studiu fezabilitate VPP

7/25/2019 studiu fezabilitate VPP

http://slidepdf.com/reader/full/studiu-fezabilitate-vpp 1/53

Examensarbete 30 hp Juni 2013

Feasibility study of a Virtual

Power Plant for Ludvika

Johanna Lundkvist



Teknisk- naturvetenskaplig fakultetUTH-enheten

Besöksadress: ÅngströmlaboratorietLägerhyddsvägen 1Hus 4, Plan 0

Postadress:Box 536751 21 Uppsala

Telefon:018 – 471 30 03

Telefax:018 – 471 30 00

Hemsida:http://www.teknat.uu.se/student

Abstract

Feasibility study of a Virtual Power Plant for Ludvika

Johanna Lundkvist

This thesis is a feasibility study of avirtual power plant (VPP) in centralSweden and part of a project withInnoEnergy Instinct and STRI. The VPPconsists of a wind park, small hydroplant as well as solar photovoltaic andenergy storage. The 50 kVsubtransmission network was modeled inorder to evaluate the network servicesthat could be provided by coordinatingexisting distributed energy resources inthe network. Simulations where performedusing measured hourly variations in

production and consumption of allnetwork nodes. The studied networkservices included both reactive andactive power control. The aim of this thesis is to evaluatethe potential contribution from the VPPfor capacity firming in order to allow abalance responsible party to meet placedbids on the day-ahead spot market,minimize peak load in order to reducesubscribed power, decrease networklosses, the contribution from reactivepower control using the power convertersis studied. Comparisons of the economicgains from spot and balance markets ofthe VPP distributed energy resources aremade for each operation case.

Sponsor: InnoEnergy Instinct and STRIISSN: 1650-8300, UPTEC ES 13015

Examinator: Kjell PernestålÄmnesgranskare: Joakim WidénHandledare: Nicholas Etherden



Populärvetenskaplig sammanfattning

El producerad fr,n inter#ittenta produ!tions!-llor" so# till e&e#pel sol och vind!raft" f.rv-ntas .!a

/etta leder till en tidss!illnad i !onsu#tion och produ!tion so# #,ste hanteras /etta proje!t

studerar enbart variationer p, ti#basis p, grund av tillg,ng till data /en inter#ittenta aspe!ten i

produ!tionen #,ste !o#penseras f.r att undvi!a sv,ra variationer i n-tsp-nning och0eller fre!vens

Proje!tet -r en del ino# InnoEnergy Instinct och -r utf.rt so# e&a#ensarbete ino#

civilingenj.rsprogra##et i energisyste#" givet vid 1ppsala universitet och S21 Proje!tet -r oc!s,

underlag till tv, !onferensarti!lar f.r !onferenser ino# s#arta eln-t

3odeller har tagits fra# f.r att unders.!a p,ver!an p, ett distributionsn-t vid anv-ndning av ett s,

!allat virtuellt !raftver! /etta s!er geno# analys av #-tv-rden ifr,n ett distributionsn-t" en #odell i

ett #ate#atis!t ber-!ningsprogra# och en #odell i ett si#uleringsprogra# f.r ele!tris!a

fl.desber-!ningar

4.r att anpassa produ!tion och !onsu#tion !an energi #ellanlagras #ed hj-lp av styrning av ett

virtuellt !raftver! i batterilager eller vattenreservoarer /et unders.!ta virtuella !raftver!et antogs

best, av batterier" solceller" en vind!raftspar! och ett litet vatten!raftver! 5ontrollsyste# f.r

!raftver!ets oli!a appli!ationer utvec!lades i ber-!ningsprogra##et och si#ulerades #ed insa#lad

#-tdata

Svaren fr,n si#uleringarna analyserades utifr,n ett e!ono#is!t perspe!tiv f.r att ge en b-ttre bild

.ver investeringsnyttan hos ett virtuellt !raftver! 6eno# en !o#bination av ver!liga data och

si#ulerade svar !unde d-rigeno# effe!ten av ett virtuellt !raftver! i ett regionn-t unders.!as

Virtuella !raftver!s#odellen innefattar f.ljande #o#ent7

1. 3-tdata sa#las in fr,n ett ver!ligt distributionsn-t

2. Effe!tfl.de fr,n solceller s!alas upp f.r att #-r!as i si#uleringarna

3. 2agernas !o#penseringsbehov ber-!nas #ed hj-lp av .ns!at energifl.de

4. /et nya energifl.det unders.!s och !vantifieras #ed hj-lp av besparingsber-!ningar

5o##uni!ations' och infor#ationsbehovet f.r i#ple#entering av ett virtuellt !raftver! unders.!tes

utifr,n de byggda #odellerna /etta #,ste vidare studeras d, #odellerna bygger p, ti#data och

styrning i ett ver!ligt syste# b.r utf.ras i realtid

8 of ++



Executive summary

E1 countries ai# to have 8%9 of their energy production fro# renewable energy resources by 8%8% This

will contribute to an increase of inter#ittent generation in the distribution grids To handle proble#s with

grid capacity" voltage control and variable load flow" active control are going to be needed in the

distribution grids This thesis ai#s to find infor#ation needed to build a virtual power plant for active

control in a distribution grid in central Sweden and to find the opti#iation application best fitted for the

grid

This thesis shows that the applications #ost fitted for the virtual power plant is to opti#ie for #eeting the

bids on the day'ahead spot #ar!et and #a!e sure that the power output is the sa#e as predicted the day

before To opti#ie for lowering the #onth highest pea! de#and and thereby lowering the tariff is also a

fitted application

:ased on this thesis two scientific conference papers has been written*" 8

* N. Etherden, M. H. Bollen and J. Lundkvist, "Quantifiation of !nillar #ervies fro$ a %irtual &o'er &lant in an E(istin)

#u*trans$ission Net'ork," in +EEE &E# +nnovative #$art rid -ehnolo)ies +#- Euro/e0, o/enha)en, 213.

8 N. Etherden, M. H. Bollen and J. Lundkvist, "o$$uniation euire$ents of a %irtual &o'er &lant usin) +E 5167 to

&rovide rid #ervies," in +EEE #$artrido$$, %anouver, 213



Acknowledgment

This project gave i#portant insight of how it is to wor! in new and uncharted territories and with new

software The project also gave insight in the i#portance of co##unication and discussions in a large

project Perfor#ing a project for real applications co#pared to acade#ic purposes gives a wider

representation of the e&isting barriers and a #ore custo#ied view

I would li!e to than! 3ath :ollen at STRI ;: for helpful discussions and advice I would also li!e to than! V:

Energi for #easure#ent data and fore e&planations and discussions about their grid ; special than!s to all

engineers at STRI ;: in particular the ones at STRI ;: in V-ster,s

1ppsala 1niversity also provided valuable assistance" and I appreciate the support and assistance of <oa!i#

=id>n and 5jell Pernest,l

;t last I want to than! ?icholas Etherden for all support and useful discussions and advice



Table of Content

*;bbreviation*

8Introduction*

8*;i#8883ethod8

8+2i#itations of scope8

+:ac!ground8

+*Virtual Power Plant8

+**E&isting de#onstration networ!s+

+8Studied grid$

+8*The $% !V and *% !V grid$

+88@ydro power plant 2oforsenA

+8+=ind far# B

+8CPhotovoltaic unitsB

+8$:attery storageB

CTheoryB

C*Virtual power plantB

C8Econo#yB

C8*:alance #ar!etB

C88TariffD

C8+2ossesD

$3ethodD

$*Si#ulation dataD

$8 Power4actory #odel*+

$+3odels of controllable units*+

$+*@ydro power plant #odel*C

$+8:attery storage #odel*$

$Cpti#iation applications for the VPP*A

$C*:ase case*A

$C8Responsible party to #eet placed bids on the day'ahead spot #ar!et *B

$C+Pea! shaving8%

$CC/ecrease networ! losses88

$C$Reactive power control8CAResults8C

A*Responsible party to #eet placed bids on the day'ahead spot #ar!et 8C



A8Pea! shaving8B

A+/ecrease of networ! losses8D

ACReactive power control8D

B/iscussion8F

B*I#ple#entation of the VPP8F

B86eneral thoughts+%

B+Possible i#prove#ents of the #odels+%

Donclusion+*

References+8

;ppendi& ; Si#ulated #iddle voltage grid

;ppendi& : Si#ulated *+% !V grid

;ppendi& Guantification of ;ncillary Services fro# a Virtual Power Plant in an E&isting

Subtrans#ission ?etwor!

;ppendi& / o##unication ReHuire#ents of a Virtual Power Plant using IE A*D$% to Provide 6rid

Services

A of ++



1 Abbreviation

:ESS :attery Energy Storage Syste#

@P o#bined @eat and Power/ER /istributed Energy Resources

/6 /istributed 6eneration

/R /e#and Response

/S /istribution Syste# perator

PV Photovoltaic

RES Renewable Energy Resources

So State f hargeTS Trans#ission Syste# perator

VPP Virtual Power Plant

2 Introduction

3ost renewable energy resources (RES) are inter#ittent sources that are hard to predict and control

such as wind and solar power /aily and seasonal variations can easily be predicted but in s#aller

ti#escales the power fluctuations #ay be large" for e&a#ple the production fro# a photovoltaic

installation can change B%9 of rated power within half a #inute but production predictions fro# day'

ahead for a specific hour is correct for around B% 9 of the predictions *J The variations do not occur

on the sa#e pattern for the different RES neither on large nor s#all ti#escale

3any of the RES are often placed as /istributed 6eneration (/6) in the #ediu# and low voltage grid

The grid is designed for radial distribution of power fro# a few large scale centralied distribution

units and can therefore have proble#s with /6 E&a#ple of proble#s is variable power flow with

unintended islanding because of the wind and solar fluctuations" and too high voltage levels in the

end of the distribution grid which cause insulation da#age to eHuip#ent and unsafe situations for

the end users 8J

In a cluster of different !inds of RES the fluctuations fro# for e&a#ple wind and solar can cancel each

other out and the variations would be reduced ; hydro power plant can for e&a#ple regulate to

#ini#ie the fluctuations and capacity fir#ing for the cluster If other !inds of /6" battery energy

storage syste# (:ESS) and /e#and Response (/R) providers would be i#ple#ented in the cluster the

fluctuations could be reduced even #ore Such aggregation of /istributed Energy Resources (/ER)

controlled fro# one location using co##unication lin!s is called Virtual Power Plants (VPP) 8J

; VPP for research could be built in 2udvi!a" in central Sweden" with solar production and e&isting

S#art 6rid Research" /evelop#ent K /e#onstration platfor# at STRI together with e&isting hydro

and wind power production in the surroundings The need for a VPP is un!nown and different

* of ++



applications are going to be investigated in this thesis to evaluate different use and the benefits of

i#ple#ent a VPP in the industrial city 2udvi!a

21 Aim

The ai# of this thesis is to find applications for VPP by a literature study" evaluate different

application for a planed research VPP in 2udvi!a and find fields of applications that can be co#bined

;n evaluation of the infor#ation need for control of the VPP in the $% !V grid of 2udvi!a will also be

#ade

22 !et"od

3easured hourly consu#ption and production data fro# the regional distribution grid is obtained by

the /istribution Syste# perator (/S) %B Eln8t, a subsidiary of %attenfall ther #easure#ent data

is obtained fro# STRI" were the solar #easure#ents is linear up scaled EHuations and the

#easure#ent data #odels for hydro power plant and battery storage is built in 3;T2;: 3odels for

the different opti#iation applications of VPP are constructed in 3;T2;: fro# the #odels for hydroand battery" eHuations and #easure#ent data 4ro# these #odels a production plan for hydro and

storage is calculated To evaluate the results the cost'saving in the #odel is calculated 3easured data

and the 3;T2;: calculated production plan is used in the si#ulation #odel of the $% !V grid created

in the /igSI2E?T Power4actory to evaluate the results

2# $imitations of scope

The following li#itations have been #ade in the study

• ?o pure #ar!et oriented opti#iation of the VPP including #a&i#iing profit fro# trading on

the spot #ar!et

• ?o controllable heat production or other de#and response

# %ackground

This chapter will focus on the virtual power plant concept and e&isting projects :ac!ground for the

e&isting grid near 2udvi!a is also presented in this chapter

#1 &irtual Po'er Plant

; virtual power plant is a cluster of several /ER that are controlled fro# one location and participatesin the spot #ar!et as a single generator Energy storage and0or de#and response" for e&a#ple heat

pu#ps and electric vehicles" can be included in a VPP The /ER in a virtual power plant can be spread

over the distribution grid but are assu#ed to be in the sa#e geographic area :ecause of the

fle&ibility of a /ER cluster VPPs are seen as a way to integrate #ore RES in the distribution and

trans#ission grids+J CJ

There are #any different uses of a virtual power plant" as seen in +** The inter#ittent energy

output of RES can cause fluctuation in the power supply fro# other plants in a passive controlled grid

:y coordinating all plants in the area this can be reduced Increase of /6 in a passively controlled

8 of ++



distribution networ! will also increase the voltage 4or econo#ical aspects the VPP can be used to

reduce the regulation cost for prediction errors or lower the tariff" described in C88 8J

#11 Existing demonstration net'orks

; few VPPs have been built as de#onstration projects around the world" but the reason to build the#

have been different in various projects ountries have different proble#s with integration of RES and

various grid codes which #a!es different interests for VPPs So#e of the e&isting de#onstration

networ!s will be described in this chapter in regard to project objectives

3.1.1.1 Cell Controller Pilot Project

The ell ontroller Pilot Project (PP) is built on the island :ornhol# in /en#ar! The networ! has a

production of around D% 3= on a voltage level at A% !V and lower The purpose of the project is to

perfor# e&tensive de#onstration of the functionality capabilities of the cell controller The

investigated functionalities of this project is #ar!et operations support" voltage control"

active0reactive power flow" freHuency control in blac! start and island operation $J

3.1.1.2 PREMIO

PRE3I is a 4rench project and an acrony# that in English stands for LIntegration and pti#iation

for /istributed 6eneration (/6) /e#and Side 3anage#ent (/S3) and Renewable Energy Resources

(RES)M The networ! consists of photovoltaic (PV) units" o#bined @eat and Power (@P) plants and

different !inds of /R The objective of the project is to opti#ie the integration of /6" storage and /R

as well as to reduce the power pea!s Secondary objective is to identify reHuire#ents for

co##unication" infor#ation and control To reach the objectives PRE3I focused on opti#iation of

the econo#ic parts such as day'ahead #ar!et and intraday #ar!et AJ

3.1.1.3 MILLENER

3I22E?ER is a VPP placed on the 4rench islands orsica" 2a R>union and 6uadeloupe" with the goals

of developing #ethods and tools to enable the integration of #ore distributed RES and to #anage

the power and energy de#and The networ! consists of 8$% PV syste#s and *%%% heating and

cooling syste#s The objectives of the de#onstration networ! are to #aintain the balance between

supply and de#and" opti#ie the power flow to enable a higher penetration of inter#ittent

generation and to decrease the e#ission of carbon dio&ide +J

3.1.1.4 Unna

In the 6er#an district 1nna is a VPP with the objectives to reduce the cost of balancing power placed

BJ The production technology co#ponents of the virtual power plant is si& @P plants with different

sies and power generation technologies" two wind far#s" a hydro power plant" a #icro gas turbine

and several s#aller PV units DJ The VPP produces +$6=h electricity and +B6=h of heat every year

