PhDThesis Hamada Almasalma

ARENBERG DOCTORAL SCHOOL
Faculty of Engineering Science
Peer-to-Peer-Based
Grid Support Functions
for Smart Inverters
Distributed Voltage and Frequency Control
P2P
Frequency Support Voltage Support

Function Function
Solar PV Panel Inverter Control
Hardware
Battery Energy
Storage System
P2P-Based Smart Inverter
Hamada Almasalma
Supervisor: Dissertation presented in partial

Prof. dr. ir. G. Deconinck fulfillment of the requirements for the
degree of Doctor of Engineering
Science (PhD): Electrical Engineering
August 2020
Peer-to-Peer-Based
Grid Support Functions
for Smart Inverters
Distributed Voltage and Frequency Control
Hamada ALMASALMA
Examination committee: Dissertation presented in partial

Prof. dr. ir. J. Berlamont, chair fulfillment of the requirements for
Prof. dr. ir. G. Deconinck, supervisor the degree of Doctor of Engineering
Prof. dr. ir. J. Driesen Science (PhD): Electrical Engineer-
Prof. dr. ir. J. Suykens ing
Prof. R. Caire
(Grenoble Alpes University, France)
Prof. J. Wu
(Cardiff University, UK)
August 2020
© 2020 KU Leuven – Faculty of Engineering Science
Uitgegeven in eigen beheer, Hamada Almasalma, Kasteelpark Arenberg 10, box 2445, B-3001 Leuven (Belgium)
Alle rechten voorbehouden. Niets uit deze uitgave mag worden vermenigvuldigd en/of openbaar gemaakt worden
door middel van druk, fotokopie, microfilm, elektronisch of op welke andere wijze ook zonder voorafgaande
schriftelijke toestemming van de uitgever.
All rights reserved. No part of the publication may be reproduced in any form by print, photoprint, microfilm,
electronic or any other means without written permission from the publisher.
“I cannot do all the good that the
world needs. But the world needs
all the good that I can do.”
Jana Stanfield
Preface
This work would have never succeeded without the dedicated support of several
people, whom I would like to thank personally.
First and foremost, I would like to sincerely thank my supervisor Prof. Geert
Deconinck for giving me the opportunity to carry out a doctoral research in
the field of peer-to-peer control and distributed optimization. During the past
four years and ten months, I had the valuable fortune of being able to enjoy his
guidance. He took the time to discuss and read, had great ideas, made numerous
useful comments, showed an open-mindedness and gave me the freedom and
confidence to delve into different areas that contributed to this PhD.
Secondly, I would like to express my gratitude to the Fonds voor Wetenschap-
pelijk Onderzoek (FWO), who believed in the successful outcome of my work
and decided back then to grant me a PhD scholarship. Thank you for the
support.
I would like to thank all the members of the examination committee, for sharing
their very useful thoughts and insights. Prof. Berlamont, thank you for chairing
the committee. Thank you Prof. Driesen and Prof. Suykens for being part
of my supervisory committee throughout the years. Thank you Prof. Caire
and Prof. Wu for being part of the jury and for the useful remarks that have
definitely made this thesis more complete.
I am also very grateful to the lab technicians Johan, Luc and Gert for assisting
me in the experimental works. I also would like to give special thanks to sweet
Electa staff: Katleen, Katja, Veerle and Veronica. They have eased my life by
fixing the administrative, financial and ICT problems.
Of course, I would like to thank all my colleagues and ex-colleagues at Electa
for making my PhD journey the most memorable one. I owe special thanks to
Jonas Engels who shared his experience in distributed and robust optimization
with me; all this work would not have carried out without his help. I am
i
ii PREFACE
also thankful for the good cooperation with Sander Claeys. Thanks for his
valuable contribution of the implementation and validation of the distributed
optimization-based voltage control algorithms. I also would like to thank Mudar
Abedrabbo for helping me with the implementation of the inverter control loops
and for his friendship.
I would like to thank all the members of the P2P-SmarTest project for their
collaboration, the constructive meetings and for sharing their insights and ideas.
I especially want to thank Ari Pouttu, Konstantin Mikhaylov and Jussi Haapola
(from Centre for Wireless Communications, University of Oulu, Finland) who
provided me with the device-to-device communication modules and the materials
of Section 4.5.
Finally, there are the people who were always there for me, who have supported
me before, during and will after. For them, I will finish in Arabic.
H@
ð QºË@ P AJ.« ùÖÞ AK. ÐY® K @ , éËAQË@
è YêË ùKAîE@ úÎ« èQºð úÍAªK é<Ë@ YÔg YªK. ð
úÍA« É¿ úÍ AÓY¯ð ,QªË@ áÓ ÕÎªË @ Ik ú¯ AQ« áK YÊË@ áK QK QªË@ ø YË@ð úÍ@ QK Y® JË@
.

Èñ¢. AÒêË ZA«YË@ B@ ½ÊÓ @ C¯ , à B@ éJË@ IÊð AÒJ¯ é<Ë@ YªK. É®Ë@ AÒêË àA¿ð ,®Kð
áÓ úæJ¢« @ AÖÏ , àAJK úæk
AëYêk. ð AîD¯ð . ÐY® K @ AÒ» .ÉÒªË@ ákð QÒªË@
. . ð P úÍ@ QºËAK
. áªÓ ð YJ Qg úÍ I KA¿ ð éJJÒJË@ Aêm' A ð
úÍ ÐY¯ ð @ , A¯ðQªÓ úÍ øY @ ð @ , àñªË@ YK úÍ YÓ áÓ É¾Ë ð
ú¯ ÑîD @ ð @ , éjJ
.QK Y® JË@ ð QºË@
ËAg úæÓ éÊ¯ ÉÒªË@ @ Yë PAm.' @
Hamada Almasalma,
August 2020.
Abstract
Control of distribution networks are facing significant challenges with the

increasing penetration of distributed energy resources. The focus of this thesis
is to develop active voltage control systems provided by photovoltaic (PV) and
PV-battery inverters to mitigate or eliminate voltage problems of distribution
networks. Various approaches that combine local and centralized voltage control
techniques have been proposed in literature. These techniques suffer from
different problems; centralized voltage control systems have poor reliability and
scalability; and local voltage control systems suffer from degraded performance.
To overcome the drawbacks of centralized and local voltage control systems,
this thesis develops novel peer-to-peer-based grid voltage support functions
(P2P-based GVSFs). Smart inverters equipped with P2P-based GVSFs interact
with each other in a P2P fashion and form a distributed voltage control (DVC)
system. Two novel methodologies are proposed in the thesis to design a P2P-
based DVC system. The first one is based on real-time distributed optimization,
while the second one is based on offline robust optimization.
In the first methodology, two gossip-based decomposition techniques are
developed: the fully distributed Dual Decomposition (DD) method and the
Jacobi-Proximal Alternating Direction Method of Multipliers (JP-ADMM). The
fully distributed DD algorithm allows for simple implementation, but in most
cases requires large number of iterations to converge. Results show that this
algorithm can be used to implement a DVC system with only few agents (e.g.
10 agents or less). The JP-ADMM algorithm requires more local computations
and communication; yet it converges much faster than the DD algorithm. The
JP-ADMM-based DVC algorithm with 50 agents, for example, needs around
1.25 minutes to converge, whereas for the same number of agents, the DD-based
DVC algorithm needs around 59.4 minutes to converge. To experimentally
validate the performance of the real-time optimization-based DVC system, the
thesis develops a novel laboratory-based P2P voltage control testbed.
There are several challenges for the decomposition-based voltage control
iii
iv ABSTRACT
technique. First major issue is its vulnerability to control instability due to

inappropriate control parameters tuning. Second, implementing a real-time
DVC system using decomposition algorithms requires the GVSFs to solve a
complex optimization problem every iteration (e.g. every 5 s). This motivates
studying the simplification of the implementation of DVC systems. In the
second methodology, the thesis simplifies the implementation of a DVC system
and reduces its computational and communication burden by applying robust
optimization to learn linear voltage control policies on a day-ahead basis. In
doing so, the inverters communicate with each other in real-time and solve only
a set of linear equations to control voltages in a distributed manner.
In the final phase of this work, the thesis combines policy-based DVC system
with frequency containment reserve (FCR), to increase value from PV-battery
systems. To this end, the thesis develops a robust mathematical optimization
program that enables PV-battery systems to simultaneously provide FCR service
to their synchronous area, and voltage regulation service to their distribution
network.
Beknopte samenvatting
De regeling en sturing van distributienetwerken staan voor grote uitdagingen

door de toename van gedistribueerde energiebronnen. De focus van dit
proefschrift is het ontwikkelen van actieve spanningsregelingen geleverd door
fotovoltaïsche (PV) omvormers en PV-batterij-omvormers om spanningsproble-
men in distributienetwerken te verminderen of te elimineren. Verschillende
benaderingen, die lokale en gecentraliseerde spanningsregelingstechnieken
combineren zijn in de literatuur voorgesteld. Deze technieken hebben nadelen:
gecentraliseerde spanningsregelsystemen hebben een beperkte betrouwbaarheid
en schaalbaarheid, en lokale systemen presteren minder goed dan centrale. Om
de nadelen van gecentraliseerde en lokale spanningsregelsystemen te overwinnen,
ontwikkelt dit proefschrift nieuwe peer-to-peer-gebaseerde functionaliteit voor
spanningsondersteuning (P2P-based grid voltage support function - GVSF).
Slimme omvormers uitgerust met P2P-gebaseerde GVSF’s werken samen in
een P2P-configuratie en vormen het gedistribueerd spanningsregelingssysteem
(distributed voltage control - DVC). In het proefschrift worden twee nieuwe
methodes voorgesteld voor dergelijke P2P-gebaseerde DVC-systemen. De eerste
is gebaseerd op realtime gedistribueerde optimalisatie, de tweede op offline
robuuste optimalisatie.
In de eerste methode zijn er twee op gossip-communicatie-gebaseerde ontbin-
dingstechnieken ontwikkeld: de volledig gedistribueerde Dual Decomposition
(DD) methode en de Jacobi-Proximal Alternating Direction Method of
Multipliers (JP-ADMM). Het volledig gedistribueerde DD-algoritme zorgt voor
een eenvoudige implementatie, maar in de meeste gevallen vereist het een groot
aantal iteraties vooraleer het convergeert. Resultaten geven aan dat dit algoritme
kan worden gebruikt om een DVC-systeem te implementeren met slechts enkele
agenten (bijv. 10 agenten of minder). Het JP-ADMM-algoritme vereist meer
lokale berekeningen en communicatie; toch convergeert het veel sneller dan het
DD-algoritme. Het op JP-ADMM gebaseerde DVC-algoritme met bijvoorbeeld
50 agenten vraagt 1,25 minuten om te convergeren, terwijl voor hetzelfde aantal
agenten het DD-gebaseerde DVC-algoritme ongeveer 59,4 minuten nodig heeft.
v
vi BEKNOPTE SAMENVATTING
Om de prestaties van het realtime op optimalisatie-gebaseerde DVC-systeem

experimenteel te valideren, heeft de proefschrift een specifiek labo testbed voor
P2P-spanningsregeling ontwikkeld.
Er zijn verschillende uitdagingen voor de ontbinding-gebaseerde spanningsrege-
lingstechnieken. Het eerste probleem is de kwetsbaarheid qua instabiliteit ten
gevolge van niet-correct afgestemde regelparameters. Ten tweede, de realtime
implementatie van het DVC-systeem met DD-algoritmen, vereist dat GVSF’s
bij elke iteratie (bijv. elke 5 s) een complex optimalisatieprobleem oplossen.
Dit suggereert om de implementatie van DVC-systemen te vereenvoudigen.
In de tweede methode, vereenvoudigt het proefschrift de implementatie van
een DVC-systeem en vermindert het de reken- en communicatielast door met
robuuste optimalisatie om een dag op voorhand het beleid voor de lineaire
spanningsregeling te leren. Hierbij communiceren de omvormers in realtime met
elkaar en lossen ze slechts een reeks lineaire vergelijkingen op om spanningen
op een gedistribueerde manier te regelen.
In het laatste deel combineert het proefschrift de DVC-regeling met frequentie-
ondersteuning (frequency containment reserve - FCR), om de waarde van
een PV-batterij-systeem te verhogen. Het proefschrift ontwikkelt hiervoor
een robuust optimalisatieprogramma dat PV-batterij-systemen in staat stelt
om tegelijkertijd de FCR-dienst te verlenen (aan hun synchrone zone) en de
spanningsregelingsdienst (aan hun distributienetwerk).
List of Abbreviations
3GPP 3rd Generation Partnership Project.
AARO Affinely Adjustable Robust Optimization.

AC Alternate Current.
ADMM Alternating Direction Method of Multipliers.
aFRR automatic Frequency Restoration Reserve.
ARO Adjustable Robust Optimization.
BCBV Branch Current to Bus Voltage.

BESS Battery Energy Storage System.
BIBC Bus Injection to Branch Current.
C-V2X Cellular Vehicle-to-Everything.

CE Continental Europe.
CLM Coupled Local Minimizers.
CVC Coordinated Voltage Control.
D2D Device-to-Device.
DC Direct Current.
DD Dual Decomposition.
DERs Distributed Energy Resources.
DisCOP Distributed Constraint Optimization.
DisCVC Distributed Coordinated Voltage Control.
DoD Depth of Discharge.
Dof Degrees of freedom.
DSO Distribution System Operator.
DVC Distributed Voltage Control.
vii
viii List of Abbreviations
FCR Frequency Containment Reserve.

FSF Frequency Support Function.
G-JP-ADMM Gossip-Based Jacobi-Proximal Alternating Di-

rection Method of Multipliers.
GS-ADMM Gauss-Seidel Alternating Direction Method of
Multipliers.
GSF Grid Support Function.
GVSF Grid Voltage Support Function.
IFS Iterative Feedback Strategy.

IoT Internet of Things.
IRENA International Renewable Energy Agency.
ITS Intelligent Transportation System.
J-ADMM Jacobi Alternating Direction Method of Multi-

pliers.
JP-ADMM Jacobi-Proximal Alternating Direction Method
of Multipliers.
KKT Karush-Kuhn–Tucker.
KPI Key Performance Indicator.
L Inductor.
LoRaWAN Long Range Wide Area Network.
LPWA Low Power Wide Area.
MGCC Microgrid Central Controller.
p.u. per unit.

P2P Peer-to-Peer.
PALF Proximal Augmented Lagrangian Function.
PCC Point of Common Coupling.
PF Power Factor.
PLL Phase-Locked-Loop.
PV Photovoltaic.
LIST OF ABBREVIATIONS ix
R Resistor.
R.Pi Raspberry Pi.
RTT Real Time Target Computer.
TSO Transmission System Operator.
UART Universal Asynchronous Receiver-Transmitter.

Nomenclature
Notation
[·](k) The subscript k indicates a value at time step k.
[·]l The subscript l indicates a household.
[·]pv The subscript pv indicates a PV installation.
[·]bat The subscript bat indicates a battery.
[·]inv The subscript inv indicates an inverter.
[·]x Index x of network’s nodes.
[·]i Index i of control nodes.
[·]j Index j of passive nodes.
[·]T Transpose of a matrix.
k · k1 First norm.
k · k2 Second norm.
acos[·] Inverse cosine.
cov[·] A covariance matrix.
diag[·] Square diagonal matrix.
E[·] Expected value.
ln[·] Natural logarithm.
Tr[·] Trace.
xi
xii NOMENCLATURE
General
v Voltage (V).
vo Nominal voltage (V).
p Active power (kW).
q Reactive power (kvar).
E Energy (kWh).
fnom Nominal frequency (Hz).
sinvi Apparent power capacity of inverter i (kVA).
v Maximum voltage limit (V).
v Minimum voltage limit (V).
∆t Time step duration (minute).
Variables and Parameters of Chapters 3 and 4
cp,invi Penalty factor of inverter active power.
cq,invi Penalty factor of inverter reactive power.
cri Active power curtailment factor of inverter i (%).
L Lagrangian Function.
LPALF Proximal Augmented Lagrangian Function.

(k)
(ppv
i ) Active power generation of PV installation i (kW).
vi Voltage magnitude of bus i.

p
vi,d Sensitivity of the voltage at bus i to the change in the
active power of bus d (V/kW).
q
vi,d Sensitivity of the voltage at bus i to the change in the
reactive power of bus d (V/kvar).
α Step size of the dual decomposition method.
γinvi Acceleration factor of inverter i.

p
γinv i
Curtailment acceleration factor of inverter i.
NOMENCLATURE xiii
q
γinv i
Reactive acceleration factor of inverter i.
ρinvi Augmented penalty factor of inverter i.
τinvi Proximal penalty factor of inverter i.

(k)
∆pinvi Change in the active power of inverter i (kW).
(k)
∆qinvi Change in the reactive power of inverter i (kvar).
(k)
∆q invi Maximum reactive power of inverter i (kvar).
λinvi Lagrangian multiplier associated with the constraint of

maximum voltage.
λinvi Lagrangian multiplier associated with the constraint of

minimum voltage.
Variables and Parameters of Chapters 5 and 6
cp Battery degradation cost (cent/kWh).
cq Reactive power compensation cost (cent/kvarh).
e bati Maximum energy capacity of battery i (kWh).
e bati Minimum energy capacity of battery i (kWh).

0
Ebati
Initial energy content of battery i (kWh).
(k)
ẼVC, bati Uncertain energy content of the VC battery i (kWh).
E 0VC, bati Initial energy content of the VC battery i (kWh).

(k)
e VC, bati Maximum energy capacity of the VC battery i (kWh).
(k)
e VC, bati Minimum energy capacity of the VC battery i (kWh).
(k)
ẼR-DN, bati Uncertain energy content of the R-DN battery i (kWh).
E 0R-DN, bati Initial energy content of the R-DN battery i (kWh).

(k)
e R-DN, bati Maximum energy capacity of the R-DN battery i (kWh).
(k)
e R-DN, bati Minimum energy capacity of the R-DN battery i (kWh).
(k)
ẼR-UP, bati Uncertain energy content of the R-UP battery i (kWh).
E 0R-UP, bati Initial energy content of the R-UP battery i (kWh).

xiv NOMENCLATURE
(k)
e R-UP, bati Maximum energy capacity of the R-UP battery i (kWh).
(k)
e R-UP, bati Minimum energy capacity of the R-UP battery i (kWh).
H, h The matrix H and the vector h are used to define the

bounds of the uncertainty set U in Chapter 6.
M R-DNi Matrix of the R-DN energy management policy i.
M R-UPi Matrix of the R-UP energy management policy i.

(k)
Lqi Coefficient of the reactive power policy i (Chapter 5).
(k)
Lpi Coefficient of the battery power policy i (Chapter 5).
(k)
pFCR FCR power (kW).
(k)
p̃lx Uncertain active power consumption of household x (kW).
(k)
p̃pvi Uncertain active power generation of PV installation i
(kW).
(k)
plx Forecasted active power consumption of household x (kW).
(k)
ppvi Forecasted active power generation of PV installation i
(kW).
(k)
plx Maximum value of the uncertain active power consumption
of household x (kW).
(k)
ppvi Maximum value of the uncertain active power generation of
PV installation i (kW).
(k)
p̃VC, bati Uncertain active power of the VC battery i (kW).
(k)
p̃R-DN, bati Uncertain active power of the R-DN battery i (kW).
(k)
p̃R-UP, bati Uncertain active power of the R-UP battery i (kW).
(k)
p̃R-DNi Uncertain downward-FCR power of battery i(kW).
(k)
p̃R-UPi Uncertain upward-FCR power of battery i (kW).
(k)
p̃m-DN, bati Uncertain active power of the R-DN energy management
policy i (kW).
(k)
p̃m-UP, bati Uncertain active power of the R-UP energy management
policy i (kW).
NOMENCLATURE xv
pbati Maximum power of battery i (kW).
pbat Minimum power of battery i (kW).

i
(k)
pm-DN, bati Power capacity of the R-DN energy management policy i
(kW).
(k)
pm-UP, bati Power capacity of the R-UP energy management policy i
(kW).
(k)
pVC, bati Power capacity of the voltage rise control (kW).
(k)
pVC, bati Power capacity of the voltage drop control (kW).
(k)
p̃R-DN, invi Uncertain active power, of inverter i, that considers
downward-FCR power (kW).
(k)
p̃R-UP, invi Uncertain active power, of inverter i, that considers
upward-FCR power (kW).
(k)
q̃R-DN, invi Uncertain reactive power, of inverter i, that takes into
account the effect of downward-FCR (kvar).
(k)
q̃R-UP, invi Uncertain reactive power, of inverter i, that takes into
account the effect of upward-FCR (kvar).
(k)
q̃lx Uncertain reactive power of household x (kvar).
(k)
rDNi Downward-FCR capacity of battery i (kW).
(k)
rUPi Upward-FCR capacity of battery i (kW).
(ṽ (k) )2R-DN Vector of squared (uncertain) nodal voltage magnitudes

that includes the effect of downward-FCR.
(ṽ (k) )2R-UP Vector of squared (uncertain) nodal voltage magnitudes

that includes the effect of upward-FCR.
W Whitening transformation matrix.
Z pi Matrix of the active power policy i (Chapter 6).

(k)
zpi Coefficient of the active power policy i (Chapter 6).
β pi Matrix of the reactive power policy i, related to the control

nodes’ active power consumption and PV power (Chapter
6).
xvi NOMENCLATURE
(k)
βpi Coefficient of the reactive power policy i, related to the
control nodes’ active power consumption and PV power
(Chapter 6).
β R-DNi Matrix of the reactive power policy i, related to the

positive normalized frequency deviation (Chapter 6).
(k)
βR-DNi Coefficient of the reactive power policy i, related to the
positive normalized frequency deviation (Chapter 6).
β R-UPi Matrix of the reactive power policy i, related to the

negative normalized frequency deviation (Chapter 6).
(k)
βR-UPi Coefficient of the reactive power policy i, related to the
negative normalized frequency deviation (Chapter 6).
∆f Frequency deviation at which maximum FCR capacity

needs to be activated (Hz).
∆fdb Frequency deadband in which no FCR reaction is required

(Hz).
(k)
∆f
f
R-DN Uncertain normalized positive frequency deviation.
(k)
∆f
f
R-UP Uncertain normalized negative frequency deviation.
(k)
∆f R-DN Maximum normalized frequency deviation.
∆f (k)
R-UP
Minimum normalized frequency deviation.
(k)
∆P lx Maximum deviation from forecasted active power
consumption of household x (kW).
∆P (k)
pvi Maximum deviation from forecasted active power of PV
installation i (kW).
ζ The vector ζ ∈ Rnζ includes all uncertain variables of the

optimization problem of Chapter 6 .
ηich Charge efficiency of battery i.
ηidis Discharge efficiency of battery i.
ξ ∆p Vector that includes uncertain sum of households’ active

power consumption and PV power generation of all control
nodes for all time steps.
NOMENCLATURE xvii
ξ R-DN Vector that includes uncertain normalized positive

frequency deviations.
ξ R-UP Vector that includes uncertain normalized negative

frequency deviations.
Counts
nnodes Number of network’s nodes.
nc Number of control nodes (inverters participating in voltage

control).
nt Number of time steps.
nζ Number of uncertain variables of the optimization problem

(Chapter 6).
Sets
I Control nodes set (inverters participating in voltage

control).
J Passive nodes set.
K Time steps set.
N Network’s nodes set.
U Uncertainty set.
Y Observers set.
Contents
Abstract iii
Beknopte samenvatting v
List of Abbreviations ix
Nomenclature xi
Contents xix
List of Figures xxv
List of Tables xxxi
1 Introduction 1
1.1 Context and Motivation . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives and Scope . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Outline and Contributions . . . . . . . . . . . . . . . . . . . . . 8
2 Peer-to-Peer Control of Distribution Networks 13

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Issues with the Integration of DERs in the Electrical Grid . . . 15
xix
xx CONTENTS
2.2.1 Voltage Issues . . . . . . . . . . . . . . . . . . . . . . . . 15

2.2.2 Frequency Issues . . . . . . . . . . . . . . . . . . . . . . 16
2.2.3 Towards a New Control Paradigm for the Distribution Grid 16
2.3 Architectures of Microgrid Control . . . . . . . . . . . . . . . . 17
2.3.1 Centralized Control Architecture . . . . . . . . . . . . . 17
2.3.2 Hierarchical Control Architecture . . . . . . . . . . . . . 19
2.3.3 Two-Level Distributed Control Architecture . . . . . . . 20
2.3.4 P2P-Based Control Architecture . . . . . . . . . . . . . 21
2.3.5 Local Control Architecture . . . . . . . . . . . . . . . . 22
2.4 Proposed P2P-Based Control Paradigm . . . . . . . . . . . . . 23
2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3 Real-Time Optimization-Based Distributed Voltage Control 27

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . 31
3.3 Branch Flow Model . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.4 Simulated Network . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4.1 PV and Household Profiles . . . . . . . . . . . . . . . . 35
3.4.2 Simulation Environment . . . . . . . . . . . . . . . . . . 36
3.5 Dual-Decomposition-Based Distributed Voltage Control . . . . 36
3.5.1 Lagrangian Relaxation . . . . . . . . . . . . . . . . . . . 37
3.5.2 Minimax of the Lagrangian Function . . . . . . . . . . . 38
3.5.3 Minimax Theorem and Strong Duality . . . . . . . . . . 40
3.5.4 The Dual Decomposition Method . . . . . . . . . . . . . 43
3.5.5 Feedback Strategy . . . . . . . . . . . . . . . . . . . . . 46
3.5.6 Fully Distributed Dual Decomposition . . . . . . . . . . 47
3.5.7 Push-Sum Gossip Protocol . . . . . . . . . . . . . . . . 48
CONTENTS xxi
3.5.8 Anti-Windup Strategy . . . . . . . . . . . . . . . . . . . 50

3.5.9 Active and Standby Modes . . . . . . . . . . . . . . . . 50
3.5.10 The Proposed Gossip-Based Dual Decomposition Algorithm 51
3.5.11 Convergence Speed of the DD-Based Distributed Voltage
Control Algorithm . . . . . . . . . . . . . . . . . . . . . 51
3.6 ADMM-Based Distributed Voltage Control . . . . . . . . . . . 53
3.6.1 Equations of the JP-ADMM Algorithm . . . . . . . . . 54
3.6.2 Acceleration Factors . . . . . . . . . . . . . . . . . . . . 56
3.6.3 The Proposed G-JP-ADMM Voltage Control Algorithm 57
3.7 Analysing the Performance of the JP-ADMM-Based Distributed
Voltage Control Algorithm . . . . . . . . . . . . . . . . . . . . . 59
3.7.1 Parameters of the Algorithm . . . . . . . . . . . . . . . 59
3.7.2 62-Bus Case Study . . . . . . . . . . . . . . . . . . . . . 60
3.7.3 124-Bus Case Study . . . . . . . . . . . . . . . . . . . . 65
3.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4 Laboratory Implementation of Peer-to-Peer Voltage Control System 71

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.2 Testbed Architecture . . . . . . . . . . . . . . . . . . . . . . . . 72
4.3 Dual-Decomposition-Based P2P Voltage Control Algorithm . . 75
4.4 Control of the Rapid Prototyping Inverter with Grid Voltage
Support Function . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.4.1 PM15FM30C Triphase Module . . . . . . . . . . . . . . 79
4.4.2 Drive of PM15FM30C Triphase Module . . . . . . . . . 80
4.5 Device-to-Device Communication . . . . . . . . . . . . . . . . . 83
4.5.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.5.2 D2D Communication Modules . . . . . . . . . . . . . . 84
4.6 Results of the Experiment . . . . . . . . . . . . . . . . . . . . . 88
xxii CONTENTS
4.6.1 First Experiment: Without P2P Voltage Control . . . . 88

4.6.2 Second Experiment: With P2P Voltage Control . . . . . 89
4.6.3 Key Performance Indicators . . . . . . . . . . . . . . . . 91
4.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5 Robust Policy-Based Distributed Voltage Control Provided by PV-

Battery Inverters 95
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.2 The Proposed Voltage Control System . . . . . . . . . . . . . . 98
5.3 Design of the Robust Offline Coordinator . . . . . . . . . . . . 99
5.3.1 Problem Formulation . . . . . . . . . . . . . . . . . . . 100
5.3.2 Uncertainty Model . . . . . . . . . . . . . . . . . . . . . 101
5.3.3 Linear Control Policies . . . . . . . . . . . . . . . . . . . 103
5.3.4 Branch Flow Model . . . . . . . . . . . . . . . . . . . . 104
5.3.5 Inverter Model . . . . . . . . . . . . . . . . . . . . . . . 105
5.3.6 Battery Model . . . . . . . . . . . . . . . . . . . . . . . 106
5.3.7 Second-Order Moment Matrix . . . . . . . . . . . . . . 107
5.3.8 Robust Optimization . . . . . . . . . . . . . . . . . . . . 108
5.4 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.4.1 Simulated Network . . . . . . . . . . . . . . . . . . . . . 109
5.4.2 The Load and PV Scenarios . . . . . . . . . . . . . . . . 110
5.4.3 Simulation Software . . . . . . . . . . . . . . . . . . . . 112
5.4.4 Simulation Results . . . . . . . . . . . . . . . . . . . . . 113
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
6 Simultaneous Provision of Voltage and Frequency Control by PV-

Battery Systems 117
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
CONTENTS xxiii
6.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . 118

6.1.2 Provision of FCR with a BESS . . . . . . . . . . . . . . 119
6.1.3 Batteries Providing Multiple Services . . . . . . . . . . 121
6.2 Contribution and General Overview of the Proposed Control
Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
6.2.1 The Proposed Control Framework . . . . . . . . . . . . 123
6.2.2 FCR Reserve Capacity . . . . . . . . . . . . . . . . . . . 125
6.3 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . 127
6.3.1 Uncertainty Model . . . . . . . . . . . . . . . . . . . . . 128
6.3.2 Linear Control Policies . . . . . . . . . . . . . . . . . . . 130
6.3.3 Allocation of BESS Capacity . . . . . . . . . . . . . . . 135
6.3.4 Battery Model . . . . . . . . . . . . . . . . . . . . . . . 138
6.3.5 Inverter Model . . . . . . . . . . . . . . . . . . . . . . . 139
6.3.6 Branch Flow Model . . . . . . . . . . . . . . . . . . . . 140
6.3.7 Logic Constraints . . . . . . . . . . . . . . . . . . . . . . 142
6.3.8 Objective Function . . . . . . . . . . . . . . . . . . . . . 143
6.3.9 Robust Optimization . . . . . . . . . . . . . . . . . . . . 145
6.3.10 Optimality Gap . . . . . . . . . . . . . . . . . . . . . . . 146
6.4 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
6.4.1 Simulated Network . . . . . . . . . . . . . . . . . . . . . 148
6.4.2 The Loads, PV and Frequency Scenarios . . . . . . . . . 149
6.4.3 Simulation Software . . . . . . . . . . . . . . . . . . . . 149
6.4.4 Simulation Results . . . . . . . . . . . . . . . . . . . . . 151
6.4.5 Loss of Optimality . . . . . . . . . . . . . . . . . . . . . 152
6.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
7 General Conclusions and Future Work 155

7.1 Summary and Contributions . . . . . . . . . . . . . . . . . . . . 155
xxiv CONTENTS
7.2 Answers to the Research Questions . . . . . . . . . . . . . . . . 156

7.3 Recommendations to Stakeholders . . . . . . . . . . . . . . . . 161
7.4 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Bibliography 167
Curriculum Vitae 187
List of Publications 189

List of Figures
1.1 Outline of the thesis. . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1 Centralized control. . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.2 Hierarchical control. . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 Two-level distributed control. . . . . . . . . . . . . . . . . . . . . 21
2.4 P2P-based control. . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.5 P2P-based control paradigm. . . . . . . . . . . . . . . . . . . . 24
3.1 General block diagram of smart PV inverter with P2P-based grid

voltage support function (PCC: point of common coupling). . . 30
3.2 Centralized optimization-based voltage control system. . . . . . 32
3.3 A simple 2-bus distribution system. . . . . . . . . . . . . . . . . 34
3.4 Schematic diagram of the simulated network. All lengths are
drawn to scale. . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.5 Decomposition of the dual problem into sub-problems controlled
by a master problem (Dual Decomposition). . . . . . . . . . . . 44
3.6 DD-based voltage control with iterative feedback strategy. . . . 47
3.7 Illustration of the DD-based voltage control with IFS applied in
the middle of the control iteration. . . . . . . . . . . . . . . . . 48
3.8 Average convergence speed of the DD-based distributed voltage
control algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . . 53
xxv
xxvi LIST OF FIGURES
3.9 Aggregated PV generation and active power consumption of the

62 households. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.10 Voltage profiles of the 62 households without voltage control. . . 61
3.11 Voltage profiles of the 62 households with voltage control. . . . 62
3.12 Voltage quality metric E of the 62 households with and without
voltage control. . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.13 (a) Lagrangian multiplier λinv62 of smart inverter No. 62, (b)
Active power curtailment and reactive power compensation of
smart inverter No. 62. . . . . . . . . . . . . . . . . . . . . . . . 63
3.14 Comparison of the convergence speed between JP-ADMM, J-
ADMM and DD. . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.15 Predicted voltage vs. actual voltage of smart inverter No. 62. . 65
3.16 (a) Voltage regulation with the feedback strategy, (b) Voltage
regulation without the feedback strategy. . . . . . . . . . . . . 66
3.17 (a) Voltage profiles of the 124-bus network without voltage
control, (b) Voltage profiles of the 124-bus network with voltage
control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.18 Voltage convergence of smart inverter No. 62 (62-bus network)
and smart inverter No. 124 (124-bus network). . . . . . . . . . 67
3.19 124-bus network: Comparison of the convergence speed between
JP-ADMM, J-ADMM and DD. . . . . . . . . . . . . . . . . . . 68
3.20 Number of agents to be communicated with per control iteration. 68
4.1 Multi-layer multi-agent architecture of the P2P voltage control

testbed (VM: voltmeter, D2D: device-to-device communication
module, GVSF: grid voltage support function, R.Pi: raspberry
pi computer, R: resistor, L: inductor, labels (1) to (11) indicate
different parts of the testbed). . . . . . . . . . . . . . . . . . . . 74
4.2 Schematic of the P2P voltage control testbed (VM: voltmeter,
D2D: device-to-device communication module, R: resistor, L:
inductor, RTT: real time target computer, R.Pi: raspberry pi
computer, L-L: line-to-line, labels (1)-(12) indicate the different
parts of the testbed, labels (1)–(11) same as in Figure 4.1). . . 75
4.3 The laboratory microgrid (labels same as in Figure 4.1) . . . . 76
LIST OF FIGURES xxvii
4.4 Circuit diagram of Triphase PM15FM30C rapid prototyping

inverter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.5 Actuator agent setup (labels same as in Figures 4.1 and 4.2). . . 81
4.6 Inner control system of the inverter integrated with the GVSF
(labels same as in Figures 4.1 and 4.2). . . . . . . . . . . . . . . 82
4.7 Structure of the D2D communication module (USB: universal
serial bus, UART: universal asynchronous receiver-transmitter,
FTDI: future technology devices international (semiconductor
device company), VCP: virtual communication port). . . . . . . 85
4.8 Structure of the embedded software. . . . . . . . . . . . . . . . 86
4.9 Illustration of the implemented synchronized protocol operation
(TX: transmission, RX: reception). . . . . . . . . . . . . . . . . 87
4.10 30 min generation profiles of inverter 1 and inverter 2 (both
inverters apply the same generation profile). . . . . . . . . . . . 88
4.11 Voltage profiles without voltage control for the 3 nodes of the
microgrid (the generation profile causes over-voltages up to 1.145
p.u.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.12 Voltage profiles with voltage control for the 3 nodes. . . . . . . 89
4.13 (a) Active power curtailment of inverter 1 and inverter 2; (b)
Reactive power compensation of inverter 1 and inverter 2; (c)
Control signals for over-voltages; (d) Control signals for under-
voltages. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.14 Voltage quality metric: sum of the surfaces above and below the
voltage limits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.15 Communication delays: The observer agents update the control
signals every 1.5 s, with very little deviation. The actuator agents
issue their updates more slowly, with a significant difference
between both actuator agents. . . . . . . . . . . . . . . . . . . . 93
5.1 General overview of the proposed voltage control system. . . . 99

5.2 Schematic diagram of the network used in this case study. . . . 110
5.3 General overview of the MATLAB simulation used in the case
study. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.4 Voltage profiles of control node 31 with no voltage control. . . . 113
xxviii LIST OF FIGURES
5.5 Voltage profiles of control node 31 with voltage control. . . . . 113

5.6 (a) Apparent power magnitude of inverter 31, (b) Reactive
power of inverter 31, (c) Energy content of battery 31, (d)
Charge/discharge power of battery 31. . . . . . . . . . . . . . . 114
6.1 The results of this figure are obtained based on the implemen-
tation of the policy-based DisCVC proposed in Chapter 5, in a
simulation over 104 scenarios. The scenarios are presented in
Subsection 6.4.2. The DisCVC is used to regulate voltage profiles
of the 62-node network presented in Subsection 6.4.1. This figure
presents the charging power of BESS number 31. . . . . . . . . 119
6.2 Energy content of the 13.5 kWh battery over 730 frequency
scenarios. Battery parameters: maximum power= 7 √kW,
initial charge= 6.5 kWh, charge/discharge efficiency= 0.9.
The energy content E is calculated based on: E (k+1) =
(k) + (k) +
E + ∆t ηi pFCR − ηdis
(k) ch 1
−pFCR , where ηich and ηidis are,
i
respectively, the charge and discharge efficiencies , ∆t is the time
+
step duration and [·] ≡ max(·, 0). . . . . . . . . . . . . . . . . . 121
6.3 General overview of the proposed control framework. . . . . . . 124
6.4 (a) Unregulated voltage profile of a prosumer in a network with
high PV penetration, (b) Expected upward and downward reserve
capacity profile. In this example, the inverter reactive power and
the VC battery power are used to solve the voltage rise problem,
therefore, the upward reserve capacity is zero in Area 2. . . . . 126
6.5 Normalized positive and negative frequency deviation ∆f of
730 days of historical frequency data (2017-2018). The dotted
lines show the maximum and minimum normalized frequency
deviations. ∆f is calculated based on (6.1b). The frequency data
are from the CE synchronous region. . . . . . . . . . . . . . . . 129
6.6 Illustration of energy and power capacity allocation of battery i. 136
6.7 Schematic diagram of the network used in this case study. . . . 149
6.8 Aggregated power capacities of the downward-FCR, upward-FCR,
voltage rise, energy management policies of R-DN and R-UP
batteries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
6.9 Power capacity reserved for charging the VC battery 31. . . . . 152
LIST OF FIGURES xxix
6.10 (a) Voltage profiles of node 31 with no control, (b) Voltage profiles
of node 31 with voltage and frequency control. The dotted lines
show the maximum and minimum voltage limits . . . . . . . . 153
6.11 (a) Apparent power magnitude of inverter 31, (b) Energy content
of battery 31. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
7.1 P2P-based platform (GSF: grid support function). . . . . . . . 165

List of Tables
4.1 Voltage quality metrics of the regulated and unregulated voltage

profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.1 Parameters of inverters, batteries and loads. . . . . . . . . . . . . 111
6.1 Parameters of inverters, batteries and loads. . . . . . . . . . . . 150
xxxi
Chapter 1
Introduction
1.1 Context and Motivation
Over the last decade, the worldwide deployment of photovoltaic (PV) systems
has been increasing exponentially due to the cost reductions in renewables,
advances in digital technologies, the supporting governments’ policies and
the increased customer awareness. According to the IEA’s1 2019’s Trends in
Photovoltaic Applications report [1], at the end of 2018 the globally cumulative
PV capacity reached over half a TW. It is also predicted that the global PV
capacity will continue to grow by 9% every year in the next 50 years [2],
according to the 2019 study released by the International Renewable Energy
Agency (IRENA). This rapid growth of PV penetration in power system helps
the nation to fulfill their electrical energy needs without increasing fossil fuel
consumption, which largely contributes to reducing greenhouse gas emissions
and mitigating climate change effects.
Although PV systems are a key technology for implementing the shift to a
decarbonized energy supply, the current distribution networks were not designed
to incorporate this type of generation. Consequently, grid operators are facing
several problems in controlling distribution networks. One of the most common
problems is voltage deviation from acceptable range defined by the current
standards (e.g. EN 50160 [3]). For example, high PV generation during low
demand might reverse power flows in the grid and create voltage rise issues,
while sudden drop in PV generation during peak demand period might create
voltage drop issues. Furthermore, intermittent and unpredictable nature of PV
1 International Energy Agency
1
2 INTRODUCTION
systems increases the complexity of regulating voltages of distribution networks

with high PV penetration [4].
The traditional solution to alleviate voltage problems is to reinforce distribution
networks (e.g. installing more cables). The main disadvantage of this solution
is the relatively high investment costs [5]. A transition towards active voltage
management strategies would be capable of maintaining voltage profiles of
distribution networks within acceptable limits while minimizing, deferring, or
even avoiding any network upgrades [6]. Smart inverters can contribute to active
voltage control, by expanding their features with added grid voltage support
functions (GVSFs) and without major hardware modifications [7, 8, 9, 10, 11].
This thesis focuses on voltage regulation support provided by smart PV inverters.
These inverters have the ability to absorb (or inject) reactive power, curtail PV
power, charge battery energy storage systems (BESSs) during PV peak period
and discharge BESSs during demand peak period.
The practice of using smart inverters to regulate grid voltages is rather new
in Europe. PV inverters were initially designed for low PV penetration levels.
They were intentionally configured to inject as much active power as available
from PV modules. These inverters have been programmed to monitor the Point
of Common Coupling (PCC) voltage and disconnect immediately after sensing
the PCC voltage exceeding certain limits, to comply with the interconnection
standards. The earlier versions of the standards IEEE 1547 and UL 1741 prevent
PV inverters from providing any type of grid support, and thus prohibit these
inverters from actively participating in distribution system operation. Not until
lately did the interconnection standards for distributed energy resources allow
smart inverters to regulate voltages. IEEE updated the standard IEEE 1547
in 2014 (1547a – Amendment 1) and UL 1741 was updated in 2016 (UL 1741
Supplement A) [12, 13]. The new standards identify inverter functions required
for grid stability.
Grid Voltage Support Functions of Smart Inverters
Control algorithms of the GVSF can be designed mainly based on three

strategies [14, 15]: communication-based centralized control, communication-
less decentralized control (local control), and communication-based distributed
control. A centralized voltage control system uses a central optimizer to
coordinate the GVSFs of the different inverters participating in voltage control.
It provides optimal coordination, taking into account the complete system
behaviour. The central optimizer has access to all information about the
inverters in its control area. Consequently, this centrally organized structure
can coordinate the GVSFs based on exact information about local and global
CONTEXT AND MOTIVATION 3
objectives. However, centralized control of a large-scale distribution networks

will reach the limits of scalability, computational complexity and communication
overhead. Moreover, centralized control is a single point of failure [16, 17].
A decentralized communication-less voltage control system, on the other hand,
mitigates the computational and communicational burden by using GVSFs
with more advanced local processing capabilities. This system does not rely
on any form of communication. Each GVSF is designed by ignoring the
interactions from other GVSFs, the control decisions are based only on the
available local information. In this case, optimal operation cannot be reached,
as it is impossible to know the complete state of the distribution network and
all operational boundary conditions of other inverters [18, 19].
A promising approach to overcome the drawbacks of centralized and
decentralized control is distributed voltage control (DVC). DVC is characterized
by absence of central optimizer. Every GVSF is considered as an autonomous
control agent. The control agents are equal and build a network of peers.
To overcome the absence of the central decision making controller the
peers communicate with each other in a peer-to-peer (P2P) fashion. With
communication, they are able to make the correct control decisions in every
particular situation, to maintain the voltage at any node in the network within
required limits. P2P-based DVC systems have greater controller redundancy
and are robust to a single point of failure; failure of one GVSF will not cause
system blackout. Furthermore, distributed voltage control systems are more
scalable and flexible for future modifications and expansions [20, 21, 22].
Distributed Constraint Optimization Algorithms
Distributed Constraint Optimization (DisCOP) techniques are mostly used in

literature to implement and optimize DVC systems [23, 24, 25]—for example,
minimizing, in a distributed manner, a set of objectives related to reactive power
provision, active power curtailment and battery ageing, while maintaining grid
voltages within accepted limits.
DisCOP algorithms deal with iterative decomposition techniques [26], which
iterate over a sequence of steps to reach a (near) globally optimal solution
(under certain assumptions). Hence, one of the main challenges in using real-
time DisCOP for implementing P2P-based GVSFs is to reduce the number of
iterations (as much as possible) to solve a voltage problem within a reasonable
time.
Iterative decomposition algorithms appearing in literature are mainly based on
the Dual Decomposition (DD) method and the Alternating Direction Method
4 INTRODUCTION
of Multipliers (ADMM). DD and ADMM have been used in many publications

to design distributed voltage control algorithms [27, 28, 29]. These algorithms
require a large number of iterations to converge. Hence, they fail in regulating
grid voltages within a reasonable time. Moreover, these algorithms have been
implemented mostly by means of simulators (e.g. MATLAB/Simulink); real
deployments of DVC systems are thus lacking. Therefore, there is a need of
developing novel optimization-based DVC algorithms that enjoy fast convergence
rate and are suitable for real-time voltage control. Additionally, there is a need
of laboratory implementations of DVC systems, to better understand the real
implementation of these systems.
Complexity of Decomposition Techniques
Despite the advantages of decomposition techniques, their implementation is

complex. There are several challenges for the decomposition-based voltage
control techniques. First major issue is their vulnerability to control instability
due to inappropriate control parameters tuning. These techniques are sensitive
to some parameters (e.g. step size and penalty factors) that should be tuned
carefully to obtain satisfactory results. A decomposition algorithm may not
converge if its parameters are not set correctly—for example, the algorithm may
diverge if the iteration step size is large, and it may converge way too slow if the
step size is small. Tuning these parameters is not trivial, and requires several
experiments of trial and error. Second, implementing a real-time distributed
voltage control system using decomposition algorithms means that the GVSFs
have to solve a complex optimization problem every iteration (e.g. every 5 s);
this requires multiple communication exchanges among inverters every iteration.
The complex implementation of the iterative decomposition algorithms motivates
studying how the implementation of real-time DVC systems can be simplified.
Value Stacking of Battery Energy Storage Systems
Behind-the-meter battery energy storage systems can play an important role in

regulating voltage profiles of distribution networks with high PV penetration
[30, 31, 32]. BESSs can be paired with smart PV inverters to control their
energy output in response to violation of grid voltage. When BESSs are used
to support PV inverters in regulating voltage profiles, they are inactive or
partially used most of the time. Their unused capacity could be used to provide
other services, allowing them to generate additional revenues. Hence, there is
a need of designing innovative control systems that enable BESSs to provide
simultaneously multiple services (so-called value stacking) [33, 34, 35, 36]. Since
frequency control has been identified as one of the highest value services for
OBJECTIVES AND SCOPE 5
BESSs [37], the authors are interested in combining distributed voltage control
with frequency control.
1.2 Objectives and Scope
This thesis aims to tackle the aforementioned challenges with respect to over-
voltage and under-voltage problems of distribution networks with high PV
penetration. Owning to the drawbacks of centralized and decentralized voltage
control systems, the thesis focuses on distributed voltage control systems. The
main objective of this thesis is twofold. The first is to develop P2P-based GVSFs
for smart PV inverters and PV-battery inverters2 . The aim is to enable smart
inverters to communicate with each other in a P2P fashion and regulate grid
voltages cooperatively; without relying on a central decision-making controller.
The second is to develop frequency support functions (FSFs) for PV-battery
inverters, and combine them with the GVSFs. The aim is to enable PV-battery
systems to provide simultaneously frequency containment reserve service (to
their synchronous area) and DVC service (to their distribution network). There
are five underlying objectives to reach the main objective, as discussed in the
following paragraphs.
Objective 1: Propose a general P2P-based control paradigm for distribution

networks
The first goal of the thesis is to help the readers understand the linkage of the
thesis work to the big picture of P2P-based concept in distribution networks.
To this end, this thesis will first look at different architectures of distribution
networks control. It will then present a general P2P-based control architecture.
The research questions addressed in this objective are:
• How P2P-based control can be imposed on the distribution level of the

grid?
• How the thesis work contributes to the proposed P2P-based control

paradigm of distribution networks?
2 A PV-battery inverter is an inverter which can simultaneously manage the electricity
outputs of both PV panels and batteries; charging batteries with either PV panels or the grid.
6 INTRODUCTION
Objective 2: Develop an optimization-based distributed voltage control

algorithm with fast convergence rate
The second goal of the thesis is to implement a fast iterative decomposition

method to enable P2P-based GVSFs to solve voltage problems within a
reasonable time. To this end, this thesis will first study the convergence
speed of DD-based and ADMM-based DVC algorithms to see if they can be
fast enough to solve a voltage problem in real-time. It then will improve the
convergence speed of these algorithms. The research questions addressed in this
objective are:
• To which extent can the number of inverters participating in the DD-based

DVC system be increased?
• What is the effect of extending the 2-block ADMM to parallel multi-block
ADMM on the convergence speed of the ADMM algorithm?
• How the convergence speed of the ADMM-based DVC algorithm can be
accelerated?
Objective 3: Develop a laboratory-based distributed voltage control system
The third goal of the thesis is to illustrate the practical implementation of DVC
systems. To this end, this thesis will develop a laboratory-based microgrid
equipped with P2P-based PV inverters. The research question addressed in
this objective is:
• How to validate experimentally a real-time optimization-based distributed

voltage control?
Objective 4: Simplify the implementation of distributed voltage control

systems
The fourth goal of the thesis is to reduce the complexity and communication
burden of DVC systems. To this end, this thesis will develop a DVC system
enabled by linear policies. The idea is to let smart inverters solve, each time step
in real-time, a set of linear equations rather than solving a complex optimization
problem. The research question addressed in this objective is:
• How forecast data of residential active power consumption and PV power

generation can be used to compute robust linear voltage control policies?
OBJECTIVES AND SCOPE 7
Objective 5: Combine distributed voltage control with frequency control
The final goal of the thesis is to maximize exploitation of PV-battery systems

that provide voltage control. To this end, this thesis will develop a control
methodology for PV-battery systems to provide simultaneously DVC and
frequency containment reserve. The research question addressed in this objective
is:
• How to design a control methodology that optimally combines policy-based

distributed voltage control with frequency containment reserve?
P2P-SmarTest Project
This research fits within the wider scope of the EU funded Peer-to-Peer Smart
Energy Distribution Networks (P2P-SmarTest) project3 , in which the author
actively contributed. The overall objective of this project was to employ P2P
approaches to ensure the integration of demand side flexibility and the optimum
operation of distributed energy resources, while maintaining power balance
and the quality and security of the supply. The focus was to develop and
experimentally validate intelligent P2P-based control, communication and
trading algorithms. The contributions of the KU Leuven, and specifically
the ELECTA research group, in this project were:
• Development of alternative P2P-based control paradigm of distribution

networks. The P2P-based control paradigm that will be proposed in
Chapter 2 has served as an input to this contribution.
• Integration of control functions to enable and facilitate P2P-based energy
trading and better network operation. The DD-based voltage control
algorithm that will be implemented in Chapter 3 has served as an input
to this contribution.
• Experimental validation of P2P-based distributed voltage control systems.

