Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
6, N
o
. 3
,
Ju
n
e
201
6, p
p
. 1
281
~ 12
93
I
S
SN
: 208
8-8
7
0
8
,
D
O
I
:
10.115
91
/ij
ece.v6
i
3.9
597
1
281
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Mitigating Electricity a Pric
e Sp
ike under P
r
e-Cooling Method
Marw
an M
a
r
w
an
1
, Pirm
an
2
1
Ele
c
tri
cal
Eng
i
n
eering
Depar
t
m
e
nt
2)
Chemical Engineer
ing
Dep
a
rtment
Poly
technic State of Ujung
Pand
ang Makassar
In
donesia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Dec 24, 2015
Rev
i
sed
Feb
26
, 20
16
Accepted
Mar 10, 2016
The growing demand for air-con
ditioning
is one of the larg
est co
ntributors to
Australia over
a
l
l
elec
tric
it
y cons
um
ption. This has started to cr
ea
te peak lo
ad
supply
problems for some elect
ricity
utilities par
t
icular
ly
in Queensland. This
research
aimed to develop a con
s
umer
demand side response model to assist
ele
c
tri
c
it
y
consum
ers to m
itigate
peak dem
a
nd on
the el
ect
ric
a
l ne
twork. Th
e
proposed model
allows consumer
s to i
ndep
e
nden
tly
and p
r
oactively
manage
air
conditioning
peak
electricity
d
e
ma
nd.
The main con
t
ribution of th
is
research
is how t
o
show
consum
ers can m
itig
at
e p
eak d
e
m
a
nds b
y
optim
izing
energ
y
costs for
air
conditioning
in a sever
a
l
cases such as no
spik
e and
spik
e
considering
to
th
e probab
ility
spike cases may
only
o
ccur
in
the middle of
the
day
for h
a
lf hou
r spikes. This model al
so inv
e
stigates how air conditioning
applied a pre-co
oling method when there is
a sub
s
tantial risk of a price spike.
The res
u
lts
indi
cat
e the poten
ti
al of the s
c
hem
e
to achiev
e
en
erg
y
s
a
vings
and reducing
ele
c
tri
c
it
y bi
lls (co
s
ts) to
the cons
um
er. The m
odel
was
tes
t
ed
with the
Queens
l
and e
l
e
c
tri
c
it
y
m
a
rket da
ta fro
m
Australian En
erg
y
Marke
t
Operator and B
r
isbane temperature da
ta from Bureau statistic during hot
day
s
.
Keyword:
Co
o
ling
Eco
nom
i
c
s
Electrical Power
M
odel
l
i
n
g
Netw
or
k
Copyright ©
201
6 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Mar
w
an
Marwan
,
Pol
y
t
echni
c
St
at
e of
U
j
un
g
P
a
nda
n
g
,
Peri
nt
i
s
Kem
e
rdekaa
n St
reet
, M
a
kassar
,
S
o
u
t
h
S
u
l
a
wesi
,
I
n
do
nesi
a.
Em
a
il: marwan_
e
n
e
rg
y@yahoo
.co
m
; marwan
@po
liu
pg
.ac.id
1.
INTRODUCTION
The Sm
art
Gri
d
i
s
a pr
om
i
s
ing c
o
ncept
t
o
cope
wi
t
h
i
n
c
r
easi
ng e
n
er
gy
dem
a
nd an
d e
nvi
ro
nm
ent
a
l
co
n
c
ern
s
.
As t
h
e m
a
in
featu
r
e and
co
n
s
t
r
u
c
tio
n
g
o
a
l
of
the sm
art g
r
id
, i
n
tellig
en
t in
teractio
n
in
cl
u
d
es two-
way in
teraction
o
f
inform
atio
n
and
en
erg
y
, to
en
co
urag
e electricity consumers to
change
the traditional usa
g
e
st
y
l
es and
pa
rt
i
c
i
p
at
e i
n
t
h
e
n
e
t
w
o
r
k
o
p
erat
i
o
n
act
i
v
el
y
(s
u
c
h as
ad
j
u
st
m
e
nt
o
f
e
n
er
gy
c
ons
um
pt
i
on
pa
t
t
e
rns
according t
o
re
al-tim
e price),
and to ac
hieve
the pl
ug-a
nd-play Grid-conne
c
tion
of t
h
e dis
t
ribute
d
gene
ra
tion.
Th
us,
dem
a
nd si
de m
a
nagem
e
nt
(DSM
) t
echn
o
l
o
gy
i
s
o
n
e o
f
t
h
e m
o
st
im
port
a
nt
pa
rt
s of t
h
e sm
art
gri
d
.
B
a
sed on
t
h
e t
r
adi
t
i
onal
fu
n
c
t
i
ons,
t
h
e
sm
art
g
r
i
d
DSM
has new
co
nt
e
n
t
s
,
i
n
cl
udi
n
g
aut
o
m
a
t
i
on
de
m
a
nd
resp
o
n
se, sm
art
cons
um
e seque
nce,
rem
o
te ener
gy
effi
c
i
ency
m
onitor & control,
energy efficient
powe
r
gene
rat
i
o
n, an
d so o
n
[
1
]
.
De
m
a
nd si
de res
p
ons
e (D
SR
) i
a
a part
h
of t
h
e s
m
art
gri
d
sy
st
em
can be defi
n
e
d as
th
e ch
ang
e
s in electr
i
c u
s
ag
e b
y
en
d-
u
s
e custo
m
er
s f
r
o
m
t
h
eir
no
r
m
al co
n
s
u
m
p
tio
n
p
a
tter
n
s in
r
e
sponse to
changes
in t
h
e
price
of electricity over tim
e
[2].
To i
m
pl
em
ent DSR
m
odel
,
cons
um
er i
s
req
u
i
r
e
d
t
o
en
rol
as a
m
e
m
b
er of a gr
o
up c
ont
r
o
l
l
e
d by
an
agg
r
e
g
at
or
. T
o
be ex
p
o
sed t
o
el
ect
ri
ci
t
y
m
a
rket
pri
ce an
d
net
w
or
k
ove
rl
oad c
o
st
s, sm
al
l
-
cons
um
ers n
eed an
aggre
g
ator to
comm
unicate an
d n
e
go
tiate th
e electricity mark
et pri
ce a
n
d
net
w
o
r
k o
v
e
rl
oa
d.
A
n
y
ch
ange
i
n
the electricity
usa
g
e for the
sm
a
ll-cons
um
e
r
is base
d
on
th
e in
form
at
io
n fro
m
th
e aggregator. As a result,
aggre
g
ators
ke
ep a
n
d m
a
intain c
o
mm
unication
bet
w
een m
a
rket
operator a
n
d cons
um
ers.
In
t
h
e co
m
p
etitiv
e electricit
y
m
a
rk
et stru
ctu
r
e, t
h
e aggreg
ator con
c
ep
t d
e
scri
b
e
s an
in
d
e
p
e
n
d
e
n
t
ag
en
t
prov
id
i
n
g
its sm
all-co
n
s
u
m
ers
with
a wi
d
e
rang
e
o
f
inn
o
v
a
tiv
e
serv
ices in
cludin
g
b
ill m
a
n
a
g
e
m
e
n
t
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E
V
o
l
.
6,
No
. 3,
J
u
ne 2
0
1
6
:
12
8
1
– 12
93
1
282
hom
e
m
a
nagem
e
nt
, hom
e elect
ri
ci
t
y
gener
a
t
i
on, a
nd
ot
he
r ser
v
i
ces [
3
,
4
]
. B
a
sed o
n
t
h
e
s
e servi
ce
pr
o
v
i
si
ons
,
the aggregat
or com
b
ines its
consum
er
s in
t
o
a
sing
le purch
asing
un
it to
n
e
go
tiate
th
e pu
rch
a
se of
electricit
y
fr
om
t
h
e ret
a
i
l
er [
3
]
.
The a
g
gre
g
at
o
r
al
so
n
e
got
i
a
t
e
s dem
a
nd
res
p
o
n
se a
n
d be
hal
f
of t
h
e
cons
um
er wi
t
h
t
h
e
retailer, d
i
stri
bu
tio
n and
tran
smissio
n
co
m
p
an
y. Man
y
ec
ono
m
i
sts b
e
liev
e
th
at th
e p
a
rticip
atio
n of
ag
gr
eg
a
t
or
with
inn
o
v
a
tive serv
ices an
d
co
nsu
m
er
ag
greg
atio
ns can
offer po
ten
tial so
lu
tion
s
fo
r small-scale co
n
s
u
m
ers
t
o
effect
i
v
el
y
m
a
nage t
h
ei
r c
ons
um
pt
i
on, a
nd t
h
e
r
e
b
y
b
e
co
m
i
n
g
activ
e p
a
rticip
an
ts in
th
e electricity
mark
et
[3-6
]. Th
e agg
r
eg
ation
m
e
t
h
od
s, t
h
e co
mb
in
ing
m
u
ltip
l
e
electricity
l
o
ad, prov
id
es th
e b
e
n
e
fits of retai
l
electric com
p
etition for the
consum
er with lo
wer electric us
age called a small cons
um
er.