FJ

3.1.1.5 Am Steinwe

;# Steinweg is a VPP placed near 3annhei# in 6er#any The project objectives are to opti#ie the

energy flow" i#prove econo#ic operation of the energy supply as well as eHualiation of the pea!

loads The VPP consists of PV installations" a gas driven @P plant and battery storage syste#s BJ

+ of ++



3.1.1.! CRES" N#UA an$ Meltem

In this project a VPP is built of /ERs and storages placed in three places in 6reece" the enter of

Renewable Energy Sources (RES)" the ?ational Technical 1niversity of ;thens (?T1;) and the

3elte#i holiday a#p The de#onstration networ! ai#s to show the advantage of a VPP by

increased fle&ibility The goal is to be reached by focusing on de#and patterns for different seasonsand #ar!et prices The production technology co#ponents of the virtual power plant is PV

installations" one @P syste# and /Rs *%J

3.1.1.% Micro&C'P (nit) at con)(mer *remi)e)

; field test of ten clustered *!= @P units to a VPP is done by E?" a /utch independent research

institute for renewable energy" and the gas co#pany 6asunie in ?etherlands The #ain goal is to

de#onstrate the ability of a cluster of #icro'@P units operated in a VPP to reduce the local pea!

de#and The field test is focused on the networ! utiliation factor of the local distribution grid" by

using #ar!et'based control to reduce the pea!'loads**J

3.1.1.+ PowerMatc,in Cit-

In the /utch project Power3atching ity the power balance is investigated The project ai#s to

reduce the i#balance caused by RES" specifically the i#balance caused by wind turbines The

production technology co#ponents of the VPP are 88 co##on households" *% of the# with a #icro'

@P and *8 with air'water heat pu#ps" both heating syste#s are each connected to a water buffer

acting as storage The cluster is supple#ented by two electric vehicles (EV) and a 8$ 3= wind

turbine *8J

3.1.1. San A()tin $el /(a$ali0

The Spanish technology de#onstration center San ;gustin del 6uadali& have a VPP consisting of PV

units" a wind turbine" a diesel generator with an installed capacity of *AB 3= and a nu#ber of

houses as loads *+J The objective of the VPP is to i#prove the power Huality and reliability in the

grid BJ

3.1.1.1 S(mmation

The projects described above su##arie the use of VPPs It is i#portant to opti#ie the econo#y to

get profitability in the invest#ent of building a virtual power plant The project use of VPP can be

divided into three categories ;ctive power control to #eet the day ahead #ar!et so no balance

power are needed by balancing the prediction error so no need to pay the balance #ar!et 2ower thetariff cost by reducing the largest load flow pea!s Reduce the losses by opti#ie the power flow and

power Huality These areas will be investigated for the 2udvi!a research VPP 3any of the projects

use both heat and power which is a better use of a VPP than only power The heat or cooling

production is a controllable load and households have a co#fort one that spans over a few degrees

which gives a freedo# to #ove the ti#e for heat and cool production This will not be further

investigated in this thesis ; list of the projects is found in Table *

C of ++



#2 (tudied grid

The grids in the surroundings of 2udvi!a have voltage levels of *% !V" $% !V" *+% !V and C%% !V The

VPP is investigated at the $% !V level and connected to the national trans#ission networ! at C%% !V

through the *+% !V grid The studied grid with generators connected to the *%!V and $%!V grid"

transfor#ers and lines are described below

#21 T"e )* k& and 1* k& grid

The infor#ation of the studied grid co#es fro# %B Eln8t a %attenfall subsidiary in central Swedenand the /istributed Syste# perator (/S) in 2udvi!a ;n overview of the VPP is shown in 4igure *

The grid in the surroundings of 2udvi!a consists #ostly of overhead lines of different di#ensions but

there are two ground cables in the studied *% !V grid The line types in the syste# are shown in Table

8 The grid has three connections to the *+% !V grid In this thesis the load flow in one of these

connection points is studied

-a*le 29 Line t/es used in the 7 k% and 1 k% )rids.

2ine type Resistance N0!# Reactance N0!#

A+% ;O5< %%CAF %*%%$

;<< *D$ %*A8 %%F*

4E;2 FF %++A %+B$F

4E;2 *$B %8*C %+B$F

4E;2 8+C %*C+ %+B$F

:2 FF %8FA %+%8D

:2 8C* %*8B %+$%

$ of ++

-a*le 19 ! su$$ation over de$onstration net'orks in the 'orld and there /ro:et o*:etives.

Country Name of project Objectives References

/en#ar! ell ontroller Pilot Project Econo#y" voltage control"active0reactive power control" island

operation" blac! start" freHuencycontrol

$J

4rance PRE3I /ay'ahead #ar!et" balance #ar!et"energy efficiency

AJ

4rance 3I22E?ER pea! shaving +J

6er#any 1nna Econo#y" balance power BJDJFJ

6er#any ;# Steinweg Econo#y" opti#iation of power flow BJ

6reece RES" ?T1; and 3elte# Econo#y (fuel" #ar!et) *%J

?etherlands 3icro'@P units at consu#erpre#ises

pea! shaving **J

?etherlands Power3atching ity ;ctive power control *8J

Spain San ;gustin del 6uadali& Power Huality" power safety *+JBJ



3ost transfor#ers in the syste# have auto#atic transfor#er tap changers which #ean that they can

vary the voltage with plus #inus so#e percent in a few steps The transfor#er lea!age flu& (u)

causes the secondary voltage to not be directly proportional to the pri#ary voltage The transfor#ers

in the grid are su##aried in Table +

-a*le 39 &ara$eters inluded in the stud.

Transfor#ers Voltage !V Tap ;pparent power 3V; u 9 onnection

@yttbac!en *+D0*%$ FQ*AB9 C% D+ ?d**

STRI *%$0%C FQ*AB9 *$ *8 T?yn%

6r-ngesberg *+D0$$ FQ*AB9 A+ **D$ ?yn%

4j-llberget $$0*%$ DQ*AB9 *C FBC ?yn%

Sa&berget $$0*%$ FQ*AB9 8% *% ?yn%

?yha##ar $$0*%$ DQ*AB9 A+ B** ?d**

Tuna @-stberg $$0*%$ DQ*AB9 D BF+ ?d**

2oforsen $$0AA 8Q$9 D8 B+8 ?d**

Sa&dalen $$0*%$ DQ*AB9 *8 B+ ?yn%

S!a!elbac!en $8$0*%$ DQ*AB9 +% *% ?yn%

#22 +ydro po'er plant $oforsen

The hydro power generator is rated BB 3V; at AA !V and have been in operation since *FF% The

generator is directly coupled to a vertical 5aplan turbine with a #a&i#u# water flow (G) of *$#+0s

through the turbine The rating of the sole #achine is low co#pared to sie of the reservoir thus

enabling pea! power production The si#ulated hydro power plant has a #a&i#u# output (P) of A

3= active power and CD 3V;r reactive power *CJ

The da# is today used as a day0wee! reservoir depending on the loads and water levels The allowed

water levels in the da#" according to Swedish water'rights court" are 8% c# under the retention

water level in su##er and $% c# under in winter The #a&i#u# head of the da# (h) is CC$ # The

A of ++

;i)ure 19 ! si$/le $odel of the investi)ated %&&, 'here the *atter and /hotovoltai are u/ saled.



reservoir area (;) is 8 !#8 and assu#ed to be constant in the allowed head span The power plant

have no #ini#u# flows if the water level is in the right span *$J*AJ

#2# ,ind farm

The wind far# on Sa&berget and 4j-llberget consists of *B Vestas VF% wind turbines The first five was

installed in 8%%A and have a hub height of *8$ #eter The outer *8 was installed in 8%%D with a hub

height of *$% #eter *BJ The turbines have a rated power of 8 3= each The generators are doubly

fed so that the turbine can vary the rotation speed without changing the power'syste# freHuency

The stator operates on AF% V and the rotor at CD% V *DJ

#2- P"otovoltaic units

The photovoltaic power plant at STRI has a capacity of +% != and was installed in su##er 8%*8 The

plant is built of *8% solar #odules rated at 8*% to 8$% = and have since the installation an average

energy production of A8C !=h0day 8%J

#2) %attery storage

The battery storage at STRI has a #a&i#u# power output of 8*A !V; at C+8 V and have #a&i#u#

energy storage of 8% !=h The battery can be charge with #a&i#u# FB8 !=" which is C$9 of the

output power 8*J

- T"eory

-1 &irtual po'er plant

; VPP is controlled fro# one co##unication center" as in 4igure *" which gets infor#ation and#easure#ents fro# the /ER in the VPP The infor#ation is analyed together with other infor#ation"

for e&a#ple spot prices or whether prognosis" to investigate the need of changes in power output

The analysis gives a production plan for the chosen application of the VPP The wanted power output

is sent to the /ERs so they regulate the power production to fit the plan The infor#ation needed for

the VPP is investigated in B*

-2 Economy

The econo#ical #ar!ets and #odels used in this thesis will be described in this chapter in regard to

the cost for the /S

-21 %alance market

Every day at *8 a# an energy producer needs to sub#it a bid on the Spot #ar!et over how #uch

energy he can deliver and at what price for each hour the following day 8CJ The balance #ar!et is

used for co#pensate prediction errors fro# the day'ahead bids

If the production in a prie area is too low the trans#ission syste# operator (TS) have to by balance

power or if the consu#ption is too low they have to pay stations to lower their production These

costs should be paid by the producers in the sa#e prie area that contributed to the i#balance

Producers that do not contribute to the i#balance do not have to pay for it The regulation cost is

B of ++



different if it is up or down regulation The producers that should pay" pays the up or down prie

ti#es the error sie in 3=h The costs in SE5 per 3=h error can be found in 4igure 8 and on the

?ord Pool Spot website 8$J" there can also infor#ation about the historical regulations be found

-22 Tariff

The tariff is paid each #onth for the #a&i#u# hour average power output in !ilowatts (!=) during

the #onth 8AJ :y lowering the hour average power pea! input fro# e&ternal grid the /S can

reduce the cost to the TS

-2# $osses

;ccording to the Swedish Electricity ;ct (*FFB7D$B) *$ of the third chapter an owner of a VPP" orother installation of energy production" can get paid for reduce energy losses in the connected grid

The paragraph can be read below

L;n owner of a generation installation is entitled to pay#ent by the networ! concessionaire to whose

cable networ! the installation is connected The pay#ent shall correspond to

* the value of the reduction of power losses that the infeed of electrical power fro#

the installation involves for the networ! concessionaires cable networ!" and

8 the value of the reduction of the networ! concessionaires charges for having his

cable networ! connected to another networ! concessionaires cable networ! thatbeco#es possible through the installation being connected to the cable networ!

M8BJ

) !et"od

)1 (imulation data

The data used in the si#ulations co#es fro# several different sources In this chapter the data

#anage#ent is described

D of ++

;i)ure 29 e)ulation ost fro$ #/ot $arket 'ith u/ and

do'n re)ulation.



The power consu#ption in the *%!V grids is si#ulated as loads in the #odels The grid #ean wee!ly

power consu#ption can be seen in 4igure + The data is given by %B Eln8t as three year hourly power

output #easure#ents on the buses connected to the lower voltage grids The power inflow fro# the

*+% !V grid is hourly #easured at the three grid connection points over the three year period The

power inflow for the studied connection point between the $% !V and *+% !V grids vary during the

year as seen in 4igure C %B Eln8t have not started to #easure reactive power in the si#ulated $% !V

grid so no data is available for the si#ulated ti#e" therefore the load power factor assu#ed to be %F"

according to an assu#ption fro# %B Eln8t

4or the water inflow to the hydro power reservoir hourly #easure#ents of production and water

level is used Two and a half year #easure#ents of the water level in the reservoir are provided by

the hydro power plant owner %B <raft The data have so#e #easure#ent errors and so#e data is

#issing (around 8%9)" these are replaced with the sa#e values as the last hour #easure#ent or for

errors over longer ti#e a #ean of the other years is calculated for these values The hydro power

production is provided by three year hourly #easure#ent of the power output on the bus connected

F of ++

;i)ure 39 !vera)e 'eekl /o'er onsu$/tion in the )rid.

;i)ure 49 &o'er flo' in one of the 7=13 k% onnetion /oints.



to the hydro power plant" given by %B Eln8t =ithout reservoir" the production would correspond to

the water inflow" is seen in 4igure $

The wind power production data is given by three year hourly #easure#ents of the power input on

the transfor#ers connected to the two wind far#s These #easure#ents are also provided by %B

Eln8t The su##aried power production fro# wind power is seen in 4igure A The figure shows that

the wind does not have a strong yearly variation

To get an effect fro# the photovoltaic installation in the si#ulated VPP it is scaled up :ased on an

assess#ent of available roof area in adjacent industrial area" the installed solar power is assu#ed to +

3= in the si#ulations

4or the sun power production #easure#ents fro# lava ener) enter 8BJ *%D != solar par! (8C%

!# fro# 2udvi!a) is scaled and used for second half of year *" whole year 8 and first half of year + 4or

*% of ++

;i)ure 59 Measure$ent data for 'ind /o'er /rodution.

;i)ure 79 &o'er inflo' to the hdro /o'er reservoir /er hour.



the second half of year three #easure#ents fro# the STRI solar par! is scaled and used This because

the solar installation at STRI was installed in su##er 8%*8 6lava and 2udvi!a are placed at al#ost

the sa#e height in Sweden and according to 8FJ they have the sa#e nor#al sunshine ti#e in a year

The data fro# both 6lava and STRI is available for a few #onths" visual inspection shows that the

data is si#ilar 4or the first half of year * #ean values for year 8 and + are used The co#bined data

used in the si#ulation is seen in 4igure B

In the si#ulations the battery storage is scaled up to D 3V; at C%% V with energy storage of +8 3=h

The charging power for the scaled storage is set to C$9 of the discharge that is +A 3= So#e

producers of battery storage clai#s to have an efficiency of around FD9 but in these si#ulations the

efficiency is assu#ed to be F%9 88J The up scaling of the storage capacity is chosen fro# the

conditions that it has to be large enough to effect the VPP and the facility should have a sie that is

realistic to build today The sa#e storage capacity as the si#ulated in this thesis is in use in a wind

energy storage project in alifornia" 1S; There facility is $D$#8 and contains batteries" transfor#ers"

power conversion syste#s" co##unication and interface eHuip#ent 8+J

@istorical hourly data for the regulation volu#es for the si#ulated three years is given by ?ord pool

spot 8$J The data for up and down regulating prices is also available at ?ord Pool Spot :ecause

2udvi!a is placed in prie area + the values for this area is used /ata for calculations and si#ulations

in this thesis are su##aried in Table C

** of ++

;i)ure >9 -he saled hourl $easure$ent data fro$ lava ener)

enter and #-+ used in the si$ulations.



-a*le 49 &rovided data for the thesis alulations and si$ulations.

