Recei ed: 12 Janua y 2021
-
Re ised: 15 Ma ch 2021
-
Accep ed: 28 Ma ch 2021
-
IET Ene gy Sys ems In eg a ion
DOI: 10.1049/esi2.12018
ORIGINAL RESEARCH PAPER
Impac o ad anced in e e unc ions on low‐ ol age powe
g ids
A jen Men ens
1,2
|Ha old R. Chamo o
2
|Valé y Ann Jacobs
3
|Da id Topolánek
4
|
Jiří D ápela
4
|Wilma Ma inez
2
1
Depa men o Enginee ing Technology (INDI),
V ije Uni e si ei B ussel, B ussels, Belgium
2
Depa men o Elec ical Enginee ing (ESAT),
Ka holieke Uni e si ei Leu en, Diepenbeek,
Belgium
3
Depa men o Elec onics and In o ma ics
(ETEC), Depa men o Applied Physics and
Pho onics (TONA), Rec o a e, V ije Uni e si ei
B ussel, B ussels, Belgium
4
Depa men o Elec ical Powe Enginee ing
(UEEN), B no Uni e si y o Technology, B no,
Czech Republic
Co espondence
A jen Men ens, Depa men o Enginee ing
Technology (INDI), V ije Uni e si ei B ussel,
B ussels, Belgium.
Email: [email p o ec ed]
Abs ac
In oday's powe g id, a g ea numbe o in e e ‐based dis ibu ed ene gy esou ces (DERs)
a e connec ed and a e mainly designed o supply powe wi hou conside ing he ol age and
equency de ia ions o he g id. The e o e, dis ibu ion sys em ope a o s (DSOs) a e
challenged wi h an inc ease in g id e en s because o he andom implemen a ion o DERs.
Vol age le els can a y beyond p ede ined limi s a he poin o connec ion and a e cu en ly
no e alua ed by DSOs. Summa ized he e is he de elopmen o a simula ion model o
e alua ing he impac o suppo unc ions in eg a ed in in e e ‐based DERs. The model
aims o help g id ope a o s simula e ol age and equency e en s and s udy he impac o
DERs o he g id wi h espec o di e en se ings o in eg a ed suppo unc ions. A model
is de eloped in MATLAB/Simulink con o ming o Eu opean s anda ds and egula ions.
G id dynamics can be e alua ed by imi a ing ol age and equency de ia ions. Suppo
unc ions can be ei he adjus ed acco ding o he si ua ion o u ned o . Toge he wi h
adjus able se ings acco ding o DSO eques , his model o e s lexibili y and insigh in he
capabili ies o DERs o sol e ol age and equency issues. Case s udies show ha he model
co esponds o expec ed beha iou and can be used o u he de elopmen .
1
|
INTRODUCTION
The wo ld is looking o oppo uni ies o p oduce clean ene gy.
While households accoun o o e 27% o o al ene gy de-
mand, hey (indi ec ly) accoun o an agg a a ion o global
wa ming [1]. The Eu ope 2020 s a egy includes a ge s o
clima e change and ene gy, and go e nmen s a e p omo ing
DERs wi h incen i es [2, 3]. Wo ldwide, all (powe ‐consuming)
sec o s con ibu e o a ound 38% o ene gy‐ ela ed CO
2
emissions. Inc easing and s imula ing pho o ol aic (PV) p o-
duc ion can signi ican ly educe hese emissions, as 1 kWh
p oduced by PVemi s as li le as 15 g/kWh CO
2
compa ed wi h
he global a e age o 475 g/kWh CO
2
[3]. While me ely 3% o
elec ici y is gene a ed by PV, i a oids a ound 4.5% o powe
sec o emissions. This is because o coun ies wi h high ca bon
elec ici y gene a ion, such as China and India, ins alling a g ea
amoun o PV powe [3–5].
In he pas , powe was only consumed bu ne e supplied
by households. And hus, o a long ime, an on‐load ap
change (OLTC) was he only mechanism necessa y o change
local ol age le els. They ely on he ac ha he e is a uni o m
ol age d op ac oss he powe lines. Un o una ely, hey no
longe su ice. Due o he implemen a ion o in e e ‐based
DERs, mos ly PV panels, he uni o m ol age d op has
become less common, and ol age le els can a y in bo h
di ec ions [6].