In this
pape
r,
the Quee
nsland electricity market
price i
s
chose
n
for t
h
e case
studie
s. Collective
bene
fits are ty
pically the pri
m
ary consi
d
eration for th
e s
m
all consum
er and aggregat
or so these
pric
es are
use
d
t
o
dem
onst
r
at
e t
h
e
m
i
nim
i
sat
i
on proce
d
u
r
e. T
h
e w
hol
esale electricity
m
a
rket
pri
ces are pu
bl
i
s
he
d
on t
h
e
Aust
ral
i
a
n E
n
e
r
gy
M
a
rket
O
p
erat
or
(
A
EM
O
)
websi
t
e
.
Det
a
i
l
e
d i
n
f
o
rm
ation
ab
o
u
t
A
E
M
O
pri
ce
dat
a
c
a
n
be
f
oun
d in
[
7
].
Seasonal climate variation
has a
si
gni
fi
ca
n
t
im
pact
on t
h
e ope
rat
i
o
n of electrical power system
s.
Du
e to
th
e temp
erat
u
r
e rises in
su
mm
er, th
e
electricity
d
e
man
d
will in
crease with
th
e lo
ad
o
f
air cond
itio
n
i
ng
o
r
o
t
h
e
r app
lian
ces. M
o
reov
er, if th
e con
s
umers all tu
rn
on
th
e air con
d
i
tio
n
i
ng
at th
e
sam
e
ti
me, th
en
th
e
to
tal d
e
m
a
n
d
will b
e
i
n
creased
.
Tem
p
erature is an
im
p
o
r
t
a
n
t
driv
er
for
electricity co
n
s
u
m
p
tio
n
.
M
o
re th
an
40
% of e
n
d
-
u
s
e ener
gy
con
s
um
pt
i
on i
s
rel
a
t
e
d t
o
t
h
e heat
i
ng an
d co
ol
i
ng
need
s i
n
t
h
e resi
dent
i
a
l
and
commercial sectors
[8].
A load surv
ey study underta
k
en
by the Quee
nsla
nd
Gove
rnm
e
nt indicated that
each
ki
l
o
wat
t
o
f
ai
r
-
co
n
d
i
t
i
oni
n
g
i
n
st
al
l
e
d i
n
Qu
eensl
an
d c
o
st
s
up t
o
$
3
0
0
0
i
n
ne
w e
n
er
gy
i
n
fra
st
ruct
ure t
o
m
e
et
p
eak d
e
m
a
n
d
[9
].
Th
erefore,
air-con
d
ition
i
n
g
u
s
ag
e
c
o
n
t
ribu
tes
g
r
eatly to
p
e
ak
l
o
ad growth in
both
th
e
com
m
e
rci
a
l
and
resi
de
nt
i
a
l
sect
ors i
n
Quee
n
s
l
a
nd
[
10]
.
A price sp
ik
e
can
b
e
g
e
n
e
rally d
e
fin
e
d
as an
abno
rm
al p
r
ice v
a
lu
e, wh
ich
is sign
ifican
tly d
i
fferent
fr
om
it
s expec
t
ed val
u
e
[1
1,
12]
. T
h
e p
r
i
c
e
spi
k
e i
n
t
h
e e
l
ectricity
market is an
abnormal
market clearing
p
r
ice at a ti
m
e
p
o
i
n
t
t an
d
is sig
n
i
f
i
can
tly d
i
f
f
e
r
e
n
t
f
r
o
m
th
e
av
er
ag
e pr
ice. I
n
a sp
i
k
e, th
e p
r
ice cou
l
d
r
i
se 100
or
10
0
0
t
i
m
e
s
hi
g
h
er t
h
a
n
t
h
e no
rm
al
pri
ce, whi
c
h b
r
i
n
g
s
a h
i
g
h
risk
for th
e m
a
rk
et p
a
rticip
an
ts [13]. Th
is
im
pact
i
s
not
j
u
st
on
t
h
e c
o
ns
um
er
but als
o
on the electrici
ty retailer.
2.
R
E
SEARC
H M
ETHOD
2.1.
Numeric
a
l Optimisation
Nu
m
e
rical
m
o
d
e
llin
g
is a feasib
le so
lu
tion
t
o
allo
w
for unp
red
i
ctab
le m
a
rk
et price ch
ang
e
s du
e to
t
h
e
i
n
t
e
r
r
u
p
t
i
o
n of
m
a
jo
r g
e
nerat
i
o
n or
ot
he
r
s
u
p
p
l
y
-s
id
e con
s
trai
n
t
s. To
co
ndu
ct th
is inv
e
stigatio
n
,
math
e
m
atica
l
m
o
d
e
ls fo
r the co
n
s
u
m
er p
a
rticip
an
t
we
r
e
devel
o
pe
d t
o
q
u
ant
i
f
y the
econom
i
c
effect of
dem
a
nd-si
de
v
a
ri
at
i
on.
A l
i
n
ear pr
o
g
ram
m
i
ng
-base
d
al
g
o
r
i
t
h
m
was devel
o
ped t
o
de
t
e
rm
i
n
e t
h
e opt
im
al
sol
u
t
i
o
n t
o
ac
h
i
eve t
h
e best
o
u
t
c
om
es. In a
d
di
t
i
on, t
h
e de
v
e
l
ope
d m
odel
was de
si
g
n
ed t
o
be a
p
pl
i
cabl
e
fo
r
l
o
ad dem
a
nd
con
s
t
r
ai
nt
s t
o
gi
ve g
o
od eco
nom
i
c
perfo
r
m
ance for el
e
c
t
r
i
c
i
t
y
generat
i
on, t
r
an
sm
i
s
si
on an
d
d
i
str
i
bu
tio
n.
The m
odel
sh
ows
h
o
w ai
r
-
c
o
n
d
i
t
i
oni
ng s
h
oul
d dec
r
ease
t
e
m
p
erat
ure l
o
ad
s i
n
hi
g
h
t
e
m
p
erat
ur
e
peri
ods
w
h
e
n
t
h
ere
i
s
a s
u
bst
a
nt
i
a
l
ri
sk
o
f
a
p
r
i
ce s
p
i
k
e,
t
h
at
i
s
, by
a
p
pl
y
i
ng
a
pre
-
c
ool
i
n
g
m
e
t
hod t
o
avoi
d
high prices in
a critical peak peri
od. Cons
umers are able to operate th
e ai
r-co
n
d
i
t
i
oni
ng
usage by
co
nt
rol
l
i
n
g
t
h
e desi
re
d l
e
vel
s
o
f
r
oom
t
e
m
p
erat
ure
,
t
u
r
n
i
n
g o
n
t
h
e
ai
r-co
n
d
i
t
i
oni
ng
w
h
en t
h
e t
e
m
p
erat
ur
e ri
s
e
s t
o
a
m
a
xim
u
m
t
h
re
shol
d (i
.e.
,
25
o
C) th
en
tu
rn
ing
it o
ff for the n
e
x
t
p
e
ri
o
d
u
n
til th
e te
m
p
eratu
r
e
d
r
o
p
s to
th
e
m
i
nim
u
m
t
h
reshol
d (i
.e., 1
9
o
C). In
add
itio
n, th
is research
in
v
e
stig
ated
ho
w con
s
u
m
ers can
o
p
tim
ise
en
erg
y
co
sts
wh
en th
ey h
a
ve
n
o
t
commi
tted
to
th
e p
e
rm
i
tted
te
mp
erat
u
r
e.
On th
is
o
p
tim
isat
io
n
p
r
o
cess,
when
t
h
e
roo
m
te
m
p
erat
u
r
e is less or
m
o
re th
an
th
e
min
i
m
u
m
o
r
max
i
m
u
m
te
m
p
e
r
atu
r
e th
resho
l
d
th
en
a
p
e
n
a
lty to
th
e
o
p
tim
izat
io
n
pro
cess will b
e
id
en
tified
.
The cyclin
g
tim
e
o
f
th
e ai
r-co
n
d
itio
n
i
ng
is
b
a
sed
o
n
th
e
resu
lt of
te
m
p
eratu
r
e opti
m
isatio
n
.
