e
dg
ements
I woul
d li
ke t
o
express
my sp
eci
a
l tha
n
k
s
of g
r
atitud
e supe
rviso
r
commi
ssion
who
gave
me golde
n g
u
idan
ce until
I am able to compl
e
te this re
se
arch,
as well a
s
the Dire
cto
r
a
t
e
Gene
ral of Education in
High
er Edu
c
ation (D
IKTI) and Com
put
ational Intelligen
ce Group
for
Advance
d
Rese
arch a
nd
Tech
nolo
g
y Innovation
s
of
Supercom
p
u
t
ing (CIGA
R
I
S
), i.e SMART-
TIN
©
Proje
c
t
w
hi
ch re
sp
ect
i
vely have co
ntributed fun
d
s
to study an
d sug
g
e
s
tion
s for me.
Referen
ces
[1]
Badan Pusat Statistik.
Statistik Daera
h
Kab
u
paten Bo
gor T
ahu
n 20
13
. No
.3201.1
1
.02. 2
013.
[2]
Sastratenay
a
A.S, Sudi, A.
Nucle
a
r en
erg
y
devel
op
me
nt in Indon
esia
.
In: Proceedi
ng
s of the IAEA
T
C
W
o
rkshop Lon
g Ra
nge Pl
ann
ing, Vi
enn
a
,
2010.
[3]
Machmu
d S.
Peng
aru
h
perb
and
ing
an so
lar
-
bio
d
ies
e
l (min
y
ak j
e
la
ntah) t
e
rha
dap em
isi
gas bua
n
g
pad
a motor di
e
s
el.
JANATEKNIKA
. 2009; 11
(2):107-
116.
[4]
Kamah
a
ra
H, Hasa
nud
in
U, W
i
di
yant
o A. T
a
chi
ban
a R, At
suta Y, Goto
N, Daim
on H,
and F
u
ji
e K
.
Improveme
n
t potentia
l for net
energ
y
bal
anc
e of bio
d
ie
s
e
l
deriv
ed from p
a
lm oil: A case
stud
y
fro
m
Indon
esi
a
n
practice.
Bio
m
ass
and
Bioe
nergy.
2
0
1
0
;
3
4
(
20
10
):1
8
1
8
-
1
824.
doi:1
0.10
16/j.bi
om
bi
oe.20
10.0
7
.014.
[5]
Aziz I, Nur
baiti
A, Ul
um B. P
e
mbu
a
tan
pro
duk
bio
d
i
e
sel
dari M
i
n
y
a
k
G
o
ren
g
Bek
a
s
d
eng
an
Car
a
Esterifikasi dan
T
r
ansesterifikasi.
Jurna
l
Val
ensi.
20
11; 2(3
)
:443-4
48.
[6]
Z
hang Y, D
u
d
e
MA, McLean
DD, Kates M. Biod
iese
l
pro
d
u
ction from
w
a
ste c
ooki
ng o
il:
2. Econom
ic
assessme
n
t an
d sensitiv
it
y
a
n
a
l
y
sis. 20
03; 9
0
(2
0
03):2
29-2
40 DOI:10.10
1
6
/S096
0-8
524(
03)0
015
0-0.
[7]
F
u
jita H, Okuh
ara K,
Nak
ano
K,
T
s
ubaki H.
Environm
ental Analisys
of
Waste Cook
ing Oil Recycling
and C
o
mpl
e
te
Use Practice in Bog
o
r, Indon
esi
a
.
Internati
ona
l Co
nferenc
e on
T
e
chno
logy,
Informatics, Mana
ge
me
nt, Engin
eeri
ng & En
viron
m
e
n
t
. Ban
dun
g. 201
3; 28
-31.
[8]
Kunca
h
y
o P, F
a
thall
ah AZ
M, Semin. Analis
a
pred
iksi
potensi b
a
h
a
n
baku b
i
o
d
ie
sel seb
aga
i
supl
emen b
a
h
an bak
ar motor
diese
l
di Ind
o
n
e
sia.
Jurnal Teknik Pom
i
ts.
20
13; 2(1): 62-6
6
.
[9
]
Ami
n
ah
S, Is
w
o
ro
T
J
. Pra
k
te
k Pe
ng
go
re
nga
n
D
a
n Mu
tu
Mi
ny
a
k
Go
re
ng
Si
sa
Pa
d
a
Ru
ma
h T
a
n
gga
RT
V/III Kedungmu
ndu T
e
mbal
ang Sem
a
r
ang.
Prosi
d
in
g
semi
nar nas
i
ona
l UNIMUS
2010
. 20
10;
ISBN: 978.97
9.704.8
83.9
[10]
Z
hang H, Oztu
rk AU, W
ang Q, Z
hao Z
.
Bio
d
ies
e
l pr
o
duc
e
d
b
y
w
a
ste co
o
k
ing o
il Rev
i
e
w
of rec
y
cl
in
g
modes i
n
Chi
n
a, the US and
Japa
n. 201
4; 38(20
14): 67
7-6
85.