Provided data Measurement period Motivation of data

@ourly #easure#ent of water level in hydro reservoir Two and a half yeardata

hanges in syste#" data only availablefro# this period

@ourly #easure#ent of active power output fro# hydro power plant Three year data

@ourly #easure#ent of active power output fro# wind power par!s Three year data

@ourly #easure#ent of power consu#ption Three year data

@ourly #easure#ent of active power output solar power fro# 6lavaenergy center

Tow year data ?o solar cells at STRI installed usedavailable data with sa#e nor#al sunshineti#e in a year

@ourly #easure#ent of active power output solar power fro# STRI @alf a year data Installed su##er 8%*8

@ourly #easure#ent of active power input to studied grid Three year data

@ourly #easure#ent of active power output fro# studied grid Three year data

3easure#ent of active power outco#e fro# power prediction ne year data 3easured in northeastern part of6er#any If sa#e forecasting #ethods isused do it not #atter were the research isdone

@istorical data of regulation in prie area + Three year data

@istorical data of up regulating pries in area + Three year data

@istorical data of down regulating pries in area + Three year data

4or the si#ulations of the *+% !V grid fi&ed values for the loads" lower voltage grids e&cept the

investigated $% !V grid" are used Typical values for the power trading are used and is provided fro#

%B Eln8rt. The values are seen in Table $

-a*le 79 ;i(ed loads in 13 k% )rid.

2oad ;ctive ower 3= Reactive ower 3V;r

* *A+%C C+$$

8 %*F$ %*8

+ $8F$ +8D+

C C A%B 8 D$B

apability curves show how #uch reactive power that can be produced0consu#ed depending on the

power output fro# a specific !ind of power plant The capability curve for a synchronous generator

can be seen in 4igure D ;s seen in the figure the power output is opti#ied to get as #uch active and

reactive power as possible The wind turbine has a power factor range of %Fcapacitive U %F$inductive with

the default set at *%% *FJ" this gives the capability curve for Vestas VF% as shown in 4igure F The

solar power plant has a power factor range of %Dcapacitive U %Dinductive which is seen in 4igure *% The

reactive power output fro# the #odeled STRI 2ithiu# Ion storage battery can be fully controlled ;s

seen in the capability curve in 4igure ** the battery converters can produce reactive power even

when the active power output is ero

*8 of ++



)2 Po'er.actory model

/igSI2E?T Power4actory is a power syste# si#ulation tool and according to their website 8BJ they

can handle Lstandard power syste# analysis needs" including high'end applications in new

technologies such as wind power and distributed generation and the handling of very large power

syste#sM 4or load flow calculation the progra# uses the ?ewton'Raphson #ethod

The si#ulated grid is showed in ;ppendi& ; The loads (arrows in the figure) are a representation of

the lower voltage grids The e&ternal grid (the chec!ered sHuares) represents the *+% !V connection

points" these uses power input data described in $* e&cept the right input that is used as slac! bus

The battery storage and the photovoltaic units are placed at STRI" at the C%% V level The si#ulated

*+% !V grid is shown in ;ppendi& :" where the e&ternal grid of C%% !V acts as a slac! bus The $% !V

grid e&ternal connections are here connected to the corresponding bus in the *+% !V grid

)# !odels of controllable units

The #odels for the controllable units the hydro power and battery storage #odeled in 3;T2;:" are

described in this chapter

*+ of ++

;i)ure ?9 a/a*ilit urve for the %estas %?.

;i)ure 19 a/a*ilit urve for the /hotovoltai units.

;i)ure 119 a/a*ilit urve for the *atterstora)e.

;i)ure 69 a/a*ilit urve for snhronous

)enerator.@34A



)#1 +ydro po'er plant model

4ro# the hydro power plant data" the efficiency (W) is calculated to be F*9 by EHuation *" were g is

the acceleration of gravity" X the density of water" P is the power output h the head and G the water

flow

4ro# the hourly water levels in the reservoir and the !nown power production (P) the change in

water level due to the hourly water inflows could be calculated with EHuation 8 There water leveli is

the level for the initial hour" water level iY* the level for the ne&t hour" h the head and ; the area of the

reservoir

The possible power output fro# the hydro power plant depends on the water level in the reservoir

The #odel starts with a calculation of the water inflow per hour The last hour inflow" production andwater level gives the head for the current hour" EHuation + :y co#paring the head with the allowed

water levels for the season" CC+'CC$ # in su##er and CC'CC$ # in winter" the #a&i#u# and

#ini#u# water flows is calculated according to EHuation C and $ It is assu#ed that the reservoir is

full when the si#ulations start 4ro# the water flow the #a&i#u# and #ini#u# possible power

output is calculated according to EHuation A and B

hi in the eHuations below is the height at the initial hour" h iY* the height at ne&t hour" h inflow the change

in water level fro# the inflow at the initial hour" G i the water flow at the initial hour" G #a& the

#a&i#u# allowed water flow at the initial hour" G #in the #ini#u# allowed water flow at the initial

hour" ; is the area of the reservoir" P2o #in0#a& is the #ini#u#0#a&i#u# allowed output power" W is theefficiency of the hydro power plant" X is the density of water and g is the acceleration of gravity

;fter calculation of the #a&i#u# possible power output for the initial hour the actual power output

is calculated by co#paring the desired grid load flow reduction with the possible #a&i#u# output If

the difference between desired reduction and possible power output is bigger than ero" the hydro

power plant can not co#pensate the whole overload by itself The hydro power plant will then set the

power output to the highest possible for the current hour If the difference is s#aller than ero the

power output fro# the power plant will be the #a&i#u# possible power output #inus the desired

reduction The #odel is su##aried in a flow diagra# in 4igure *8

*C of ++

20hourly water inflow = water level i+1

− water level i+ 3600 P

ηρ g h A

10 P =ηρ g h Q

40Qmini=

A(h i h inflow−h max)3600

30hi+1=hi+hinflow−3600Qi

A

50 P Lo max=ηρ g h i Qmax i

>0 P Lo min=ηρ g hi Qmini

70Qmaxi=

A(hi hinflow−hmin)3600



)#2 %attery storage model

To find the #a&i#u# possible power output the battery state of charge (So) is #ultiplied with the

energy in a fully loaded battery (+8 3=h)" which gives the a#ount of energy re#aining If the energy

is bigger than the battery #a&i#u# power output (D 3=) for one hour the possible output is put to

the #a&i#u# output ti#es the efficiency If the product is s#aller than the #a&i#u# power output"

but bigger than ero" the possible power output is set to the product ti#es the efficiency ;t last if the

product is s#aller or eHual to ero the possible power output is set to ero because it #eans that the

battery storage is e#pty The #odel is su##aried in the flow diagra# in 4igure *+

The battery charging #odel is described as a flow diagra# in 4igure *C If the grid load flow is lower

than a chosen li#it the battery is supposed to charge and thereby behave as a load The energy

*$ of ++

;i)ure 129 ;lo' dia)ra$ for the hdro /o'er $odel.

;i)ure 139 ;lo' dia)ra$ for the *atter dishar)in) $odel.

Yes No

Yes No

Calculating the energy left in the battery

Calculating maximal power output for the state of charge

Energy left in battery > maximal output for one hour

Possible output = maximal output times the efficiency Possible output = energy left times efficiency

Compare possible power output with need for compensation

Desired load flow reduction from grid > possible output

Power output = possible output Power output = reduction

Calculating new state of charge

Reduction > max power Reduction min power

!ydro power output = reduction !ydro power output = maximum

"inimum hydro power output reduction maximum hydro power output

!ydro power output = minimum

Calculate min#max allowed water flow

Calculate min#max allowed output power

Calculate desired load flow reduction

Compare need with min#max output power

Calculate water inflow

Decide min water le$els depending on season

Calculate head



needed to get a fully loaded battery is e#itted and co#pared to the #a&i#u# input power If the

needed energy is higher than #a&i#al loading power during the hour it is investigated if the grid load

flow with the battery charging at #a&i#u# input power is still s#aller than the low li#it In so#e of

the VPP #odels the high and low li#its is the sa#e If the new load flow is s#aller than the low li#it

then the battery input is put to the #a&i#u# input power If the new load flow is bigger than the low

li#it the battery input is set to the desired load flow reduction #inus the low li#it If the needed

energy is s#aller than the #a&i#u# loading power input it is investigated if the su# of desired

reduction and needed energy is s#aller or eHual to the low li#it If that condition is true" battery

input is set to the needed energy If the condition is not true" battery input is set to desired reduction

#inus the low li#it ;s the last step in the #odel the new state of charge is calculated

The state of charge calculates by subtract the latest So value with the Huota of the battery power

output and the energy in a fully loaded battery

)- /ptimi0ation applications for t"e &PP

4our different algorith#s for the VPP are progra##ed in 3;T2;: and the resulting power flows

loaded into the Power4actory #odel The different controls answered to different opti#iations

#eeting the day'ahead Spot #ar!et" described in $C8" lower the tariff ($C+)" lower the losses

($CC) and reactive power co#pensation ($C$) and is co#pared to a base case($C*) The hydro

power plant and storage wor!s the sa#e in all opti#iations Therefore the sa#e #odels" described

in $+" can be used even if prioritiation of which storage option is used first differs for all cases e&cept

for the base case

)-1 %ase case

The hydro power plant is regulated in the real grid To not get confusing results due to the regulation"

the hydro power plant is assu#ed to !eep the water level stable by useing the sa#e water volu#e as

the hour water inflow The produced energy then corresponds to the inflow in 4igure $

The wind and solar production and the loads are assu#ed as non adjustable and therefore the

#easured production" seen in 4igure A" B and +" is used in all calculations The battery storage is in

this case assu#ed to be disconnected fro# the grid

*A of ++

;i)ure 149 ;lo' dia)ra$ for the *atter har)in) $odel.

YesNo

%attery input = reduction & lower limit

Reduction ' maximum loading power = lower limit

%attery input = ( maximum loading power %attery input = ( needed energy %attery input = reduction & lower limit

)tate of charge *

Desired load flow reduction from grid Needed battery energy > maximal loading power %attery input = +

Yes No

Reduction ' needed battery energy = lower limit

Yes No

Yes No

Calculate state of charge



)-2 esponsible party to meet placed bids on t"e daya"ead spot market

The goal of this opti#iation is to #ini#ie the prediction error and reduce the regulation cost In this

second application two #odels are investigated to see the difference between having a VPP with both

hydro reservoir and battery storage and with only battery In the first #odel the hydro reservoir and

battery storage are wor!ing together to reduce the prediction error fro# the day'ahead spot #ar!etand in the other #odel only battery storage is used to co#pensate the prediction error fro# the wind

and solar

/ata for predicted production and actual production for wind power in the northeastern part of

6er#any during 8%%F is used to create a #odel for the prediction error" 4igure *$" *J It was found

fro# +%J that the prediction error of the solar" 4igure *A" power have si#ilar behavior as the wind

power so sa#e #odel for prediction error is used The prediction outco#e data is divided into eleven

parts to find the probability curves for different a#ount of production" the eleven parts is then

nor#al distributed" 4igure *B :ecause the actual production for the wind and sun is !nown and the

prediction data is not available the predicted production is rando#ly chosen fro# the nor#al

distributed curve fitted for the a#ount of production The prediction fault is then calculated for the

different /ER and then su##aried to get the power needed for co#pensating the error

The battery can regulate both over and under production and therefore the battery should have the

goal not to be fully charged @ydro power is used to co#pensate over consu#ption The battery is

considered to be the pri#ary regulation source and the hydro reservoir the secondary

*B of ++

;i)ure 179 Measure$ents of /redited and atual /rodution for 'ind

/o'er in the northeastern /art of er$an.



3odel with both hydro and storage7

*D of ++

;i)ure 169 ;lo' dia)ra$ for the $eetin) the daahead $arket 'ith hdro and stora)e re)ulation $odel.

,ctual production

Chose normal distribution cur$e

-et a random number from the distribution cur$e for the predicted wind and solar production

.ind the prediction error by subtract the actual power with the predicted power

Error + Error = + Error > +

!ydro = minimum output%attery /ero or loading to compensate

the error from hydro

!ydro =production corresponding tocalculated inflow

%attery not re0uired e0ual /ero

!ydro = minimum output

%attery output to compensate the rest

of the error1 "ore hydro if necessary


;i)ure 159 o$/aration of atual &% /rodution "Leistun)

)e$essen"0 and &% /ro)nosis "Leistun) *erehnet"0.

;i)ure 1>9 Nor$al distri*ution urve of /rodution

o$/ared to /redition, hi)h /rodutions.



If the production of wind and solar is bigger than predicted for solar" wind and hydro then the hydro

power produce #ini#u# power output and the battery is charging If the #ini#u# power output

fro# the hydro power plant is bigger than ero the battery have to try to co#pensate for that as well

If the battery is fully charged nothing can be done to reduce the prognosis error in the solar and wind

production

If the production is s#aller than predicted the #ini#u# hydro power is e#itted and it is investigated

if the battery can co#pensate for the prediction error If the battery is insufficient the hydro plant

produce #ore to help co#pensate for the rest of the prediction error If prediction and actual

production is the sa#e then the power output fro# hydro is corresponding to the power inflow and

the battery is not reHuired This #odel is su##aried in 4igure *D

3odel with only storage7

In this #odel it is assu#ed that the hydro power plant has no prediction error The day'ahead Spot#ar!ed bid is assu#ed to correspond to the water inflow The water reservoir is not used forco#pensate the prognosis error of wind and solar production

4irst a rando# prediction error is added to the actual production If the production is bigger than

predicted the battery is charged If the battery beco#es full despite the control algorith#

opti#iation there is no longer anything that can be done to reduce the prognosis error If the

production is s#aller than predicted the battery co#pensates for as #uch as it can 4inally if

prediction and actual production is the sa#e the battery output is ero This calculation is

su##aried in the flow diagra# in 4igure *F

In both #odels different capacities of battery storage power output is used The used storage

capacities are % 3=" 8 3=" C 3= and D 3= with 8 and C hour storage

To calculate the regulation cost the infor#ation fro# C8* is used The prediction error is co#pared

to the regulation history fro# the spot #ar!et to see if the errors will reHuire the plant owner to pay

for the regulation or not If the error is the sa#e as for the entire price area then the price is applied

fro# the balance #ar!et data for up or down regulation" depending on the error" and #ultiplied by

the error volu#e to get the cost This calculation is su##aried in the flow diagra# in 4igure 8%

*F of ++

;i)ure 1?9 ;lo' dia)ra$ for $eetin) daahead $arket 'ith a %&& 'ith stora)e re)ulation.

,ctual production

Chose normal distribution cur$e

-et a random number from the distribution cur$e for the predicted production

.ind the prediction error by subtract the predicted power with the actual

Error + Error = + Error > +

%attery discharge to compensate the error %attery output e0ual + %attery loading to compensate the error




)-# Peak s"aving

The pea! shaving #odel investigates the potential of the VPP to restrict #a&i#u# power at one

connection point to the *+% !V grid This is done to reduce the subscribed power and obtain a lower

tariff for the operator of a 3V grid In Sweden the tariff is only based on the hour of each #onth with

#a&i#u# load flow as described in C88

4orecasts for CD'B8 hours are often used for energy trading and longer ti#e scales" up to $'B days

ahead" #ay be used for e&a#ple planning the #aintenance of wind far#s" conventional power plants

or trans#ission lines +*J It is therefore hard to predict when during the #onth pea! flow will occur

and therefore historical data to set li#its for when charge and discharge shall occur is used

The power input needed fro# the *+% !V networ! is found by co#paring the loads with the

production of the wind far# and photovoltaic cells To define li#its the highest value for each #onth

is found and then the #ean pea! of the #onth over the three years is obtained Si#ilarities between

so#e #onths is obtained" for e&a#ple /ece#ber to 4ebruary and <uly to Septe#ber" are grouped In

/ece#ber the second year the pea! is e&tre#ely low co#pared to the other years and co#pared to

pea!s in <anuary and 4ebruary" therefore the #ean of the #ean pea! for these #onths is calculated

; high li#it is choosen to be between A and D 3= lower than the #ean pea!" how #uch lower

depends on the a#ount of high pea!s The choice of A'D 3= is due to the storage #a&i#al output

The low li#it is chosen to be $ 3= lower than the high li#it because the charging of the battery

should not contribute to the highest pea! of the #onth The goal is to reduce the highest pea! for

every #onth" a table over the pea! values and the chosen li#its can be seen in Table A

8% of ++

;i)ure 29 ;lo' dia)ra$ for alulation of re)ulation ost.

