TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 14, No. 2, May 2015, pp. 280 ~ 28
7
DOI: 10.115
9
1
/telkomni
ka.
v
14i2.759
7
280
Re
cei
v
ed Fe
brua
ry 24, 20
15; Re
vised
Ap
ril 18, 201
5; Acce
pted
April 29, 201
5
Power Optimization of Spark Ignition Engine by Fuzzy
Logic Ignition Controller Based on Knock Detection
Agus Sujon
o
*
1
, R. Soekrisno
2
, Eka Firmans
y
ah
3
, O
y
as
Wah
y
u
nggoro
3
1
Electrical En
gi
neer
ing D
e
p
a
rtment, Sebel
as
Maret Univ
ersit
y
, Surak
a
rta, Indon
esia
2
Mechan
ical E
ngi
neer
in
g Dep
a
rtment, Gadja
h
Mada U
n
iver
sit
y
, Yog
y
a
k
art
a
, Indon
esia
3
Electrical En
gi
neer
ing D
e
p
a
rtment, Gadjah
Mada U
n
ivers
i
ty, Yog
y
a
k
arta, Indon
esi
a
*Corres
p
onding author, e-mail: agus.sujon
o@ft.uns.ac.id, soekris
no67@
gmail.c
o
m,
eka.firmans
ya
h
@
ugm.ac.i
d
, oyas
@u
gm.ac.i
d
A
b
st
r
a
ct
Spark
ig
nitio
n
eng
ines have ma
ny
adv
anta
ges,
but
to inc
r
ease the
pow
er and effici
en
cy, they
have a pr
ob
le
m to set the rig
h
t igniti
on ti
mi
n
g
, at w
h
ic
h the max
i
mu
m p
o
w
e
r can be
ac
hi
eved. In rea
lity, the
opti
m
u
m
i
gniti
o
n
timi
ng is i
mmedi
ately pri
o
r to, or clos
e to knock (deto
nati
on). On
the oth
e
r han
d, the va
lu
e
of this o
p
ti
mu
m i
g
n
i
tion
timi
ng is
de
pe
nde
nt pri
m
ar
ily o
n
the rotati
on
of the cra
n
k sh
a
ft and the
lev
e
l of
throttle openings. To provide the right
timing, it is very difficult if
only us
ing mechanic
a
l control system
as
found i
n
conve
n
tion
al en
gi
nes
. So, in this stu
d
y, a
new
elec
tro-mec
h
a
n
ica
l
metho
d
bas
ed
on F
u
zz
y
Lo
gi
c
Ignitio
n
Co
ntrol
l
er (F
LIC) w
h
ich follow
s
the p
a
ttern of
the ti
mi
ng i
n
the kn
ock char
t was created. In fact the
results of the F
L
IC study, was abl
e to pro
v
ide corr
ec
tion
for each ign
i
ti
on, accord
in
g to the data o
n
th
e
chart of th
e o
p
timal
ign
i
tio
n
ti
mi
ngs for
al
l c
o
mb
in
ations
of e
ngi
ne r
o
tatio
n
s
an
d throttle
o
p
eni
ngs. F
r
o
m
t
h
e
final d
a
ta reco
rded, it can be
said that the F
L
IC
able to p
u
sh the pow
er
up to 15% ab
ove nor
ma
l, whil
e
eli
m
i
nati
ng the
knock.
Ke
y
w
ords
: kn
ock, deton
ation
,
spark advanc
e,
fu
zz
y
lo
gic c
ontrol, en
gi
ne
pow
er
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
The po
we
r from the en
gin
e
dep
end
s o
n
the air to
fu
el ratio, the
q
uality of the fuel mix,
the turbul
en
ce level of the flow, com
b
u
s
tion timi
ng, the temperature in
the co
mb
ustion
cha
m
b
e
r,
the
engi
ne cooling and
l
u
brication sy
st
ems, a
nd
so
on. [1]. Ho
we
ver, the big
g
e
st
challe
nge
is
how to get th
e optimal igni
tion timing in orde
r to
get best perfo
rm
, but it is always fa
ced t
he
probl
em of kn
ock, whi
c
h is
a seri
ou
s pro
b
lem until tod
a
y.
Control of th
e optimal ig
n
i
tion timing (spa
rk
advan
ce) i
s
very co
mplicate
d
be
cau
s
e
of
somethi
ng de
pend
s on the
current op
erating co
ndi
tio
n
s, namely: e
ngine
spe
ed, throttle openi
ng
(load
), e
ngin
e
tempe
r
atu
r
e,
fuel type, a
n
d
so o
n
. So,
at any time,
whe
n
the
op
e
r
ating
co
nditi
ons
cha
nge, the spa
r
k a
d
van
c
e must also cha
nge.
In a conventio
nal
engine, co
ntrol is do
ne b
y
mech
ani
cal
control syste
m
s, namely by the
governo
r system
and vacuu
m
system in
the
manifold [2].