In t
h
i
s
resea
r
ch
, a pre-c
o
ol
i
ng
m
e
t
hod wa
s ex
am
i
n
ed as a way
t
o
m
i
nim
i
se energy
co
st
s. Pre-c
o
ol
i
n
g
i
s
a
m
e
t
hod t
o
red
u
ce t
h
e
r
o
o
m
t
e
m
p
erat
ure
i
n
ad
vance
o
f
a po
ssi
bl
e s
p
i
k
e. Thi
s
m
e
t
hod
i
s
consi
d
ere
d
t
o
be
effective bec
a
u
se it can
m
i
nimise energy
costs and
ca
n kee
p
room te
m
p
eratures
com
f
ortable for the
consum
er. Howeve
r,
pre-c
o
oling is
only underta
ken
when
there is a s
u
bs
tantial risk
of
a price s
p
ike
becaus
e
it co
sts a lo
t a
n
d
th
e sp
ik
e
may n
o
t
alway
s
o
ccur on
th
e syste
m
. Ho
wev
e
r, wh
ile app
l
yin
g
th
is m
e
th
od
is
ex
p
e
n
s
i
v
e, it is m
o
re efficien
t
th
an
switch
i
n
g
o
n
th
e ai
r-co
nditio
n
i
n
g
at all times d
u
ring
th
e critical ti
m
e
.
The
objective
is to minim
i
se
en
er
gy
co
st
s by
o
p
t
i
m
i
si
ng ro
om
t
e
m
p
eratures
. T
h
e ene
r
gy cost i
s
base
d
on
t
h
e
ai
r-co
n
d
i
t
i
oni
n
g
st
at
u
s
, t
h
at
i
s
,
no
c
o
st
wh
en t
h
e ai
r
-
c
o
n
d
i
t
i
oni
n
g
st
at
u
s
i
s
of
f
(
U
0
) a
n
d
m
a
rket
cost
i
f
t
h
e ai
r-c
o
ndi
t
i
oni
ng
st
at
us i
s
on
(
U
1
). T
o
ac
hieve this
object
iv
e, an optimisation
packa
g
e
su
ch
as MATLAB allo
ws th
e
u
s
er t
o
carry ou
t op
timisa
tio
n
with
in
op
erati
o
n
a
l con
s
train
t
s su
ch
as a
p
e
rmit
ted
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Mitig
a
tin
g Electricity a
Price Sp
ike und
er Pre-Coo
ling
Meth
od
(Marwan
Ma
rwan
)
1
283
t
e
m
p
erat
ure
ra
nge
. I
n
t
h
e
opt
im
i
s
ati
on p
r
oc
ess, t
h
e M
A
TL
AB
o
p
t
i
m
i
s
at
ion t
ool
bo
x f
u
n
c
t
i
on
fmin
con
and
t
h
e
or
di
na
ry
di
ffe
r
e
nt
i
a
l
eq
uat
i
o
n
sol
v
er
O
D
E
4
5
we
re
used
.
In
o
r
de
r t
o
f
o
r
m
ul
at
e t
h
e part
i
c
i
p
at
i
on
of t
h
e co
nsum
er i
n
t
h
e D
S
R
p
r
o
g
r
am
, t
h
e ene
r
gy
cost
m
odel
wh
ich
represen
ts th
e chang
i
n
g
tem
p
erature and electric
ity price
was
de
vel
o
pe
d as
re
po
rt
ed
he
re
. T
h
e
opt
i
m
i
s
ati
on
pr
obl
em
can t
h
e
n
be
re
prese
n
t
e
d
as m
i
nim
i
sed
ener
gy
c
o
st
(Z
)
,
o
r
m
a
t
h
em
at
ical
l
y
[4,
5]
:
Zt
St
.
P
t.
Dt.
U
td
t
(1
)
Su
bject
t
o
c
o
ns
t
r
ai
nt
s [
1
4,
1
5
]
:
.
.
.
(2
)
whe
r
e:
Z
= M
i
ni
m
i
sed ener
gy
c
o
st
(
A
$
)
S
= Electricity price (A$/kWh)
P
=
Rating p
o
we
r of AC (k
W)
D
= Du
ration
time fo
r
op
erating
AC during
a
d
a
y (hou
rs)
U
= C
o
nt
i
n
u
o
u
s t
i
m
e
bi
nary
vari
abl
e
(
1
or
0
)
Q
= Heat tra
n
s
f
er coe
fficient
from
floor
walls a
n
d ceiling
(W
/
m
2
o
C)
B
= Heat tra
n
sm
ission from
the
AC (W)
A
= Total area (m
2
)
H
= Heat ca
pacity of the
room
(J/
o
C)
To
= Tem
p
erature
outside
(
o
C)
Tt
= Tem
p
eratu
r
e in
sid
e
th
e
room
a
t
ti
m
e
t (
o
C)
n
= in
terv
al time t (h
ou
r)
During
th
e
op
ti
m
i
zat
io
n
,
if t
h
e
r
o
o
m
t
e
mp
e
r
a
t
u
r
e
i
s
mo
r
e
o
r
l
e
s
s
th
a
n
t
h
e
ma
x
i
mu
m
o
r
mi
n
i
mu
m
te
m
p
erature
(
Tm
a
x
or
Tm
in
) t
h
resho
l
d
,
th
e
min
i
mizatio
n
will ad
d a
p
e
nalty to
th
e co
mp
u
t
ed
co
st.
If
T
t
orIfT
t
ThenP
e
nalty
K
(3)
ElseP
e
nalty
0
(4)
There
f
ore, the
energy c
o
st
wi
l
l
be cal
c
u
l
a
t
e
d
by
:
Zt
St
.
P
t.
Dt.
U
td
t
K
(5
)
2.
2.
D
a
t
a
Pro
cessi
ng
2.2.1
Price Spike in the Electricity
Marke
t
In t
h
e
prese
n
t
r
e
search
, aft
e
r a
n
al
y
s
i
s
of t
h
e h
i
st
ori
cal
dat
a
, a t
h
resh
ol
d
val
u
e of A
$
75
per
M
W
h
was
u
s
ed
for an
alysis of th
e
Qu
een
s
land
electri
city
m
a
rket
d
u
r
i
n
g week
day
peri
ods
.
T
h
i
s
m
eans
any
re
g
i
onal
refe
rence
p
r
i
ce m
o
re t
h
an
A
$
75
pe
r M
W
h i
s
cal
l
e
d a
pri
c
e
spike
.
T
h
e a
v
erage
of th
e el
ectricity prices unde
r
A$
7
5
pe
r M
W
h i
s
cal
l
e
d t
h
e
no
n
-
spi
k
e p
r
i
c
e, w
h
i
c
h i
n
t
h
i
s
peri
od
was
A
$
3
0
.
6
9 pe
r M
W
h
.
Fi
g
u
r
e
1 i
ndi
cat
es
th
e RRP o
f
th
e electricity
m
a
r
k
et in
Qu
eenslan
d
du
ring
h
o
t
d
a
ys in
20
11-2
012
[7
]. Th
e
d
a
ta p
r
esen
ted
in
th
e
table shows cle
a
rly that any
price above
t
h
e
red line is
a s
o
-called price s
p
ike.
Fig
u
r
e
1
.
Electr
i
city
m
a
r
k
et pr
ice in
Qu
een
sl
an
d dur
ing
h
o
t
d
a
ys in 201
1-
20
12
[7
]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E
V
o
l
.
6,
No
. 3,
J
u
ne 2
0
1
6
:
12
8
1
– 12
93
1
284
2.
2.
2
Ho
t
D
a
y
s
an
d O
u
tsi
d
e
T
empera
ture
I
n
t
h
is r
e
sear
ch
, an
y d
a
y
on
w
h
ich
th
e av
e
r
age
daily te
m
p
erature
was
more t
h
an 30
o
C
i
s
cal
l
e
d a
hot
day, as
gi
ven in Fi
gure
2. Figure
3 summ
arises an exam
ple of classic
fluctuations
in
outside tem
p
eratures
i
n
B
r
i
s
ba
ne o
n
2
9
Fe
br
uary
20
12
. T
h
e t
e
m
p
erat
ure i
s
t
y
pi
c
a
l
l
y
at
it
s l
o
west
l
e
vel
du
ri
n
g
t
h
e m
o
rni
n
g
and
a
t
night tim
e
. Norm
ally, the dai
l
y outside tem
p
erat
ure i
n
cr
ea
ses in t
h
e m
i
ddle of the day. In the e
x
am
ple, the
hi
g
h
est
t
e
m
p
erat
ure
occu
rre
d
fr
om
arou
nd
12:
00 t
o
1
7
:
00. The
outside t
e
m
p
eratur
e around the
s
e tim
e
s was
mo
r
e
t
h
a
n
3
0
o
C
.
T
h
e ma
x
i
m
u
m t
e
mp
e
r
a
t
u
r
e
o
f
3
5
o
C
occ
u
rre
d at
15:
00
.
The m
i
nim
u
m
t
e
m
p
erat
ure
o
f
1
8
o
C
occu
rre
d i
n
th
e
early
m
o
rnin
g
(0
5:0
0
t
o
05:
45). T
h
e a
v
era
g
e te
m
p
erature
was
25
o
C.
