Internati
o
nal
Journal of P
o
wer Elect
roni
cs an
d
Drive
S
y
ste
m
(I
JPE
D
S)
Vol.
6, No. 4, Decem
ber
2015, pp. 712~
722
I
S
SN
: 208
8-8
6
9
4
7
12
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
/
IJPEDS
Partial Shading in Building Integrated PV System: Causes,
Effects and Mitigating Techniques
Za
i
n
a
l
S
a
la
m
1,
2
, Z
u
lkifli Ramli
3
, Jubaer
Ahmed
1
, Muh
a
mm
ad Am
ja
d
4
1
Center of
Electrical En
erg
y
S
y
s
t
ems, Faculty
of
Electrical Eng
i
n
eer
ing
,
Univ
eristi Teknologi Malay
s
ia,
81310 Johor Bahru, Malay
s
ia
2
Insitute of
Futu
re En
erg
y
, Univ
eristi Te
kno
logi
Malay
s
ia, 81310
J
ohor Bahru
,
M
a
lay
s
ia
3
Facul
t
y
of
El
ec
tric
al
Engin
eerin
g, Universi
ti
Tek
n
ikal
Mala
ysi
a
,
Melaka
, Air
Ker
oh, Mel
a
ka
, Ma
l
a
y
s
i
a
4
College o
f
Eng
i
neer
ing and
Technolog
y
,
The Islami
a University
of Bahawal
pur, Bahawalpur
,
Pakistan
Article Info
A
B
STRAC
T
Article histo
r
y:
Received
J
u
l 22, 2015
Rev
i
sed
Sep
15
, 20
15
Accepte
d Oct 2, 2015
This paper
is aimed to provide
a ho
listic unders
tanding on th
e issues related
to parti
a
l shadin
g: its causes,
th
e theor
e
ti
cal
and
ph
y
s
ic
al re
ason
s behind it,
its im
plic
ations
on the BIPV s
y
stem
. Furtherm
ore th
e possible
m
itigati
o
n
techn
i
ques using
the softwar
e
(M
PPT) a
nd hardw
a
re solutions ar
e discussed.
Finally
an
example is given to
illustrate
the impact of partial shad
ing and th
e
econom
ic ben
e
fits of em
plo
y
ing v
a
rious
parti
a
l shading
m
itigation
techn
i
ques into t
h
e BIPV sy
stem
To ai
d the unfa
m
iliar read
ers in this subject
,
a brief bu
t
comprehensive overv
iew of im
portant
PV concepts
are
also giv
e
n.
Keyword:
In
verte
r
MPPT
Partial Sh
ad
ing
Pho
t
ov
o
ltaic
Solar Ene
r
gy
Copyright ©
201
5 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
:
Zain
al Salam
,
Center
of Electrical Ene
r
gy
Syste
m
s,
Faculty of Electrical Engineeri
n
g,
Un
i
v
eristi Tekn
o
l
o
g
i
Malaysia,
8
131
0 Joho
r B
a
h
r
u
,
Malaysia.
Em
a
il: zain
a
ls
@fk
e
.u
tm
.
m
y
1.
INTRODUCTION
As a rene
wa
ble source, the s
o
lar photovolt
a
ic (PV)
has attracted consi
d
erable interest
for se
veral
reason
s:
1
)
th
e con
tin
uou
s
d
e
clin
e in
t
h
e
pri
ce of PV m
o
dules,
2) the
mature a
n
d
reliable electronic
powe
r
co
nv
ersion
tech
no
log
y
, 3) the si
m
p
le in
sta
llatio
n
p
r
o
cess an
d
low m
a
i
n
ten
a
n
ce co
st
4
)
en
v
i
ron
m
en
tally
fri
en
dl
y
.
I
n
ad
di
t
i
on t
o
t
h
ese
,
m
a
ny
count
ri
es have i
n
t
r
o
d
u
ced
vari
ous i
n
cent
i
v
e
pr
o
g
r
a
m
s
such as fe
ed-i
n-
tariff, tax brea
ks a
n
d initial inve
stm
e
nt reba
tes to acceler
ate the
growth of this indus
t
r
y. In
recent
years,
one
ap
p
lication
th
at h
a
s
g
a
in
ed
i
mmen
s
e atten
tio
n
is th
e
bu
ild
ing
in
teg
r
ated PV (BIPV)
[1]. Besid
e
s
p
r
ov
id
eing
ener
gy
t
o
t
h
e
bui
l
d
i
n
g
,
B
I
P
V
has
ot
he
r i
n
here
nt
eco
n
o
m
i
c pot
ent
i
a
l
:
t
h
e
m
odul
es ca
n
serve as t
h
e st
ruct
ura
l
el
em
ent
s
and
red
u
ces c
once
n
t
r
at
i
o
ns o
f
a
dde
d
wei
g
ht
on
roofs. In cases where
the feed-in-t
a
riff is
i
m
p
l
e
m
en
ted
,
t
h
e bu
ild
ing
can
g
e
n
e
rate sufficien
t
in
co
m
e
t
o
p
a
y-b
ack
its PV inv
e
stm
e
n
t
co
st—u
su
ally
wit
h
sizeable profit
margin.
Despite the
s
e
adva
ntage
s
, the BIPV in
urban area
s faces
considera
b
le c
h
allenge
s—
pri
m
arily due to
p
a
rtial sh
ad
ing fro
m
th
e tall
er adj
acen
t
build
in
g
s
, el
ectrical u
tility to
w
e
rs, tran
sm
issi
o
n
cab
les, telep
hone
poles
, a
n
tennas
, trees
etc.
In
many cases, t
h
e structure
s
a
r
e
n
o
t
prese
n
t
d
u
r
i
n
g t
h
e
pl
an
ni
ng
a
n
d
de
si
g
n
st
ages,
but are erecte
d
after the PV s
y
ste
m
has been commissione
d. As a res
u
lt,
the BIPV
has to endure the s
h
adi
ng
t
h
r
o
u
g
h
o
u
t
i
t
s
servi
ce l
i
f
et
i
m
e. D
u
ri
ng
part
i
a
l
shadi
n
g
,
i
.
e.
whe
n
one
or
m
o
re
m
odul
es
expe
ri
ence
di
f
f
e
rent
i
rradi
a
n
ce t
h
a
n
t
h
e rest
of t
h
e array
,
t
h
e by
pass di
ode
of
t
h
e
m
odul
e i
s
act
i
v
at
ed, t
h
us
causi
n
g
t
h
e shade
d
m
odul
e t
o
be
by
passe
d.
T
h
e
o
p
erat
i
o
n ca
u
s
es t
h
e
P-V
c
u
rve
t
o
c
o
nt
ai
n
num
ero
u
s m
a
xim
a
poi
nt
s
wi
t
h
on
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Pa
rtia
l
S
had
ing
in Bu
ild
ing
In
teg
r
a
t
ed PV
System: C
a
u
s
es, Effects a
nd Mitig
a
tin
g .... (Zain
a
l
S
a
l
a
m
)
71
3
g
l
ob
al p
e
ak
.
If th
e inv
e
rter is no
t equ
i
pp
ed with
an
MPPT th
at cou
l
d
differen
tiate b
e
t
w
een th
e lo
cal
and
g
l
ob
al p
eaks, th
ere is a p
o
s
sib
ility fo
r th
e alg
o
rith
m
to
b
e
trap
p
e
d
in
one o
f
th
e fo
rm
e
r
, th
us cau
sing th
e
p
o
wer to
dro
p
. It is statistical
ly rep
o
rted
in
t
h
at th
e en
erg
y
lo
ss du
e to
p
a
rtial sh
asd
i
ng
can
v
a
ry fro
m
1
0
to
7
0
% [2
].
Desp
ite
b
e
ing
well-do
cu
m
e
n
t
ed
in
literatures [3
], m
a
n
y
are n
o
t
aware
o
f
th
e adv
a
n
t
ag
es th
at can
b
e
obt
ai
ne
d
by
i
m
pl
em
ent
i
ng
cert
a
i
n
part
i
a
l
sha
d
i
n
g m
i
ti
gat
i
on m
easures
i
n
t
o
t
h
ei
r B
I
P
V
sy
st
em
. W
i
t
h
t
h
i
s
h
i
nd
sigh
t in
p
e
rsp
ectiv
e, th
is p
a
p
e
r d
i
scu
sses th
e cau
se
s, affects of p
a
rtial sh
ad
ing
and
m
o
re i
m
p
o
r
tan
tly
th
e
strateg
i
es fo
r i
t
s
m
i
tig
atio
n
.
A case stud
y is g
i
v
e
n
t
o
illu
strate th
e
b
e
n
e
fits of em
p
l
o
y
in
g
v
a
ri
o
u
s p
a
rtial
sha
d
i
n
g m
i
ti
gat
i
on t
ech
ni
ques
.
It
i
s
e
n
vi
sage
d t
h
at
t
h
i
s
wo
r
k
wo
ul
d
be
a
v
a
l
u
abl
e
on
e-st
o
p
refe
rence
so
u
r
ce t
o
en
ab
le PV research
ers, system d
e
sig
n
e
rs an
d
p
r
ac
titio
n
e
rs
m
a
k
i
n
g
m
o
re in
form
ed
d
ecisio
n
s
in
desig
n
i
ng
BIPV system
t
h
at is su
bj
ected
to p
a
rtial sh
ad
ing
.