At a certai
n throttle po
sitio
n
, the engin
e
w
ill
produ
ce maximum
po
wer, whe
n
th
e
sp
ark
advan
ce in th
e optimal po
sition, or wh
at is
kn
own as
Maximum Brake T
o
rq
ue (MBT) timing,
or
prod
uce the
maximum IM
EP (Indicated
Mean Effe
ct
ive Pressu
re
). Whe
n
the
spark a
d
van
c
e
is
less or
greater, it
will result
in a decrease of power [3]. Fu
rthermore, when
t
he engine
speed
cha
nge
s, the
MBT timing
also
ch
ang
es [4]. Basi
call
y, the greate
r
spark
adva
n
ce, the
grea
ter
power, but if
it is too large, t
here will be knock, whi
c
h will result
in loss of power, wasted fuel,
exce
ss e
ngin
e
heat and t
he engi
ne m
a
y quickly da
maged [4]. T
herefo
r
e the
optimum ignit
i
on
timing is th
e i
gnition timing
whe
n
the
so
und of th
e kn
ock is sta
r
ting
to be h
e
a
r
d
and the
n
p
u
shed
back 2 CA
D (cra
nk a
ngle d
egre
e
s). This moment is
the critical time margin to the MBT and thi
s
is called
the knock-limited
maximu
m
brake
torq
ue
(KL-MBT) [5]. Thi
s
me
an
s
that at the
sa
me
time one ca
n avoid the kn
o
ck h
app
en
s.
A conventio
n
a
l me
chani
cal
cont
rol
syste
m
for
the sp
a
r
k advan
ce control wa
s
no
t
able
to
provide
the
M
B
T. For a
co
n
v
entional
eng
ine, which
i
s
still wi
dely u
s
ed, if the
ignit
i
on
system
was
repla
c
e
d
it would be ve
ry
co
stly and d
i
fficult. So
in this stu
d
y, a new m
e
thod
whi
c
h is
mo
re
approp
riate i
s
propo
se
d. It prov
ides t
he spark a
d
v
ance
co
rrec
tion for the
existing
syst
em
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
QCA and
CM
OS Nanote
c
h
nolog
y Based
Desi
gn an
d De
velo
pm
ent… (S. Deven
d
ra K. Verm
a)
281
electroni
cally by
Fuzzy Log
ic
Ignition Control (FLIC), a
sy
stem
that will
control t
he ignition ti
me
to be optimal (KL-MBT
)
.
The optimum
ignition time is a unique
cha
r
a
c
teri
stic of a gasolin
e engine, whi
c
h takes
the form
of a
3D
data m
a
p whi
c
h
is th
e location of
the ignition ti
ming (sp
a
rk
advan
ce) on
the
vertical
axis : high
revoluti
ons (engin
e
speed
),
a
nd th
rottle o
penin
g
(lo
a
d) on
the
two
ho
rizont
al
axis , such as sho
w
n in Fig
u
re 1.
Figure 1. Cha
r
acte
ri
stics of the 3D
diag
ram of the spa
r
k adva
n
ce [2]
In fact the
ch
ara
c
teri
stics
of an e
ngin
e
in
the 3
D
ch
a
r
t of spark
ad
vance
are no
t linear
and uniq
ue. Therefore, thi
s
FLIC control mech
ani
sm
starts fro
m
reality to
cre
a
te the d
e
sired
control and analysi
s
of the system. The existing co
ndition is used as an i
n
itial
data to facilit
ate
the an
alysis.
Although
th
e dia
g
ram
is com
p
lex,
n
o
t linea
r a
n
d
the
system
analysi
s
i
s
v
e
ry
difficult, but the fuzzy logic can overco
m
e
of it.
The b
a
si
c
co
figuration
of t
he fu
zzy
syst
em [6]
is to
i
n
crea
se th
e f
u
zzyfier of th
e input,
the fuzzy inf
e
ren
c
e
and t
he defu
zzyfie
r
of the
outp
u
t, where the real valu
e
of the fuzzif
i
er
transfo
rm
s th
e input vari
ab
les into fu
zzy
sets,
while
th
e defu
zzyfier
transfo
r
m
s
th
e fuzzy set to
a
r
eal value for the output var
i
able. Fuzzy infer
e
nc
e
is us
ed to pr
ocess
the input data to pr
oduc
e
the right de
ci
sion
s ba
se
d on the out
put
of the basi
s
o
f
the fuzzy se
tting (fuzzy ru
le base) give
n.
In the appli
c
ation of the contro
l, the fuzzy logic sy
stem was pr
esented in Figure 2, which
sho
w
s t
h
e
a
r
chit
e
c
t
u
re
of
t
he f
u
zzy
co
nt
rolle
r
that
co
nne
cts to
geth
e
r th
e in
put, the p
r
o
c
e
s
s a
n
d
the output.