Fig
u
r
e
2
.
Classif
i
catio
n
o
f
ho
t
d
a
ys, 201
1 to
2
012
Fi
gu
re
3.
B
r
i
s
b
a
ne
out
si
de t
e
m
p
erat
ure (
T
o
)
o
n
29
Febr
u
a
r
y
201
2
3.
R
E
SU
LTS AN
D ANA
LY
SIS
There are
vari
ous
way
s
t
o
i
m
pl
em
ent
t
h
e
DSR
i
n
t
h
e us
e of ai
r-c
on
di
t
i
oni
ng
. The M
a
rk
o
v
bi
rt
h an
d
deat
h p
r
oc
ess has bee
n
de
vel
ope
d t
o
m
a
nage sm
all
packag
e ai
r-co
ndi
t
i
o
n
e
r l
o
ads ba
sed
on a q
u
eui
ng s
y
st
em
.
Th
is m
o
d
e
l en
ab
les resid
e
n
t
s with
sm
a
ll a
i
r-co
n
d
itio
n
e
r lo
ads to
p
a
rtici
p
ate in
v
a
ri
o
u
s lo
ad
m
a
n
a
g
e
men
t
program
s
whereby they can receive in
centives and lower t
h
eir electricity
bills while their conve
n
ienc
es are
t
a
ken i
n
t
o
acc
o
unt
[
16]
. T
h
i
s
m
odel
pro
v
i
d
e
s
effect
i
v
e an
d
con
v
e
n
i
e
nt
l
o
a
d
m
a
nagem
e
nt
m
easures t
o
b
o
t
h t
h
e
p
o
wer co
m
p
any an
d th
e con
s
u
m
er. In
cen
tives and
co
m
p
en
satio
n
are recog
n
i
sed
b
y
th
e
utili
ty co
m
p
an
y
b
a
sed
on t
h
e l
e
vel
of
part
i
c
i
p
at
i
o
n of t
h
e c
o
nsum
ers [
16]
. I
n
t
h
i
s
m
odel
,
t
h
e el
ect
ri
ci
t
y
pri
ce was n
o
t
base
d
on t
h
e
electricity
m
a
r
k
et price. T
h
erefore,
th
e aggreg
ator
w
a
s no
t
r
e
qu
ir
ed
to
co
nt
r
o
l
sm
al
l con
s
um
ers. On t
h
e
ot
her
hand, these m
odels a
r
e
not
approp
riate for anticipating
a price
spi
k
e
and seas
onal
clim
ate changes in
Au
st
ralia. As a resu
lt, th
ese mo
d
e
ls were
n
o
t
co
nsid
ered
a
s
a
p
r
e-c
ool
i
n
g m
e
t
h
o
d
t
o
av
oi
d
hi
g
h
c
o
st
s.
A sim
p
le co
n
t
ro
l strateg
y
is also
u
s
ed to m
a
n
a
g
e
th
e air-con
d
ition
i
ng
i
n
a DSR
program in
Kuwait.
Du
e to th
e no
rmal o
p
e
rati
o
n
o
f
air-con
d
ition
i
ng
i
n
Kuwai
t
on
a 24
ho
ur b
a
sis, th
e co
ntro
l
syste
m
p
r
o
v
i
d
e
s
co
m
f
o
r
tab
l
e con
d
ition
s
d
u
ring th
e o
c
cup
a
n
c
y
p
e
ri
o
d
on
ly. Fo
r ex
am
p
l
e, the syste
m
is ap
plied
for fi
v
e
p
e
riod
s
d
u
ring
a d
a
y: (i) 0
3
:
3
0
-04
:
0
0
, (ii) 1
2
:
0
0
-13
:
0
0
, (iii) 15
:15
-
1
6
:
0
0
, (iv) 18
:
0
0-1
8
:
3
0
, and
(v) 20
:00
-
21
:00
.
To
achieve
accept
a
ble c
o
m
f
ort c
o
nditions
duri
ng these
pe
riod
s, a
pre
-
cooling m
e
thod is
a
pplied [17].
T
h
e
pre
-
co
o
ling
m
e
th
od
is ap
p
lied b
y
ex
tend
ing
t
h
e
o
p
e
ration
tim
e
of t
h
e air-conditioning. T
h
e
pre-c
ooling m
e
thod is
not
base
d
on
t
h
e su
bst
a
nt
i
a
l
ri
sk
of
t
h
e
pri
ce
spi
k
e
.
T
h
ere
f
o
r
e, t
h
i
s
m
e
t
hod
i
s
not
e
ffe
ct
i
v
e
t
o
be a
ppl
i
e
d
on
t
h
e
syste
m
if a pric
e spi
k
e
happe
n
s.
Th
e pro
c
ess o
f
co
n
s
u
m
er to
min
i
mize en
erg
y
will
wh
ile k
eep
i
n
g
co
m
f
o
r
t with
in
sp
eci
ficatio
n
is to
co
n
t
ro
l each
switch
i
ng
in
stan
t of th
e
AC.
Wh
en
t
h
e to
ta
l
cost is m
i
nimized the
n
i
n
m
a
ny cases
whe
r
e pric
e
risk is low
it is appropriate to
m
a
ke no speci
al prepa
r
atio
n. Howev
e
r wh
en th
e p
r
ice risk
is h
i
g
h
th
e op
timal
swi
t
c
hi
n
g
pat
t
e
rn m
a
kes use of pre
-
co
ol
i
n
g.
At
t
h
e l
e
vel
of opt
i
m
i
zati
on w
i
t
h
const
r
ai
nt
s no
ot
her
out
c
o
m
e
is
feasi
b
l
e
.
Ot
he
r
m
e
t
hods
of c
o
st
re
duct
i
o
n
are exam
i
n
ed
abo
v
e
but
no
n
e
of t
h
e
s
e ha
d
t
h
e p
o
t
e
nt
i
a
l
fo
r t
h
e
sam
e
lev
e
l o
f
sav
i
ng
s
foun
d in
th
is research.
C
ons
um
ers sh
oul
d st
art
t
o
a
ppl
y
t
h
e
D
S
R
pr
o
g
ram
t
o
o
p
t
i
m
i
se t
h
e ai
r con
d
i
t
i
oni
ng
as
soo
n
a
s
t
h
ey
receive i
n
form
ation
from
the aggre
g
ator. Due t
o
the
pa
ttern
of high outside
tem
p
erature, t
h
e c
o
nsumer is
requ
ired
to
p
a
rticip
ate in
th
e DSR prog
ram
startin
g
from
10:
00 t
o
19:00. The e
n
ergy cost was calculated
whe
n
t
h
e ai
r c
o
n
d
i
t
i
oni
ng
w
a
s on
, an
d t
h
e
cost
was zer
o
whe
n
t
h
e ai
r con
d
i
t
i
oni
n
g
was o
f
f
.
Thi
s
m
e
t
hod
co
n
tinu
e
d
un
til th
e ti
m
e
o
f
o
p
eratin
g
the air co
nd
itio
n
i
ng
had
exp
i
red
.
To
m
a
k
e
th
e te
mp
erat
u
r
e co
m
f
o
r
tab
l
e
for the c
o
nsumer, the room te
m
p
eratur
e
was
onl
y
al
l
o
wed t
o
b
e
bet
w
een
1
9
o
C and
25
o
C. Th
is
mean
s th
e
t
e
m
p
erat
ure
w
a
s n
o
t
al
l
o
we
d
t
o
reac
h t
h
e
m
a
xi
m
u
m
and m
i
ni
m
u
m
perm
i
t
t
ed t
e
m
p
erat
ure
s
. F
o
r t
h
e
pu
r
p
ose
of
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Mitig
a
tin
g Electricity a
Price Sp
ike und
er Pre-Coo
ling
Meth
od
(Marwan
Ma
rwan
)
1
285
th
e sim
u
latio
n
,
th
e starting
poin
t
te
m
p
erature o
f
22
o
C
was c
hos
en
with t
h
e
air conditioning status
off. Ta
ble
1
su
mm
arises th
e p
a
ram
e
ters o
f
th
e typ
i
cal roo
m
an
d
th
e air
co
nd
itio
n
i
ng
used
i
n
th
is
op
timisatio
n
.
Tab
l
e
1
.
Param
e
ter of th
e Ex
am
p
l
e Ro
o
m
Used
in th
is
An
alysis
3.
1.