Si ua ions e en exis whe e PV panels a e p ohibi ed in
pa s o he g id [7]. Ins ead o p ohibi ing hem, hey can
become pa o he solu ion. Households a e supplying an
amoun o powe ha can no longe be igno ed. I was ound
ha ol age de ia ions will no occu when he a e age
pene a ion pe household lies below 2.5 kW [8]. The s udy
in [8] assumed a DER pene a ion le el o 0%–11.25%, bu
pene a ion le els ha e isen o 22% [9, p. 13]. As a esul ,
ol age de ia ions a e occu ing mo e equen ly and wi h a
highe ampli ude bu only impac he local g id [10].
Ins ead o ein o cing he g id, PV in e e s can become
an impo an pa o g id suppo . Fo his eason,
This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion‐NonComme cial‐NoDe i s License, which pe mi s use and dis ibu ion in any medium, p o ided he
o iginal wo k is p ope ly ci ed, he use is non‐comme cial and no modi ica ions o adap a ions a e made.
© 2021 The Au ho s. IET Ene gy Sys ems In eg a ion published by John Wiley & Sons L d on behal o The Ins i u ion o Enginee ing and Technology and Tianjin Uni e si y.
426
-
IET Ene gy Sys . In eg . 2021;3:426–436. wileyonlinelib a y.com/jou nal/esi2
egula ions ha e been implemen ed [11]. While PV is a
clean al e na i e, i con ibu es only 2.9% o global elec-
ici y demand. This indica es ha PV is no a comp e-
hensi e solu ion o abandoning pollu ing powe ‐gene a ing
acili ies. Ul ima ely, de elopmen o ene gy s o age (elec-
ical, he mal, hyd ogen e c.) can play an impo an ole in
s imula ing in es men s in enewable ene gy esou ces in
gene al. The exponen ial g ow h o ins alled PV capaci y is
a i s a gumen o how hese ins alla ions could impac
he dis ibu ion g id and also why hey can and should be
used as g id suppo . The wo ldwide cumula i e ins alled
capaci y in 2008 was only 14.5 GW, while i exceeded
100 GW in 2012. Howe e , in 2018, a i e old le el was
al eady ins alled. Following his end, i can be expec ed o
each 1 TW by 2022 [3].
Dynamic models wi h ad anced unc ionali ies o
con e e ‐based gene a ion a e c ucial o unde s anding he
beha iou o he g id unde s essed ci cums ances [12, 13].
Howe e , such unc ionali ies b ing a ious challenges associ-
a ed wi h inc eased pene a ion o DERs and hei g id in e -
ac ion [14]. Sma in e e s wi h ol age and equency con ol
abili ies a e aluable o DERs so hey can con ibu e o he
g id wi h suppo unc ions and ancilla y se ices, such as
eac i e powe con ol, aul ide‐ h ough, and ha monic
compensa ion [15]. Many esea ch pape s ha e been published
in ecen yea s ha discuss he ol age iola ion issues ha
eme ge om he high pene a ion o in e e gene a ion in o
he powe sys ems [16]. Fo ins ance, a es sys em adap ed
om he medium‐ ol age dis ibu ion sys em in On a io,
Canada, is s udied in [17], p o iding g id ol age suppo
unc ionali ies. An op imized con ol s a egy o manage he
eac i e powe esou ce gene a ed by in e e ‐based gene a ion
is p esen ed in [18] o imp o e he quali y o he ol age dis-
ibu ion ne wo k and ul il he la es echnical equi emen s
lis ed by dis ibu ion sys em ope a o s (DSOs) in hei g id
codes. A case s udy dealing wi h long‐ e m ol age ins abili y in
sys ems hos ing ac i e dis ibu ion ne wo ks is epo ed in [19].
The documen ed simula ions show he e ec o he es o a ion
o dis ibu ion ne wo k ol age. A hyb id con ol s a egy o
suppo ing he ol age unde aul condi ions is p esen ed in
[20], demons a ing he simul aneous mi iga ion o ol age sags
by injec ing ac i e and eac i e powe o ide h ough he
pe u ba ion and main ain g id ol age. Simila ly, a ol age
egula ion scheme using a deadbea con olle ha helps o
mi iga e as ol age dis u bances is p esen ed in [21] ha
supp esses he ansien s in he sys em.