2f price area error +

2f prediction error + 2f prediction error > +

Price = +Price = up price

Yes YesNo No

Cost = price 3 prediction error

2f price area error = + 2f price area error > +

Price = down price

-a*le 59 #hoes the hi)hest /eak /er $onth under the $easure$ent /eriod and the hosen li$its for the /o'er in/ut fro$ the

e(ternal )rid.

Month Peak Year 1[MW]

Peak Year 2[MW]

Peak Year 3[MW]

Mean peak [MW]

High limit [MW]

Low limit [MW]

<anuary C+*8 CADA C%*$ C++D +A +*

4ebruary C+FA C$AD +F%$ C8F% +A +*

3ars +$88 +A%$ +8$% +C$F 8A 8*

;pril 8BC+ 8DA8 ++DC 8FFA 8+ *D

3ai 8++$ +%FD 8AFD 8B*% 8* *A

<une 8C8F 8$AA ++$F 8BD$ 8* *A

<uly 88AA 8C8+ 8+C% 8+C+ *$ *%

;ugust +**C 8*8$ 8C$C 8$AC *$ *%

Septe#ber 8DDA *BA$ 8*F8 88D* *$ *%

ctober 8F*B 8B8A 8B** 8BD$ 8* *A

?ove#ber +AC8 +%*F 8FD* +8*C 8A 8*/ece#ber CD$% +%BA +F*+ +FCA +A +*



The #odel is based on three different cases The first case represents if the power input fro# the

e&ternal grid is between the power li#its" the second case if the input is too high and the third if the

power input is s#aller than the lower li#it The following calculations are #ade for all hours in the

three year period and is su##aried in 4igure 8*

:etween li#its7

In the first case the power input fro# the e&ternal grid is found acceptable and then there is no need

for co#pensation The power output fro# the hydro power plant is put to the #ini#u# poweroutput value to avoid too high water levels in the reservoir If the #ini#u# producing value eHual

ero then the battery is also put to ero If the hydro power plant has to produce so #uch that the

load flow co#es under the low li#it then the battery storage is charging

;bove high li#it7

In the second case where the power input fro# the e&ternal grid is over the high li#it the VPP

atte#pt to increase the production to lower the power input The atte#pted power input reduction is

here called the overload

If the hydro power plant can not co#pensate for the co#plete overload it is investigated how #uch

the battery can co#pensate The co#pensation is calculated by the difference between the overload

and the battey possible power output If this difference is bigger than ero the battery cannot

co#pensate for all The battery output is put to the highest possible therwise" if the difference is

s#aller than ero" the battery can co#pensate for the rest The power output is then put to the

difference between the possible output and the re#aining overload The new So is then calculated

by subtract the latest state of charge value with the Huota of the power output and the energy in a

fully loaded battery If the co#plete overload cannot be co#pensated is the #onth high li#it

updated to the pea! value after the co#pensation

8* of ++

;i)ure 219 ;lo' dia)ra$ for the /eak shavin) $odel.

NoYes

NoYes

NoYes

Calculate difference between production and consumption

4oad flow between limits 4oad flow under lower l imit 4oad flow abo$e higher limit

Calculate hydropower head5 min#max power flow Calculate the desired load flow reduction

Calculate hydropower head5 min#max power flow

Calculate hydro power output and non fulfilled reduction

Non fullfilled reduction

Calculate battery power output %attery output = +


"inimum hydropower flow > +

!ydropower output = min flowReduction = ( min flow

!ydropower output = +Reduction = +

Reduction ' compensation lower limit

4oad battery %attery output = +



:elow low li#it7In the third case the power input fro# the e&ternal grid is lower than the lower li#it In this case no

power production is needed fro# the hydro power plant" so its power output is put to ero (if the

reservoir allows it) and otherwise to the #ini#u# power output The battery #ight need to be

loaded The used battery energy is calculated in percent If the power input fro# e&ternal grid plus the

product of the percent and #a&i#u# power input in the battery is s#aller or eHual to the lower li#it

then the battery power output is put to #inus the percent ti#es the #a&i#u# input The #inus sign

shows that the battery is loading which can be seen in the So calculations described in $+8 If the

su# is bigger than the low li#it the power input to the battery is put to the power input fro# e&ternal

grid #inus the lower li#it If the battery is fully loaded the battery power is put to ero The new So

is then calculated

)-- 3ecrease net'ork losses

The goal of this opti#iation is to reduce the losses in the overlying networ! without increasing the

losses in the own grid too #uch The application will only be econo#ical if the loss reduction is largerthan the increasing losses in the VPPZs grid plus the battery losses

?etwor! losses consist of series losses and shunt losses =ith energy storage it is the series losses

that can be reduced The average series losses are described in EHuation D where it is assu#ed that

the series resistance R and the voltage 1 are constant EHuation D uses the standard deviation of a

series #easure#ent" EHuation F In the eHuation I sy#bolie the current" P the active power" G the

reactive power" 1 the voltage" R the resistance and [ the standard deviation +8J

The eHuations show that there are three contributions to the series losses7

* 2osses due to the energy transfer proportional to the sHuare of the trans#itted energy (P)

8 2osses due to variation in the transferred energy" proportional to the sHuare of the standard

deviation of the active power (P U P)

+ 2osses due to reactive power

Point nu#ber one is hard to affect because no de#and response is used in the #odel In this chapter

point nu#ber two is investigated and nu#ber three is studied in chapter $C$

To reduce the losses due to variation the daily variation of the load is studied The pea!s and troughs

are identified and the available energy for the battery is calculated The level for reducing the

#orning or evening pea! is chosen so the battery will be fully loaded in the ti#e between the pea!s

The goal is to #eet the ne&t day with a full battery so it could reduce the #orning pea! even if it

would co#e early In 4igure 88 it is shown how the unevenness in the transfer of energy is handled by

#oving the need of energy inflow fro# the pea! hours to the troughs hours The battery regulation

for this #odel is seen in 4igure 8+

88 of ++

?0σ p2=( P − P )2

60 I 2

R= P

2+Q2

U 2 R=

R

U 2(( P )

2

σ p

2

+Q2

)



=hen the power schedule for the battery is decided it investigates if the load with the battery

reduction is larger than the high li#its chosen in chapter $C+ If so the hydro power plant will try to

reduce the need of power inflow to the high li#it as described in $+* In the base case the hydro

power plant produce only power corresponding to the water inflow If the hydro production is bigger

in the base case than in this #odel when the power inflow fro# *+% !V grid is over the high li#it" the

production in this #odel will be eHual to the production in the base case (if the water level will

allowed it) If the needed power inflow is s#aller than the high li#it the power output fro# the hydro

is put to %%$ 3= less than in the base case" if the water level described in $+* allowed it" so that

the reservoir will have a chance to recover The flow diagra# for the hydro power output in this

#odel is seen in 4igure 8C

8+ of ++

;i)ure 229 !n e(a$/le fro$ 17 ;e*ruar 211 of the load dail

variations 'ith and 'ithout *atter re)ulation.

;i)ure 239 ;lo' dia)ra$ for the *atter re)ulation in the loss redution

$odel.

.ind e$ening pea6

.ind the balance between the pea6 and the trough until midnight

Put the high and low limit to the balance power

Calculate the battery output and input

.ind morning pea6

.ind the balance between the pea6 and between pea6s

Put the high and low limit to the balance power

Calculate the battery output and input



; Power4actory si#ulation with and without the *+% !V grid calculates the losses in the grids The

si#ulation is then co#pared to the si#ulation of the base case to investigate the loss reduction fro#

the VPP

)-) eactive po'er control

This #odel is an e&tension of the #odel in $CC The possible production of reactive power is

calculated fro# the active power output at the production sources in previous chapter and the

capability curves in $*

In the Power4actory si#ulation the capability curves is used to analye the loss reduction in the

overlying grid The grid is si#ulated for the base case ($C*) and the loss reducing case ($CC) with

and without the *+% !V grid described in $8

4 esults

In this chapter the si#ulations result for the different opti#iations are presented and analyed

41 esponsible party to meet placed bids on t"e daya"ead spot market

The si#ulations with the control for #eeting the bids on the day'ahead spot #ar!et described in

$C8 show that it is better to use both hydro power and battery storage as regulation to co#pensatefor the prediction error than only use battery storage

In the #odel with both hydro and storage regulation the hydro power plant can co#pensate the

prediction error if the production outco#e is s#aller than predicted" then the hydro power plant and

the battery can co#pensate for the error together If the production is less than predicted the hydro

power can only co#pensate a little by producing less than the inflow" if the water level in the

reservoir allow it" and the battery has to do the rest =hen changing the storage capacity the hydro

power plant can not co#pensate for #ore than for the other cases The si#ulation is #ade for both

two and four hour storage capacity to evaluate that dependency ;s seen in 4igure 8$ and 8A the

difference in the #ean absolute error between the two and four hour storage s#all The largest

8C of ++

;i)ure 249 Hdro /o'er out/ut for the loss reduin) $odel.

Calculate difference between production and consumption

4oad flow under high limit

Power output without regulation & +1+7

minimum output

4oad flow abo$e high limit

Calculate hydropower head5 min#max power flow

Calculate the desired load flow reduction

Calculate hydro power output

NoYes

2f output without regulation > with regulation

!ydropower output = without regulation !ydropower output = with regulation

"inimum output

Power output without regulation & +1+7

"aximum output

Power output without regulation & +1+7>

minimum output

Power output = Power output without regulation & +1+7

Power output = minimum output Power output = maximum output



difference occur at the D 3= storage where the four hour storage lower the absolute #ean error

with *$ != #ore than the two hour storage

The regulation cost is calculated for the different #odels of storage and hydro power" the result for

the C hour storage is seen in 4igure 8B The cost showed in the figure is a #ean yearly cost in #illion

Swedish crowns based on the three years of si#ulation ;ccording to these si#ulations a VPP with an

D 3= 0+8 3=h storage and a hydro power plant with A 3= capacity can reduce the regulation cost

with over $% 9 ; VPP with only the D 3=0+8 3=h storage regulations can reduce the cost with just

under $% 9 :ecause of the rando#ly chosen predicted production the results are based on the #ean

of three si#ulations

8$ of ++

;i)ure 279 -he $ean of a*solute /redition fault in MC 'ith different

stora)e a/ait 'ith 2 hour stora)e and hdro re)ulation.

;i)ure 259 -he $ean of a*solute /redition fault in MC 'ith different

stora)e a/ait 'ith 4 hour stora)e and hdro re)ulation.



To evaluate the #odel the prediction errors larger than the regulation capacity is su##aried and the

share of all prediction errors is e#itted :y adding the share of reduced prediction errors with the

share of the errors that are not possible to co#pensate due to power li#its an efficiency of the #odel

is esti#ated to around F% 9 for all cases In 4igure 8D the reduction of prediction error (green solid

line) is shown with the error larger than the regulation (red dotted line)

;n evaluation if the regulation cost can be further reduced by only co#pensating for the prediction

error liable for pay#ent shows no cost saving This is because the battery will not be discharged as

often as when it co#pensate for all errors and then it cannot co#pensate as #uch for the case where

the production is higher than predicted

This #odel reduces the regulation cost with average 8D 3SE5 and the tariff with average *C% !SE5

per year

8A of ++

;i)ure 269 -he /redition error o$/ensated * the %&& 'ith 4 hour

*atter stora)e 'ith different a/ait and hdro re)ulation and the

errors that is not /ossi*le to o$/ensate 'ith the %&&.

;i)ure 2>9 -he re)ulation ost over the $easure$ent /eriod 'ith 4 hour

stora)e *ut different a/ait and hdro re)ulation.



42 Peak s"aving

The pea! shaving #odel described in chapter $C+ is used to restrict #a&i#u# power at the

connection point to the *+% !V networ! The power de#and in the connection point with and

without a virtual power plant can be seen in 4igure 8F The VPP reduced the #onth highest pea!

de#and on average by 8% 9 over the three years of #easure#ent

The corresponding saving in tariff to the sub trans#ission networ! operator is esti#ated by the /S

to be in average CC% !SE5 per year The tariff saving per #onth is seen in 4igure +%

This #odel will reduce the tariff with CC% !SE5 for the /S but increase the regulation cost for the

VPP owner with +A% !SE5

8B of ++

;i)ure 2?9 &o'er in/ut in the 13=7 k% onnetion 'ith and 'ithout a

virtual /o'er /lant.

;i)ure 39 -he %&& redution of the tariff for the D#.



4# 3ecrease of net'ork losses

The si#ulation results for the loss reduction #odel are seen in Table B The difference in total losses

for the two cases in the *+% !V grid is co#pared to the difference in total losses for the $% !V grid

The losses in the *+% !V grid reduces with *%C 3=h with the loss reducing #odel but the losses in

the $% !V grid increases with 8C* 3=h The #odel does decrease the losses in the overlying grid butit is not efficient because the loss increase in the grid is larger than the reduction

-a*le >9 #i$ulation results for the t'o ases si$ulated in the 7 k% and 13 k% )rid.

:ase case 6=hJ 2oss reduction case 6=hJ /ifference 3=hJ

Total E&ternal Infeed *+% !V

grid

*+%+ *+%% '+8C*

Total 2osses *+% !V grid *A$8 *AA$ *+B%

Total E&ternal Infeed $% !V grid +%8B 8B*+ '+*+B

Total 2osses $% !V grid 888F 88$+ 8C**

4- eactive po'er control

:y using the converters ability to produce or consu#e reactive power according to the capability

curves in chapter +8 reactive power de#and in the *+% !V connection point can be reduced The

supply of reactive power fro# the sources of production co#pared to the load" gives that the reactive

power de#and fro# superior networ! on average decreases by B$9 4igure +* shows grid reactive

power de#and (upper red curve) and the reactive power de#and if the VPP regulate the need with

the converters (lower blue curve)

The si#ulation results for the loss reduction #odel with reactive power production are seen in Table

D The difference in total losses for the two cases in the *+% !V grid is co#pared to the difference in

total losses for the $% !V grid o#pared to the #odel without reactive production reduced losses in

the overlying grid is larger in this #odel The needed reactive power transferred through the

overlying grid is lowered in this #odel when the converters produce and the VPP regulate the

8D of ++

;i)ure 319 -he reative /o'er de$and 'ith and 'ithout %&& re)ulated

reative /o'er /rodution.



reactive power :ecause the reactive power causes losses" a production near the need will reduce the

losses in the connected grid The #odel is however still not efficient because the losses in the *+% !V

grid reduces with *%$+ 3=h but the losses in the $% !V grid increases with 8+++ 3=h

-a*le 69 #i$ulation results for the t'o ases si$ulated in the 7 k% and 13 k% )rid 'ith /rodution of reative /o'er.