The laboratory implementation of Chapter 4 has served as an input to
this contribution.
• Development of qualitative and quantitative models for a more resilient
P2P operation of distribution networks.
3 Project website: www.p2psmartest-h2020.eu
8 INTRODUCTION
1.3 Outline and Contributions
Main Contributions
The main contributions of the thesis are:
• Proposing a high-level functional architecture for P2P-based control

paradigm of distribution networks.
• Implementation of a novel fast real-time optimization-based DVC
algorithm.
• Development of a multi-layer multi-agent P2P-based voltage control
testbed.
• Implementation of a novel policy-based DVC algorithm.
• Implementation of a novel control methodology for PV-battery systems

to provide simultaneously DVC and FCR.
Outline
The outline of this thesis is visualized in Figure 1.1. In Chapter 2, a P2P-based

control paradigm is proposed for controlling the distribution grid. Chapter 2
creates a framework for the rest of the chapters. In Chapter 3, P2P-based voltage
control algorithms are developed based on distributed optimization techniques.
The work of Chapter 3 is validated experimentally in Chapter 4. The work
of Chapter 5 is motivated by the complex implementation of the distributed
optimization algorithms applied in Chapter 3. In Chapter 5, a novel control
methodology is proposed to implement a distributed voltage control system.
The methodology is based on robust optimization techniques and linear decision
rules. The robust optimization problem developed in Chapter 5 is extended
in Chapter 6 to enable the PV-battery systems to provide simultaneously
voltage and frequency control. The overall conclusions and recommendations
for future research are discussed in Chapter 7. The chapters of this thesis can
be summarized as follows:
Chapter 2: Peer-to-Peer Control of Distribution Networks
This chapter classifies distribution networks control from highly centralized to

decentralized control architectures. The chapter proposes a P2P-based control
OUTLINE AND CONTRIBUTIONS 9
Chapter 2
Towards a new control paradigm for distribution networks
Chapter 3 Chapter 5
Real-time
Optimization-based Offline robust Policy-based
distributed
distributed voltage optimization distributed voltage
optimization
control control
Development of a new
methodology to simplify the
Chapter 6
Chapter 4 implementation of distributed
Simultaneous provision
Laboratory voltage control systems
of voltage and
implementation frequency control
Chapter 7
Conclusions and future work
Figure 1.1: Outline of the thesis.
paradigm for distribution networks with a high penetration of distributed energy

resources. The work of this chapter is the result of a collaboration with my
KU Leuven colleague Jonas Engels, and has been published in proceedings
of the 8th IEEE Benelux Young Researchers symposium in electrical power
engineering, 2016 [38].
Chapter 3: Real-Time Optimization-Based Distributed Voltage Control
The contribution of this chapter is four-fold. First, the chapter modifies the state-
of-the-art DD algorithm to implement a fully distributed DD method. It then
applies the fully distributed DD method to design a DVC system. Second, the
chapter speeds up the convergence of the DVC system significantly by applying
the Jacobi-Proximal ADMM method (JP-ADMM). Third, three strategies are
proposed in this chapter to improve the performance of the proposed algorithms:
feedback strategy; anti-windup strategy; and active/standby modes. Fourth,
the chapter integrates a push-sum gossip protocol with the DD-based and JP-
10 INTRODUCTION
ADMM-based voltage control systems, to enable P2P data interchange between

inverters. A real low voltage residential semiurban feeder from the region of
Flanders (Belgium) is used as a case study to analyze the performance of the two
algorithms. The DD-based DVC algorithm has been published in CIRED-IET
Open Access Proceedings Journal, vol. 2017, no. 1, pp. 1718–1721 [39]. The
JP-ADMM-based DVC algorithm has been published in Applied Energy, vol.
239, pp. 1037–1048, 2019 [40].
Chapter 4: Laboratory Implementation of Peer-to-Peer Distributed Voltage

Control System
This chapter presents the design, development and hardware setup of a

laboratory-based P2P voltage control testbed. The testbed is used to evaluate
the performance of the proposed P2P-based GVSFs, and show how DVC
systems can be deployed in real implementations. This chapter is the result of
a collaboration with my KU Leuven colleague Sander Claeys, and researchers
from Centre for Wireless Communications (University of Oulu): Konstantin
Mikhaylov; Jussi Haapola; and Ari Pouttu. The work has been published in
Energies, vol. 11, pp. 317–337, 2018 [41].
Chapter 5: Robust Policy-Based Distributed Voltage Control Provided by

PV-Battery Inverters
This chapter proposes a DVC system enabled by robust linear policies. The
proposed voltage control consists of an offline coordinator and real-time DVC.
In the offline coordinator, the voltage control problem is formulated as a
chance-constrained stochastic optimization problem, where the control actions
are expressed as linear function of the uncertain households’ active power
consumption and PV power. In real-time, the inverters communicate with each
other and apply the robust policies to regulate voltage profiles within limits.
Simulations over 36500 scenarios are used to demonstrate the robustness of the
proposed voltage control system. The work of this chapter has been published
in IEEE ACCESS, vol. 8, pp. 124939-124948, 2020 [42].
Chapter 6: Simultaneous Provision of Voltage and Frequency Control by

PV-Battery Systems
This chapter proposes a novel control methodology for PV-battery systems to

provide simultaneously DVC and frequency containment reserve. The control
methodology is structured in two phases. In the day-ahead phase, the control
OUTLINE AND CONTRIBUTIONS 11
problem is formulated as a robust optimization problem. The aims of this

optimization problem are to allocate fractions of the energy and power capacity
of each BESS to the two services, minimize the expected cost of reactive
power compensation and batteries degradation, maximize profits from frequency
control, and compute a set of linear control policies. The optimization problem
also aims to immunize against service unavailability, and violating operational
limits. In the real-time phase, the linear policies are applied to regulate voltage
profiles, and keep energy contents of batteries within limits while providing
frequency control. Simulations over 104 scenarios are used to demonstrate the
robustness of the proposed control methodology. The work of this chapter has
been published in IEEE ACCESS, vol. 8, 2020 (Early Access) [43].
Chapter 7: Conclusions and Future Work
This chapter concludes the work and presents recommendations for future
research.
This research was funded by the P2P-SmarTest project from its beginning in
November 2015 until the end of 2016. After that, the research was awarded with a
PhD fellowship by the Research Foundation – Flanders (Fonds Wetenschappelijk
Onderzoek -Vlaanderen, FWO). The work of Chapter 2 and the development of
the DD-based DVC algorithm in Chapter 3 were funded by the P2P-SmarTest
project. The development of the JP-ADMM-based DVC algorithm in Chapter
3 and the work of Chapters 5 and 6 were funded by FWO. The work of Chapter
4 was funded by both the P2P-SmarTest project and FWO.
Chapter 2
Peer-to-Peer Control of
Distribution Networks
This chapter classifies distribution networks control from highly centralized to

decentralized control architectures. The chapter then proposes a P2P-based
control paradigm to overcome the complexity of controlling distribution networks
with a high penetration of distributed energy resources (DERs). The aim of
this chapter is to help the readers understand the linkage of the thesis work to
the big picture of distribution networks control and P2P-based control. The
contents of this chapter are based upon the paper:
• H. Almasalma, J. Engels, and G. Deconinck, “Peer-to-peer control of

microgrids,” in Proceedings of the 8th IEEE Benelux Young Researchers
symposium in electrical power engineering: YRS2016, 12-13 May,
Eindhoven University of Technology, 2016.
2.1 Introduction
Growing concerns about energy sustainability, security of supply and an

increasing penetration of renewable energy and other DERs, such as storage
systems and electrical vehicles, are impacting the operation and the architecture
of the electricity system. While currently placing a burden on the distribution
grid, it is generally agreed that DERs could also be used for active grid control,
thereby contributing to a stable and secure grid. To accomplish this, new control
systems have to be designed that are able to fully harness the potential of the
13
14 PEER-TO-PEER CONTROL OF DISTRIBUTION NETWORKS
installed DERs [44]. This active monitoring and control of the distribution grid
is commonly referred to as an essential part of the smart grid, which is regarded
to be key in the future integration of electricity consumers, generators and those
that do both (prosumers).
Current distribution networks are not designed to accommodate a large amount
of DERs. A large penetration of DERs may create problems to maintain the
quality of supply to all customers connected to the distribution network. Besides,
the intermittent nature of DERs can create issues with the second-by-second
balance of demand and supply. However, by coordinating the DERs, these
issues could be resolved without the need for additional investments in grid
infrastructure [45].
In literature [46, 47, 48], the idea of a microgrid is an often mentioned alternative
to controlling the whole distribution grid with a large amount of DERs. The
main idea is that, when there are many DERs in a wide network, it can be very
complex and difficult to control. Thus, a potential way to manage this complexity
is by breaking down the entire grid into smaller microgrids, containing only
a limited amount of DERs. This chapter elaborates this idea and proposes
an operational control paradigm for the future distribution grid, based on the
concept of microgrids.
When considering such a microgrid and the coordination of multiple microgrids,
different control methods can be found in literature [49, 50, 51]. This chapter
proposes a taxonomy of these control methods, from fully centralized to
completely decentralized. A fully distributed control method, the P2P control
architecture, is further elaborated in this chapter, as it is a promising way for
future control of the distribution grid. Since the DERs are typically highly
distributed, operated by many different owners and with different objectives, it
is desirable that the microgrid control system operates in a highly distributed
way as well. Besides, a robust control system is needed that does not depend
on a single point of failure, as most of the more centralized control methods do.
The rest of the chapter is organized as follows: Section 2.2 introduces the most
prominent issues with regard to the integration of DERs in the electrical grid,
together with an elaboration of the microgrid concept that would be able to
overcome these issues. Section 2.3 proposes a categorization for the different
architectures of microgrid control. Section 2.4 then proposes a new control
paradigm for the distribution grid, based on the microgrid concept and the P2P
architecture. Finally, the chapter is concluded in Section 2.5.
ISSUES WITH THE INTEGRATION OF DERS IN THE ELECTRICAL GRID 15
2.2 Issues with the Integration of DERs in the

Electrical Grid
This section summarizes the most prominent issues with regard to the integration
of DERs in the current electrical grid. Both voltage issues and frequency or
stability issues are discussed. Other type of issues, such as harmonics, security
issues and power fluctuations are not discussed here. The microgrid concept is
presented as a possible solution to overcome these issues.
2.2.1 Voltage Issues
A large penetration of DERs may create problems to maintain the voltage

quality of all customers connected to the same part of the distribution network.
Up till now, voltage quality in the distribution grid is achieved based on the
lay out of grid infrastructure that is capable of operating within limits even in
worst case scenarios, with the assumption of unidirectional power flows. The
planning of the infrastructure is quite straightforward: minimum and maximum
load conditions are considered and minimum and maximum voltages in the grid
are examined. The network is dimensioned in such a way that the minimum
voltage is near the lower limit of the allowed voltage range and the maximum
voltage is near the upper limit of the allowed voltage range. When connecting
significant amounts of distributed generation to the network, the assumption of
unidirectional power flows is not always valid any more and the voltage profile of
the network can be quite different than in the case without any generation. With
maximum load conditions, distributed generation increases the voltage level
in the network and, hence, enhance the voltage quality in the grid. However,
when the load on the network is at a minimum, the generated power of the
distributed generation can reverse the power flows in the grid, what could lead
to a rise of the voltage profile beyond its allowed limits [52, 53, 54].
Therefore, the hosting capacity for renewable generation of many traditional
distribution networks is limited by the voltage variations that occur between
maximum and minimum load conditions [55, 56]. The traditional solution to
this problem is to reinforce the local distribution grid by installing more cables
[57]. However, generally this is quite expensive, as new infrastructure has to
be installed in residential neighbourhoods. Another approach is by using the
already installed infrastructure in a more optimal way, by coordination of the
local generation, on-load tap-changers or other equipment that can be used
to control voltage in distribution networks. A comprehensive literature review
on active voltage control methods for distribution networks is presented in
[15, 58, 59, 60].
2.2.2 Frequency Issues
The consumption and generation in the electricity system has to be balanced on

a second-by-second basis. Traditionally, this balance is maintained by flexible
generation units that are standby and are able to regulate their generated
amount of electricity as required. However, the intermittent and unpredictable
nature of new renewable energy sources creates issues with this traditional
approach, as these new sources are usually not dispatchable.
Any deviation from the demand and supply balance results in a deviation of
the system frequency from its nominal value, while large frequency deviations
will affect the frequency stability. Therefore, the system operators maintain
frequency within strict limits by using ancillary services for balancing of the
grid, that are able to respond within various time frames. Up till now these
balancing services are exclusively organized by the transmission system operator
(TSO). However, with the high penetration of DERs, maintaining the supply
and demand balance and thus the system frequency within limits becomes
more challenging. Power system operators are increasingly seeking new reserves
for frequency response from DERs (e.g. battery energy storage systems) and
demand flexibility. As these sources are connected to the distribution grid, this
means that the distribution system operators (DSOs) will be involved in this
process. A comprehensive literature review on control methods of DERs and
demand flexibility for frequency support is presented in [61, 62, 63].
2.2.3 Towards a New Control Paradigm for the Distribution

Grid
One of the key solutions to overcome the above mentioned challenges of

integrating DERs in the distribution grid is the design of a smart grid, containing
control systems for the coordination of these DERs, thereby ensuring a reliable,
secure and economical operation of the distribution networks at all times.
Controlling distribution networks to be able to utilize the emerging diversity of
DERs at significant levels of penetration, means that the control system has
to be able to manage a wide and dynamic set of resources. Controlling such a
complex network is not a trivial task. A potential way to manage this complexity
is by breaking the entire grid down into smaller microgrids, containing only a
limited amount of DERs. These microgrids should be able to control their local
resources as optimal as possible, while being connected to the rest of the grid
through a point of common coupling (PCC) [64].
ARCHITECTURES OF MICROGRID CONTROL 17
These microgrids can be coordinated on a higher level into a system of multi-

microgrids. The main idea is to design the control of the microgrid in such
a way that the microgrid is perceived by the main grid as a single element
responding to appropriate control signals.
A conventional way of controlling such a microgrid, is by a hierarchy of three
control levels, each operating on a different timescale and with a different
priority: primary, secondary and tertiary [65, 66]. Primary control is focused
on keeping the grid stable in all circumstances, thus it needs the largest priority
and should act on the smallest time scale. A robust control method is needed,
so that in case e.g. the communication network fails, the primary control is still
able to maintain stability of the microgrid. Therefore, as little communication
as possible is desired. As primary control is often implemented as some kind
of proportional controller, a steady state error remains, that shall have to be
eliminated with the use of a secondary control method. Secondary control is
activated on a slower time scale, e.g. every 15 minutes, and with a lower priority
than primary control. Finally, tertiary control is implemented as the slowest
level of control, with as purpose the economically optimal operation of the
microgrid. Both secondary and tertiary controls usually require at least some
kind of communication, as knowledge about the state of the entire system is
needed.
These three control levels can be implemented through various organizational
architectures of the microgrid. In the following section, this chapter proposes a
classification for these different architectures.
2.3 Architectures of Microgrid Control
The control of the DERs in a single microgrid can be organised according to

many different control architectures [67]. They can range from fully centralized
control where all decisions are made by a single central controller, to completely
decentralized controls where all decisions are made by the local DERs. The
required communication architecture changes accordingly. This chapter identifies
five different approaches: centralized control; hierarchical control; two-level
distributed control; fully distributed P2P-based control; and local control.
2.3.1 Centralized Control Architecture
In an fully centralized design shown in Figure 2.1, all available measurements

of the considered microgrid are gathered in a central controller that determines
the control actions for all units [68]. The advantage of a centralized control
system is that the central system receives all necessary data of the microgrid,
and based on all available information the multi-objective controller can achieve
globally optimal performance. As there is only one controller, this results in a
high controllability of the system. However, this high performance comes at
a cost. First of all, the computational burden is heavy, as the optimization is
computed based on a large amount of information. Moreover, a centralized
controller is a single point of failure and redundancy of the central controller is
expensive. The loss of communication with the central controller may cause
a shutdown of the overall system. Besides, as all system states and boundary
conditions have to be known at the central point, this requires a high quality
of communication from all DERs to the central point of control. There is also
the concern that the owners of the different DERs are not willing to hand over
control of their resources to a third party. Finally, central systems are usually
regarded as not being very scalable and system maintenance requires complete
shutdown. To overcome these issues, more distributed control architectures are
developed, as described in the subsections below.
Centralized controller
Optimization
Prosumer Prosumer
Prosumer Prosumer
Prosumer
Prosumer Prosumer
Figure 2.1: Centralized control.
Centralized control architecture has been used widely in literature to implement

centralized voltage control systems [69, 70, 71]; centralized frequency control
systems [71, 72]; and centralized coordination of DERs [73, 74, 75].
2.3.2 Hierarchical Control Architecture
A first step towards a more distributed control architecture is the introduction

of a hierarchical system, as shown in Figure 2.2. In this case, there exists
some kind of aggregation of the local DERs towards the central controller [76].
Typically, the characteristics of the DERs are represented by a few heuristics
or parameters that are combined by an aggregator, who is able to offer these
aggregated resources to a central optimizer. The central controller is then able
to dispatch the necessary resources, through a hierarchical system of aggregators,
who determine which DERs should be used at which moment. Therefore, these
methods are also referred to as aggregate and dispatch methods. As the resources
are offered to the central optimizer in an aggregated way, there is considerably
less information needed at the central controller, which results in a more
scalable system. However, the single point of failure remains, and the points of
aggregation might even become new points of failure.
Centralized controller
: Optimization
Aggregator Aggregator
Prosumer Prosumer
Prosumer Prosumer
Prosumer
Prosumer Prosumer
Figure 2.2: Hierarchical control.
Nowadays, this approach is used for exploiting demand response resources.

Examples are the Intelligator algorithm implemented in [77], or the demand
response reserves offered by aggregators to a TSO for e.g. automatic or manual
frequency restoration reserves (aFRR or mFRR). The coordination of multiple,
centrally controlled microgrids can also be organized in a hierarchical way. In
that case, a central coordinator coordinates multiple microgrids, each controlled
by a local Microgrid Central Controller (MGCC). The MGCC of a single

microgrid tries to reach an optimal operation point using only its local resources.
If the internal resources are not sufficient, the MGCC shall ask the microgrid
central coordinator for external resources from other microgrids [78].
2.3.3 Two-Level Distributed Control Architecture
In both architectures described above, the DERs are controlled by a third

party. However, since the owners of the DERs impose the operational boundary
conditions, one can argue that it might be better to keep the control of the DERs
locally. Besides, not all owners want to exchange all their information with a
third party for privacy reasons. However, to reach a (near) optimal operation of
the grid, these DERs should be coordinated. It is at this point that distributed
control architectures come into play. The idea behind distributed control is
to divide the centralized problem into a certain number of local controllers or
agents (low-level problems). Therefore, each agent does not have a global vision
of the problem, but by means of correct coordination (high-level problem) they
can reach a globally (near) optimal state.
Coordination is organised by a central agent that is able to communicate global
constraints, such as the power limit of a transformer, or exceeding voltage limits.
This can be done by the communication of Lagrange multipliers. Examples
of algorithms that are suited for this approach are the Dual Decomposition
(DD) methods or the Alternating Direction Method of Multipliers (ADMM)1
[79, 80]. Both are based on the dual ascent method, where price vectors are
sent iteratively from the central controller to the DERs. The DERs optimize
their control decisions towards such a price vector and return demand vectors
to the central agent. The central agent then analyses the demand vectors with
regard to operational grid constraints, and updates the prices when constraints
are being violated. The DERs optimize again according to this new price
vector. This iteration goes on until a steady state solution is found. Figure 2.3
represents such a two-level distributed control scheme, consisting of local DERs
that optimize and a central agent that controls the global constraints.
Distributed approaches have important advantages that justify their use. As
the global optimization problem is divided into several sub-problems, the
computational requirements are lower. Besides, the information exchange
between local and central agents is limited, which relaxes the requirements of
the communication system. This approach results in a scalable method. As the
1 In Chapter 3, the thesis eliminates the master problem of the DD and ADMM methods
by letting each agent responsible for computing the Lagrangian multipliers associated with
the local PCC voltage.
Coordinator
: Optimization
Prosumer Prosumer
Prosumer Prosumer
Prosumer Prosumer
Figure 2.3: Two-level distributed control.
local DERs perform an optimization by themselves. However, a central agent

still exists, inherently resulting in a single point of failure.
2.3.4 P2P-Based Control Architecture
To eliminate the problems that a more centralized control method possesses,

having a single point of failure, the idea of P2P-based microgrids has been
developed. This type of architecture, inspired by P2P computer networking [81],
is characterized by the complete absence of a central controller. All local DERs
or agents, are equally important and can communicate to other agents [82] in a
P2P fashion, as shown in Figure 2.4. The absence of a central controller leads to
the term of autonomous control. P2P communication is used for dissemination
of the grid state to all required agents in the microgrid. The grid-supporting
agents can then act according to the received information, in cooperation with
each other. In this way they should be able to reach a (near) optimal operation
of the considered microgrid. Examples of algorithms that could be used for such
P2P-based control are gossip-based [83] and consensus-based control algorithms

[84]. This architecture will be elaborated further in Section 2.4.
: Optimization
Prosumer
Prosumer
Prosumer
Prosumer
Prosumer
Figure 2.4: P2P-based control.
In this architecture, there is a clear absence of a single point of failure. In the

case a single agent fails, the other agents can still operate the grid in a stable
way. Also when a single communication channel fails, the required information
can still reach all necessary participants, via other agents. These properties
makes this architecture a robust way of controlling a microgrid.
2.3.5 Local Control Architecture
Finally, there also exist control architectures without any form of communication.
This chapter classifies them as local control architectures. In this case, optimal
operation of the microgrid is rather difficult, as it is impossible to know the
complete state of the grid and all operational boundary conditions of the DERs.
Nevertheless, the absence of any communication in this architecture will ensure
that the grid is still controlled when all communication channels fail. As primary
frequency control should be able to operate even when communication fails,
this is often implemented as a local control architecture [85]. Additionally, local
PROPOSED P2P-BASED CONTROL PARADIGM 23
control architecture has been used by many publications to implement local

voltage control techniques [18, 19, 86].
2.4 Proposed P2P-Based Control Paradigm
A P2P-based control architecture, as introduced in Subsection 2.3.4, seems to

be a good method for controlling DERs. It is a robust method and is able to
work without the need for a central controller [87].
It is impossible to impose this architecture on the whole distribution grid, as
it incorporates thousands of DERs that are geographically very dispersed. To
deal with this, breaking the complete grid down into a smaller microgrids,
containing only a limited amount of DERs, can be a solution. These microgrids
operate then according to the P2P-based control architecture. Points of common
coupling are used to connect the different microgrids.
The proposed scheme is shown in Figure 2.5. The distribution network is
divided into several microgrids, hierarchically organized on different voltage
levels. A microgrid can consist of a couple of low voltage feeders, physically
connected to the same transformer, or a part of the medium voltage network
on the same voltage level, for example. Each microgrid consists of several
autonomous agents. Such an agent could be a renewable generation unit, a
group of intelligent controllable loads, a digital substation, or any other form of
DERs. On the connection points of two microgrids there is a coupling agent
which serves as gateway of one microgrid to the other microgrid, the point of
common coupling. As the microgrids represented in Figure 2.5 are separated by
transformers or substations, these would be good candidates for such a coupling
agent. Such a coupling agent represents the characteristics of the whole lower
level microgrid (e.g. a low voltage feeder) on the higher level microgrid (e.g. a
medium voltage distribution grid).
The agents are each able to communicate with some other neighbouring agents
in a P2P way, creating possibilities to disseminate data about the state of the
grid without the need for one central point of information. Epidemic algorithms
(e.g. gossiping protocols) seem to be particularly appropriate for this goal [88].
These algorithms are designed to disseminate and aggregate data in distributed
networks in a quick and robust way, making them good candidates to enable
the P2P-based microgrid control.
In this paradigm, each agent should have some local intelligence that determines
the action of the agent based on the perceived state of the grid. The agents
decide their own actions, that will result in a new state of the grid. The
High voltage
Load
G Distributed
generator
Medium voltage Coupling agent
P2P communication
G Microgrid
G
G G
G
Low voltage
G
G G
G
G G
G
Figure 2.5: P2P-based control paradigm.
agents obtain new information about the new state of the grid through the
P2P communication, and decide on a new action based on this new state. The
successive grid states should converge to a desired steady state of the grid.
Appropriate control algorithms are needed to achieve this.
Epidemic Algorithms
Epidemic or gossiping algorithms are used for scalable and efficient data
dissemination in distributed P2P networks, without a central controller [89].
They mimic the spread of a contagious disease. Each agent in the distributed
system sends new information it has received to other agents rather than to
a server or cluster of servers in charge of forwarding it. In turn, each of these
agents forwards the information to other selected agents, and so on. Gossiping is
often used as a synonym for epidemic algorithms [90]. Gossiping protocols have
been used in many publications to enable fully distributed control systems. In
PROPOSED P2P-BASED CONTROL PARADIGM 25
[91], a gossip-like P2P-based distributed optimization algorithm is presented for

optimal reactive power flow control in a microgrid. A P2P gossiping algorithm
is employed in [92] to enable primary and distributed secondary and tertiary
control.
P2P Energy Trading
The proposed P2P-based architecture can be used to enable P2P energy trading.
P2P energy trading concept provides local and regional energy producers with
options to trade energy fairly within the neighbourhood, within the community
and within the vicinity of the distribution system. This will fundamentally
change the current peer-to-grid paradigm where any surplus of local production
can only be sold to transmission grids, and transform consumers position from
energy/price takers to energy/price makers.
P2P energy trading has been discussed widely in literature. A comprehensive
review of the state-of-the-art in research on P2P energy trading techniques is
presented in [93]. A four-layer system architecture of P2P energy trading is
proposed in [94] to identify and categorize the key elements and technologies
involved in P2P energy trading. In [95], a motivational psychology framework is
introduced for P2P energy trading with an objective to improve the participation
of prosumers. The study proposed in [96] investigates the optimal sizing problem
of battery energy storage systems in a P2P energy sharing network considering
different ownership structures. An optimization model is proposed in [97] to
maximize the economic benefits for rooftop PV-battery system in a P2P energy
trading environment. In [98], a methodology is proposed based on sensitivity
analysis to assess the impact of P2P transactions on the grid and to guarantee
an exchange of energy that does not violate grid constraints.
Thesis Contribution to the Proposed Paradigm
This thesis contributes to the P2P-based control paradigm by developing novel

P2P-based grid voltage and frequency support functions for smart inverters.
The proposed functions enable PV and PV-battery systems to carry out a
distributed voltage control in each microgrid, and participate in supporting the
grid frequency.
2.5 Conclusion
This chapter discusses the major issues with regard to the integration of DERs
in the electrical grid and the concept of microgrids as a possible solution
to overcome the complexity of controlling distribution networks. As there
are different methods for organizing the control of these microgrids found in
literature, this chapter presents a classification of these methods, from highly
centralized to fully decentralized architectures. The drawbacks of the centralized
control and the advantages of the distributed control architectures have been
discussed as a motivation to propose a new control paradigm for the distribution
grid, P2P-based control paradigm. In this paradigm, the distribution grid is
decomposed into coupled microgrids. P2P-based control is deployed in each
microgrid to coordinate the local DERs in a fully distributed manner, and
enable the microgrids to coordinate themselves, through coupling agents, in a
distributed manner as well.
Chapter 3
Real-Time
Optimization-Based
Distributed Voltage Control
Smart photovoltaic (PV) inverters can contribute to active grid control, by

expanding their features with added grid support functions. In this chapter, we
develop novel grid voltage support algorithms for smart PV inverters, based
on distributed optimization and peer-to-peer (P2P) communication. The dual
decomposition method (DD), and the Jacobi-Proximal Alternating Direction
Method of Multipliers (JP-ADMM) are applied in this chapter to locally optimize
reactive power compensation and active power curtailment of each inverter
participating in voltage control. We propose the use of a push-sum gossip
protocol to enable P2P data interchange between inverters.
A real low voltage residential semiurban feeder from the region of Flanders
(Belgium) is used as a case study to analyse the performance of the two
algorithms. Results show that for a voltage control system with many agents
(e.g. more than 10), the DD-based distributed voltage control fails in regulating
voltage profiles of the simulated network within a reasonable time. Compared
to the DD method, JP-ADMM achieves much faster convergence, and provides
satisfactory regulation of voltage profiles. A JP-ADMM-based voltage control
system with 50 agents, for example, needs around 1.25 minutes to converge to
a solution that puts voltage profiles of the simulated network within accepted
limits, whereas the DD-based voltage control system needs around 59.4 minutes.
The contents of this chapter are based upon the two journal papers:
27
28 REAL-TIME OPTIMIZATION-BASED DISTRIBUTED VOLTAGE CONTROL
• H. Almasalma, S. Claeys, and G. Deconinck, “Peer-to-peer-based

integrated grid voltage support function for smart photovoltaic inverters,”
Applied Energy, vol. 239, pp. 1037–1048, 2019.
• H. Almasalma, J. Engels, and G. Deconinck, “Dual-decomposition-based
peer-to-peer voltage control for distribution networks,” CIRED-IET Open
Access Proceedings Journal, vol. 2017, no. 1, pp. 1718–1721, 2017.
3.1 Introduction
In order to overcome the drawbacks of centralized and decentralized (local)

voltage control systems, distributed voltage control has been proposed in
literature. The distributed voltage control system is characterized by the
complete absence of a central coordinator. A comprehensive survey of distributed
control algorithms for smart grids can be found in [23, 26, 82, 99, 100]. In
[101], a distributed reactive power feedback control for voltage regulation is
proposed. In the proposed strategy, distributed energy resources (DERs) are
smart agents that can measure their voltages, share these measurements with
the other agents, and compute the amount of reactive power needed to regulate
voltages within limits, according to a feedback control policy that descends from
duality-based methods. In [102], a consensus-based voltage control algorithm is
implemented to enable smart loads to manage grid voltages in a distributed way.
In [103], a distributed coordinated voltage control algorithm is proposed. The
algorithm uses a simplification method to deal with the lack of impedance and
power information in distribution networks. A control methodology is proposed
in [104] to coordinate transformer tap-changer and DERs in a distributed
manner, to regulate distribution system voltages. The Lyapunov technique is
adopted in [105] to design adaptive distributed voltage control for inverter-based
microgrids. In [106], distributed feedback control laws are designed for active
and reactive powers to keep voltages within specified limits.
The above mentioned methods apply different optimization and heuristic
techniques to implement a distributed voltage control system. Distributed
voltage control, provided by PV inverters, can be formulated as a distributed
optimization problem [24], to minimize the cost of reactive power compensation
and active power curtailment. Distributed optimization-based control algorithms
have the potential to reach a (near) globally optimal solution under certain
assumptions, thus achieving nearly the same control quality of centralized
schemes. Distributed optimization algorithms iterate over a sequence of steps
that are needed to compute a desirable solution. Hence, one of the main
challenges in distributed optimization-based voltage control algorithms is to
reduce the number of iterations (as much as possible) to solve a voltage problem
INTRODUCTION 29
within a reasonable time, making voltage profiles comply with the European
standard EN 50160 [3].
Distributed optimization-based voltage control algorithms appearing in literature
are mostly based on the DD method and the Alternating Direction Method of
Multipliers (ADMM) [80]. DD and ADMM decompose a coupled optimization
problem into sub-problems, suitable for distributed control. DD and ADMM
apply the theory of Lagrangian multipliers and duality. ADMM also applies
penalty function methods. DD and ADMM have been used in many publications
to decompose a centralized optimization-based voltage control system into sub-
systems that interact with each other in a distributed manner [27, 28, 107, 108].
Most of these algorithms require a large number of iterations to converge.
Hence, these proposed algorithms fail in regulating voltage profiles within a
reasonable time. To tackle this issue, our work develops a novel fast distributed
optimization-based voltage control algorithm. The algorithm uses a change in
reactive power and active power curtailment of some participating PV inverters
installed in the grid to regulate voltage profiles within allowed limits.
The work of this chapter is four-fold. First, we modify the state-of-the-art
DD algorithm to implement a fully distributed DD method. We then apply
the fully distributed DD method to design a fully distributed voltage control
system. Second, we speed up the convergence of the distributed voltage control
system significantly by applying the JP-ADMM method. Third, three strategies
are proposed in this chapter to improve the performance of our algorithms:
feedback strategy; anti-windup strategy; and active/standby modes. Fourth, we
integrate a push-sum gossip protocol with the DD-based and JP-ADMM-based
voltage control systems, to enable P2P data interchange between the inverters.
A real low voltage network is used to study the convergence speed of the DD-
based, Jacobi ADMM-based, and JP-ADMM-based distributed voltage control
algorithms, to see if they are fast enough to solve a voltage problem in real-
time. Based on the convergence speed study, we propose the use of JP-ADMM
(with the push-sum gossip protocol) for distributing a centralized optimization-
based voltage control algorithm. As shown in Figure 3.1, the gossip-based
JP-ADMM (G-JP-ADMM) algorithm is hosted by the grid voltage support
function (GVSF), and integrated with inner control loops of the PV inverter.
In G-JP-ADMM, the GVSFs are updated in parallel; proximal penalization
functions are used to preserve the convergence; and acceleration factors are used
to speed up the convergence. Two factors for accelerating the convergence of
the JP-ADMM are proposed in this chapter: one related to the reactive power
compensation; and one related to the active power curtailment. To the best of
our knowledge, this work is the first work proposing the use of G-JP-ADMM
for designing a P2P-based GVSF.
Smart inverter
P2P-based
communication
Voltage support
function Scope of the
G-JP-ADMM chapter
DC AC
Control Control
DC AC PCC
PV Filter
Bus DC
Grid
Figure 3.1: General block diagram of smart PV inverter with P2P-based grid
voltage support function (PCC: point of common coupling).
The rest of this chapter is organized as follows. The centralized optimization-

based voltage control problem treated in this chapter is formulated in Section 3.2.
To avoid non-convexity, a linearised branch flow model is presented in Section
3.3. Section 3.4 presents the simulated network and PV/loads data used to study
the performance of the proposed algorithms. The design and implementation
of the DD-based distributed voltage control system is presented in Section 3.5.
Section 3.5 also analyses the convergence speed of the DD-based distributed
voltage control system. Section 3.6 presents the design and implementation
of the JP-ADMM-based distributed voltage control system. Section 3.6 also
demonstrates the fast convergence of the JP-ADMM-based distributed voltage
control system, and its ability to provide satisfactory regulation of voltage
profiles. Finally, the chapter is concluded in Section 3.8.
PROBLEM FORMULATION 31
3.2 Problem Formulation
For each time step k, a centralized optimization-based voltage control problem

can be formulated as follows:
2 2
(k) (k)
X
minimize cp,invi ∆pinvi + cq,invi ∆qinvi (3.1a)
(k) (k)
∆pinv ,∆qinv ,∀i∈I i∈I
i i
Subject to: ∀i ∈ I
(k)
v ≤ vi ≤v (3.1b)

(k) (k) (k)
f vi , ∆pinvd , ∆qinvd = 0, ∀d ∈ I (3.1c)
(k) (k)
(−cri ) (ppv
i ) ≤ ∆pinvi ≤ 0 (3.1d)
(k) (k) (k)
− ∆q invi ≤ ∆qinvi ≤ ∆q invi (3.1e)
r 2
(k)

2 (k)
∆q invi = (sinvi ) − (ppv
i ) (3.1f)
where I is the set of inverters participating in the voltage control. i and d are used
to indicate inverters within the set I. The objective function (3.1a) minimizes
the total cost of all changes in inverter active power pinvi and inverter reactive
power qinvi needed to maintain voltages within the maximum voltage limit v
and the minimum voltage limit v. The total cost is the sum of the quadratic
2
(k)
cost functions of the individual inverters: cp,invi ∆pinvi represents the cost
(k)
function of a change in active power of inverter i with an amount ∆pinvi , while
2
(k)
cq,invi ∆qinvi represents the cost function of a change in reactive power of
(k) (k) (k)
inverter i with an amount ∆qinvi . ∆pinvi and ∆qinvi have to be within specified
limits, defined by the constraints (3.1d)-(3.1f), to respect the maximum apparent
power magnitude sinvi of inverter i, and the maximum allowed curtailment
cri . ppv
i is the active power generation of the PV module connected to inverter
(k)
i. vi in (3.1b)-(3.1c) is the expected magnitude of the PCC voltage after
(k) (k)
applying ∆pinvi and ∆qinvi . The constraint (3.1c) represents the non-linear
(k)
relation between vi and the active/reactive power changes of the inverters
participating in voltage control. The non-linear relation can be obtained from
the AC power flow equations [109].
cp,invi and cq,invi are constant factors used to prioritize the use of reactive power,
while active power curtailment is performed only as a last resort. In this chapter,
cp,invi is set to be greater than cq,invi in order to use most reactive power before
starting to use active power curtailment [110].
A centralized controller solves the voltage control problem (3.1a)-(3.1f) in three
steps (see Figure 3.2). First, at time step k, all input variables are gathered;
these include PCC voltages and PV active power generation measured by the
GVSFs. Secondly, the central optimizer solves the optimization problem (3.1a)-
(3.1f) to optimally coordinate the inverters by minimizing the total change
in reactive power compensation and active power curtailment. Finally, the
controller sends the new set-points to the GVSFs. The major drawback of this
approach is that a failure in the central coordinator completely disables the
voltage control. Therefore, a methodology is proposed in this chapter to convert
the centralized optimization-based voltage control problem into a distributed
one.
Household with PV installation
2-way communication
1 2 i
pv (k)
sinv1, ( p1 ) , v1(k) pinv(k) , qinv(k) pinv(k) , qinv(k) pinv(k) , qinv(k)
1 1 2 2 i i
pv (k)
sinv2, ( p2 ) , v2 (k)
(k)
pv
sinvi , ( pi ) , vi(k)
Figure 3.2: Centralized optimization-based voltage control system.
It is worth to point out that network losses and voltage imbalances are not
considered in the optimization problem. Two constraints can be added to
the optimization problem to limit network losses incurred by reactive power
provision, and minimize voltage imbalances. Minimizing network losses and
voltage imbalances is out of scope of this chapter.
BRANCH FLOW MODEL 33
3.3 Branch Flow Model
Since the DD method and ADMM are iterative methods, the time step k of the
distributed optimization-based voltage control system is considered as a control
iteration number.
The use of the AC power flow model in (3.1c) results in a non-convex optimization
problem that is difficult to handle in distributed optimization. To avoid the non-
convexity, many works use linearised power flow equations [111, 112]. Voltage
sensitivities can be used to formulate a linear relation between controlled voltages
(k) (k) (k)
vi , and control variables ∆pinvd and ∆qinvd (∀d ∈ I). In Figure 3.3, a simple
distribution system is shown. The system consists of an inverter d connected to
a bus i via a power line having a resistance Rdi and a reactance Xdi . Based on
the distributed power flow equations of Baran and Wu [113], it can be shown
(k)
that the voltage magnitude vi can be described by the following equation.
(k)
vi = (3.2a)
v
u
u 2 + X 2 ) (p(k) )2 + (q (k) )2
(Rdi
u (k) di inv d inv d
t(v )2 − 2 Rdi p(k) + Xdi q (k) +
d invd invd (k) 2
(vd )
(k) (k−1) (k)

pinvd = pinvd + ∆pinvd (3.2b)
(k) (k−1) (k)
qinvd = qinvd + ∆qinvd (3.2c)
(k−1) (k)
Here, pinvd and pinvd denote active powers of inverter d at the control iterations
(k−1) (k)
k − 1 and k, respectively. qinvd and qinvd denote reactive powers of inverter d
at the control iterations k − 1 and k, respectively.
(k) (k)
For a small change ∆pinvd and ∆qinvd , a first order approximation can be used
to simplify (3.2a) as follows:
(k) (k−1) ∂vi (k) ∂vi (k)

vi ≈ vi + ∆pinvd + ∆q (3.3)
∂pd ∂qd invd
(k−1)
where vi is the voltage of bus i at the control iteration k − 1. As the partial
derivatives are not constant and dependent on the system state, these are often
approximated by [112]:
vd(k) vi (k)
Rdi Xdi
AC
DC
d pinv(k) +
(k)
j qinv i
d d
Figure 3.3: A simple 2-bus distribution system.
∂vi Rdi ∂vi Xdi