The au ho s in [22] p opose a con ol scheme wi ha dynamic
injec ion egion o he in e e sys em ha adap s o he se ‐
poin s assigned by a cen alized con olle . Simila ly, a con ol
scheme ha op imizes he eal‐ ime ope a ion o ac i e dis i-
bu ion ne wo ks while also conside ing he p o ision o ol age
suppo as an ancilla y se ice o he ne wo k equi emen s in
Swi ze land and in es iga e he ope a ional modes o he DER
in e e s is p esen ed in [23]. The s a ic ol age con ol consid-
e ing ol age‐ eac i e powe mode and dynamic and ex ensi e
ol age con ol wi h maximum u iliza ion o DER capaci y and
sys em s abili y a e s udied in [24].
As he p e ious con ibu ions discussed, a ious con ol
s a egies exis and a e able o unc ion well unde di e en
g id condi ions. Also, i shows ha applying a con ol s a egy
depends hea ily on he chosen con ol me hod and
pa ame e s.
This pape discusses he de elopmen o a simula ion
model o p o ide g id ope a o s wi h mo e insigh ega ding
he e ec o in e e ‐based DERs. Since he con ol pa am-
e e s can ha e a g ea impac on he g id esponse, he p o-
posed simula ion model o e s lexibili y ega ding pa ame e
choice.
The es o his pape is o ganized as ollows. Sec ion 2
desc ibes he possible me hods o p o iding g id suppo .
Sec ion 3p esen s he model design and implemen a ion in
MATLAB and explains he impo ance o se ing he co ec
ime cons an s. The simula ion esul s and discussions a e
p esen ed in Sec ion 4. Finally, he conclusions a e gi en in
Sec ion 5.
2
|
METHODS FOR PROVIDING GRID
SUPPORT
2.1
|
Ac i e and eac i e powe
compensa ion using in e e s
As b ie ly men ioned abo e, he in eg a ion o DERs will esul
in an inc eased ol age a he poin o connec ion (POC). Due
o luc ua ing injec ion o powe (sola and wind powe a e no
cons an ), he need o au oma ed solu ions is g owing, which
implies ha (e en au oma ed) OLTCs a e no longe su icien .
Using hese DERs o compensa e o low o high ol age is
one o he mos commonly discussed me hods [6, 7, 25–30].
Figu e 1depic s an equi alen schema ic o a powe line.
The esis ance and induc ance o he line co espond o he
eplacemen R and L alue.
The ol age in a ce ain poin is gi en by
U¼ ðP;QÞ ð1Þ
whe e
dU¼∂U
∂P dPþ∂U
∂QdQð2Þ
FIGURE 1 Powe line wi h R and L componen s o indica e esis ance
and induc ance alues
MENTENS ET AL.
-
427
and
ΔU¼U1−U2¼R⋅I2⋅cosφ2þX⋅I2⋅cosφ2ð3Þ
wi h φ
2
he phase shi be ween ol age Uand cu en I
2
. To
ew i e Equa ion (4) as a unc ion o P and Q,
ΔU¼R⋅PþX⋅Q
U2ð4Þ
Equa ion (4) indica es ha he ol age d op ΔUis ela ed
o he ac i e and eac i e powe . I can also be seen ha powe
lines wi h a high R/X a io will expe ience mo e impac om a
change in ac i e powe (P) han a change in eac i e powe (Q).
Th ee main ypes o in e e —a s ing in e e , a mic o‐
in e e , and a cen al in e e —exis [31, 32]. A s ing
in e e is based on sola panels connec ed in se ies. When one
PV panel is shaded o mal unc ions, he en i e powe ou pu is
limi ed by his one panel. A mal unc ioning PV panel can be
eplaced, bu shade caused by ees can o en no be con olled
by he owne . To o e come his, a mic o‐in e e can be
ins alled ins ead. The PV panels a e connec ed in pa allel, and
he e o e only he shaded o mal unc ioning panels a e limi ed
in ou pu powe . The di e ence wi h a cen al in e e is i s
size. Cen al in e e s a e mainly used in indus ial ins alla ions
wi h ypical powe anges om 100 kW o 1 MW [32]. Due o
i s size, hey a e no conside ed he e. The s udy in [31] also
shows ha mic oin e e sys ems p esen be e pe o mances
a bo h shaded and no ‐shaded condi ions. The main d awback
o a mic oin e e is he highe cos . Howe e , acco ding o
[33, p. 2885],
he s ing in e e appea s o ha e a lowe pe ‐
wa capi al cos when jus he in e e is
conside ed. Howe e , he in e e ep esen s only
abou 15% o he en i e PV sys em cos whe eas
he ins alla ion labou (…) cos accoun s o 40%,
depending on he sys em con igu a ion and
in e e echnology. These ac o s ha e made i
di icul o pe o m a compa a i e cos s udy.