2.
SYSTE
M
O
V
ERVIEW
2.
1.
BIPV S
y
s
t
em
Configur
ation
The B
I
PV i
s
t
h
e i
n
t
e
grat
i
o
n
of
PV cel
l
s
/
m
od
ul
es i
n
t
o
t
h
e
bui
l
d
i
n
g en
ve
l
ope t
o
bec
o
m
e
part
of t
h
e
st
ruct
u
r
e [4]
.
Fi
gu
re 1 sh
o
w
s
a sim
p
li
fi
ed di
agram
of a B
I
PV fo
r a t
y
pi
cal
resi
de
nt
i
a
l
ho
m
e
. The
m
odul
es are
arra
nge
d i
n
ser
i
es st
ri
ngs t
o
a
c
hi
eve t
h
e re
q
u
i
r
e
d
w
o
r
k
i
n
g di
rect
(dc
)
v
o
l
t
a
ge. T
o
i
n
crea
se t
h
e po
wer
,
severa
l
of the
s
e strings
are connecte
d
in paralle
l to form
an array.
Since m
o
st ele
c
trical ap
p
liances run
on
altern
ating
(ac)
v
o
l
t
a
ge,
t
h
e dc
p
o
we
r e
x
t
r
act
ed
fr
om
t
h
e m
odul
es i
s
c
o
nve
rt
ed
t
o
ac
u
s
i
n
g
an
i
n
vert
e
r
[
5
]
.
Fi
gu
re
1.
A T
y
pi
cal
B
I
PV
sy
st
em
confi
g
u
r
a
t
i
o
n
2.
2.
The
I–V
an
d
P–V
Cur
v
e an
d
MPPT
In i
t
s
m
o
st
bas
i
c fo
rm
, a PV
m
odul
e (
o
r cel
l
)
can
be
m
odel
l
e
d as a c
u
r
r
e
n
t
s
o
u
r
ce t
h
at
i
s
de
pe
nde
n
t
on t
h
e s
o
lar irradiance
(
G
) a
n
d tem
p
erature
(
T
)
.
T
h
e va
ri
at
i
on i
n
G
and
T
r
e
su
lts in non-
lin
ear
I–
V
and
P–V
cu
rv
es as sh
ow
n in
Figu
r
e
2
(
a)
an
d
(b
),
resp
ectiv
ely.
A
t
an
y tim
e, th
ere ex
ists a
un
iqu
e
o
p
e
rating
po
in
t, at
wh
ich
th
e
p
o
wer is at t
h
e
p
e
ak
(MPP).
Du
e
to
th
e d
y
n
a
m
i
c
s
of
G
an
d
T
,
a
MPP trac
ke
r
(MPPT) is nee
d
ed to
ens
u
re m
a
xim
u
m
power is ext
r
acted
from
th
e m
odules
unde
r a
n
y environm
ental condition.
The M
PPT ca
n be cat
e
g
o
r
i
z
ed i
n
t
o
t
h
e co
nve
nt
i
o
nal
an
d
no
n c
o
n
v
e
n
t
i
onal
.
The
wo
r
k
i
n
[
6
]
ha
ve
revie
w
ed the c
o
nve
n
tional M
PPT m
e
thods
whic
h incl
ude
the Pert
urb a
nd Observe
d
(P&
O
)
[7], Hill Clim
bing
(
H
C) an
d In
cre
m
en
tal Co
ndu
ctan
ce (IC),
as well as th
e lesser
kno
wn Rip
p
l
e Correlatio
n
,
Op
en
Circu
it
Vo
ltag
e
, Sh
ort
Circu
it Cu
rren
t and
Slid
i
n
g
Co
n
t
ro
l.
Th
e
no
n-
co
nv
en
tio
nal MPP m
e
th
od
s ar
e
b
a
sed
on
so
f
t
-
com
put
i
ng,
w
h
i
c
h rel
i
e
s m
a
i
n
l
y
on t
h
e sea
r
c
h
an
d
opt
i
m
i
z
at
i
on ap
p
r
oac
h
[8]
-
[
1
0
]
.
Am
o
ng t
h
e co
n
v
ent
i
onal
MPPT, t
h
e P&O is th
e m
o
st wid
e
ly u
s
ed
[11
]
. Th
e
g
o
a
l
o
f
th
e algorith
m
is to
po
sition
t
h
e
o
p
e
rating
po
in
t as
clo
s
est as
po
ssib
le to
t
h
e MPP b
y
clim
b
i
n
g
th
e slop
e
of the
P-V
curve.
It
calculates the
powe
r (
P
) by
s
e
nsi
n
g
th
e vo
ltag
e
(
V)
o
f
t
h
e
P
V
ar
ra
y
;
t
h
en i
t
p
r
ovi
des a
pe
rt
u
r
bat
i
on
(
ϕ
) i
n
V
,
ba
sed
o
n
t
h
e
cha
nge
o
f
p
o
we
r, i
.
e.
new
ol
d
o
l
d
new
o
l
d
o
l
d
=X
+
(
i
f
>
)
[
]
=-
(
i
f
<
)
X
sl
ope
P
P
Whe
r
e
X
=
V
or
I
or
D
X
X
sl
ope
P
P
(1
)
whe
r
e, the sl
ope indicates the direction
of t
h
e pe
rturbation. Clearly, the size of
ϕ
is cru
c
ial; if
ϕ
is larg
e, th
e
co
nv
erg
e
n
ce is fast,
bu
t it resu
lts in
larg
e fluctu
atio
n
i
n
P
a
n
d vice
ve
rsa.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l.
6
,
No
.
4
,
D
ecem
b
er
2
015
:
71
2 – 722
71
4
(a)
(b
)
Figu
re
2.
(a
) T
h
e
I–V
and
P–V
cu
rve
s
un
der
va
ry
i
n
g
G
.
(b)
I–
V
an
d
P–
V
c
u
r
v
es u
nde
r di
f
f
ere
n
t
T
3.
PV CH
AR
A
C
TERISTIC
S
DU
RI
NG PA
RTIAL SH
A
D
IN
G
3.
1.
Parti
a
l Shadi
n
g
Phenomena
In
B
I
PV, th
e
main
cau
se
o
f
p
a
rtial sh
ad
ing is th
e ob
stru
ct
io
n
s
fro
m
th
e su
rroun
d
i
ng
bu
ild
ing
s
, and
o
t
h
e
r
tall str
u
ctu
r
es
su
ch
as tr
ees, telephon
e
p
o
l
e, tr
an
smissio
n
lin
es
an
d an
ten
n
a
s. Par
tial sh
ad
i
n
g-
lik
e
characte
r
istics is also exhibited by m
odule
irregularitie
s s
u
ch a
s
crac
ks. Practically, every m
odule i
s
fitted
wi
t
h
a
by
pass
di
o
d
e t
o
a
voi
d
a h
o
t
s
pot
(i
.e.
conce
n
t
r
at
i
o
n
of
cu
rre
nt
i
n
t
o
one
m
odul
e)
i
f
any
of
t
h
e m
odul
e i
s
sha
d
ed.
When the entire PV string
e
xpe
ri
e
n
ce a u
n
i
f
orm
i
rradi
a
n
ce (
w
hi
ch
is th
e
n
o
rm
al
co
n
d
ition), th
e
b
y
p
a
ss
d
i
od
e re
m
a
in
s o
ff
d
u
e
to
th
e eq
u
i
po
t
e
n
tial acro
ss its ter
m
in
als. In
th
is case, th
e
P–V
curve e
xhi
bits the
no
rm
al
si
ngl
e peak c
h
a
r
act
er
i
s
t
i
c
s, as sho
w
n by
C
u
r
v
e (
1
)
of
Fi
g
u
re
3(c
)
.
When
pa
rt
i
a
l
sha
d
i
n
g occ
u
r
s
, t
h
e
pot
e
n
t
i
a
l
di
ffe
rence t
r
i
gge
rs
t
h
e by
pas
s
di
o
d
e, ca
usi
n
g t
h
e cur
r
ent
t
o
b
e
di
vert
e
d
aw
ay
from
t
h
e shade
d
m
odule. Since
the s
h
ade
d
m
odule is
short circuited,
the
vol
t
age across
it ceases to
ze
ro. Conse
q
uently, the
P–
V
cu
rv
e is ch
aracterized
b
y
sev
e
ral lo
cal and
a g
l
o
b
a
l p
e
ak
, as illu
strated
b
y
Cu
rv
e (2). A goo
d
MPPT is
expect
e
d
t
r
ac
k
t
h
e gl
obal
,
w
h
i
l
e
av
oi
di
n
g
a
n
y
o
f
t
h
e l
o
cal
pea
k
..
U
n
f
o
rt
unat
e
l
y
, f
o
r t
h
e co
nve
nt
i
o
nal
M
P
P
T
(suc
h as P&O), once it locates a particul
ar peak (re
ga
rdless
whethe
r it is gl
o
b
a
l or lo
cal),
th
e alg
o
rith
m
f
o
rces
th
e op
eratin
g po
in
t to go
fo
rth and
b
ackwards aro
und
th
e p
e
rceiv
e
d
MPP, cau
s
ing
po
wer l
o
ss.