≈ , and ≈ (3.4)
∂pd vo ∂qd vo
where vo is the nominal voltage. Equation (3.4) is a good approximation when
the angle between the voltages at different nodes is small [114], which is the
case in distribution networks. For a set of inverters I, equation (3.3) can be
written as:
(k) (k−1) (k) (k)
X p q
vi ≈ vi + vi,d ∆pinvd + vi,d ∆qinvd (3.5)
d∈I
p q
where the coefficients vi,d and are the approximated partial derivatives.
vi,d
p
These coefficients are considered as voltage sensitivity coefficients. vi,d and
q
vi,d represent the influence of active power and reactive power (respectively) of
inverter d on the controlled voltage vi . Equation (3.5) represents a constant
linear voltage model, since the voltage sensitivities are considered as constant
coefficients. The voltage sensitivity coefficients can be calculated from the
topology of the grid and the knowledge of the impedances of the lines. In
[114, 115], a direct load flow approach is presented. The approach is based on
the calculation of two matrices named bus injection to branch current (BIBC),
and branch current to bus voltage (BCBV). These matrices allow for easy
calculation of the voltage sensitivity coefficients.
It is worth to mention that there are other methods that can be used to estimate
the voltage sensitivities in distribution networks. If historical data of voltages,
active powers and reactive powers are available, then linear regression can
be used to estimate the voltage sensitivities, as detailed in [111]. The voltage
sensitivities can be also estimated by applying small perturbations (small change
SIMULATED NETWORK 35
in active and reactive powers) into a given network, the voltage sensitivities are
then determined by measuring the change in voltage.
3.4 Simulated Network
Before we proceed presenting the DD-based and ADMM-based voltage control

algorithms, we present in this section the simulated network that will be used
to study the performance of the proposed voltage control systems.
The simulated network used in this chapter represents a typical Flemish
(Belgium) semi-urban distribution network1 . This three-phase network operates
with a nominal voltage of 230/400 V. It connects 62 households, as shown in
Figure 3.4. All main feeder cables are of type EAXVB 1 kV 4 × 150 mm2
(impedance: 0.206+j0.0778 Ω/km) except for the cable between node A and
node B, which is of type EAXVB 1 kV 4 × 95 mm2 (impedance: 0.320+j0.0778
Ω/km). The cables connecting each household with the feeder are of type
EXVB-Cu 1 kV 4 × 16 mm2 (impedance: 1.15+j0.0828 Ω/km).
32 31 30 26 28 29 0 300m
Household with PV installation
25
F E
27
13 15 17 19 21 23 33 34 36 38 40 41 43
G
D
10/0.4 kV
250 kVA 14 16 18 20 22 24 35 37 39 42 44
A B
47 49 51 53 55 57 59 61
3 5 7 9 11
1 H
C 45
46 48 50 52 54 56 58 60 62
2 4 6 8 10 12
Figure 3.4: Schematic diagram of the simulated network. All lengths are drawn
to scale.
3.4.1 PV and Household Profiles
Active power consumption profiles of the households are generated by load

profiles presented in [117]. The reactive power consumption is calculated by
assuming a constant power factor of 0.85. Every household is equipped with a PV
1 The network of Figure 3.4 was used as semi-urban reference network in the LINEAR
project [116].
installation. PV active power profiles are based on the measured PV generation

data from the rooftop PV installation at the department of electrical engineering
(ESAT) of the KU Leuven. The PV inverters have different capacities ranging
from 5 to 13 kVA.
To comply with the European standard EN 50160 on power quality [3], the
voltage limits v and v are enforced to be ±10% of the nominal phase voltage
230 V, resulting in v = 253 V and v = 207 V. The curtailment factor cri is
set to 1 (∀i ∈ I), meaning that each inverter would curtail its whole active
power, if needed. To prioritize the use of reactive power, while active power
curtailment is performed only as a last resort, the reactive power penalization
factor cq,invi is set to 1 (∀i ∈ I) and the curtailment penalization factor cp,invi
is set to 200 (∀i ∈ I). The tuning is based on trial and error.
3.4.2 Simulation Environment
The proposed algorithms are implemented in MATLAB R2018a and applied to

the network of the case study to evaluate its performance. The MATPOWER
package 6.0 for power flow analysis is used to calculate the measured voltages
[118]. The CVX for solving convex optimization problems is used to solve the
optimization problem of each GVSF [119]. All of the numerical experiments are
run on a workstation with an Intel Core i7-7700HQ CPUs (2.80GHz) and 32
GB of RAM. In normal operation, each control iteration k is updated every 5
seconds. The push-sum gossip algorithm is executed with 100 ms communication
latency.
3.5 Dual-Decomposition-Based Distributed Volt-

age Control
The main goal of this chapter is to design a voltage control system that does not
rely on a centralized controller. To do so, the centralized optimization problem
(3.1a)-(3.1f) has to be decomposed into sub-optimization problems that can be
solved locally by the GVSFs.
The objective function (3.1a) is basically a sum of separate cost functions, one
for each participating inverter. The constraints (3.1d)-(3.1f) are local, meaning
(k) (k)
that they only influence the local decision variables ∆pinvi and ∆qinvi , and
therefore these constraints can be distributed easily, one for each inverter. On
the other hand, the constraints (3.1b) and (3.1c) are complicating (coupling)
constraints and cannot be distributed, as the voltage magnitude of each PCC
DUAL-DECOMPOSITION-BASED DISTRIBUTED VOLTAGE CONTROL 37
voltage i depends on the change in reactive/active powers of the local inverter i

and the other inverters participating in the voltage control system (See Equation
(3.5) ).
3.5.1 Lagrangian Relaxation
The linearization of vi makes it possible for the whole optimization problem

(3.1a)-(3.1f) to be solved in a distributed way based on Lagrangian relaxation [79,
120, 121]. The basic idea in Lagrangian relaxation is to relax the original problem
(3.1a)-(3.1f) by transferring the complicating constraints to the objective function
in the form of a weighted sum. The method penalizes violations of complicating
constraints using Lagrangian multipliers, which imposes a cost on violations.
These added costs are used instead of the complicating constraints in the
optimization problem. The Lagrangian relaxation of (3.1a)-(3.1f) is defined as:
For each control iteration k :
minimize L(k) (3.6a)
Subject to: ∀i ∈ I
(k) (k)
λinvi ≥ 0 , λinvi ≥ 0 (3.6b)
Local constraints (3.1d)-(3.1f) (3.6c)
Here, L(k) is the Lagrangian function, which is given by:

2 2
(k) (k)
X
L(k) = cp,invi ∆pinvi + cq,invi ∆qinvi +
i∈I
(k)

(k) (k) (k)
λinvi vi − v + λinvi −vi + v (3.7)
(k)
where vi is given by (3.5). λinvi is the Lagrangian multiplier associated with
the complicating constraint vi ≤ v, whereas λinvi is the Lagrangian multiplier
associated with the complicating constraint −vi ≤ −v. Because of the Karush-
Kuhn–Tucker conditions (KKT), the Lagrangian multipliers cannot be smaller
than zero [122]. One can notice that the optimization problem (3.6a)-(3.6c) is
penalized when the complicating constraints are violated, and rewarded when
these constraints are satisfied strictly.
Keeping the Lagrangian multipliers fixed makes the Lagrangian function L(k)
a sum of separate objective functions, one for each participating inverter.
The relaxed optimization problem (3.6a)-(3.6c) can be decomposed into sub-

optimization problems that can be solved locally, considering P2P interchange
of Lagrangian multipliers between the GVSFs. We postpone the discussion of
decomposing the Lagrangian function till Subsection 3.5.4, to focus on improving
the solution of the relaxed optimization problem.
The solution of the relaxed optimization problem (3.6a)-(3.6c) is an approximate
solution to the original problem (3.1a)-(3.1f). We show in the next subsection
how minimizing the maximum of the Lagrangian function, subject to local
constraints, leads to a solution equals to the solution of the original problem
(3.1a)-(3.1f).
3.5.2 Minimax of the Lagrangian Function
In this subsection, we modify the decomposable optimization problem (3.6a)-

(3.6c) by minimizing the maximum of the Lagrangian function (subject to local
constraints), to obtain a solution that is equal to the solution of the original
problem (3.1a)-(3.1f):
minimize maximize L(k) (3.8a)

(k) (k) (k)
∆pinv ,∆qinv λinv ≥0
i i i
∀i∈I (k)
λinv ≥0
i
∀i∈I
Subject to local constraints (3.1d)-(3.1f) , ∀i ∈ I (3.8b)
The optimization problem (3.8a)-(3.8b) can be rewritten as follows:
minimize maximize L(k) (3.9)

(k) (k)
∆pinv ∈Υ λinv ≥0
i i
(k) (k)
∆qinv ∈Υ λinv ≥0
i i
∀i∈I ∀i∈I
where Υ is a feasible region defined by the local constraints set:

(k) (k) (k)
Υ = ∆pinvi ≤ 0, −∆pinvi ≤ cri (ppv i ) ,
(k) (k) (k) (k)

∆qinvi ≤ ∆q invi , −∆qinvi ≤ ∆q invi , ∀i ∈ I (3.10)
Below we prove that the solution of (3.9) equals the solution of the original
optimization problem (3.1a)-(3.1f).
Proof : We define the feasible region Ψ by the complicating constraints set
(k) (k) (k) (k)
{vi ≤ v, −vi ≤ −v, ∀i ∈ I}. If ∆pinvi and ∆qinvi (∀i ∈ I) are within the
feasible region Ψ, then all the complicating constraints are non-positive. In this
case, the Lagrangian function L(k) can be written as:
2 2
(k) (k)
X
i∈I
(k) (k)
λinvi × negative value + λinvi × negative value (3.11)
(k) (k)
If ∆pinvi and ∆qinvi (∀i ∈ I) are not within the feasible region Ψ, then all the
complicating constraints are positive. In this case, the Lagrangian function L(k)
can be written as:
2 2
(k) (k)
X
i∈I
(k) (k)
λinvi × positive value + λinvi × positive value (3.12)
Based on the above cases, maximum of the Lagrangian function can be computed
as follows:
(
(k) (k)
f (k) ∆pinvi , ∆qinvi , ∀i ∈ I ∈ Ψ,
maximize L = (k)
(k) (k) (3.13a)
(k)
λinv ≥0 ∞ /Ψ
∆pinvi , ∆qinvi , ∀i ∈ I ∈
i
(k)
λinv ≥0
i
∀i∈I
2 2
(k) (k)
X
f (k) = cp,invi ∆pinvi + cq,invi ∆qinvi (3.13b)
i∈I
To maximize the Lagrangian function in (3.11) over the Lagrangian multipliers,

these multipliers should be zero to get rid of the negative values in (3.11)
that decrease the value of L(k) . When the decision variables are not within
the feasible region Ψ, maximizing the Lagrangian function in (3.12) over the
Lagrangian multipliers leads to an infinite value. The ∞ in (3.13a) can be
excluded by taking the minimum of the maximum of the Lagrangian function.
The minimum of (3.13a) subject to local constraints (3.1d)-(3.1f) results in:
minimize maximize L(k) = minimize f (k)
(k) (k) (k)
∆pinv ∈Υ λinv ≥0 ∆pinv ∈Υ
i i i
(k) (k) (k)
∆qinv ∈Υ λinv ≥0 ∆qinv ∈Υ
i i i
∀i∈I ∀i∈I ∀i∈I
(k) (k)
Subject to: ∆pinvi , ∆qinvi , ∀i ∈ I ∈ Ψ (3.14)
(3.14) can be rewritten as follows:
minimize maximize L(k) = minimize f (k)
(k) (k)
∆pinv ∈Υ λinv ≥0
i i
(k) (k)
∆qinv ∈Υ λinv ≥0
i i
∀i∈I ∀i∈I
Subject to:
(k) (k)
∆pinvi , ∆qinvi , ∀i ∈ I ∈ Ψ
(k) (k)
∆pinvi , ∆qinvi , ∀i ∈ I ∈ Υ (3.15)
One can notice that (3.15) is the same as the original problem (3.1a)-(3.1f).
Hence, the optimization problem (3.9) gives the same solution as the original
optimization problem (3.1a)-(3.1f).
3.5.3 Minimax Theorem and Strong Duality
The optimization problem (3.9) is a decomposable problem and gives the same
solution as the original optimization problem (3.1a)-(3.1f). However, minimizing
maximize L(k) is not an easy task, as maximize L(k) is ill-conditioned 2 . As
(k) (k) (k) (k)
λinv ,λinv ≥0 λinv ,λinv ≥0
i i i i
∀i∈I ∀i∈I
shown in (3.13a), if the decision variables are slightly outside the feasible region,
then this would lead to maximize L(k) → ∞.
(k) (k)
λinv ,λinv ≥0
i i
∀i∈I
In this subsection, we modify the ill-conditioned decomposable optimization

problem (3.9), to obtain a decomposable optimization problem that can be
solved. The minimax theorem [124] is used here to modify the optimization
problem (3.9) based on the following relation:
minimize maximize L(k) ≥ maximize minimize L(k) (3.16)

(k) (k) (k) (k)
∆pinv ∈Υ λinv ≥0 λinv ≥0 ∆pinv ∈Υ
i i i i
(k) (k) (k) (k)
∆qinv ∈Υ λinv ≥0 λinv ≥0 ∆qinv ∈Υ
i i i i
∀i∈I ∀i∈I ∀i∈I ∀i∈I
The solution of maximize minimize L(k) gives a lower bound on the problem
(k) (k)
λinv ≥0 ∆pinv ∈Υ
i i
(k) (k)
λinv ≥0 ∆qinv ∈Υ
i i
∀i∈I ∀i∈I
(3.9). The above relation is called the minimax theorem. Below we prove that
relation (3.16) holds.
Proof : For fixed Lagrangian multipliers, one can always claim that the value
(k) (k)
of the Lagrangian function for any given ∆pinvi and ∆qinvi (∀i ∈ I) ∈ Υ is
2 A mathematical problem is ill-conditioned if a small change in input leads to a large change
in the output. This can lead to computational problems. For example, if an optimization
problem is ill-conditioned, the solution exists, but it is very difficult to find [123].
(k) (k)
greater than or equal to its minimum over ∆pinvi and ∆qinvi (∀i ∈ I) ∈ Υ:
(k) (k)
∀∆pinvi , ∆qinvi (∀i ∈ I) ∈ Υ L(k) (∆p, ∆q, λ) ≥ minimize L(k) (∆p, ∆q, λ)
(k)
∆pinv ∈Υ
i
(k)
∆qinv ∈Υ
i
∀i∈I
| {z }
g (k) (λ)
(3.17)
minimize L(k) (∆p, ∆q, λ) can be defined as the function g (k) (λ), that depends
(k)
∆pinv ∈Υ
i
(k)
∆qinv ∈Υ
i
∀i∈I
(k) (k) (k) (k)
only on λinvi , λinvi (∀i ∈ I). Now we fix ∆pinvi and ∆qinvi (∀i ∈ I), to
(k) (k)
maximize both sides of (3.17) over λinvi ≥ 0, λinvi ≥ 0 (∀i ∈ I):
maximize L(k) (∆p, ∆q, λ) ≥ maximize g (k) (λ) (3.18)

(k) (k)
λinv ≥0 λinv ≥0
i i
(k) (k)
λinv ≥0 λinv ≥0
i i
∀i∈I ∀i∈I
| {z } | {z }
Ĺ(k) (∆p,∆q) g (k) (λ∗ )
maximize L(k) (∆p, ∆q, λ) can be defined as the function Ĺ(k) (∆p, ∆q), that
(k)
λinv ≥0
i
(k)
λinv ≥0
i
∀i∈I
(k) (k)
depends only on ∆pinvi , ∆qinvi (∀i ∈ I). g (k) (λ∗ ) in (3.18) is a constant value.
(k) (k)
As Ĺ(k) (∆p, ∆q) ≥ g (k) (λ∗ ) for all possible values of ∆pinvi , ∆qinvi (∀i ∈ I)
(k) (k)
∈ Υ, its minimum over all possible values of ∆pinvi and ∆qinvi (∀i ∈ I) ∈ Υ is
greater than g (k) (λ∗ ):
minimize Ĺ(k) (∆p, ∆q) ≥ g (k) (λ∗ ) (3.19)

(k)
∆pinv ∈Υ
i
(k)
∆qinv ∈Υ
i
∀i∈I
Relation (3.19) can be rewritten as follows:

minimize maximize L(k) ≥ maximize minimize L(k) (3.20)

(k) (k) (k) (k)
i i i i
(k) (k) (k) (k)
i i i i
Relation (3.20) is the same as (3.16)

(k) (k)
Minimizing the Lagrangian function over ∆pinvi , ∆qinvi (∀i ∈ I) is much easier
than maximizing the Lagrangian function over the Lagrangian multipliers. The
problem in the right side of relation (3.16) is called the dual problem [125]:
maximize minimize L(k)

(k) (k) (k)
λinv ≥0 ∆pinv ,∆qinv
i i i
(k)
λinv ≥0 ∀i∈I
i
∀i∈I
Subject to local constraints (3.1d)-(3.1f) , ∀i ∈ I (3.21)
The gap between the dual problem (3.21) and the primal problem (3.1a)-(3.1f)
(or the equivalent relaxed problem (3.9)) is called the duality gap [126]. If the
equality
minimize maximize L(k) = maximize minimize L(k) (3.22)

(k) (k) (k) (k)
i i i i
(k) (k) (k) (k)
i i i i
holds, then we say strong duality holds. Strong duality holds if and only if the
duality gap is equal to zero [127]. This is as opposed to weak duality, which
holds if the duality gap is not zero:
minimize maximize L(k) > maximize minimize L(k) (3.23)

(k) (k) (k) (k)
i i i i
(k) (k) (k) (k)
i i i i
Fortunately, Boyd et al proved that if the objective function of the primal

problem is convex and all the constraints are linear, then strong duality holds
[128]. Since the objective function (3.1a) is convex and the constraints (3.1b)-
(3.1f) are linear, we can state that the dual problem (3.21) is equivalent to the
original primal problem (3.1a)-(3.1f). In the next subsection, we decompose the
dual optimization problem (3.21) into sub-optimization problems.
3.5.4 The Dual Decomposition Method
The basic idea of the DD method is to decompose the dual problem into
sub-problems, which are then coordinated by a high level master problem, by
means of the Lagrangian multipliers. This approach results in a distributed
control system that has a degree of centralization. The local controllers (GVSFs)
perform an optimization by themselves. However, a central agent still exists,
inherently resulting in a single point of failure.
In this subsection, we modify the state-of-the-art DD method by letting the
GVSFs compute the Lagrangian multipliers locally, to coordinate themselves
in a P2P manner. We apply the push-sum gossip protocol to enable the P2P
coordination [129].
State-of-the-Art Dual Decomposition:
One can show that the Lagrangian function L(k) can be decomposed into nc local
Lagrangian functions, where nc is the number of smart inverters participating
(k−1) (k−1)
in voltage control. For λinvi , λinvi (∀i ∈ I) computed at the control iteration
k − 1, and (vimeas )(k−1) (∀i ∈ I) measured at the control iteration k − 1, L(k)
can be decomposed as follow:
(k)
X
L(k) = Linvi (3.24a)
i∈I
2 2
(k) (k) (k)
Linvi = cp,invi ∆pinvi + cq,invi ∆qinvi +
(k−1)

(k−1) (k) (k)
X p q
λinvd − λinvd vd,i ∆pinvi + vd,i ∆qinvi +
d∈I
(k−1)

(k−1)
λinvi − λinvi 2(vimeas )(k−1) − v − v (3.24b)
p q
Here, vd,i and vd,i represent the influence of active power and reactive power
(respectively) of inverter i on the PCC voltage vd . As shown in Figure 3.5, the
DD method seperates the dual problem (3.21) into two levels. At the lower
level, the GVSFs solve the sub-problems in parallel:
∀i ∈ I :
(k)
minimize Linvi (3.25a)
(k) (k)
∆pinv ,∆qinv
i i
Subject to:
(k−1) (k)
(−cri ) (ppv
i ) ≤ ∆pinvi ≤ 0 (3.25b)
(k) (k) (k)
− ∆q invi ≤ ∆qinvi ≤ ∆q invi (3.25c)
r 2
(k)

2 (k−1)
i ) (3.25d)
(k−1)
where (ppv
i ) is the active power generation (of the PV installation i)
measured at the control iteration k − 1.
Master problem
(k-1) (k-1)
λinv 1
λinv n
c
Original Decomposition (k-1)

λinv pinv(k) pinv(k) (k-1)
λinv
1 n
c
problem 1 nc
qinv(k) qinv(k)
1 nc
Sub-problem Sub-problem
1 nc
Figure 3.5: Decomposition of the dual problem into sub-problems controlled by

a master problem (Dual Decomposition).
At the higher level, the central agent solves the master problem:
maximize g (k) (λ) (3.26)

(k)
λinv ≥0
i
(k)
λinv ≥0
i
∀i∈I
where g (k) (λ) is the dual Lagrangian function defined in (3.17).

In general, the dual ascent method [130] is used to solve the DD problem. The
dual ascent solves the DD problem iteratively, by moving in the direction of
the steepest ascent defined by the gradient applied to the dual Lagrangian
function g (k) (λ). The dual ascent algorithm starts with moderate values of the
Lagrangian multipliers, and successively obtains new Lagrangian multipliers
with improved Lagrangian function value, with the aim to find the optimum
(∗) (∗)
Lagrangian multipliers λinvi , λinvi (∀i ∈ I), and substitute them in Linvi (∀i ∈ I)
(∗) (∗)
to find the optimum control variables ∆pinvi , ∆qinvi (∀i ∈ I).
The gradient of g (k) (λ) consists of 2nc components, since g (k) (λ) has 2nc
Lagrangian multipliers. It can be shown that the gradient of g (k) (λ) is:
(k)
∇λinv g (k) (λ) = vi − v , (∀i ∈ I)
i

(k) (k)
X p q
= (vimeas )(k−1) + vi,d ∆pinvd + vi,d ∆qinvd − v (3.27a)
d∈I
(k)
∇λinv g (k) (λ) = −vi + v , (∀i ∈ I)
i

(k) (k)
X p q
= −(vimeas )(k−1) − vi,d ∆pinvd + vi,d ∆qinvd + v
d∈I
(3.27b)
When we use the dual ascent to solve the DD-based voltage control problem,
we get the following simple algorithm:
repeat
Local problem: solve the sub-problems in parallel:
(k) (k) (k)
Find ∆pinv1 and ∆qinv1 that minimize Linv1 .
..
.
(k) (k) (k)

Find ∆pinvnc and ∆qinvnc that minimize Linvnc .
Master problem:
update the Lagrangian
multipliers:

(k) (k−1)
λinv1 = max 0, λinv1 + α ∇λinv g
k (k)
.
1
(k) (k−1)
λinv1 = max 0, λinv1 + αk ∇λinv g (k) .
1
..
.
(k)
(k−1)

λinvnc = max 0, λinvnc + αk ∇λinv g (k) .
nc
(k) (k−1)
λinvnc = max 0, λinvnc + αk ∇λinvn g (k) .
c
Here, α is the step size of the dual ascent that may vary throughout the
k
iterations. The max operator in the master problem is used to guarantee that the
Lagrangian multipliers are greater than or equal to zero (KKT conditions). With
suitable choice of αk and certain assumptions, the DD method is guaranteed to
converge to an optimal solution [131]. However, as will be demonstrated later
in this chapter, the convergence of the DD method often tends to be slow in
practice.
One can notice that the update of the Lagrangian multipliers, at the control
iteration k, depends on the local PCC voltages (vimeas )(k−1) (∀i ∈ I), measured
at the control iteration k − 1. Meaning that the GVSFs should apply the
control variables at each control iteration, and not at the end of the convergence.
Applying the control variables and updating voltages measurement at each
control iteration is called the iterative feedback strategy (IFS).
3.5.5 Feedback Strategy
There are two approaches regarding applying the control variables. First,
applying the control variables only after the control algorithm converges. Second,
applying the control variables at each iteration. The latter is called IFS (see
Figure 3.6). IFS has been used in [132], and is applied in this chapter to
improve the performance of the distributed voltage control algorithm. In IFS, at
each control iteration k, the GVSF i dynamically adjusts the inverter outputs,
measures the PCC voltage (vimeas )(k) . IFS helps in increasing the voltage
quality. Without IFS, the voltage during the convergence would stay beyond
the accepted limits, without getting closer to these limits until the end of
convergence. Moreover, IFS helps in correcting the error of linearizing the
branch flow model (3.5). Additionally, if the PV generation and/or consumption
changes during the computation, the GVSFs would implement outdated control
variables.
pinv(k) , qinv(k)
Sub-problems i i Master problem
i=1,2, ,nc
Dual decomposition
(k-1) (k-1)
λinvi
, λinv i
i=1,2, ,nc
Feedback
(k-1)
(vi meas ) Grid
i=1,2, ,nc
Figure 3.6: DD-based voltage control with iterative feedback strategy.
3.5.6 Fully Distributed Dual Decomposition
We eliminate the master problem of the DD method by letting each GVSF i

responsible for computing the Lagrangian multipliers (λinvi and λinvi ) associated
with the local PCC voltage vi . We apply the push-sum gossip protocol to enable
P2P interchange of the Lagrangian multipliers between the GVSFs. Additionally,
we apply the IFS in the middle of the control iteration (as shown in Figure 3.7)
rather than at the end of the control iteration, to update voltage measurements
before computing the Lagrangian multipliers. This makes the update of the
Lagrangian multipliers depend on the locally measured voltages, and eliminate
the need of communicating the control variables:
∀i ∈ I
(k)
(k−1)

λinvi = max 0, λinvi + αk (vimeas )(k) − v (3.28a)

(k) (k−1)
λinvi = max 0, λinvi + αk −(vimeas )(k) + v (3.28b)
Here, (vimeas )(k) is the magnitude of the PCC voltage i measured in the middle
of the control iteration k.
(k-1) (k-1)
λinvi
, λinv i
i=1,2, ,nc
Dual decomposition Update of the

Sub-problems Lagrangian
multipliers
pinv(k) , qinv(k)
i i
i=1,2, ,nc
Feedback
(k)
Grid (vi meas )
i=1,2, ,nc
Figure 3.7: Illustration of the DD-based voltage control with IFS applied in the
middle of the control iteration.
From Figure 3.7, one can notice that there is no feedback from the grid to the
(k−1) (k−1)
sub-problems. This is because the part (λinvi − λinvi )(2(vimeas )(k−1) − v − v)
(k) (k)
in (3.24b) is constant, and can be removed when we minimize Linvi over ∆pinvi
(k)
and ∆qinvi .
3.5.7 Push-Sum Gossip Protocol
A communication layer is needed to disseminate the Lagrangian multipliers

among the GVSFs. Each GVSF needs the Lagrangian multipliers to substitute
them in (3.25a). This work proposes the use of a P2P gossip-based push-sum
algorithm presented in [90]. It is a fully distributed asynchronous communication
protocol. It is scalable, fault-tolerant, resistant to packet losses and delays, and
simple to implement. The algorithm is modified in this chapter so that it can
be used for dynamic disseminations. The modified algorithm is presented in
Algorithm 1. It is an iterative algorithm with gossiping iteration kG . Each

(k )
GVSF tries to fill the matrix Si G before reaching the deadline. The deadline
is used so that the GVSF does not need to wait forever when one of the GVSFs
(or more) fails to communicate with the others. As shown in step 2 of the
(k )
algorithm, the GVSF starts filling Si G by initializing the matrix with its
(k)
own Lagrangian multipliers and power limit indicator µinvi . The power limit
(k)
indicator µinvi indicates whether the inverter reaches its power limits or not.
(k )
The weight vector Wi G is initialized to the unit vector ei , with 1 on the i-th
position and 0 everywhere else.
Algorithm 1 Push-Sum Gossip Protocol

1: Reset gossiping clock TG and start it
2: initialize:
(k ) (k )
kG = 0, σinvGi = 0, Si G = zeros(nc , 4),
h (k) i
(k ) (k ) (k) (k) (k)
Wi G = ei , (Si G )(i,:) = λinvi λinvi µinvi σinvi
(k )
3: while σ invGi has zero entry (or entries) and TG 6= deadline, do
4: kG ← n kG + 1 o
5: Let Ŝ(r) , Ŵ(r) be all pairs sent to the GVSF i at the iteration
kG − 1
(k ) (k )
Si G ← r Ŝ(r) , Wi G ← r Ŵ(r)
P P
6:
(k )
7: if all the entries of the vector Wi G are non-zero, then
(k )
8: σinvGi = 1
9: end if
(k ) (k )
10: Send 12 Si G and 12 Wi G to a random other GVSF and to the local
GVSF i (yourself)
11: hend
(k)
while i
(k) (k) (k ) (k )
12: λinvi λinvi µinvi = Si G /Wi G
(:,1:3)
(k ) (kG ) (k) (k ) (k )
Si G consists of four vectors: λinvi , λinvi , µinvGi and σ invGi . Each vector has nc
entries, each entry belongs to one of the GVSFs. As will be discussed later, the
(k )
vector µinvGi is used to detect the Lagrangian windup. In step 3 of the algorithm,
(k )
the vector σ invGi is used to make sure that the other GVSFs have filled their
matrices before the GVSF stops the gossiping iterations of the control iteration
(k )
k. In step 7, the GVSF detects a complete filling of its own Si G by observing
(kG )
the entries of the vector Wi .
(k ) (k )
Each entry of Si G represents a sum and Wi G is the weight of this sum
(entries of the same row have the same weight). After stopping the gossiping
iterations, the correct values of the Lagrangian multipliers and the power limit
indicators can be obtained by dividing each entry of the first three vectors by its
(k ) (k )
own weight, as shown in step 12. The matrix Si G and the vector Wi G are
always transmitted in pairs, so if a packet gets lost, all the other pairs continue
to have the correct values.
3.5.8 Anti-Windup Strategy
The proposed control method continuously repeats two steps. First, the GVSFs
optimize their control decisions in parallel; exceeding the voltage limit has a
fixed price, set by the Lagrangian multipliers. Secondly, the prices are updated;
if one of the controlled voltages exceeds the voltage limit, then the corresponding
price will be increased. This makes the over-voltage (or under-voltage) more
expensive, and in the next iteration the GVSFs should decide to use more
compensation to reduce the over-voltage. However, the optimization might
be infeasible; there might not be enough compensation available to resolve
the over-voltage issues. In such a case, the price will keep on increasing until
the inverters saturate. Past this point, increasing the price further has no use
and will slow down the response when the problem becomes feasible again.
This effect is similar to ’windup’ in classical control theory. The update of the
Lagrangian multipliers are disabled when all inverters reach their operational
limits, which is referred to hereafter as the anti-windup strategy.
3.5.9 Active and Standby Modes
Active and standby modes are proposed in this chapter to decide when to
activate/deactivate the GVSFs. There is no need to operate the GVSFs
continuously. When the controlled voltages are within the accepted limits,
then there is no need to activate the GVSFs, whereas all the GVSFs should be
activated when one (or more) of the controlled voltages is beyond the accepted
limits. After activating the GVSFs, the Lagrangian multipliers can be used to
know when to go to standby mode. The GVSFs can be deactivated when the
Lagrangian multipliers of all the GVSFs return back to zero. Zero values of
the Lagrangian multipliers indicate that the complicating constraints (3.1b) are
satisfied strictly. This means that voltages are below the maximum limit, and
above the minimum limit.
3.5.10 The Proposed Gossip-Based Dual Decomposition Al-

gorithm
In this subsection, we combine all the ideas and equations, presented in the
previous subsections, in one algorithm. Algorithm 2 is proposed in this chapter
to implement a distributed voltage control system provided by GVSFs of smart
PV inverters.
The algorithm includes different strategies that improve the performance of
the fully distributed DD-based voltage control system: active and standby
modes; iterative feedback; and anti-windup. The push-sum gossip Algorithm 1
is applied in Algorithm 2 to enable P2P data interchange between the GVSFs.
(k)
In step 14, the GVSF sets µinvi = 1 when the inverter reaches its power limits.
(k)
In case of voltage rise, µinvi is set to 1 when maximum reactive power absorption
and maximum active power curtailment are used. In case of voltage drop, on
(k)
the other hand, µinvi is set to 1 when maximum reactive power injection is used.
In step 16, when all the inverters reach their power limits, each GVSF freezes
its Lagrangian multipliers to avoid windup.
3.5.11 Convergence Speed of the DD-Based Distributed

Voltage Control Algorithm
Numerical tests on the simulated network of Figure 3.4, over 104 scenarios of
daily residential demand and PV profiles (5 s resolution), are performed to
study convergence speed of the DD-based distributed voltage control system
(Algorithm 2). The daily residential demand and PV profiles are generated
for a weekday in July, to incorporate the effect of high PV generation on grid
voltages. The time step of each control iteration k is 5 s, and the time step of
each gossiping iteration is 100 ms. The fastest convergence of Algorithm 2 is
achieved when the step size α is set to a constant value of 0.3. Algorithm 2
diverges when α is higher than 0.3.
Figure 3.8 shows the average convergence speed of the DD-based distributed
voltage control system with number of agents3 varying from 10 to 60. The
average convergence speed is calculated by taking the mean of convergence
speeds of different time steps for the 104 scenarios. One can notice that the
average convergence speed varies from 8.4 minutes to 65 minutes4 . When 60
agents participate in the voltage control, Algorithm 2 needs more than one hour
3 Distributed voltage control system with 10 agents, for example, means that only the PV
inverters of the last 10 nodes of Figure 3.4 (nodes 53-62) participate in the voltage control.
4 One minute represents 12 iterations (60 s/5 s (5 s is the time step of one control iteration))
Algorithm 2 DD-Based Fully Distributed Voltage Control

1: Each GVSF i carries out the following procedure:
2: vimeas ← measure the PCC voltage
3: if vimeas > v (or < v), or an active signal(s) is received, then
4: Send the received active signal(s) and your active signal using the standard
push-sum gossip presented in [90]
5: Go to active mode
6: initialize:
(0) (0)
λinvi = λinvi = zeros(nc , 1)
(0)
µinvi = 0, (vimeas )(0) = vimeas , k = 1
7: begin the control iterations:
8: (ppv
i )
(k−1)
← measure the PV generation
(k) (k)
9: Update ∆pinvi and ∆qinvi based on (3.25a)-(3.25d)
(k) (k)
10: Send ∆pinvi and ∆qinvi to the inner control loops of the inverter
11: (vimeas
n ) ← measure the PCC voltage (feedback ostrategy)
(k)
(k) (k) (k)
12: if ∆qinvi =−∆q invi and ∆pinvi = −cri (ppv
i )
(k−1)
(k)
n o
(k)
13: or ∆qinvi = ∆q invi , then
(k)
14: µinvi ← 1 (power limit indicator)
15: end if
(k)
16: if all the entries of the vector µinvi are 1, then
(k) (k−1) (k) (k−1)
17: λinvi ← λinvi , λinvi ← λinvi (anti-windup)
18: else
(k) (k)
19: Update λinvi and λinvi based on (3.28a) and (3.28b)
20: end if
(k) (k) (k)
21: Execute Algorithm 1 to update the vectors λinvi , λinvi and µinvi
(k) (k)
22: if all the entries of the vectors λinvi and λinvi are zero, then
23: Go to step 2 (standby mode)
24: else
25: k ← k + 1, repeat steps 8-25
26: end if
27: else
28: Stay in standby mode
29: end if
to converge and solve a single voltage problem. When 10 agents participate in

the voltage control, Algorithm 2 needs 8.3 minutes to converge.
To ensure satisfactory regulation of voltage profiles, the real-time distributed
voltage control system should solve a voltage problem within few minutes (e.g.
ADMM-BASED DISTRIBUTED VOLTAGE CONTROL 53
less than 10 minutes). It does not make sense to spend one hour, or even 30
minutes, to solve a single voltage problem in real-time. Therefore, we conclude
from this study that the DD method can be used to implement a distributed
voltage control system with few agents (e.g. 10 agents or less). We also see
from this study that the DD method has more potential in offline distributed
optimization (e.g. offline coordination of the agents) than online distributed
optimization.
Figure 3.8: Average convergence speed of the DD-based distributed voltage

control algorithm.
To tackle the slow convergence of the DD method, in the next section, we

modify the Lagrangian function by adding quadratic and proximal penalty
functions. Additionally, we accelerate the convergence of the Lagrangian
multipliers by using reactive and active acceleration factors.
3.6 ADMM-Based Distributed Voltage Control
An effective decomposition technique can be designed based on the Alternating

Direction Method of Multipliers (ADMM). ADMM utilizes the augmented
Lagrangian function, which incorporates a quadratic penalty (augmentation)
of the complicating constraints into the Lagrangian function. Including the
penalty term yields to a better convergence. ADMM technique allows a problem
with x and z variables to be solved by first solving for x with z fixed, and then
solving for z with x fixed. These variables are thus updated in an alternating
fashion. Hence the name alternating direction method of multipliers. Boyd
et al. [80] describe the ADMM algorithm and its benefits as “an algorithm
that is intended to blend the decomposability of dual ascent with the superior
convergence properties of the method of multiplier.”
The classic ADMM algorithm [133, 134] is a two-block system, it cannot be used
to decompose a centralized optimization problem into sub-problems that can be
solved in parallel. The 2-block ADMM can be extended to: 1) multi-block Gauss-
Seidel ADMM (GS-ADMM); and 2) multi-block Jacobi ADMM (J-ADMM).
Different forms of ADMM are reviewed in [135]. GS-ADMM is not amenable
for parallelization as the blocks are updated one after another (sequentially).
Additionally, the convergence is not guaranteed [136]. J-ADMM updates all
the blocks in parallel. The convergence of J-ADMM can be guaranteed under
certain assumptions [137], but it has a slow rate of convergence. Our simulation
results show that the convergence of the J-ADMM algorithm is even slower than
the convergence of the DD algorithm. In [138], a mathematical method called
JP-ADMM is developed. The work of [138] shows that the 2-block ADMM
can be extended to parallel multi-block ADMM and preserve a convergence
(at a rate of o(1/k)) by adding proximal terms to the augmented Lagrangian
function. The use of proximal terms to preserve the convergence of the parallel
multi-block ADMM has been also discussed in [139].
In this section, the Jacobi-Proximal ADMM (JP-ADMM) method is applied to
relax the optimization problem (3.1a)-(3.1f), and decompose it into sub-problems
that can converge quickly. These sub-problems are then solved separately by
the GVSFs in parallel, imposing only the local constraints. The motivation
behind using the ADMM is driven by its superior convergence, and its ability
to solve voltage problems within a reasonable time.
JP-ADMM applies an iterative optimization, in a Jacobi-fashion, of the Proximal
Augmented Lagrangian Function (PALF), followed by a steep ascent update
of the Lagrangian multipliers. In this chapter, JP-ADMM is integrated with
the push-sum gossip protocol to enable the GVSFs to communicate with each
other, and exchange their control variables and Lagrangian multipliers in a P2P
fashion. With P2P communication, the GVSFs coordinate themselves and make
the correct control decision in every particular situation to maintain voltages
within the required limits.
3.6.1 Equations of the JP-ADMM Algorithm
The main difference between the Augmented Lagrangian Function (considered

in 2-block ADMM and J-ADMM) and the PALF (considered in JP-ADMM)
is the addition of proximal penalty terms and acceleration factors for each
sub-problem. As will be demonstrated later, these additions help to speed up
the convergence of the algorithm. This subsection presents the derivation of the
PALF, and shows how the PALF can be used to decompose the optimization
problem (3.1a)-(3.1f) into sub-problems that can be solved in parallel. To derive

the PALF, let’s first define:
(k) (k) (k) (k)

U invi = vi − v , U invi = −vi +v (3.29)
2 2
(k) (k) (k)
finvi = cp,invi ∆pinvi + cq,invi ∆qinvi (3.30)
The PALF of the objective function (3.1a) and the complicating constraints
(3.1b)-(3.1c) can be formulated as:
(k) X (k) (k−1) (k) (k−1) (k)

LPALF = finvi + λinvi U invi + λinvi U invi
i∈I
(k) 2
ρinvi ρinvi 2
(k)
+ max 0, U invi + max 0, U invi
2 2
!
τinvi (k) (k−1) 2 τinvi (k) (k−1) 2
+ ∆pinvi − ∆pcinvi + ∆qinvi − ∆qinv
c
2 2 i
(3.31)
(k) 2

Here, ρinvi > 0 is the augmented penalty factor. max 0, U invi and
2
(k)
max 0, U invi are one-sided quadratic penalty functions used to penalize
(k)
the objective function only when vi is higher than v or less than v.

(k) (k−1) 2
τinvi > 0 is the proximal penalization factor. ∆pinvi − ∆pcinvi and
2
(k) (k−1)
∆qinvi − ∆qinv are the proximal terms. These terms penalize the
c

i
(k) (k)
deviation of the control variables at the k-th iteration (∆pinvi and ∆qinvi ) from
the control decisions (∆pcinvi )(k−1) and (∆qinv
c
i
)(k−1) that have been calculated
at the previous iteration (superscript c: constant). The proximal terms are used
to preserve the convergence of the extended parallel multi-block JP-ADMM. To
derive the sub-problems, let’s first define the following:
(k) p q (k) (k)

Yd,i = vd,i ∆pinvi + vd,i ∆qinvi (3.32)
(k)
X p q

Cd,i = vd,j (∆pcinvj )(k−1) + vd,j (∆qinv
c
j
)(k−1) (3.33)
j∈I
j6=i
To decompose the PALF into local Proximal Augmented Lagrangian Functions,

each GVSF fixes the control variables and Lagrangian multipliers of the other
GVSFs by considering their control variables and Lagrangian multipliers that
have been calculated at the previous control iteration k − 1. Based on this, and
based on the branch flow model (3.5), it can be shown that the local PALF of
the GVSF i can be given by equation (3.34). The constants that do not affect
the minimization of the local PALF are not included in (3.34).
(k)
(k−1)

(k) (k−1) (k)
X
LPALF
invi = finvi + λinvd − λinvd Yd,i
d∈I
ρinvd 2
(k) (k)
+ max 0, (vdmeas )(k−1) + Yd,i + Cd,i − v
2
!
ρinvd 2
(k) (k)
+ max 0, −(vdmeas )(k−1) − Yd,i − Cd,i + v
2
τinvd (k) (k−1) 2 τinvd (k) (k−1) 2

+ ∆pinvi − ∆pcinvi + ∆qinvi − ∆qinv
c
(3.34)
2 2 i
To reach the same solution as solving the optimization problem (3.1a)-(3.1f)

with a constant linear voltage model, the JP-ADMM algorithm iterates as
shown in Algorithm 3. The factor γinvi > 0 in the algorithm is the acceleration
factor. At each control iteration k, the local Proximal Augmented Lagrangian
Functions are minimized separately, in parallel, followed by an update of the
Lagrangian multipliers. To guarantee the convergence, as demonstrated in
[138], the proximal penalization factor should be tuned to respect the following
relation:

nc
τinvi > ρinvi − 1 (3.35)
2 − γinvi
3.6.2 Acceleration Factors
The acceleration factor γinvi is used to have a steep ascent update of the
Lagrangian multipliers, which helps in increasing the speed of convergence. In
Algorithm 3 JP-ADMM iterations

for k = 0, 1, ... do
(k) (k)
Update ∆pinvi and ∆qinvi for i = 1, ..., nc in parallel by:
(k)
argmin{∆p(k) ,∆q(k) } LPALF
invi (3.36a)
invi invi
Subject to the local constraints:

(k−1) (k)
(−cri ) (ppv
i ) ≤ ∆pinvi ≤ 0 (3.36b)
(k) (k) (k)
(k) (k)
Update λinvi and λinvi for i = 1, ..., nc in parallel by:
(k)
(k−1)

λinvi = max 0, λinvi + γinvi ρinvi (vimeas )(k) − v (3.37a)

(k) (k−1)
λinvi = max 0, λinvi − γinvi ρinvi (vimeas )(k) − v (3.37b)
end for
this work, two acceleration factors are proposed: 1) the reactive acceleration
q p
factor γinv i
; and 2) the curtailment acceleration factor γinv i
. The acceleration
q
factor γinvi is used to accelerate the convergence of the controlled voltages to
p
v (or v) when reactive power compensation is used, whereas γinv i
is used
to accelerate the convergence when active power curtailment is used. As
active power curtailment is penalized more than reactive power compensation,
convergence of the controlled voltages to the accepted limits using curtailment
is slower than the convergence when using reactive power compensation. Hence,
p q
γinv i
should be greater than γinv i
.
3.6.3 The Proposed G-JP-ADMM Voltage Control Algorithm
Here, we extend Algorithm 3 to include: acceleration for reactive power

compensation; acceleration for active power curtailment; the feedback strategy
of Figure 3.7; the anti-windup strategy presented in Subsection 3.5.8; and the
active/standby mode discussed in Subsection 3.5.9. These additional features
improve the performance of the G-JP-ADMM algorithm. The gossiping push-
sum algorithm (Algorithm 1) is applied in Algorithm 4 to disseminate the control
variables and the Lagrangian multipliers among the GVSFs. Each GVSF needs
these data to substitute them in equations (3.36a)-(3.37b). Algorithm 1 is

slightly changed here to include the dissemination of the control variables.
(k) (k) (k) (k) (k)
In step 6, ∆pinvi , ∆qinvi , λinvi , λinvi , and µinvi are vectors with nc entries,
each entry represents a control variable, Lagrangian multiplier or power limit
indicator for one of the GVSFs.
Algorithm 4 G-JP-ADMM Algorithm

1: Each GVSF i carries out the following procedure:
2: vimeas ← measure the PCC voltage
3: if vimeas > v (or < v), or an active signal(s) is received, then
4: Send the received active signal(s) and your active signal using the standard
push-sum gossip presented in [90]
5: Go to active mode
6: initialize:
(0) (0) (0) (0)
∆pinvi = ∆qinvi = λinvi = λinvi = zeros(nc , 1)
(0)
µinvi = 0, (vimeas )(0) = vimeas , k = 1
7: begin the control iterations:
8: (ppv
i )
(k−1)
← measure the PV generation
(k) (k)
9: Update ∆pinvi and ∆qinvi based on (3.36a)-(3.36c)
(k) (k)
10: Send ∆pinvi and ∆qinvi to the inner control loops of the inverter
11: (vimeas
h ) ← measure the PCC
(k)
i voltageh (feedback strategy)
(k)
i
(k) (k) (k−1)
12: if | ∆qinvi |> 0.9 × ∆q invi and ∆pinvi < 0.1 × (−cri ) (ppvi ) ,
then
p
13: γinvi ← γinv i
(acceleration of the curtailment)
14: else
q
15: γinvi ← γinv i
(acceleration of the reactive power)
16: end if
(k) (k) (k) (k−1) (k) (k)
17: if ∆qinvi = −∆q invi and ∆pinvi = (−cri ) (ppv i ) , or ∆qinvi = ∆q invi ,
then
(k)
18: µinvi ← 1 (power limit indicator)
19: end if
(k)
20: if all the entries of the vector µinvi are 1, then
(k) (k−1) (k) (k−1)
21: λinvi ← λinvi , λinvi ← λinvi (anti-windup)
22: else
(k) (k)
23: Update λinvi and λinvi based on (3.37a) and (3.37b)
24: end if
(k) (k) (k) (k)
25: Execute Algorithm 1 to update the vectors ∆pinvi , ∆qinvi , λinvi , λinvi
(k)
and µinvi
ANALYSING THE PERFORMANCE OF THE JP-ADMM-BASED DISTRIBUTED VOLTAGE CONTROL
ALGORITHM 59
(k) (k)
26: if all the entries of the vectors λinvi and λinvi are zero, then
27: Go to step 2 (standby mode)
28: else
29: k ← k + 1, repeat steps 8-29
30: end if
31: else
32: Stay in standby mode
33: end if
p
In step 13, the switching to the curtailment acceleration factor γinv i
occurs
when most of the available reactive power capacity is used (90%) and the active
power curtailment has begun to be used (10%). In step 18, the GVSF sets
(k)
µinvi = 1 when the inverter reaches its power limits. In case of voltage rise,
(k)
µinvi is set to 1 when maximum reactive power absorption and maximum active
(k)
power curtailment are used. In case of voltage drop, on the other hand, µinvi
is set to 1 when maximum reactive power injection is used. In step 20, when
all the inverters reach their power limits, each GVSF freezes its Lagrangian
multipliers to avoid windup.
3.7 Analysing the Performance of the JP-ADMM-