The ollowing unc ions a e also known as ad anced
in e e unc ions and a e discussed in [6, 34]. The se poin s a
which hese unc ions a e deployed can di e acco ding o he
local equi emen s.
2.1.1
|
Ac i e powe compensa ion
The possibili y o he in e e o abso b P when he e is
o e ol age in he low‐ ol age (LV) g id is desc ibed as ac i e
powe compensa ion. The in e e is se o s a abso bing
ac i e powe when a h eshold ol age limi is me (e.g. a 3%
o e ol age, he in e e shall s a his compensa ion). Fi s , i
should be no ed ha his is only possible i a s o age sys em is
p esen o abso b ac i e powe . I no , he in e e can educe
i s P ou pu , and i necessa y, be disconnec ed om he g id.
This will only happen in ex eme si ua ions. Also, being
disconnec ed om he g id will cause a loss o income o he
ene gy p oduce , so his should be a oided as much as
possible. Second, his compensa ion is only a ailable un il he
s o age sys em is ully cha ged o un il a h eshold cha ge is
eached.
2.1.2
|
Reac i e powe compensa ion
In medium‐ ol age (MV) o LV g ids whe e he eac ance Xis
impo an , Q(U) compensa ion is used. In o de o alle ia e a
ol age d op caused by a g id e en , he in e e needs o
p o ide a ce ain amoun o Q o he g id [26, 35]. A deadband
a ound he nominal ol age le el is in oduced o p e en he
in e e om swi ching be ween abso bing o deli e ing Q in a
sho ime span [36].
2.2
|
Demand‐side managemen esul s in a
educ ion o use com o
Ins ead o limi ing he ou pu o an in e e , demand‐side
managemen (DSM) ocuses on limi ing powe usage in case
o high load and does he opposi e in case o high injec ion.
Washing machines, ho wa e bu e s, and (in he u u e)
elec ic ehicles may pose issues du ing peak hou s [37]. The
pu pose is o pos pone he usage o hese appliances. In his
manne , he load will be sp ead ac oss a g ea e amoun o
ime. The amoun o pos poned powe is ep esen ed as
lexibili y [38, 39].
DSM is seen as an impo an me hod o help mi iga e he
e ec s o he inc easing sha e o unp edic able enewable en-
e gy p oduc ion, he inc eased elec ical load due o ossil uel
powe ed equipmen being eplaced by elec ical equipmen ,
and he dec easing in es men s in di ec ly con ollable ( ossil
uel) plan s. To cla i y he p os and cons o DSM, he neces-
si ies o a success ul implemen a ion and i s con ibu ion o
blackou s a e discussed.
2.2.1
|
Necessi ies
Compa ed wi h using in e e s, he equi emen s a e mo e
challenging. Fi s , sma appliances a e needed o con ol he
powe usage acco ding o he ol age le el a he POC, which
was measu ed by a sma me e . These appliances consis o
pos ponable appliances, such as dishwashe s, washing ma-
chines and umble d ye s, and bu e ed appliances such as ho
wa e bu e s and elec ical ehicles. Ho wa e bu e s a e
conside ed o ha e he mos in luence on lexibili y. Second,
es amilies a e equipped wi h a home ene gy managemen
sys em. In a case s udy, one g oup was asked o al e hei
usage based on di e en ene gy a i s du ing he day and he
428
-
MENTENS ET AL.
o he g oup was equipped wi h an Au oma ed Home Ene gy
Managemen Sys em [38]. Appliances wi hou sma capabil-
i ies we e e o i ed wi h communica ion de ices o ensu e
sma con ol. Sma appliances we e u ned on o o au o-
ma ically, while basic com o , such as always being able o ake
a ho showe , was s ill p o ided. Final, ene gy s o age can be
in e es ing o PV owne s o p omo e sel ‐consump ion. I is
no eally seen as a necessi y, as he p ice pe kWh as well as he
kWh pe olume is s ill imp o ing.
2.2.2
|
Demand‐side managemen as a solu ion
o blackou s?