:ase case 6=hJ 2oss reduction case 6=hJ /ifference 3=hJ

Total E&ternal Infeed *+% !V grid *+%+ *+%% '+8$%

Total 2osses *+% !V grid *AFD *B** *8D*

Total E&ternal Infeed $% !V grid +%$C 8B+F '+*C$

Total 2osses $% !V grid 88$A 88D% 8+++

5 3iscussion

In this chapter the i#ple#entation of the VPP is discussed in regard to the need of infor#ation

6eneral thoughts of the result are also discussed in this chapter

51 Implementation of t"e &PP

The reHuired infor#ation to perfor# the applications described in $C consists of both co##ercial

data fro# #ar!et traders" real'ti#e infor#ation fro# the included /ER and #easure#ent of power

flow and voltages in the distribution grid The VPP needs to co##unicate with either the /Ss

S;/; syste# or eHuip#ent in the grid in order to gather infor#ation There is need for infor#ation

e&change between the VPP and #ar!et actors The reHuired infor#ation to perfor# the applications

described in $C is su##aried in Table F So#e infor#ation reHuired with a \] is not essential for thebasic application" but can be used to i#prove perfor#ance as discussed in B+

The need for fast real'ti#e infor#ation is not anticipated in this VPP /ata on the grid power flow and

/ER production are probably sufficient on #inute scale 3inute average of voltages in the grid is

li!ewise seen as sufficient for the studied applications as well as for #easure#ents of the energy in

the reservoir and battery storage

To act as balancing party and co#pensate for prediction errors in day ahead and intraday bids" *$

#inute update intervals are e&pected to be enough for production and storage set'points as well as

for reactive power #easure#ent Infor#ation on the latest regulating volu#es can be retrieved fro#

?ord Pool Spot website ++J with *'8 hours delay

The highest pea! so far in the #onth will be stored and updated when a #ajor pea! is reached ;t the

end of the #onth the highest pea! is added to the historical values and a new start high value is

updated for the sa#e #onth ne&t year

8F of ++



52 6eneral t"oug"ts

; battery energy storage syste# is an e&pensive invest#ent To buy storage of the sie studied in this

thesis for the purpose of using it in a virtual power plant will not be econo#ical today =ith further

develop#ent of battery storage and increased need for controllable syste#s in distribution networ!s

to handle the inter#ittent distributed energy sources a VPP can be financially viable

5# Possible improvements of t"e models

o#pensation need could be detected in advance with a co#parison of weather data that production

prediction were based on and real ti#e weather data" or short ti#e prediction" would give an

indication if the predicted production is reasonable or the direction of the change

+% of ++

-a*le ?9 +nfor$ation needed for i$/le$entation of the different a//liations for the %&&.

Information Application escription ata source

CC8 CC+ CCC CC$

Predicted ProductionO O O

O

:attery storage State'of'hargeO O O

;#ount of Stored energy in :ESS

@ydro reservoir water level

O O O

Real 'ti#e production of /ER

O O O O

]

]

Electricity retai le rs

Real'ti#e weather data

]

@istorical levels of power flow

O O

Power flow in grid location(s )

O O

Real'ti#e voltage at /ER

]

O

Real'ti#e voltage in grid

]

:id on the da y'ahea d spot #ar!et

(8C bids placed *8 a# for ne&t day)

4orecasting tool us ed

by Electricity retail ers

Intraday correction of predicted

Production

:id on intraday #ar!et (* bid pe r

hour" la test an hour before period

starts)

4orecasting tool used

by Electricity retailers

:attery storage

controller

;#ount of Stored e nergy in hydro

reservoir

@ydro p roducer or

controller of hydro

plant

To deter#ine accu#ulated

production wi thin pres ent hour

;lso used together with latest bid

to esti#a te production for present

hour

S;/; or /ER

controllers

2atest Regula ting volu#e for

price area

To predict if over or under

production (co#pared to predi ction)

will result in econo#ic cost

Electricity #ar!et

bro!er

=eather data used for retailersprediction

=ind spee d" rainfal l" irradiation etcon whi ch production predi ction

where ba sed 1sed to esti#ate i f

bids during present hour are

accurate

;ctual wea ther data within hour to

enable esti#ation of prediction

error for present hour

3easure#ents at

/ER or #etrological

bureau

/eter#ine i f VPP shal l a tte#pt to

lower pea ! flows to reduce tariff or

loss es :ased on hi storical hourly

power flows or logged data

VPP controlle rs l og of

real ti #e production

@ighes t power flow during

current #onthO

:loc! power flow reduction i f

#onthly #a&i#u# will not be

e&ceeded

VPP controlle rs l og of

real ti #e production

/eter#ine if pea! power flow and

high losses are occurring

S;/; or IE/

#easure#ent withingrid

alculate losses #ore accurately

when voltage varies considerably at

/ER

S;/;" /ER control le r

or IE/ nea r /ER

Reactive power in selected

location(s)

Reactive power flow a t a point(s) in

the grid for which reactive power

e&change is to be opti#ied

S;/; or IE/

#easure#ent within

grid

alculate ΔQ reHuired fro# /ER to

hold voltages in s ele cted point(s) of

the grid wi thin l i#its

S;/;" IE/ i n grid o r

s#a rt #etering

syste#



The #odel in $C+ could be i#proved not only by the use of historical data but also by forecasts for

s#aller ti#e scales" li!e weather and consu#ption predictions If historical data fro# #ore years is

used the #odel could be better opti#ied

In the si#ulated *+% !V grid the consu#ption fro# the lower voltage grid is put to fi&ed values as

seen in Table $ The #odel could be i#proved by using the #easure#ent values at the connectionbuses =ith data for reactive power in the si#ulated buses the #odel for the *+% !V and $% !V grid

would be i#proved

7 Conclusion

This #aster thesis has showed possible applications for a virtual power plant in 2udvi!a :enefits of

several #illion SE5 per year have been found

:y using active control to #eet the bids on the day ahead #ar!et the hydro power plant can reduce

the regulation cost by 8% 9 The hydro power plant together with the D 3= four hour battery storagecan reduce the regulation cost with over $% 9 The regulation cost will not be further reduced by only

co#pensate for the prediction errors with the sa#e error as the price area

The regulation cost reduction is linearly dependent of the battery power output capacity The

difference between two and four hour storage is s#all if the storage is used to co#pensate prediction

errors If the storage is used for a pea! shaving application the sie of the storage is #ore i#portant

The VPP reduces the power de#and by FC 9 of the #onthly pea!s with the si#ulated #odel The

#onth highest pea! de#and is reduced on average by 8% 9 This #eans a reduction of the /S tariff

with average CC% !SE5 per year

The loss decreasing #odel reduces the losses in the overlying grid The application is not efficientbecause the loss increase in the $% !V grid is larger than the loss reduction in the *+% !V grid Even if

the VPP could s#other the load curve to one value for the day" the application would not be efficient

because the loss increase in the $% !V grid would still be larger than the reduction in the *+% !V grid

To produce0consu#e reactive power in the $% !V grid the losses in the overlying grid decreased #ore

than without production0consu#ption" but the #odel is still not efficient The production

0consu#ption can instead be opti#ied to reduce the reactive power de#and fro# the *+% !V grid

The average reactive power de#and can al#ost be reduced to ero with the VPP

The applications #ost fitted for a VPP in 2udvi!a is" according to this thesis" to co#pensate for

prediction errors The cost for regulation is reduced with $% 9 for the owner of the VPP and the tariff

is reduced with average *C% !SE5 for the /S with this application

+* of ++



eferences

*J 3 @ :ollen" +nte)ration of Distri*uted eneration in the &o'er #ste$ <ohn =iley K Sons" 8%**8J 5 E :a!ari and = 2 5ling" LVirtual power plants7 ;n answer to increasing distributed generation"M in

+nnovative #$art rid -ehnolo)ies onferene Euro/e +#- Euro/e0, 21 +EEE &E#" ct" pp *UA+J :ang" 4 4oc!" and 3 Togeby" L/esign of a real ti#e #ar!et for regulating power"M Ea Energy

;nalyses" ?o +" /ec 8%**CJ L4eni&M nlineJ ;vailable7 http700feni&iwesfraunhoferde0ht#l0whatht# ;ccessed7 8D'4eb'8%*+J$J Lell ontroller verview and 4uture Perspectives"M Energinetd!" 8%B+$0*8 ase *%0ABD" Sep 8%*8AJ R :elho##e" Tranchita" ; Vu" < 3aire" and @uet" Lverview and goals of the clusters of s#art

grid de#onstration projects in 4rance"M in 211 +EEE &o'er and Ener) #oiet eneral Meetin)" 8%**"pp * UD

BJ : :uchhol and 1 Schluec!ing" LEnergy #anage#ent in distribution grids European cases"M in +EEE

&o'er En)ineerin) #oiet eneral Meetin), 25" 8%%A" p 8 ppDJ Prof /r hristian <-nig" LVo# Virtuellen 5raftwer! u# vernetten Energiesyste#"M Stadtwer!e 1nna

6#b@" 8%%AFJ LruhrenergyM nlineJ ;vailable7

http700wwwruhrenergyde0visitorcentre0proje!te0proje!tdetailphp]lang^enKid^CF ;ccessed7 8B'4eb'8%*+J

*%J ? / @atiargyriou" ; 6 Tsi!ala!is" E 5arfopoulos" T 5 To#tsi" 6 5aragiorgis" hristodoulou" and; Poulli!!as" LEvaluation of Virtual Power Plant (VPP) operation based on actual #easure#ents"M in>th Mediterranean onferene and E(hi*ition on &o'er eneration, -rans$ission, Distri*ution and

Ener) onversion Med&o'er 210" 8%*%" pp * UD**J : Roossien" 3 @o##elberg" =ar#er" 5 5o!" and <'= Tur!stra" LVirtual power plant field

e&peri#ent using *% #icro'@P units at consu#er pre#ises"M in #$artrids for Distri*ution, 26. +E-

+ED. +ED #e$inar " 8%%D" pp * UC

*8J : Roossien" ; van den ?oort" R 5a#phuis" 4 :lie!" 3 Eijgelaar" and < de =it" L:alancing wind powerfluctuations with a do#estic Virtual Power Plant in Europes 4irst S#art 6rid"M in &o'er-eh, 211 +EEE

-rondhei$" 8%**" pp * U$*+J : :1@@2_" ? @;T_I;R6RI1" I 41R?ES 4;RTS" and 1 S@21E5I?6" L2ESS?S 2E;R?E/7

E1RPE;? PI2T I?ST;22;TI?S 4R /ISTRI:1TE/ 6E?ER;TI? ' ;? VERVIE= : T@E IRE/21STER"M 532" vol I6RE 8%%A

*CJ S :erlijn" S 3ar!alous" and 5 Stro#" L=eb'based P/ #onitoring of a generator in 2oforsen Sweden"Min +nternational onferene on ondition Monitorin) and Dia)nosis, 26. MD 26" 8%%D" pp *ABU*B%

*$J 3attias @ol#stro#" LV: !raft"M %D';pr'8%*+*AJ Vattendo#stolen Stoc!hol#s tingsr-tt" %attenhushFllnin) oh kontroll *FDD

*BJ L4j-llberget0Sa&berget `Stena Renewable"M %indkraft 8r nGdv8ndi)t nlineJ ;vailable7http700stenarenewablese0proje!t0fjallberget0 ;ccessed7 8D'4eb'8%*+J*DJ L6eneral Specification VF% U *D08% 3= ptispeedT3 U =ind TurbineM Vestas =ind Syste#s ;0S"

%B'Sep'8%%$*FJ IES" LSyste# I#pact ;ssess#ent Report (;ddendu#)M Vestas =ind Syste#s ;0S" 8C'Sep'8%*88%J S3; Solar Technology ;6" LReactive Power and 6rid Integration with S1?? 3I?I E?TR;2 and

S1?? TRIP=ERM *A';pr'8%*%8*J G2pe etal" L/etailed /esign :ESS onverterM inergia" *$'<an'8%*+88J Pia :org" L4.rstudie Energilager anslutet till vind!raft"M Elfors!" *87CC" Sep 8%*88+J Southern alifornia Edison" LTechachapi =ind Energy Storage Project 2ithiu#'ion battery and s#art

inverter technologies will i#prove grid perfor#ance and wind integrationM 1S /epart#ent of

energy" ct'8%*88CJ L?ord Pool SpotM nlineJ ;vailable7 http700wwwnordpoolspotco#0@ow'does'it'wor!0/ay'ahead'#ar!et'Elspot'0 ;ccessed7 88'3ay'8%*+J



8$J L?ord Pool SpotM nlineJ ;vailable7 http700wwwnordpoolspotco#03ar!et'data*0/ownloads0@istorical'/ata'/ownload*0/ata'/ownload'Page0 ;ccessed7 %A'3ay'8%*+J

8AJ LPrislistor S.der fr,n 8%*+'%C'%*M Vattenfall" %*';pr'8%*+8BJ LElectricity ;ct Energi#ar!nadsinspe!tionenM nlineJ ;vailable7

http700eise0en0Publications02aws'and'regulation0electricity'act0 ;ccessed7 %8'3ay'8%*+J

8DJ 6lava Energy enter" LSolpar! * och 8M nlineJ ;vailable7http700wwwglavaenergycenterse0inde&php0proje!t0sol#odulspar!en ;ccessed7 *D';pr'8%*+J

8FJ L?or#al sols!enstid f.r ett ,r Str,lning 3eteorologi 5li#atdata S3@IM nlineJ;vailable7 http700wwws#hise0!li#atdata0#eteorologi0stralning0*+%$8 ;ccessed7 *A'3ay'8%*+J

+%J hristoph 6RISS etal" LP@TV2T;I5'ER_E161?6 4R EI?E RE6E?ER;TIVEV22VERSR61?6 STERREI@SM Sy#posiu# Energieinnovation" %8'/ec'8%*%

+*J S5 ;ggarwal and 3eenu 6upta" L=ind Power 4orecasting7 ; Review of Statistical 3odels"M+nternational Journal of Ener) #iene" vol +" no *" 4eb'8%*+

+8J ?icholas Etherden and Susanne ;c!eby" L;cceptansgr-ns vid anslutning av distribuerad

generering"M Elfors! rapport *87CC U;ppendi& *" Sep 8%*8++J L?ord Pool SpotM nlineJ ;vailable7 http700wwwnordpoolspotco#03ar!et'data*0Regulating'Power*0Regulating'Volu#es*0;220@ourly0 ;ccessed7 88'3ay'8%*+J

+CJ LIngeniera e InvestigacinM nlineJ ;vailable7http700wwwrevistasunaleduco0inde&php0ingeinv0article0view08$8*80++B*A ;ccessed7 %C';pr'8%*+J

++ of ++



Appendix A (imulated middle voltage grid

;i)ure !.19 -he si$ulated M% )rid 'ith the virtual /o'er /lant.



Appendix % (imulated 1#* k& grid

;i)ure B.19 -he si$ulated 13 k% )rid.

*

8 +

C



Quantification of Ancillary Services from a VirtualPower Plant in an Existing Subtransmision Network

Nicholas Etherden, Math H.J. Bollen

STRI ABGothenburg, Sweden

[email protected]

Johanna Lundkvist

Uppsala UniversityUppsala, Sweden

Abstract — This paper presents the results of a feasibility study of

a virtual power plant (VPP) in central Sweden designed to

provide ancillary services to a 50-kV distribution network. The

VPP consists of a wind park, hydro plant and reservoir as well

as solar PVs and battery energy storage. The 50-kV

subtransmission network was modeled in order to evaluate the

ancillary services that could be provided by coordinating

existing distributed energy resources in the network.