Figure 2. Fuzzy cont
rolle
r architectu
re [
7
]
The u
s
e
s
of the FL
C (F
uzzy Logi
c Con
t
rol) meth
od
whi
c
h al
rea
d
y
exist are: for setting
the hydro
gen
injection fo
r
diesel motor
[8];
Method for optimu
m
valve openi
ng
has al
so u
s
ed
FLC for fuel
pump
moto
r
at a
co
nsta
nt sp
eed
[9]; F
uel th
rottle
settings. al
ong
with
the i
gnit
i
on
angle. are very effectively u
s
ed to control
the engine to
rque [10].
Other related
studies in
clu
de: A study of k
nocking o
n
the fuel mix
of gasolin
e with 10%
ethanol
havin
g the
be
st
ch
ara
c
teri
stics [
11]; Usin
g
DMF mixture
a
nd
spa
r
k timi
ng o
p
timization,
increa
sed
efficien
cy, better exhau
st e
m
issi
on
s
[12]
. Optimizatio
n
of spa
r
k ti
ming an
d EGR
(Exhau
st Ga
s Re
circul
ation), increa
se
effici
ency a
nd red
u
ce knock [13]; The new dyn
a
mic
cal
c
ulatio
n method to cal
c
ulate the thresh
old of
the kno
ck inten
s
ity, which is a complex and
difficult thing
wh
en
usi
n
g
the n
o
rm
al
way [14]; An
alysis of m
u
l
t
idimensi
onal
sim
u
lation
s
to
redu
ce
kno
c
k is by timing, EGR, the
ratio of
the mixture and
the shap
e of the combu
s
tion
cham
be
r [15]; The com
put
er mod
e
ls
wa
s used to p
r
e
d
ict the kno
c
k limited op
erating co
nditio
n
s
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 14, No. 2, May 2015 : 280 – 287
282
[16]; Knock
d
e
tection m
e
th
od ba
sed
on
non-i
n
tru
s
iv
e thermal sign
a
l
transi
ents o
n
the
cylind
e
r
wall was m
a
de [17]. Knock ca
n be d
e
tected u
s
in
g
an optical im
aging te
chni
q
ue ba
sed o
n
th
e
natural
emission from the f
l
ame and
UV
spe
c
tro
s
cop
y
. Radical
cla
ssifi
cation
s such a
s
O
H
a
nd
HCO a
r
e d
e
tected
and
co
rrel
a
ted
with t
he on
set
and
duration of t
he kno
c
k an
d
the p
r
esen
ce of
hot-spots by
the end-ga
s [18]. A study of
knocking was d
o
ne usin
g LES (Larg
e
Eddy
Simulation) couple
d
with d
e
tailed
ch
emi
c
al
kin
e
tics [1
9]. A misfire (kno
ck) ha
s
a
uniqu
e p
a
ttern
of vibration si
gnal
s asso
cia
t
ed with a pa
rticular
cylinde
r whi
c
h can b
e
extracte
d a
nd analy
z
ed t
o
detect it by usin
g a stati
s
tical a
pproa
ch an
d i
denti
f
y it by using
deci
s
ion tre
e
algorith
m
[20].
Whe
n
there i
s
a
kno
c
k, combu
s
tion th
at occurs
wil
l
cau
s
e
a ra
pid ri
se a
nd
fluctuate mo
re
stron
g
ly, so t
hat the
signal
can
be
pro
c
essed
with a
wavelet p
a
cket tran
sform
a
tion to dete
c
t
the
knock [21].
Optimization of spark timing w
ill give the maximum
output power, high thermal
efficien
cy, an
d re
stri
ct the
occurre
n
ce of
kno
c
k
[22].
Setting the spark timing
a
nd the a
dditio
n
of
CO a
nd hyd
r
ogen
ga
s will
red
u
ce the o
c
curren
ce
of kno
c
k [23]. Variation
s
in
th
e co
mpressio
n
ratio d
one
by
varying
the
spark ig
nition t
i
ming g
a
ve th
ese
re
sult
s: t
he hi
ghe
r
co
mpre
ssion
rat
i
o,
the highe
r th
e indicated th
ermal effici
en
cy, whic
h lea
d
s to high
er
bra
k
e torque
and lo
wer
break
spe
c
ific fuel
con
s
um
ption
[24]. The calcul
at
ed re
sults demo
n
st
rated
that the kno
c
k onset
defined by p
r
essu
re o
s
cillations
cha
r
a
c
teri
stics he
re wa
s app
ro
ximately con
s
iste
nt with that
defined
by
heat
rele
ase. The
kno
c
k inten
s
ity,
chara
c
te
rized
by kn
ock m
e
trics, Maxi
mum
Amplitude of Pressu
re O
s
cillations
(MA
P
O)
and
KI
2
0
a
s
th
e m
e
asu
r
e
of the
severity of
kn
ock
[25].
So the pu
rp
o
s
e of thi
s
stu
d
y is to
crea
te a ne
w m
e
thod to b
o
o
s
t engin
e
p
o
wer a
nd
efficien
cy, by avoidin
g
the
kno
c
k o
c
cu
rs, by me
a
n
s o
f
optimization
of
the
spark advan
ce ba
sed
on the
FLIC
system that
store
s
d
a
ta in
a 3D char
act
e
risti
c
s chart
whe
n
the i
gni
tion timing of
the
engin
e
is
opti
m
al, in orde
r to ma
ke a
co
rrection to
th
e
conve
n
tional i
gnition, while
eliminating th
e
kno
c
k.