Conside ing ha 18% o Flemish households hea hei wa e
using elec ici y (ho wa e bu e s ha e he mos impac as
men ioned abo e), ex apola ing his o he whole popula ion
o Belgium, delayed powe usage would p o ide 207 MW o
powe . Taking in o accoun he o he appliances (washing
machines, elec ic ehicles e c.) adds up o 267.9 MW.
Compa ing hese alues wi h he 725 MW s a egic ese e ha
Belgian T ansmission Sys em Ope a o (TSO) Elia has o
c ea e, i can be assumed ha , e en wi h a pa icipa ion g ade
o 100% o households wi h ho wa e bu e s, he equi emen
will no be me [38, 40]. Ne e heless, DSM can become an
impo an pa o he solu ion.
In Belgium, mos o he ol age suppo is p o ided by
OLTCs. Once pe yea (o mo e, depending on he necessi y)
he ap s and o he ans o me is changed o mee ol age
limi s. This is done manually, bu mo e au oma ed solu ions a e
being implemen ed. To complemen OLTC suppo , o e en
ully eplace hem, in e e ‐based DERs can be used. Bo h P
and Q suppo unc ions can be implemen ed o alle ia e
ol age and equency de ia ions.
Gi en he abo e, i is clea ha addi ional g id suppo
unc ions should be implemen ed in in e e s o con ol
ol age le els a he POC [11]. This should be ex ended om
household DERs (e.g. PV panels) o in e e ‐based powe
plan s. Be o e deploymen o suppo unc ions, ex ensi e
es ing needs o be pe o med in o de o p e en e o s and
op imize e ec i eness o implemen a ion. The lack o
comp ehensible simula ion models makes i mo e di icul o
pe o m plausible es s [25].
3
|
DESIGN AND IMPLEMENTATION
USING MATLAB/SIMULINK
3.1
|
In oduc ion
Tes ing and modelling will be pe o med in MATLAB/Simu-
link. Be o e i s implemen a ion, a basic LV g id model has o
be de eloped. This model, as illus a ed in Figu e 2, consis s o
a ol age sou ce, se e al loads ha simula e household o in-
dus ial loads, and one o mo e DERs. The ope a o will be
able o imi a e g id e en s by swi ching loads on and o o
cause ol age de ia ions o by se ing he equency le el so
ha he g id dynamics can be e alua ed. The ol age as a
unc ion o he line leng h p o ides us insigh on he impac o
DERs. An ac i e DER will cause a aised ol age nea he
POC. In Figu e 2 his is shown as a posi i e e ec while he
ol age le el s ays wi hin i s bounda ies o a longe line leng h.
Issues will occu when mul iple DERs a e connec ed in an a ea
whe e hey can ein o ce each o he 's beha iou . This may
cause he maximum ol age le el o be exceeded. To esol e
his, DERs should implemen unc ions o suppo he g id
and change hei ou pu acco ding o ol age and equency
le els [11, 25, 41].
Figu e 3depic s he simula ion model in one block dia-
g am. Again, colou codes a e used o indica e he o igin o he
se ings. The p ese cha ac e is ics consis o he P(U), P( ), Q
(P), and Q(U) blocks. The measu emen s a e he ou pu o he
equi alen g id model (see Figu e 4). The ime cons an s a e
used as an inpu o he p ese cha ac e is ics and can be
changed acco ding o DSO eques . The use se ings a e
simula ion speci ic. On he one hand, a ol age and equency
e o can be simula ed o compensa e o measu ing issues. On
he o he hand, he minimum powe ac o can also be se .
Needless o say, his will in eal li e be de i ed om he
connec ion con ac be ween DSO and he owne o he
in e e .
3.2
|
Equi alen g id model
The equi alen single‐phase g id model, as illus a ed in
Figu e 4, is used as a ealis ic ep esen a ion o a g id. Since he
ol age sou ce block is ideal ( his is he s anda d se ing in
Simulink) a sou ce induc ance is added o compensa e o he
sho ‐ci cui impedance o he second ans o me winding.