Based Distributed Voltage Control Algorithm
3.7.1 Parameters of the Algorithm
The maximum and minimum voltage limits are set to 207 V and 253 V,
respectively. The sensitivity of the controlled voltages to the change in the
active/reactive power is calculated based on the constant linear voltage model
presented in Section 3.3.
To prioritize the use of reactive power, while active power curtailment is
performed only as a last resort, the reactive power penalization factor cq,invi is
set to 1 and the curtailment penalization factor cp,invi is set to 200 (the tuning
is based on trial and error).
q p
To tune γinv i
and γinv i
, we did two experiments. The first experiment was
q p
to tune γinvi , whereas the second experiment was to tune γinv i
. In the first
experiment, we disabled the active power curtailment, to regulate voltage profiles
q
using only reactive power. γinv i
was initialized to 1. After that, we increased
q
γinvi gradually till undesirable oscillation was noticed in voltage profiles. In the
second experiment, we disabled the reactive power compensation, to regulate

voltage profiles using only the active power curtailment. After that, we increased
p
γinv i
gradually till undesirable oscillation was noticed in voltage profiles. For
q p
the case study of this chapter, γinv i
is set to 500, whereas γinv i
is set to 104 .
The proximal penalization factor τinvi in (3.35) depends on the number of
inverters nc participating in the voltage control, the factor γinvi and the
augmented penalization factor ρinvi . It is desirable to have small τinvi , that
respects the relation (3.35), to increase the speed of convergence by being less
conservative in moving from one control variable to another, e.g. to have a bigger

(k−1) (k)
step from ∆qinvi to ∆qinvi . In our work, we set τinvi = ρinvi 2−γ nc
invi
−1 +1
and ρinvi is set to be 0.001.
(k )
The deadline for each GVSF to fill Si G is set to 1 minute. For 50 GVSFs,
for example, each GVSF needs in average around 12 gossiping iterations (see
the convergence rate in [90]) to fill the matrices. With 100 ms latency, the
GVSFs need around 1.2 s to fill the matrices. Hence, the time needed to fill the
matrices is expected to be much less than 1 minute.
3.7.2 62-Bus Case Study
Algorithm 4 is used to implement a fully distributed voltage control system with

20 agents to regulate voltage profiles of the 62-bus network of Figure 3.4. The
twenty agents participating in voltage control are smart inverters No. 11, 12,
23, 24, 29, 32, 39-44, and 55-62. These inverters are located at the end of the
cables. The inverters at the end of the cables are selected to control voltages,
because they have higher impact on voltage variation than the inverters at the
beginning of the cables.
The Algorithm is executed for a low consumption summer day in July to be able
to incorporate the effect of high PV generation. The simulation is performed for
the entire day. Figure 3.9 shows the total PV generation and the total active
power consumption of the 62 households.
Voltage Rise Mitigation
The low load and high PV generation result in reversed power flows. The
reversed power flows cause a voltage rise beyond the EN 50160 limits. Figure
3.10 shows voltage profiles of the 62 households (phase to neutral voltages).
One can see that most of voltage profiles exceed the voltage limit v = 253 V.
To solve the voltage rise problem, the P2P-based GVSFs of the 20 inverters
ALGORITHM 61
Generation
Consumption
Figure 3.9: Aggregated PV generation and active power consumption of the 62

households.
are activated. The regulated voltage profiles are shown in Figure 3.11. The
existing violations of the over-voltage limit are mitigated. The voltage peaks,
resulting from steep change in PV generation, fall beyond the limits, because
the inverters need some time to react to the change (convergence time).
Figure 3.10: Voltage profiles of the 62 households without voltage control.
To quantify the performance of the algorithm, a voltage quality metric E is

proposed in this chapter. The metric E, as illustrated in (3.38), integrates the
over-voltages over time. This means that both the duration of the voltage rise
problem and its severity will increase the metric E. A value of zero is the
best possible value and indicates that there are no over-voltage issues. The
higher the value of E, the worse the voltage problem. Figure 3.12 shows the
voltage quality metric E of the 62 households with and without voltage control.
The voltage problem starts at household No. 18 and gets worse as we move
towards the end of the feeder. One can notice that the voltage quality metric
of the different households is reduced by a factor of 100 (approximately) after
Figure 3.11: Voltage profiles of the 62 households with voltage control.
activating the GVSFs, which means that the voltage quality of the voltage
profiles has been improved significantly.
Z tend =24:00
Ei = max (vimeas − v , 0) dt (3.38)
tstart =00:00
without control
with control
Figure 3.12: Voltage quality metric E of the 62 households with and without
voltage control.
Figure 3.13a shows the Lagrangian multipliers λinv62 5 of smart inverter No. 62.
The GVSF of inverter 62 is in standby mode between 00:00 and 08:40; when
voltages are within the limits. The GVSF starts to regulate voltages at 08:40
(active mode), when voltages start to violate the maximum limit. The GVSF
returns back to standby mode and the Lagrangian multiplier returns back to
5λ is zero for the entire day, since there are no voltage drop problems.
inv62
ALGORITHM 63
zero at 16:35, when voltages return back to normal values due to solar irradiance
reduction. Figure 3.13b shows the control variables of smart inverter No. 62.
The average reactive power inverter 62 absorbs during the day is around -4
kvar, whereas the average active power the inverter curtails is around -0.5 kW.
∆pinv62 and ∆qinv62 return to zero when the Lagrangian multipliers return back
to zero.
pinv [kW]
62
qinv [kvar]
62
(a) (b)
Figure 3.13: (a) Lagrangian multiplier λinv62 of smart inverter No. 62, (b)
Active power curtailment and reactive power compensation of smart inverter
No. 62.
The Convergence of JP-ADMM, J-ADMM and DD
Figure 3.14 shows a comparison of the convergence speed of the proposed JP-
ADMM algorithm (Algorithm 4), the J-ADMM algorithm, and the DD algorithm
(Algorithm 2). The J-ADMM algorithm can be implemented as presented in
Algorithm 4, but without including the proximal terms and the acceleration
factors [138]. As discussed earlier, the best results for the JP-ADMM algorithm
(in terms of convergence speed) occur when γ q = 500, γ p = 104 , and ρ = 0.001
(for all the GVSFs). For the case study of this subsection, the best results for
the J-ADMM algorithm occur when ρ is set to 0.1 for all the GVSFs. The
J-ADMM algorithm does not converge when ρ is higher than 0.1. The fastest
convergence of the DD algorithm is achieved when the step size α is set to 0.3
for all the GVSFs. The DD algorithm diverges when α is higher than 0.3.
The three algorithms are implemented at each GVSF of the 20 inverters, and
are executed at 14:00, the time when voltage profiles have the highest voltage
rise. Figure 3.14 presents the convergence of the PCC voltage of smart inverter
JP-ADMM, q =500, p
=10 4 , =0.001
J-ADMM, =0.1
DD, =0.3
Figure 3.14: Comparison of the convergence speed between JP-ADMM, J-

ADMM and DD.
No. 62 (worst voltage). We can see that the JP-ADMM algorithm is clearly
the fastest one among the compared algorithms. The JP-ADMM algorithm
returns the PCC voltage back to the limit in less than 10 iterations, whereas the
J-ADMM algorithm needs around 500 iterations, and the DD algorithm needs
arround 370 iterations to bring the PCC voltage back to the limit. From this
study, we can also conclude that the J-ADMM algorithm is even slower than
the DD algorithm. It is worth to mention again that the JP-ADMM enjoys a
fast preserved convergence thanks to the proposed acceleration factors and the
applied proximal penalization functions.
Accuracy of the Linear Voltage Model
The linear voltage model presented in Section 3.3 is used to predict voltages of
(k) (k)
the control iteration k before applying the control decisions ∆pinvi and ∆qinvi
(∀i ∈ I). The optimization of the control decisions depends highly on how
accurate the prediction is. The prediction made by equation (3.5) is quite
accurate. Figure 3.15 shows a comparison between the predicted voltage and
the actual one of smart inverter No. 62. The algorithm is executed for two hours,
from 14:00 till 16:00 (period of worst voltage problem). The predicted voltage
is calculated by equation (3.5), whereas the measured voltage is calculated by
exact power flow equations (using MATPOWER). One can notice the high
accuracy of the linear voltage model. The maximum error is around 0.3%.
The high accuracy of the prediction is due to two reasons; first, the algorithm of
the proposed method is an iterative one. The system takes small steps towards
the final solution. At each iteration, the change in reactive power and active
ALGORITHM 65
Predicted voltage
Measured voltage
Figure 3.15: Predicted voltage vs. actual voltage of smart inverter No. 62.
power curtailment is small. For a small change, the first order approximation of
equation (3.2a) is quite accurate, and second, thanks to the feedback strategy,
the system corrects the error in its decision at each control iteration. This means
that the error does not accumulate. Figure 3.16 presents a comparison between
the predicted voltage and actual voltage for two cases, with and without the
feedback strategy. With the feedback strategy the error is around 0.3%, whereas
without the feedback strategy the error is around 1.2%. One can notice that
without the feedback strategy the algorithm brings the actual voltage to a value
lower than the maximum limit. This means that the voltage is over-regulated,
and the amount of the absorbed reactive power and curtailed active power is
more than needed.
3.7.3 124-Bus Case Study
To test the performance of the proposed G-JP-ADMM algorithm on a large

system, we have extended the network of Figure 3.4 from 62-bus (with 20
smart inverters participating in voltage control) to 124-bus (with 50 smart
inverters participating in voltage control). We extended the following feeders
Predicted voltage Predicted voltage

Measured voltage Measured voltage
(a) (b)
Figure 3.16: (a) Voltage regulation with the feedback strategy, (b) Voltage
regulation without the feedback strategy.
to host more households: feeder DE to host 25 households, feeder DG to host

32 households, and feeder DH to host 40 households. Additionally, we set the
size of the 124 inverters to 5 kVA. Figure 3.17a shows the voltage profiles of
the 124 households without voltage control, whereas Figure 3.17b shows the
voltage profiles with voltage control. Again we notice that our proposed method
succeeds in improving the quality of voltage profiles significantly.
(a) (b)
Figure 3.17: (a) Voltage profiles of the 124-bus network without voltage control,
(b) Voltage profiles of the 124-bus network with voltage control.
Figure 3.18 demonstrates the scalability of the proposed G-JP-ADMM method.

It shows a comparison between voltage convergence of smart inverter No. 62
ALGORITHM 67
(62-bus network) and voltage convergence of smart inverter No. 124 (124-bus
network)6 , at 14:00. In case of 20 agents (62-bus network), the algorithm takes
8 iterations to bring the voltage back to the limit, whereas the algorithm takes
15 iterations to bring the voltage back to the limit in case of 50 agents (124-bus
network). Hence, increasing the number of agents does not lead to exponential
increase in the number of iterations. This is because in G-JP-ADMM, each
agent treats the decision variables of other agents as constants.
62-bus network
124-bus network
Figure 3.18: Voltage convergence of smart inverter No. 62 (62-bus network)

and smart inverter No. 124 (124-bus network).
Figure 3.19 shows a comparison between JP-ADMM, DD and J-ADMM, at

14:00 (124-bus network). The figure presents the convergence of PCC voltage
of smart inverter No. 124. Again we notice that JP-ADMM method is much
faster than DD and J-ADMM.
Scalability of the Push-Sum Gossip Protocol
Figure 3.20 shows the number of GVSFs each GVSF needs to communicate
with, to know the control decisions and the Lagrangian multipliers of all the
GVSFs. In case of 20 smart inverters participating in voltage control, each
GVSF needs to communicate with 8 to 13 GVSFs. In case of 50 smart inverters
participating in voltage control, each GVSF needs to communicate with 10 to
14 GVSFs. The number (of GVSFs to be communicated with) is not fixed since
each agent contacts random agent at each gossiping iteration. The results of
6 Smart inverter No. 124 is at the end of the extended feeder DH.
JP-ADMM
J-ADMM
DD
Figure 3.19: 124-bus network: Comparison of the convergence speed between

JP-ADMM, J-ADMM and DD.
Figure 3.20 demonstrate the fact that each GVSF needs to communicate with
some GVSFs (not all) to know the status of all other GVSFs. The results also
demonstrate the scalability of the proposed push-sum gossip protocol. The
gossip-based protocol enjoys exponential rapid spread of information. Increasing
the number of smart inverters does not significantly increase the communication
burden.
20 agents
50 agents
Figure 3.20: Number of agents to be communicated with per control iteration.

CONCLUSION 69
Discussion
The time needed to finish one control iteration of JP-ADMM, J-ADMM and
DD is around 5 seconds7 . Based on the results presented in the 124-bus case
study, the proposed JP-ADMM needs around 1.25 minutes (5s × 15 iterations)
to bring the voltage of bus No. 62 to the accepted limit, whereas J-ADMM
needs around 86.6 minutes (5s × 1040 iterations), and DD needs around 59.4
minutes (5s × 713 iterations). We conclude from this analysis that the proposed
JP-ADMM algorithm succeeds in regulating the voltage profiles to comply with
the European standard EN 50160, whereas DD and J-ADMM fail to comply
with the standard EN50160, since they fail to maintain all 10 min mean values
of voltages within the range [207 V, 253 V].
The problem treated in this chapter is a convex problem. For non-convex
problems, we would like to refer the readers to the coupled local minimizers
(CLM) method [140, 141]. CLM is intended for fast global optimization
of non-convex problems and speeds up simple gradient descent schemes in
convex problems. In CLM, a cooperative search mechanism is set up using a
population of local optimizers, which are coupled during the search process by
synchronization constraints.
3.8 Conclusion
This chapter presents two novel distributed voltage control methods for PV
inverters to expand their features with added grid voltage support functions.
In the first method, the state-of-the-art dual decomposition (DD) method is
modified to a fully distributed dual decomposition, by decomposing its master
problem into local sub-problems. The fully distributed dual decomposition is
then used to convert a centralized optimization-based voltage control system
into a distributed one. Simulation results show that the DD-based distributed
voltage control system suffers from slow convergence.
In the second method, we succeeded in speeding up the convergence of the
distributed voltage control system, by applying the Jacobi-Proximal Alternating
Direction Method of Multipliers (JP-ADMM). The JP-ADMM enjoys a fast
preserved convergence thanks to the proposed acceleration factors and the
applied proximal penalization functions. For a distributed voltage control
7 Based on the performance of the workstation, and assuming the following: a)1 second as
the time needed to measure the PV generation and the voltage, b) 50 ms as the response
time of the inverter inner control loop, c) 100 ms communication latency
system with 50 agents, for example, the JP-ADMM needs 15 iterations to

converge, whereas the DD needs 713 iterations to converge.
Although the JP-ADMM method is much faster than the DD method, its
implementation is more complex than the DD method. The JP-ADMM is
sensitive to four parameters that should be tuned carefully to obtain satisfactory
results: the augmented penalty factor; the proximal penalty factor; the reactive
acceleration factor; and the curtailment acceleration factor. Tuning these
parameters is not an easy task, and requires several experiments of trial and error.
The DD method is sensitive to only one parameter, the step size. Moreover,
including the quadratic augmented and proximal penalty terms in the Proximal
Augmented Lagrangian Function requires each agent (in JP-ADMM) to know
the control decisions and the Lagrangian multipliers of all the other agents. In
DD, each agent needs to know only the Lagrangian multipliers of all the other
agents. Hence, communication overhead per iteration is more in JP-ADMM
than DD.
To simplify the implementation of real-time distributed voltage control systems,
in Chapter 5, we use forecast data and forecast error of residential demand and
PV profiles to learn robust linear voltage control policies on a day-ahead basis.
We then let the GVSFs apply the voltage control policies in real-time. The
idea is to let the GVSFs solve, at each time step in real-time, a set of linear
equations rather than solving a complex optimization problem.
Chapter 4
Laboratory Implementation of
Peer-to-Peer Voltage Control
System
A hardware setup of a laboratory-based testbed is presented in this chapter,

to better understand the real implementation of distributed voltage control
provided by peer-to-peer-based grid voltage support functions. The architecture
of the setup consists of four layers: microgrid, control, communication, and
monitoring. The contents of this chapter are based upon the journal paper:
• H. Almasalma, S. Claeys, K. Mikhaylov, J. Haapola, A. Pouttu, and G.

Deconinck, “Experimental Validation of Peer-to-Peer Distributed Voltage
Control System,” Energies 2018, vol. 11, no. 5, pp. 1–22, 2018.
4.1 Introduction
There are few studies in existing literature addressing experimental imple-

mentation of distributed control algorithms. Laboratory implementation of
distributed control systems are thus lacking. In [142], a gossip-based peer-to-
peer (P2P) voltage control system has been tested in a pilot site; the work
is part of the European Commission FP7 DREAM project. Six households
were equipped with smart control agents, which measure households’ active
power consumption, and control households’ flexible loads. Each agent is
71
72 LABORATORY IMPLEMENTATION OF PEER-TO-PEER VOLTAGE CONTROL SYSTEM
connected to a local internet gateway, and a virtual private network is then

used to enable P2P communication between agents. In [143], a multi-agent
platform has been implemented and used to test a dual-decomposition-based
optimization method for controlling the prosumers’ flexibility. The distributed
agents are implemented in Raspberry Pi computers. The agent-based control
algorithm of each agent is implemented in Python and executed via MATLAB
calls. The setup is part of the LINEAR project [144]. In [145], a gossiping
P2P semantic overlay network is implemented to enable P2P communication
between agents. The toolbox has been used to implement a distributed tertiary
control system, which allows group of generators to operate at an economical
optimum. In [146], distributed reactive power control system has been tested
experimentally using real power inverters; coordination between inverters is
obtained by exchanging information via an IP-based communication network.
The work in [147] presents an approach for ADMM-based DERs coordination
coupled with real-time simulator and cluster of Raspberry Pi computers and
PV inverters.
This chapter presents the design, development and hardware setup of a
laboratory-based P2P voltage control testbed1 . The dual decomposition method
is used in this chapter to solve a voltage optimization problem in a fully
distributed way. The dual decomposition is used here and not the JP-ADMM
method because the setup consists of few agents. LoRaWAN-based device-to-
device communication modules are integrated with the P2P voltage control
system, to enable P2P data interchange between agents of the laboratory
microgrid.
The rest of this chapter is organized as follows. The laboratory-based P2P
voltage control testbed is described in Section 4.2. Section 4.3 presents the
P2P-based voltage control algorithm. Setup of the inverters control and their
inner control system are presented in Section 4.4. The device-to-device (D2D)
communication modules are described in Section 4.5. Section 4.6 presents the
experimental results. Finally, the chapter is concluded in Section 4.7.
4.2 Testbed Architecture
The architecture of the P2P voltage control testbed is depicted in Figure 4.1.
The testbed consists of four different layers which interact with each other: (1)
microgrid layer; (2) control layer; (3) communication layer; and (4) monitoring
1 The testbed was developed in the P2P-SmarTest project. The communication modules of
the testbed have been built and supplied by the Centre for Wireless Communications (CWC,
University of Oulu).
TESTBED ARCHITECTURE 73
layer. The microgrid layer consists of programmable inverters (label 1 in Figure

4.1); connected to DC power supplies (label 2). The inverters emulate prosumers
with photovoltaic (PV) installations, they are connected to the grid by resistors
in series with inductors (label 3). The resistors and inductors are used to
emulate a low voltage feeder. The control layer consists of inner control systems
(label 4) that drive power inverters, and grid voltage support functions (GVSFs)
that control voltage profiles of the micogrid (label 6). The communication layer
consists of D2D communication modules (label 7) that are used to disseminate
the status of voltage profiles in a P2P fashion. The monitoring layer consists of
voltmeters (label 9) and data acquisition platform (label 11).
The P2P voltage control testbed consists of three types of agents: (1)
actuators; (2) observers; and (3) a monitor. Each GVSF is connected to a
D2D communication module, and together they form an actuator agent (label
5). Actuator agents are connected to the programmable inverters through inner
control systems (control loops) and participate actively in voltage control by
calculating the change in reactive power and active power that each inverter
should follow to maintain voltage profiles within specified limits. The setpoints
of the change in reactive and active powers of inverters are determined based
on an optimization problem solved in a fully distributed way.
The observer agent (label 8) consists of a voltmeter connected to a D2D
communication module through a Raspberry Pi (R.Pi) computer (label 10). The
voltmeter periodically measures the voltage of its bus, and the R.Pi fetches the
latest reading. The R.Pi of each voltmeter hosts a software that was developed
for interfacing with both the communication module and the voltmeter. The
R.Pi calculates control signals (further referred to as Lagrangian multipliers)
based on the latest voltage measurement, according to a procedure described
later. These control signals are broadcasted through the D2D modules. The
actuators communicate with the observers in a P2P fashion to receive the
control signals. The actuators then determine how to react based on these
control signals and based on their impact on the observed voltages (the impact
on voltages is expressed by voltage sensitivities). They also take into account
the cost of dispatching a change in active power and reactive power.
The third type of agent, the monitoring agent (label 11), represents a data
acquisition platform. This additional agent is not required for the operation
of the P2P voltage control algorithm. Observers and actuators record several
variables from the algorithm that they execute, together with timestamps. These
recordings are cached locally. Periodically, observers and actuators transfer
the cached recordings to the monitoring agent in a robust way. Therefore,
even if the data acquisition network is temporarily offline, no data will be lost.
The monitoring agent hosts a web service, through which all recorded data is
visualized in several dashboards.
Monitoring agent
Data Acquisition
Monitoring
(11)
Observer agent Observer agent Observer agent
VM R. Pi VM R. Pi VM R. Pi
(9) (10) (9) (10) (9) (10)
D2D D2D D2D

(7) (7) (7)
Communication
(8) (8) (8)
P2P communication
D2D D2D
Actuator agent
(7) (7)
Actuator agent
(5)
Voltage measurement
Voltage measurement
Voltage measurement
(5)
Control
GVSF GVSF
(6) (6)
Inner control Inner control

(4) system
(4) system
L R L R
Microgrid
(3) (3)
Bus 1 Bus 2 Bus 3
DC power DC power
(2) supply
(1) (2) supply
(1)
Inverter Inverter
Figure 4.1: Multi-layer multi-agent architecture of the P2P voltage control

testbed (VM: voltmeter, D2D: device-to-device communication module, GVSF:
grid voltage support function, R.Pi: raspberry pi computer, R: resistor, L:
inductor, labels (1) to (11) indicate different parts of the testbed).
DUAL-DECOMPOSITION-BASED P2P VOLTAGE CONTROL ALGORITHM 75
The overall schematic of the testbed2 is depicted in Figure 4.2. The microgrid,
shown in Figure 4.3, is connected to the main grid through 400 V (line-to-line
voltage (L-L)), 64 A busbar.
(7)
D2D
Actuator agent
(7) (7) (7)
D2D D2D D2D
(5)
(8) (8) (8)
Observer agent Observer agent Observer agent
VM VM VM RTT (12)
1
R. Pi 2 R. Pi 3 R. Pi
(10) (10) (10)
(9) (9) (9)
L1 R1 L2 R2 (1)
Inverter 2
Node 0
Node 1
Node 2
(3) (3)
(7)
D2D
Grid (400 V L-L) DC Power
From other (1) Supply
agents Router
(2)
R. Pi Inverter 1
RTT
(12) Actuator agent

Monitoring (5)
agent
DC Power
(11) Supply
(2)
Figure 4.2: Schematic of the P2P voltage control testbed (VM: voltmeter, D2D:
device-to-device communication module, R: resistor, L: inductor, RTT: real
time target computer, R.Pi: raspberry pi computer, L-L: line-to-line, labels
(1)-(12) indicate the different parts of the testbed, labels (1)–(11) same as in
Figure 4.1).
4.3 Dual-Decomposition-Based P2P Voltage Con-

trol Algorithm
The P2P voltage control algorithm regulates voltages within allowed limits
based on an optimization problem. The algorithm uses a minimum change
2 Short video on the testbed: https://www.youtube.com/watch?v=CgZn-A5Ejww
(12) (2)
Grid
(7)
(1) (10)
(5)
(3)
(9)
(2)
(7) (10) (3)

(5) (1)
(12)
(9)
Figure 4.3: The laboratory microgrid (labels same as in Figure 4.1)
in reactive and active powers of some participating inverters installed in the

laboratory microgrid to control the voltage. The derivation of the algorithm is
presented in Chapter 3; here we present the algorithm in a more practical way.
Without compensation, each inverter i injects a certain amount of active power
into the system. In reality, this active power originates from the solar energy
received by the PV module. The inverter can additionally inject reactive power,
as long as the total apparent power does not exceed the inverter rating. The
inverter has an additional degree of freedom; it can curtail a fixed percentage of
DUAL-DECOMPOSITION-BASED P2P VOLTAGE CONTROL ALGORITHM 77
the active power. Therefore, the actuator agent can take two actions: reducing
(k)
the active power (by an amount ∆pinvi ) and injecting or absorbing reactive
(k)
power (by an amount ∆qinvi ).
(k)
Each actuator agent i solves the following optimization problem to find ∆pinvi
(k)
and ∆qinvi , at each time step k:
2 2
(k) (k)
argmin cp,invi ∆pinvi + cq,invi ∆qinvi + (4.1a)
(k) (k)
∆pinv ,∆qinv
i i
X (k−1)

p (k) q (k)

λy − λ(k−1)
y vy,i ∆pinvi + vy,i ∆qinvi
y∈Y
Subject to:
(k) (k)
(−cri ) (ppv
i ) ≤ ∆pinvi ≤ 0 (4.1b)
(k) (k) (k)
r 2
(k)

2 (k) (k)
i ) + ∆pinvi (4.1d)
Here, i ∈ I is the number (index) of the actuator agent (I is the set of actuators
(control nodes) participating in voltage control). y ∈ Y is the number (index)
of the observer agent (Y is the set of observers participating in voltage control).
2
(k)
cp,invi ∆pinvi represents the quadratic cost of a change in active power
2
(k) (k)
of inverter i with an amount ∆pinvi at time step k, while cq,invi ∆qinvi
represents the quadratic cost of a change in reactive power of inverter i with an
(k)
amount ∆qinvi at time step k. cp,invi and cq,invi are constant factors used to
(k) (k)
penalize the control variables ∆pinvi and ∆qinvi . These factors define priorities
for the control actions. It is supposed that reactive power control of the inverter
is cheaper than cutting its active power. Therefore, cp,invi should be greater
than cq,invi in a sense that gives priority of the control action to the reactive
power. When the reactive power of the inverter is not sufficient, active power
curtailment of the inverter will be used to regulate system voltages. In our
control system, we set cp,invi = 200 and cq,invi = 1, ∀i ∈ I. Active power
curtailment can be penalized more to minimize its use, but having higher cp,invi
would decrease the speed of convergence when the curtailment is used to return
voltages back to limits.
p q
vy,i and vy,i are the sensitivity of the voltage at bus y to the change in the
active power and reactive power of inverter i, respectively. As discussed in
Chapter 3, these voltage sensitivities can be calculated based on the impedance
and topology of the microgrid. cri is the curtailment factor. In this chapter,
cri is set to 30%, ∀i ∈ I. In reality, cri can be set based on how much the
(k)
prosumer would like to curtail the active power. (ppv i ) represents the active
power generation of the emulated PV module connected to inverter i, at time
step k. sinvi is the rated apparent power of inverter i.
(k−1)
λy and λ(k−1)
y are the control signals of violating the maximum and minimum
(respectively) allowed voltages at bus y. They are calculated at the previous time
step k − 1 and considered in the optimization of time step k. Mathematically
speaking, they represent the Lagrangian multipliers. Each observer measures
the voltage at its bus and updates these control signals based on the following
equations:
(k)
(k−1)
(k)
λy = max 0, λy + α vymeas −v
(4.2)
meas (k)
λy = max 0, λy
(k) (k−1)

− α vy −v
(k)
where λy and λ(k)
y are the updated control signals calculated at time step k,
(k)
and considered in the optimization of time step k +1. vymeas is the measured
(k) (k)
voltage at bus y after applying the decisions ∆pinvi and ∆qinvi . v and v are
the maximum and minimum allowed voltages, respectively. We set v = 1.1 p.u.
(per unit) and v = 0.9 p.u., according to the European standard EN 50160 [3].
The parameter α is the step size of the dual decomposition method. Because of
the Karush-Kuhn–Tucker conditions (KKT), the Lagrangian multipliers cannot
be smaller than zero. This explains the use of maximum operator in (4.2).
The control algorithm goes through the following steps:
1. Each observer agent measures the voltage. If the voltage exceeds the upper
voltage limit v, it will increase λy . If the voltage is lower than the upper
limit, it will decrease λy , at most until it reaches zero. A similar procedure
applies to λy . The parameter α determines how large the updates to the
control signals will be.
2. Actuator agents receive updates of λy and λy periodically. They will
adjust their compensation to take new values of the control signals into
account.
3. The voltage changes due to the actions of actuator agents. Observer
agents update again their λy and λy , and the whole process repeats.
CONTROL OF THE RAPID PROTOTYPING INVERTER WITH GRID VOLTAGE SUPPORT FUNCTION
79
The communication from observer to actuator takes place through D2D

communication modules, while the feedback path goes through the
electrical network.
From this explanation, it is clear that this process is based on feedback. As long
as the voltage problem persists, observer agents will increase control signals to
get more compensation from actuator agents. The effect of α is similar to a
gain in control theory. The trade-off in its selection is similar: a low value can
lead to slow convergence, while a too large value can lead to instability.
4.4 Control of the Rapid Prototyping Inverter with

Grid Voltage Support Function
4.4.1 PM15FM30C Triphase Module
DC/AC (Direct Current/Alternate Current) PM15FM30C Triphase rapid

prototyping inverter modules are used in the testbed to emulate prosumers with
PV installations. A schematic diagram of the PM15FM30C circuit is depicted
in Figure 4.4. The PM15FM30C module mainly consists of:
1. A 15 kVA three phase inverter, consisting of three half-bridges with

insulated-gate bipolar transistors (IGBTs).
2. A rectifier that can be connected directly to the AC voltage of the
microgrid.
3. An inductor-capacitor-inductor filter (LCL filter).
4. Three bypass resistors to limit the inrush current at the beginning of
operation; these resistors are bypassed with a relay when the rapid
prototyping module is running.
5. Current sensors to measure the current before and after the LCL filter.
6. Voltage sensors to measure the DC bus voltage and the AC voltage after
the LCL filter at the grid side.
7. Control board to drive the IGBTs, control the switches K1-K6, and the
fan of the module.
The PM15FM30C inverter is programmed and operated through MAT-

LAB/Simulink running on a computer. The computer communicates over
LCL Filter
Current sensors Current sensors IGBTs Rectifier
Voltage sensors
Bypass resistors
DC Voltage sensor DC Bus
Figure 4.4: Circuit diagram of Triphase PM15FM30C rapid prototyping inverter.
Ethernet with an on-board PC-based Real Time Target (RTT), which controls
the Triphase power electronics as shown in Figure 4.5. Python has been used to
code a software that manages the interface with D2D communication modules,
fetches the Lagrangian multipliers from D2D modules, stores the PV profiles,
and solves the quadratic optimization problem (4.1a)-(4.1d). The software also
manages the interface with MATLAB.
MATLAB exposes an interface to the Python software, which allows the Python
software to directly execute scripts in MATLAB. The Python software uses a
(k) (k)
MATLAB script to push updates on the PV profiles, and ∆pinvi and ∆qinvi
setpoints to Simulink. MATLAB also manages the interface with the RTT to
control the switches and fan of the PM15FM30C.
4.4.2 Drive of PM15FM30C Triphase Module
The P2P voltage control algorithm represents a high level control system to
coordinate inverters in a distributed way. For the inverter to be able to follow
the regulation of the P2P control algorithm, an internal control system has
to be implemented and integrated with the GVSF. We have implemented a
state-of-the-art current control loop, Phase-Locked-Loop (PLL), and Kalman
filter to drive the inverter. The inner control system of the inverter is depicted
in Figure 4.6.
The inverter uses current-mode control to control the active and reactive power.
The line current is tightly regulated by the current control loop, through the
inverter AC-side terminal voltage. Then, the active and reactive power are
controlled by the phase angle and the amplitude of the inverter current with
respect to a rotating frame that is synchronized with the point of common
CONTROL OF THE RAPID PROTOTYPING INVERTER WITH GRID VOLTAGE SUPPORT FUNCTION
81
(7)
Interface with D2D modules

PV Profiles
USB cable
Python GVSF (6)
Triphase Interface with MATLAB
PV Profiles Decisions of GVSF

Interface with python
MATLAB
Interface with RTT
Interface with SIMULINK
Non-real time PV Profiles Decisions of GVSF

network Simulink Inner control system (4)
Real time network

RTT
(12)
(1)
Figure 4.5: Actuator agent setup (labels same as in Figures 4.1 and 4.2).
coupling (PCC) voltage using PLL. A Kalman filter is placed in front of the PLL
in order to ensure that the PLL input at all times matches an ideal sinusoidal
waveform as closely as possible, even when the voltage is highly distorted by
the presence of harmonics. This ensures fast and low distortion operation of
the PLL. Kalman filter is used in this work because it efficiently deals with the
uncertainty of tuning its parameters.
The reference setpoints of active and reactive power are calculated based on
the PV profiles and the decisions of the GVSF as shown in Figure 4.6; then
the reference setpoints of active and reactive power are converted into d-q
(direct-quadrature) reference setpoints of the three phase current, and these
d-q setpoints are used by the current control loop as a reference to control the
d-component and q-component of the three phase current in order to follow the
reference setpoints of active and reactive power.
The details of the design and tuning of Kalman filter are presented in [148].
Chapter 8 in [149] presents details of the design and tuning of proportional-
integral (PI) controllers used by the current control loop and PLL. The d-q
transformation (of the current, voltage and power) can be also found in the
same chapter.
It is worth mentioning that we have not used a voltage control loop, because
the voltage of the DC bus is fixed by the DC source. For the inverter to be able
to inject power from the DC side to the AC side, the DC bus has to be charged
LCL Filter
DC Bus Label(1) 2.3 mH 0.93 mH

Aa
Ab
Vdc
Ac
PWM
10 Vabc
PCC
Iabc Fundamental+Harmonics
Vabc
abc
wt w
Kalman Filter
dq
Fundamental
Vabc
PWM gating pulses wt abc
dq
ma , mb , mc Id Iq Vd Vq
Vq
VDC Current control loop wt PLL
PI controller PI controller
f
Idref Iqref
Saturation block
P ref Idref
Q ref Iqref
P ref Q ref
Inner control system
PV profile + (4)
+
(k) (k)
p inv q inv
Label(7) Actuator agent
GVSF (5)
Lagrangian multipliers (6)
A Current meter. VDC Voltage of the DC bus. f Frequency

Current of phase a, phase b and phase c, w Angular frequency
Iabc
V Voltage meter.
before the filter. wt Angle
d-component and q-component of Iabc , Pulse width
Id , Iq PWM
modulation
respectively.
PCC Point of Common Coupling.
m a , mb , m c
Reference setpoints of d-component and
Id , Iq
ref ref
Voltage of phase a, phase b and phase c, q-component of Iabc , respectively. Modulation indices related
Vabc to phase a, phase b and
after the filter.
Reference setpoints of active and reactive phase c, respectively.
d-component and q-component of Vabc , P ref ,Q ref
Vd ,Vq power, respectively.
respectively.
Figure 4.6: Inner control system of the inverter integrated with the GVSF
(labels same as in Figures 4.1 and 4.2).
to a DC voltage higher than 650.6 V, which is the peak-to-peak

√ voltage (phase
to neutral) of the grid connection (Vpp = 2 × 230 × 2 = 650.6).
DEVICE-TO-DEVICE COMMUNICATION 83
4.5 Device-to-Device Communication3
4.5.1 Background
Device-to-device communications typically refer to cellular communications

technologies enabling direct transmission between proximate devices, without
relaying information through the cellular base station [152]. However, D2D
communications is not the exclusive domain of cellular networks, and generally
relates to the ability of peer devices to directly communicate with one
another without having to relay the actual data through a central coordinator
device, as e.g., used in [153, 154]. The paper [152] presents a survey of the
current state of the art for cellular D2D communications and points out that
cellular D2D communications are much more efficient than communications
on unlicensed spectrum as the communication interference is controllable at
the licensed spectrum. The paper categorizes cellular D2D communications
into four categories based on the level of control the base station has
on them. The first category is device relaying with base station assisted
controlled link. Here the base station allocates the channel resources for
user equipment communications so that user equipment in poor coverage can
maintain connectivity with the network. Direct communication between devices
with base station assisted controlled link is the second category, where user
equipment exchange data directly and some of these features have already
been standardized by 3rd Generation Partnership Project (3GPP) in release
13 proximity services [155], and in release 14 for cellular vehicle-to-everything
(C-V2X) communications [156]. The C-V2X uses out-band communications as
the actual data communications occur at the intelligent transportation system
(ITS), and operates in the licensed radio frequency band and not in the cellular
bands. The third category is relaying device with device assisted controlled
link, where the user equipment communicate with one another using relays and
without base station control. The fourth category is direct D2D with device
assisted controlled link, where the user equipment communicate directly with
one another without base station provision of control links. The paper [157]
proposes a solution combining categories three and four for smart grid demand
response scenarios for increased resiliency of smart grid operations.
The D2D communications required for P2P voltage control need not be
based on cellular technologies and currently, no commercial-of-the-shelf cellular
D2D chipsets are available. The key criteria for the selection of appropriate
communication technology to adopt arise from the distances and placement of
3 This section describes work not done by the author; the work is presented for completeness.
Konstantin Mikhaylov and Jussi Haapola (Centre for Wireless Communications, University of
Oulu) have provided the materials of this section [150, 151].
the observer and actuator agents. Common unlicensed band communication

technologies like the IEEE 802.11 family (WiFi) or the IEEE 802.15.4 family
(low rate personal area network) can be utilized if the distances between agents
are within a few hundred meters, and they have been installed outdoors or inside
buildings near the exterior walls. Even then mesh type network where devices
communicate in ad hoc fashion are required to ensure reliable connectivity. In
other cases, low power wide area (LPWA) communication technologies need to
be utilized. The work in [158] provides a survey on LPWA networks and claim
that they represent a novel communication paradigm, which will complement
traditional cellular and short range wireless technologies in addressing diverse
requirements of Internet of Things (IoT) applications. This applies to smart
grids in particular as LPWA technologies offer unique sets of features including
wide-area connectivity for low power and low data rate devices, not provided by
legacy wireless technologies. As an example, [154] proposes a gateway assisted
D2D communications solution utilizing LoRaWAN technology, and the work is
a basis for the D2D communications scheme used in this chapter. The work in
[154], and the communications solution of this chapter are not the same though;
the work in [154] is similar to the second category whereas the solution applied
in this chapter is similar to the fourth category of cellular D2D communications,
both using LoRaWAN technology.
4.5.2 D2D Communication Modules
The D2D communication modules are implemented based on a modular

WSAN/IoT platform (wireless sensor and actuator network/Internet of Things)
[159]. Each module is composed of three submodules stacked on top of each
other, as shown in Figure 4.7. The radio submodule (the top submodule)
hosts the RN2483 LoRaWAN radio transceiver. The main submodule (the
middle submodule) includes the microcontroller, the power circuitry, and other
peripherals. The USB submodule (the lower submodule) hosts an FTDI USB-
UART chip4 . Additionally, each D2D module needs to have an 868 MHz SMA
(SubMiniature version A) antenna. Also a mini or micro USB cable (any of
these two, but only one at a time) should be connected to the USB submodule
to interface with the agents (actuators and observers). The power required
for the module’s operation is also provided via the USB interface (maximum
consumption is in the order of 200–300 mW).
4 FTDI: future technology devices international (semiconductor device company), USB:
universal serial bus, UART: universal asynchronous receiver-transmitter

Power switch Voltage Antenna

switch 868 MHz
Stack Stack
on on
bottom top
Mini USB
connector (gives Reset Switch
power and acts
as data
interface
Fully assembled devices
during testing
USB-UART
USB UART Main UART Radio
submodule
cable board submodule
[FTDI VCP]
Figure 4.7: Structure of the D2D communication module (USB: universal

serial bus, UART: universal asynchronous receiver-transmitter, FTDI: future
technology devices international (semiconductor device company), VCP: virtual
communication port).
Implemented Embedded Firmware
The application software is written in C and operates on top of the FreeRTOS

embedded operation system. The software has been developed using Eclipse, and
compiled with GNU Compiler Collection (GCC) for Advanced RISC Machine
(ARM) processors. The high level structural diagram of the embedded firmware
is depicted in Figure 4.8, and it is composed of the three threads: main thread;
radio thread; and UART thread. The main thread is initialized after the basic
initialization procedures (setting clock, checking the module and configuring
the peripherals, blinking LEDs). The main thread initializes the UART thread
for communicating with the physically connected agent, the radio thread for
controlling the radio transceiver, and the server data structure for storing the
data from the agents. The UART drivers are implemented based on direct
memory access (DMA) and use a timer to detect end of a packet. Due to this
reason, agents should enable for at least a 5 ms idle time between the sequential
UART packets.
The server data structure is implemented as a table listing the identifiers of the
agents and the most recent data from them. The server structure is accessed
and can be modified by either the radio thread or the UART thread. The D2D
module can be configured to periodically report the complete table (i.e., the
data from all other agents) to its agent.
The developed firmware implements a multi-stage error detection and correction
system. In case of non-critical errors (e.g., wrong format of UART commands
from the physically connected agent), the module recovers automatically. In
case of severe mistakes (detected by the software or if the software hangs), the
module reset procedure is initiated. After reset, the most recent state of the
module is recovered.
Initializations
Serial over USB interface to

Serial interface to radio
power agent
Main thread
Radio thread UART thread
Read ID & mode from Flash
Initialization Initialization
Initialize server & other

threads
Listen UART & forward packets
Operate according to a
between main thread & radio
synchronized protocol
thread
Commands If periodic report enabled – Commands
periodically send table to UART
Critical Critical
errors errors
Critical errors
Read Commands
Data
Data from packets Data from packets

Server data structure
Reboot
Figure 4.8: Structure of the embedded software.
Synchronized Protocol
The radio thread handles control over the radio transceiver and implements
a synchronized radio protocol. The synchronized protocol is a simple slotted
protocol, where each of the D2D modules is assigned a periodic time slot for
transmission of its data and receiving the transmissions from the other modules
in their respective slots, as shown in Figure 4.9.
The parameters of the protocol, namely the number of slots (M) in the
superframe and the duration of each slot (T-slot) are hardcoded in the firmware
and cannot be changed without reprogramming the module. Each module
uses for its transmission the slot with the number equaling to its programmed
identifier (i.e., a module with ID 1 will send in slot 1, etc.). Empirically it was
found out that the need of using low-speed UART interface between the main
module and the radio transceiver chipset and the slow operation of the chipset
itself introduces substantial overheads (e.g., packet transmission, switching
between transmit and receive, etc.). Due to this reason, the duration of one slot
cannot be set below 150 ms.
When enabled the first time, the D2D module based on this protocol first
scans the radio channel for several superframe periods. If it does not find
any transmissions and it has data to send, it will start the transmission right
away. If during scanning a module finds some transmissions ongoing, it will
use this transmission as a reference for defining its designated slot. After each
superframe, a module adjusts its synchronization. As a reference point for
adjusting the synchronization, each module uses the timestamp of the packet
with minimum identifier not exceeding the identifier of the module. If such a
reference is not available, no compensation is applied. As a practical example,
module 1 transmitting in slot 1 never adjusts its synchronization. If modules 2,
3 and 4 hear transmission of module 1, they will adjust their synchronization
based on it. If module 5 does not hear module 1 but hears modules 2 and 3, it
will adjust its synchronization based on the transmission of module 2.
T-superframe
T-slot T-slot
slot 1 slot 2 slot 3 slot M slot 1
Time
Module 1
TX RX TX
Time
Module 2
RX TX RX
Time
Figure 4.9: Illustration of the implemented synchronized protocol operation

(TX: transmission, RX: reception).
4.6 Results of the Experiment
To test the performance of the P2P distributed voltage control system, one
needs to create a voltage rise (or drop) problem and solve it in a P2P fashion. To
create a voltage rise problem in a laboratory-based microgrid, a high-power
injection from the inverters back to the grid can be used. Alternatively, the
impedance of the feeder depicted in Figure 4.2 can be oversized to create such
a problem with low-power injection. In the following experiments, R1 and R2
(of the microgrid) are set to 8 Ω, L1 and L2 (of the microgrid) are set to 5 mH.
Figure 4.10 shows the generation profile applied at both inverters. The active
power generation starts at zero, and increases to a maximum of 1200 W. At the
higher generation, the voltage is expected to rise above the maximum voltage
limit. To comply with the European standard EN 50160, voltage limits v and v
are enforced to be ±10% of the nominal phase voltage.
Two experiments are carried out to compare voltage profiles with and without
voltage control. The comparison helps in quantifying the performance of the
P2P voltage control.
Ac�ve power [W]
Time [s]
Figure 4.10: 30 min generation profiles of inverter 1 and inverter 2 (both

inverters apply the same generation profile).
4.6.1 First Experiment: Without P2P Voltage Control
The generation profile described by Figure 4.10 is applied at both inverters of

the setup. Figure 4.11 shows that this leads to voltages exceeding the upper
limit of 1.1 p.u. at both the first and second node. The agents remained idle
during this experiment.
RESULTS OF THE EXPERIMENT 89
node 0
node 1
node 2
Maximum voltage
Voltage [V]
Time [s]
Figure 4.11: Voltage profiles without voltage control for the 3 nodes of the
microgrid (the generation profile causes over-voltages up to 1.145 p.u.)
4.6.2 Second Experiment: With P2P Voltage Control
The inverters apply the same generation profile, but now the agents execute the
distributed voltage control algorithm. This leads to the voltage profile shown
by Figure 4.12. When an increase in generation causes an over-voltage issue,
the agents bring voltages back to the defined limits (±10%) within 3 min.
node 0
node 1
node 2
Maximum voltage
Voltage [V]
Time [s]
Figure 4.12: Voltage profiles with voltage control for the 3 nodes.
The actions of the observer and actuator agents are reflected in Figure 4.13.
The evolution of the control signals over time are presented in Figure 4.13c,
and 4.13d. The control signals for under-voltages (λ) are zero, because no
under-voltages beyond the limits occur during this experiment. The control
signals for over-voltages (λ) however, increase sharply after an increase in the
voltages above the maximum voltage limit. One can notice that the control
signals λ return back to zero when the voltages return back to normal values
without compensation, due to a decrease in the generation profiles. Figure 4.13a,

and Figure 4.13b show the actions taken by the actuator agents. As soon as
an over-voltage occurs, nearly all reactive power is dispatched. This behaviour
depends on the values of α, cp,invi and cq,invi . The step size α controls mainly
how fast the control signals will increase, and hence how fast compensation is
dispatched. Since the penalty factor of active power is set to be a lot higher
than the penalty factor of reactive power, the algorithm will dispatch first the
available reactive power. α is set high enough to get a fast response in the
active power dispatch. However, this causes the dispatch of reactive power to be
nearly instantaneous. ∆pinvi and ∆qinvi return back to zero when the voltages
return back to normal values without compensation.
inverter 1 inverter 1
inverter 2
[Var]
inverter 2
Δpinv[W]
Δq inv
350 700 1050 1400 350 700 1050 1400