2. Rese
arch
Metho
d
This
study was p
e
rfo
r
med
on a
station
a
ry engi
ne m
ounted
on a
n
ETB (Engi
ne Te
st
Bed), in
a la
borato
r
y, u
s
i
ng a
150
0
cc, 4-cylind
e
r,
4
stro
ke
en
gine
with a
carbu
r
ettor
a
n
d
conve
n
tional
ignition sy
stem. Figure
3 sho
w
e
d
th
e installatio
n
of the engi
ne, whe
r
e it
is
equip
ped wit
h
a cooli
ng system,
tempe
r
ature
read
er
,
dwell mete
r,
flow mete
r, p
r
essure g
aug
e
s
,
vibration sensors,
digital stor
age oscilloscope, interface and
timing light (stroboscope). The
measured dat
a will be re
co
rded o
n
the
compute
r
thro
ugh data a
c
q
u
isition.
(a)
(b)
Figure 3. Engine Test Bed
(ETB) Sy
ste
m
: (a) s
c
h
e
m
a
tic (b
) pictu
r
e
The e
ngine
chara
c
te
risti
c
s take th
e form of a data
cha
r
t sho
w
in
g the optim
u
m
ignition
timing (Kl-MB
T) whi
c
h i
s
d
one by ru
nnin
g
the engi
n
e
with varie
d
lo
ads a
nd revs,
until detonati
o
n
(kn
o
ck)
starts to occu
r. Th
en, at this po
int,
data on
e
ngine
sp
eed,
throttle o
pen
ings and
spa
r
k
advan
ce i
s
measured
as the m
a
ximu
m ignitio
n
ti
ming
(MBT)
with the
ai
d
of a timin
g
li
ght
(strobo
scop
e). The rev
s
are varied f
r
om
1,000 rpm
to
3,000 rpm; the throttle i
s
varied from 1
0
%
to 70%. The optimum ignit
i
on timing (K
L-MBT
)
is
the ignition timing at the sta
r
t of detonation
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
QCA and
CM
OS Nanote
c
h
nolog
y Based
Desi
gn an
d De
velo
pm
ent… (S. Deven
d
ra K. Verm
a)
283
(kn
o
ck) de
creased by 2
CAD, whilst
avoiding t
he con
d
ition
s
where a
large detonatio
n
(knock)
occ
u
rs
.
A data ch
art
of the optimal
ignition timin
g
(KL-MBT can be the
r
efo
r
e p
r
epa
re
d u
s
ing the
following s
t
eps
.
The firs
t s
t
ep in this
study was
(1)
to set the tim
e
of combu
s
t
i
on a
s
the i
n
i
t
ial
con
d
ition; the
n
(2
)
open
th
e throttle; a
n
d
(3
)
sele
ct a
parti
cula
r e
n
g
ine
spe
ed
a
nd regul
ate t
h
e
engin
e
loa
d
; then
(4) g
r
adu
ally cha
ngin
g
the igniti
o
n
ti
ming u
n
til the
detonatio
n
(knock
)
starts to
be audi
ble; then, wh
en d
e
tonation i
s
d
e
tected, (5
) t
he firing time
shoul
d be
re
duced sli
ghtl
y
(2
CAD) to get the KL-MBT value; then (6) re
co
rd the
ignition timing (spark ad
vance
)
, throttle
openi
ng an
d
engine
sp
e
ed. Next, re
peat this p
r
o
c
ed
ure:
cha
n
ge the throttl
e openi
ng, then
repe
at step
s
3, 4, 5, and 6.
When all
condition
s are
record
ed,
a 3D chart
can
be plotted. The
cha
r
a
c
teri
stics of th
e e
n
g
i
ne a
r
e
arra
nged
in
2
condition
s, n
a
m
ely: spa
r
k
advan
ce
setti
ngs
norm
a
l (conv
entional
) or
NBT (Norm
a
l
Brake To
rqu
e
) and o
p
timum spa
r
k ad
vance
setting
s or
KL-MBT
(Ma
x
imum Bra
k
e
Limited-K
n
o
ck
Toq
ue). T
he differen
c
e
betwe
en th
e
two
setting
s
(KL-
MBT vs. NBT) is the
correction of the
spa
r
k
advan
ce to make th
e engin
e
pro
duce maximu
m
torque
or ma
ximum power at a certai
n
engine
spe
e
d
and throttle openi
ng, a
s
sh
own in the
Figure 4.