The line impedance is calcula ed based on he line leng h, wi h
a esis ance o 0.38 Ω pe km and an induc ance o 0.72 mH
pe km. While his model ocuses on simula ing ol age de-
ia ions by se ing he sou ce main ol age, only wo main
loads o 9.2 kW a e used. A u he segmen a ion o loads
should be made when he mu ual dis ance be ween households
is o impo ance, and hus a line impedance be ween
FIGURE 2 Basic in e p e a ion o dis ibu ed ene gy esou ce (DER)
impac (Z =line impedance), (a) is wi h DER and has a posi i e impac ,
(b) is wi hou DER and ol age d ops below he limi a he end o he line
MENTENS ET AL.
-
429
households should also be added. The a o emen ioned alues
o impedance and loads a e always unique o a speci ic si -
ua ion and should be changed acco dingly when using his
simula ion model o o he se ups. The alues o loads and PV
sizes can be ob ained by alues gi en by he digi al sma me e .
Simulink p o ides many use ul s anda d blocks. Simple
cha ac e is ics, such as P(U), Q(P), and Q(U) cu es a e
implemen ed using 1‐D lookup ables. The P( ) cha ac e is ic,
implemen ed in laye 2, equi es ex a unc ionali ies. Mo e
complex unc ions a e he e o e implemen ed using a combi-
na ion o 1‐D lookup ables and MATLAB unc ions.
3.3
|
Cha ac e is ics o con ol scheme
3.3.1
|
P(U) cha ac e is ic
I is necessa y o calcula e P(U) and Q(U) o calcula e I
ampli ude
and i
phase
. P is calcula ed by using a P(U) cha ac e is ic shown
in Figu e 5. The ac ual implemen a ion is shown in Figu e 6.
The ol age [p.u.] alues used o limi ing P can di e ac-
co ding o DSO equi emen s.
While mos egula ions a e based on p.u. alues, his
model also uses he ol age p.u. as an inpu o he P(U)
cha ac e is ic. The implemen a ion also equi es a limi in
ou pu o p e en alues lowe han 0 and highe han 1.
A e his, he P(U) cha ac e is ic ou pu is mul iplied by he
a ailable P (P
nominal
mul iplied by an a ailabili y ac o ,
depending on uncon ollable a iables, e.g. sunligh ). The
a ailabili y ac o can be used by he ope a o o limi he
nominal powe ou pu caused by shadow, lack o sunligh ,
and so o h
3.3.2
|
P( ) cha ac e is ic
DERs can also ha e an impac on equency le els. A single
DER will no ha e a isual impac , bu adding hem all
FIGURE 3 This lowcha ep esen s he en i e simula ion model. All inpu s, measu emen s and calcula ions a e summa ized in one block diag am
FIGURE 4 Equi alen single‐phase g id model. A isual
ep esen a ion makes i s aigh o wa d o add o edi pa ame e s
430
-
MENTENS ET AL.

oge he will. To compensa e o his change in equency
le el, he powe ou pu can be adap ed i o e ‐o unde -
equency is p esen . The P( ) cha ac e is ic shown in
Figu e 7is desc ibed in [41]. A alue h eshold is applied o
main ain 0.6 p.u. P ou pu when equency eaches 51.1 Hz.
To p e en equency le els om ising oo as again when
equency d ops (P ou pu will also ise again), he P ou pu
is limi ed o 0.6 p.u. un il he equency d ops below
50.1 Hz.
3.3.3
|
Q(P) modes
The ou pu o he P(U) o P( ) cha ac e is ic is used as an
inpu o calcula ing Q(P). This can be calcula ed using
di e en use ‐speci ic Q modes. The ollowing pa ag aphs
explain all he con igu ed modes. In he simula ion model, a
swi ch selec o de e mines which mode is cu en ly p e e ed.
The applicabili y o Q(P) modes is no u he discussed, as
his is beyond he scope o his pape . A speci ic case s udy
wi h only changing he Q(P) mode could de e mine he mos
sui able mode.
Q(P) mode 1
The i s Q‐mode ou pu s he minimum alue be ween
Q
nominal
and Q
PFmin
. The minimum powe ac o (PF) in he
model is 0.85 bu can be changed by he ope a o . No e
ha he P inpu o bo h subsys ems is di e en . Fo
calcula ing Q
nominal
, a limi a ion in Q is no aken in o ac-
coun ( he limi a ion being he a ailabili y ac o , depending
on sunligh e c.).
Q(P) mode 2
The second mode uses a a ying PF as a unc ion o P ou pu .
Figu e 8illus a es ha he PF a ies om 0.9 o 1. The slope
om 0.5 o 1.0 p.u. P can be changed acco ding o DSO o
local equi emen s.