Fi
gu
re
3
(a)
P
V
un
de
r
pat
i
a
l
sha
d
i
n
g (
b
) t
h
e
res
u
l
t
i
ng
I–
V
(c)
P–
V
cu
rv
e
4.
PARTI
A
L SH
ADI
NG
M
I
TI
GATIO
N
To m
i
tigate the effect of pa
rt
ial shading, t
h
ree ap
p
r
oaches
are p
o
ssi
bl
e
.
F
i
rst
i
s
by
by
addi
ng m
o
re
in
tellig
en
ce to th
e P&O to
en
ab
le it t
o
d
i
fferen
tiate b
e
t
w
een
th
e lo
cal an
d g
l
o
b
a
l
p
e
ak. Altern
ativ
ely, soft
co
m
p
u
tin
g
tech
n
i
q
u
e
s can
b
e
u
tilized
to
scan
th
e en
tire
(or p
a
rt
o
f
t
h
e)
PV curv
e in
o
r
der to
locate th
e g
l
ob
al
p
eak. In
bo
th
cases, on
ly
mo
d
i
f
i
catio
n
s
in th
e so
f
t
w
a
r
e
, i.e. th
e MPPT co
d
e
s ar
e necessar
y
. Th
e
second
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Pa
rtia
l
S
had
ing
in Bu
ild
ing
In
teg
r
a
t
ed PV
System: C
a
u
s
es, Effects a
nd Mitig
a
tin
g .... (Zain
a
l
S
a
l
a
m
)
71
5
ap
pro
ach
is to retro
f
it th
e in
v
e
rter/conv
ert
e
r with
d
e
dicated energy rec
ove
ry ci
rcu
it to
h
a
rv
est wh
atev
er
energy is available from
the
sha
d
ed
m
odule
s
. It ens
u
res that powe
r
from
the shade
d
m
o
dule is utilized by
b
a
lan
c
ing th
e to
tal curren
t
t
h
at flow i
n
to
t
h
e PV stri
n
g
. Th
e t
h
ird ap
proach
is to
fit a
sm
a
ll in
v
e
rter
to
each
m
odul
e,
kn
ow
n as
t
h
e m
i
cro-
i
nve
rt
er.
I
n
t
h
i
s
m
e
t
hod, t
h
e c
e
nt
ral
i
n
ve
rt
er
i
s
n
o
l
o
n
g
er
re
qui
red
beca
use
eve
r
y
micro
-
inv
e
rter
is in
d
i
v
i
du
ally co
nn
ected d
i
rectly to
th
e
g
r
id.
4.
1.
Mo
dified P
&
O
Pat
e
l
&
Aga
r
w
a
l
[1
2]
pr
op
ose
d
a
t
w
o-m
ode
m
odi
fi
ed P&
O
t
h
at
di
vi
des t
h
e M
P
P t
r
ac
ki
ng
i
n
t
o
m
a
i
n
p
a
rts: th
e
g
l
obal an
d
th
e l
o
cal
track
ing
m
o
d
e
. Th
e
form
er is u
s
ed
to
b
r
ing
t
h
e op
eratin
g
po
in
t to
the v
i
ci
n
ity o
f
th
e g
l
ob
al
p
e
ak
,
wh
ile th
e lat
t
er is d
e
sign
ed to
m
a
in
tain
the o
p
e
rating
po
i
n
t on
ce th
e
g
l
ob
al p
e
ak
is d
e
t
ected
.
To
a
v
oi
d
t
h
e s
canni
ng
of
t
h
e ent
i
r
e vol
t
a
ge spa
n
,
t
w
o
cr
itical steps a
r
e ta
ken:
1) t
h
e
pea
k
s a
r
e a
ssum
e
d to
be
at th
e
m
u
ltip
les o
f
80
% of
V
oc
and 2
)
t
h
e m
i
ni
m
u
m
di
spl
acem
e
nt
bet
w
ee
n successi
ve pe
aks is approxi
m
atel
y
80
% of
V
oc
. F
u
rt
herem
o
e, t
h
e det
ect
i
on o
f
t
h
e part
i
a
l
sha
d
i
n
g i
s
t
r
i
gge
r
e
d by
a t
h
res
h
ol
d val
u
e
∆
P
cri
t
. If the
p
o
wer ch
an
g
e
is larg
er th
an
∆
P
crit
th
en
p
a
rt
ial sh
ad
ing
is assu
m
e
d
to
h
a
v
e
o
c
cu
rred
, pro
m
p
tin
g
algo
rith
m
to
j
u
m
p
in
to
th
e
g
l
ob
al m
o
d
e
. On
ce
with
in
the v
i
cin
ity o
f
the g
l
ob
al p
eak,
th
e alg
o
rith
m
switch
e
s
o
v
e
r
to
th
e
lo
cal m
o
d
e
, wh
ich
is t
h
e co
nv
en
tion
a
l P&O. It stays
th
ere un
til a n
e
w partial sh
ad
i
n
g
co
nd
itio
n is d
e
tected.
Sim
ilar tracking m
e
thod is also presente
d in [13]. Ho
we
ver, unlike the duty cycle
is
utilized as the
m
a
in
cont
rol
varia
b
le; thus it can
be categor
ized a
s
the hill cli
m
bing.
In anothe
r work [14], t
h
e
P&O is
re
placed by
the IC t
o
perform
the search
mechan
i
s
m
,
whi
l
e
t
h
e
pr
oce
d
ure t
o
di
st
i
n
g
u
i
s
h
bet
w
ee
n l
o
c
a
l
and
gl
obal
p
eak i
s
si
m
ilar to
[13
]
.
4.
2.
Soft Compu
t
i
n
g
So
ft co
m
p
u
tin
g
is an
opti
m
izatio
n
tech
n
i
q
u
e
th
at
ex
h
i
b
its n
a
tural ab
ility to
search
fo
r
m
i
nim
a
/m
i
n
ima poi
nt
s [
1
5]
. Depe
n
d
i
n
g o
n
t
h
e al
go
ri
t
h
m
,
i
t
perfo
rm
s a search f
o
r t
h
e g
l
obal
pea
k
o
v
e
r
t
h
e
en
tire sp
an
(0
to
V
oc
) or o
n
l
y certain
sectio
ns o
f
vo
ltag
e
ran
g
e
. Salam
et. al.
[8]
an
d Is
h
a
que
et. al.
[
10]
have
summ
arized the soft c
o
m
puting MPPT
whi
c
h incl
ude
Art
i
ficial Ne
u
r
al
Net
w
or
k
(A
N
N
)
[
16]
,
F
u
zzy
Lo
gi
c
[17] and Pa
rticle Swarm
Optimizati
on (PSO) [18],[19],
Diffe
rential Evol
ution (DE) [20].
More recentl
y the
Ant
C
o
l
o
ny
O
p
t
i
m
i
zat
i
on (A
C
O
)
[2
1]
an
d
C
h
aot
i
c
Sea
r
ch
[2
2]
, C
u
c
k
oo
Searc
h
[
23]
h
a
ve al
so
bee
n
use
d
t
o
han
d
l
e
t
h
e
part
i
a
l
shadi
n
g
co
n
d
i
t
i
on.
H
o
we
v
e
r, si
nce m
a
ny
o
f
t
h
e
m
e
t
hod
p
u
rs
ue
si
m
i
l
a
r p
r
oces
ses,
o
n
l
y
one
m
e
t
hod,
nam
e
l
y
PSO
i
s
co
nsi
d
ere
d
.
In
PSO, a
num
b
er of
parti
c
les roam
within
t
h
e searc
h
-space
accordi
ng t
o
thei
r position a
nd
m
ovem
e
nt
vel
o
ci
t
y
[2
4]
. T
h
e
i
d
ea i
s
de
pi
ct
ed
by
Fi
g
u
re
4
;
t
h
e p
o
si
t
i
on
of eac
h
part
i
c
l
e
i
s
det
e
rm
i
n
ed
by
i
t
s
o
w
n
b
e
st
po
sitio
n and
t
h
e
g
l
ob
al b
e
st po
sition
s
.
Th
e po
sition
o
f
th
e i
n
d
i
v
i
du
al p
a
rticle is giv
e
n
b
y
11
kk
k
ii
i
x
xv
(2
)
whe
r
e
v
i
corresp
ond
s t
o
th
e v
e
lo
city co
m
p
o
n
en
t.
It is calcu
l
a
ted
b
y
u
s
i
n
g th
e
fo
llowing
relatio
n
s
h
i
p
:
1
11
2
2
()
(
)
kk
k
k
ii
b
e
s
t
i
b
e
s
t
i
vw
v
c
r
P
x
c
r
G
x
(3
)
I
n
(3
),
w
is th
e in
ertia weig
ht,
c
1
and
c
2
are
the acceleration c
onsta
nts,
while
P
best
and
G
best
are the
pers
o
n
al
an
d g
l
obal
best
posi
t
i
ons
, res
p
ect
i
v
el
y
.
To st
art
t
h
e opt
i
m
i
z
at
i
on pr
ocess
,
a vect
or
of
dut
y
cy
cl
es are
in
itialized
an
d
th
e algo
rith
m
tran
sm
its th
e du
ty cycles to
t
h
e
p
o
wer con
v
erter. Th
es
e d
u
ty
cycles
(rep
r
esen
ted
by
x
i
in
(2
) serv
e as th
e in
itia
l p
a
rticles in
th
e first iteratio
n
.