Figure 4. Con
v
entional an
d
FLIC system
combin
ation
in kno
c
k co
rrection
The control system will u
s
e a 3D
ch
art, as a
re
feren
c
e or
corre
c
tio
n
s to the
con
v
entional
system, see
Figure
4. Wh
enever
t
he d
r
iver of
the car cha
nge
th
e
throttle op
ening
for a
f
e
w
possibl
e reasons, such as
to speed up
or slow do
wn, the FLIC will set the ignition timing (spark
advan
ce) a
ccordin
g to the 3-D
cha
r
t. Co
rre
ction
s
are
made to cha
nge the conv
entional ig
nition
timing in orde
r to get the o
p
timum ignition timi
ng that gives the engine its be
st perfo
rman
ce.
The
FLIC rea
d
s
throttle openi
n
g
an
d spe
ed as
i
nput,
a
nd it
therefo
r
e calcul
ates
the Fuzzy
Inferen
c
e
to provide opt
imum ignition
timing values.
Figure 5. FLIC sp
ark adva
n
ce cor
r
e
c
t
i
o
n
sy
st
em
Figure 5 sho
w
s
a sche
ma
tic diag
ram o
f
the installati
on of a fuzzy
logic
(FLIC)
ignition
control
syste
m
for
co
rrecti
on of th
e ignit
i
on timing.
Th
e micro
co
ntroller
doe
s
not
dire
ctly a
c
tivate
the ignitio
n
coil, but d
o
e
s
so th
ro
ugh
el
ectro
n
ic
d
r
ive
r
s a
s
the
eq
u
i
pment i
n
terf
ace.
Thi
s
circuit
will ke
ep the
curre
n
t micro
c
ontrolle
r sm
all enou
gh
so that it will be preserved
and not q
u
ickly
bro
k
en. Th
e Cha
nge Ove
r
Switch (CO
S
) is the tr
an
sfer
swit
ch from the conv
entional igniti
on
s
y
s
t
em to the FLIC s
y
s
t
em, in order to
shift the system used at an
y given time.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 14, No. 2, May 2015 : 280 – 287
284
3. Results a
nd Analy
s
is
The cha
r
a
c
te
ristics of the
intensity of
detonatio
n (knock)
gen
era
t
ed usin
g th
e eddy
curre
n
t dyna
mometer
on the ETB, which mea
s
ures t
h
e torq
ue o
r
power versu
s
spa
r
k a
d
van
c
e
according
to
the en
gine
speed
variatio
ns, a
s
s
hown
in Fi
gure
6,
sho
w
e
d
that:
the
power a
t
a
given spee
d
of rotation
(rpm) al
way
s
i
n
crea
se
s
with an i
n
crea
se in
spa
r
k a
d
vance, but i
f
the
ignition timin
g
is advan
ce
d again, det
onation (kn
o
c
k) will ari
s
e
and gro
w
from small to ver
y
large.
Figure 6. Engine ch
ara
c
te
ri
stic
in kno
ck
(detonatio
n) vs. power an
d spa
r
k a
d
van
c
e variation in
the engin
e
rot
a
tion
Furthe
rmo
r
e,
data from Fi
gure
6 can b
e
used
to mak
e
: the
c
h
arac
teris
t
ics
of a normal
spa
r
k a
d
van
c
e whi
c
h is a
convention
a
l spark adv
an
ce
with the normal default settings (NBT) as
sho
w
n
in Fi
g
u
re
7(a). T
h
e
optimum
spa
r
k advan
ce
(KL-MBT) is the MB
T
sp
ark a
d
van
c
e
(when
the en
gine
re
ach
e
s its
ma
ximum po
we
r) min
u
s 2
C
A
D
pr
es
e
n
t
ed
in
F
i
gu
r
e
7(
b
)
. F
i
g
u
r
e
7(c
)
is
the calculatio
n of the
co
rre
c
t
ed
sp
ark a
d
v
ance
whi
c
h
i
s
the
NBT
sp
ark adva
nce
redu
ce
d by t
h
e
KL-MBT spark advan
ce.
(a)
Normal sp
ark a
d
van
c
e (NBT)
convent
ional
(
b
)
Op
timu
m
s
p
ar
k
ad
va
nce
(
K
l-
MBT)
(c) The
corre
c
tion of sp
ar
k advance (KL
-
MBT to NBT
)
Figure 7. Spark adva
n
ce chara
c
te
risti
c
: (a), (b), (c)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
QCA and
CM
OS Nanote
c
h
nolog
y Based
Desi
gn an
d De
velo
pm
ent… (S. Deven
d
ra K. Verm
a)
285
Figure 7
(
a
)
shows th
e
ch
ara
c
teri
stics
of the e
ngin
e
at n
o
rm
al (co
n
vention
a
l
)
ignitio
n
setting
s
(NBT
), indi
cating
t
hat the i
n
cre
a
se
in
sp
ark
advan
c
e
occurs
when
the
r
e i
s
an i
n
cre
a
se
in the revs at
a certai
n throttle openin
g
, but if
the throttle openin
g
is increa
sied
, then the sp
ark
advan
ce d
e
cl
ines. T
h
is i
n
d
i
cate
s that th
e engi
ne
ha
s a gove
r
no
r
and a
vacuu
m
to control
the
size of th
r
spark a
d
van
c
e. In a
conv
entional
i
gniti
on
system, t
he g
o
vern
or
is a
me
cha
n
i
c
al
sy
st
em,
whi
c
h ma
ke
s t
he
incr
ea
se in
s
par
k a
d
v
ance
if the revs a
r
e in
crea
sed.