Q(P) mode 3
The hi d mode uses PF
nominal
o calcula e he Q(P) ou pu . In
his mode, he Q ou pu is always p opo ional o he P ou pu .
Q(P) modes 4, 5, and 6
The emaining h ee modes a e he ollowing:
�Q
nominal
calcula ed wi h PF
nominal
and P
nominal
,
�cons an Q,
�ze o Q when no Q suppo is expec ed.
All six modes will ha e a di e en impac on ol age
le els. The bes mode will di e acco ding o he si ua ion
and he gene al ol age p o ile o he eede . Cu en ly, he
p e e ed mode is chosen manually o be e e alua e he
impac in speci ic si ua ions. Selec ing he Q‐mode wi h he
lowes eac i e powe ou pu will be bene icial o he PV
owne , bu less bene icial o suppo ing ol age le els a
he POC.
3.4
|
Q(U) cha ac e is ic
The ou pu o he Q‐mode selec o is used o calcula e he
ac ual Q ou pu . This calcula ion is done using a Q(U) cha -
ac e is ic, shown in Figu e 9.
FIGURE 5 P(U) cha ac e is ic wi h a linea ol age limi om 1.09
p.u. o 1.11 p.u.
FIGURE 6 The me hod o implemen ing he P(U) cha ac e is ic in
he model
FIGURE 7 P( ) cha ac e is ic, p oposed in [41, p. 29] and
implemen ed in simula ion model
FIGURE 8 PF(P) cha ac e is ic indica ing a dec ease in PF when mo e
han 0.5 P p.u. is deli e ed
MENTENS ET AL.
-
431
3.5
|
Use o a iable ime cons an s in P(U),
P( ), and Q(U) cha ac e is ics
In o de o slow down he esponse o gene a ing uni s,
addi ional delay in he o m o a i s o de low‐pass il e is
in oduced. The P(U), P( ) and Q(U) cha ac e is ics use a
di e en ime cons an . This ime cons an is changed manu-
ally, bu in eal li e ope a ion i is eques ed by he ele an
DSO.
To o e mo e lexibili y ega ding ime cons an s, he
Model Disc e ize (Simulink app) is used. A con inuous ime
ans e unc ion can be con igu ed and is used o compu e he
disc e e ans e unc ion. The ze o‐o de hold me hod is
chosen, since his me hod uses he exac con inuous alue and
holds i o (in his case) 0.02 s.
The possibili y o changing he ime cons an , e en
du ing simula ion, is in e es ing o compa ing he impac o
di e en ime cons an s. Also, his can simula e he eques
o DSOs. As men ioned abo e, a speci ic ime cons an can
be eques ed by he DSO o in luence he impac o he
DER. A smalle ime cons an will also b ing mo e isk, as
his can cause oscilla ions due o sudden changes o a ail-
able P and Q.
4
|
SIMULATION RESULTS
To alida e he simula ion model, andom alues o he
ol age and equency se poin s (see Table 1) a e se o
explain he ou pu and indica e he accu acy o imple-
men a ion. Table 1summa izes he es condi ions. A ela-
i ely la ge s ep size o 0.2 s is chosen o imp o e simula ion
ime. Howe e , la ge sys em s udies can be pe o med wi h
espec ing slow dynamics (in o de o seconds). These a e
alid o bo h case s udies. An R/X a io o a ound 1 o 3 is
usual in LV g ids [5]. In his case, a a io o a ound 2 is used.