All p
a
rticles
are h
ead
i
n
g
to
ward
s th
eir lo
cal b
e
st
p
o
s
ition
P
best
.
Am
ong t
h
es
e p
a
rt
i
c
l
e
s, one
o
f
t
h
em
i
s
t
h
e gl
obal
best
G
best
.
It
gi
ves t
h
e be
st
fi
t
n
ess val
u
e
.
Aft
e
r
calcu
latin
g
th
e v
e
lo
city, wh
ich
serv
es as a pertu
r
b
a
tion
to
t
h
e vo
ltag
e
, a
new po
sitio
n
o
f
th
e vo
ltag
e
is fo
und
.
Thr
o
ug
h s
u
cce
ssi
ve i
t
e
rat
i
on
al
l
part
i
c
l
e
s
m
ove t
o
wa
rds
gl
obal
be
st
posi
t
i
on.
As t
h
e pa
rt
i
c
l
e
s appr
oac
h
t
h
e
M
P
P
,
th
e
y
g
e
t c
l
o
s
er
t
o
t
h
e
G
best
po
sition
.
Co
rrespon
d
i
n
g
ly, th
e
P
best
f
a
cto
r
and
G
best
fa
ct
or i
n
vel
o
ci
t
y
t
e
rm
m
oves towards zero. E
v
ent
u
ally a zero velocity is
achi
eve
d
an
d t
h
e vol
t
a
ge p
o
si
t
i
on rem
a
i
n
s
alm
o
st
unc
ha
nge
d. Under this c
o
ndition, th
e
PV syst
e
m
reaches
at
MPP.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l.
6
,
No
.
4
,
D
ecem
b
er
2
015
:
71
2 – 722
71
6
Fi
gu
re
4.
M
o
v
e
m
e
nt
of
Part
i
c
l
e
s i
n
P
S
O
To c
o
m
p
are t
h
e pe
rf
orm
a
nce
o
f
t
h
e c
o
n
v
e
n
t
i
onal
P
&
O
an
d P
S
O
du
ri
n
g
part
i
a
l shadi
n
g, a scenario
in
vo
lv
ing fo
urn
series-co
n
n
e
cted
PV m
o
du
les, lab
e
lled as
A, B
,
C
a
n
d
D
i
s
si
m
u
l
a
t
e
d. The s
p
eci
fi
cat
i
ons
of
the m
odules at the Standa
rd Test Condition
1
(STC
)
of t
h
e
are gi
ve
n i
n
T
a
bl
e 1. T
h
e si
m
u
l
a
t
i
on i
s
car
ri
ed o
u
t
using the sim
u
lator develope
d by [2
5], which utilizes the two-diode m
o
del for the sol
a
r cel. Initially each
m
odule receives an
uni
form
irra
diance
of
1000W/m
2
. Co
n
s
eq
u
e
n
tly, th
ere ex
ists on
ly one MPP at 44
0
W, as
sho
w
n by
C
u
r
v
e 1 o
f
Fi
g
u
re
5(a)
. Aft
e
r a lapse o
f
o
n
e secon
d
, m
odul
es
A, B
,
C
and D are i
rra
di
at
ed wi
t
h
1
000
W
/
m
2
, 800
W/
m
2
, 500
W
/
m
2
and 30
0
W
/
m
2
resp
ectiv
ely. Du
e to
these p
a
rtial sh
ad
ing
s
, m
u
ltip
le p
eaks
are
g
e
n
e
rated in
th
e
P
–
V
curv
e, as sho
w
n
by Cu
rv
e
2
of
t
h
e sam
e
f
i
gu
r
e
. Th
ese ar
e
115
W
,
220
W
,
2
80
W
an
d
2
4
0
W
.
As th
e o
p
e
rati
ng
po
in
t sh
ifts fro
m
Cu
rv
e 1
to
Cu
rv
e 2
,
the P&O alg
o
rith
m
wil
l
cl
i
m
b
to
th
e
nearest
peak, i.e. 240
W as
directed
by the a
r
rows i
n
Figure 5(a
)
. Clearly,
th
is p
e
ak
is local. Howev
e
r,
in
th
e
case of PSO,
whe
n
pa
rtial shadi
n
g is
detected, the
algorith
m
begins the
searc
h
for the global
pea
k
. Aft
e
r
successi
ve i
t
e
r
a
t
i
ons, t
h
e
gl
o
b
al
M
PP
(i
.e.
28
0
W)
i
s
t
r
ac
ked
,
as
di
rect
e
d
by
t
h
e
part
i
c
l
e
s
m
ovem
e
nt
s (u
si
n
g
t
h
e di
rect
i
o
n
of a
r
r
o
w) i
n
Fi
gu
re
5(
b)
. T
h
e di
ffe
rence
bet
w
ee
n t
h
e
g
l
obal
a
nd l
o
ca
l
peak i
s
4
0
W,
or
approxim
a
tely 14%
of t
h
e peak power
.
In the context of PV system
, suc
h
powe
r loss is consi
d
ere
d
ve
ry
si
gni
fi
ca
nt
. Th
e out
p
u
t
p
o
w
er
t
i
m
i
ng di
ag
ra
m
s
for b
o
t
h
ca
ses are sh
ow
n
i
n
Fi
gu
re 6
.
A
s
can be o
b
se
r
v
ed
, at
t
h
e poi
nt
of
pa
rt
i
a
l
shadi
n
g o
ccur
r
en
ce
(at one second), the P&O
quic
k
ly ge
t trappe
d
at the local peak, i.e.
240
W. On t
h
e
othe
r
ha
nd
PSO s
u
ccess
f
ully tracks
the
gl
obal MPP at
280 W. This res
u
lt
is
consiste
nt
with
th
e ob
ser
v
ati
o
n of
th
e
P
–
V
c
u
rve in Figure
5(a)
an
d (b
),
r
e
spectiv
ely.
Tab
l
e
1
.
M
o
dule sp
ecifications at Stand
a
rd
Test Co
nd
itio
n (STC)
Para
m
e
ters
Sy
m
b
o
l
Valu
e
Po
wer at
MPP
P
M
PP
110 W
Voltage at M
PP
V
M
PP
16.
7 V
Cu
rren
t
at
MPP
I
M
PP
6.
6 A
Open circuit volta
ge
V
O
C
20.
7
V
Short circuit curre
n
t
I
S
C
7.
5
A
(a)
(b
)
Fi
gu
re 5.
P
–
V
cu
rv
e for
MPP track
ing
un
d
e
r p
a
rtial
sh
ad
i
n
g
(a)
Conv
en
tion
a
l
P&O ( b
)
PSO
1
Standar
d
T
e
st
Co
ndition
(STC): Ir
radian
ce: 1000
w/m
2
, Tem
p
era
t
ure= 25
o
C,
P
r
e
s
su
r
e
=
1
A
T
M.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Pa
rtia
l
S
had
ing
in Bu
ild
ing
In
teg
r
a
t
ed PV
System: C
a
u
s
es, Effects a
nd Mitig
a
tin
g .... (Zain
a
l
S
a
l
a
m
)
71
7
Fi
gu
re
6.
M
PP
t
r
acki
n
g
by
P
S
O
a
n
d
P&
O
u
n
d
er
pa
rt
i
a
l
sha
d
i
n
g
4.
3.
Energy Recovery Circuit
Alth
oug
h th
e
m
o
d
i
fied
P&O and
so
ft co
m
p
u
tin
g MPPT are ab
le to
t
r
ack
th
e g
l
o
b
a
l
p
e
ak
, it m
u
st b
e
not
e
d
t
h
at
as l
o
ng as t
h
e s
h
ade
d
m
odul
e i
s
be
i
ng s
h
o
r
t
-
ci
rc
u
i
t
e
d by
t
h
e by
p
a
ss di
o
d
e, t
h
at
part
i
c
ul
a
r
m
odul
e i
s
t
o
t
a
l
l
y
unusa
b
l
e
. Thi
s
i
s
des
p
i
t
e
t
h
e fact
t
h
at
t
h
e m
odul
e rec
e
i
v
es cert
a
i
n
a
m
ount
of
ene
r
gy
w
h
i
l
e
i
t
i
s
shade
d
.
To m
a
ke t
h
e s
h
ade
d
m
odul
e
usa
b
l
e
, t
h
e e
n
e
r
gy
rec
o
very
c
i
rcui
t
i
s
pr
o
pos
ed [
2
6]
,[
27]
. T
h
e i
d
ea i
s
t
o
c
a
pt
u
r
e
the energy from
the non-s
ha
ded m
odules a
n
d the
n
s
h
a
r
e i
t
with t
h
e s
h
a
d
ed m
odul
e
until the
power delivere
d
by each m
odule in the string is balanced. The circuit can
be easily retrofitted to the central inve
rter s
y
stem
with m
i
nim
u
m changes i
n
the
electrical
wi
ri
ngs
. Ty
pi
cal
l
y
bi
di
rect
i
o
nal
b
u
ck
-
b
o
o
st
,
fl
y
b
ack o
r
c
uk c
o
n
v
ert
e
r
is
u
s
ed
. Th
ere are
sev
e
ral v
a
riatio
n
s
[28
]
, bu
t
th
e co
n
c
ep
t
remain
s as illu
strated
in
Figu
re
7
(
a).