Time [s] Time [s]
(a) (b)
x
node 0 node 0
node 1 node 1
node 2 node 2
λ λ
350 700 1050 1400 350 700 1050 1400

Time [s] Time [s]
(c) (d)
Figure 4.13: (a) Active power curtailment of inverter 1 and inverter 2; (b)
Reactive power compensation of inverter 1 and inverter 2; (c) Control signals
for over-voltages; (d) Control signals for under-voltages.
When the measured voltage vymeas is less than v, λy starts to decrease till it
reaches zero. One can explain this based on (4.2). The Lagrangian multipliers
drop back to zero because the underlying profile of the inverters change. The
active power injection drops, and the voltage drops with it. The Lagrangian
multipliers adapt to the new situation. When both λy and λy are zero at each
observer, the optimization problem (4.1a)-(4.1d) can be written as:
2 2
(k) (k)
argmin cp,invi ∆pinvi + cq,invi ∆qinvi + zero
(k) (k)
∆pinv ,∆qinv
i i (4.3)
Subject to the local constraints (4.1b)-(4.1d)

(k)
One can notice that the solution of the above optimization problem is: ∆pinvi = 0
(k)
and ∆qinvi = 0. Hence, a stop mechanism can be designed to stop the solver of
the optimization problem whenever the Lagrangian multipliers are zero at each
observer. This should decrease the computational burden of the algorithm.
4.6.3 Key Performance Indicators
There are three key performance indicators (KPIs) considered in this work:
convergence time; voltage quality; and communication delays.
The first KPI, convergence time, is a measure for how long it takes the algorithm
to solve the voltage problem. Voltage quality reflects how well the control
algorithm can mitigate the voltage rise (or drop) problems. Finally, the
communication delays depend on the communication infrastructure. Below
follows an explanation of how each of these KPIs is quantified in practice.
Convergence Time
The voltage control algorithm is online and adjusts itself continuously. When
a change in the generation profile occurs, there are two possibilities: either
there is a voltage problem or not. If there is no voltage problem, the control
algorithm stays idle. However, if there is a voltage problem, then the agents
start to undertake action. The observer agents change the control signals until
the voltage problems are resolved. If they succeed, then the control signals
converge to a stable value, and the voltages converge to a value within the
limits. In this chapter, we define the convergence time as the time it takes
from a moment when the voltage exceeds the limits until the moment when the
voltage is restored within the limits. As demonstrated in Figure 4.12, it takes
the algorithm around 3 min to regulate the voltages within the defined limits,
which is an acceptable time for voltage problems.
It is worth mentioning that the intervention time of the interface protection relay
of Triphase inverter is much less than the convergence time. The intervention
time of the interface protection relay of Triphase inverter is less than 1 ms.
The Triphase inverter is configured to trip at 280 V. This means that there
is 27 V as a voltage margin, since the algorithm starts regulating the voltage
when the PCC voltage is higher than 253 V. Hence, the inverter in our setup is
able to correct the voltages before reaching 280 V. If an inverter trips at 253 V
(maximum voltage defined by the standard EN 50160), then v of the proposed
algorithm should be set to a value lower than 253 V (i.e., 240 V), in a way to
make sure that the convergence time is sufficient to correct the PCC voltages
before reaching the maximum voltage at which the inverter trips.
Voltage Quality
The voltage quality is quantified by the metric E ≥ 0 defined by Equation

(4.4). The metric E integrates the over and under voltages as shown in Figure
4.14. This means that both the duration of a voltage problem and its severity
will increase the metric E. A value of zero is the best possible value and indicates
that there are no over or under voltage issues: the higher the E, the worse the
voltage problem.
XZ tend
E= max vymeas − v , 0 + max v − vymeas , 0 dt (4.4)

y∈Y tstart
Maximum E
voltage
Minimum
Vymeas
voltage
t
Figure 4.14: Voltage quality metric: sum of the surfaces above and below the
voltage limits.
Table 4.1 shows a comparison between the regulated and the unregulated voltage
profiles based on the voltage quality metric. E is the sum of the metrics E0 , E1
and E2 of the nodes 0, 1, and 2, respectively. The P2P voltage control reduced
the metric E from 58.724 to 2.633. E of the regulated voltage profiles is slightly
higher than zero, because it takes the algorithm some time until it has resolved
the voltage issues.
Communication Delays
For the observer agents, the delay is defined as the time between consecutive
updates of their control signals, which they broadcast periodically to the
actuator agents. For the actuator agents, the delay is defined as the time
between consecutive updates of the setpoints which are sent to the Triphase
power hardware. Figure 4.15 shows the delays between the iterations of the
control algorithm, for each agent individually.
Table 4.1: Voltage quality metrics of the regulated and unregulated voltage
profiles
Ey
vymeas E
Node 0 Node 1 Node 2
without control 0 19.719 39.005 58.724
with control 0 0.081 2.552 2.633
Figure 4.15: Communication delays: The observer agents update the control
signals every 1.5 s, with very little deviation. The actuator agents issue their
updates more slowly, with a significant difference between both actuator agents.
The observer agents are implemented by dedicated single-board computers with

few other processes running in the background. They manage to update the
control signals every 1.5 s, with little deviation. The actuator agents however
experience longer control delays, with large differences between both actuators.
There are two main causes for these additional delays. Firstly, the actuator
agents solve an optimization problem at each iteration. Secondly, the actuator
agents are implemented by laptops. These laptops run additionally control
software for the Triphase Rapid Prototyping Inverter System, which requires
rather heavy processing. The laptop running actuator 2 is older, which shows
in the performance. Adapting the implementation of the algorithm for the

actuator agents can lower the delays. The lower limit for the delays is 1.5 s,
which is the period with which the observer agents send updates of the control
signal.
Overall, the delays are as expected. Only the delays for actuator 2 could be
shorter to be in line with the other devices. Upgrading actuator 2 to hardware
similar to actuator 1, should resolve these additional delays.
4.7 Conclusion
The testbed presented in this chapter provides realistic and pragmatic solution
for evaluating P2P smart grid applications. The testbed is used to evaluate
the performance of the P2P voltage control system experimentally. It can also
be used to implement and evaluate other P2P systems. For example, a dual-
decomposition-based P2P trading algorithm can replace the dual-decomposition-
based voltage control algorithm, to study the performance of a P2P trading
system.
Chapter 5
Robust Policy-Based
Distributed Voltage Control
Provided by PV-Battery
Inverters
In this chapter, we show how forecast data of active power consumption

and photovoltaic power generation can be used to simplify the design and
implementation of a distributed coordinated voltage control system. To this
end, this chapter proposes a distributed voltage control system enabled by
robust linear policies. The chapter focuses on voltage control provided by
photovoltaic-battery inverters.
The proposed voltage control consists of an offline coordinator and real-time
policy-based distributed voltage control. In the offline coordinator, the voltage
control problem is formulated as a chance-constrained stochastic optimization
problem, where the control actions are expressed as linear function of the
uncertain households’ active power consumption and photovoltaic power. Robust
optimization technique is applied to solve the optimization problem. In real-
time, the inverters communicate with each other and apply the robust policies
to compute the amount of reactive power and battery power needed to maintain
voltages within limits. A 62-node real low voltage network is used as a case
study. Simulation over 36500 scenarios is used to demonstrate the robustness
of the proposed voltage control system. The contents of this chapter are based
upon the journal paper:
95
96 ROBUST POLICY-BASED DISTRIBUTED VOLTAGE CONTROL PROVIDED BY PV-BATTERY
INVERTERS
• H. Almasalma, and G. Deconinck, “Robust Policy-Based Distributed

Voltage Control Provided by PV-Battery Inverters,” IEEE ACCESS, vol.
8, pp. 124939-124948, 2020.
5.1 Introduction
Renewable power capacity is forecast to increase by 50% between 2020 and

2024 [160]. The market for battery energy storage systems is expected to grow
at a compound annual growth rate of approximately 50% during the period
2020-2025 [161]. This transition towards distributed energy resources (DERs)
is impacting the operation of distribution networks [4]. Distribution networks
were originally designed as passive networks, containing mainly loads. Therefore,
a high penetration of DERs leads to several technical challenges in controlling
distribution networks. One of the main problems is voltage rise issue due to
the feed-in power from DERs.
Fortunately, DERs in general and photovoltaic (PV) inverters in particular
offer different technical features, such as reactive and active power control.
Applying these features properly can positively influence the grid voltage.
To accomplish this, new control algorithms have to be developed that are
able to coordinate DERs and allow them to participate in voltage control by
efficiently using their active and reactive power control capabilities. As has been
discussed in the previous chapters, coordinated voltage control (CVC) methods
of different complexity and data transfer needs have been proposed in literature
[38]. The techniques for CVC can be categorized based on communication and
computation requirements as centralized and distributed [21, 26].
As the conventional networks are shifting toward DERs, the technology to control
the networks must also adapt. Since DERs are typically highly distributed,
operated by different owners and with different objectives, it is desirable that
CVC operates in a highly distributed way as well. Besides, a robust control
system is needed that does not depend on a single point of failure, as most of
the more centralized control methods do. Moreover, centralized control might
not be able to deal with the high computational demand of voltage control
systems with a large number of DERs.
Distributed coordinated voltage control (DisCVC), as an alternative approach to
solve challenges of a centralized coordination, has attracted increasing attention
recently. However, the task of implementing DisCVC is more complex than
that of implementing the centralized one. DisCVC algorithms appearing in
literature are mainly designed based on distributed constraint optimization
algorithms (DisCOP) [23]. DisCOP makes it possible for DERs to cooperatively
INTRODUCTION 97
solve a voltage problem through local computations and peer-to-peer (P2P)

communications. DisCOP relies on iterative decomposition techniques, such
as dual decomposition and the alternating direction method of multipliers
(ADMM) [162, 27]. Hence, one of the main challenges of using DisCOP for
distributed coordinated voltage control is to reduce the number of iterations (as
much as possible) to solve a voltage problem within a reasonable time. Most of
DisCOP algorithms proposed in literature require a large number of iterations
to converge and multiple communication exchanges [28]. Hence, these proposed
algorithms fail in regulating voltage profiles within a reasonable time. The
gossip-based JP-ADMM algorithm, presented in Chapter 3, succeeds in speeding
up the convergence of DisCOP used for distributed coordinated voltage control.
However, the practical implementation of this algorithm is complex, as the grid
voltage support functions need to a solve (in real-time) a complex optimization
problem every time step, and the tuning of the penalization and acceleration
factors is not straightforward.
To reduce the computational and communicational burden of a DisCVC system
and to design a real-time DisCVC that can be implemented easily, this chapter
proposes a DisCVC enabled by robust linear policies computed on a day-ahead
basis. The chapter focuses on DisCVC provided by PV-battery inverters. The
control actions considered in this chapter are reactive power from inverters and
charge/discharge power from batteries. Two types of policies are considered:
reactive power policy that gives reactive power of each inverter; and battery
power policy that gives charge/discharge power of each battery. The idea is
to let PV-battery inverters solve, each time step in real-time, a set of linear
equations rather than solving a complex optimization problem.
Related Work
Based on our experience with coordinated control, using real-time distributed

optimization to coordinate several inverters distributed over a feeder is a very
difficult task. As a middle-ground solution, a centralized coordinator can be
run on an offline basis (i.e. day-ahead) to compute voltage control policies.
A DisCVC can then be applied in real-time enabled by the computed control
policies.
In [163, 164, 165, 166], different techniques are used for computing linear voltage
control policies: chance constrained programming in [163]; distributionally
robust chance constraints in [164]; affinely adjustable robust optimization in
[165]; and conic robust optimization in [166]. The aforementioned control policies
are designed for local voltage control, they are not meant for coordinated voltage
control. This chapter builds on the techniques presented in [163, 164, 165, 166],
INVERTERS
but extends the voltage control policies to be used for enabling real-time
distributed voltage control. To the best of our knowledge, this work is the
first work proposing the use of linear policies for enabling real-time distributed
coordinated voltage control.
The remainder of the chapter is organised as follows: Section 5.2 presents a
general overview of the proposed voltage control system. The optimization
problem treated in this chapter is formulated in Section 5.3 based on a robust
optimization technique and linear models of control policies, voltages, inverters
and batteries. Section 5.4 demonstrates the robustness of the proposed voltage
control system and its ability to solve voltage problems. Finally, the chapter is
concluded in Section 5.5.
5.2 The Proposed Voltage Control System
As shown in Figure 5.1, the proposed voltage control system consists of a

robust offline coordinator and a policy-based DisCVC. In the offline coordinator,
the voltage control problem is formulated as a robust optimization problem,
where the control actions are expressed as a linear function of the uncertainty.
The uncertainty includes households’ active power consumption and PV power
generation. Constraints related to voltages, inverters and batteries are included
as probabilistic constraints (chance constraints).
The aim of the robust optimization problem is to compute a set of linear control
policies that minimize the expected cost of operating the voltage control system
and guarantee (with a predefined probability) that voltages will not exceed
pre-defined bounds and inverters/batteries constraints are respected.
It is well known in voltage control that designing a control policy based only on
local information yields poor performance [26]; this is because the voltage of
each network’s node depends on its power injection and the power injections of
the other network’s nodes. Therefore, the proposed voltage control policy of
each inverter is based on local (load and PV) measurements and measurements
of the other control nodes. Each PV-battery inverter needs to know the sum
of the households’ active power consumption and PV power generation of the
nodes participating in voltage control to be able to compute its own control
actions. To compute the total PV power generation and households’ active
power consumption in a fast and efficient way, a P2P-based push-sum gossip
protocol is applied in this chapter [90]. The push-sum gossip protocol enables
the PV-battery inverters to perform a distributed summation with a moderate
communication overhead.
DESIGN OF THE ROBUST OFFLINE COORDINATOR 99
PV/load forecast Control policies to

Offline Coordinator
Maximum forecast error other systems
Control policies
Day-ahead
Real time control
p p
PV Grid voltage invi i
ppv support function q q
i invi i
Measurement
point Measurement
p point
p Inverter control li
bat i q Control
li
Smart PV-Battery Inverter Node
Battery
Household Load
Control policies
Control policies
PV-Battery
PV-Battery System
System
Figure 5.1: General overview of the proposed voltage control system.
In the policy-based DisCVC, the linear control policies are deployed at each
individual PV-battery inverter. Once the uncertainty is revealed, the inverters
communicate with each other and apply the push-sum gossip protocol to compute
the sum of the PV power generation and households’ active power consumption
of their nodes. Each inverter then applies its control policy to compute the
reactive power and battery power needed to maintain voltages within accepted
limits.
5.3 Design of the Robust Offline Coordinator
This Section presents the optimization problem the offline coordinator solves
on a day-ahead basis.
INVERTERS
5.3.1 Problem Formulation
The objective of the offline coordinator is to compute reactive power and battery
power control policies that minimize the expected cost of operating the voltage
control system while keeping voltage profiles within accepted limits; inverters
within their capacity constraints; and batteries within their energy and power
constraints. The expected cost of operating the voltage control system is the
sum of two costs: expected cost of the reactive power compensation; and
batteries degradation cost. As we want to make sure that the risk of breaching
the constraints (of voltages, inverters and batteries) is as small as possible,
the probability of violating these constraints is set on the order of 10−4 . This
results in the following stochastic optimization problem with chance constraints
(probabilistic constraints):
" #
X 2 2
min E c2qi ∆t2 kQinvi k2 + c2pi ∆t2 kPbati k2 (5.1a)
i∈I
subject to Qinvi , Sinvi , Pbati , Ebati ∈ Rnt ,
V2 ∈ Rnt ×nnodes , ∀i ∈ I :
Pr v 2 6 V2 6 v 2 > 1 − (5.1b)

Pr pbat 6 Pbati 6 pbati > 1 − (5.1c)
i
Pr ebati 6 Ebati 6 ebati > 1 − (5.1d)

Pr sinvi 6 Sinvi 6 sinvi > 1 − (5.1e)

Here, E[·] denotes the expected value operator. nt is the number of time steps
with duration ∆t (K is the time steps set). nnodes is the number of network’s
nodes. We consider two types of nodes in this chapter: control nodes where
the PV-battery inverters and loads are connected; and passive nodes with only
loads. Control nodes belong to the set I, passive nodes belong to the set J and
network’s nodes belong to N .
h iT h iT
(1) (n ) (1) (n )
Qinvi = qinvi , · · ·, qinvti , Pbati = pbati , · · ·, pbatt i denote reactive power
(k)
vector of inverter i and battery power vector of battery i, respectively. qinvi
(k)
and pbati are respectively the reactive power value (of inverter i) and active
power value (of battery i) at time step k ∈ K. The objective function (5.1a)
minimizes the expected total quadratic cost of reactive power injection and
battery degradation cost. cqi and cpi in (5.1a) are respectively the reactive
power compensation cost (cent/kvarh) and battery degradation cost per kWh
(cent/kWh). As shown in subsection 5.3.7, the expected value of the objective
function is approximated using a second-order moment matrix. The constraints
(5.1b)-(5.1e) are the chance constraints and is the probability of violating
these constraints.
The constraint (5.1b) is used to maintain nodal voltage magnitudes within
the maximum voltage limit v and the minimum voltage limit v. V2 =
T
(v ) , · · ·, (v(nht ) )2 denotes vectori of squared nodal voltage magnitudes,
(1) 2
(k) (k) (k) (k)
where (v )2 = (v1 )2 , · · ·, (vnnodes )2 . vx is the voltage magnitude of node
x ∈ N at time step k ∈ K. Computation of voltages is presented in subection
5.3.4.
The constraints (5.1c), (5.1d) are used to keep batteries within their energy
h iT
(1) (n )
and power limits. Ebati = Ebati , · · ·, Ebatti is the energy content vector of
(k)
battery i, where Ebati is the energy content of battery i at time step k ∈ K.
ebati and ebati are respectively the maximum and minimum energy capacity of
battery i. pbati and pbat are the maximum and minimum power of battery i,
i
respectively. The computation of the battery energy content, at different time
steps, is discussed in subsection 5.3.6.
The constraint (5.1e) is used to keep inverters within their capacity limits.
h iT
(1) (n )
Sinvi = Sinvi , · · ·, Sinvti is the apparent power vector of inverter i, where
(k)
Sinvi is the apparent power magnitude of inverter i at time step k ∈ K. sinvi and
sinvi are respectively the maximum and minimum apparent power magnitude
of inverter i. The inverter capacity model is presented in subsection 5.3.5.
The models, approximations, and reformulation needed to solve the optimization
problem (5.1a)-(5.1e) are presented in the following subsections. Trading off
modelling accuracy against computational tractability, we resort to linearized
models. Robust optimization is then used to construct a tractable optimization
problem.
5.3.2 Uncertainty Model
The sources of uncertainty in the optimization problem (5.1a)-(5.1e) include

households’ active and reactive power consumption and PV power generation.
INVERTERS
The uncertain households’ active power consumption and PV power generation

can be defined as:
(k) (k) (k) (k)

p̃lx = plx + ξlx ∆plx , ∀x ∈ N , ∀k ∈ K (5.2)
(k)
pvi = ppvi + ξpvi ∆ppvi , ∀i ∈ I , ∀k ∈ K
p̃(k) (5.3)
(k) (k)
A tilde sign is used to denote uncertain household’s active power consumption

(k) (k) (k) (k)
p̃lx and PV power generation p̃pvi . plx and ppvi are respectively the day-ahead
forecasted household’s active power consumption and PV power generation at
(k)
time step k. ∆plx and ∆p(k) pvi are the maximum deviation from the forecasted
household’s active power consumption and PV power generation at time step k.
(k) (k)
∆plx and ∆p(k) pvi can be approximated based on historical forecast error. ξlx
(k)
and ξpvi are random variables distributed in the range [−1, 1]. Since there is a
degree of correlation between households’ load forecast errors [167] and between
(k) (k)
PV forecast errors [168], ξlx and ξpvi are considered as dependent random
variables.
The uncertain households’ reactive power consumption can be defined as function
of the households’ active power consumption and power factor (PF), as shown
in (5.4).
(k) (k)
q̃lx = tan (acos (PF)) p̃lx , ∀x ∈ N , ∀k ∈ K (5.4)
The uncertainty definition (5.2)-(5.4) allows the forecast error to vary between
their minimum and maximum values, leading to conservative solutions.
Therefore, budget uncertainty is applied in this chapter to control the degree
of conservatism. As we would like to formulate the constraints (5.1b)-(5.1e)
as linear constraints, a polyhedral uncertainty set Ξ with budget Γ is used to
model the uncertainty [169].
n o
U = P̃l,pv ∈ Rnt (nnodes +nc ) |P̃l,pv = Pl,pv + I∆P ξ l,pv (5.5)
Where ξ l,pv ∈ Ξ , ∀k ∈ K :
h iT h iT
(1) (nt ) (1) (nt )
P̃l,pv = P̃l,pv , · · · , P̃l,pv , Pl,pv = Pl,pv , · · ·, Pl,pv ,
h i
(k) (k) (k)
P̃l,pv = p̃l1 , · · · , p̃ln , p̃(k) (k)
pv1 , · · ·, p̃pvn ,
nodes c
h i
(k) (k) (k)
Pl,pv = pl1 , · · ·, pln , p(k) (k)
pv1 , · · ·, ppvn ,
nodes c
h iT
(1) (nt )
I∆P = diag [∆Pl,pv ] = diag ∆Pl,pv , · · · , ∆Pl,pv ,
h i
(k) (k) (k)
∆Pl,pv = ∆pl1 , · · · , ∆pln , ∆p(k) (k)
pv1 , · · · , ∆ppvn ,
nodes c
n o
Ξ = ξ l,pv ∈ Rnt (nnodes +nc ) |kξ l,pv k1 6 Γ ,
h iT
(1) (nt )
ξ l,pv = ξ l,pv , · · · , ξ l,pv ,
h i
(k) (k) (k)
ξ l,pv = ξl1 , · · · , ξln (k)
, ξpv 1
(k)
, · · · , ξpv n
nodes c
Here, nc is the number of control nodes. The operator diag [∆Pl,pv ] creates
a square diagonal matrix with the elements of the maximum deviation vector
∆Pl,pv on the main diagonal. The vector P̃l,pv includes the uncertain households’
active power consumption and PV power generation for nt time steps. The
(k)
elements of P̃l,pv are defined in (5.2), (5.3). The households’ active power
consumption and PV power generation forecast data for nt time steps are
contained in the vector Pl,pv .
The convention in this chapter is that the power is positive when it is consumed
(k)
and negative when it is injected. In our simulations, the elements of P̃l,pv and
(k) (k)
Pl,pv are positive. Hence, a minus sign will be introduced in front of p̃pvi and
(k)
ppvi .
5.3.3 Linear Control Policies
An effective approach to reduce the complexity of the optimization problem

(5.1a)-(5.1e) is to restrict the functional form of decision rules to be linear
policies. This approach has been proposed by Ben-Tal et al. in a robust
optimization context [170]; and was later extended to stochastic programming
by Chen et al. [171] and Shapiro and Nemirovski [172]. To regulate voltage
profiles, each PV-battery inverter adjusts its reactive power and battery power
linearly as function of control nodes’ active power consumption p̃l and PV power
generation p̃pv :
INVERTERS
(k)
X (k)

qinvi = L(k)
qi p̃li − p̃(k)
pvi (5.6)
i∈I
Qinvi = Lqi P̃∆ , ∀i ∈ I, (5.7)

" #T
X (1)
X (n )

P̃∆ = p̃li − p̃(1)
pvi ,··· , p̃li t − (nt )
p̃pv i
i∈I i∈I
(k)
Where Lqi is the coefficient of the reactive power policy of inverter i at time
step k, contained in the diagonal matrix Lqi ∈ Rnt ×nt . The vector P̃∆ includes
the total summation of households’ active power consumption and PV power
(k)
generation of all control nodes at each time step. With Lpi as the coefficient
of battery power policy (of battery i at time step k) contained in the diagonal
matrix Lpi ∈ Rnt ×nt , the battery power policy can be defined as:
Pbati = Lpi P̃∆ , ∀i ∈ I (5.8)
It is worth to point out that to reduce the measurements and communication

overhead of the real-time DisCVC, the control policies are designed as a linear
function of only the households’ load and PV power of control nodes. In real-
time, the control policies do not rely on the households’ load of passive nodes,
but we do consider the effect of passive nodes’ load on voltages in the offline
optimization, as will be shown in the next subsection.
5.3.4 Branch Flow Model
In this chapter, the linear branch flow model shown in (5.9) is applied. The
model is based on the DistFlow method developed in [173]. According to [174],
the linearized branch flow model tends to introduce a small relative error of
1-5% when used for calculating power flows of real distribution networks.
(k) 2
(v ) = RP(k) + XQ(k) + vo2 1nnodes , ∀k ∈ K (5.9)
Here, vo is the substation voltage, which is considered constant. 1nnodes ∈

h iT
(k) (k)
Rnnodes is a vector of all ones. The vectors P(k) = p1 , · · · , pnnodes and
h iT
(k) (k)
Q(k) = q1 , · · · , qnnodes include, respectively, active power and reactive
power of network’s nodes at time step k. If i is a control node and j is a passive
node, then their active and reactive power can be defined as:

(k) (k) (k) 
pi = p̃li − p̃(k)
pvi + pbati 
∀i ∈ I (5.10)
(k) (k) (k)
qi = q̃li + qinvi



(k) (k)
pj = p̃lj 
∀j ∈ J (5.11)
(k) (k) 
qj = q̃lj 
The matrices R and X (∈ Rnnodes ×nnodes ) depend, respectively, on the resistance

and reactance of the network’s feeders. We ask the readers to refer to [166] for
details about the derivation of the linearized model (5.9) and the computation
of the matrices R and X.
5.3.5 Inverter Model
The magnitude of the inverter’s apparent power at time step k can be given by
the following equation:
q
(k) (k) (k)
Sinvi = (pinvi )2 + (qinvi )2 , ∀i ∈ I , ∀k ∈ K (5.12)
(k) (k) (k)

Where pinvi = pbati − p̃pvi is the inverter’s active power at time step k. The active
and reactive power passing through the inverter are limited by its capacity:
q
(k) (k)
(pinvi )2 + (qinvi )2 6 sinvi , ∀i ∈ I, ∀k ∈ K (5.13)
As shown in [165], the conic constraint (5.13) can be approximated by a number

of linear constraints nS with a relative error of 1 − cos(π/2nS ):
(k) (k)
− sinvi 6 cos(ϑφ)pinvi + sin(ϑφ)qinvi 6 sinvi (5.14)
π
φ= , ϑ = 1, · · · , nS
nS
Regarding the reactive power cost cqi (cent/kvarh), it corresponds to the

approximative compensation cost of the additional inverter losses due to the
reactive power utilization and is provided by [175].
INVERTERS
5.3.6 Battery Model
We consider a simple discrete battery model:

1 h (k) i+
h i+
(k+1) (k) (k)
Ebati = Ebati + ∆t ηi pbati − dis −pbati
ch
(5.15)
ηi
Here, ηich and ηidis are respectively the charge and discharge efficiency of battery
+
i. The operator [·] ≡ max(·, 0) introduces integer variables to the constraint
(5.1d), which results in a mixed integer chance constraint. As mixed integer
chance constraints lead to a high computational complexity, a heuristic approach
is proposed in [176] to get rid of integer variables of a battery model providing
frequency containment reserve. In this chapter, we adapt the heuristic approach
to be used for a battery model providing voltage control:
!
+ +
(k+1) (k) (k) (k)
Ebati = Ebati + ∆t pbati (k) − − pbati (k) , (5.16)
η η

(k+1) (k) (k)
Ebati = Ebati + ∆t pbati ,
η (k)
(k) (k)
(

(k)
ch (k)
ηi pbati if pli 6 ppvi
pbati = (k) (k)
η (k)
1 (k)
p
η dis bati
if pli > ppvi
i
The if condition in (5.16) is based on the fact that the controller tends to
discharge the battery when the load is higher than the PV generation (to
mitigate voltage drop) and charge the battery when the PV generation is higher
than the load (to mitigate voltage rise). Our simulation results show that this
heuristic approach does not lead toviolation of the constraint
(5.1d).
In fact,
(k) (k) (k)
(5.16) is the same as (5.15) if sign p̃li − p̃pvi = −sign pbati . To include
the initial energy content Ebat
0
i
, we can reformulate (5.16) as:
k−1
X
(k) (t)
Ebati = Ebat
0
i
+ ∆t pbati , k = 2, 3, · · · , nt (5.17)
η (t)
t=1
(1)
Ebati = Ebat
0
i
Degradation cost: The per-kWh battery degradation cost proposed in [177] is

applied in this chapter:
creplacementi
cpi = (5.18)
Ncycle life × DoD × 2ebati
Where creplacementi is the battery replacement cost, Ncycle life is the battery cycle
life and DoD is the depth-of-discharge. The denominator of (5.18) gives the
charging and discharging kWh the battery can tolerate over its serviceable life.
5.3.7 Second-Order Moment Matrix
Based on the defined linear control policies, the objective function (5.1a) can
be reformulated as:
" #
X 2 2
E c2qi ∆t2 kQinvi k2 + c2pi ∆t2 kPbati k2 = (5.19)
i∈I
X h 2
i h 2
i
c2qi ∆t2 E kLqi P̃∆ k2 + c2pi ∆t2 E kLpi P̃∆ k2
i∈I
A closed-form of the expected value in (5.19) is not readily available. Therefore,

a second-order moment matrix can be used
to approximate the expected value,
as discussed in [178]. If M = E P̃∆ P̃∆T is a second-order moment matrix of
the random vector P̃∆ , then:
h 2
i
E kLqi P̃∆ k2 = E Tr LTqi Lqi P̃∆ P̃∆T = Tr LTqi Lqi M

h 2
i
E kLpi P̃∆ k2 = E Tr LTpi Lpi P̃∆ P̃∆T = Tr LTpi Lpi M

Based on this, (5.19) can be written as:

X
c2qi ∆t2 Tr LTqi Lqi M + c2pi ∆t2 Tr LTpi Lpi M (5.20)

i∈I
Where M ∈ Rnt ×nt is assumed to be positive definite, and can be computed

based on the following formula:
M = P̃∆T,mean P̃∆ ,mean + cov P̃∆scen (5.21)

Here, the matrix P̃∆scen ∈ Rnscen ×nt contains nscen scenarios of the random vector
P̃∆ . P̃∆ ,mean ∈ R1×nt is the sample mean of P̃∆scen , and cov P̃∆scen ∈ Rnt ×nt is

the covariance matrix of P̃∆ .

INVERTERS
5.3.8 Robust Optimization
Based on the previous linear models, one can show that the chance constraints
(5.1b)-(5.1e) can be written as:
Pr APl,pv + BI∆P ξ l,pv 6 C > 1 − (5.22)

A, B ∈ R(2nt (2nc +nc ns +nnodes ))×(nt (nc +nnodes ))
C ∈ R(2nt (2nc +nc ns +nnodes ))
Where the matrices A, B and C can be defined to represent the constraints

(5.1b)-(5.1e). Robust optimization is used in this chapter to construct a tractable
approximation to chance constraints [176]. The idea is to find a solution that
remains feasible for any realization of ξ l,pv in the uncertainty set Ξ. To this
end, we consider the following robust counterpart of the constraint (5.22):
APl,pv + max BI∆P ξ l,pv 6 C (5.23)

ξl,pv ∈Ξ
By correct design of Ξ, the solution of (5.23) can ensure that the probability in
(5.22) is larger than or equal to 1 − . Unfortunately, the probability guarantee
is only applicable to independent variables with zero mean, which is not the
case when considering ξ l,pv . However, one can obtain independent variables
with zero mean by applying a whitening transformation [179]:

ξ indp = l,pvmean = W ξ l,pv
zero mean
(5.24)

l,pv W ξ l,pv − ξ
Where the vector ξ indp

l,pv contains independent random variables, ξ l,pvmean is
the mean of ξ l,pv and W is the whitening matrix satisfying the condition
W T W = cov−1 (ξ zero
l,pv
mean
). The whitening matrix can be calculated as the
Cholesky decomposition of cov−1 (ξ zero
l,pv
mean
). Based on (5.24), the constraint
(5.23) can be written as:

b + max BI −1 indp
A ∆P
W ξ l,pv 6 C, (5.25)
ξindp
l,pv
∈Ξ
b = APl,pv + BI ξ
A ,
∆P l,pvmean
n o
Ξ = ξ indp
l,pv ∈ R nt (nnodes +nc )
|kξ indp
k
l,pv 1 6 Γ
Using standard duality techniques, it can be shown that the constraint (5.25)
with a polyhedral uncertainty set can be reformulated as linear constraints
CASE STUDY 109
(5.26b)-(5.26d). We ask the readers to refer to property 4.3 in [169] for details
about robust counterpart of linear constraints with polyhedral uncertainty set.
For more robust approaches on BESSs and DERs, we ask the readers to refer to
[180]. Below is the complete optimization problem the robust offline coordinator
solves on a day-ahead basis:
X
min c2qi ∆t2 Tr LTqi Lqi M + c2pi ∆t2 Tr LTpi Lpi M (5.26a)

i∈I
subject to ∀ϕ = 1, 2, · · · , 2nt (2nc + nc ns + nnodes )
aϕ + Γ λϕ 6 cϕ (5.26b)
λϕ > I ∆P W −1 µϕ (5.26c)
− µϕ 6 bTϕ 6 µϕ (5.26d)
Where Γ = −4nt (3nc + nnodes ) ln (based on theorem 3.1 in [181] ). The

p
h iT
scalar variable λϕ and the variables of the vector µϕ = µϕ1 , · · · , µϕnt (nc +nnodes )
are auxiliary variables. aϕ and cϕ are, respectively, the ϕ-th element of the
vector Ab and C. bϕ is the ϕ-th row of the matrix B.
5.4 Case Study
5.4.1 Simulated Network
We demonstrate the robust performance of the proposed policy-based DisCVC

on a real low voltage Belgian network [182]. This three-phase semi-urban radial
distribution network operates with a nominal voltage of 230/400 V. The network
consists of 62 three-phase customers (nnodes = 62) and is depicted in Figure
5.2. All main feeder cables are of type EAXVB 1 kV 4 × 150 mm2 (impedance:
0.206+j0.0778 Ω/km), except for the cable between node I and node II, which
is of type EAXVB 1 kV 4 × 95 mm2 (impedance: 0.320+j0.0778 Ω/km). The
cables connecting each household with the feeder are of type EXVB-Cu 1 kV
4 × 16 mm2 , with a fixed length of 30 m (impedance: 1.15+j0.0828 Ω/km).
The numbered nodes in the network are control nodes (nc = 31) and the other
nodes are passive nodes. The households’ load, inverters and batteries between
II and III in Figure 5.2 belong to group G1, between II and IV belong to
group G2, between IV and V (IV and VI) belong to group G3, between IV and
INVERTERS
Group G3
16 14 15
13
Control node
i V VI
Passive node
VII
IV
7 8 9 10 11 12 17 18 19 20 21 22
Grid Group G2 Group G4
II
I
III VIII
1 2 3 4 5 6 23 24 25 26 27 28 29 30 31
Group G1 Group G5
Figure 5.2: Schematic diagram of the network used in this case study.
VII belong to group G4 and between IV and VIII belong to group G5. The
parameters of each group are listed in Table 5.1.
230 V, resulting in v= 253 V and v = 207 V. The probability of violating
voltages, inverters and batteries constraints is set to 10−4 . The number of
linear constraints ns used to approximate the conic constraint (5.13) is set to
10 (relative error = 0.0123).
5.4.2 The Load and PV Scenarios
An open-source CREST DEMAND MODEL [183] is used in this chapter to

generate one year (365 days) households’ load profiles with one-minute resolution.
The generated 365 daily load profiles for each node are considered as the
forecasted load profiles. The maximum load forecast error is set to 20 %.
For each day, we generate 100 load scenarios for each node based on eq. (5.2),
resulting in 36500 load scenarios for each node. The random dependent variables
in (5.2) are generated using the following procedure. First we generate random
independent variables. Then we compute the covariance matrix of the normalized
forecasted load profiles. Next, we apply the Cholesky decomposition to transform
the random independent variables into random dependent variables. We ask
CASE STUDY 111
Table 5.1: Parameters of inverters, batteries and loads.
G1 G2 G3 G4 G5
sinv (kVA) 5 8 10 12 15
ebat (kWh) 4.8 7.2 8.2 9.8 13.5
ebat (kWh) 0.96 1.44 1.64 1.96 2.7
Ebat
0
(kWh) 2.4 3.6 4.1 4.9 6.75
pbat (kW) 2 3 3.3 5 7
pbat (kW) -2 -3 -3.3 -5 -7
Ncycle life 3500 3500 3500 5000 5000
DoD (%) 80 80 80 80 80
creplacementi () 2400 3600 4100 4900 6750
cp (cent/kWh) 8.92 8.92 8.92 6.25 6.25
cq (cent/kvarh) 0.29 0.29 0.29 0.29 0.29

√ √ √ √ √
ch
ηi 0.9 0.9 0.9 0.9 0.9
√ √ √ √ √
dis
ηi 0.9 0.9 0.9 0.9 0.9
Load kW-peak 3.5 5.6 7 8.4 10.5
Power factor 0.85 0.85 0.85 0.85 0.85
PV kW-peak 4.8 7.68 9.6 11.52 13.44
the reader to refer to [179] for details about generating scenarios of random
dependent variables using Cholesky decomposition.
365 daily PV profiles (with one-minute resolution) are generated for each control
node, based on the model presented in [184]. The generated PV profiles are
considered as the forecasted PV profiles. The maximum PV forecast error is set
to 40 %. 100 PV scenarios for each control node for each day of the simulation
are generated based on eq. (5.3). The random dependent variables in (5.3)
are generated based on the covariance matrix of the normalized forecasted PV
profiles and the Cholesky decomposition.
INVERTERS
5.4.3 Simulation Software
The proposed voltage control system is implemented in MATLAB R2018a. The

robust optimization problem is solved using YALMIP [185] toolbox with Mosek
9.0 [186]. The MATPOWER package 7.0 [187] for power flow analysis is used
to compute voltage profiles of network’s nodes with and without the proposed
voltage control system. The different tools used in the simulations are depicted
in Figure 5.3. First, the robust optimization is executed (365 times) for the
considered 365 forecast profiles of the households’ load and PV power, resulting
in 365 reactive power policies and 365 battery power policies for each PV-battery
inverter. Then, the generated 36500 load and PV scenarios of control nodes
are sent to the push-sum gossip protocol [90], which calculates the summation
of households’ load and PV power for each scenario. The summations are
then substituted in the linear control policies to calculate reactive power of
inverters and charge/discharge power of batteries for each scenario. Finally,
MATPOWER is used to compute voltage profiles of network’s nodes. The
power flow analysis of MATPOWER is executed (36500 times) for the generated
36500 scenarios.
Forecasted
data
Yalmip/ Scenario Network

Mosek Load and PV generation Load and PV model
scenarios scenarios
Control nodes
scenarios
Push-sum MATPOWER
gossip
Summations of
households load
and PV power Voltage
profiles
Active and reactive
Control policies power of voltage control
Figure 5.3: General overview of the MATLAB simulation used in the case study.
15-minute time step is considered in Yalmip simulations. To have load and

PV forecasted profiles with 15-minute resolution, we take the average over 15
minutes for each one-minute resolution scenario. In MATPOWER simulations,
CASE STUDY 113
one-minute time step is considered for each day. One control policy is considered
for each 15 minutes simulation in MATPOWER.
5.4.4 Simulation Results
First, the voltage profiles are computed for the 36500 scenarios without applying
the linear control policies. Figure 5.4 shows voltage profiles of control node 31
(phase to neutral voltage). One can see that most of voltage profiles of different
scenarios exceed the maximum voltage limit. When the linear control policies
are applied, it is noticed that the existing violations of the over-voltage limit
are eliminated for the 36500 scenarios, which demonstrates the robustness of
the proposed policy-based DisCVC. Figure 5.5 shows voltage profiles of control
node 31 (at the end of the feeder) after applying the policies.
Figure 5.4: Voltage profiles of control node 31 with no voltage control.
Figure 5.5: Voltage profiles of control node 31 with voltage control.

INVERTERS
Figure 5.6a and Figure 5.6b show, respectively, the apparent power magnitude
and reactive power of inverter 31 for the 36500 scenarios. Figure 5.6c and
Figure 5.6d show the energy content and charge/discharge power of battery 31
for all the scenarios. One can see that limits of the battery and inverter are
respected. As the cost of reactive power compensation cq is cheaper than the
cost of battery degradation cp , we notice that the reactive power of inverter 31
used in regulating voltage profiles is higher than the charge/discharge power of
battery 31.
(a) (b)
(c) (d)
Figure 5.6: (a) Apparent power magnitude of inverter 31, (b) Reactive power
of inverter 31, (c) Energy content of battery 31, (d) Charge/discharge power of
battery 31.
CONCLUSION 115
5.5 Conclusion
A novel robust policy-based distributed voltage control is proposed in this

chapter to regulate voltage profiles of distribution networks with high penetration
of photovoltaic systems. The proposed system consists of offline coordinator
and real-time distributed voltage control. The offline coordinator computes
linear voltage control policies on a day-ahead basis using robust optimization.
During the day, the PV-battery inverters communicate with each other and
apply the robust policies to compute the amount of reactive power and battery
power needed to maintain voltages within the limits.
The proposed policy-based distributed voltage control keeps the same advantages
as real-time optimization-based distributed voltage control techniques, but it is
easier to implement in real-time and does not rely on iterative methods. The
PV-battery inverters solve cooperatively in real-time a set of linear equations
rather than solving a complex optimization problem.
The proposed policy-based distributed voltage control is applied on a real
low voltage network. 36500 load and PV power scenarios are generated to
demonstrate the robustness of the linear control policies.
Chapter 6
Simultaneous Provision of
Voltage and Frequency
Control by PV-Battery
Systems
With the rise of distributed energy resources, photovoltaic-battery systems

are needed to maintain voltages within limits, and balance between demand
and supply. These systems can be exploited more by controlling them to
provide multiple, stacked services. In this chapter, we propose a novel
control methodology for photovoltaic-battery systems to provide simultaneously
distributed voltage control and frequency containment reserve. The control
methodology is structured in two phases. In the day-ahead phase, the control
problem is formulated as a robust optimization problem. The aims of this
optimization problem are to allocate fractions of the energy and power capacity
of each battery energy storage system to the two services, minimize the expected
cost of reactive power compensation and batteries degradation, maximize profits
from frequency control, and compute a set of linear control policies. The
optimization problem also aims to immunize against service unavailability, and
violating operational limits. This immunity is accomplished by considering the
uncertainty in the households’ active power consumption, photovoltaic power
generation, and grid frequency. In the real-time phase, the linear policies are
applied to regulate voltage profiles, and keep energy contents of batteries within
limits while providing frequency control. A 62-node low voltage network is used
117
118 SIMULTANEOUS PROVISION OF VOLTAGE AND FREQUENCY CONTROL BY PV-BATTERY
SYSTEMS
as a case study. Simulations over 104 scenarios are used to demonstrate the
robustness of the proposed control methodology. The contents of this chapter
are based upon the journal paper:
• H. Almasalma, and G. Deconinck, “Simultaneous Provision of Voltage

and Frequency Control by PV-Battery Systems,” IEEE ACCESS, vol. 8,
2020 (Early Access).
6.1 Introduction
6.1.1 Motivation
Smart inverters with advanced grid support mechanisms and data communica-
tion capabilities are being deployed to tackle the negative impact of increased
PV penetration on grid voltage [188]. These inverters have the ability to absorb
(or inject) reactive power and curtail PV power to maintain grid voltage within
accepted limits. Behind-the-meter battery energy storage systems (BESSs)
can also play an important role in regulating voltage profiles of distribution
networks with high PV penetration [31]. BESSs can be paired with smart PV
inverters to control their energy output in response to violation of grid voltage
limits. Smart inverters, for example, can charge BESSs during PV peak period
to mitigate voltage rise.
To enable PV-battery inverters to actively participate in voltage regulation,
various voltage control strategies of different complexity and data transfer needs
have been proposed in literature. As has been discussed in the previous chapters,
one of the effective control strategies that enables smart PV-battery inverters
to regulate grid voltage is distributed coordinated voltage control (DisCVC)
[40]. In DisCVC, smart inverters communicate with each other in a peer-to-peer
(P2P) fashion to solve a voltage control problem in a distributed way without
relying on a central decision-making controller.
As shown in Figure 6.1, when BESSs are used to support PV inverters in
regulating voltage profiles, they are inactive or partially used most of the time.
Their unused capacity could be used to provide other services, allowing them
to generate additional revenues. Since frequency control has been identified
as one of the highest value services for BESSs [189], we are interested in
combining voltage control with frequency containment reserve (FCR) (also
referred to as primary frequency control). Moreover, it can be a necessity for
PV-battery systems that provide frequency control to provide voltage control
as well; providing frequency control services with several BESSs connected to
INTRODUCTION 119
a distribution network can lead to voltage issues [37]. Combining frequency

control with voltage control can help in avoiding these issues. Reactive power
control of smart inverters, for example, can support BESSs in providing more
FCR capacity without causing the grid violating voltage operation limits.
Figure 6.1: The results of this figure are obtained based on the implementation
of the policy-based DisCVC proposed in Chapter 5, in a simulation over 104
scenarios. The scenarios are presented in Subsection 6.4.2. The DisCVC is used
to regulate voltage profiles of the 62-node network presented in Subsection 6.4.1.
This figure presents the charging power of BESS number 31.
6.1.2 Provision of FCR with a BESS
The objective of the FCR is to maintain active power balance between

consumption and generation, within a synchronous area, at a frequency close to
the nominal frequency fnom . FCR reserve capacity is activated automatically in
response to the grid frequency deviations from the nominal frequency. FCR is
the fastest reserve procured by transmission system operators (TSOs); providers
are required to reach their full committed reserve capacity within 30 seconds.
In many countries, TSOs procure their FCR capacity via a periodic auctions in
which third parties, such as aggregators, can bid a certain amount of reserve
capacity at a certain price. When accepted, the committed FCR capacity should
be provided continuously with 100% availability during the contracted period
[190]. Until the middle of 2019, FCR auctions in Europe were put out on a
weekly basis. From July of 2019, the contracted period was reduced from one
week to one day. This shortened period will help FCR providers to design their
control systems in a more robust way, as uncertainty is considered for only one
day [191].
SYSTEMS
The FCR power p(k) FCR

of a certain asset, participating in frequency control,
should be adjusted proportional to the relative grid frequency deviations, as
shown in (6.1a) 1 .
p(k)
FCR
= r∆f (k) , (6.1a)
 (k)
−fnom
 f
if ∆fdb <| f (k) − fnom |< ∆f ,
∆f



1 if f (k) − fnom ≥ ∆f ,

∆f (k) = (6.1b)
−1

 if f (k) − fnom ≤ −∆f ,
0 otherwise.