Simulations were performed using measured hourly variations

in production and consumption at all network nodes. Thestudied ancillary services include both reactive and active power

control. Contribution from the VPP is evaluated for 1)

balancing, to enable a producer to meet spot markets bids and

avoid purchases of balancing power 2) minimize peak load in

order to reduce subscribed power and tariff to the regional 130-

kV network 3) decrease network losses, 4) the contribution from

reactive power control using the power converters to reduce the

reactive power flow to the overlying network. Quantification of

the economic gains from each operation case is provided.

I ndex Terms -- Distributed power generation, Energy storage,

Power generation planning, Reactive power control, Virtual

power plant

I.

I NTRODUCTION The use of electricity from Distributed Energy Resources

(DER) like wind and solar power will impact the performanceof the electricity network and this sets a limit to the amount ofrenewables that can be connected [1] [2]. If designed to reactto minor fluctuations in the network DER units could becomean important asset in the effort to keep the power networkstable and allow for more variable energy resources. However,each DER is itself often too small to be effectively controlledand managed by power utilities and energy market actors.

A Virtual Power Plant is a term used for aggregation ofDER units in order to make them appear as a single, larger,

power plant [3]. Whereas the market participation of the VPP

is obtained by the joined production of all the DER units, theinteraction with the network is different for each unit anddepending on their location in the network. When evaluatingthe opportunities that a VPP offers, the electricity networkshould therefore be considered as well. This is not only

because the network can limit the ability of the VPP to participate in the electricity market. The VPP units may also provide multiple ancillary services, supporting the network.

The value of these ancillary services goes beyond the meremarket value of the produced energy. For example many DERunits have power electronic converters that are able to operatein all four quadrants. Often these units are set to keep reactive

power near zero at the DER connection point. The reactive

power capabilities of the DER units are in this way not fullyutilized. A more efficient operation of the grid would be possible by controlling the reactive-power flow between theDER units and the grid. The goal of this study is to quantifythe potential and economic value of such ancillary services.

There is a need to develop and evaluate the control,communication and operational stability of the VPP. Theambition is to implement the VPP described in this paperwithin an existing Smart Grid Research, Development andDemonstration (RD&D) unit at STRI in Sweden by providingreal time communication between the DER units. For thisreason the paper concludes with a section about theimplementation of the proposed VPP.

II.

STUDIED NETWORK AND DER CHARACTERISTICS

A.

The Network

The interaction between the local network and a VPP has been studied for an existing 50-kV network in central Sweden,shown in Fig. 1. Next to the network data, we had access tothree years of hourly data of consumption at all nodes over athree-year period. Hourly production data is used also from a34-MW wind park and a 6-MW hydro-power unit within thesame network. The availability of such hourly data is common

place in Sweden, for these voltage levels.

For a number of locations, active and reactive-power datais available with a higher time resolution, from installed

power-quality monitors. B.

Participating DER Units

The DER units listed below are included in the proposedVPP. All units are existing, even if the size of existing PVinstallation and Li-Ion battery storage is increased in the

This work was part of a master thesis within a joint development projectat the HVV (www.highvoltagevalley.se) titled “Smart Grid Energy Storage”.

The work was financed by the Swedish Governmental Agency for InnovationSystems as well as IET InnoEnergy INSTINCT.

N. Etherden, and M H.J. Bollen, are with Luleå University ofTechnology, 931 87, Skellefteå, Sweden and with STRI AB, 771 80 Ludvika,Sweden. J. Lundkvist is with Uppsala University, Sweden and with STRIAB, 771 80 Ludvika, Sweden.



simulations. No demand response capability is included in thestudy. The DER units are envisioned to be monitored andcontrolled from the RD&D for closed loop testing of the VPPapplications.

A 17 turbine, 34 MW wind park, which cannot currently be extended due to limiting capacity in its 10/50-kVtransformer. Currently the turbine converters set reactive

power to zero in the two network connection points of the park.

A hydro plant whose generator is rated 7.7 MVA at 6.6 kVand directly coupled to a vertical Kaplan turbine at 500rpm. It features a rotating brushless excitation system.Rated power is 6 MW with an average production during2010-2012 of just over 2 MWh per hour. The rating of thesole generator is low compared to the size of the reservoirthus enabling peak-power production. The local utility alsocontrols all the upstream reservoirs in the catchment area[4]. Regulation is allowed between 44.3 and 44.5 m insummer and 44.0 and 44.5 m in winter, corresponding to amaximum “stored” production of 111 MWh.

The VPP will be studied together with the STRI RD&Dfacilities. These facilities include small (30 kW) solar

power and Li-Ion storage for one hour maximum solar production (30 kWh). Fuel cells, electrolyser and hydrogenmetal hybrid storage as well as two electric vehicles arealso part of the facility, but these are not included in thisstudy. Based on an assessment of available roof area in thenearby industrial park the solar power is scaled up to 3MW in the simulations. The battery storage size isincreased in the studies up to 8 MW /32 MWh. The

response from PV installation and Li-Ion battery storage atthe RD&D facility is increased to represent the largerstorage and PV in the studies when simulating the VPPeffect on the network.

III. EVALUATED A NCILLARY SERVICES

Based on a literature study and discussions with the localDistribution System Operator (DSO) three applications ofnetwork services were selected. A fourth application,where the aggregator operating the VPP is a balance-responsible party, was also included. The applications are

presented in Table I. Studies of similar applications ofVPP for reactive power compensation are described in [5],

for improved system efficiency in [6] and to optimize power flow and minimize the peak network load in [7] [8].

Figure 1. The VPP uses existing DER within the 10 (green) and 50-kv (red) networks. The 130-kV network was also simulated but not shown in figure.

6 MW Hydro

34 MW Wind farm

7*2 MWwind turbines

10*2 MWwind turbines

Studied 130 kV gridconnection point

8 MW/32 MWhattery storage

3 MW solar PV

STRI RD&D(VPP controller)



TABLE I. SELECTED A NCILLARY SERVICES FROM THE VPP

Appl. Network servi ce

B

Balancing: Compensate for prediction error from non dispatchableDER. The hydro reservoir is used with battery storage to allow theVPP to act as a balance responsible party and meet the forecasted

production of the day-ahead spot market of the wind and solarunits. In this way the next day stated capacity of the DER unitscan be confirmed. The limitations set by the 50-kV network onmarket transactions are specifically part of the assessment.

P

Peak monthly power reduction: Reduce subscribed power toregional 130-kV network in order to obtain a lower tariff for theoperator of the distribution network. This application also reducesthermal overload, increasing the networks hosting capacity.

LMinimize network losses: By smoothing peaks in the power flowthe losses in the overlying sub-transmission network are reduced.

RControl reactive power flow: Lower the reactive power in theconnection between 50- and 130-kV networks by coordinatingDER reactive power control possibilities.

A detailed description of the implementation of eachapplication can be found in [9]. Pure market orientedoptimization of the VPP including maximizing profit fromtrading on the spot market was not included in the assessment.

However, a commercial value of each ancillary service is presented in section IV.A to IV.D. For application B this isnon-purchased balance power, for application P lowered tariffto regional network, application L reduction in costs to coverlosses in the local network. For the fourth application thevalue is in the increased network hosting capacity (i.e. abilityof the network to host more DER [10]) as well as lowernetwork losses in the overlying network. The gain is alsodependent on national regulation regarding e.g. balanceresponsibility and tariffs [11]. For this reason no assumptionson VPP ownership are made.

IV. R ESULTS

A. Balancing

This application allows a balance responsible party to usethe storage and controllable production units of the VPP tomeet placed bids on the day-ahead spot market from wind andsolar units. The application compensates for prediction errorsfrom wind and solar production. The prediction of productionis randomly chosen from a normal distributed curve based ondata from [10]. The distribution of production error is given inFig. 2. It was deduced from [12] that the prediction error ofthe solar power has similar error and distribution.

Figure 2. Distribution of prediction error for wind power in North-Eastern part of Germany 2009. The average production in each 15 minute period is1754 MW.

This application uses the battery as primary storage andthe hydro reservoir as secondary. With higher production than

predicted the hydro power plant lowers the power output andthe battery is charged. The battery is discharged if the

production is lower than predicted. The hydro power output isincreased when the battery storage is not sufficient.

A simulation is made for increasing power ratings of the

storage. Maximum charging power is 45 % of the maximumdischarge power in accordance with the characteristics of themodeled Li-Ion storage of the RD&D. The reduction in

prediction error (i.e. need for balancing power) as a functionof battery size is shown in Fig. 3. Two curves are shown: for2-hour and 4-hour ratio between storage capacity and rated

power. Fig. 3 reveals only a minor improvement with thelarger 4 hour capacity. As the cost of energy storage today islargely influenced by the battery cost it is important that thesame converters with fewer batteries can achieve nearly thesame effect.

Figure 3. Reduction of prediction error from wind and solar production

when included in a VPP with hydro reservoir and different battery storagesizes. The two upper (solid/red) curves are with only battery storage.

The hydro plant and reservoir are able to eliminate 20% ofthe prediction error by itself. Together with an 8 MW/32MWh battery storage, up to 60% of the requirement for

balancing power can be eliminated. In order to quantify theassociated economical gain from balancing, the cost of

balancing power on the Nord Pool Spot market wascalculated. The economic gain is up to 2 million Swedishcrowns (SEK), currently about 230 000 EUR as seen in theright vertical axis scale of Fig. 3.

B. Peak Monthly Power Reduction

In this application the aim is to reduce subscribed power tothe regional 130-kV subtransmission network in order for theDSO to obtain a lower tariff. The model is made to reducesubscribed power when the power consumption is higher thana limit chosen from historical data. The level is updated duringthe month so as not to compensate for high power flows thatwill not result in the hour with the monthly largest MWh

power flow that governs the tariff. The application uses primarily the hydro reservoir and secondarily the batteries.

The possible reduction of the maximum hourly power flowduring each month is up to 14 MW, as shown in Fig. 4.

-4000 -2000 0 2000 4000Predicted error (MW)



Figure 4. Reduction of maximum monthly power flow with the VPP. Upper(red) curve is the monthly maximum flow without VPP and lower ( blue) withthe VPP optimised to reduce peak power flow.

The DSO tariff for highest monthly power from theoverlying network is equivalent to 70 kSEK (8300 EUR) per

MW and year. Thus the annual saving would be 440 kSEK peryear. The savings for an end customer from reduced powertariff to the DSO would be three times as large.

C. Minimize Network Losses

In this application the battery and hydro storages are usedto reduce the power flow variations. This in turn will decreasethe transformer and line losses according to equation (1).

(1)

With the measured values of active power flow and a power factor of 0.9, 80% of the network losses are due toactive power and 20 % due to reactive power. It was found

that the reduction of losses in the network with active powercompensation from primarily the battery storage wasinsignificant.

To determine the maximum potential of this applicationsimulations have also been done in which the active powerwas kept constant during each full day. Even this only resultedin a small reduction in losses in the 130-kV grid, see Table II.As will be shown in Section IV.D, reactive-power control has

bigger opportunities to reduce losses.

D. Reactive Power Flow Management:

This application controls reactive power flow in theconnection point between the 50 and 130-kV networks. The

reactive power capabilities of the VPP’s DER units are shownin Fig. 5. In Section IV.C the reactive-power control of theDER units was used to maintain voltages. In this applicationthe reactive-power control is used to minimize the reactive

power in the 50/130 -kV network connection point. Thereactive power flow in the 50/130 -kV network connectionwas simulated with and without optimization of active powerflow. The reactive power requirement results in maximalreactive power output from the storage and generation sourcesfor most of the hours.

Figure 5. Capability curves of the VPP inverters

The VPP is able to eliminate the reactive power demand ofthe 50-kV network with 75%, see Fig. 6. This reduction

almost eliminates the contribution to network losses from thereactive power of the 50-kV network in the 130-kV networkas shown in Table II in which a reduction of losses has anegative sign.

Figure 6. Reduction of reactive power in the 50-/130-kV connection point.

TABLE II. OVERVIEW OF NETWORK LOSSES WITH AND WITHOUT VPP

Case

Annual losses /

reduction [MWh]

Maximum annual

reduction [MWh]

50-kV 130-kV 130-kV

Without VPP1 4656 5610 N/A

VPP with active powercompensation

14 -1 -452

VPP with reactive powercompensation

-28 -90 -145 3

VPP with both active andreactive compensation

-17 -86 -132 4

1 No battery storage or hydro reservoir (hydro production given by water inflow)2 Based on unlimited size of VPP that each hour can bring power exchange with130-kv network to its daily average3 Losses with Q equal to zero in the 50/130-kv connection point4 Losses with P according to 2 and Q according to 3



A reduction of losses in the 50-kV network was alsofound. However violation of voltage limits was shown tooccur for some hours. Keeping the voltage within limits willrequire the reactive-power flows to be limited. This will inturn reduce the ability of the VPP to reduce reactive-powerexchange and to reduce harmonics.

With the average spot market price 2010-2012 the annual

savings in losses of 28 + 90 MWh would amount to only 50kSEK or 5.7 kEUR per year.

V. IMPLEMENTATION

The different applications of the VPP compete with eachother for the priority of the DER. Application P uses primarilythe hydro reservoir. Applications B and L use first the batterystorage, only using the hydro storage when the battery storageis insufficient. As the secondary storage is used less often andthe primary storage source varies, the applications cansuccessfully coexist with little performance reduction. This isshown in Fig. 7.

Figure 7. Time use of the hydro and battery storages for each studiedancillary service. Letters represent the applications of Table I. In thesimulations only one application is applied at a time.

It may also be possible for these network services tocoexist with commercial spot market trading optimizationwhere they are only initiated during contingency situations.Because the prioritization of the services is internal to the VPPcontroller are not further discussed in this paper. Theinformation requirements of the VPP have been furtherassessed in [13] which also evaluate the extent to which SmartGrid standards can fulfill the VPP information requirements.

VI. CONCLUSSIONS

The application most profitable for the Virtual Power Plantwas to optimize use of dispatchable production and storageresources in order to meet placed bids on the day-ahead spotmarket of (non dispatchable) wind and solar resources. Thiscould halve the cost for balance power, a saving of 2.5 millionSEK, but it would require relatively large battery storage. Tominimize peak load and thereby reduce the tariff could saveon average 440 kSEK per year for the DSO. Benefits fromreduced network losses and control of reactive power flowwhere also found but are greatly dependent on the networkcharacteristics and therefore these results cannot be

generalized. Especially when using reactive powercompensation care must be taken not to violate voltage limitsin the network. As reactive power product that cannot betransported over large distances, it is difficult to establish a

proper market [14]. Therefore regulatory measures tomaximize the hosting capacity for DER may be a moreappropriate evaluation criteria then the economic metrics usedhere when evaluating such ancillary services.

Although gains from the describe VPP amount to 2-3million Swedish crowns per year the current costs for grid-scale energy storage of these sizes would still mean paybacktimes of several decades. Today the BESS cost is mainlyrelated to the price of its batteries. Thus applications like

balancing that rely mainly on a storages power rating are morecost efficient than using the battery storage for tariff or lossreduction (where the capacity of the storage is relatively moreimportant than the power ratings).

R EFERENCES

[1] M. H. Bollen, Adapting the Power System to New Challenges,Morganand Claypool, 2011.

[2] L. Ferris and D. Infield, Renewable Energy in Power Systems,Chichester, UK: Wiley, 2008.