[11]
Amalia
F
,
Ret
nan
ings
ih, J
o
h
an IR.
20
10.
Pe
rilak
u
pe
ng
gun
aa
n mi
n
y
a
k
gor
en
g serta
pe
ng
aruh
n
y
a
terhad
ap ke
iku
t
sertaan pro
g
r
a
m pen
gum
pul
an min
y
a
k
jel
a
ntah di kot
a
Bogor.
Jurn
al Il
mu Ke
lu
arg
a
Dan Ko
nsu
m
e
n
3(2):18
4
-1
89
.
[12]
Widod
o PSP. Anal
isis kel
a
y
a
kan usa
ha p
e
n
gol
aha
n min
y
a
k
jela
ntah me
n
j
adi
bio
d
ies
e
l.
Bogor (ID).
Institut Pertani
an Bog
o
r. 201
1.
[13]
Martini S,
Mo
del
inv
e
stasi f
u
zz
y u
n
tuk a
nalis
is k
e
la
ya
k
an fi
nans
ia
l u
s
aha
div
e
rsifik
asi i
n
d
u
str
i
berbasis tebu.
Jurna
l
Sosia
l
Ekono
mi Pertani
an da
n Agri
bis
n
is
. 201
0; 10(2
)
:134-1
40.
[14]
Shamsh
irba
nd
S, Petkovic
D,
Coj
basic
Z, Niko
lic V, A
n
uar NB, S
h
u
i
b
NLM, Ki
ah M
L
M, Akib S
.
Adaptiv
e n
eur
o-fuzz
y o
p
timi
zation
of
w
i
n
d
farm
proj
e
c
t net pr
ofit.
En
e
r
g
y
Co
nve
r
si
on
and
Mana
ge
me
nt.
2014; 80(2014)
:229-2
37. DOI: 10.1016/j.enc
onman.2014.01.03.
[15]
Loa
n, Vior
ica.
F
i
nanc
ing
Inve
sment Pro
j
ect t
he R
e
l
a
tions
hi
p Bet
w
e
e
n
F
e
a
s
ibil
it
y Stud
y
a
nd B
u
sin
e
s
s
Plan. Ann
a
ls
of Dunar
ea d
e
Jos Universiti
. F
a
scicle i:
Econo
mic and
Appl
ies Infor
m
atics
. 2010;
1(20
10): 12
5-1
30.
[16] Shee
n
NJ.
Fu
z
z
y
Econ
om
i
c
De
ci
si
on
m
o
de
l
s
fo
r In
fo
rm
a
t
io
n Security Investm
en
t. Proce
e
d
in
gs of the
9th WSEAS Int.
Confere
n
ce
o
n
Instrument
ati
on, Meas
ur
em
ent, Circu
its an
d S
y
stems. T
a
iw
a
n
. 2
0
0
7
;
ISBN: 978-96
0
-
474-
174-
8. 14
1-14
7.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 13, No. 3, March 2
015 : 546 – 5
5
4
554
[17]
Nasuti
on H, J
a
lam
a
lu
ddi
n H,
S
y
er
iff MJ. 2011. E
ner
g
y
An
al
ysis for Air
Con
d
itio
nin
g
S
y
stem Us
in
g
F
u
zz
y
Log
ic
C
ontrol
l
er.
T
E
LK
OMNIKA Indon
esia
n Jo
urna
l
of Electrica
l
En
gin
eeri
n
g
. 2
0
1
1
; 9(1): 1
39-
150.
[18]
Chiu CY a
nd
Park CS. Fuzz
y
cash fl
o
w
ana
l
y
sis usi
n
g
present
w
o
rth
criterion.
T
h
e
enge
neer
in
g
econ
o
m
ist.
199
4;
39(2): 11
3-1
37.
[19]
Kahra
h
man C,
T
o
lga E, Uluka
n
Z
.
Justification of manufactu
ring te
ch
nol
og
i
e
s usin
g fuzz
y
ben
efit/cost
ratio an
al
ysis.
Internati
o
n
a
l J. Producti
on Eco
n
o
m
ics
. 20
00; 66. 45-5
2
.
[20]
Jang JS
R. AN
F
I
S: Adaptive
Net
w
ork B
a
se
d F
u
zz
y Infere
nce S
y
stem.
I
EEE Transaction on System,
Man, and Cy
be
rnetics
. 199
3; 23(3): 66
5-6
8
5
.
[21]
Kaufman
n
A,
Gupta MM.
Int
r
oducti
on t
o
fu
zz
y
arith
m
etic
:
theor
y a
nd a
pplic
atio
ns.
N
e
w
York:
V
a
n
Nostran
d
Rei
n
hol
d. 198
8;
Evaluation Warning : The document was created with Spire.PDF for Python.