The
ope
rat
i
o
n
of t
h
e e
n
er
gy
reco
very
ci
r
c
ui
t
i
s
sho
w
n i
n
Fi
gu
re
7(
b)
. T
h
e
basi
c u
n
i
t
c
o
m
p
ri
ses o
f
fo
ur
m
odul
es,
w
h
i
c
h
i
s
di
v
i
ded
i
n
t
o
t
w
o
g
r
o
u
p
s.
G
r
o
u
p
1 i
n
v
o
l
v
e
s
PV
1 a
n
d
P
V
2
,
t
o
get
h
e
r
wi
t
h
t
h
ei
r
corres
ponding powe
r electroni
cs circuit, com
p
rises of
S
1
,
D
1
,
L
1
,
C
1
,
S
2
,
D
2
and
C
2
. Gr
ou
p 2 i
n
cl
u
d
es
PV3
,
PV4
with
S
3
,
S
4
,
D
3
,
D
4
,
L
2
,
C
3
and
C
4
. In
o
r
d
e
r to
conn
ect
th
e two
group
s to
geth
er, t
h
e cap
acito
r
C
5
is
u
s
ed
.
Ass
u
m
i
ng t
h
at
PV1 i
s
sha
d
e
d
an
d PV
2 re
cei
ves ful
l
i
r
r
a
di
at
i
on, P
V
2
del
i
v
ers hi
gh
er cur
r
e
n
t
t
h
an PV
1.
Ho
we
ver
,
si
nc
e t
h
e
m
odul
es
are co
nnect
e
d
i
n
seri
es, t
h
e st
r
i
ng c
u
r
r
ent
wi
l
l
be l
i
m
i
t
e
d t
o
t
h
e am
ount
del
i
vere
d
by
PV
1.
Du
ri
n
g
pa
rt
i
a
l
shadi
n
g
,
pa
rt
of t
h
e
cur
r
ent
f
r
om
PV
2 i
s
di
ve
rt
ed t
o
t
h
e e
n
er
g
y
recove
ry
ci
r
c
ui
t
(by
tu
rn
ing
S
2
o
n
)
and t
h
e ene
r
gy
i
s
st
ored t
e
m
pora
r
i
l
y
i
n
L
1
. B
y
doi
n
g
s
o
, t
h
e
st
ri
ng c
u
r
r
e
n
t
can be m
a
i
n
t
a
ined at
th
e lev
e
l g
e
n
e
rated
b
y
PV1 an
d
h
e
n
ce th
ere
is n
o
n
e
ed
fo
r
PV1
to
b
e
b
ypassed
.
As a
resu
lt, PV1
is still ab
le to
actively producing
powe
r (al
b
eit in lesser a
m
ount, de
pe
ndi
ng
on the s
h
a
d
ing c
o
ndition)
because its voltage is
not zero. Mea
n
while, the energy stored i
n
L
1
will b
e
released
b
a
ck
to
t
h
e ou
tpu
t
v
i
a
D
1
(
b
y turn
ing o
f
f
S
2
).
Thus,
no P
V
powe
r is
waste
d
exce
pt for the
n
on-idealities in active
and t
h
e passi
ve c
o
m
pone
nts.
T
r
i
gge
r
Ci
r
c
u
i
t
T
r
i
gge
r
Ci
r
c
u
i
t
En
e
r
g
y
Re
co
v
e
r
y
C
i
r
c
ui
t
In
v
e
rt
e
r
wi
t
h
MP
P
T
U
t
ilit
y G
r
id
P
r
op
os
e
d
C
i
r
c
u
i
t
PV1
PV2
By
pa
s
s
Di
o
d
e
S
1
S
2
L
1
C
1
C
2
D
1
D
2
Lo
a
d
C
3
C
4
L
2
S
3
S
4
D
3
D
4
C
5
B
oos
t
co
n
v
e
r
t
e
r
Lo
ad Se
ct
i
o
n
V
pv
I
dc
Gr
o
u
p 1
Gr
o
u
p
2
PV
1
PV
2
PV
3
PV
4
(a)
(b
)
Fi
gu
re
7.
(a
) T
h
e
ove
ral
l
bl
oc
k
di
ag
ram
of t
h
e ene
r
gy
reco
v
e
ry
ci
rcui
t
.
(
b
)
Det
a
i
l
of
t
h
e e
n
er
gy
rec
o
very
circu
it
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l.
6
,
No
.
4
,
D
ecem
b
er
2
015
:
71
2 – 722
71
8
4.
4.
Micro-inver
te
r
The conce
p
t is totally differe
nt from
the energy r
ecove
ry circuit because it
doe
s not utilised a central
inve
rter [29].
Instead, sm
all
dedicated invert
ers are
directly connected
t
o
t
h
e ac g
r
i
d
[
3
0]
, as sh
ow
n i
n
F
i
gu
re
8
.
Th
is con
f
i
g
u
r
ation
is v
e
ry attractiv
e fo
r
lo
w
v
o
ltag
e
g
r
id
, wh
ich
p
a
rti
c
u
l
arly su
its the resid
e
n
tial B
I
PV.
Each i
nve
rt
er i
s
eq
ui
p
p
ed
wi
t
h
i
t
s
o
w
n M
P
P
T
co
nt
r
o
l
l
e
r;
t
hus
t
h
e
o
u
t
p
ut
fr
om
t
h
e i
nvert
er ca
n be
co
nt
r
o
l
l
e
d
inde
pende
n
tly. For e
x
am
ple, if PV1
is
shaded,
other modules a
r
e not
a
ffecte
d
by because they are not
connected as
one
string. De
spite its
attractiveness
,
during norm
al
opera
tion, t
h
e i
nve
rter c
o
nducts t
h
e
ful
l
lo
ad
curren
t
, resu
ltin
g
i
n
h
i
gh
co
ndu
ctio
n and
switch
i
ng
losses. Fu
rt
h
e
rm
o
r
e, th
e reliab
ility o
f
th
e electro
n
i
cs
com
pone
nts is
re
duce
d
due t
o
thei
r e
x
pos
u
re to ha
rs
h e
n
vironm
ent conditions
,
pa
rticularly high
ope
rating
t
e
m
p
erat
ure
.
F
o
r a l
a
rge sy
st
em
wi
t
h
hi
g
h
n
u
m
b
er o
f
m
odul
es, t
h
e
cost
o
f
i
n
vert
ers,
wi
ri
n
g
a
n
d t
h
e
com
p
lexity of t
h
e system
increases ve
ry ra
pi
dly.
U
til
ity
g
r
id
U
t
i
l
i
t
y gr
i
d
(b)
M
i
cr
o i
nver
t
er
DC-
B
u
s
(a
)
DC
-
D
C
PV
2
DC
-
D
C
PV
3
DC
-
D
C
PV
n
DC
-
D
C
DC-
A
C
In
v
e
rt
e
r
PV
1
DC-
A
C
PV
2
DC-
A
C
PV
3
DC-
A
C
PV
n
DC-
A
C
PV
1
Fi
gu
re
8.
M
i
cr
o i
n
ve
rt
er i
n
t
e
r
c
on
nect
i
o
n
5.
CASE ST
UDY
5.
1.
BI
P
V
Sy
st
em
s
e
t
-
up
A si
m
u
l
a
ti
on o
f
a B
I
PV sy
st
e
m
i
s
carri
ed o
u
t
usi
ng s
o
ft
war
e
devel
ope
d
by
[2
5]
. The set
-
up i
s
a t
w
o-
st
ri
ng
P
V
ar
ra
y
com
p
ri
ses o
f
1
6
m
o
d
u
l
e
s,
co
nfi
g
u
r
ed
b
y
ei
ght
m
odul
es pe
r st
ri
ng
.
The m
o
d
u
l
e
h
a
s t
h
e
fo
l
o
owing
STC
sp
ecifications:
P
MPP
= 24
0 W
,
V
MPP
=
19
.0
8 V,
I
MPP
= 8.
22
A
,
V
OC
=
37.25 V and
I
SC
= 8.
28
A
.
Based
on
th
ese d
a
ta, th
e th
eoretical o
u
t
p
u
t
po
wer of th
e syste
m
is
3
.
80
kWp
(i.e. 16
×
2
40 W
)
. To
q
u
an
tify
th
e effect of partial sh
ad
ing
,
six
arb
itrarily sh
ad
in
g patte
rns are im
pos
ed, as shown
in Table 2; they are
lab
e
lled
as co
n
d
ition
s
PS1
th
rou
g
h
PS5
.
In
th
e first co
l
u
m
n
, th
e “n
o
-
sh
ad
ing
”
is g
i
v
e
n
as th
e b
e
nch
m
ark
.
Fu
rt
h
e
rm
o
r
e, fo
r sim
p
licit
y,
sev
e
ral log
i
cal assu
m
p
tio
n
s
are m
a
de:
1) onl
y
fo
ur i
rra
di
at
i
on val
u
es, i
.
e. 1.
0
,
0.
7, 0
.
5 a
nd
0.