[3] K. El Bakari, W. L. Kling, “Virtual power plants: An answer to

increasing distributed generation,” in Innovative Smart Grid

Technologies Conference Europe (ISGT), Gothenburg, 2010

[4] S. Berlijn, S. Markalous and K. Ström, "Web-based PD monitoring of aGenerator in Loforsen Sweden," in International Conference on

Condition Monitoring and Diagnosis, Beijing, 2008.

[5] P. Nyeng, J. Østergaard, C. A. Andersen, J. Dall and C. Strunge,"Reactive power control with CHP plants - A demonstration,"

Proceedings of the CIGRE session, 2010.

[6] D. Pudjianto, C. Ramsay and G. Strbac, "Virtual power plant andsystem integration of distributed energy resources," Renewable Power

Generation, IET, vol. 1, no. 1, pp. 10-16, March 2007.

[7] N. Ruiz, I. Cobelo and J. Oyarzabal, "A Direct Load Control Model for

Virtual Power Plant Management," IEEE Transactions on PowerSystems, vol. 24, no. 2, pp. 959-966, May 2009.

[8] J. Kumagai, "Virtual Power Plants, Real Power," IEEE Spectrum, vol.49, no. 3, pp. 13-14, March 2012.

[9] J. Lundkvist, "Feasibility study of a Virtual Power Plant in Ludvika,"Uppsala University, M.Sc. thesis, Uppsala, 2013.

[10] M. H. Bollen and F. Hassan, “Integration of Distributed Generation inthe Power System”, Hoboken, USA: John Wiley & Sons, Inc., 2011.

[11] S. Ackeby, L. Ohlsson, N. Etherden, “Regulatory Aspects of EnergyStorage in Sweden,” in CIRED, Stockholm, 2013

[12] C. Groiss, et.al, “Photovoltaik -Erzeugung für einte RegenerativeVollversorgung Österreichs,” Symposium Energieinnovation, Graz,2010

[13] N. Etherden, M. H. Bollen and J. Lundkvist, "CommunicationRequirements of a Virtual Power Plant using IEC 61850 to Provide GridServices," in IEEE SmartGridComm, Vancouver, 2013 (submitted).

[14] Energinet.dk, "Energinet.dk’s ancillary services strategy," Energinet.dk,

Doc. No. 77566/11 v1, Case No. 10/5553, Fredericia, Denmark, 2011.



Communication Requirements of a Virtual Power

Plant using IEC 61850 to Provide Grid Services

Nicholas Etherden and Math H.J. BollenSTRI AB

Gothenburg, Sweden

[email protected]

Johanna Lundkvist

Uppsala UniversityUppsala, Sweden

Abstract — A feasibility study of a Virtual Power Plant (VPP)

consisting of hydro and wind plants, solar PV and storage

facilities is presented. Several applications to provide grid

services from the VPP are described and the information and

communication requirements evaluated. The information

requirements from the proposed VPP in an existing 50 kV grid

are analyzed and mapped against services and information

models of IEC 61850. It is found that with planned extensions of

the IEC 61850 most information requirements between VPP

controller and the distributed energy resources can be met.

Keywords — Distributed power generation, Energy storage,

Power generation planning, Reactive power control, Vir tual power

plant

I. I NTRODUCTION ( Heading 1)

A Virtual Power Plant (VPP) is a term used for aggregationof Distributed Energy Resources (DER) in order to make themappear as a single, larger, power plant. Whereas the market

participation of the VPP is obtained by the joined production ofall the DER units, the interaction with the grid is different foreach unit and depending on their location in the grid. When

evaluating the opportunities that a VPP offers, the electricitygrid should therefore be considered as well. This is not only because the grid can limit the ability of the VPP to participatein the electricity market. The VPP units may also providemultiple grid services. If designed to react to fluctuations in thegrid the coordinated control of DER units could become animportant asset in the effort to keep the power grid stable andallow for more variable energy resources.

International standards will play an important role to allowoptimised control of multiple small units that would otherwise

be left to operate in an uncoordinate way. For a VPP to be costeffective in integrating also small DER units, standardizedcommunication interfaces are needed for a broad variety of

different DER units. Not only is interoperability required frommany different kinds and sizes of DER units, support of welldefined engineering workflows and procedures can be just asimportant in lowering the overall integration cost.

There is a need to asssess the applicability of emergingSmart Grid communication architecture standards against theneeds of the VPP. In the first half of this paper (Section II andIII) an assement is made of available Smart Grid standards forVirtual Power Plants. In the second half of the paper (SectionIV and V) the information requirements of a proposed VPPusing existing DER in a 50 kV grid are presented. The extent to

which information models and functionality in IEC 61850 meetthe VPP requirements is assessed.

II. ASSESSMENT OF SMART GRID STANDARDS FOR VPP

A VPP requires information from a large number ofapplication domains (wind, hydro, solar PVs, storage, marketoperation, demand response schemes etc.). In order to enablean open communication infrastructure within the VPP

interoperability is required on several levels.

In Europe the Smart Grid Architecture Model (SGAM) [1]has been developed by CEN, CENELEC and ETSIstandardization bodies in response to the EU Commissionmandate M/490 [2]. The SGAM model consists of fiveconsistent layers representing business objectives and

processes, functions, information models, communication protocols and components. Using the five interoperabilitydimensions of the SGAM model, the requirements of the VPPcan be expressed as in Fig. 1.

Generation

Transmission

Distribution

DER

Customer

Premisses

Process

Field

Station

Operation

Enterprise

Market

Data Model

Data Model

Protocol Protocol

IEC 61400-25

wind turbine

Data ModelIEC 61850-7-410

Hydro Data Model

Grid monitoring

Reactive power

control

DER control &

supervision

Commercial VPP

Component

Communication

Information

Function

Business La er

I n t e r - o p e r a b i l i t y

D i m e n s i o n

Domains

Zones

Fig. 1. Mapping of VPP requirements to the interoperability dimensions of

the Smart Grid Architecture Model (SGAM)

The VPP provides interoperability on business layers for anumber of applications (four of which are analyzed in sectionIV.B.) These applications utilize functions that act on multipleDER units. The information from the various DER isinteroperable if it uses a common semantic and syntax. Finally

The feasibility study of the Virtual Power Plant (VPP) was undertakenwithin a joint development project at the HVV (www.highvoltagevalley.se)

titled “Smart Grid Energy Storage” and financed by the SwedishGovernmental Agency for Innovation Systems.

The assessment of open communication standards for VPPs was part of

an IET InnoEnergy INSTINCT project.



the information is to be obtained from the individual DERcontrollers using interoperable protocols.

A VPP based on open communication standards requires alarge spectrum of the technologies, collectively grouped underthe term smart Grid. Several assessments of standards for theSmart Grid have been performed. In the US the NationalInstitute of Standards and Technology Smart GridInteroperability Panel (NIST SGIP) has been appointed the taskto provide a framework for coordinating all Smart Gridstakeholders in an effort to accelerate standards harmonizationand advance the Interoperability of Smart Grid devices andsystems. Their first assessment gave five "foundational" sets ofstandards, all IEC TC 57 standards [3], but has since beenextended to some 50 protocols.

The IEC Smart Grid Standardization Roadmap [4] andreference architecture [5] advocate the use of IEC 61850 forfield and station communication while IEC 61970/61968Common Information Model (CIM) is to be used on enterpriseand market zones. The areas for IEC 61850 and CIM areshown on the component layer of the SGAM model in Fig. 2.

Fig. 2. SGAM Smart Grid planes domains and hierarchical zones. Reproduced

from [1] with areas for the CIM and IEC 61850 standards overlaid

Alternatives proposed for implementing VPPs includemiddle ware software, web service, the power utilityautomation standard IEC 61850, the control center applicationexchange interface model IEC 61968 (CIM). Also OPC and theuse of pure protocol solutions that do not provide semanticdescription of the data can be alternatives. An overview ofdifferent international standards for integrating DER is given in[6]. The overview concludes that IEC 61850 is most suitablefor device to application communication while CIM is

preferable for inter application information exchange. Use ofDPWS web services as protocol for IEC 61850 in order toachieve a Service Oriented Architecture (SAO) has also been

proposed [7] [8]. However, several of the alternative proposalsare better understood as alternative carriers of the informationon the communication interoperability dimension of the SGAMmodel, rather than a full scale alternative to the semanticallydefined IEC 61850 data model on the information layer andassociated engineering process for the integration of the DER

provided by IEC 61850 on the functional layer.

III. USE OF IEC 61850 FOR VPPS

A. The IEC 61850 Standard

IEC 61850 is an information model and communicationarchitecture standard developed to allow interoperability within

power utility automation systems [9]. The intent is to define aflexible and future-proof standard that supports interoperabilityin power utility automation and communication. IT-security

procedures, quality codes and error handling can be common.The protocols used to send IEC 61850 information aregenerally not defined in the standard itself, only the mapping toa limited number of protocols standardized elsewhere isincluded in the standard. Dedicated parts of the standard series

exist for hydro and thermal power, wind, DER resources; intersubstation and substation to control center communication tostate only a few.

IEC 61850 was originally intended solely for the use withinsubstations, primarily at transmission level. The standard has

become the most common communication solution at highervoltage levels in many countries, including Sweden. In the

beginning of 2013 some 80 manufacturers in over 20 countrieshad certified devices for the standard and solutions exist to runIEC 61850 software protocol stacks on microprocessorscosting as little as 50 EUR.

Considerable work has been ongoing the last three to fiveyears to extend IEC 61850 to new domains and make itapplicable for new application areas and lower voltage levels.Assessments of the requirements from Smart Grids have beenanalyzed and new work items created by the IEC in response toidentified demands. Several of the IEC 61850 parts referred toin Table II are therefore still under development.

In IEC 61850 Logical Nodes are used to group data that isto be exchanged. The data can contain multiple attributesincluding quality, time stamp and even currency information.The Logical Nodes represent a function, measurements or avirtualization of a real object like a circuit breaker or battery.Logical Nodes are grouped into Logical Devices implementedin Intelligent Electronic Devices (IEDs). The second edition ofthe standard allows the grouping of Logical Nodes into

functions and sub functions in a topology related data model.The same information can be grouped into component and subcomponents in a parallel product view of the IEDs. A gatewaydevice may also hold a representation of non IEC 61850devices to which it communicates by other means. This allowsa flexible modeling where information from simple and diversIEDs of the participating DER can be efficiently structured in ahierarchical model within the VPP.

B. IEC 61850 for Hierarchical DER

Today IEC 61850 covers data models for a broad range ofVPP components, allowing them to interact through a coherentinformation model using standardized communication and

service-oriented functions. IEC 61850 enables well definedsystem interfaces between different DER units usingstandardized semantic and syntax where information isexchanged over a limited number of mapped protocols.Potentially IEC 61850 allows the interoperability on thecommunication; information and functional dimensions of theSDGAM model as shown in Fig. 1.

The model and system configuration can be specified on afunctional level allowing free allocation of the functionality tovarious Intelligent Electronic Devices (IEDs) used for

protection, control and monitoring. Information from the IEDs

CIM

IEC 61850



can be sent to gateways or directly to SCADA and EMSsystems. Therefore it is very appropriate for modeling ofcomplex distributed systems. Another advantage is the

potential for seamless integration with IEC 61968 CIM (used between applications in control and dispatch centers) as well asfor the information exchange with energy market actor systems(defined in the IEC 62325 standard series). This allows thetopological origin and information content of the IEC 61850

data to be converted to the CIM.IEC 61850-90-15 is an upcoming technical report titled

“ Management and Aggregation of Hierarchical DER System “.According to this document the VPP is intermediary forrequests from clients seeking information from the DER unitson process level. Several interfaces exist between the VPPManagement System, the DER units and market actors as givenin Fig. 3.

Fig. 3. Interfaces of the VPP to grid operastor and market, according to IEC

technical committee 57, working group 17

At the bottom is the plant management interface where IEC61850 is used to interface with DER units. The interface to thegrid operator (left) could be implemented by either IEC 61850or IEC 60870-6 TASE.2 (that uses the same MMS protocol asIEC 61850). Another possibility would be an IEC 61968 CIMgateway making the conversion of data to this module alreadyin the VPP management system. The intention of IEC is thatenterprise and market application (top) exchange is handledwith the CIM.

IEC 61850-90-15 distinguishes between commercial(participation on wholesale markets) and technical VPP(support for the local grid). Commercial VPP is mainly part of

the enterprise and market zones of the SGAM model and thusassociated to the IEC 61968 CIM (se Fig. 2). The technicalVPP covers communication between field, substation andoperation zones associated with IEC 61850.

The control of the individual DER units is anticipated tofollow principles of IEC 61850-7-420 “ Distributed energyresources logical nodes”. IEC 61850-90-15 will also describe anumber of services required by a VPP. This will includeadministration functions that allow the VPP to maintain aregistry of the VPP DER units. Other parts of the standardseries are also intendeded to develop dynamic system

management services to allow a DER to register to a clientspecified with only its IP-address. The client can then query thecharacteristics and properties of a DER. In a longer perspectivestandardized services such as those described for PV and otherDER inverters in IEC 61850-90-7 could be subscribed toautomatically. Such functions include freq-watt, volt-wattcontrol of active power and fixed power factor, fixed VAr,Volt-VAr and Watt-power factor control of reactive power. An

example of such a standardized service is provided in Fig. 4.

Fig. 4.

Frequency-watt mode operation from IEC 61850-90-7

The VPP can optimize the settings of the various DER andchange set points for the DER when required. Based on itsdatabase the VPP could then respond to requests from a markettrading system (or grid operator power system managementtool) with requested characteristics and properties and providea response with available capacity. The database may need to

be updated with real value parameters and estimate theavailable capacities in real time.

C. Experiences with Using IEC 61850 for VPPs

There exists practical experience from use of IEC 61850from a few Smart Grid demonstration projects. For example

reactive power support for frequency control and ElectricVehicle charging is planned to be set under full scale trialsusing IEC 61850 in the Danish Island of Bornholm [8]. Asimilar VPP as that described in section IV was constructedand implemented with IEC 61850 in Germany [10].

IV. DESCRIPTION OF THE PROPOSED VPP

In order to verify the maturity of both the standards andavailable equipment implementing the standard the extensionof an existing Smart Grid Research, Development andDemonstration (RD&D) unit at STRI into a VPP test bench has

been studied. The intention is to include real-timecommunication from the MV distribution grid as well as anearby hydro power plant and wind farm. This will enable

development and verification of control algorithms for VPPsand to gain experience in their practical deployment andoperation.

The required interaction between the local grid and a VPPhas been studied for an existing 50-kV grid in central Sweden.The VPP that was studied consisted of four elements: a 34 MWwind park; a 3 MW PV installation; a 6 MW hydro-power

plant with a maximum of 111 MWh stored energy available inthe reservoir; and a 8 MW grid-size battery storage with atmost 32 MWh storage capacity. The wind park and the PVinstallation are not controllable; their production is determined



by the availability of sun and wind. Variations and predictionerrors from these two sources can be compensated by thehydro-power plant and the battery storage. For a number oflocations within the grid, active and reactive-power data isavailable with a higher time resolution, from installed power-quality monitors. The participating DER units and detailedsimulation results from a study of the VPP using three-yearshourly consumption and production data can be found in [11].

The DER units are to be controlled from the RD&D. Theexisting facilities at the RD&D that includes 30 kV solar PVwith IEC 61850 upgradable inverters, a 30 kV lithium-ion

battery storage modeled and controlled by IEC 61850 (se Fig.5). IEC 61850 capable energy and power quality meters arealso installed. The response from PV installation and Li-Ion

battery storage at the research facility is increased to representthe larger storage and PV in the studies when simulating theVPP effect on the grid. The hydro and wind farm exist todayand actual hourly production data is available.