C
o
s
t
as a Functi
on o
f
a Pri
ce
Spi
ke W
i
thou
t DS
R
Pr
ogr
am
Th
e aim
o
f
th
e co
n
t
ro
ller is t
o
m
a
in
tain
th
e te
m
p
erature
of the room
betw
een s
o
m
e
lower and upper
te
m
p
eratu
r
es i
n
ord
e
r t
o
k
e
ep it with
in
co
m
f
o
r
tab
l
e li
m
i
ts. Fo
r th
is sim
u
la
tio
n
,
the starting
po
in
t
o
f
22
o
C was
chosen with the air c
o
nditioni
ng st
at
us of
f.
The
l
o
we
r
a
n
d
u
ppe
r
t
e
m
p
er
at
ures we
re 22
o
C to
2
4
o
C. T
h
e air
co
nd
itio
n
i
ng
was tu
r
n
ed
on
on
ce th
e
te
m
p
erature rose to the selected
m
a
xi
m
u
m
.
Next
,
t
h
e ai
r con
d
i
t
i
oni
n
g
was tu
rn
ed
off o
n
ce th
e temp
erat
ure dropped to the selected
m
i
nim
u
m
.
W
i
t
h
t
h
e ai
r con
d
i
t
i
oni
ng
of
f, t
h
e
te
m
p
eratu
r
e cou
l
d
i
n
crease and
rise to th
e selected
m
a
x
i
m
u
m
.
Th
e typ
i
cal o
p
e
ration
o
f
the air con
d
ition
i
n
g
i
s
co
n
tinuo
us w
i
t
h
ou
t co
n
t
ro
l
by th
e D
S
R
p
r
og
r
a
m
.
To
o
p
e
rate th
e air
cond
itio
n
i
ng
i
n
th
i
s
case, th
e consu
m
er
d
i
d
no
t co
n
s
i
d
er a
p
r
ice sp
i
k
e. Figure
5
illustrates th
e cycl
in
g
tem
p
erature and
t
h
e m
a
rket co
st if a sp
i
k
e ma
y
occu
r i
n
t
h
e
m
i
ddl
e
o
f
t
h
e
day
.
In
t
h
is sim
u
lati
o
n
, th
ere are
20
switch
edg
e
s to
com
pute the energy cost
f
o
r t
h
e ai
r c
o
n
d
i
t
i
oni
ng
. I
f
S
is th
e electricity p
r
ice
wh
en
a sp
i
k
e
o
c
cu
rs, K is th
e p
e
n
a
l
t
y, th
en th
e to
t
a
l m
a
rk
et co
st
for the sp
ik
e cas
e
(
MC
is d
e
term
in
ed
b
y
th
e fo
llowin
g
equ
a
tion
:
MC
,,
t
S
t
.
P
t.
Dt.
U
tdt
K
(6
)
Eq
uat
i
ons
(1
) t
o
(
6
)
were
use
d
t
o
com
put
e t
h
e res
u
l
t
s
of
si
m
u
l
a
t
i
on wi
t
h
o
u
t
DSR
pr
og
ra
m
when a
hal
f
ho
u
r
spi
k
e
m
a
y
occur
i
n
t
h
e m
i
ddl
e o
f
t
h
e
day
,
as
sh
ow
n i
n
Fi
g
u
r
e
4.
Cy
cling T
e
m
p
er
atur
e without DSR Pr
ogr
am
Half hour
spike case
Figure
4. Cycling tem
p
erature
and m
a
rket cost without
DSR
As sh
own
in
Fig
u
re 4, th
e calcu
latio
n
of th
e electricity co
st d
u
ring
th
is
period
was b
a
sed
o
n
t
h
e air
conditioni
ng status. T
h
e electricity co
st increased
whe
n
the te
m
p
erature
wa
s bei
n
g
re
du
ced
by
ha
vi
n
g
t
h
e
ai
r
co
nd
itio
n
i
ng
on
.
Howev
e
r, there was
no
electricity co
st
when t
h
e air c
o
nditioning
wa
s
off or electricity costs
10
11
12
13
14
15
16
17
18
19
19
20
21
22
23
24
25
26
T
e
m
p
era
t
ure (
o
C)
10
11
12
13
14
15
16
17
18
19
0
2
4
6
M
a
rk
e
t
C
o
s
t
(A
$
)
No Para
m
e
ters
Unit
Value
1
Heat tr
ansf
er coef
ficient f
r
o
m
f
l
oor w
a
ll and ceiling (Q
)
1
W/
m
2 o
C
2
Total are
a
(A)
54
m
2
3
Heat capacit
y
of
th
e roo
m
(H)
48
J/
o
C
4
Heat transfer
fr
o
m
the air
conditioning (
B
)
900
W
5
Ref
e
rence of
te
m
p
erature
22
o
C
6 Hyster
esis
3
o
C
7 M
a
xim
u
m
te
m
p
er
atur
e
25
o
C
8 M
i
nim
u
m
te
m
p
er
atur
e
19
o
C
9
Rating power
of air
conditioning (
P
)
2.
6
kW
10
Nu
m
b
er
of switch
change events
20
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E
V
o
l
.
6,
No
. 3,
J
u
ne 2
0
1
6
:
12
8
1
– 12
93
1
286
were
not
cal
cu
l
a
t
e
d whe
n
t
h
e
ai
r con
d
i
t
i
oni
ng
was o
f
f
.
The electricity c
o
st calculation started from
s
w
itch
num
ber
1 t
o
n
u
m
ber 2.
The
n
,
t
h
e ai
r
-
co
n
d
i
t
i
oni
ng
wa
s t
u
r
n
ed
of
f a
g
ai
n
be
t
w
een
swi
t
c
h
n
u
m
b
ers 2
t
o
3
,
whe
n
the electricity cost is zero. The type of
ope
ration
was co
n
tinuo
u
s
fo
r all switch
i
n
g
and
all ti
mes. Th
e
con
s
um
er/
a
gg
r
e
gat
o
r pay
s
t
h
e
cost
acc
or
di
n
g
t
o
t
h
e
no
rm
al
price
before
and afte
r a s
p
ike
happe
n
s.
The
price
spike
was only calculated when
t
h
e spi
k
e ha
ppe
ne
d i
n
t
h
e
m
i
ddl
e of t
h
e
day
.
I
n
t
h
i
s
case, t
h
re ki
nd
s of
spi
k
e
were c
o
nside
r
e
d
,
only
f
o
r t
h
e
half h
o
u
r s
p
ike
.
If a
half
ho
u
r
spike
(M
C
1
) is defi
ned
fo
r t
h
e total m
a
rket cost
with
ou
t th
e DSR p
r
og
ram
,
th
en
th
e to
tal m
a
rk
et co
sts of e
v
ery spi
k
e a
r
e a
s
prese
n
ted i
n
Table
2.
Tab
l
e 2
.
To
tal Mark
et
C
o
st W
i
t
h
ou
t DSR Program
Half Hour Spike (MC
1
)
T
o
tal M
a
r
k
et Cost
(
A
$)
5.
99
3.
2.
Co
st a
s
a
Functi
o
n
of a
Price
Spike Under DS
R Prog
ram
Co
nsidering
to the Pro
b
a
b
ility Spike
The c
ont
rol
sy
st
em
opt
im
i
s
ed t
h
e r
oom
t
e
m
p
erat
ure
of
t
h
e
ai
r co
n
d
i
t
i
oni
ng t
o
defi
ne t
h
e ener
gy
c
o
st
for con
s
u
m
ers. Th
e aim
o
f
th
e co
n
t
ro
ller is to
m
a
in
tain
th
e te
m
p
eratu
r
e
b
e
tween
th
e
p
e
rmit
ted
m
a
x
i
m
u
m an
d
minim
u
m
te
mperat
ures
in
orde
r to
provide a com
f
orta
ble room
te
m
p
er
ature
for the consum
er. In thi
s
o
p
t
i
m
i
s
a
t
i
o
n
,
th
e
ma
x
i
mu
m a
n
d
mi
n
i
mu
m t
e
mp
e
r
a
t
u
r
e
s
w
e
r
e
2
5
o
C
an
d
21
o
C. Tem
p
erature
starting of
22
o
C
was c
hose
n
.
U
nde
r
DSR
p
r
og
ram
t
h
e cy
cli
ng t
e
m
p
erat
ure
ro
om
was l
o
n
g
e
r t
h
a
n
wi
t
h
o
u
t
DSR
pr
og
ram
.
Thi
s
is to
g
i
v
e
m
o
re o
p
tion
and
mo
re
flex
i
b
ility
for th
e
op
ti
m
i
s
a
tio
n
.
In
add
itio
n, sin
ce t
h
e price sp
ike m
a
y
o
ccur
in
th
e m
i
d
d
l
e
of th
e d
a
y, th
e co
n
s
u
m
er is requ
ired
to
optimise to ac
hieve
minim
u
m
expe
cted ene
r
gy costs.