Here, k is the time step of the frequency control, ∆f is the frequency deviation
at which maximum FCR capacity r needs to be activated, and ∆fdb is the
frequency deadband in which no FCR reaction is required. In the Continental
Europe (CE) synchronous area, ∆f = 200 mHz and ∆fdb = 10 mHz.
Non-delivery or non-availability of a committed FCR reserve capacity will result
in penalties charged by the TSO. Providers can lose all revenues made from
FCR in case of not meeting the FCR requirements, or can be excluded from
further participation in the market in case delivery is consistently not available
or insufficient.
As BESSs have proven great potential in providing FCR [37], many TSOs
opened their networks for BESSs and released new FCR regulations for assets
with limited energy reservoirs [192]. BESSs are expected to entirely take over
the role of FCR providers from thermal power plants. Nevertheless, designing
a control system that enables BESSs to provide FCR is not straightforward.
Due to the limited capacity of batteries, ensuring the FCR reserve capacity r is
available during the contracted period is not possible without a management
strategy that keeps energy content of BESSs within limits. This is because the
frequency signal has a non-zero energy content over short time periods, and
efficiency losses of batteries decrease their energy content when being charged
or discharged. Figure 6.2 demonstrates the fact that a BESS cannot be used to
provide FCR without an energy management strategy. In Figure 6.2, a 13.5
kWh residential battery is used to provide FCR based on (6.1a). Two years of
frequency data (2017-2018) with 1 minute resolution, from the CE synchronous
region [193], are used to study the energy content of the 13.5 kWh battery
while providing FCR. One can clearly notice the violation of the maximum and
minimum energy content limits. Hence, an energy management strategy is a
must.
1 The convention in this chapter is that the power is positive when it is consumed (e.g.,
charging power of the BESS) and negative when it is injected (e.g., discharging power of the
BESS).
INTRODUCTION 121
Figure 6.2: Energy content of the 13.5 kWh battery over 730 frequency
scenarios. Battery parameters: maximum
√ power= 7 kW, initial charge= 6.5
kWh, charge/discharge efficiency=
0.9. The energy content E is calculated
(k) + (k) +
based on: E (k+1)
= E + ∆t ηi pFCR − ηdis −pFCR
(k) ch 1
, where ηich and
i
ηidis are, respectively, the charge and discharge efficiencies , ∆t is the time step
+
duration and [·] ≡ max(·, 0).
The German TSOs have defined control strategies [194], also referred to as
degrees of freedom (Dof), which can be utilized to keep energy content of
batteries within operational range during FCR provision. These strategies
presume that BESSs can deviate slightly from the required FCR Power. These
strategies are: 1) overdelivery: providers are allowed to provide FCR power that
is anywhere between 100% and 120% of the instantaneous FCR requirement,
2) deadband utilization: usually FCR assets are not obliged to activate their
reserve capacity within the range ±10 mHz around the nominal frequency, a
battery can use this
deadband for charging and discharging by using a power
within the range 0, p(k) in case of charging, or within the range , 0 in
(k)
FCR
−p FCR
case of discharging. A comparison of these control strategies is made in [194],
in which authors conclude that overdelivery and deadband utilization are not
sufficient to regulate the energy contents of BESSs within limits while providing
FCR. Providers, who wish to use these strategies, should combine them with an
additional management strategy, or use a single battery management strategy
that can regulate energy content of batteries within limits while providing FCR.
6.1.3 Batteries Providing Multiple Services
BESSs create value for prosumers and grid operators, but leave significant
untapped value on the table. Currently, most BESSs are deployed for one of
SYSTEMS
three single applications: maximization of PV self-consumption, demand charge

reduction, or backup power. This results in batteries sitting unused for over
half of their lifetime [33]. Hence, there is a need for designing innovative control
systems that enable batteries to provide simultaneously multiple services (so-
called value stacking). The challenges of designing such a control system are to
decide how much of batteries energy and power capacities to allocate to each
service, minimize the risk of service unavailability, and minimize the risk of
violating battery constraints.
A control framework is designed in [195] for a BESS to provide simultaneously
FCR and dispatch of the operation of an active distribution feeder. The work
of [196] presents a methodology for evaluating benefits of batteries for multiple
services, including balancing service, energy arbitrage, distribution system
equipment deferral, and power outage mitigation. Model predictive control is
used in [197] to dynamically co-optimize the allocation of batteries energy and
power capacities over three services: FCR, minimization of PV curtailment,
and demand smoothing. A dynamic programming approach is applied in [198]
to co-optimize a storage device for energy arbitrage and frequency regulation.
A joint optimization framework is proposed in [199] for batteries to perform
peak shaving and provide frequency regulation service. Stochastic dual dynamic
programming is applied in [200] to implement a control system that enables
PV-battery systems to provide FCR, automatic frequency restoration reserve
(aFRR), and maximization of PV self-consumption. A controller that allows
batteries to be used simultaneously for self-consumption and FCR is presented
in [176]; stochastic and robust optimization techniques are used to maximize the
expected profit from combining these two services; the controller also computes
a linear management policy that regulates battery energy content within limits
while providing FCR.
To the best of our knowledge, simultaneous provision of voltage control and FCR
by BESSs has not been discussed in literature yet. There are some methods
in literature that combine voltage and frequency control. These methods use
active power control to provide frequency control and reactive power control
to provide voltage control. They do not use active power control to provide
simultaneously voltage and frequency control as we propose in this chapter.
For example, in [201], active power demand is controlled to support the grid
frequency and reactive power compensation is used to support the voltage.
The remainder of the chapter is organised as follows: Section 6.2 presents the
chapter’s contribution and general overview of the proposed control system.
The optimization problem treated in this chapter is formulated in Section
6.3 based on a robust optimization technique and linear models of control
policies, voltages, inverters and BESSs. Results that validate the proposed
mathematical optimization problem are presented in Section 6.4. Finally, the
CONTRIBUTION AND GENERAL OVERVIEW OF THE PROPOSED CONTROL FRAMEWORK 123
chapter is concluded in Section 6.5.
6.2 Contribution and General Overview of the

Proposed Control Framework
In this chapter, we complement previous work on BESSs providing multiple

services by designing a novel control system for residential PV-battery systems
(working together) to provide simultaneously FCR service (to their synchronous
area), and voltage control service (to their distribution network). The main
contributions of this chapter are summarized in the following subsections.
6.2.1 The Proposed Control Framework
We propose a robust optimization-based control framework for PV-battery

systems to simultaneously provide policy-based DisCVC and FCR. General
overview of the proposed control framework is shown in Figure 6.3. The control
framework consists of two phases. In the day-ahead phase, the combined voltage
and frequency control problem is formulated as a robust optimization problem.
The aims of this optimization problem are to allocate fractions of the energy and
power capacity of each BESS to the two services, compute a set of linear control
policies (voltage control policies and battery management policies), minimize
the expected cost of reactive power compensation and batteries degradation cost,
and maximize profits from FCR. The robust optimization problem also aims
to immunize against service unavailability, and violating batteries, inverters
and voltage constraints. This immunity is accomplished by considering the
uncertainty in the households’ active power consumption, PV power generation,
and grid frequency.
In the real-time phase, the grid support functions of smart PV-battery inverters
apply the linear control policies to regulate voltage profiles, and keep energy
content of BESSs within limits while providing FCR service. Each battery is
divided into 3 virtual batteries: voltage control battery (VC battery) that
regulates grid voltage together with the inverter reactive power; upward-
FCR battery (R-UP battery) that provides upward frequency regulation; and
downward-FCR battery (R-DN battery) that provides downward frequency
regulation. Each inverter contains two grid support functions, grid voltage
support function (GVSF) and frequency support function (FSF). The GVSF
hosts two voltage control policies: reactive power policy that gives reactive
power setpoints of the inverter, and active power policy that gives setpoints
of the VC battery power. As designing a voltage control policy based only on
SYSTEMS
PV/load forecast Control policies and

Offline controller reserve capacity to
Frequency data
other systems
Control policies
and reserve capacity Day-ahead
Real time control
p p
Frequency Grid voltage invi i
PV support function support function q q
invi i
ppv
i
p
li
Inverter control q Control
p li node
bat i Smart PV-battery inverter
VC battery p
VC, bat i
Household load
R-UP battery p
R-UP, bati
p
R-DN, bati
R-DN battery
Control policies,
Control policies, reserve capacity
reserve capacity Control node
Control node
Figure 6.3: General overview of the proposed control framework.
local information yields to a poor performance [26], each voltage control policy
is designed as a linear function of the local active powers of the household’s
load and PV installation, and active powers of the households’ loads and PV
installations connected to the nodes participating in voltage control (control
nodes). Each GVSF needs to know the sum of the households’ active power
consumption and PV power generation of the nodes participating in voltage
control, to be able to compute its own control actions. To compute the sum
of the households’ active power consumption and PV power generation in a
fast and efficient way, a P2P-based push-sum gossip protocol is applied in this
chapter [90]. The push-sum gossip protocol enables the PV-battery inverters to
perform a distributed summation with a moderate communication overhead.
The FSF computes the upward-FCR power of the R-UP battery, and the
downward-FCR power of the R-DN battery. The FSF hosts two linear
management policies: an R-UP battery management policy that maintains
energy content of the R-UP battery within limits while providing upward
frequency regulation; and an R-DN battery management policy that maintains
CONTRIBUTION AND GENERAL OVERVIEW OF THE PROPOSED CONTROL FRAMEWORK 125
energy content of the R-DN battery within limits while providing downward
frequency regulation.
6.2.2 FCR Reserve Capacity
In this work, we consider the case of end-prosumers, with PV-battery systems,

providing FCR service to the TSO, possibly through an intermediary, for
example an aggregator. In distribution networks with high PV penetration,
the FCR capacity is affected by the magnitude of voltages, the capability
of the voltage control system, and inverters capacity. Batteries located in a
distribution network with a voltage problem cannot commit to provide a fixed
reserve capacity over the entire contracted period; they can commit to provide
a variable reserve capacity. Moreover, they cannot commit to provide the same
reserve capacity for the upward-FCR and downward-FCR2 . Hence, we propose
a variable FCR capacity profile. The FCR is proposed to be auctioned daily
in the form of 15 minutes products. The upward and downward reserves are
proposed to be procured independently. Prosumers (or their aggregator) shall
send day-ahead two FCR profiles to the TSO: upward-FCR capacity profile; and
downward-FCR capacity profile. The example of Figure 6.4 shows the reason
why we propose a variable FCR capacity for batteries located in distribution
networks with voltage problems.
Figure 6.4a shows a voltage profile of a prosumer, with a PV-battery system,
located in a distribution network with high feed-in power from PV installations;
Figure 6.4b shows the expected upward and downward reserve capacity profiles
of the prosumer’s battery that provide simultaneously voltage and frequency
control. If the amount of available reactive power capacity is sufficient to solve
the voltage rise problem (of Area 2 in Figure 6.4a) without the need to charge
the VC battery, the R-UP battery can be discharged to provide upward-FCR.
In this case, the upward-FCR capacity depends on the amount of remaining
reactive power capacity that can support the R-UP battery to inject more FCR
power without causing the network violating voltage operation limits. On the
other hand, if the amount of available reactive power capacity is not sufficient to
solve the voltage rise problem, the controller will charge the VC battery. In this
case, the R-UP battery cannot be discharged to provide upward-FCR3 , as the
2 In the upward-FCR, batteries are discharged in response to the negative frequency
deviation, whereas in the downward-FCR, batteries are charged in response to the positive
frequency deviation.
3 It is worth to point out that the priority of the control system is to solve voltage rise
problems. This is to avoid inverters disconnection due to overvoltage.

SYSTEMS
(a)
Upward reserve capacity

Downward reserve capacity
(b)
Figure 6.4: (a) Unregulated voltage profile of a prosumer in a network with high
PV penetration, (b) Expected upward and downward reserve capacity profile.
In this example, the inverter reactive power and the VC battery power are used
to solve the voltage rise problem, therefore, the upward reserve capacity is zero
in Area 2.
battery4 cannot be charged and discharged at the same time. Since there is no
voltage rise problem in Area 1 and Area 3, the R-UP battery can be discharged
in these areas to provide upward-FCR. The voltage in Area 1 is closer to the
maximum limit than the voltage in Area 3, therefore, the upward-FCR capacity
in Area 1 is expected to be less than the upward-FCR capacity in Area 3.
As there is no voltage drop problem in the example of Figure 6.4, the R-DN
battery can be charged in the three areas to provide downward-FCR. The
downward-FCR capacity depends on the amount of reactive power capacity
4 In this chapter, if the word “battery” is not preceded by the three abbreviations: VC;
R-UP; and R-DN, then this word means the real battery and not one of the three virtual
batteries.
that can support the R-DN battery to absorb more FCR power without causing
the network violating voltage operation limits. In Area 2 of Figure 6.4a, the
downward-FCR capacity is expected to be at minimum, since most of the
inverter capacity is expected to be occupied by the PV power and the reactive
power of the voltage control system. The downward-FCR capacity in Area 3 is
expected to be less than the downward-FCR capacity in Area 1. This is because
the voltage in Area 3 is closer to the minimum voltage limit than the voltage in
Area 1.
For sake of simplicity, the power capacity reserved for the energy management
policy of the example of Figure 6.4b is set to a constant value of 1 kW. In the
proposed optimization problem, the power capacity of the energy management
policy is considered as a variable capacity to compute the minimum sufficient
power capacity. It is clear from Figure 6.4b that the energy management policy
will have to be designed carefully, as more power capacity for the management
policy will mean that less FCR capacity can be sold to the TSO, while on the
other hand the power capacity for the management policy should be sufficient
to ensure the energy content of the battery remains within limits.
6.3 Problem Formulation
In the following subsections, we present the constraints and objective function

of the optimization problem the offline controller solves on a day-ahead basis.
Trading off modelling accuracy against computational tractability, we resort to
linearized models. Robust optimization is then used to construct a tractable
optimization problem for the uncertain optimization problem of the offline
controller.
The time horizon of the optimization problem is one day, discretized into nt
time steps of duration ∆t. The time steps belong to the set K = {1, · · · , nt }.
We consider two types of network’s nodes: control nodes where the PV-
battery inverters and households’ loads are connected; and passive nodes with
only households’ loads. Control nodes belong to the set I = {1, · · · , nc },
passive nodes belong to the set J , and network’s nodes belong to the set
N = {1, · · · , nnodes }. nc is the number of control nodes, and nnodes is the
number of network’s nodes. A tilde sign (∼) is used in this chapter to indicate
a variable subject to uncertainty.
SYSTEMS
6.3.1 Uncertainty Model
The sources of uncertainty in the control problem of the offline controller include
households’ active and reactive power consumption, PV power generation, and
(k)
the grid frequency. The uncertain household’s active power consumption p̃lx ,
(k)
and the uncertain PV power generation p̃pvi can be bounded as:
(k) (k)
0 ≤ p̃lx ≤ plx


∀x ∈ N , ∀k ∈ K (6.2)
(k) (k) (k) 
plx = plx + ∆plx
0 ≤ p̃(k) (k)

pvi ≤ ppvi 
∀i ∈ I, ∀k ∈ K (6.3)
pvi = ppvi + ∆ppvi
p(k) (k) (k) 
(k) (k)
Where plx and ppvi are, respectively, the maximum uncertain household’s
active power consumption, and the maximum uncertain PV power generation.
(k)
plx is defined in (6.2) as the sum of the forecasted household’s active power
(k) (k)
consumption plx ∈ R≥0 and the maximum deviation ∆plx from the forecast of
(k)
the household’s active power consumption. In (6.3), ppvi is defined as the sum
(k)
of the forecasted PV power generation ppvi ∈ R≥0 and the maximum deviation
(k) (k) (k)
∆ppvi from the PV forecast. ∆plx and ∆ppvi can be approximated based on
historical data of forecast error.
(k)
The uncertain household’s reactive power consumption q̃lx can be defined as
(k)
function of p̃lx and a power factor (PF):
(k) (k)
q̃lx = tan (acos (PF)) p̃lx , ∀x ∈ N , ∀k ∈ K (6.4)
(k)
A normalized positive frequency deviation ∆f
f
R-DN (related to the downward-
(k)
FCR), and normalized negative frequency deviation ∆f f
R-UP (related to the
upward-FCR) can be bounded as:
(k) (k)
0 ≤ ∆f
f
R-DN ≤ ∆f R-DN , ∀k ∈ K (6.5a)
(k)
∆f (k)
R-UP
≤ ∆f
f
R-UP ≤ 0, ∀k ∈ K (6.5b)
(k)
Here, ∆f R-DN and ∆f (k)
R-UP
are the maximum normalized frequency deviation
and minimum normalized frequency deviation at time step k, respectively. In
(k)
this chapter, ∆f R-DN and ∆f (k)
R-UP
are set based on historical frequency data.
(k)
One can set ∆f R-DN to 1, and ∆f (k)
R-UP
to −1 ∀k ∈ K, which is correct. However,
from our experience, this will lead to conservative solutions, e.g., low contracted
reserve capacity. As shown in Figure 6.5, the historical maximum normalized
frequency deviation can be much less than 1, and the historical minimum
normalized frequency deviation can be much higher than −1.
Figure 6.5: Normalized positive and negative frequency deviation ∆f of 730 days
of historical frequency data (2017-2018). The dotted lines show the maximum
and minimum normalized frequency deviations. ∆f is calculated based on
(6.1b). The frequency data are from the CE synchronous region.
For given historical frequency measurements with nf measurements per time

(k)
step k, ∆f R-DN and ∆f (k)
R-UP
can be calculated as:
(κ)
∆f (k) = (6.6)
 (k) (κ)
(f ) −fnom
 if ∆fdb <| (f (k) )(κ) − fnom |< ∆f ,
∆f



1 if (f (k) )(κ) − fnom ≥ ∆f ,



−1 if (f (k) )(κ) − fnom ≤ −∆f ,
0 otherwise.


κ = 1, · · · , nf ,
!
(k)
(1) + (nf ) +
∆f R-DN = max ∆f (k)
,··· , ∆f (k)
,
SYSTEMS
!
(1) − (nf ) −
∆f (k)
R-UP
= min ∆f (k)
,··· , ∆f (k)
(κ)
Here, ∆f (k) is the normalized frequency deviation number κ related to
the frequency measurement number κ (f (k) )(κ) at time step k. The operator
+ −
[·] ≡ max(·, 0), and [·] ≡ min(·, 0).
To get immunity against service unavailability, and violating operational
constraints, we look for solutions which are feasible for any realization of the
uncertain data in a predefined uncertainty set U. In this chapter, we consider a
polyhedral description of the uncertainty set of the form:
U = {ζ ≥ 0 : Hζ ≤ h} , (6.7)
H ∈ R2nζ ×nζ and h ∈ R2nζ can be defined to represent the constraints (6.2)-
(6.5b), where nζ = nt (2 + nc + nnodes ) is the number of uncertain variables for nt
times steps. The vector ζ ∈ Rnζ includes the uncertain households’ active power
consumption, PV power generation, normalized positive frequency deviations,
and normalized negative frequency deviations for nt time steps:
h (1) (nt ) (1) (nt )
ζ = P̃ l , · · · , P̃ l , P̃ pv , · · · , P̃ pv , (6.8a)
(1) (n ) (1) (n ) iT
∆f f t f t
R-DN , · · · , ∆f R-DN , ∆f R-UP , · · · , ∆f R-UP
f f
(k)
h i
(k) (k)
P̃ l = p̃l1 , · · · , p̃ln , ∀k ∈ K (6.8b)
nodes
(k)
h i
P̃ pv = p̃(k)
pv1 , · · · , p̃(k)
pvn , ∀k ∈ K (6.8c)
c
In subsection 6.3.9, we present the robust counterpart formulation of uncertain

linear constraints. The robust counterpart formulation is derived based on the
uncertainty set U defined in (6.7).
The convention in this chapter is that the power is positive when it is consumed
(k) (k)
and negative when it is injected. In (6.3), p̃pvi and ppvi are defined as positive
(k) (k)
values. Hence, a minus sign will be introduced in front of p̃pvi and ppvi .
6.3.2 Linear Control Policies
Robust optimization is a technique to solve mathematical optimization problems

in which the data are uncertain and are only known to belong to some uncertainty
set. The goal of robust optimization is to find solutions that are immune to
uncertainty. In static robust optimization, decisions must be made before the
realization of the uncertain data. Ben-Tal et al. extended the robust optimization
framework to dynamic settings by proposing the adjustable robust optimization
(ARO) technique [170], in which some decision variables are allowed to be
computed after the realization of the uncertain data, thus leading to better
objective value and less conservative solutions. This performance is achieved
at higher computational burden, as ARO is usually NP-hard to solve. Ben-Tal
et al. proposed the method of affinely adjustable robust optimization (AARO)
to construct a tractable optimization problem. In AARO, decision variables
are restricted to be affine functions of the uncertain data [202]. Accordingly,
control policies proposed in this chapter are designed as linear functions of the
uncertain data.
As has been discussed in subsection 6.2.1, we consider four linear control policies
for each PV-battery system: reactive power policy that gives reactive power
setpoints; active power policy that gives charge/discharge power setpoints of
the VC battery; R-UP battery management policy that gives charge power
setpoints needed to maintain the energy content of the R-UP battery within
limits while providing upward-FCR; and R-DN battery management policy that
gives discharge power setpoints needed to maintain the energy content of the
R-DN battery within limits while providing downward-FCR.
Active and Reactive Power Policies
To regulate voltage profiles, each PV-battery system adjusts its reactive power
and VC battery power linearly as function of control nodes’ active power
consumption and PV power generation, and normalized frequency deviations.
Given the possibility that the frequency deviation can be either positive or
negative for each time step, we propose two types of reactive power policy:
reactive power policy that takes into account the effect of responding to positive
frequency deviations; and reactive power policy that takes into account the
effect of responding to negative frequency deviations:
∀i ∈ I, ∀k ∈ K :
(k)
X (k)

(k) (k)
q̃R-DN, invi = βp(k)
i pvi + βR-DNi ∆f R-DN
p̃li − p̃(k) f (6.9a)
i∈I
(k)
X (k)
(k) (k)
q̃R-UP, invi = βp(k)
i pvi + βR-UPi ∆f R-UP
p̃li − p̃(k) f (6.9b)
i∈I
SYSTEMS
(k) (k)
where q̃R-DN, invi and q̃R-UP, invi are the uncertain reactive powers of inverter
i (at time step k) that take into account the effect of downward-FCR and
(k)
upward-FCR, respectively, on mitigating voltage problem. βpi is the coefficient
of the reactive power policy (of inverter i at time step k) related to the active
powers of households’ loads and PV installations connected to the control nodes.
(k) (k)
βR-DNi and βR-UPi are, respectively, the coefficients of the reactive power policy
related to the positive and negative normalized frequency deviations at time
step k. The two types of the reactive power policy can be defined in a vectorized
form as:
∀i ∈ I :
h iT
(1) (nt )
Q̃R-DN, invi = q̃R-DN, invi , · · · , q̃R-DN, invi
= β pi ξ ∆p + β R-DNi ξ R-DN (6.10a)

h iT
(1) (nt )
Q̃R-UP, invi = q̃R-UP, invi , · · · , q̃R-UP, invi
= β pi ξ ∆p + β R-UPi ξ R-UP (6.10b)

" #T
X (1)
X (n )

ξ ∆p = p̃li − p̃(1)
pvi ,··· , p̃li t − p̃(nt)
pvi ,
i∈I i∈I
T
(1) (n )
ξ R-DN = ∆f
f , · · · , f t
∆f ,
R-DN R-DN
T
(1) (nt )
ξ R-UP = ∆f R-UP , · · · , ∆f R-UP
f f
(k)
where β pi , β R-DNi and β R-UPi are diagonal matrices ∈ Rnt ×nt with the βpi ,
(k) (k)
βR-DNi and βR-UPi coefficients (∀k ∈ K) contained in their diagonal elements,
respectively. When the frequency deviation is positive, the real-time controller
uses the reactive power policy (6.9a), whereas when the frequency deviation is
negative, the real-time controller uses the reactive power policy (6.9b). The two
policies are equal when the frequency deviation is zero, or within the deadband
The active power policy that is responsible for charging and discharging VC
batteries can be defined as:
h iT
(1) (nt )
P̃ VC, bati = p̃VC, bati , · · · , p̃VC, bati = Z pi ξ ∆p , ∀i ∈ I (6.11)
(k)
where p̃VC, bati is the uncertain active power of the VC battery i at time step k.
(k)
Z pi is a diagonal matrix ∈ Rnt ×nt , with zpi (∀k ∈ K) the coefficients of the
active power policy contained in its diagonal elements.
For each time step k, the coefficients of the reactive and active power policies
are calculated considering the following cases:
1. In the case the offline controller predicts a voltage rise problem:

(a) If VC batteries are predicted not to be charged, the offline controller
(k) (k)
sets zpi to zero, and maximizes the upward-FCR capacity (rUPi )
(k)
and downward-FCR capacity (rDNi ) considering limits of batteries,
(k) (k) (k)
inverters and voltages. βpi , βR-DNi and βR-UPi are calculated
considering the following cases:
(k)
i. Negative frequency deviation case (upward-FCR): βpi is
calculated to solve the voltage rise problem predicted to be
(k)
caused by reverse power flow from PV units, whereas βR-UPi is
calculated to solve the voltage rise problem that may arise from
discharging R-UP batteries.
(k)
ii. Positive frequency deviation case (downward-FCR): βpi is
calculated to solve the voltage rise problem taking into account
the effect of charging R-DN batteries on mitigating voltage rise.
(k)
βR-DNi is included in the calculation of reactive power to let
the real-time controller be aware that charging R-DN batteries
helps in mitigating voltage rise problem and less reactive power
is needed.
(b) If VC batteries are predicted to be charged:
i. Negative frequency deviation case: the offline controller sets the
upward-FCR capacity to zero, as it is impossible to charge and
discharge the real battery at the same time. Logic constraints
(presented in subsection 6.3.7) are used to set the upward-FCR
(k) (k)
capacity at time step k and βR-UPi to zero. βpi is calculated
to solve the predicted voltage rise problem considering the help
(k)
from VC batteries (zpi < 0).
ii. Positive frequency deviation case: the offline controller maxi-
mizes the downward-FCR capacity considering limits of batteries,
(k) (k)
inverters and voltages. βpi and zpi are calculated to solve the
voltage rise problem taking into account the effect of charging
the R-DN batteries on mitigating voltage rise (which is included
(k)
in βR-DNi ).
SYSTEMS
2. In the case the offline controller predicts a voltage drop problem: the
same logic as above is applied.
3. In the case the offline controller predicts no voltage problem: the upward-
FCR and downward-FCR capacities are maximized considering constraints
related to batteries, inverters and voltages. The offline controller sets
(k) (k) (k)
βpi and zpi to zero. βR-UPi is calculated to avoid any voltage rise
(k)
problem that may arise from discharging R-UP batteries, whereas βR-DNi
is calculated to avoid any voltage drop problem that may arise from
charging R-DN batteries.
It is worth to point out that to reduce the measurements and communication

overhead of the real-time controller, the reactive and active power policies are
designed as linear functions of only the households’ active power consumption
and PV power of control nodes. In real-time, the control policies do not consider
the households’ active power consumption of passive nodes, but we do consider
the effect of passive nodes’ active power consumption on voltages in the branch
flow model of the offline optimization, as will be shown in subsection 6.3.6.
Energy Management Policies
In [176], a linear energy management policy to regulate the state of charge of a

battery while providing FCR is presented. For each time step k, the power of
this energy management policy is designed asa linear function of normalized

(1) (k−1)
frequency deviations that belong to the set ∆f , · · · , ∆f
f f . In this
chapter, we adapt this policy to be used for regulating the energy content of
R-UP batteries and R-DN batteries. For each time step k, the power of the
(k)
R-UP battery management policy p̃m-UP, bati is expressed as a linear function of

(1) (k−1)
negative frequency deviations that belong to the set ∆f R-UP , · · · , ∆f R-UP ,
f f
(k)
whereas power of the R-DN battery management policy p̃m-DN, bati is expressed
as
a linear function of positive frequency deviations that belong to the set
(1) (k−1)
∆f
f , · · · , ∆f
R-DN
f
R-DN :
k−1 (k)
(k)
X
(κ) (κ)
p̃m-DN, bati = mR-DNi ∆f
f
R-DN , ∀i ∈ I, ∀k ∈ K
κ=1
(k)
h iT
(1) (nt )
P̃ m-DN, bati = p̃m-DN, bati , · · · , p̃m-DN, bati
= M R-DNi ξ R-DN , ∀i ∈ I (6.12)
(k)
(κ)
with mR-DNi , κ = 1, · · · , k − 1 (∀k ∈ K), the coefficients of the R-
DN battery management policy i, contained in the lower triangular matrix
M R-DNi ∈ Rnt ×nt with zeros on the diagonal.
(k)
(κ)
With mR-UPi , κ = 1, · · · , k − 1 (∀k ∈ K), as the coefficients of the R-
UP battery management policy i, contained in the lower triangular matrix
M R-UPi ∈ Rnt ×nt with zeros on the diagonal, the power of the R-UP battery
(k)
management policy p̃m-UP, bati (∀k ∈ K) can be defined as:
(k)
h iT
(1) (nt )
P̃ m-UP, bati = p̃m-UP, bati , · · · , p̃m-UP, bati
= M R-UPi ξ R-UP , ∀i ∈ I (6.13)
6.3.3 Allocation of BESS Capacity
As shown in Figure 6.6, the offline controller allocates fractions of the energy
and power capacity of each BESS to the voltage and frequency control services.
For each time step k, the energy content of each battery is divided into three
parts that provide upward-FCR, downward-FCR, and voltage control. The
constraints (6.14a)-(6.14h) allocate the maximum, minimum, and initial energy
capacity for each part:
∀k ∈ K, ∀i ∈ I
Stochastic constraints:
(k) (k) (k)
eVC, bati 6 ẼVC, bati 6 eVC, bati (6.14a)
(k) (k) (k)

e R-UP, bati 6 ẼR-UP, bati 6 eR-UP, bati (6.14b)
SYSTEMS
pbat e bat i
i
(k)
r DN
Charge power
(k) (k)
i Downward-FCR Downward-FCR e R-DN, bat i e R-DN, bat i
(k)
pm-UP, bat
i R-UP management policy R-DN management policy
p (k)
VC, bat i Voltage rise
Voltage rise
(k) (k) (k)
pVC, bat
0 e VC, bat i e VC, bat
i Voltage drop i
Voltage drop
(k)
Discharge power
p m-DN, bati R-DN management policy R-UP management policy
r (k)
UP Upward-FCR Upward-FCR (k) (k)
i e R-UP, bat i e R-UP, bat i
p bat i e bat i
Power capacity allocation Energy capacity allocation
Figure 6.6: Illustration of energy and power capacity allocation of battery i.
(k) (k) (k)

e R-DN, bati 6 ẼR-DN, bati 6 eR-DN, bati (6.14c)
Other constraints:
(k) (k) (k)
e VC, bati ≥ 0, e R-UP, bati ≥ 0, e R-DN, bati ≥ 0 (6.14d)
(k) (k) (k)

e VC, bati ≥ 0, e R-UP, bati ≥ 0, e R-DN, bati ≥ 0 (6.14e)
(k) (k) (k)

e VC, bati + e R-UP, bati + e R-DN, bati ≥ e bati (6.14f)
(k) (k) (k)

e VC, bati + e R-UP, bati + e R-DN, bati ≤ e bati (6.14g)
E 0VC, bati + E 0R-UP, bati + E 0R-DN, bati ≤ Ebat

0
i
(6.14h)
(k) (k) (k)

Here, ẼVC, bati , ẼR-UP, bati and ẼR-DN, bati are, respectively, the uncertain
energy contents of the VC battery i, R-UP battery i, and R-DN battery i
at time step k. The model of the uncertain energy contents of the virtual
batteries is presented in subsection 6.3.4. e bati and e bati denote, respectively,
(k) (k)
the maximum and minimum energy capacity of battery i. e VC, bati , e R-UP, bati
(k)
and e R-DN, bati denote maximum energy capacities of the VC battery i, R-
(k) (k)
UP battery i, and R-DN battery i at time step k. e VC, bati , e R-UP, bati and
(k)
e R-DN, bati denote minimum energy capacities of the VC battery i, R-UP
battery i, and R-DN battery i at time step k. Ebat 0
i
, E 0VC, bati , E 0R-UP, bati
and E R-DN, bati denote the initial energy contents of the battery i, VC battery
0
i, R-UP battery i, and R-DN battery i, respectively.

For each time step k, the maximum charge power capacity pbati of battery i is
(k) (k)
divided into 3 parts: downward-FCR capacity rDNi ; power capacity pm-UP, bati
(k)
of the R-UP battery management policy; and power capacity pVC, bati of the
voltage rise control. The maximum discharge power capacity (or minimum
power capacity) of battery i pbat is divided, for each time step k, into 3 parts:
i
(k) (k)
upward-FCR capacity rUPi ; power capacity pm-DN, bati of the R-DN battery
(k)
management policy; and power capacity pVC, bati of the voltage drop control.
The constraints (6.15a)-(6.15f) allocate the power capacity for each part:
∀k ∈ K, ∀i ∈ I
Stochastic constraints:
(k) (k)
p(k)
VC, bat
6 p̃VC, bati 6 pVC, bati (6.15a)
i
(k) (k)
0 6 p̃m-UP, bati 6 pm-UP, bati (6.15b)
(k)
p(k)
m-DN, bat
6 p̃m-DN, bati 6 0 (6.15c)
i
Other constraints:
(k) (k)
rUPi ≥ 0, rDNi ≥ 0 (6.15d)
(k)
p(k)
VC, bat
− rUPi + p(k)
m-DN, bat
≥ pbat (6.15e)
i i i
(k) (k) (k)

pVC, bati + rDNi + pm-UP, bati ≤ pbati (6.15f)
The power of the R-UP management policy is defined as a charge power in

(6.15b), to avoid discharging the R-UP battery to less than its minimum energy
capacity while providing upward-FCR. The power of the R-DN management
policy is defined as a discharge power in (6.15c), to avoid charging the R-DN
battery to more than its maximum energy capacity while providing donward-
SYSTEMS
FCR. In (6.15a), the power of the VC battery varies between negative power
(k) (k)
capacity pVC, bati (discharge power), and positive power capacity pVC, bati
(charge power). The VC battery can be charged to reduce the reverse power
flow from PV installations during PV peak period, which helps mitigate voltage
rise. During load peak period, the VC battery can be discharged to feed a local
load, which helps mitigate voltage drop.
6.3.4 Battery Model
We consider a simple discrete battery model for the three virtual batteries:
1 h (k)
h i+ i+
(k+1) (k) (k)
ẼVC, bati = ẼVC, bati + ∆t ηi p̃VC, bati − dis −p̃VC, bati
ch
(6.16a)
ηi
1 (k)

(k+1) (k) ch (k)
ẼR-DN, bati = ẼR-DN, bati + ∆t ηi p̃R-DNi + dis p̃m-DN, bati ,
ηi
(k) (k) f (k)

p̃R-DNi = rDNi ∆f R-DN (6.16b)
1 (k)

(k+1) (k) (k)
ẼR-UP, bati = ẼR-UP, bati + ∆t ηich p̃m-UP, bati + dis p̃R-UPi ,
ηi
(k) (k) f (k)

p̃R-UPi = rUPi ∆f R-UP (6.16c)
Here, ηich and ηidis are, respectively, the charge and discharge efficiencies of
(k) (k)
battery i. p̃R-DNi and p̃R-UPi are, respectively, the uncertain downward-FCR
+
power and upward-FCR power at time step k. In (6.16a), the operator [·] ≡
max(·, 0) introduces integer variables to the stochastic constraint (6.14a), which
results in a mixed integer stochastic constraint. As mixed integer stochastic
constraints lead to a high computational complexity, a heuristic approach is
proposed in [176] to get rid of integer variables of a battery model providing
FCR. In this chapter, we adapt the heuristic approach to be used for a battery
model providing voltage control:
+ +
(k+1) (k) (k) (k)
ẼVC, bati = ẼVC, bati + ∆t p̃VC, bati − − p̃VC, bati ,
η (k) η (k)

(k+1) (k) (k)
ẼVC, bati = ẼVC, bati + ∆t p̃VC, bati , (6.17)
η (k)
(k) (k) (k)

if pli 6 ppvi
(
ch

(k)
ηi p̃VC, bati
p̃VC, bati = (k) (k) (k)
η (k)
1
p̃
ηidis VC, bati
if pli > ppvi
The if condition in (6.17) is based on the fact that the controller tends to
discharge the battery when the load is higher than the PV generation (to
mitigate voltage drop) and charge the battery when the PV generation is higher
than the load (to mitigate voltage rise). Our simulation results show that
this heuristic approach does not lead to violation of the constraint
(6.14a).
In

(k) (k) (k)
fact, (6.17) is the same as (6.16a) if sign pli ,meas − ppvi ,meas = −sign pbati ,
(k) (k)
where pli ,meas and ppvi ,meas are, respectively, the real-time measured active
power consumption and PV power generation at time step k.
Degradation Cost
The per-kWh battery degradation cost proposed in [177] is applied in this

chapter:
creplacementi
cpi = (6.18)
Ncycle life × DoD × 2ebati
Here, cpi is the battery degradation cost per kWh (cent/kWh), creplacementi
is the battery replacement cost, Ncycle life is the battery cycle life, and DoD
is the depth-of-discharge. The denominator of (6.18) gives the charging and
discharging kWh the real battery can tolerate over its serviceable life.
6.3.5 Inverter Model
The active and reactive power passing through inverter i is limited by its
apparent power capacity:
q
(k) (k)
(p̃R-DN, invi )2 + (q̃R-DN, invi )2 6 sinvi , ∀i ∈ I, ∀k ∈ K,

(k) (k) (k)
p̃R-DN, invi = −p̃(k)
pvi + p̃R-DNi + p̃VC, bati +
(k) (k)
p̃m-DN, bati + p̃m-UP, bati (6.19a)
q
(k) (k)
(p̃R-UP, invi )2 + (q̃R-UP, invi )2 6 sinvi , ∀i ∈ I, ∀k ∈ K,
SYSTEMS

(k) (k) (k)
p̃R-UP, invi = −p̃(k)
pvi + p̃R-UPi + p̃VC, bati +
(k) (k)
p̃m-DN, bati + p̃m-UP, bati (6.19b)
The stochastic constraint (6.19a) considers the possibility of a positive frequency

deviation case, whereas the stochastic constraint (6.19b) considers the possibility
of a negative frequency deviation case. The two constraints are equal if the
frequency deviation is zero, or within the deadband. sinvi is the maximum
(k)
magnitude of apparent power of inverter i. p̃R-DN, invi is the uncertain active
power of inverter i at time step k that considers the downward-FCR power,
(k)
whereas p̃R-UP, invi is the uncertain active power of inverter i at time step k that
(k) (k)
considers the upward-FCR power. p̃R-DNi and p̃R-UPi cannot be included in one
(k) (k)
equation, as both rDNi and rUPi can be greater than zero, but only one of them
is applied in real-time. This is because the frequency deviation in real-time can
be either positive or negative.
As shown in [165], the conic stochastic constraints (6.19a) and (6.19b) can be
approximated by a number of linear stochastic constraints nS with a relative
error of 1 − cos(π/2nS ):
(k) (k)
− sinvi 6 cos(ϑφ)p̃R-DN, invi + sin(ϑφ)q̃R-DN, invi 6 sinvi ,
(k) (k)
− sinvi 6 cos(ϑφ)p̃R-UP, invi + sin(ϑφ)q̃R-UP, invi 6 sinvi ,
π
φ= , ϑ = 1, · · · , nS (6.20)
nS
Reactive Power Cost
The reactive power cost cqi (cent/kvarh), considered in this chapter, corresponds
to the approximative compensation cost of the additional inverter losses due to
the reactive power utilization and is provided by [175].
6.3.6 Branch Flow Model
In this chapter, a linear branch flow model is applied. The model is based on
the DistFlow method developed in [173]. According to [174], the linearized
branch flow model tends to introduce a small relative error of 1-5% when used
for calculating power flows of real distribution networks.
∀k ∈ K :
(k) (k)
(ṽ (k) )2R-DN = RP̃I + X Q̃I + vo2 1nnodes (6.21a)
(k)
h (k) (k)
i (k) h (k) (k)
i
P̃I = P̃ R-DN, node ; P̃ passive , Q̃I = Q̃R-DN, node ; Q̃passive
(k) (k)
(ṽ (k) )2R-UP = RP̃II + X Q̃II + vo2 1nnodes (6.21b)
(k)
h (k) (k)
i (k) h (k) (k)
i
P̃II = P̃ R-UP, node ; P̃ passive , Q̃II = Q̃R-UP, node ; Q̃passive
Here, vo is the substation voltage, which is considered constant. 1nnodes ∈

h iT
(k) (k) (k)
Rnnodes is a vector of all ones. (ṽ )2R-DN = (ṽ1 )2R-DN , · · · , (ṽnnodes )2R-DN
h iT
(k) (k) (k)
and (ṽ )2R-UP = (ṽ1 )2R-UP , · · · , (ṽnnodes )2R-UP denote vectors of squared
nodal voltage magnitudes at time step k. The effect of charging R-DN
(k)
batteries on voltage profiles is included in (ṽ )2R-DN , whereas the effect
(k)
of discharging R-UP batteries on voltage profiles is included in (ṽ )2R-UP .
The two effects cannot be included in one equation, as only one of them
occurs in real-time, but in the offline computation, both of them can be
non-zero at time step k. For each time step, we consider the two effects
in the offline optimization, to prepare the real-time controller for the two
cases: positive frequency deviation; and negative frequency deviation. The
(k)
h iT (k)
(k) (k)
vectors P̃ R-DN, node = p̃R-DN, node1 , · · · , p̃R-DN, noden and P̃ R-UP, node =
c
h iT
(k) (k)
p̃R-UP, node1 , · · · , p̃R-UP, noden denote active powers of control nodes (at
c
time step k) that consider, respectively, the downward-FCR power and upward-
(k)
h iT
(k) (k)
FCR power. The vectors Q̃R-DN, node = q̃R-DN, node1 , · · · , q̃R-DN, noden
c
(k)
h iT
(k) (k)
and Q̃R-UP, node = q̃R-UP, node1 , · · · , q̃R-UP, noden denote reactive pow-
c
ers of control nodes (at time step k) that consider, respectively, the
(k)
downward-FCR power and upward-FCR power. The vectors P̃ passive =
h iT (k)
h iT
(k) (k) (k) (k)
p̃passiven +1 , · · · , p̃passiven and Q̃passive = q̃passiven +1 , · · · , q̃passiven
c nodes c nodes
include, respectively, active powers and reactive powers of passive nodes at time
step k. If i is a control node and j is a passive node, then their active and
SYSTEMS
reactive powers can be defined as:

(k) (k) (k)
p̃R-DN, nodei = p̃li + p̃R-DN, invi 






(k) (k) (k)
q̃R-DN, node = q̃l + q̃R-DN, inv 


i
i
 i
∀i ∈ I (6.22)
(k) (k) (k)
p̃R-UP, nodei = p̃li + p̃R-UP, invi 







(k) (k) (k)

q̃R-UP, nodei = q̃li + q̃R-UP, invi

(k) (k)

p̃passivej = p̃lj 
∀j ∈ J (6.23)
(k) (k) 
q̃passivej = q̃lj 
The matrices R and X (∈ Rnnodes ×nnodes ) depend, respectively, on the resistance

and reactance of the network’s feeders. We ask the readers to refer to [166] for
details about the derivation of the above linearized branch flow model, and the
computation of the matrices R and X.
To maintain nodal voltage magnitudes within the maximum voltage limit v and
the minimum voltage limit v, we consider the following stochastic constraints:
v 2 6 (ṽ (k) )2R-DN 6 v 2 , ∀k ∈ K (6.24a)
v 2 6 (ṽ (k) )2R-UP 6 v 2 , ∀k ∈ K (6.24b)
6.3.7 Logic Constraints
As has been discussed in subsection 6.3.2, in the case VC batteries are predicted
to be charged at time step k (to help in solving voltage rise problems), the offline
(k) (k)
controller sets rUPi and βR-UPi to zero. In the case VC batteries are expected
to be discharged at time step k (to help in solving voltage drop problems), the
(k) (k)
offline controller sets rDNi and βR-DNi to zero. The following constraints are
used to implement the aforementioned settings:

(k)
If pVC, bati > 0 and p(k)
VC, bat
= 0 , (6.25a)
i
(k) (k)
then rUPi = 0 and βR-UPi = 0