Fig. 5.

IEC 61850 datamodel of inverter controller of the battery storage

Through a dedicated VLAN within the local utility networkthe facilities will have access to equipment in the distributiongrid. An IEC 61850 capable power quality metering system is

being installed in a dozen locations within this distribution gridallowing measurements from locations in the grid to be used bythe VPP.

A. Selected Applications

Based on a literature study and discussions with the localDistribution System Operator (DSO) three applications of gridservices were selected. A fourth application where the

aggregator operating the VPP is a balance-responsible partywas also included. The intention was to limit the functions towhat IEC 61850-90-15 calls a technical VPP. The value ofthese grid service go beyond the mere market value of the

produced energy. For example many DER units have powerelectronic converters that are able to provide or consumereactive power to help maintain voltage levels. Often theseunits are set to keep reactive power near zero at the DERconnection point. The potential to coordinate these reactive

power capabilities to optimize the overall efficiency of thegrid(s) in which the DER units are located are then not fully

utilized. Pure market oriented optimization of the VPPincluding maximizing profit from trading on the spot marketwas therefore not included in the assessment. The applicationsare presented in Table I. Studies of similar applications of VPPfor reactive power compensation are described in [12], forimproved system efficiency in [13] and to optimize power flowand minimize the peak network load in [14] [15].

TABLE I. SELECTED GRID SERVICES FROM THE VPP

# Grid service

B

Balance responsible party (B): Compensate for prediction error

from non dispatchable DER. The hydro reservoir and power plant

is used together with battery storage to allow the VPP to act as a balance responsible party and meet the forecasted production of

the day-ahead spot market of the wind and solar units. In this way

the next day stated capacity of the DER units can be confirmed.

The limitations set by the 50 kV grid on market transactions are

specifically part of the assessment.

P

Decrease peak monthly power (P): Reduce subscribed power toregional 130 kV subtransmission grid in order to obtain a lower

tariff for the operator of the distribution grid. This application also

reduces thermal overload, increasing the grids hosting capacity.

LMinimize grid losses (L): By smoothing peaks in the power flow

the losses in the overlying sub-transmission grid can be reduced.

R

Control reactive power flow (R): Lower the reactive power in the

connection between 50 and 130 kV grids by coordinating DERreactive power control possibilities.

Although maximization of economic return of the VPP wasnot the primary goal of the study a commercial value of eachgrid service can be ascribed as presented in [11]. Forapplication B this is non-purchased balance power, forapplication P lowered tariff to regional grid, application Lreduction in costs to cover losses in the local grid. For thefourth application the value is in the increased grid hostingcapacity (i.e. ability of the grid to host more DER see [16]).

The different applications of the VPP compete with each

other for the priority of the DER. Application P uses primarilythe hydro reservoir. Application B and L use first the batterystorage, only using the hydro storage in extreme circumstanceswhen other applications are down prioritized. As the secondarystorage is used less often and the primary storage source variesthe applications can successfully coexist with little

performance reduction, as shown in [11]. It would also be possible for these grid services to coexist with commercial spotmarket trading optimization where they are only initiatedduring contingency situations. Because the prioritization of theservices is internal to the VPP controller they do not affect therequired information and are not further discussed in this paper.

B. Information Requirement

The required information to perform the applicationsdescribed in Table I consists of both commercial data frommarket traders, real-time information from the included DERand measurement of power flow and voltages in thedistribution grid. The VPP needs to communicate with eitherthe DSOs SCADA system or equipment in the grid in order togather information. There is need for information exchange

between the VPP and market actors.

The required information to perform the applicationsdescribed in the previous section is summarized in Table II.



Some information required with a ‘?’ is not essential for the basic application, but can be used to improve performance.

C. Communication Requirements

The VPP is planned to send set points to battery storage andcontrollable production units using the Logical Nodes for DERsupervisory control, controller status and controllercharacteristics (DRCC, DRCS, DRCT) defined in IEC 61850-

7-420.The need for fast real-time information is not anticipated in

this VPP. Data on the grid power flow and DER production are probably sufficient on minute scale. Minute average ofvoltages in the grid was likewise seen as sufficient for thestudied applications.

To acts as balancing party and compensate for predictionerrors in day ahead and intraday bids, 15 minute updateintervals are expected to suffice for production and storage set-

points as well as for reactive power measurement. Payment for balancing power is only due if the deviation contributes to thedeviation of the entire price area. In Scandinavia more than70% of the total consumption of electrical energy is traded

through Nord Pool Spot which is currently the world’s largestelectricity broker, measured in traded TWh. Information on thelatest regulating volumes can be retrieved from their website

with 1-2 hours delay. This delay may still be acceptable as onaverage the direction of regulating volumes changed only aboutonce a day in 2012 with less than 9 % of the changes occurringwithin 2 hours. However, regulation power is bought only forhalf the hours so it would be important to know if paymentswill be charged during the current hour, which requires quickerdata access.

V.

MAPPING TO IEC 61850 DATA MODELS AND SERVICES Included in Table II are proposals of appropriate IEC 61850

Logical Nodes, data and services to be used in theimplementation of the VPP applications described in sectionIV.A. It was found that semantically defined data exists or is

planned within IEC 61850 for all required information betweenthe VPP controller and the DER units. The VPP is defined sothat mainly generic measurements that will be mapped to thestandard MMXU Logical Nodes from IEC 61750-7-4 can beused in most IEDs. The services defined in IEC 61850-7-4 forstatistical and historical data allow the VPP controller to keeptrack of power flows and limits. Although the studied VPP useslimited amount of data from IEC 61850 domain extensions the

presence of these extensions enables a flexible VPP controllerwhere additional information available in DER controllers can

be browsed and included in control algorithm to improve performance as required.

TABLE II. R EQUIRED I NFORMATION AND POTENTIAL MAPPING TO IEC 61850 FOR THE APPLICATIONS OF TABLE I

InformationApplication

Description Data sourceIEC 61850

B P L R Logi cal Nodes seri es part

Predicted Production X X X Bid on the day-ahead spot market. (24 bids placed

12 a.m. for next day)

Forecasting tool used by

Electricity retailersa

Use of IEC 61850 not anticipated,

although schedules from upcoming

IEC61850-90-10 could be used.Intraday correction of

predicted Production

X Bid on intraday market. (1 bid per hour, latest an

hour before period starts)

Forecasting tool used by

Electricity retailersa

Battery storage State-

of-Charge

X X X Amount of Stored energy in BESS Battery storage controller DBMS To be defined in

upcoming -90-9

Hydro reservoir water

level

X X X Amount of Stored energy in hydro reservoir Hydro producer or

controller of hydro plant

HLVL.LevM 7-410

Real-time production

of DER

X X X X To determine accumulated production within

present hour. Also used together with latest bid toestimate production for present hour.

SCADA or DER

controllers

MMXU.W 7-4

Latest Regulating

volume for price area

? To predict if over or under production (compared to

prediction) will result in economic cost

Electricity market broker Use of IEC 61850 is not anticipated

Weather data used forretailers prediction

? Wind speed, rainfall, irradiation etc on which production prediction where based. Used to estimate

if bids during present hour are accurate.

Electricity retailers Use of IEC 61850 is not anticipated

Real-time weatherdata

? Actual weather data within hour to enableestimation of prediction error for present hour

Measurements at DER ormetrological bureau

MMET. HorWdSpd,DctInsol, CloudCvr

7-4

Long term production

forecast

? Optimize SoC and hydro reservoir levels in over

longer than 12-36 h period forecasted in spot bids)

Electricity retailers or

DSO/TSO forecasts

Use of IEC 61850 not anticipated,

although schedules from 90-10 possible

Historical levels of

power flow

X X Determine if VPP shall attempt to lower peak flows

to reduce tariff or losses. Based on historical hourly

power flows or logged data.

VPP controllers log of

real time production

VPP historical

statistical information

model of MMXU.W

7-2

Highest power flowduring current month

X Block power flow reduction if monthly maximumwill not be exceeded

VPP controllers log ofreal time production

VPP statisticalinformation model of

MMXU.W

7-4 (and 7-1)

Power flow in gridlocation(s)

X X Determine if peak power flow and high losses areoccurring.

SCADA or IEDmeasurement within grid

MMXU.W 7-4

Real-time voltage at

DER

? Calculate losses more accurately when voltage

varies considerably at DER .

SCADA, DER controller

or IED near DER

MMXU-PhV 7-4

Reactive power in

selected location(s)

X Reactive power flow at a point(s) in the grid for

which reactive power exchange is to be optimized.

SCADA or IED

measurement within grid

MMXU.Q 7-4

Real-time voltage ingrid

? Calculate ΔQ required from DER to hold voltages inselected point(s) of the grid within limits.

SCADA, IED in grid orsmart metering system

MMXU-PhV 7-4

a. In contrast to other sources like IEC Smart Grid Standardization Roadmap [4] the VPP is not by itself expected to make more than very-short-term

forecasts (1 hour). The energy traders are expected to use existing software for prognosis and the results thereof fed into the VPP controller.



VI. DISCUSSION

In this study the number of DER units to be integrated inthe VPP was limited and only a small amount of informationwas required by the VPP controller. Often the small size ofDER units does not allow them to individually participatedirectly in the energy market. Likewise each DER is in itselftoo small to be effectively controlled and managed by powerutilities and energy market actors. The different ownership also

contributes to making the effective coordination of theoperation of the DER a challenge. The VPP tasks may also besubordinated requirements of individual DER to support thelocal grid. The complexity of the VPP is increased further, assmaller DER may be desirable to group hierarchically so thatsmall VPPs are aggregated into larger ones. This concept has

been developed and field tested within the Danish CellController project [17] [18]. Due to this complexity theintegration costs for providing grid services from divers anddispersed units that do not have a common ownership is amajor concern. In order to access the capabilities of small homePVs with individual storage capabilities, a VPP would need toconnect to several hundred units. If each manufacturer uses its

own communication interfaces the cost of integration may addconsiderably towards the cost of the VPP, possibly making ituneconomical.

With the introduction of low cost gateways and integrationsoftware in the last couple of years the cost of implementingIEC 61850 has been reduced considerably. Several commercialSCADA systems today support IEC 61850 and the standard is

being developed to cover also substation to control centercommunication. IEC 61850 therefore promises to provide acost effective, flexible, open and extendable solution to a VPP

providing grid services. This VPP could use bi-directional IEC61850 communication to MV and LV substation, DER units aswell as demand-response providers. Because access to the grid

related data is often only available within the DSO network theVPP controller may need to be placed in their dispatch center.This limits the ability for market actors other than the DSO tooperate a VPP for providing grid services.

European Transmission System Operators (TSOs) havedecided to use CIM/IEC 61970 for exchanges of systemoperations and system studies [19]. The CIM model could bean alternative for information exchange with the VPPaggregator. Due to the lack of commercial gateways able tomap information from IEC 61850 data to CIM the use of CIMdoes not seem like a realistic alternative at operational level forcommunication to VPP participating DER units. CIM could bean alternative for communication between VPP application and

other utility and market traders’ information, although othersolutions exist for this information exchange. For examplethere exists a simple low cost solution to provide spot market

price to heat pumps in Sweden via internet. This solution could be extended to provide regulating power information.However, due to IT-security reasons it would be preferable touse existing electronic data interchange platform for marketactors on the Nordic market. This is called Ediel and uses theEDIFACT/ISO 9735 standard and is operated by the SwedishTSO. A commercial VPP also exist where capacity can be

bought ahead of time at fixed via Nord Pool Spot [20], but the

aim and market structure of this VPP has little in common withthe grid support services described in this paper.

VII. CONCLUSSION

The IEC 61850 standard covers a wide area of domains.Including ongoing extensions of the standard, IEC 61850 will

be able to support interoperable exchange of information froma very broad variety of DER, as well as scheduling and the

aggregation of DER required by a VPP. The IEC standard cancontribute to seamless integration of the DER data to a VPPand ease integration to market applications due to ongoingharmonization with IEC 61970/IEC61968 CIM model.

REFERENCES

[1] CEN-CENELEC-ETSI Smart Grid Coordination Group, "First set ofstandards", 2012.

[2] European Commission, , "Mandate M/490; Standardization mandate toEuropean standardization organizations (ESOs) to support EuropeanSmart Grid deployment," 2011

[3] [G. Arnold, "NIST-identified standards for consideration by regulators,"6 October, 2010.

[4] [IEC, "IEC Smart Grid Standardization Roadmap," Smart Grid StrategicGroup (SG3), Geneva, 2010.

[5] IEC, "IEC 62357-1 Reference architecture for power system informationexchange," Technical Committee 57, Geneva, 2011.

[6] S. Jaloudi, et.al., "Integration of distributed energy resources in the grid by applying international standards to the inverter as a multifunctionalgrid interface," in IEEE PowerTech, Trondheim, 2011.

[7] S. Sucic, B. Bony, L. Guise, F. Jammes and A. Marusic, "IntegratingDPWS and OPC UA device-level SOA features into IEC 61850applications," in IEE IECON, Montreal, 2012.

[8] B. A. Pedersen, E. B. Hauksson and P. B. Andersen, "Facilitating ageneric communication interface to distributed energy resources:mapping IEC 61850 to RESTful services," in IEEE SmartGridCommunications, Gaithersburg, 2010.

[9] IEC, "IEC 61850 Communication networks and systems for powerutility automation," Technical Committee 57, Geneva, 2004-2013.

[10]

B. M. Buchholz, C. Brunner, A. Naumann and A. Styczynski, "ApplyingIEC standards for communication and data management as the backboneof smart distribution," in IEEE PES General Meeting, San Diego, 2012.

[11] N. Etherden, M. H. Bollen and J. Lundkvist, "Quantification of NetworkServices from a Virtual," in IEEE PES Innovative Smart GridTechnologies (ISGT Europe), Copenhagen, 2013.

[12] P. Nyeng, J. Østergaard, C. A. Andersen, J. Dall and C. Strunge,"Reactive power control with CHP plants - A demonstration,"Proceedings of the CIGRE session, 2010.

[13] D. Pudjianto, C. Ramsay and G. Strbac, "Virtual power plant and systemintegration of distributed energy resources," Renewable PowerGeneration, IET, vol. 1, no. 1, pp. 10-16, March 2007.

[14] N. Ruiz, I. Cobelo and J. Oyarzabal, "A Direct Load Control Model forVirtual Power Plant Management," IEEE Transactions on PowerSystems, vol. 24, no. 2, pp. 959-966, May 2009.

[15] J. Kumagai, "Virtual Power Plants, Real Power," IEEE Spectrum, vol.49, no. 3, pp. 13-14, March 2012.

[16] M. H. Bollen and F. Hassan, "Integration of Distributed Generation inthe Power System", Hoboken: John Wiley & Sons, Inc., 2011.

[17] P. Lund, "The Danish Cell Project - Part 1: Background and generalapproach," in IEEE PES General Meeting, Tampa, 2007.

[18] N. Martensen, P. Lund and N. Mathew, "The Cell Controller pilot project: from surviving system black-out to market support," in CIRED,Frankfurt, 2011.

[19] ENTSO-E, "Common Information Model (CIM)," [Online]www.entsoe.eu/major-projects/common-information-model-cim/, 2013.

[20] Nord Pool Spot, " Virtual power plant in Jutland," [Online]

studiu fezabilitate VPP

Documents

Transcript of studiu fezabilitate VPP