Un
de
r t
h
e DS
R
pro
g
r
am
, t
h
e cont
r
o
l
sy
st
em
appl
i
e
d t
h
e
pre-c
o
ol
i
n
g m
e
t
hod t
o
a
v
o
i
d hi
g
h
cost
s
whe
n
a spi
k
e hap
p
e
n
s. Si
m
i
lar t
o
t
h
e p
r
evi
ousl
y
desc
ri
be
d m
e
t
hod
, t
h
e
ai
r con
d
i
t
i
oni
n
g
was t
u
rne
d
o
n
o
n
ce
th
e te
m
p
eratu
r
e ro
se to
th
e
max
i
m
u
m
p
e
rmit
ted
te
m
p
er
ature. T
h
en, it was turne
d
off whe
n
the temperat
ure
dr
o
ppe
d t
o
t
h
e
m
i
nim
u
m
perm
i
t
t
e
d t
e
m
p
erat
ure. T
h
e c
ont
r
o
l
sy
st
em
kept
t
h
e r
oom
t
e
m
p
erat
ure
bet
w
e
e
n t
h
e
m
a
x
i
mu
m a
n
d
mi
n
i
mu
m p
e
r
m
i
t
t
e
d
t
e
mp
e
r
a
t
u
r
e
s
.
I
f
S
is the
electricity pri
ce whe
n
a
spi
k
e occ
u
rs,
K is
the
penalty, then the total m
a
rket cost
for t
h
e s
p
ike case
(
MC
is d
e
termin
ed
b
y
the fo
llo
wi
n
g
equ
a
tio
n
:
MC
t
S
t
.
P
t.
Dt.
U
td
t
K
(7
)
To
con
s
id
er the case wh
ere th
ere is a fin
ite
p
r
ob
ab
ility th
at a p
r
ice sp
ik
e may o
ccu
r at t
h
e mid
d
l
e o
f
t
h
e day
,
t
h
e
n
we nee
d
ed
a n
e
w fo
rm
of o
p
t
i
m
i
sati
on. He
r
e
we
di
d not
k
n
o
w
w
h
at
t
h
e
pri
ce w
oul
d be
,
s
o
we
forced th
e swit
ch
ing
t
o
b
e
t
h
e sam
e
u
p
un
til th
e sp
ik
e
ti
m
e
. Th
e switch
i
ng wo
u
l
d
b
e
d
i
fferen
t
fro
m
th
at ti
me
on
wa
rds
,
depe
ndi
ng
o
n
w
h
et
her t
h
e s
p
i
k
e
real
l
y
occu
rre
d.
In
t
h
i
s
case
,
t
h
e
r
e we
re
f
o
u
r
s
w
i
t
c
hes
whi
c
h
characte
r
ised t
h
e tim
e up to t
h
e
price
eve
n
t
and a
rem
a
i
n
i
ng
16
s
w
i
t
c
hes
un
de
r t
h
e
t
w
o
swi
t
c
hi
n
g
s
cen
ari
o
s
.
Th
is
g
a
v
e
a total o
f
3
6
switch
ed
g
e
s to
o
p
t
i
m
ise. In
th
is
research, th
e
pro
b
a
b
ility o
f
a
h
a
lf
h
our sp
ike was
2
.
2
%
. To
con
s
id
er wh
en
th
ere is a fin
ite p
r
o
b
a
b
ility (
P
) that a p
r
ice sp
ike will o
ccu
r i
n
th
e system
,
we
com
puted
MC
as t
h
e t
o
tal m
a
rket cost
without
a
s
p
ike
oc
curri
ng a
nd
MC
as t
h
e to
tal co
st assumin
g
a sp
ike
o
ccurs. Th
e total m
a
rk
et co
st
co
n
s
i
d
eri
n
g t
h
e prob
ab
ility for
h
a
lf
ho
ur sp
ik
e (
TMC
i
s
t
h
us
gi
ve
n
as
t
h
e
fo
llowing
equ
a
tio
n
:
TMC
M
C
t
∗P
M
C
t
∗
1
P
(8
)
Su
bject
t
o
c
o
ns
t
r
ai
nt
s:
MC
t
S
t
.
P
t.
Dt.
U
td
t
K
(9
)
MC
t
S
t
.
P
t.
Dt.
U
tdt
K
(1
0)
3.
2.
1 H
a
lf H
o
ur Spike
Case
Eq
uat
i
ons
(
1
)
t
o
(
5
) a
n
d
(
7
)
t
o
(
1
0)
we
re
us
ed t
o
com
put
e
t
h
e
num
eri
cal
res
u
l
t
s
o
f
opt
i
m
i
s
at
i
on of
th
e air con
d
ition
i
ng
co
nsid
eri
n
g th
e
prob
ab
ility th
at a h
a
lf
h
our sp
ik
es m
a
y o
ccur i
n
th
e
mid
d
l
e of th
e
d
a
y, as
sh
own
i
n
Fi
gu
r
e
s 6 an
d Tab
l
e II
I.
Figu
res 5 and 6 ind
i
cate th
e
numer
ical r
e
su
lts of
air cond
itio
n
i
n
g
o
p
tim
isat
io
n
if
a sp
ik
e m
a
y o
ccu
rs
(wit
h
2.2% p
r
ob
ab
ility) an
d
no
sp
i
k
e may o
ccurs (with
97
.8
%
p
r
ob
ab
ility)
i
n
t
h
e
m
i
ddl
e of t
h
e day
.
T
h
e i
n
si
de ro
om
tem
p
erat
ur
es w
e
re un
de
r t
h
e m
a
xim
u
m
perm
i
t
t
e
d t
e
m
p
erat
ure an
d
ab
ov
e th
e m
i
n
i
m
u
m
p
e
r
m
it
ted
te
m
p
eratu
r
e. Figu
res
5
an
d
6
illu
strat
e
th
at th
e ti
me an
d
i
n
sid
e
roo
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Mitig
a
tin
g Electricity a
Price Sp
ike und
er Pre-Coo
ling
Meth
od
(Marwan
Ma
rwan
)
1
287
te
m
p
erature
for switch
num
bers
1 to 4 we
re
equal. This
is because, during
this
tim
e,
the proba
b
ility of
a spi
k
e
was no
t con
s
idered. In
con
t
rast, d
u
e
to
t
h
e prob
ab
ility o
f
th
e sp
ik
e in
th
e
mid
d
l
e o
f
th
e
day, th
e ch
aract
eristics
of s
w
i
t
c
h n
u
m
b
ers
5 t
o
2
0
w
e
re n
o
t
i
d
ent
i
c
al
. Fi
gu
re 6 d
e
m
onst
r
at
es t
h
e appl
i
cat
i
o
n of
a pre-c
o
ol
i
n
g m
e
t
hod
bef
o
re
a
spi
k
e
ha
p
p
ens
i
n
t
h
e m
i
ddl
e of
t
h
e day
.
T
h
e
room
te
m
p
erature dropped to less tha
n
20
o
C. Th
is
te
m
p
eratu
r
e is called
a p
r
e-coo
lin
g
tem
p
eratu
r
e. Th
e air
con
d
ition
i
ng
statu
s
wh
en
t
h
e spik
e started
to
hap
p
en
was off
u
n
til th
e du
ration
of
th
e sp
ik
e
was
n
early exp
i
red
.
As a resu
lt, the roo
m
te
m
p
eratu
r
e wh
en
th
e sp
ik
e
hap
p
e
n
ed
o
n
l
y
dr
op
pe
d t
o
2
1
.
5
o
C. In
co
n
t
rast, as th
e sp
i
k
e p
r
o
b
a
b
ility was n
o
t
co
n
s
i
d
ered
for th
e
n
o
-sp
i
k
e
case, t
h
e
r
oom
t
e
m
p
erat
ure
d
r
op
pe
d t
o
2
1
o
C
,
as s
h
o
w
n i
n
Fi
gu
re
6.
It is clear from
Figures
5 that a pre-c
o
oling m
e
t
hod wa
s
necessa
ry to
m
i
nim
i
se the energy cost if
co
nsid
eri
n
g
t
h
e sp
ik
e
p
r
ob
abilit
y. Th
e con
t
ro
l syste
m
ap
p
l
ied
a pre-coo
lin
g
m
e
th
o
d
to
av
o
i
d
exp
e
n
s
ive co
st
w
h
en
th
e
sp
ik
e h
a
pp
en
s
.