(k)
If pVC, bati = 0 and p(k)
VC, bat
< 0 , (6.25b)
i
(k) (k)
then rDNi = 0 and βR-DNi = 0
The solver used in the case study (YALMIP) transforms the above logic
constraints into mixed integer linear constraints using Big-M strategy [203].
6.3.8 Objective Function
The objective function of the proposed optimization problem aims at minimizing

the expected cost of operating the voltage control system, and maximizing profits
from FCR. The expected cost of operating the voltage control system is the sum
of two costs: expected cost of the reactive power compensation; and batteries
degradation cost. This results in the following objective function:
"
X 2 2
min E cqi ∆t kβ pi ξ ∆p k + kβ R-DNi ξ R-DN k2 +
2 2
2
i∈I | {z } | {z }
1 2

2 2
kβ R-UPi ξ R-UP k2 + c2pi ∆t2 kZ pi ξ ∆p k +
| {z } | {z }2
3 4
2
k M R-DNi + IRDNi ξ R-DN k +

| {z }2
5
#
2 X
k M R-UPi + IRUPi ξ R-UP k (RUPi + RDNi ) (6.26)

−
| {z }2 i∈I
6 | {z }
7
Here, E[·] denotes the expected value operator, and k · k2 is the second norm.
Parts 1, 2 and 3 of the objective function represent sum of quadratic uncertain
reactive powers of inverter i for nt time steps. Uncertain reactive powers of
parts 1, 2 and 3 are responsible for solving voltage problems expected to be
SYSTEMS
caused by the reverse power flow from PV units, charging R-DN batteries, and
discharging R-UP batteries, respectively. Parts 4, 5 and 6 of the objective
function represent sum of quadratic uncertain active powers of battery i for
nt time steps. Uncertain active powers of part 4 are the uncertain charging
and discharging powers of the VC battery i. Uncertain active powers of part 5
represent the uncertain charging powers of the downward-FCR and uncertain
discharging powers of the R-DN management policy. Uncertain active powers
of part 6 represent uncertain discharging powers of the upward-FCR and
uncertain charging powers of the R-UP management policy. IRDNi in part
5 denotes a diagonal matrix ∈ Rnt ×nt with the vector of downward reserve
h iT
(1) (n )
capacities RDNi = rDNi , · · · , rDNt i contained in its diagonal elements. IRUPi
in part 6 denotes a diagonal matrix ∈ Rnt ×nt with the vector of upward reserve
h iT
(1) (n )
capacities RUPi = rUPi , · · · , rUPti contained in its diagonal elements. Part
7 is included in the objective function to maximize the upward and downward
reserve capacities.
It can be shown that the objective function (6.26) can be written as:
h i X
2
min E kzζk2 − (RUPi + RDNi ) (6.27)
i∈I
where z ∈ R6nc nt ×nζ can be defined to represent the first six parts of the
objective function. A closed-form of the expected value in (6.27) is not readily
available. A second-order moment matrix can be used h to
i approximate the
expected value, as discussed in [178]. If M = E ζζ T
is a second-order
moment matrix of the random vector ζ, then:
h i h i
2
E kzζk2 = E Tr zT zζζ T = Tr zT zM (6.28)

where M ∈ Rnζ ×nζ is assumed to be positive definite, and can be computed

based on the following formula:
M = µTζ µζ + cov (ζ scen ) (6.29)
Here, the matrix ζ scen ∈ Rnscen ×nζ contains nscen scenarios of the random vector
ζ. µ ∈ R1×nζ is the sample mean of ζ scen , and cov (ζ scen ) is the covariance
matrix of ζ scen .
6.3.9 Robust Optimization
Based on the previous linear models, one can show that the stochastic constraints
(6.14a)-(6.14c), (6.15a)-(6.15c), (6.20), (6.24a), and (6.24b) can be written as:
Aζ ≤ B, (6.30)
A ∈ R(2nc nt (6+2ns )+4nnodes nt )×nζ ,
B ∈ R(2nc nt (6+2ns )+4nnodes nt )
where the matrices A and B can be defined to represent the stochastic

constraints mentioned above. To find a solution that remains feasible for
any realization of ζ in the uncertainty set U, we consider the following robust
counterpart of the stochastic constraint (6.30):
maxAζ 6 B (6.31)
ζ∈U
Using standard duality techniques, it can be shown that the constraint (6.31)
with a polyhedral uncertainty set U can be reformulated as the linear constraints
(6.32b)-(6.32d). We ask the readers to refer to lemma 2 in [204] for details
about robust counterpart of linear constraints with polyhedral uncertainty set.
Below is the complete optimization problem the offline controller solves on a
day-ahead basis:
X
min Tr zT zM − (RUPi + RDNi ) (6.32a)
i∈I
subject to:
Robust constraints :
πh ≤ B (6.32b)
πH ≥ A (6.32c)
π≥0 (6.32d)
Logic constraints: (6.25a) and (6.25b) (6.32e)
Other constraints: (6.14d)-(6.14h), (6.15d)-(6.15f) (6.32f)
The elements of the matrix π ∈ R(2nc nt (6+2ns )+4nnodes nt )×2nζ are auxiliary
variables.
SYSTEMS
6.3.10 Optimality Gap
In this subsection, we investigate the loss of optimality of the robust optimization

problem due to the use of linear control policies.
Robust optimization involving adaptive decision rules (control policies) is
generally computationally intractable. To reduce its complexity, in subsection
6.3.2 we restrict the decision variables to be linear function of the uncertain
data. In [178], an efficient method is proposed to estimate the approximation
error introduced by this rather drastic means of complexity reduction. We apply
the technique presented in [178] to estimate the optimality gap of the proposed
robust optimization problem (6.32a)-(6.32f).
To compute the optimality gap, one needs to compute a lower and upper
bounds on the optimal solution. The robust optimization problem (6.32a)-
(6.32f) provides an upper bound (i.e., a conservative approximation) on the
optimal solution, since it is formulated by reducing the underlying feasible
set. The duality theorem can be used to obtain a lower bound on the optimal
solution. The dual problem of the problem
X
i∈I
subject to:
Áζ + Θ (ζ) = 0 (6.33b)
Logic constraints: (6.25a) and (6.25b) (6.33c)
Other constraints: (6.14d)-(6.14h), (6.15d)-(6.15f) (6.33d)
can be derived as follows:

X
min max Tr zT zM − (RUPi + RDNi ) +
λ≥0
i∈I
h i
T
E λ (ζ) Áζ + Θ (ζ) (6.34a)
subject to:
Θ (ζ) ≥ 0 (6.34b)
Logic constraints: (6.25a) and (6.25b) (6.34c)
Other constraints: (6.14d)-(6.14h), (6.15d)-(6.15f) (6.34d)

where Áζ = Aζ − B. The elements of the vector Θ (ζ) are auxiliary variables,
and λ (ζ) is a vector of dual decisions. The dual feasible set of the problem
(6.34a)-(6.34d) can be restricted to combine only linear dual decisions. To this
end, we require the dual decisions to be representable as λ (ζ) = Λζ. Based on
this, the objective function (6.34a) can be reformulated as:
X
min max Tr zT zM − (RUPi + RDNi ) +
Λ≥0
i∈I
h i
E ζ T ΛT Áζ + Θ (ζ) (6.35a)
X
= min max Tr zT zM − (RUPi + RDNi ) +
Λ≥0
i∈I
h i
Tr ΛT E Áζ + Θ (ζ) ζ T (6.35b)
The maximization in (6.35b) can be carried out explicitly. This yields to the
following approximate problem.
X
i∈I
subject to:
h i
E Áζ + Θ (ζ) ζ T = 0 (6.36b)
Θ (ζ) ≥ 0 (6.36c)
Logic constraints: (6.25a) and (6.25b) (6.36d)
Other constraints: (6.14d)-(6.14h), (6.15d)-(6.15f) (6.36e)
It is proved in [178] that the problem (6.36a)-(6.36e) provides a lower bound

on the optimal solution. It can be shown that a tractable reformulation of this
problem can be given as follows:
X
i∈I
subject to:
Á + θ = 0 (6.37b)
H − hoT1 M θ T ≤ 0 (6.37c)

SYSTEMS
Logic constraints: (6.25a) and (6.25b) (6.37d)
Other constraints: (6.14d)-(6.14h), (6.15d)-(6.15f) (6.37e)

Here, o1 denotes a vector whose first element is 1 while all the others are 0.
The matrix H and the vector h are defined in subsection 6.3.1 to model the
uncertainty set U. The second-order moment matrix M defined in subsection
6.3.8 is used to approximate the expected value in (6.36b). This approximation
should respect the constraint (6.37c). The elements of the matrix θ are auxiliary
variables that respect the following relation:
h i
θM = E Θ (ζ) ζ T (6.38)
The optimality gap can be computed by comparing the solution of the problem
(6.32a)-(6.32f) (upper bound) to the solution of the problem (6.37a)-(6.37e)
(lower bound). We ask the reader to refer to [178] for an in-depth analysis of
the sub-optimality of robust optimization with linear decision rules.
6.4 Case Study
6.4.1 Simulated Network
We demonstrate the robust performance of the proposed control system on a real

low voltage Belgian network. This three-phase semi-urban radial distribution
network operates with a nominal voltage of 230/400 V. The network consists of
62 three-phase customers (nnodes = 62) and is depicted in Figure 6.7. All main
feeder cables are of type EAXVB 1 kV 4 × 150 mm2 (impedance: 0.206+j0.0778
Ω/km), except for the cable between node I and node II, which is of type
EAXVB 1 kV 4 × 95 mm2 (impedance: 0.320+j0.0778 Ω/km). The cables
connecting each household with the feeder are of type EXVB-Cu 1 kV 4 × 16
mm2 , with a fixed length of 30 m (impedance: 1.15+j0.0828 Ω/km).
The numbered nodes in the network are control nodes (nc = 31) and the other
nodes are passive nodes. The households’ loads, inverters and batteries between
II and III in Figure 6.7 belong to group G1, between II and IV belong to
group G2, between IV and V (IV and VI) belong to group G3, between IV and
VII belong to group G4 and between IV and VIII belong to group G5. The
parameters of each group are listed in Table 6.1.
230 V, resulting in v= 253 V and v = 207 V. The number of linear constraints
ns used to approximate the conic constraints (6.19a) and (6.19b) is set to 5.
CASE STUDY 149
Group G3
16 14 15
13
Control node
i V VI
Passive node
VII
IV
7 8 9 10 11 12 17 18 19 20 21 22
Grid Group G2 Group G4
II
I
III VIII
1 2 3 4 5 6 23 24 25 26 27 28 29 30 31
Group G1 Group G5
Figure 6.7: Schematic diagram of the network used in this case study.
6.4.2 The Loads, PV and Frequency Scenarios
For a weekday in June, we generate 104 daily profiles of households’ active

power consumption and PV power generation, with one-minute resolution. We
consider the mean of the daily profiles as the forecasted households’ active power
consumption and PV power generation, and the maximum of the generated
profiles as the maximum uncertain households’ active power consumption and
PV power generation. An open-source CREST DEMAND MODEL [183] is used
in this chapter to generate the households’ active power consumption profiles.
The daily PV profiles are generated based on the model presented in [184].
We use locally measured frequency data in the CE synchronous region with a
resolution of 1 minute over a period of four years (2015-2018), to generate 104
daily frequency profiles.
6.4.3 Simulation Software
The robust optimization problem (6.32a)-(6.32f) is solved using YALMIP [185]

toolbox with Mosek 9.0 [186]. The MATPOWER package 7.0 [187] for power flow
analysis is used to compute voltage profiles of network’s nodes with and without
the proposed control system. 15-minute time step resolution is considered in
Yalmip simulations. To have daily profiles with 15-minute resolution, we take
the average over 15 minutes for each one-minute resolution profile. One-minute
SYSTEMS
Table 6.1: Parameters of inverters, batteries and loads.
G1 G2 G3 G4 G5
sinv (kVA) 5 8 10 12 15
ebat (kWh) 4.8 7.2 8.2 9.8 13.5
ebat (kWh) 0.96 1.44 1.64 1.96 2.7
Ebat
0
(kWh) 2.4 3.6 4.1 4.9 6.75
pbat (kW) 2 3 3.3 5 7
pbat (kW) -2 -3 -3.3 -5 -7
Ncycle life 3500 3500 3500 5000 5000
DoD (%) 80 80 80 80 80
creplacementi () 2400 3600 4100 4900 6750
cp (cent/kWh) 8.92 8.92 8.92 6.25 6.25
cq (cent/kvarh) 0.29 0.29 0.29 0.29 0.29

√ √ √ √ √
ch
ηi 0.9 0.9 0.9 0.9 0.9
√ √ √ √ √
dis
ηi 0.9 0.9 0.9 0.9 0.9
Load kW-peak 3.5 5.6 7 8.4 10.5
Power factor 0.85 0.85 0.85 0.85 0.85
PV kW-peak 4.8 7.68 9.6 11.52 13.44
time step resolution is considered in MATPOWER simulations, therefore, one

active control policy, one reactive control policy, one R-DN energy management
policy, and one R-UP energy management policy are considered for each 15
minutes simulation in MATPOWER. We select 5000 daily profiles (of loads,
PV and frequency) randomly as training data (offline computation). All the
generated profiles are considered in the validation (MATPOWER simulations).
CASE STUDY 151
6.4.4 Simulation Results
Figure 6.8 shows aggregated power capacities of the 31 BESSs reserved for
the downward-FCR, upward-FCR, voltage rise, energy management policies
of R-DN and R-UP batteries. The upward reserve capacity is defined as a
positive value in (6.15d); it is presented as a negative power in Figure 6.8 to
compare it to the maximum discharging power. One can notice that the upward
reserve capacity is zero from 10:00 till 14:30, this is because VC batteries are
expected to be charged over this period. The highlighted positive power area
between 10:00 and 14:30 shows the aggregated power capacity reserved for
solving voltage rise problems. Figure 6.9, for example, shows the power capacity
reserved for charging the VC battery 31 over the period 10:00-14:30. The
aggregated downward reserve capacity is at minimum at 10:45, since most of the
inverters capacity is occupied by the PV power, VC batteries power, and reactive
power. The highlighted negative power area in Figure 6.8 shows the aggregated
power capacity reserved for the management policy of R-DN batteries. The
highlighted positive area between 00:00 and 10:00 and between 14:30 and 24:00
shows the aggregated power capacity reserved for the management policy of
R-UP batteries. There is no power capacity reserved for the management policy
of R-UP batteries between 10:00 and 14:30. Reactive power is expected to solve
the minor voltage drop problem shown in Figure 6.10a, therefore, there is no
power capacity reserved for solving voltage drop problems.
Voltage profiles are computed for the 104 scenarios without activating the
voltage and frequency control system. Figure 6.10a shows voltage profiles of
control node 31 (phase to neutral voltage). One can see that most of voltage
profiles of different scenarios exceed the maximum voltage limit. When linear
voltage control policies are applied, with frequency control enabled, it is noticed
that the existing violations of the over-voltage limit are eliminated for the 104
scenarios, which demonstrates the robustness of the proposed voltage control
policies. Figure 6.10b shows voltage profiles of control node 31 (at the end of the
feeder) after applying voltage control policies, with frequency control enabled.
Figure 6.11a and Figure 6.11b show, respectively, the apparent power magnitude
of inverter 31 and energy content of battery 31 for all the scenarios. One can
see that the capacity limits of the inverter and battery are respected for all
the scenarios. The R-UP and R-DN energy management policies succeed in
maintaining the energy content of battery 31 within limits while providing
downward-FCR and upward-FCR, which demonstrates the robustness of the
proposed energy management policies.
SYSTEMS
Aggregated downward reserve capacity

R-UP management policy
Aggregated upward reserve capacity
R-DN management policy
Aggregated power capacity of voltage rise
Figure 6.8: Aggregated power capacities of the downward-FCR, upward-FCR,

voltage rise, energy management policies of R-DN and R-UP batteries.
Figure 6.9: Power capacity reserved for charging the VC battery 31.
6.4.5 Loss of Optimality
We compute upper bounds on the optimal solutions of the 104 scenarios by

solving the robust optimization problem (6.32a)-(6.32f). We solve the dual
problem (6.37a)-(6.37e) for the 104 scenarios to find the lower bounds. The
optimality gaps of the different scenarios are computed by comparing the lower
CASE STUDY 153
(a) (b)
Figure 6.10: (a) Voltage profiles of node 31 with no control, (b) Voltage profiles
of node 31 with voltage and frequency control. The dotted lines show the
maximum and minimum voltage limits
(a) (b)
Figure 6.11: (a) Apparent power magnitude of inverter 31, (b) Energy content
of battery 31.
bounds to the upper bounds. Results show that the optimality gap of the
proposed robust control methodology varies between 8% and 13% for the 104
scenarios, which is an acceptable optimality gap. We would like to remind the
reader that obtaining an optimal solution is intractable. Hence, we trade-off
optimality for computational tractability.
SYSTEMS
6.5 Conclusion
In this chapter, we complement previous work on BESSs providing multiple

services by designing a novel control system for residential PV-battery systems to
provide simultaneously FCR service to their synchronous area, and distributed
voltage control service to their distribution network. We are interested in
combining voltage control with frequency control, as extending the control
framework of PV-battery systems that provide voltage control to provide FCR
helps these systems generate additional revenues from FCR. On the other hand,
extending the control framework of BESSs that provide frequency control to
provide voltage control helps these systems maximize their frequency reserve
capacity without causing the grid violating voltage operation limits.
In this chapter, we develop a novel robust mathematical optimization program to
schedule the energy and power budgets of PV-battery systems to the voltage and
frequency control, and learn a set of linear control policies. The optimization
problem is solved on a day-ahead basis. In real-time, grid voltage support
functions of smart inverters apply voltage control policies to compute, in a P2P
fashion, the amount of reactive power and batteries power needed to maintain
voltages within accepted limits. Frequency support functions of smart inverters
compute FCR powers, in real-time, based on frequency measurements and the
upward and downward reserve capacity profiles (computed through the solution
of the optimization problem). Frequency support functions also apply linear
energy management policies to maintain energy contents of BESSs within limits
while providing FCR. Simulations over 104 scenarios are used to demonstrate
the robustness of the proposed control system.
Chapter 7
General Conclusions and

Future Work
7.1 Summary and Contributions
The increasing interest in deploying photovoltaic (PV) systems in distribution

networks presents major challenges from the viewpoints of reliable operation
and control. One of the main challenges is the voltage rise problem due to the
feed-in power from the PV systems. Fortunately, PV inverters could, under the
correct operational control, provide numerous benefits to distribution networks,
including active power control and reactive power compensation. Applying these
features properly can positively influence the grid voltage. The main focus of
this thesis is to develop active voltage control systems provided by PV and
PV-battery inverters to mitigate or eliminate voltage problems of distribution
networks with high PV penetration.
There are mainly three strategies for implementing active voltage control:
centralized control; local control; and distributed control. In a centralized
voltage control system, a global optimization problem is formulated and solved
by a centralized controller to determine optimal control decisions for the
overall system. In contrast, a local voltage control system sacrifices optimality
in favour of less computational burden. This system does not rely on any
form of communication, as control decisions are based only on the available
local information. Generally, centralized voltage control systems have poor
reliability and scalability, and local voltage control systems suffer from degraded
performance. To overcome the drawbacks of centralized and local voltage control
155
156 GENERAL CONCLUSIONS AND FUTURE WORK
systems, distributed voltage control (DVC) systems are implemented in this

thesis. In DVC systems, control functionality of the centralized controller is
distributed among intelligent agents (smart inverters); local control decisions of
each agent are computed using only local measurements augmented with limited
information from other agents through a peer-to-peer (P2P) communication.
The thesis proposes a new concept of grid support functions of smart inverters
called P2P-based grid voltage support functions (GVSFs). Smart inverters
equipped with P2P-based GVSFs interact with each other in a P2P fashion
and form a P2P-based DVC system. Two novel methodologies are proposed
in this thesis to design a P2P-based DVC system. The first one is based on
real-time distributed optimization, while the second one is based on offline
robust optimization.
In the first methodology (Chapter 3), two gossip-based decomposition techniques
are developed to implement a real-time optimization-based DVC system: the
fully distributed Dual Decomposition (DD) method and the Jacobi-Proximal
Alternating Direction Method of Multipliers (JP-ADMM). The fully distributed
DD algorithm allows for simple implementation, but in most cases requires
significantly more iterations to converge. The JP-ADMM algorithm requires
more local computations and communication; nevertheless, it converges much
faster than the DD algorithm. To better understand the real implementation of
DVC and experimentally validate the performance of the real-time optimization-
based DVC system, the thesis develops a novel laboratory-based P2P voltage
control testbed in Chapter 4.
In the second methodology (Chapter 5), the thesis simplifies the implementation
of a DVC system by applying robust optimization to learn linear voltage
control policies on a day-ahead basis. This methodology depends on the
availability of forecast data and forecast error of residential demand profiles
and PV generation. Thus, the non-availability of these data leaves us with one
option for implementing a DVC system: distributed optimization.
In the final phase of this work, the thesis studies in Chapter 6 how to combine
policy-based DVC system with frequency containment reserve (FCR), to increase
value from PV-battery systems.
7.2 Answers to the Research Questions
This section summarizes the answers the thesis provides to the research questions
posed in Section 1.2.
ANSWERS TO THE RESEARCH QUESTIONS 157
Chapter 2
• How P2P-based control can be imposed on the distribution level of the

grid?
• How the thesis work contributes to the proposed P2P-based control paradigm
of distribution networks?
To eliminate the problems that a more centralized control method possesses,

having a single point of failure, this thesis adopts P2P-based control techniques.
Additionally, as distributed energy resources (DERs) are typically highly
distributed in distribution networks, operated by different owners and with
different objectives, it is desirable that control of distribution networks operates
in a highly distributed way as well. In this perspective, a P2P-based control
paradigm is proposed in the thesis.
It is impossible to impose this paradigm on the whole distribution grid, as it
incorporates thousands of DERs that are geographically very dispersed. To deal
with this complexity, the thesis proposes to break the complete distribution grid
down into a smaller P2P-based microgrids, containing only a limited amount
of agents. Such an agent could be a PV system, a group of intelligent loads,
a digital substation, or any other form of DERs. With P2P communication
and local intelligence, agents are able to communicate with each other and
operate their microgrid in a distributed manner. To ensure overall system
stability, P2P-based microgrids can coordinate themselves through coupling
agents; substations would be good candidates for such coupling agents.
The thesis contributes to the P2P-based control paradigm by developing novel
P2P-based grid voltage and frequency support functions for smart inverters.
The proposed functions enables PV and PV-battery systems to carry out a
distributed voltage control in each microgrid, and participate in supporting the
grid frequency.
Chapter 3
• To which extent can the number of inverters participating in the DD-based

DVC system be increased?
In Subsection 3.5.11, the thesis studies convergence speed of the DD-based DVC
system. The thesis performs 104 numerical tests on the simulated network of
Figure 3.4. For each test, the number of smart inverters participating in voltage
control varies from 10 to 60. Results show that the average convergence speed
of the different experiments varies linearly from 8.4 minutes (10-agent case)
to 65 minutes (60-agent case). Although convergence speed does not increase

exponentially, the algorithm is not fast enough to be implemented for real-time
voltage control with many agents. In general, the algorithm can be used to
implement a DVC system with only few agents (e.g. 10 agents or less).
• What is the effect of extending the 2-block ADMM to parallel multi-block

ADMM on the convergence speed of the ADMM algorithm?
One of the main findings in the thesis is that the direct extension of the sequential
2-block ADMM to parallel multi-block ADMM (the so-called Jacobi ADMM)
without additional modifications or assumptions slows down the convergence
significantly. As demonstrated in Subsection 3.7.2, Jacobi ADMM is even slower
than the DD algorithm.
• How the convergence speed of the ADMM-based DVC algorithm can be

accelerated?
The ADMM-based DVC algorithm can be accelerated by adding the proximal

2 2
(k) (k−1) (k) (k−1)
terms ∆qinvi − ∆qinvi and ∆pinvi − ∆pinvi for each sub-problem (see
Equation (3.34)), and acceleration factors for the update of the Lagrangian
multipliers (see Section 3.6.2). Two acceleration factors are proposed in this
thesis: one for accelerating the convergence when reactive power compensation
is used, and the other for accelerating the convergence when active power
curtailment is used. The ADMM with the aforementioned additions is called
Jacobi-Proximal ADMM (JP-ADMM). The convergence of the JP-ADMM is
guaranteed, as proved in [138], by respecting Relation (3.35). As demonstrated
in Subsection 3.7.2, the JP-ADMM-based DVC algorithm with 50 agents needs
around 1.25 minutes to converge, whereas for the same number of agents, the
DD-based DVC algorithm needs around 59.4 minutes to converge and the Jacobi
ADMM-based DVC algorithm needs around 86.6 minutes to converge.
Chapter 4
• How to validate experimentally a real-time optimization-based distributed

voltage control?
To validate experimentally the proposed real-time optimization-based DVC

system, the thesis develops a laboratory-based microgrid equipped with
programmable inverters, P2P communication capabilities, and data acquisition.
ANSWERS TO THE RESEARCH QUESTIONS 159
The P2P-based microgrid consists of three types of agents: P2P-based PV

systems; P2P-based voltmeters; and a monitor.
A P2P-based PV system is emulated by a programmable inverter connected
to a DC power supply. Each programmable inverter is operated through
MATLAB/Simulink running on a computer. The computer communicates over
Ethernet with an on-board PC-based Real Time Target, which controls the
power electronics of the inverter. The thesis implements a state-of-the-art
current control loop, Phase-Locked-Loop, and Kalman filter to drive each
inverter. The thesis then integrates the proposed P2P-based GVSF with the
inner control system of each inverter. LoRaWAN-based device-to-device (D2D)
communication module is integrated with the P2P-based GVSF of each inverter.
The P2P-based voltmeter is connected to a D2D communication module. The
voltmeter periodically measures the voltage of its bus and calculates control
signals (Lagrangian multipliers). The inverters communicate with the voltmeters
in a P2P fashion to receive the control signals. The inverters then determine
how to react based on these control signals. Inverters calculate the change in
reactive power and active power needed to maintain voltages within limits by
solving the DD-based distributed optimization problem. The DD method is
used in the setup and not the JP-ADMM because the setup consists of few
agents.
The monitor is built to visualize and record the data. This agent hosts a web
service, through which all recorded data are visualized in several dashboards.
Experimental results show the ability of the real-time optimization-based DVC
to solve voltage rise problems within 3 minutes.
Chapter 5
• How forecast data of residential active power consumption and PV power

generation can be used to compute robust linear voltage control policies?
In Chapter 5, the thesis develops a methodology for implementing a policy-based

DVC system. The control methodology is structured in two phases. In the day-
ahead phase, a voltage control problem is formulated as a chance-constrained
stochastic optimization problem, where the control actions are expressed as a
linear function of the uncertain active power consumption and PV generation.
The aim of the chance-constrained stochastic optimization problem is to compute
robust linear voltage control policies that minimize the expected cost of operating
the voltage control system, guarantees (with a predefined probability) that
voltages will not exceed pre-defined limits, and inverters/batteries constraints
are respected. The thesis applies robust optimization technique to construct a
tractable approximation to the chance-constraints. In the real-time phase, the

inverters communicate with each other and apply the linear policies to compute
the amount of reactive power and batteries power needed to maintain voltages
within limits.
Robust optimization transforms the complex chance constraints into simple
robust linear constraints (robust counterpart formulation). To attain a robust
solution, robust optimization looks for solutions which are feasible for any
realization of the uncertain active power consumption and PV generation in a
predefined uncertainty set. To model the uncertainty set, as detailed in Section
5.3.2, forecast data and forecast error are needed. Hence, the thesis uses forecast
data of active power consumption and PV generation to formulate a distributed
voltage control problem as a robust optimization problem, which is then solved
to compute a set of linear voltage control policies.
Chapter 6
• How to design a control methodology that optimally combines policy-based

distributed voltage control with frequency containment reserve?
In chapter 6, the thesis implements a control methodology for PV-battery

systems to simultaneously provide voltage and frequency control. The control
methodology consists of two phases. In the day-ahead phase, a robust
optimization problem is formulated. The aims of this problem are to allocate
fractions of the energy and power capacity of each battery to the two services,
compute a set of linear control policies, minimize the expected cost of reactive
power compensation and batteries degradation, and maximize profits from
frequency control. The robust optimization problem also aims to immunize
against service unavailability, and violating batteries, inverters and voltage
constraints. In the real-time phase, the grid support functions of smart PV-
battery inverters apply the control policies to regulate voltage profiles, and keep
energy contents of batteries within limits while providing frequency control.
Logic constraints are implemented to avoid a situation in which the voltage
control system requires batteries be charged at the same time that the frequency
control requires discharging the same batteries (or vice versa).
The real-time control is operated by two grid support functions, the GVSFs and
the frequency support functions (FSFs). The GVSFs host the voltage control
policies, whereas the FSFs compute the FCR power and host linear management
policies that maintain energy contents of batteries within limits while providing
frequency control.
RECOMMENDATIONS TO STAKEHOLDERS 161
7.3 Recommendations to Stakeholders
Inverter Manufacturers
Today’s smart inverter are equipped with communication-less local grid support
functions or centralized grid support functions or both. As has been discussed
in this thesis, local grid support functions suffer from degraded performance and
centralized grid support functions have poor reliability and scalability. Hence,
it is recommended for inverter manufacturers to place emphasis on the design,
development and experimental validation of P2P-based grid support functions.
Distribution System Operators
Future distribution networks are expected to host thousands of DERs and

flexible loads. Consequently, communication of distribution system operators
(DSOs) with each distributed unit (to manage and control it) is out of reach,
due to the high costs that the technology capability would entail. Distributed
management is perhaps the only viable strategy that can potentially overcome
the complexity of operating future distribution networks, and here is where the
P2P-based control paradigm (proposed in Chapter 2) comes into the picture.
Therefore, it is recommended for the DSOs to adopt the P2P-based paradigm
to gradually replace the centralized control and management systems. One way
to build P2P-based distribution networks is by mandating DERs and flexible
loads to comply with certain P2P-based communication and local intelligence
requirements. By optimally managing DERs and smart loads in a P2P fashion,
DSOs will be able to solve voltage problems, avoid congestion, participate in
balancing the grid, and defer or avoid costly network investments.
Transmission System Operators
In many countries, Transmission System Operators (TSOs) procure their FCR

capacity via a periodic auctions in which third parties can bid a certain amount
of reserve capacity at a certain price. As has been demonstrated in Chapter 6,
battery energy storage systems located in distribution networks with voltage
problems cannot commit to provide a fixed reserve capacity over the entire
contracted period. Moreover, they cannot commit to provide the same reserve
capacity for the upward-FCR and downward-FCR. This is because FCR capacity
of assets located in distribution networks (with high PV penetration) is affected
by the magnitude of voltages. Therefore, it is recommended for TSOs to design
a new FCR auction, in which FCR capacity can be auctioned daily in the form
of 15 minutes products. Additionally, the upward and downward reserves are

recommended to be procured independently.
Providing FCR service with several assets connected to distribution networks
can lead to voltage issues. TSOs and DSOs should collaborate to investigate
how voltage problems caused by FCR assets can be solved. In the author’s
view, TSOs and DSOs should require FCR providers (with assets located in
distribution networks) to provide FCR simultaneously with voltage control;
FCR providers should dedicate part of their reserve capacity to solve voltage
problems caused by their assets. Hence, it is recommended to mandate grid
voltage support as one of the requirements a market player (with assets located
in distribution networks) has to respect to become a qualified FCR provider.
Control Software Developers
Based on our experience with distributed voltage control, using real-time

distributed optimization to coordinate several agents, participating in voltage
control, is a challenging task. Therefore, it is recommended for the control
software developers, interested in the implementation of distributed voltage
control algorithms, to use forecast data to coordinate agents on an offline basis
(e.g. day-ahead), by learning a set of linear control policies. The aim is to design
a distributed voltage control system in which agents solve in real-time in a P2P
fashion a set of linear equations, rather than solving a complex optimization
problem.
If forecast data are not available, then it is recommended to use the ADMM-
based decomposition techniques (e.g. the Jacobi-Proximal ADMM), as they
can be implemented with fast convergence. In this case, attention has to be
paid to the tuning of the penalization, acceleration and step size parameters.
Inappropriate settings of these parameters will degrade the performance of the
algorithms.
Standardization Bodies
To enable DERs, electric vehicles, and smart loads to interact with each other
and control distribution networks in a P2P fashion, appropriate regulatory
frameworks must be developed first. Hence, it is recommended for the
standardization bodies to develop technical specifications and amend grid codes
for the provision of P2P-based control and management services. Roles and
responsibilities of the DSOs and prosumers in the P2P-based paradigm should
be clearly defined.
FUTURE WORK 163
7.4 Future Work
Robustness to Communication Failures and Delays
Performance of the proposed distributed voltage control algorithms depends on

the availability and quality of the P2P communication between agents. The
thesis tests the performance of the proposed algorithms assuming communication
performs perfectly. It is essential to quantify the impact of communication
failures and delays on the performance of the proposed algorithms, and provide
solutions to cope with communication failures and delays. These issues are out
of scope of this thesis, hence recommended for further research.
Robustness to the Quality of Data
The proposed algorithms depend on the quality of data exchanged between

agents. It is interesting to study the behaviour of the algorithms if the data is
incorrect, missing, or tampered with. Future research should come up with a
solution to cope with these problems.
Voltage Unbalance
This thesis focuses on solving over-voltages and under-voltages of distribution

networks with three-phase PV inverters. Distribution networks with single-phase
PV inverters are not considered in the case studies of this thesis.
Different amounts of penetration of single-phase PV inverters in different phases
of a distribution network may result in unbalanced voltages. For example, one
of the phases may experience reverse power flow and over-voltage problem, while
the other two phases may not be affected at all. Moreover, uneven distribution
of single-phase loads or random usage of these loads could further increase the
voltage unbalance. Extending the objectives of the proposed distributed voltage
control algorithms to solve voltage unbalance problems is recommended for
future research.
Distribution Networks with Incomplete Data
The prosed algorithms of the thesis require perfect knowledge of feeder

characteristics, such as network topology and line segment impedances. The
information that DSOs have about their networks is usually limited. Therefore, it
is highly recommended for future research to augment the proposed algorithms
with machine learning techniques to learn the topology and impedances of

distribution networks, or to directly learn voltage sensitivities of the linearized
branch flow model (see Section 3.3).
Voltage Regulation Market
Market design for the P2P-based voltage control is highly recommended

for future research. There are, at the moment, no markets for P2P-based
voltage control in distribution networks. Establishing an appropriate market
framework is a prerequisite for enabling the P2P-based voltage control. Moreover,
the willingness of prosumers to participate in the P2P-based voltage control
essentially depends on the economic incentives they receive, i.e., the existence
of a business case and the benefits it brings. Hence, there is a need of designing
and developing market and business models for the P2P-based voltage control.
The models should clearly define how the P2P voltage control technology can
be used in a real world.
P2P-Based Platform
This thesis focuses on developing grid support functions for PV and PV-battery
systems. An interesting future research would be to investigate how the proposed
grid support functions can be extended to build a scalable P2P-based platform
(Figure 7.1) that integrates renewable energy sources, energy storage systems,
smart controllable loads, electric vehicles, on-load tap changer, etc. Enabling
different types of DERs and network assets to communicate with each other in
a P2P fashion would create a very powerful control system.
Validation in Real-Life Applications
Chapter 4 shows the validation of the P2P-based grid support functions in

a laboratory environment. Test conditions may vary substantially between
laboratory testbed systems and real-life applications. It is recommended for
future work to validate the obtained results in real-life environments and
applications. A good starting point can be based on the testing procedures
proposed in [142, 205]. In [142], demonstrations were performed in the Meltemi
community smart grid pilot site. The work is part of the EU FP7 project
DREAM. In [205], a dual decomposition-based transactive control system
has been deployed in residential buildings in the Netherlands, to validate its
performance under real-life conditions. The work is part of the EU-funded
REnnovates project.
FUTURE WORK 165
Grid
On-load
tap-changer
Controlling active GSF Adjusting tap changing

and reactive power transformer
GSF
Inverter Controlling
GSF active power
Solar panel
P2P
Inverter
Controlling smart
loads Controlling active GSF
GSF Battery
and reactive power
Inverter
Smart House Wind

turbine
Figure 7.1: P2P-based platform (GSF: grid support function).

Bibliography
[1] International Energy Agency, “Trends in PV applications 2019.”

https://iea-pvps.org/wp-content/uploads/2020/02/5319-iea-pvps-report-
2019-08-lr.pdf. Accessed: 05-05-2020.
[2] International Renewable Energy Agency, “Future of solar photo-
voltaic.” https://www.irena.org/publications/2019/Nov/Future-of-Solar-
Photovoltaic. Accessed: 05-05-2020.
[3] Cenelec, “European standard EN 50160 on power quality: Voltage
characteristics of electricity supplied by public distribution networks,”
2007.
[4] M. Nijhuis, M. Gibescu, and J. Cobben, “Assessment of the impacts of
the renewable energy and ICT driven energy transition on distribution
networks,” Renewable and Sustainable Energy Reviews, vol. 52, pp. 1003–
1014, 2015.
[5] Y. Zhang, F. Li, Z. Hu, and G. Shaddick, “Quantification of low voltage
network reinforcement costs: A statistical approach,” IEEE Transactions
on Power Systems, vol. 28, no. 2, pp. 810–818, 2012.
[6] T. Stetz, M. Kraiczy, M. Braun, and S. Schmidt, “Technical and
economical assessment of voltage control strategies in distribution grids,”
Progress in Photovoltaics: Research and Applications, vol. 21, no. 6,
pp. 1292–1307, 2013.
[7] T. Fawzy, D. Premm, B. Bletterie, and A. Goršek, “Active contribution
of PV inverters to voltage control—From a smart grid vision to full-scale
implementation,” e & i Elektrotechnik und Informationstechnik, vol. 128,
no. 4, pp. 110–115, 2011.
[8] Y. Xue, M. Starke, J. Dong, M. Olama, T. Kuruganti, J. Taft, and
M. Shankar, “On a future for smart inverters with integrated system
167
168 BIBLIOGRAPHY
functions,” in 2018 9th IEEE International Symposium on Power

Electronics for Distributed Generation Systems (PEDG), pp. 1–8, IEEE,
2018.
[9] D. Dong, M. S. Agamy, M. Harfman-Todorovic, X. Liu, L. Garces, R. Zhou,

and P. Cioffi, “A PV residential microinverter with grid-support function:
Design, implementation, and field testing,” IEEE Transactions on Industry
Applications, vol. 54, no. 1, pp. 469–481, 2017.
[10] A. Hoke, S. Chakraborty, and T. Basso, “Testing advanced photovoltaic

inverters conforming to IEEE standard 1547-Amendment 1,” in 2014
IEEE 40th Photovoltaic Specialist Conference (PVSC), pp. 1014–1021,
IEEE, 2014.
[11] J. Smith, W. Sunderman, R. Dugan, and B. Seal, “Smart inverter volt/var
control functions for high penetration of PV on distribution systems,”
in 2011 IEEE/PES Power Systems Conference and Exposition, pp. 1–6,
IEEE, 2011.
[12] X. Zhao, L. Chang, R. Shao, and K. Spence, “Power system support
functions provided by smart inverters—A review,” CPSS Transactions on
Power Electronics and Applications, vol. 3, no. 1, pp. 25–35, 2018.
[13] T. Basso, S. Chakraborty, A. Hoke, and M. Coddington, “IEEE 1547

Standards advancing grid modernization,” in 2015 IEEE 42nd Photovoltaic
Specialist Conference (PVSC), pp. 1–5, IEEE, 2015.
[14] Z. Li, R. Bai, W. Zheng, Y. Zhi, X. Zhang, S. Xun, G. Hu, Y. Shen, H. Xiao,
and D. Wu, “Research of Voltage Control Strategy for Distribution
Network with PV Connected,” in 2019 IEEE 2nd International Conference
on Electronics Technology (ICET), pp. 446–450, IEEE, 2019.
[15] H. Sun, Q. Guo, J. Qi, V. Ajjarapu, R. Bravo, J. Chow, Z. Li, R. Moghe,
E. Nasr-Azadani, U. Tamrakar, et al., “Review of challenges and research
opportunities for voltage control in smart grids,” IEEE Transactions on
Power Systems, vol. 34, no. 4, pp. 2790–2801, 2019.
[16] P. N. Vovos, A. E. Kiprakis, A. R. Wallace, and G. P. Harrison,
“Centralized and distributed voltage control: Impact on distributed
generation penetration,” IEEE Transactions on Power Systems, vol. 22,
no. 1, pp. 476–483, 2007.
[17] N. Itaya, “Power-distribution-system voltage control system, power-

distribution-system voltage control method, and centralized voltage control
apparatus,” July 9 2019. US Patent 10,345,842.
BIBLIOGRAPHY 169
[18] K. De Brabandere, B. Bolsens, J. Van den Keybus, A. Woyte, J. Driesen,

and R. Belmans, “A voltage and frequency droop control method for
parallel inverters,” IEEE Transactions on Power Electronics, vol. 22,
no. 4, pp. 1107–1115, 2007.
[19] J. Von Appen, T. Stetz, M. Braun, and A. Schmiegel, “Local voltage
control strategies for PV storage systems in distribution grids,” IEEE
Transactions on Smart Grid, vol. 5, no. 2, pp. 1002–1009, 2014.
[20] F. Guo, C. Wen, and Y.-D. Song, Distributed control and optimization
technologies in smart grid systems. CRC Press, 2017.
[21] Z. Cheng, J. Duan, and M.-Y. Chow, “To centralize or to distribute:
That is the question: A comparison of advanced microgrid management
systems,” IEEE Industrial Electronics Magazine, vol. 12, no. 1, pp. 6–24,
2018.
[22] G. Qu and N. Li, “Optimal distributed feedback voltage control under
limited reactive power,” IEEE Transactions on Power Systems, vol. 35,
no. 1, pp. 315–331, 2019.
[23] D. K. Molzahn, F. Dörfler, H. Sandberg, S. H. Low, S. Chakrabarti,
R. Baldick, and J. Lavaei, “A survey of distributed optimization and
control algorithms for electric power systems,” IEEE Transactions on
Smart Grid, vol. 8, no. 6, pp. 2941–2962, 2017.
[24] Y. Wang, S. Wang, and L. Wu, “Distributed optimization approaches for
emerging power systems operation: A review,” Electric Power Systems
Research, vol. 144, pp. 127–135, 2017.
[25] J. Lavaei, S. Low, R. Baldick, B. Zhang, D. Molzahn, F. Dörfler, and
H. Sandberg, “Guest editorial distributed control and efficient optimization
methods for smart grid,” IEEE Transactions on Smart Grid, vol. 8, no. 6,
pp. 2939–2940, 2017.
[26] K. E. Antoniadou-Plytaria, I. N. Kouveliotis-Lysikatos, P. S. Georgilakis,
and N. D. Hatziargyriou, “Distributed and decentralized voltage control
of smart distribution networks: Models, methods, and future research,”
IEEE Transactions on Smart Grid, vol. 8, no. 6, pp. 2999–3008, 2017.
[27] H. J. Liu, W. Shi, and H. Zhu, “Distributed voltage control in distribution
networks: Online and robust implementations,” IEEE Transactions on
Smart Grid, 2017.
[28] P. Šulc, S. Backhaus, and M. Chertkov, “Optimal distributed control of
reactive power via the alternating direction method of multipliers,” IEEE
Transactions on Energy Conversion, vol. 29, no. 4, pp. 968–977, 2014.
170 BIBLIOGRAPHY
[29] J. Li, Z. Xu, J. Zhao, and C. Zhang, “Distributed online voltage control
in active distribution networks considering PV curtailment,” IEEE
Transactions on Industrial Informatics, vol. 15, no. 10, pp. 5519–5530,
2019.
[30] I. Ranaweera, O.-M. Midtgård, and M. Korpås, “Distributed control
scheme for residential battery energy storage units coupled with PV
systems,” Renewable Energy, vol. 113, pp. 1099–1110, 2017.
[31] M. Zeraati, M. E. H. Golshan, and J. M. Guerrero, “Distributed control

of battery energy storage systems for voltage regulation in distribution
networks with high PV penetration,” IEEE Transactions on Smart Grid,
vol. 9, no. 4, pp. 3582–3593, 2016.
[32] C. A. Hill, M. C. Such, D. Chen, J. Gonzalez, and W. M. Grady,
“Battery energy storage for enabling integration of distributed solar power
generation,” IEEE Transactions on Smart Grid, vol. 3, no. 2, pp. 850–857,
2012.
[33] G. Fitzgerald, J. Mandel, J. Morris, and H. Touati, “The Economics of
Battery Energy Storage: How multi-use, customer-sited batteries deliver
the most services and value to customers and the grid,” Rocky Mountain
Institute, p. 6, 2015.
[34] S. P. Forrester, A. Zaman, J. L. Mathieu, and J. X. Johnson, “Policy
and market barriers to energy storage providing multiple services,” The
Electricity Journal, vol. 30, no. 9, pp. 50–56, 2017.
[35] Y. Tian, A. Bera, M. Benidris, and J. Mitra, “Stacked revenue and

technical benefits of a grid-connected energy storage system,” IEEE
Transactions on Industry Applications, vol. 54, no. 4, pp. 3034–3043, 2018.
[36] S. J. Hossain, R. Bhattarai, M. Ahmed, S. Abdelrazek, and
S. Kamalasadan, “Operational cost value assessment and value based
stacked energy storage management for active power distribution systems,”
in 2017 IEEE Industry Applications Society Annual Meeting, pp. 1–8,
IEEE, 2017.
[37] J. Engels, B. Claessens, and G. Deconinck, “Techno-economic analysis and
optimal control of battery storage for frequency control services, applied
to the German market,” Applied Energy, vol. 242, pp. 1036–1049, 2019.
[38] H. Almasalma, J. Engels, and G. Deconinck, “Peer-to-peer control of

microgrids,” in 8th IEEE Benelux PELS/PES/IAS Young Researchers
Symposium, pp. 1–6, 2016.
BIBLIOGRAPHY 171
[39] H. Almasalma, J. Engels, and G. Deconinck, “Dual-decomposition-based

peer-to-peer voltage control for distribution networks,” CIRED-Open
[40] H. Almasalma, S. Claeys, and G. Deconinck, “Peer-to-peer-based
[41] H. Almasalma, S. Claeys, K. Mikhaylov, J. Haapola, A. Pouttu, and
G. Deconinck, “Experimental validation of peer-to-peer distributed voltage
control system,” Energies, vol. 11, no. 5, p. 1304, 2018.
[42] H. Almasalma and G. Deconinck, “Robust Policy-Based Distributed
Voltage Control Provided by PV-Battery Inverters,” IEEE Access, vol. 8,
pp. 124939–124948, 2020.
[43] H. Almasalma and G. Deconinck, “Simultaneous Provision of Voltage and
Frequency Control by PV-Battery Systems,” IEEE Access, vol. 8, 2020.
[44] R. O’Gorman and M. Redfern, “Enhanced autonomous control of
distributed generation to provide local voltage control,” in 2008 IEEE
Power and Energy Society General Meeting-Conversion and Delivery of
Electrical Energy in the 21st Century, pp. 1–8, IEEE, 2008.
[45] J. P. Lopes, N. Hatziargyriou, J. Mutale, P. Djapic, and N. Jenkins,
“Integrating distributed generation into electric power systems: A review
of drivers, challenges and opportunities,” Electric power systems research,
vol. 77, no. 9, pp. 1189–1203, 2007.
[46] R. H. Lasseter and P. Paigi, “Microgrid: A conceptual solution,” in 2004
IEEE 35th Annual Power Electronics Specialists Conference (IEEE Cat.
No. 04CH37551), vol. 6, pp. 4285–4290, IEEE, 2004.
[47] F. Katiraei, R. Iravani, N. Hatziargyriou, and A. Dimeas, “Microgrids
management,” IEEE Power and Energy Magazine, vol. 6, no. 3, pp. 54–65,
2008.
[48] R. H. Lasseter, “Smart distribution: Coupled microgrids,” Proceedings of
the IEEE, vol. 99, no. 6, pp. 1074–1082, 2011.
[49] N. Hatziargyriou, Microgrids: architectures and control. John Wiley &
Sons, 2014.
[50] D. E. Olivares, A. Mehrizi-Sani, A. H. Etemadi, C. A. Cañizares, R. Iravani,
M. Kazerani, A. H. Hajimiragha, O. Gomis-Bellmunt, M. Saeedifard,
R. Palma-Behnke, et al., “Trends in microgrid control,” IEEE Transactions
on Smart Grid, vol. 5, no. 4, pp. 1905–1919, 2014.
172 BIBLIOGRAPHY
[51] K. De Brabandere, K. Vanthournout, J. Driesen, G. Deconinck, and