25
k
W
/
m
2
are
u
s
ed
to
represen
t th
e d
i
fferen
t
lev
e
ls
o
f
sh
adin
g
in
tensities, 2
)
th
e sh
ad
ings are
assu
m
e
d
to
b
e
con
s
isten
t
(in term
s o
f
in
ten
s
ity and
tim
e
p
r
ofile),
3
)
th
e
p
o
wer electron
ics circu
its are
assum
e
d t
o
be 10
0% ef
fi
ci
ent
and 4
)
t
h
e m
o
dul
e t
e
m
p
erat
ure i
s
uni
f
o
rm
l
y
t
a
ken t
o
be co
nst
a
nt
, i
.
e. at
4
0
o
C.
As an exam
ple, Figure
9 s
h
ows t
h
e m
odul
e-invert
er c
onnection
for c
o
ndition PS
5.
T
h
e corre
spondi
ng
P-V
cur
v
e i
s
sh
ow
n i
n
Fi
g
u
re
9(
b)
. It
ex
hi
bi
t
s
t
h
ree
peak
s, na
m
e
l
y
P1, P2,
P3,
wi
t
h
P
3
be
i
ng t
h
e
gl
o
b
al
.
On t
h
e
ot
he
r ha
n
d
,
wh
en t
h
e e
n
e
r
gy
r
ecove
ry
m
e
t
hod i
s
ap
pl
i
e
d
,
t
h
e m
u
l
t
i
p
l
e
peak cu
r
v
e i
s
t
r
a
n
sfo
r
m
e
d i
n
t
o
a
si
ngl
e
peak. Clearly, it has
an advantage
beca
us
e the
pea
k
po
wer
i
s
hi
g
h
er.
Fo
r t
h
e m
i
cro-i
nve
rt
er,
t
h
e
di
rec
t
connection of
all sixteen inve
rters to t
h
e gri
d
is m
a
de si
m
i
l
a
r to
Fi
g
u
re 9.
Th
e to
tal power av
ailab
l
e is t
h
e su
m
of
p
o
w
e
r
ha
rve
s
t
e
d
by
t
h
e i
ndi
vi
d
u
al
m
i
cro-i
nve
rt
er
fr
om
i
t
s res
p
ect
i
v
e m
o
d
u
l
e
.
5.
2.
Results
The
resul
t
s
a
r
e
sh
ow
n i
n
Ta
bl
e 3.
Fo
r t
h
e
n
o
n
-s
ha
di
n
g
c
o
n
d
i
t
i
on,
t
h
e
ge
n
e
rat
e
d
po
we
r i
s
l
o
we
r t
h
a
n
th
e th
eo
r
e
tical
v
a
lu
e, i.e.
3
.
16 kW
(
i
n
s
tead of
3.80
kW
). Th
is is ex
p
ected as th
e m
o
dule
s are s
u
bje
c
ted to a
higher tem
p
erature (40
o
C),
wh
ile th
e sp
ecifi
catio
n
in
T
ABLE
II is at STC (2
5
o
C). F
u
rthe
rm
ore, fo
r sim
p
licity
it is assu
m
e
d that the powe
r
harvested
by each m
i
tiga
tion
m
e
thod with t
h
e abse
nce of
sha
d
ing (be
n
c
h
m
a
rk
case) i
s
eq
ual
t
o
t
h
e c
o
n
v
e
n
t
i
onal
P
&
O
.
F
o
r
t
h
e m
odi
fi
ed
P&O as
wel
l
as t
h
e so
ft
com
put
i
n
g m
e
t
hod,
si
nce
th
e
P-V
c
u
rve
exhi
bits a
uni
que MPP, t
h
e al
go
rith
m
sh
ou
ld track th
e sam
e
v
a
lu
e.
The
fi
rst
r
o
w
o
f
Ta
bl
e 3
sh
o
w
s t
h
e
har
v
est
e
d
p
o
we
r
usi
n
g t
h
e c
o
n
v
e
n
t
i
ona
l
P&O
.
De
pe
n
d
i
n
g
on t
h
e
lo
catio
n
o
f
MPP p
r
i
o
r t
o
th
e
p
a
rtial sh
ad
ing o
ccurren
ce,
t
h
e tran
sitio
n
of
th
e op
erating
po
in
t fro
m
th
e sin
g
l
e
p
eak
t
o
th
e m
u
lti p
eak
curv
e resu
lts in
o
n
e
t
h
e fo
llowing
p
o
ssib
ilities: th
e P&O al
g
o
rithm fin
d
s
itself l
o
cated
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Pa
rtia
l
S
had
ing
in Bu
ild
ing
In
teg
r
a
t
ed PV
System: C
a
u
s
es, Effects a
nd Mitig
a
tin
g .... (Zain
a
l
S
a
l
a
m
)
71
9
at the 1)
globa
l
peak,
2) t
h
e lowest
peak, or 3) at one
of t
h
e pea
k
in-b
et
ween t
h
e gl
obal and the l
o
west. For
ex
am
p
l
e, in
Fig
u
re 9(b
)
, it co
u
l
d
b
e
P1
, P2 o
r
P3
, resp
ect
iv
ely. In
th
is stu
d
y
, th
e selected
op
erating
po
in
t is
the lowe
st am
o
ng t
h
e availabl
e peaks. Conse
que
ntly, fo
r the co
nv
en
tio
n
a
l
P&O, th
e resu
l
t
s sh
own
in
th
e tab
l
e
are consi
d
ere
d
as the
worst
case. If
P
no_shading
i
s
t
h
e po
w
e
r ge
ne
rat
e
d
b
y
t
h
e PV
sy
st
em
whe
n
s
h
ad
i
ng i
s
ab
sen
t
, wh
ile
P
with_shade
i
s
t
h
e po
wer
ge
nerat
e
d w
h
e
n
a pa
rt
i
c
ul
ar sha
d
i
n
g
pat
t
e
rn i
s
i
m
posed, t
h
en t
h
e p
o
w
e
r
l
o
ss i
s
c
o
m
put
ed
by
Th
e lo
ss du
e
to
p
a
rtial sh
ad
ing
varies acco
rd
ing
to
t
h
e sh
ad
ing
conditio
n
s
and
t
h
e
m
itig
atio
n
approach.
For instance, with
the abse
nce of
m
i
tigation, for conditio
n PS
1 (with only two m
odules s
h
ade
d
),
t
h
e po
wer l
o
ss i
s
appr
o
x
i
m
at
el
y
17%. I
n
t
h
e case of PS
5 (
w
i
t
h
10 m
odul
es
shade
d
)
,
t
h
e l
o
ss i
n
crea
ses t
o
o
v
er
73
%. H
o
we
ve
r
,
by
obse
r
vi
n
g
t
h
e t
a
bl
e, t
h
ere i
s
no st
rai
g
ht
f
o
r
w
ar
d m
a
them
ati
cal
rel
a
ti
ons
hi
p
bet
w
e
e
n t
h
e
sha
d
i
n
g pat
t
e
r
n
an
d t
h
e out
put
p
o
w
er. F
u
rt
herm
ore, t
h
e
l
o
sses coul
d not
be easi
l
y
qua
nt
i
f
i
e
d
due
t
o
t
h
e
in
fin
ite po
ssi
bilit
ies fo
r t
h
e
sh
ad
ing
p
a
ttern
.
Desp
ite th
is fact, th
e effectiv
en
ess
o
f
t
h
e m
i
tig
atio
n
can
b
e
gene
ral
i
zed i
n
t
h
e f
o
l
l
o
wi
n
g
or
der:
t
h
e
best
effi
ci
ency
i
s
obt
ai
ne
d
usi
n
g
t
h
e m
i
cro-i
n
v
e
rt
er,
fol
l
owe
d
by
t
h
e
ener
gy
rec
ove
r
y
ci
rcui
t
,
t
h
en
t
h
e m
odi
fi
ed P
&
O o
r
so
ft
co
m
put
i
ng. The
per
f
o
r
m
a
nce of m
i
cro-i
n
ve
rt
er i
s
t
o
be expected
because each inverter is
able
to harvest the
energy from
the indi
vidual m
odule—eve
n
if the
m
odul
e i
s
sha
d
ed.
Tabl
e
2. T
h
e
s
h
adi
n
g
pat
t
e
r
n
s
im
pose
d
on
t
h
e m
odul
es.
N
o
t
e
:
Fo
r
benc
hm
arki
ng
,
no
sha
d
i
ng i
s
i
m
posed.
M
odule
Nu
m
b
e
r
/
Conditio
n
No Shading
(bench-
m
a
r
k
ing)
Irradiation for sha
d
ing conditi
ons
( k
W
/
m
2
)
PS1
PS2 PS3
PS4 PS5
PV1,
PV9
1.
00
1.
00
1.
00
0.
25
0.
70
0.
25
PV2,
PV10
1.
00
1.
00
1.
00
0.
25
1.
00
0.
25
PV3,
PV11
1.
00
1.
00
1.
00
0.
50
1.
00
0.
25
PV4,
PV12
1.
00
0.
70
0.
25
0.
70
1.
00
1.
00
PV5,
PV13
1.
00
1.
00
0.
25
1.
00
0.
70
1.
00
PV6,
PV14
1.
00
1.
00
1.
00
1.
00
0.
50
0.
50
PV7,
PV15
1.
00
1.
00
1.
00
1.
00
0.
50
0.
50
PV8,
PV16
1.
00
1.
00
1.
00
1.
00
1.
00
1.
00
0
50
10
0
15
0
20
0
25
0
30
0
0
20
0
40
0
60
0
80
0
10
00
12
00
14
00
16
00
18
00
Vo
lt
age (V)
P
o
wer(W
)
P1
11
80
.8
W
84
.0
1V
P2
10
91
W
15
0.