In the vacuu
m
system, th
e p
r
essu
re
is m
easure
d
at
th
e inta
ke
mani
fold, whi
c
h
wi
ll ma
ke th
e d
e
clin
e of
sp
ark
advan
ce in case of a de
crease in vacu
um or throttle being op
ene
d
wide
r.
The KL-MBT
characte
ri
stic in Figu
re 7
(
b) i
s
an opti
mal spa
r
k ad
vance at whi
c
h the
engin
e
ca
n p
r
odu
ce m
a
ximum po
wer.
Corre
c
tion of
the spa
r
k a
d
v
ance i
s
a bi
g differen
c
e f
r
om
the value of
NBT an
d the
value of KL-MBT. This i
s
a co
rrectio
n
o
f
a conv
en
tion
a
l
sp
a
r
k
ad
va
n
c
e
(NBT
) to KL-MBT spa
r
k a
d
vance so th
at the
engine
can op
erate
in a conditio
n
of maximum
power. Th
e
operating
co
ndition
s of the KL-MBT
spa
r
k
advan
c
e me
an
s al
so avoi
ding
the
occurre
n
ce of the conditio
n
s
wh
ere
the
r
e
is large d
e
to
nation (kno
ck).
The
sp
ark
ad
vance
corre
c
tion, in
Figu
re
7(c),
takes t
he fo
rm
of da
ta that
can
b
e
turned
into a cha
r
t. It is this co
rre
c
tion value d
a
ta cha
r
t that is pro
c
e
s
sed
by the FLC to give a de
ci
sion
on the value of the optim
um
spark advance,
and
it will
be the
basi
s fo
r the FLC
setting,
as
sho
w
n in Ta
b
l
e 1.
Table 1. Rul
e
table of the corr
ectio
n
of the spa
r
k a
d
va
nce (
δ
- de
gree)
Thro
ttle ope
nin
g
(
θ
-
%
)
VS S
M
B
VB
Engine
speed (n
- rpm
)
VB
N N N N
N
B
N
N VS N
N
M
N VS
M
S
VS
S
N S
B
M
S
VS
M B
VB
B
M
VB = very big, B = big, M = medium, S =
small, VS = very small, N
= non (zero)
The Fu
zzy
membe
r
ship
function
of e
a
ch i
nput a
n
d
output va
ri
able i
s
st
ru
ctured
a
s
follows
: The f
u
zz
y members
h
ip func
tion
of the ope
n throttle carburettor (
θ
-%) is: VS =
30%, S =
40%, M = 50%, B = 60%,
VB = 70 %. F
u
zzy membe
r
ship fun
c
tion
s for engin
e
rotation (n - rp
m)
is: VS = 10
00rp
m
, 150
0
r
pm
= S, M = 20
00
rpm,
B = 25
00rpm, VB = 3
0
00rpm. Fu
zzy
membe
r
ship
functio
n
s for spark
advan
ce
(a
ngle
)
i
s
: V
C
=
0
o
, VS =
4
0
, S =
8
o
, M
= 12
0
, B =
16
o
,
VB =
20
o
, Calcul
ation
s
a
nd de
ci
sion
s made
usin
g
Fuzzy Lo
gi
c Control
are ba
sed
on
the
membe
r
ship functio
n
s a
nd
fuzzy rules ta
ble, as sho
w
n
in Figure 8.
Figure 8. The
combin
ation
of membershi
p
func
tion
with table rules
of the FLIC
The result of the co
nventi
onal ignitio
n
sy
stem com
b
ined with
FL
C
for spa
r
k advan
c
e
correction, from now on
will be referred to as
a F
L
IC combination syst
em, and will
certainly
provide a n
e
w
level of eng
ine perfo
rma
n
ce. Th
e re
su
lts are a
s
follows:
δ
=
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 14, No. 2, May 2015 : 280 – 287
286
(a)
(b)
(c
)
(d)
(e)
Figure 9. The
compa
r
i
s
on
of the performance of FL
IC vs
. Conventional: with the throttle open:
(a) 2
0
%
,
(b) 3
0
%
,
(c) 4
0
%
,
(d) 5
0
%
,
(e) 6
0
%
In Figure 9
(a)-(e
), with a
throttle ope
n
i
ng of
20% t
o
60%, is a
picture of the
engin
e
perfo
rman
ce
in conventio
n
a
l conditio
n
s
with n
o
rm
al
spark
advan
ce
setting
s a
s
compa
r
ed
to t
h
e
engin
e
whe
n
equippe
d wi
th the FLIC ignition syste
m
. It appears that the FLIC system gi
ves
about
1
5
%
g
r
eater po
wer esp
e
ci
ally
at high spe
ed,
whi
c
h m
ean
s that an
engi
ne with
FLI
C
has
better pe
rformance.