The eac ance X depends on he induc ance L and he
equency . I is gi en by
X¼ωLð5Þ
whe e
ω¼2π ð6Þ
4.1
|
Case 1: unde ol age wi h
o e equency
Figu e 10 depic s he P( ) cha ac e is ic and he o e w i ing o
he P(U) cha ac e is ic when he ol age le el is 0.94 p.u. o
lowe (see Figu e 3). The o e w i e is implemen ed o p e en
a bigge ol age d op when bo h unde ol age and o e -
equency a e p esen a he same ime. A =4.58 s, he
cu en limi is eached (see di e ence be ween calcula ed and
measu ed P and Q ou pu ). This indica es one o he e-
s ic ions o he in e e . No e ha his is also he bes ‐case
scena io wi h an a ailable powe o 100%, hus ou pu po-
we can e en be mo e es ic ed. A 2.00 s, o e equency
occu s and P ou pu d ops acco ding o P( ) cha ac e is ic (see
Figu e 7). A 4.04 s, ol age d ops below he le el, acco ding o
Figu e 9, ha ac i a es he Q(U) cha ac e is ic. Vol age keeps
d opping, and a 4.14 s, i eaches 0.94 p.u. and indica es an
o e w i ing o he P( ) cha ac e is ic by he P(U) cha ac e is ic
(see ∗bo om le in Figu e 3). Finally, a 4.58 s, a sa u a ion in
FIGURE 9 Q(U) cha ac e is ic wi h he deadband as discussed in 2.1.2
TABLE 1Model se ings used in he case s udies
Desc ip ion Value o se ing
S ep size 0.02 s
PF
nominal
0.90
PF
minimum
0.85
Q‐mode Q‐mode 1 (see 3.3.3)
P
nominal
10.00 kW
P
a ailable
1.00 p.u.
P(U) τ0.40 s
P( ) τ0.40 s
Q(U) τ0.40 s
Main sou ce ol age 414.00 V
Line leng h 1.00 km
Line esis ance 0.38 Ω/km
Line induc ance 0.72 mH/km
Two ex a loads Bo h o
O e ol age se poin s [1.00 1.05 1.09 1.10 1.11]
Unde ol age se poin s [1.00 0.98 0.96 0.94 0.92]
Vol age e o 0.00%
O e equency se poin s [50.00 50.50 51.10 50.50 50.00 50.00]
Unde equency se poin s [50.00 49.80 49.50 49.00 50.00 50.00]
F equency e o 0.00 Hz
432
-
MENTENS ET AL.
he in e e occu s and cu en is limi ed. The second pane
indica es his as a di e ence be ween he calcula ed and ac ual
powe ou pu .
4.2
|
Case 2: o e ol age wi h o e equency
Figu e 11 depic s he use o he minimum be ween P(U) and P
( ) cha ac e is ics o ol age le els o 0.94 p.u. o highe .
S a ing om 4.04 s, he P(U) cha ac e is ic ou pu s less P and
is he e o e de e mining he calcula ed P ou pu . A 2.00 s,
o e equency occu s and P ou pu d ops acco ding o P( )
cha ac e is ic (see Figu e 7). A 4.04 s, he ol age exceeds 1.05
and ac i a es he Q(U) cha ac e is ic (see Figu e 9). No e ha
case 2 ac i a es he opposi e side o he cha ac e is ic han case
1. A 6.04 s and 8.04 s, ol age eaches 1.11 p.u., and P and Q
d op acco dingly o mi iga e he ol age iola ion (see
Figu e 5). I also depic s he o e w i ing o he P( ) cha ac-
e is ic by he P(U) cha ac e is ic (see ∗bo om le in
Figu e 3). Finally, a 10.06 s, ol age is be ween he limi s—P
FIGURE 10 Case 1: when bo h unde ol age and o e equency occu , he in e ac ion o he P(U) and P( ) cha ac e is ic can be deno ed. The simula ion
pa ame e s a e se as de ined in Table 1
MENTENS ET AL.
-
433
and Q ou pu s a e un es ic ed. This case is simila o o e -
ol age wi h o e equency, as he P( ) cha ac e is ic is only
ac i e be ween 2.00 s and 4.04 s.
5
|
CONCLUSIONS
This pape gi es a b ie o e iew o he ol age con ol
me hods on LV powe g ids. The mos common suppo ,
ac i e and eac i e powe compensa ion using in e e s, is
u he discussed. DSM is p omising bu equi es sma ap-
pliances and household pa icipa ion. The e o e, i cu en ly is
no a su icien solu ion.
A li e a u e s udy iden i ies he p esen issues ha ol age
con ol echniques a e acing. Con ol pa ame e s (such as he
abili y o choose a Q(P) mode) a e o en ixed, which does no
allow adequa e a ia ion when he simula ion model is used o
de e mine he mos app op ia e solu ion o g id issues. The
de elopmen in a isual simula ion model such as Simulink is
he e o e ecommended.
FIGURE 11 Case 2: when bo h o e ol age and o e equency occu , he minimum ou pu be ween P(U) and P( ) is chosen. Q ou pu is s ill con olled by
he Q(U) cha ac e is ic. The simula ion pa ame e s a e se as de ined in Table 1
434
-
MENTENS ET AL.