7V
P3
82
1.
6W
24
1.
2V
P4
16
43
.2
W
21
5.
5V
(a)
(b
)
Fi
gu
re
9.
(a
) S
h
adi
n
g
pat
t
e
r
n
PS5
, (
b
) It
s
co
r
r
esp
o
ndi
ng
P-V
curve
.
Re
d trace: for the M
PPT al
gortith t
o
track t
h
e
globa
l
peak. Blue
tra
ce: curve
re
s
u
l
t
i
ng
fr
om
t
h
e en
ergy
recove
ry circuit.
PV
1
PV
2
PV
3
PV
4
PV
5
PV
6
PV
7
PV
8
PV
9
PV
10
PV
11
PV
12
PV
13
PV
14
PV
15
PV
16
+
Utilitt
y
Gr
id
In
v
e
r
t
e
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l.
6
,
No
.
4
,
D
ecem
b
er
2
015
:
71
2 – 722
72
0
Tabl
e
3. T
h
e
p
e
rf
orm
a
nce o
f
vari
ous
m
i
t
i
g
at
i
on m
e
t
hod
s
u
nde
r
di
f
f
ere
n
t
s
h
adi
n
g
pat
t
e
r
n
s
.
T
h
e
po
we
r gene
rat
e
d wi
t
h
o
u
t
s
h
di
ng
i
s
3.
16
k
W
.
M
e
thods/
Shading
Pattern
Shading Pattern: P
S
1
Shading Pattern:
PS2
Shading Pattern: P
S
3
Shading Pattern: P
S
4
Shading Pattern:
PS5
Pwr
Gen.
(kW)
Pwr Loss
(Pl
oss
)
Pwr
Gen.
(kW)
Pwr Loss
(Pl
oss
)
Pwr
Gen.
(kW)
Pwr Loss
(Pl
oss
)
P
wr Gen.
(kW)
Pwr Loss
(Pl
oss
)
Pwr
Gen.
(kW)
Pwr Loss
(Pl
oss
)
kW
%
kW
%
kW
%
kW
%
kW
%
Conventio
n
al P&O
MPP
2.
61
0.
54
17.
2
0.
91
2.
25
71.
2
0.
89
2.
27
71.
8
1.
18
1.
98
62.
6
0.
82
2.
33
73.
4
M
odified
P&O/Sof
t
Co
m
puting
MPPT
2.
75
0.
41
12.
8
2.
36
0.
80
25.
2
1.
59
1.
56
49.
5
1.
72
1.
43
45.
5
1.
18
1.
97
62.
6
En
erg
y
Recover
y
Circuit
2.
91
0.
25
7.
8
2.
41
0.
74
23.
5
2.
05
1.
10
35.
0
2.
27
0.
88
28.
0
1.
64
1.
51
47.
9
Micro
-
inver
t
er
3.
02
0.
13
4.
3
2.
52
0.
64
20.
1
2.
19
0.
97
30.
7
2.
48
0.
68
21.
5
1.
79
1.
36
43.
2
6.
CO
NCL
USI
O
N
Hav
i
n
g
do
n
e
all th
ese an
alysis, it sh
ou
ld b
e
rea
lized t
h
at the e
n
ergy reco
v
e
ry circu
it requ
ires
ad
d
ition
a
l h
a
rd
ware to
b
e
fitted
in
to
th
e ex
istin
g
PV syste
m
. Cu
rren
tly, th
e real co
st
o
f
th
is h
a
rdware is
un
k
n
o
w
n
beca
use m
o
st
of
t
h
e pr
ot
ot
y
p
es a
r
e o
n
l
y
avai
l
a
b
l
e in
t
h
e research
lab
s
. Desp
ite th
is fact, lo
ok
ing
in
to
th
e
p
r
o
s
p
e
ct, it is lik
ely t
h
at th
ese id
eas wou
l
d
b
e
translated
in
to
commercial p
r
o
d
u
c
t soon
. Fo
r t
h
e case
of m
i
cro-i
n
ve
r
t
er, t
h
e t
o
t
a
l
cost
f
o
r t
h
e si
xt
een i
n
vert
e
r
u
n
i
t
s
(a
nd t
h
ei
r
B
O
S c
o
m
pone
nt
s) m
i
ght
be
hi
g
h
e
r
than one cent
r
al inverter. Furtherm
or
e, th
e
m
i
cro
-
in
v
e
rter is relativ
ely
new, while the central inve
rter has
been
dom
i
n
at
ing
t
h
e
m
a
rket
fo
r s
o
m
e
t
i
m
e
. H
o
we
ve
r,
as
wi
t
h
ot
he
r t
e
c
h
nol
ogy
,
as
t
h
e
vol
um
e of t
h
e
m
i
cro-
in
v
e
rter grows, th
e p
r
ice will d
r
op
. Mo
reov
er, it h
a
s to
b
e
realized
th
at PV syste
m
is
a l
o
ng
term
in
v
e
st
m
e
n
t
.
Thu
s
th
e
fin
a
l
d
ecision
to
install ad
d
itio
n
a
l
h
a
rdware can
b
e
ju
stified
b
y
tak
i
n
g
in
to
acco
un
t th
e l
o
ng ter
m
p
r
o
f
itab
ility.
ACKNOWLE
DGE
M
ENTS
The aut
h
ors
would like to thank
Universiti Teknol
ogi
Malaysia an
d the Ministry of Highe
r
Edu
catio
n, Malaysia fo
r pro
v
i
d
i
ng
th
e facilities an
d
fin
a
n
c
ial suppo
rt (Research
Un
iv
ersity G
r
ant N
o
.
2
509
.0
6H7
8
)
t
o
co
ndu
ct th
is
r
e
sear
ch
.
REFERE
NC
ES
[1]
DHW. Li, SKH. Chow, EWM.
L
ee, "An analy
s
is
of a medium size grid
-
c
onnected
building
integrated photovo
ltaic
(BIP
V) s
y
s
t
em
u
s
ing m
eas
ured d
a
ta",
En
ergy and
Buildings,
vol.
60, pp
. 383-387
, 2013.
[2]
CR. Sulliv
an, J. Awerbuch, AM. Latham
, "D
ecrease
in
phot
ovoltai
c
power
output from
rip
p
le:
Sim
p
le gen
e
ral
calculation and
effect of partial shading", Applied
Power Electronics Conferen
ce and Exposition (APEC), 201
1
Twenty
-Sixth
Annual I
EEE, pp
.
1954-1960, 201
1.
[3]
YJ. Wang, PC.
Hsu, "An investigation
on p
a
rtial shading of PV modules with
different connecti
on configuration
s
of P
V
cel
ls
",
En
ergy,
vo
l. 36, pp. 3069-3078, 201
1.
[4]
C. Peng, Y. Huang,
Z. Wu, "B
uilding-in
tegr
ated phot
ovoltaics
(BIPV) in ar
chitectur
al d
e
sign
in China",
En
ergy
and Buildings,
v
o
l/issue: 4
3
(12),
pp. 3592-3598
,
2011.
[5]
M. Yam
a
guchi,
T. Tak
a
m
o
to,
K. Araki,
N. E
k
ins-Daukes, "Multi-junc
tion I
II–V solar cells:
current sta
t
us a
nd
future po
tent
ia
l",
Solar En
ergy,
v
o
l. 79
, pp
. 78-85
, 2005
.
[6]
T. Esr
a
m, PL.
Chapman, "Co
m
parison of Photovolta
ic Arr
a
y
Max
i
mum Po
wer Point Tr
acking Techniques",
Energy Con
version, IE
EE Transactions on
,
vo
l. 2
2
, pp
. 439-449
,
2007.
[7]
J.
Ahme
d,
Z.
Sa
l
a
m,
"A
n impr
oved per
t
urb an
d observe (P&O
) maximu
m power point
track
ing (MPPT)
algorithm for
higher ef
ficiency
"
,
Applied Energy,
vol. 150, pp. 97
-108, 2015
.
[8]
Z. Salam, J. Ahmed, BS.
Merugu, "The app
l
ication of soft compu
ting methods for MPPT of PV sy
stem: A
techno
logical an
d status r
e
view",
Applied Energy,
vol. 107, pp. 13
5-148, 2013
.
[9]
J. Ahmed, Z. Salam, "A critical
ev
alu
a
tion on m
a
ximum power point tracking
methods for partial shading in PV
s
y
ste
m
s",
Renew
able and
Sustain
able
Energy Reviews,
vo
l. 47, pp. 933-953, 2015.
[10]
K. Ishaque, Z. Salam, "A review
of
maximu
m po
wer point tracki
ng techniqu
es of PV
sy
stem for uniform insolatio
n
and par
tial shad
ing condition",
Renewable and
Su
stainable Energy Rev,
vol/issue:
19(3), pp
. 475-4
88, 2013
.
[11]
N. Femia, G. Petrone
, G. Spagn
uolo, M.
Vitelli, "A techniqu
e fo
r improvi
ng P&O MPPT performances of double-
stage gr
id-conn
ected
photovoltaic s
y
s
t
ems",
Ind.