Whe
n
the e
ngine
con
d
ition ch
ang
es,
it means t
hat the FLIC, req
u
ire
manual
adju
s
tments.
Howeve
r, this cont
rolle
r is only suitab
le for the sa
me or simila
r engin
e
s. F
o
r
different en
gine, this cont
rolle
r sh
ould
be adju
s
te
d to fit
the c
h
arac
teris
t
ics
of the propos
ed
engin
e
.
4. Conclusio
n
From thi
s
study it can
b
e
co
ncl
ude
d
that
the co
m
b
ination
of a
conve
n
tional
ignition
system coupl
ed with the
spark advan
ce
correctio
n
, also kn
ow
n as
the FLIC syst
em, which is a
combi
nation
of
mechani
ca
l
and
ele
c
tro
n
ic co
ntrol systems, ca
n provide
the optimum
i
gnit
i
on
timing (KL-M
B
T) at any time and
con
d
i
tion, thus
in
crea
sing
engin
e
perfo
rma
n
ce by about 1
5
%
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
QCA and
CM
OS Nanote
c
h
nolog
y Based
Desi
gn an
d De
velo
pm
ent… (S. Deven
d
ra K. Verm
a)
287
and at the
same time it
can
avoid th
e kn
ock
(det
onation
)
, Ho
pefully future
studie
s
can
be
upgrade
d to a sma
r
ter
system.
Referen
ces
[1]
Soel
aima
n T
A
F
.
Ne
w
Strate
g
y
for Det
e
ctin
g Knock
in S
p
ark Igniti
on E
n
gin
e
.
PhD Dis
ertation.
T
he
Univers
i
t
y
of Mines
ota; 199
2.
[2] Bosch.
Automo
tive Electric.
El
ectronic Syste
m
. German
y
. 1
993.
[3]
Saras
w
at
i S, Agar
w
a
l PK, Ch
and S. N
eur
al
net
w
o
r
ks
an
d fuzz
y
lo
gic-b
a
s
ed sp
ark a
d
va
nce co
ntrol
o
f
SI engin
e
s.
Expert Systems
w
i
th Applicati
o
ns
. 2011; 3
8
: 6916-
692
5.
[4]
He
yw
o
o
d
, J
o
h
n
B. Inter
nal
Com
bustio
n
Engi
ne F
und
a
m
ental.
Ne
w
York: Mc-Graw
Hi
ll
Bo
o
k
Comp
an
y. 19
8
8
.
[5]
Dani
el
R, T
i
an G, Xu S, Sh
uai S. Ig
nitio
n
timi
ng s
ensiti
v
ities of o
x
yg
e
nated bi
ofuels
compar
ed t
o
gaso
lin
e in a d
i
rect-injecti
on S
I
engin
e
.
F
uel 99
. 201
2: 72-8
2
.
[6] W
ang
L
X
.
A Course i
n
F
u
zz
y
S
y
stem a
nd C
ontro
l. Lo
nd
on:
Prentice-H
a
ll I
n
t. Inc. 1997.
[7]
Passin
o
KM, Yurkovic
h S. F
u
zz
y
Co
ntrol. C
a
lifor
nia, USA: Addis
on-W
e
sl
e
y
L
ongm
an Inc
.
1998.
[8]
Bose PK, De
b
M, Banerj
ee R,
Majumd
er A. Multi
ob
jectiv
e optimiz
ation of
performa
nce p
a
rameters
of
a sing
le c
y
li
nd
er dies
el e
ngi
n
e
runn
in
g
w
i
th
h
y
dro
gen usin
g
a
T
aguchi
-fu
z
z
y
b
a
sed
app
roach.
Ener
g
y
63
. 201
3: 375-
386.
[9]
Vachtsevanos G,
Farinw
ata SS,
Pirovolo
u
DK. F
u
zzy
L
ogic Co
ntrol o
f
Automotive Engi
ne.
IEEE
Contro
l System
. 19
93.
[10]
Mosk
w
a
JJ. A
u
tomotive E
n
g
i
ne M
ode
lli
ng
for Re
a
l
T
i
me Contro
l. PhD
Dissertati
on.
USA: MIT
.
Massach
usetts; 1988.
[11]
Rotham
er DA, Jenni
ngs JH.
St
ud
y
of the
knockin
g
pro
pens
it
y
of 2,5-
dimeth
ylf
u
ran
–
gaso
lin
e an
d
ethan
ol
–gas
oli
ne bl
en
ds.
F
uel 98
. 201
2: 203
-212.
[12]
Dani
el R, T
i
an G, Xu H, W
y
s
z
y
nsk
i
ML, W
ua X,
Hu
ang Z
.
Effect of spark timing an
d lo
a
d
on a DISI
eng
ine fu
ell
ed
w
i
t
h
2,5-d
i
meth
ylfur
an.
F
uel 9
0
. 2011: 44
9-4
58.