R. Belmans, “Control of microgrids,” in 2007 IEEE Power Engineering
Society General Meeting, pp. 1–7, IEEE, 2007.
[52] M. Mahmud, M. Hossain, H. R. Pota, et al., “Analysis of voltage rise effect
on distribution network with distributed generation,” IFAC Proceedings
Volumes, vol. 44, no. 1, pp. 14796–14801, 2011.
[53] C. Masters, “Voltage rise: the big issue when connecting embedded
generation to long 11 kV overhead lines,” Power Engineering Journal,
vol. 16, no. 1, pp. 5–12, 2002.
[54] M. Mahmud, M. Hossain, H. Pota, and A. Nasiruzzaman, “Voltage control
of distribution networks with distributed generation using reactive power
compensation,” in IECON 2011-37th Annual Conference of the IEEE
Industrial Electronics Society, pp. 985–990, IEEE, 2011.
[55] M. Rossi, G. Viganò, and D. Moneta, “Hosting capacity of distribution
networks: Evaluation of the network congestion risk due to distributed
generation,” in 2015 International Conference on Clean Electrical Power
(ICCEP), pp. 716–722, IEEE, 2015.
[56] N. Etherden and M. H. Bollen, “Increasing the hosting capacity of
distribution networks by curtailment of renewable energy resources,” in
2011 IEEE Trondheim PowerTech, pp. 1–7, IEEE, 2011.
[57] M. Resch, J. Buehler, M. Klausen, and A. Sumper, “Impact of operation
strategies of large scale battery systems on distribution grid planning in
germany,” Renewable and Sustainable Energy Reviews, vol. 74, pp. 1042–
1063, 2017.
[58] T. T. Hashim, A. Mohamed, and H. Shareef, “A review on voltage control
methods for active distribution networks,” Przeglad Elektrotechniczny
(Electrical Review), vol. 88, no. 6, 2012.
[59] T. Xu and P. Taylor, “Voltage control techniques for electrical distribution
networks including distributed generation,” IFAC Proceedings Volumes,
vol. 41, no. 2, pp. 11967–11971, 2008.
[60] S. M. Mirbagheri, D. Falabretti, and M. Merlo, “Voltage control in active
distribution grids: A review and a new set-up procedure for local control
laws,” in 2018 International Symposium on Power Electronics, Electrical
Drives, Automation and Motion (SPEEDAM), pp. 1203–1208, IEEE, 2018.
[61] K. Dehghanpour and S. Afsharnia, “Electrical demand side contribution
to frequency control in power systems: a review on technical aspects,”
Renewable and Sustainable Energy Reviews, vol. 41, pp. 1267–1276, 2015.
BIBLIOGRAPHY 173
[62] Z. A. Obaid, L. M. Cipcigan, L. Abrahim, and M. T. Muhssin, “Frequency

control of future power systems: reviewing and evaluating challenges
and new control methods,” Journal of Modern Power Systems and Clean
Energy, vol. 7, no. 1, pp. 9–25, 2019.
[63] S. K. Pandey, S. R. Mohanty, and N. Kishor, “A literature survey on
load–frequency control for conventional and distribution generation power
systems,” Renewable and Sustainable Energy Reviews, vol. 25, pp. 318–334,
2013.
[64] R. D’hulst, J. M. Fernandez, E. Rikos, D. Kolodziej, K. Heussen,

D. Geibelk, A. Temiz, and C. Caerts, “Voltage and frequency control
for future power systems: the ELECTRA IRP proposal,” in 2015
International Symposium on Smart Electric Distribution Systems and
Technologies (EDST), pp. 245–250, IEEE, 2015.
[65] A. Bidram and A. Davoudi, “Hierarchical structure of microgrids control

system,” IEEE Transactions on Smart Grid, vol. 3, no. 4, pp. 1963–1976,
2012.
[66] A. M. Bouzid, J. M. Guerrero, A. Cheriti, M. Bouhamida, P. Sicard,
and M. Benghanem, “A survey on control of electric power distributed
generation systems for microgrid applications,” Renewable and Sustainable
Energy Reviews, vol. 44, pp. 751–766, 2015.
[67] R. Zamora and A. K. Srivastava, “Controls for microgrids with storage:
Review, challenges, and research needs,” Renewable and Sustainable
Energy Reviews, vol. 14, no. 7, pp. 2009–2018, 2010.
[68] A. G. Tsikalakis and N. D. Hatziargyriou, “Centralized control for

optimizing microgrids operation,” in 2011 IEEE Power and Energy Society
General Meeting, pp. 1–8, IEEE, 2011.
[69] M. Kabir, Y. Mishra, G. Ledwich, Z. Xu, and R. Bansal, “Improving
voltage profile of residential distribution systems using rooftop PVs and
battery energy storage systems,” Applied Energy, vol. 134, pp. 290–300,
2014.
[70] A. Zakariazadeh, O. Homaee, S. Jadid, and P. Siano, “A new approach
for real time voltage control using demand response in an automated
distribution system,” Applied Energy, vol. 117, pp. 157–166, 2014.
[71] A. Cagnano and E. De Tuglie, “Centralized voltage control for distribution

networks with embedded PV systems,” Renewable Energy, vol. 76, pp. 173–
185, 2015.
174 BIBLIOGRAPHY
[72] P. T. de Godoy, P. Poloni, A. B. de Almeida, and D. Marujo, “Centralized

secondary control assessment of microgrids with battery and diesel
generator,” in 2019 IEEE PES Innovative Smart Grid Technologies
Conference-Latin America (ISGT Latin America), pp. 1–6, IEEE, 2019.
[73] N. L. Diaz, A. C. Luna, J. C. Vasquez, and J. M. Guerrero, “Centralized
control architecture for coordination of distributed renewable generation
and energy storage in islanded ac microgrids,” IEEE Transactions on
Power Electronics, vol. 32, no. 7, pp. 5202–5213, 2016.
[74] K. Tan, X. Peng, P. L. So, Y. C. Chu, and M. Z. Chen, “Centralized control
for parallel operation of distributed generation inverters in microgrids,”
[75] M. M. A. Abdelaziz, M. F. Shaaban, H. E. Farag, and E. F. El-Saadany, “A
multistage centralized control scheme for islanded microgrids with PEVs,”
IEEE Transactions on Sustainable Energy, vol. 5, no. 3, pp. 927–937,
2014.
[76] K. Kok, “Multi-agent coordination in the electricity grid, from concept
towards market introduction,” in Proceedings of the 9th International
Conference on Autonomous Agents and Multiagent Systems: Industry
track, pp. 1681–1688, International Foundation for Autonomous Agents
and Multiagent Systems, 2010.
[77] S. Vandael, B. Claessens, M. Hommelberg, T. Holvoet, and G. Deconinck,
“A scalable three-step approach for demand side management of plug-
in hybrid vehicles,” IEEE Transactions on Smart Grid, vol. 4, no. 2,
pp. 720–728, 2012.
[78] L. Che, M. Shahidehpour, A. Alabdulwahab, and Y. Al-Turki,
“Hierarchical coordination of a community microgrid with AC and DC
microgrids,” IEEE Transactions on Smart Grid, vol. 6, no. 6, pp. 3042–
3051, 2015.
[79] D. P. Palomar and M. Chiang, “A tutorial on decomposition methods
for network utility maximization,” IEEE Journal on Selected Areas in
Communications, vol. 24, no. 8, pp. 1439–1451, 2006.
[80] S. Boyd, N. Parikh, E. Chu, B. Peleato, and J. Eckstein, “Distributed
optimization and statistical learning via the alternating direction method
of multipliers,” Foundations and Trends in Machine Learning, vol. 3, no. 1,
pp. 1–122, 2011.
[81] R. Schollmeier, “A definition of peer-to-peer networking for the
classification of peer-to-peer architectures and applications,” in Proceedings
BIBLIOGRAPHY 175
First International Conference on Peer-to-Peer Computing, pp. 101–102,

IEEE, 2001.
[82] S. D. McArthur, E. M. Davidson, V. M. Catterson, A. L. Dimeas, N. D.

Hatziargyriou, F. Ponci, and T. Funabashi, “Multi-agent systems for power
engineering applications—Part I: Concepts, approaches, and technical
challenges,” IEEE Transactions on Power systems, vol. 22, no. 4, pp. 1743–
1752, 2007.
[83] J. Engels, H. Almasalma, and G. Deconinck, “A distributed gossip-

based voltage control algorithm for peer-to-peer microgrids,” in
2016 IEEE International Conference on Smart Grid Communications
(SmartGridComm), pp. 370–375, IEEE, 2016.
[84] R. Olfati-Saber, J. A. Fax, and R. M. Murray, “Consensus and cooperation
in networked multi-agent systems,” Proceedings of the IEEE, vol. 95, no. 1,
pp. 215–233, 2007.
[85] T. Borsche, A. Ulbig, M. Koller, and G. Andersson, “Power and energy
capacity requirements of storages providing frequency control reserves,”
in 2013 IEEE Power & Energy Society General Meeting, pp. 1–5, IEEE,
2013.
[86] E. Demirok, P. C. Gonzalez, K. H. Frederiksen, D. Sera, P. Rodriguez,

and R. Teodorescu, “Local reactive power control methods for overvoltage
prevention of distributed solar inverters in low-voltage grids,” IEEE
Journal of Photovoltaics, vol. 1, no. 2, pp. 174–182, 2011.
[87] K. Eger, C. Gerdes, and S. Öztunali, “Towards p2p technologies for

the control of electrical power systems,” in 2008 Eighth International
Conference on Peer-to-Peer Computing, pp. 180–181, IEEE, 2008.
[88] P. T. Eugster, R. Guerraoui, A.-M. Kermarrec, and L. Massoulié,
“Epidemic information dissemination in distributed systems,” Computer,
vol. 37, no. 5, pp. 60–67, 2004.
[89] A. Demers, D. Greene, C. Hauser, W. Irish, J. Larson, S. Shenker,

H. Sturgis, D. Swinehart, and D. Terry, “Epidemic algorithms for
replicated database maintenance,” in Proceedings of the sixth annual
ACM Symposium on Principles of distributed computing, pp. 1–12, 1987.
[90] D. Kempe, A. Dobra, and J. Gehrke, “Gossip-based computation of

aggregate information,” in 44th Annual IEEE Symposium on Foundations
of Computer Science, 2003. Proceedings., pp. 482–491, IEEE, 2003.
176 BIBLIOGRAPHY
[91] S. Bolognani and S. Zampieri, “A gossip-like distributed optimization

algorithm for reactive power flow control,” IFAC Proceedings Volumes,
vol. 44, no. 1, pp. 5700–5705, 2011.
[92] K. De Brabandere, Voltage and frequency droop control in low voltage

grids by distributed generators with inverter frond-end. PhD thesis, KU
Leuven, 2006.
[93] W. Tushar, T. K. Saha, C. Yuen, D. Smith, and H. V. Poor, “Peer-to-
peer trading in electricity networks: an overview,” IEEE Transactions on
Smart Grid, 2020.
[94] C. Zhang, J. Wu, Y. Zhou, M. Cheng, and C. Long, “Peer-to-Peer energy
trading in a Microgrid,” Applied Energy, vol. 220, pp. 1–12, 2018.
[95] W. Tushar, T. K. Saha, C. Yuen, T. Morstyn, M. D. McCulloch, H. V.
Poor, and K. L. Wood, “A motivational game-theoretic approach for
peer-to-peer energy trading in the smart grid,” Applied Energy, vol. 243,
pp. 10–20, 2019.
[96] D. L. Rodrigues, X. Ye, X. Xia, and B. Zhu, “Battery energy storage
sizing optimisation for different ownership structures in a peer-to-peer
energy sharing community,” Applied Energy, vol. 262, p. 114498, 2020.
[97] S. Nguyen, W. Peng, P. Sokolowski, D. Alahakoon, and X. Yu, “Optimizing
rooftop photovoltaic distributed generation with battery storage for peer-
to-peer energy trading,” Applied Energy, vol. 228, pp. 2567–2580, 2018.
[98] J. Guerrero, A. C. Chapman, and G. Verbič, “Decentralized P2P energy
trading under network constraints in a low-voltage network,” IEEE
[99] M. Yazdanian and A. Mehrizi-Sani, “Distributed control techniques in
2909, 2014.
[100] A. N. Venkat, I. A. Hiskens, J. B. Rawlings, and S. J. Wright, “Distributed

MPC strategies with application to power system automatic generation
control,” IEEE Transactions on Control Systems Technology, vol. 16,
no. 6, pp. 1192–1206, 2008.
[101] S. Bolognani, R. Carli, G. Cavraro, and S. Zampieri, “Distributed reactive
power feedback control for voltage regulation and loss minimization,”
IEEE Transactions on Automatic Control, vol. 60, no. 4, pp. 966–981,
2014.
BIBLIOGRAPHY 177
[102] J. Lai, X. Lu, R.-L. Tang, X. Li, and Z. Dong, “Delay-tolerant

distributed voltage control for multiple smart loads in AC microgrids,”
ISA Transactions, vol. 86, pp. 181–191, 2019.
[103] X. Wang, C. Wang, T. Xu, H. Meng, P. Li, and L. Yu, “Distributed

voltage control for active distribution networks based on distribution
phasor measurement units,” Applied Energy, vol. 229, pp. 804–813, 2018.
[104] H. E. Farag and E. F. El-Saadany, “A novel cooperative protocol
for distributed voltage control in active distribution systems,” IEEE
Transactions on Power Systems, vol. 28, no. 2, pp. 1645–1656, 2012.
[105] A. Bidram, A. Davoudi, F. L. Lewis, and S. S. Ge, “Distributed adaptive
voltage control of inverter-based microgrids,” IEEE Transactions on
Energy Conversion, vol. 29, no. 4, pp. 862–872, 2014.
[106] S. Magnússon, G. Qu, and N. Li, “Distributed optimal voltage control

with asynchronous and delayed communication,” IEEE Transactions on
Smart Grid, 2020 (Early Access).
[107] T. Xu and W. Wu, “Accelerated ADMM-based Fully Distributed Inverter-
based Volt/Var Control Strategy for Active Distribution Networks,” IEEE
Transactions on Industrial Informatics, 2020.
[108] Z. Tang, D. J. Hill, and T. Liu, “Distributed control of active distribution
networks to support voltage control in subtransmission networks,”
International Journal of Electrical Power & Energy Systems, vol. 117,
p. 105715, 2020.
[109] D. Das, D. Kothari, and A. Kalam, “Simple and efficient method for load
flow solution of radial distribution networks,” International Journal of
Electrical Power & Energy Systems, vol. 17, no. 5, pp. 335–346, 1995.
[110] F. Olivier, P. Aristidou, D. Ernst, and T. Van Cutsem, “Active
management of low-voltage networks for mitigating overvoltages due
to photovoltaic units,” IEEE Transactions on Smart Grid, vol. 7, no. 2,
pp. 926–936, 2016.
[111] S. Weckx, R. D’Hulst, and J. Driesen, “Voltage sensitivity analysis of a
laboratory distribution grid with incomplete data,” IEEE Transactions
on Smart Grid, vol. 6, no. 3, pp. 1271–1280, 2015.
[112] M. Brenna, E. De Berardinis, L. D. Carpini, F. Foiadelli, P. Paulon,

P. Petroni, G. Sapienza, G. Scrosati, and D. Zaninelli, “Automatic
distributed voltage control algorithm in smart grids applications,” IEEE
178 BIBLIOGRAPHY
[113] M. E. Baran and F. F. Wu, “Optimal capacitor placement on radial

distribution systems,” IEEE Transactions on Power Delivery, vol. 4, no. 1,
pp. 725–734, 1989.
[114] B. B. Zad, J. Lobry, and F. Vallée, “A centralized approach for voltage
control of MV distribution systems using DGs power control and a direct
sensitivity analysis method,” in Energy Conference (ENERGYCON), 2016
IEEE International, pp. 1–6, IEEE, 2016.
[115] G. Díaz, J. Gómez-Aleixandre, and J. Coto, “Direct backward/forward
sweep algorithm for solving load power flows in AC droop-regulated
2217, 2016.
[116] E. Peeters, C. Develder, J. Das, J. Driesen, and R. Belmans, “LINEAR:
towards a Breakthrough of Smart Grids in Flanders,” in Innovation for
Sustainable Production: i-SUP 2010, pp. 3–6, 2010.
[117] W. Labeeuw and G. Deconinck, “Residential electrical load model based
on mixture model clustering and Markov models,” IEEE Transactions on
Industrial Informatics, vol. 9, no. 3, pp. 1561–1569, 2013.
[118] R. D. Zimmerman and C. E. Murillo-Sánchez, “Matpower 6.0 User’s
Manual,” 2016.
[119] M. Grant and S. Boyd, “Graph implementations for nonsmooth convex
programs,” in Recent Advances in Learning and Control (V. Blondel,
S. Boyd, and H. Kimura, eds.), Lecture Notes in Control and Information
Sciences, pp. 95–110, Springer-Verlag Limited, 2008.
[120] H. Everett III, “Generalized Lagrange multiplier method for solving
problems of optimum allocation of resources,” Operations Research, vol. 11,
no. 3, pp. 399–417, 1963.
[121] K. C. Kiwiel, T. Larsson, and P. O. Lindberg, “Lagrangian relaxation
via ballstep subgradient methods,” Mathematics of Operations Research,
vol. 32, no. 3, pp. 669–686, 2007.
[122] R. K. Sundaram, Inequality Constraints and the Theorem of Kuhn and
Tucker, p. 145–171. Cambridge University Press, 1996.
[123] A. Neumaier, “Solving ill-conditioned and singular linear systems: A
tutorial on regularization,” SIAM Review, vol. 40, no. 3, pp. 636–666,
1998.
[124] K. Fan, “Minimax theorems,” Proceedings of the National Academy of
Sciences of the United States of America, vol. 39, no. 1, p. 42, 1953.
BIBLIOGRAPHY 179
[125] S. Boyd, L. Xiao, A. Mutapcic, and J. Mattingley, “Notes on decomposition

methods,” Notes for EE364B, Stanford University, pp. 1–36, 2007.
[126] J.-P. Aubin and I. Ekeland, “Estimates of the duality gap in nonconvex
optimization,” Mathematics of Operations Research, vol. 1, no. 3, pp. 225–
245, 1976.
[127] D. Li, “Zero duality gap for a class of nonconvex optimization problems,”
Journal of Optimization Theory and Applications, vol. 85, no. 2, pp. 309–
324, 1995.
[128] S. Boyd, S. P. Boyd, and L. Vandenberghe, Convex optimization, ch. 5.
Cambridge university press, 2004.
[129] F. Blasa, S. Cafiero, G. Fortino, and G. Di Fatta, “Symmetric push-
sum protocol for decentralised aggregation,” in Proceedings of the Third
International Conference on Advances in P2P Systems, pp. 27–32, Citeseer,
2011.
[130] A. Balakrishnan, T. L. Magnanti, and R. T. Wong, “A dual-ascent
procedure for large-scale uncapacitated network design,” Operations
Research, vol. 37, no. 5, pp. 716–740, 1989.
[131] N. Z. Shor, Minimization methods for non-differentiable functions, vol. 3.
Springer Science & Business Media, 2012.
[132] E. Dall’Anese, S. V. Dhople, and G. B. Giannakis, “Photovoltaic inverter
controllers seeking AC optimal power flow solutions,” IEEE Transactions
on Power Systems, vol. 31, no. 4, pp. 2809–2823, 2016.
[133] R. Glowinski and A. Marroco, “Sur l’approximation, par éléments finis
d’ordre un, et la résolution, par pénalisation-dualité d’une classe de
problèmes de dirichlet non linéaires,” Revue française d’automatique,
informatique, recherche opérationnelle. Analyse numérique, vol. 9, no. R2,
pp. 41–76, 1975.
[134] D. Gabay and B. Mercier, “A dual algorithm for the solution of nonlinear
variational problems via finite element approximation,” Computers &
Mathematics with Applications, vol. 2, no. 1, pp. 17–40, 1976.
[135] M. Ma, L. Fan, and Z. Miao, “Consensus ADMM and Proximal ADMM
for economic dispatch and AC OPF with SOCP relaxation,” in 2016 North
American Power Symposium (NAPS), pp. 1–6, IEEE, 2016.
[136] C. Chen, B. He, Y. Ye, and X. Yuan, “The direct extension of ADMM for
multi-block convex minimization problems is not necessarily convergent,”
Mathematical Programming, vol. 155, no. 1-2, pp. 57–79, 2016.
180 BIBLIOGRAPHY
[137] B. He, L. Hou, and X. Yuan, “On full jacobian decomposition of the
augmented Lagrangian method for separable convex programming,” SIAM
Journal on Optimization, vol. 25, no. 4, pp. 2274–2312, 2015.
[138] W. Deng, M.-J. Lai, Z. Peng, and W. Yin, “Parallel multi-block ADMM
with o(1/k) convergence,” Journal of Scientific Computing, vol. 71, no. 2,
pp. 712–736, 2017.
[139] B. He, H.-K. Xu, and X. Yuan, “On the proximal Jacobian decomposition
of ALM for multiple-block separable convex minimization problems and
its relationship to ADMM,” Journal of Scientific Computing, vol. 66,
no. 3, pp. 1204–1217, 2016.
[140] J. A. Suykens, J. Vandewalle, and B. De Moor, “Intelligence and
cooperative search by coupled local minimizers,” International Journal of
Bifurcation and Chaos, vol. 11, no. 08, pp. 2133–2144, 2001.
[141] J. Suykens and J. Vandewalle, “Coupled local minimizers: alternative
formulations and extensions,” in Proceedings of the 2002 International
Joint Conference on Neural Networks, vol. 3, pp. 2039–2043, IEEE, 2002.
[142] I. Kouveliotis-Lysikatos, D. Koukoula, I. Vlachos, A. Dimeas,
N. Hatziargyriou, and S. Makrynikas, “Decentralised distribution system
operation techniques: results from the Meltemi community smart grids
pilot site,” CIRED-Open Access Proceedings Journal, vol. 2017, no. 1,
pp. 1673–1677, 2017.
[143] S. Iacovella, P. Vingerhoets, G. Deconinck, N. Honeth, and L. Nordstrom,
“Multi-agent platform for grid and communication impact analysis of
rapidly deployed demand response algorithms,” in Energy Conference
(ENERGYCON), 2016 IEEE International, pp. 1–6, IEEE, 2016.
[144] B. Dupont, P. Vingerhoets, P. Tant, K. Vanthournout, W. Cardinaels,
T. De Rybel, E. Peeters, and R. Belmans, “LINEAR breakthrough project:
Large-scale implementation of smart grid technologies in distribution
grids,” in 3rd IEEE PES Innovative Smart Grid Technologies Europe
(ISGT Europe), pp. 1–8, IEEE, 2012.
[145] K. Vanthournout, K. De Brabandere, E. Haesen, J. Van den Keybus,
G. Deconinck, and R. Belmans, “Agora: Distributed tertiary control
of distributed resources,” in Proceedings of the 15th Power Systems
Computation Conference, pp. 1–7, 2005.
[146] K. J. Macken, K. Vanthournout, J. Van den Keybus, G. Deconinck, and
R. J. Belmans, “Distributed control of renewable generation units with
integrated active filter,” IEEE Transactions on Power Electronics, vol. 19,
no. 5, pp. 1353–1360, 2004.
BIBLIOGRAPHY 181
[147] T. L. Nguyen, Agent-based distributed control and optimization in

microgrids with Hardware-in-the-Loop implementation. PhD thesis,
Université Grenoble Alpes, 2019.
[148] K. De Brabandere, T. Loix, K. Engelen, B. Bolsens, J. Van den Keybus,
J. Driesen, and R. Belmans, “Design and operation of a phase-locked
loop with Kalman estimator-based filter for single-phase applications,” in
IEEE Industrial Electronics, IECON 2006-32nd Annual Conference on,
pp. 525–530, IEEE, 2006.
[149] A. Yazdani and R. Iravani, Voltage-sourced converters in power systems:
modeling, control, and applications. John Wiley & Sons, 2010.
[150] A. Pouttu, J. Haapola, P. Ahokangas, Y. Xu, M. Kopsakangas-Savolainen,
E. Porras, J. Matamoros, C. Kalalas, J. Alonso-Zarate, F. D. Gallego,
et al., “P2P model for distributed energy trading, grid control and ICT
for local smart grids,” in 2017 European Conference on Networks and
Communications (EuCNC), pp. 1–6, IEEE, 2017.
[151] J. Haapola, S. Ali, C. Kalalas, J. Markkula, N. Rajatheva, A. Pouttu,
J. M. M. Rapun, I. Lalaguna, F. Vazquez-Gallego, J. Alonso-Zarate, et al.,
“Peer-to-peer energy trading and grid control communications solutions’
feasibility assessment based on key performance indicators,” in 2018 IEEE
87th Vehicular Technology Conference (VTC Spring), pp. 1–5, IEEE, 2018.
[152] P. Gandotra, R. K. Jha, and S. Jain, “A survey on device-to-device (D2D)
communication: Architecture and security issues,” Journal of Network
and Computer Applications, vol. 78, pp. 9–29, 2017.
[153] P.-K. Huang, R. J. Stacey, Q. Li, and R. Yang, “Wireless device, method,
and computer readable media for spatial reuse for device-to-device links,”
Tech. Rep. US9794790B2, United States Patent and Trademark Office,
2017.
[154] K. Mikhaylov, J. Petäjäjärvi, J. Haapola, and A. Pouttu, “D2D
Communications in LoRaWAN Low Power Wide Area Network: From Idea
to Empirical Validation,” in International Conference on Communications
(ICC), May 2017.
[155] S.-Y. Lien, C.-C. Chien, G. S.-T. Liu, H.-L. Tsai, R. Li, and Y. J.
Wang, “Enhanced LTE device-to-device proximity services,” IEEE
Communications Magazine, vol. 54, no. 12, pp. 174–182, 2016.
[156] R. Molina-Masegosa and J. Gozalvez, “LTE-V for sidelink 5G V2X
vehicular communications: A new 5G technology for short-range vehicle-
to-everything communications,” IEEE Vehicular Technology Magazine,
vol. 12, no. 4, pp. 30–39, 2017.
182 BIBLIOGRAPHY
[157] J. Markkula and J. Haapola, “Ad hoc LTE method for resilient smart
grid communications,” Wireless Personal Communications, vol. 98, no. 4,
pp. 3355–3375, 2018.
[158] U. Raza, P. Kulkarni, and M. Sooriyabandara, “Low power wide area
networks: An overview,” IEEE Communications Surveys & Tutorials,
vol. 19, no. 2, pp. 855–873, 2017.
[159] K. Mikhaylov and J. Petäjäjärvi, “Design and implementation of the
plug&play enabled flexible modular wireless sensor and actuator network
platform,” Asian Journal of Control, vol. 19, no. 4, pp. 1392–1412, 2017.
[160] IEA, “Renewables 2019: Market analysis and forecast from 2019 to 2024,
IEA, Paris.” https://www.iea.org/reports/renewables-2019. Accessed:
24-12-2019.
[161] IEA (2020), “Tracking energy integration: Energy storage, IEA, Paris.”
https://www.iea.org/reports/tracking-energy-integration-2020/energy-
storage. Accessed: 28-07-2020.
[162] H. Almasalma, J. Engels, and G. Deconinck, “Dual-decomposition-based
peer-to-peer voltage control for distribution networks,” CIRED - Open
[163] K. S. Ayyagari, N. Gatsis, and A. F. Taha, “Chance constrained
optimization of distributed energy resources via affine policies,” in
2017 IEEE Global Conference on Signal and Information Processing
(GlobalSIP), pp. 1050–1054, IEEE, 2017.
[164] R. A. Jabr, “Robust volt/var control with photovoltaics,” IEEE
Transactions on Power Systems, vol. 34, no. 3, pp. 2401–2408, 2019.
[165] R. A. Jabr, “Linear decision rules for control of reactive power by
distributed photovoltaic generators,” IEEE Transactions on Power
Systems, vol. 33, no. 2, pp. 2165–2174, 2017.
[166] W. Lin and E. Bitar, “Decentralized stochastic control of distributed
energy resources,” IEEE Transactions on Power Systems, vol. 33, no. 1,
pp. 888–900, 2017.
[167] E. E. Elattar, J. Goulermas, and Q. H. Wu, “Electric load forecasting
based on locally weighted support vector regression,” IEEE Transactions
on Systems, Man, and Cybernetics, Part C (Applications and Reviews),
vol. 40, no. 4, pp. 438–447, 2010.
[168] N. Zhang, C. Kang, E. Du, and Y. Wang, Analytics and Optimization for
Renewable Energy Integration, ch. 5, pp. 96–98. CRC Press, 2019.
BIBLIOGRAPHY 183
[169] Z. Li, R. Ding, and C. A. Floudas, “A comparative theoretical and

computational study on robust counterpart optimization: I. robust linear
optimization and robust mixed integer linear optimization,” Industrial &
Engineering Chemistry Research, vol. 50, no. 18, pp. 10567–10603, 2011.
[170] A. Ben-Tal, A. Goryashko, E. Guslitzer, and A. Nemirovski,
“Adjustable robust solutions of uncertain linear programs,” Mathematical
Programming, vol. 99, no. 2, pp. 351–376, 2004.
[171] X. Chen, M. Sim, P. Sun, and J. Zhang, “A linear decision-based

approximation approach to stochastic programming,” Operations Research,
vol. 56, no. 2, pp. 344–357, 2008.
[172] A. Shapiro and A. Nemirovski, “On complexity of stochastic programming
problems,” in Continuous Optimization, pp. 111–146, Springer, 2005.
[173] M. E. Baran and F. F. Wu, “Network reconfiguration in distribution

systems for loss reduction and load balancing,” IEEE Power Engineering
Review, vol. 9, no. 4, pp. 101–102, 1989.
[174] M. Farivar, L. Chen, and S. Low, “Equilibrium and dynamics of local
voltage control in distribution systems,” in 52nd IEEE Conference on
Decision and Control, pp. 4329–4334, IEEE, 2013.
[175] N. Efkarpidis, T. De Rybel, and J. Driesen, “Optimization control
scheme utilizing small-scale distributed generators and OLTC distribution
transformers,” Sustainable Energy, Grids and Networks, vol. 8, pp. 74–84,
2016.
[176] J. Engels, B. Claessens, and G. Deconinck, “Combined stochastic

optimization of frequency control and self-consumption with a battery,”
[177] K. Abdulla, J. De Hoog, V. Muenzel, F. Suits, K. Steer, A. Wirth, and
S. Halgamuge, “Optimal operation of energy storage systems considering
forecasts and battery degradation,” IEEE Transactions on Smart Grid,
vol. 9, no. 3, pp. 2086–2096, 2016.
[178] D. Kuhn, W. Wiesemann, and A. Georghiou, “Primal and dual linear
decision rules in stochastic and robust optimization,” Mathematical
Programming, vol. 130, no. 1, pp. 177–209, 2011.
[179] A. Kessy, A. Lewin, and K. Strimmer, “Optimal whitening and

decorrelation,” The American Statistician, vol. 72, no. 4, pp. 309–314,
2018.
184 BIBLIOGRAPHY
[180] B. P. Swaminathan, Operational planning of active distribution networks-

convex relaxation under uncertainty. PhD thesis, Université Grenoble
Alpes, 2017.
[181] Z. Li, Q. Tang, and C. A. Floudas, “A comparative theoretical and

computational study on robust counterpart optimization: II. probabilistic
guarantees on constraint satisfaction,” Industrial & Engineering Chemistry
Research, vol. 51, no. 19, pp. 6769–6788, 2012.
[182] J. Tant, F. Geth, D. Six, P. Tant, and J. Driesen, “Multiobjective battery

storage to improve PV integration in residential distribution grids,” IEEE
Transactions on Sustainable Energy, vol. 4, no. 1, pp. 182–191, 2012.
[183] I. Richardson, M. Thomson, D. Infield, and C. Clifford, “Domestic
electricity use: A high-resolution energy demand model,” Energy and
buildings, vol. 42, no. 10, pp. 1878–1887, 2010.
[184] J. Bright, C. Smith, P. Taylor, and R. Crook, “Stochastic generation
of synthetic minutely irradiance time series derived from mean hourly
weather observation data,” Solar Energy, vol. 115, pp. 229–242, 2015.
[185] J. Löfberg, “Yalmip: A toolbox for modeling and optimization in matlab,”
in Proceedings of the CACSD Conference, vol. 3, Taipei, Taiwan, 2004.
[186] Mosek, APS, “The MOSEK optimization software, 2018 (Version 9.0)
[Software].” Available: https://www.mosek.com/.
[187] R. D. Zimmerman, C. E. Murillo-Sanchez (2019), “MATPOWER (Version
7.0) [Software].” Available: https://matpower.org/.
[188] A. M. Howlader, S. Sadoyama, L. R. Roose, and Y. Chen, “Active power

control to mitigate voltage and frequency deviations for the smart grid
using smart PV inverters,” Applied Energy, vol. 258, p. 114000, 2020.
[189] A. Oudalov, D. Chartouni, C. Ohler, and G. Linhofer, “Value analysis
of battery energy storage applications in power systems,” in 2006 IEEE
PES Power Systems Conference and Exposition, pp. 2206–2211, IEEE,
2006.
[190] Elia, the Belgian TSO, “FCR service design note.” Available
at: https://www.elia.be/en/electricity-market-and-system/system-
services/keeping-the-balance/fcr. Accessed: 7 February 2020.
[191] T. Kern, S. Roon, “New auction design for frequency containment reserve
since 1 July 2019.” Available at: https://www.ffegmbh.de/en/areas-
of-expertise/scientific-analysis-of-system-and-energy-markets/106-
BIBLIOGRAPHY 185
electricity-market/892-new-auction-design-for-frequency-containment-
reserve-since-1-july-2019-first-interim-results. Accessed: 10 March
2020.
[192] T. Thien, D. Schweer, D. vom Stein, A. Moser, and D. U. Sauer, “Real-

world operating strategy and sensitivity analysis of frequency containment
reserve provision with battery energy storage systems in the German
market,” Journal of Energy Storage, vol. 13, pp. 143–163, 2017.
[193] Elia, the Belgian TSO, “Data download.” Available at:

https://www.elia.be/en/grid-data/data-download-page. Accessed:
10 January 2019.
[194] J. Fleer and P. Stenzel, “Impact analysis of different operation strategies
for battery energy storage systems providing primary control reserve,”
Journal of Energy Storage, vol. 8, pp. 320–338, 2016.
[195] E. Namor, F. Sossan, R. Cherkaoui, and M. Paolone, “Control of battery
storage systems for the simultaneous provision of multiple services,” IEEE
[196] D. Wu, C. Jin, P. Balducci, and M. Kintner-Meyer, “An energy storage
assessment: Using optimal control strategies to capture multiple services,”
in 2015 IEEE Power & Energy Society General Meeting, pp. 1–5, IEEE,
2015.
[197] O. Mégel, J. L. Mathieu, and G. Andersson, “Scheduling distributed
energy storage units to provide multiple services under forecast error,”
International Journal of Electrical Power & Energy Systems, vol. 72,
pp. 48–57, 2015.
[198] B. Cheng and W. B. Powell, “Co-optimizing battery storage for the
frequency regulation and energy arbitrage using multi-scale dynamic
programming,” IEEE Transactions on Smart Grid, vol. 9, no. 3, pp. 1997–
2005, 2016.
[199] Y. Shi, B. Xu, D. Wang, and B. Zhang, “Using battery storage for peak
shaving and frequency regulation: Joint optimization for superlinear gains,”
IEEE Transactions on Power Systems, vol. 33, no. 3, pp. 2882–2894, 2017.
[200] O. Mégel, J. L. Mathieu, and G. Andersson, “Stochastic dual dynamic
programming to schedule energy storage units providing multiple services,”
in 2015 IEEE Eindhoven PowerTech, pp. 1–6, IEEE, 2015.
[201] J. Chen, R. Zhu, M. Liu, G. De Carne, M. Liserre, F. Milano, and
T. O’Donnell, “Smart transformer for the provision of coordinated
186 BIBLIOGRAPHY
voltage and frequency support in the grid,” in IECON 2018-44th Annual

Conference of the IEEE Industrial Electronics Society, pp. 5574–5579,
IEEE, 2018.
[202] İ. Yanıkoğlu, B. L. Gorissen, and D. den Hertog, “A survey of adjustable

robust optimization,” European Journal of Operational Research, vol. 277,
no. 3, pp. 799–813, 2019.
[203] J. Löfberg. Yalmip, “Big-M and convex hulls.”
https://yalmip.github.io/tutorial/bigmandconvexhulls/. Accessed:
13-03-2020.
[204] D. Bertsimas and F. J. de Ruiter, “Duality in two-stage adaptive linear
optimization: Faster computation and stronger bounds,” INFORMS
Journal on Computing, vol. 28, no. 3, pp. 500–511, 2016.
[205] A. Soares, O. De Somer, D. Ectors, F. Aben, J. Goyvaerts, M. Broekmans,

F. Spiessens, D. van Goch, and K. Vanthournout, “Distributed
optimization algorithm for residential flexibility activation—results from a
field test,” IEEE Transactions on Power Systems, vol. 34, no. 5, pp. 4119–
4127, 2018.
Curriculum Vitae
Hamada Almasalma
Born on 27th December, 1987 in Beit Jala (Bethlehem Governorate), Palestine.
2001 - 2006 Secondary School,

Terra Santa Boys College,
Bethlehem, Palestine.
2006 - 2011 B.Sc. in Electrical Engineering,

Double Major in Power and Communication Engineering,
Birzeit University,
Birzeit, Palestine.
2011 - 2013 Systems Engineer,

Building Automation and Control Systems,
Bethlehem Advanced Technologies CO, Ltd.
Bethlehem, Palestine.
2013 - 2015 M.Sc. in Electrical Engineering,

Smart Grids and Buildings,
Ense3-Grenoble INP,
Grenoble, France.
2015 - 2020 Ph.D. in Engineering Science,

Department of Electrical Engineering,
KU Leuven,
Leuven, Belgium.
2020 - ... R&D Professional Electrical and Power-Electronics Energy

Systems,
The Flemish institute for technological research (VITO),
Genk, Belgium.
187
List of Publications
Journal publications
• H. Almasalma, and G. Deconinck, “Simultaneous Provision of Voltage
and Frequency Control by PV-Battery Systems,” IEEE ACCESS, vol. 8,
2020 (Early Access).
• H. Almasalma, and G. Deconinck, “Robust Policy-Based Distributed
Voltage Control Provided by PV-Battery Inverters,” IEEE ACCESS, vol.
8, pp. 124939-124948, 2020.
• H. Almasalma, S. Claeys, and G. Deconinck, “Peer-to-peer-based
• H. Almasalma, S. Claeys, K. Mikhaylov, J. Haapola, A. Pouttu, and G.

Deconinck, “Experimental Validation of Peer-to-Peer Distributed Voltage
Control System,” Energies, vol. 11, no. 5, pp. 1–22, 2018.
• H. Almasalma, J. Engels, and G. Deconinck, “Dual-decomposition-based
peer-to-peer voltage control for distribution networks,” CIRED-IET Open
International Conferences
• J. Haapola, S. Ali, C. Kalalas, J. Markkula, N. Rajatheva, A. Pouttu, J.
M. M. Rapún, I. Lalaguna, F. V. Gallego, J. A. Zarate, G. Deconinck,
H. Almasalma, J. Wu, C. Zhang, E. P. Muñoz, and F. D. Gallego.
Peer-to-peer energy trading and grid control communications solutions.
IEEE 87th Vehicular Technology Conference (VTC Spring), Porto, 2018.
189
190 LIST OF PUBLICATIONS
• H. Almasalma, J. Engels, and G. Deconinck. Dual-decomposition-based

peer-to-peer voltage control for distribution networks. CIRED, paper
0282, Glasgow, 12-15 June 2017.
• J. Engels, H. Almasalma, and G. Deconinck. A Distributed Gossip-

based Voltage Control Algorithm for Peer-to-Peer Microgrids. IEEE
International Conference on Smart Grid Communications (SmartGrid-
Comm), pp. 376-381, Sydney, 6-9 November 2016.
• A. Pouttu, J. Haapola, P. Ahokangas, Y. Xu, M. K. Savolainen, E. Porras,

J. Matamoros, C. Kalalas, J. A. Zarate, F. D. Gallego, J. M. Martín, G.
Deconinck, H. Almasalma, S. Clayes, J. Wu, M. Cheng, F. Li, Z. Zhang,
D. Rivas, and S. Casado. P2P model for distributed energy trading, grid
control and ICT for local smart grids. European Conference on Networks
and Communications (EuCNC), Oulu, 2017.
• H. Almasalma, J. Engels, and G. Deconinck. Peer-to-Peer Control of

Microgrids. 8th IEEE Benelux Young researchers symposium in Electrical
Power Engineering, Eindhoven, 12-13 May 2016.
FACULTY OF ENGINEERING SCIENCE
DEPARTMENT OF ELECTRICAL ENGINEERING (ESAT)
DIVISION ELECTRICAL ENERGY SYSTEMS AND APPLICATIONS (ELECTA)
Kasteelpark Arenberg 10, box 2445
B-3001 Leuven
Belgium

PhDThesis Hamada Almasalma

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PhDThesis Hamada Almasalma

Uploaded by

Copyright:

Available Formats

ARENBERG DOCTORAL SCHOOL

Faculty of Engineering Science

Frequency Support Voltage Support

Solar PV Panel Inverter Control

Supervisor: Dissertation presented in partial

Examination committee: Dissertation presented in partial

Control of distribution networks are facing significant challenges with the

technique. First major issue is its vulnerability to control instability due to

De regeling en sturing van distributienetwerken staan voor grote uitdagingen

Om de prestaties van het realtime op optimalisatie-gebaseerde DVC-systeem

3GPP 3rd Generation Partnership Project.

AARO Affinely Adjustable Robust Optimization.

BCBV Branch Current to Bus Voltage.

C-V2X Cellular Vehicle-to-Everything.

FCR Frequency Containment Reserve.

G-JP-ADMM Gossip-Based Jacobi-Proximal Alternating Di-

IFS Iterative Feedback Strategy.

J-ADMM Jacobi Alternating Direction Method of Multi-

MGCC Microgrid Central Controller.

p.u. per unit.

TSO Transmission System Operator.

UART Universal Asynchronous Receiver-Transmitter.

[·](k) The subscript k indicates a value at time step k.

[·]l The subscript l indicates a household.

[·]pv The subscript pv indicates a PV installation.

[·]bat The subscript bat indicates a battery.

[·]inv The subscript inv indicates an inverter.

[·]x Index x of network’s nodes.

[·]i Index i of control nodes.

[·]j Index j of passive nodes.

[·]T Transpose of a matrix.

acos[·] Inverse cosine.

cov[·] A covariance matrix.

diag[·] Square diagonal matrix.

E[·] Expected value.

ln[·] Natural logarithm.

vo Nominal voltage (V).

p Active power (kW).

q Reactive power (kvar).

fnom Nominal frequency (Hz).

sinvi Apparent power capacity of inverter i (kVA).

v Maximum voltage limit (V).

v Minimum voltage limit (V).

∆t Time step duration (minute).

Variables and Parameters of Chapters 3 and 4

cp,invi Penalty factor of inverter active power.

cq,invi Penalty factor of inverter reactive power.

cri Active power curtailment factor of inverter i (%).

LPALF Proximal Augmented Lagrangian Function.

vi Voltage magnitude of bus i.

α Step size of the dual decomposition method.

γinvi Acceleration factor of inverter i.

ρinvi Augmented penalty factor of inverter i.

τinvi Proximal penalty factor of inverter i.

λinvi Lagrangian multiplier associated with the constraint of

λinvi Lagrangian multiplier associated with the constraint of

Variables and Parameters of Chapters 5 and 6

cp Battery degradation cost (cent/kWh).

cq Reactive power compensation cost (cent/kvarh).

e bati Maximum energy capacity of battery i (kWh).

e bati Minimum energy capacity of battery i (kWh).

E 0VC, bati Initial energy content of the VC battery i (kWh).

E 0R-DN, bati Initial energy content of the R-DN battery i (kWh).

E 0R-UP, bati Initial energy content of the R-UP battery i (kWh).