Th
is w
a
s
b
e
c
a
u
s
e th
ere is a sub
s
t
a
n
tial risk
of th
e price sp
ik
e. Th
e to
tal co
st
co
u
l
d
t
h
en
be
n
u
m
e
ri
cal
l
y
opt
i
m
i
s
ed
by
v
a
ry
i
n
g t
h
e
36
s
w
i
t
c
h e
dge
s. T
h
e
r
e
we
re
20
swi
t
c
hes
of
spi
k
e case
s
and
2
0
swi
t
c
hes o
f
n
o
-
spi
k
e, wi
t
h
a r
e
m
a
i
n
i
ng 1
6
swi
t
c
hes f
o
r b
o
t
h
of t
h
em
unde
r di
ffe
re
nt
scen
ari
o
s. Si
m
i
l
a
r
to t
h
e
p
r
o
cess
d
e
scrib
e
d
abo
v
e, th
e co
st cou
l
d
b
e
calcu
lated
acco
r
d
i
ng
to
t
h
e air con
d
ition
i
ng statu
s
. Th
e cost o
f
a
price
spi
k
e cas
e was
m
o
re expensive t
h
an the cost
of a
no-spike
case. T
h
e total m
a
rket cost (TMC
30
), m
a
rke
t
cost s
p
ike
(MC
30
) a
n
d m
a
rket cost
no s
p
ike
(MC
n
) are
gi
ve
n i
n
Ta
bl
e 3
.
Fig
u
re
5
.
Nu
merical resu
lt of
h
a
lf
ho
ur sp
ik
e prob
ab
ilty case
Fig
u
re
6
.
Nu
merical resu
lt of
n
o
-sp
i
k
e
prob
ab
ility case (h
al
f
h
o
u
r
sp
i
k
e)
Tab
l
e
3
.
To
tal
Mark
et C
o
st for Half
Hour Sp
ik
e Case C
o
n
s
i
d
eri
n
g to
t
h
e Prob
ab
ility Sp
ike
TMC
30
(A$
)
MC
30
(A$
)
MC
n
(A$
)
3
.
0
6
4
.
2
7
3
.
0
3
10
11
12
13
14
15
16
17
18
19
19
20
21
22
23
24
25
26
Hal
f
H
o
ur
S
p
i
k
e of
P
r
ob
ab
i
l
i
t
y
Unde
r
DS
R P
r
og
r
a
m
T
i
m
e
(ho
u
r)
T
e
m
p
er
a
t
ur
e (
o
C)
10
11
12
13
14
15
16
17
18
19
0
2
4
6
T
i
m
e
(ho
u
r)
Mark
et
C
o
s
t
(A
$)
10
11
12
13
14
15
16
17
18
19
19
20
21
22
23
24
25
26
N
o
S
p
i
k
e
o
f
P
r
ob
abi
l
i
t
y
U
nder
D
S
R
P
r
og
ram
(
H
al
f
H
our
S
p
i
k
e)
T
i
m
e
(
hour
)
T
e
m
p
erat
ure
(
o
C)
10
11
12
13
14
15
16
17
18
19
0
1
2
3
4
T
i
m
e
(
hour
)
M
a
rk
et
C
o
s
t
(A
$
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E
V
o
l
.
6,
No
. 3,
J
u
ne 2
0
1
6
:
12
8
1
– 12
93
1
288
3.
3.
Benefi
t of
DS
R
Based
o
n
th
e
resu
lts of the op
ti
m
i
satio
n
rep
o
rted
above,
the cons
um
er and aggregat
or could
gai
n
co
llectiv
e b
e
n
e
fits wh
en
th
e co
n
s
u
m
er co
n
t
ro
ls th
e
ai
r c
o
n
d
i
t
i
oni
n
g
un
de
r t
h
e
DS
R program
.
The coll
ective
bene
fi
t
(C
B
)
i
s
ex
pres
sed
by
t
h
e
fol
l
o
wi
n
g
e
quat
i
o
n:
CB
30
= MC
1
- TM
C
30
(1
1)
Th
e
p
e
rcen
tag
e
of co
llectiv
e ben
e
fit is illu
strated
b
y
t
h
e
fo
ll
o
w
i
n
g equ
a
tion
s
:
%CB
$
$
(1
2)
Eq
uat
i
ons
(
1
1
)
t
o
(1
2
)
we
r
e
use
d
t
o
c
o
m
put
e t
h
e col
l
ect
i
v
e bene
fi
t
.
Tabl
e
4 s
u
m
m
a
ri
ses t
h
e
collective be
ne
fit for t
h
e c
ons
um
er and aggregator
wh
e
n
t
h
e cons
um
er applied the
DSR
program
if a
spike
may o
n
l
y o
c
cur in th
e m
i
d
d
l
e
o
f
th
e
d
a
y co
n
s
id
ering
t
o
th
e prob
ab
ility sp
ike.
Tab
l
e
4
.
C
o
llectiv
e Ben
e
fit if
Sp
ik
e May On
l
y
Occur i
n
th
e
Mid
d
l
e
o
f
Day
Spike Duration
W
ithout DSR
MC
1
(A$
)
Under DSR
TMC
30
(A$
)
Collective Benefit
(A$
)
(%
)
Half
Ho
u
r
Sp
ik
e
5
.
9
9
3
.
0
6
2
.
9
3
4
8
.
9
1
%
As
prese
n
ted in Ta
ble 4, the
consum
er and a
g
gr
egat
or c
a
n earn c
o
llective be
nefits i
f
the
DSR
p
r
og
ram
is app
lied
to m
eet a
p
r
ice
sp
i
k
e con
s
id
ering
th
e sp
ik
e
prob
ab
ilit
y; fo
r ex
am
p
l
e, 2.93
A$
(48
.
9
1
%)
fo
r a
hal
f
ho
ur
spi
k
e
.
T
h
i
s
re
sul
t
i
ndi
cat
es t
h
e p
r
e-
co
ol
i
n
g
m
e
t
hod
was e
ffect
i
v
e t
o
m
i
n
i
m
i
se t
h
e ener
gy
cost
whe
n
a s
p
ike
happe
n
s. E
v
e
n
though t
h
e spi
k
e proba
b
ility was sm
aller, the pre-cooling
m
e
thod
was
requi
red
t
o
ant
i
c
i
p
at
e hi
gh c
o
st
s w
h
en
a spi
k
e ha
ppe
ns. T
h
e p
r
e-c
o
ol
i
ng m
e
t
hod
was o
n
l
y
appl
i
e
d w
h
en t
h
e
r
e
was a
su
bstan
tial risk of a
p
r
ice
sp
i
k
e.
4.
CO
NCL
USI
O
N
Thi
s
pa
per
has
dem
onst
r
at
ed t
h
at
t
h
e pr
op
os
ed DS
R
m
odel
al
l
o
ws cons
u
m
ers t
o
m
a
nage and co
nt
r
o
l
air
cond
itio
n
i
ng
for
ev
er
y
p
e
r
i
od
b
a
sed on
th
e electr
i
c
ity mark
et price.
Th
e m
o
d
e
l is
ap
p
licab
le fo
r
b
o
t
h
reside
ntial and comm
ercial c
ons
um
ers
to
min
i
m
i
se th
e co
st o
f
flu
c
t
u
atin
g
energy prices.
The proposed m
odel
can
assist th
e
co
nsu
m
er to
op
ti
m
i
se th
e en
erg
y
cost o
f
ai
r con
d
ition
i
ng to
m
eet a p
r
ice sp
ik
e.
Th
is resu
lt
i
ndi
cat
es t
h
at
,
t
h
e co
ns
um
er sho
u
l
d
ap
pl
y
t
h
e p
r
e-c
o
ol
i
n
g
m
e
t
hod t
o
m
i
ni
m
i
se ener
gy
cost
s by
a
n
t
i
c
i
p
at
i
n
g
the electricity
price s
p
ike
when we
know the spike m
a
y occur i
n
the m
i
ddl
e
of t
h
e da
y
.
In a
d
di
t
i
on,
a pre
-
cool
i
n
g m
e
t
h
o
d
s
h
o
u
l
d
be a
ppl
i
e
d
t
o
a
voi
d hi
g
h
el
ect
ri
ci
t
y
pri
ces at
cri
t
i
cal
t
i
m
e
s. Ho
we
ver
,
p
r
e-
cool
i
n
g
sho
u
l
d
onl
y
be un
de
rt
ake
n
wh
en
t
h
e
r
e
i
s
a
s
u
bst
a
nt
i
a
l
ri
s
k
o
f
a pri
ce spi
k
e.
REFERE
NC
ES
[1]
Singh.
, S
.
K. a
n
d P.
S.S
.
M
o
re,
"
Dem
a
n
d
Sid
e
Man
a
g
e
men
t
Po
ten
tial at th
e B
h
arati Hosp
ital and
Research Ce
ntre
,
"
I
n
t
e
rn
at
i
o
nal
J
our
n
a
l
of
El
ect
ri
cal
an
d
C
o
m
put
er E
n
g
i
neeri
n
g (
I
JEC
E
)
. 2
:
p
p
.