Electr. IEEE Trans.
vol. 56, pp. 4
473-4482, 2009
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Pa
rtia
l
S
had
ing
in Bu
ild
ing
In
teg
r
a
t
ed PV
System: C
a
u
s
es, Effects a
nd Mitig
a
tin
g .... (Zain
a
l
S
a
l
a
m
)
72
1
[12]
H. Pate
l,
V. Ag
a
r
wal,
"Maxim
um
Power
Point Track
ing
Sc
hem
e
for P
V
S
y
s
t
em
s
Operating Under P
a
rtial
l
y
S
h
ad
ed
Conditions",
Ind
u
strial Electronics,
IEEE Transactions on
,
vo
l. 5
5
, pp
. 1689-169
8, 2008
.
[13]
A. Kouchaki, H. Iman-Eini, B. A
s
aei
, "A new maximum power point tr
acking
strateg
y
for PV arr
a
y
s
under
unifor
m
and non-uniform
insolation
conditions",
Solar
En
ergy,
vo
l. 91, pp. 221-232, 2013.
[14]
K. Punitha, D. D
e
varaj, S.
Sakthivel, "Artificial n
e
ural n
e
twork b
a
sed modified in
cremental conductance algorithm
for m
a
xim
u
m p
o
wer point track
i
ng in photovolt
a
ic
s
y
s
t
em
under partial shading
conditions",
En
ergy,
vol. 62, pp
.
330-340, 2013
.
[15]
PP. Bonissone, "
S
oft computing:
the
converg
en
ce
of emerging r
eas
oning technolog
ies",
Soft Computing,
vo
l. 1
,
pp
.
6-18, 1997
.
[16]
AZ. Alab
edin
, E
.
E
l
-S
aadan
y,
M
.
S
a
l
a
m
a
, "M
axi
m
u
m
power
point tracking
for P
hotovo
ltaic s
y
stems using fuzzy
logic and
artificial n
e
ural networks", Power and
En
erg
y
Soc. Gen
e
ral Meeting, 20
11 IEEE, pp. 1-9
,
2011
.
[17]
S. Subiy
a
nto
,
A. Mohamed,
MA. Ha
nnan, "Intelligen
t
maximum power poi
nt tr
ack
ing for
PV s
y
stem using
Hopfield n
e
ural
network optim
i
z
e
d
fuzzy
l
ogi
c cont
rol
l
e
r",
Energ
y
and
Build
ings,
vol. 51
, pp
. 29-3
8
, 2012
.
[18]
K. Ishaque, Z. Salam, M.
Amjad, S. Mekhilef, "A
n Im
proved Particle Swarm Optimi
zation (PSO)–Based MPPT for
PV W
ith Reduc
e
d
Stead
y-St
ate
O
s
cilla
tion",
Pwr
Ele
c
t.
, I
E
E
E
T
r
ans
.
vol. 27, pp. 3
627-3638, 2012
.
[19]
K. Ishaque
,
Z.
Salam
,
"A De
term
inistic
Part
i
c
le
Swarm
Optim
ization
Maxi
m
u
m
Power Point
Track
er fo
r
Photovoltaic S
y
s
t
em Under Par
t
ial Shad
ing Cond
ition",
Ind,
E
l
ec
t
r
,, I
EEE
T
r
ans
,
vol. 60
, pp
. 3195
-3206, 2013
.
[20]
MFN. Tajuddin
,
SM. A
y
ob,
Z.
Salam,
MS. Saad, "Evolutionar
y
based maxi
mum power point tracking
techniqu
e
using differ
e
ntial evo
l
ution
algor
ithm",
En
ergy a
nd Buildings,
vo
l. 67
, pp
. 245-25
2, 2013
.
[21]
LL. Jiang
,
DL.
Maskell, JC. Patra,
"A novel an
t co
lon
y
opt
imization-based
maximum
power point track
ing f
o
r
photovoltaic s
y
stems under par
tially
shad
ed
conditions",
En
ergy a
nd Buildings,
vo
l. 58
, pp
. 227-23
6, 2013
.
[22]
Z.
Lin,
C. Yan,
G.
Ke, J.
Fangche
ng, "New Ap
proach for
MPPT Contro
l
of Ph
otovolta
ic
S
y
st
e
m
W
ith Mutativ
e-
S
cale
Dual-C
arri
er Chao
tic
S
ear
c
h
",
Pow
e
r El
ec
tronics,
I
E
EE Trans. on,
vol. 26, p
p
. 1038-1048
, 2
011.
[23]
J. Ahmed, Z. Sala
m, "A Maximum Power Point Track
ing (MPPT) for
PV s
y
stem using Cuckoo Search
with p
a
r
tial
shading c
a
pab
ilit
y",
Applied
En
ergy,
vo
l. 119, pp.
118-130, 2014
.
[24]
JJ. Soon, KS.
Low, "Photovoltaic mode
l iden
tification using particle swar
m optimization with inverse barrier
constrain
t
",
Pow
e
r
El
ectr
oni
cs
,
I
EEE Transactio
ns on,
vol. 27
, p
p
. 3975-3983
, 2
012.
[25]
K. Ishaque, Z. S
a
lam, H
.
Taheri,
S
y
afaruddin, "Modeling
and simulation
of phot
o
voltaic (PV) s
y
stem during par
t
ial
shading based
o
n
a
two-diode model",
Simul. Mo
delling
Practice
and Theory,
vol. 19, pp. 1613-16
26, 2011
.
[26]
W. Yanzhi, L.
Xue, K. Youngh
y
u
n
,
C. Naeh
y
u
ck
, M.
Pedram, "Enhancing
effici
ency
and robustness
of
a
photovoltaic po
wer s
y
stem unde
r par
tial shad
in
g", Quality
Electronic De
sign, 2
012 13th In
ternational S
y
mposium
on, pp
. 592-600
, 2012.
[27]
M. Z. Ramli, Z. Salam, "A Simple
En
erg
y
R
e
cover
y
Scheme
to Harves
t th
e
Energ
y
from Shaded Photovoltaic
Modules During
Partial Shad
ing"
,
Pow
e
r Electronics, I
E
EE Transactions on
,
vo
l.
29, pp
. 6458-64
71, 2014
.
[28]
T. Shimizu, O.
Hashimoto,
G. Kimura, "A novel high-perfo
rmance utility
-
intera
ctiv
e photovoltaic inverter s
y
stem",
Power
E
l
ec
tr
oni
cs
, IE
EE Transactions on
,
vo
l. 1
8
, pp
. 704-711
,
2003.
[29]
M
B
. Kalas
h
ani
,
M
.
F
a
rs
adi, "
N
ew S
t
ructure
for P
hot
ovoltai
c
Sy
st
em
s with Maxim
u
m
Power Point Tra
c
ki
ng
Abilit
y",
International
Journal o
f
Power
Elect
ron
i
cs and Drive S
y
stems (
I
JPEDS)
,
vol. 4, pp. 489-
498, 2014
.
[30]
R. Orduz, J
.
Solórzano, M. Á.
Egido,
E. Roman, "Analy
tical stu
d
y
and ev
al
uatio
n results of pow
er optimizers for
distributed pow
er conditioni
ng in
photovoltaic array
s
",
Progress in Photovolta
ics:
Research an
d Applica
tions,
2011.
BIOGRAP
HI
ES OF
AUTH
ORS
Z
a
inal Salam
re
ceiv
e
d the B.S
c
.
degree in
ele
c
tr
onics
engine
erin
g from
the Calif
ornia S
t
at
e
Universit
y
, Chi
c
o, CA, USA, the M.E.E. degr
ee i
n
elect
ric
a
l engi
neering from
the
Universiti
Teknologi Malay
s
ia (UTM)
,
Johor Bahru,
Malay
s
ia, and
th
e Ph.D. degree in power
electronics, from the University
o
f
Birmi
ngham, Birmingham, U.K
., in 1985,1989
, and 1997,
res
p
ect
ivel
y.
He
is
curr
entl
y
the
P
r
ofes
s
o
r in power el
ectr
onics
and ren
e
wabl
e e
n
erg
y
a
t
t
h
e
Facult
y
of El
ect
rica
l Engine
erin
g UTM.. Since
2011, he has bee
n
the Editor of I
EEE Tr
ans.
Sust. Energ
y
. H
e
repr
esents th
e
countr
y
as th
e expert for
the Intern
ational Energ
y
Agen
cy
(IEA) PV Powe
r Sy
stems Task
13
Working Gr
oup, which focuses on the reliability
an
d
perform
ance
of
t
h
e P
V
power s
y
s
t
em
.
Mohd. Zulk
ifli
Ramli was born
in Terengganu
,
Ma
lay
s
ia in 197
8. He r
e
ceived
the B.Sc.
and
M.Eng. d
e
grees
from
the Univer
siti T
e
knologi
Ma
lay
s
ia (UTM),
Johor
Bahru
,
Malay
s
ia,
in
2000 and 2004,r
e
spectively
,
all
in electrical eng
i
n
eering
.
He is
currently
working
toward th
e
Ph.D. degr
ee
at
UTM in th
e ar
ea of pho
tovoltaic
. He is
curren
t
ly
a Sen
i
or Lecturer at th
e
Universiti T
e
kni
kalMel
aka, Me
l
a
ka, Mal
a
y
s
ia
. His prim
ary
r
e
search in
ter
e
sts include th
e
hardware design
of all power
con
v
erters
and their
control
s
y
s
t
em
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.