[13]
Z
hen
X, W
ang
Y, Xu S, Z
hu Y. Numerical
ana
l
y
sis on k
n
ock for a hig
h
compressi
on
ratio spark-
igniti
on meth
an
ol en
gin
e
.
F
uel
103.
20
13: 89
2-89
8.
[14]
Gallo
ni E. D
y
namic kn
ock det
ectio
n
an
d qua
ntificati
on i
n
a s
park ig
ni
tion en
gi
ne b
y
means of
a
pressur
e
base
d
method.
En
e
r
gy Conv
ersio
n
and Man
a
g
e
m
ent
. 201
2; 64: 256-
262.
[15]
Z
hen
X, W
a
ng
Y,
Xu S, Z
h
u
Y.
Stud
y of k
n
ock i
n
a
hi
gh
compressi
on
r
a
tio s
park-i
gnit
i
on
metha
n
o
l
eng
ine
b
y
m
u
lti
-
dimens
io
nal si
mulati
on.
Ener
gy 50.
20
13: 1
50-1
59.
[16]
So
ylu S. Pr
ed
i
c
tion of k
nock
l
i
mited
op
eratin
g con
d
iti
ons
of a n
a
tural
g
a
s
eng
ine.
E
ner
gy
Conv
ersio
n
and Ma
na
ge
ment
. 200
5; 46: 121-
138.
[17]
Ollivier E, B
e
ll
ettre J,
T
a
zero
ut M, Ro
y
GC.
Detectio
n of k
nock oc
c
u
rren
c
e in
a gas
SI eng
ine from
a
heat transfer a
nal
ysis.
En
ergy
Convers
i
on a
n
d
Mana
ge
ment
. 2006; 47: 87
9
-
893.
[18] Merola
SS, Va
gliec
o
BM. Kn
ock inv
e
stigati
on b
y
fl
ame
an
d rad
i
cal s
pec
i
e
s detecti
on
in
spark i
gniti
o
n
eng
ine for d
i
fferent fuels.
Ener
gy Conv
ersio
n
and Ma
na
ge
ment
. 200
7; 48: 289
7-29
10.
[19]
Z
hen
X, W
a
n
g
Y, Z
hu Y. Stu
d
y
of kn
ock in
a hi
gh c
o
mpre
ssion r
a
tio SI
methan
ol
eng
i
ne us
in
g LES
w
i
t
h
deta
ile
d chemic
al kin
e
tic
s
.
Energy Con
v
ersio
n
and M
ana
ge
me
nt
. 20
13; 75: 52
3-53
1.
[20]
Sharma
SA, S
ugum
aran
V,
Devas
ena
pati
SB. Misfire
det
ection
in
a
n
IC
en
gin
e
usin
g vibrati
on
s
i
g
nal
and d
e
cisi
on tr
ee al
gorithms.
Measur
e
m
ent
. 201
4; 50: 370-
380.
[21]
Hou J, Qi
ao
X, W
ang Z, Li
u
W
,
Huan
g Z. Char
acteriz
a
ti
on
of knock
i
ng
combusti
on
in
HCCI D
M
E
eng
ine
usin
g
w
a
vel
e
t packet transform.
Appl
i
ed Ener
gy
. 201
0; 87: 123
9-12
46.
[22]
Li
H, Karim
GA. Knock
in
spar
k ig
nitio
n
h
y
dr
oge
n
eng
in
es.
Internatio
na
l J
ourn
a
l
of Hy
dr
oge
n E
nergy
.
200
4; 29: 859-
865.
[23]
Bika AS, F
r
an
klin
L, Kittelso
n
DB. En
gin
e
kno
ck a
nd c
o
mbustio
n
ch
ar
acteristics of
a
spark i
g
n
i
tio
n
eng
ine
op
erati
ng
w
i
t
h
var
y
i
n
g h
y
dr
og
en
an
d carb
on m
o
n
o
xid
e
pr
oporti
o
n
s.
Internati
o
n
a
l j
ourn
a
l
o
f
hydro
gen ener
gy
. 2011; 3
6
: 5143-
515
2.
[24]
Ma F
,
Li S, Z
h
ao J, Qi Z
,
D
eg J, Naeve N, H
e
Y, Z
hao S. Effect
of compression rati
o an
d
spark timin
g
on the
po
w
e
r p
e
rformanc
e an
d comb
usti
on c
haracter
i
stics o
f
an HCNG
en
gin
e
.
Internati
o
nal j
our
nal
o
f
hydro
gen ener
gy
. 2012; 3
7
: 1848
6-18
49
1.
[25]
Shu G, P
an J,
W
e
i H.
Ana
l
ysis of
ons
et a
nd s
e
ve
rit
y
of knock in
SI
e
ngi
ne bas
ed
o
n
i
n
-c
yli
nde
r
pressur
e
oscil
l
a
tions.
Ap
pl
ie
d Th
e
r
m
a
l
En
gi
ne
e
r
in
g
. 20
13; 5
1
: 1297-
13
06.
Evaluation Warning : The document was created with Spire.PDF for Python.