511-
5
1
8
,
201
2
[2]
Ali M
a
nso
u
ri,
et al.,
" Ev
alu
a
tio
n
of Power Syste
m
Rel
i
ab
ility
C
o
n
s
id
ering
Direct Lo
ad
Co
n
t
ro
l Effects
,
"
In
tern
a
tiona
l
Jo
urna
l
o
f
Electrica
l
an
d Compu
t
er Eng
i
n
e
erin
g (IJECE)
.
Vo
l. 3
,
: pp
. 25
4-2
5
9
,
201
3
[3]
Duy
Tha
n
h N
guy
e
n
, "Dem
and R
e
s
p
o
n
se
eXcha
n
ge i
n
a Dereg
u
l
a
t
e
d
Envi
r
o
nm
ent
,
" i
n
Scho
ol
o
f
En
gi
neeri
n
g
Un
i
v
ersity of
Tasm
an
ia: Tasman
ia, 20
12
[4]
M
a
rwa
n
,
G.
L
e
dwi
c
h, a
n
d A
.
G
hos
h,
" Dem
a
nd
-si
d
e
res
p
o
n
se m
odel
t
o
a
voi
d s
p
i
k
e
of
e
l
ect
ri
ci
t
y
pri
c
e
,
"
Jour
n
a
l
of
Pr
ocess C
o
nt
rol
.
24
: pp
. 782
-789
,
20
14
[5]
M
a
rwa
n
, "Sm
a
rt Gri
d
-
D
em
and side
res
p
o
n
s
e
m
odel
to m
i
tigate price a
nd peak im
pact on the electrical
syste
m
," in
Sci
n
ce
an
d e
n
gi
ne
eri
n
g
f
a
c
u
l
t
y
Q
u
een
sl
an
d
Uni
v
ersi
t
y
o
f
Tec
h
nol
ogy
:
B
r
i
s
ba
ne,
2
0
1
3
[6]
Su,
C.-L
., "
O
ptim
al dem
a
n
d
-si
d
e
particip
a
tion in day
-
ahead electricity
m
a
rkets," in
Fa
cu
lty o
f
En
gi
neeri
n
g a
n
d
P
h
ysi
c
al
Sc
i
e
nces
M
a
nc
hes
t
er:
M
a
nche
st
r,
U.
K,
2
0
0
7
[7]
Aust
ral
i
a
n E
n
e
r
gy
M
a
r
k
et
O
p
erat
or
, "C
u
rre
n
t
Tradi
n
g
Int
e
r
v
al
Pri
ce a
nd
Dem
a
nd G
r
ap
h Q
u
ee
nsl
a
n
d
,
"
"Aust
r
al
i
a
n E
n
ergy
M
a
rket
Op
er
at
o
r
, Ed
ito
r, 20
12
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Mitig
a
tin
g Electricity a
Price Sp
ike und
er Pre-Coo
ling
Meth
od
(Marwan
Ma
rwan
)
1
289
[8]
Christian Crowley, "
W
eathe
r Effects on Electric
ity Loads: Modeling and Fo
recasting," Departm
e
nt of
Econ
o
m
ics The G
e
o
r
g
e
W
a
sh
ing
t
on
U
n
i
v
ersity: W
a
sh
ing
t
o
n
,
2
005
[9]
Quee
nsl
a
nd
G
ove
r
n
m
e
nt
, "B
ui
l
d
i
n
g C
l
im
ate Sm
art
i
n
Qu
eensl
an
d," "
D
e
p
art
m
ent
of E
nvi
ro
nm
ent
and
R
e
sou
r
ce M
a
n
a
gem
e
nt
, Edi
t
o
r Q
u
ee
nsl
a
n
d
Go
ve
rnm
e
nt
:
B
r
i
s
ba
ne,
2
0
0
8
[
1
0
]
P
e
ter
s
on
, M.
, "Man
ag
in
g P
e
ak
D
e
m
a
n
d
,
" in
EE-
OZ A
n
nu
al
C
o
nf
erence
:
Go
ld
C
o
ast Br
isb
a
n
e
,
2
010
[1
1]
B
o
rl
i
c
k,
R
.
,
" Pay
i
ng fo
r D
e
m
a
nd-
Si
de R
e
sp
onse at
t
h
e
Wh
ol
esal
e Level
:
The Sm
al
l
C
onsum
ers’
Persp
ectiv
e
,
"
The Electricity Jour
nal
.
24
: p
p
.
8-1
9
. 2
011
[12]
Jia,
W., C.
Ka
ng, a
n
d Q
.
Chen,
" An
alysis on
d
e
m
a
n
d
-
si
d
e
in
teractiv
e
respon
se cap
a
b
ility fo
r
po
wer
syste
m
dispatch in a sm
art gri
d
fram
e
work
,
"
Electric Power
Syste
m
s Rese
arch
.
9
0
:
pp
. 11-
17
.
201
2
[1
3]
Xu
, Y.
a
n
d K.
Na
gasa
ka,
"
A Resea
r
ch on Spi
k
e J
u
m
p
of Electri
city Price in
th
e Dereg
u
l
ated Power
Mar
k
ets
,
"
In
tern
a
tiona
l Jou
r
na
l o
f
Electrica
l an
d Po
wer Eng
i
n
eeri
n
g
.
3
:
pp
. 99
-10
4
. 2
009
[
1
4
]
A
i
, X., et al. "Bid
-
s
ch
edu
ling o
f
d
e
m
a
n
d
sid
e
r
e
ser
v
e
b
a
sed
on
dem
a
n
d
r
e
spon
se con
s
id
er
ing
carbon
e
m
issio
n
trad
i
n
g
i
n
sm
art g
r
id
." in
Critica
l
In
fra
s
tru
c
t
u
re (CRIS
)
, 2
0
1
0
5
t
h
In
tern
a
t
i
o
n
a
l
C
o
n
f
eren
ce
on
. 2
010
.
[1
5]
Perf
u
r
m
o
, C
.,
J.K.
War
d
, an
d Jul
i
o
H
.
B
r
a
s
l
a
vsky
. "R
ed
u
c
i
ng Ene
r
gy
U
s
e and O
p
erat
i
onal
C
o
st
o
f
A
i
r
C
o
n
d
i
t
i
oni
ng
Sy
st
em
s W
i
t
h
M
u
l
t
y
Object
i
v
e Ev
ol
ut
i
ona
r
y
Al
go
ri
t
h
m
s
,." i
n
C
onf
ere
n
ce Evol
ut
i
o
n
a
r
y
C
o
m
put
at
i
o
n (
C
EC)
. 2
010
.
[16]
Lee, S.C.
, S.J.
Kim
,
and a.S
.
H. Kim
,
" Deman
d
Sid
e
Man
a
g
e
m
e
n
t
W
ith
Air C
o
nd
ition
e
r Lo
ad
s Based
on the
Queui
n
g System
Model
,
"
IE
EE T
r
ansactions
on
P
o
wer
Systems
,
.
26
: p
p
. 6
61-
66
8. 2
011
[
1
7
]
H
a
db
an
,
Y
.
A
., "D
em
an
d
-
Side Man
a
g
e
m
e
n
t
in
Of
f
i
ce B
u
ild
ing
s
i
n
Ku
w
a
it thro
ugh an
I
ce-Stor
age
Assisted HVAC System
with
M
o
d
e
l
Pred
ictiv
e Con
t
ro
l,," i
n
Sc
ho
ol
of
E
ngi
neeri
n
g
Cran
fil
e
d
Un
i
v
ersity: Kuwait, 200
5.
BIOG
RAPHI
E
S OF
AUTH
ORS
Dr. Ma
r
w
an
Mar
w
an
receiv
ed th
e B.Eng
degree fro
m Hasanuddin Univ
ersity
Mak
a
ssar
Indonesia, th
e M.Eng degree and the Ph.D fro
m
Queensland
University
of Technolog
y
(QUT)
Brisbane Australia, in 2000
, 200
6 and 2013
resp
ectiv
ely
,
all
in electr
i
cal
power
engineer
ing. He
has been
a lectur
er with
the State Poly
tec
hnic of
Ujung Pandang
Makassa
r Indon
esia sin
c
e 2001
until
curren
t
l
y
,
i
n
El
ectr
i
c
a
l
Engi
neering
Depar
t
m
e
nt.
Dr.
Pirman, M.Si
recived
the B
.
Eng d
e
gree fro
m Hasa
nuddin University
Mak
a
ssar Indonesia,
the m
a
ster
dan
doctor d
e
grees
from
Bandung I
n
stitute T
echnol
og
y
in 1987
, 1
994 and 2001
res
p
ect
ivel
y,
a
ll
in Chem
ica
l
En
gineer
ing. He h
a
s
been a l
e
c
t
urer
at P
o
l
y
t
echni
c
S
t
ate of Ujung
Pandang since 1
989 until cur
r
ently
in
Chemical Engineer
ing
Dep
a
rtment
Evaluation Warning : The document was created with Spire.PDF for Python.