A-2 P2P Based Probabilistic

Ref.
Ares(2016)7203385 - 31/12/2016
Peer to Peer Smart Energy Distribution Networks
D5.2 P2P-based probabilistic and predictive control functions
Document ID: P2P-SMARTEST-WP5-D5.2-V4.0
Deliverable Title: P2P-based Probabilistic and

Predictive Control Functions
Responsible beneficiary: Cardiff University (CU)
On 07/05/2014 the Project Coordinator submitted the proposal for the call for proposal
H2020-LCE-2014-3. The Specific Call is contemplated in the Work Programme 2014-
2015, 2020-LCE-2014-2015/H2020-LCE-2014-3. The Project with the contract number
646469 has been declared awarded in the framework of the Specific Call.
Start Date of the Project: 1 January 2015

Duration: 36 Months
Dissemination Level: Confidential
PROPRIETARY RIGHTS STATEMENT

This document contains information, which is proprietary to the P2P-SMARTEST Consortium.
Neither this document nor the information contained herein shall be used, duplicated or
communicated by any means to any third party, in whole or in parts, except with prior written
consent of the P2P-SMARTEST consortium.
This project has received funding from the European Union's Horizon 2020 research and
innovation programme under grant agreement No 646469
P2P-SMARTEST 646469 Name of Deliverable
Document Information
Document ID: P2P-SMARTEST- WP5-5.2-151227-V4

Version: V4.0
Version Date: 21.12.2016
Authors: CU: Dr. Chao Long, Dr. Meng Cheng, Mr. Chenghua Zhang, Prof.
Jianzhong Wu,
KUL: Mr. Hamada Almasalma, Mr. Jonas Engels, Mr. Sander Claeys, Prof.
Geert Deconinck,
BU: Mr. Heng Shi, Prof. Furong Li.
Security: Confidential
Approvals
Name Organization Date Visa
Coordinator Ari Pouttu UOULU 31.12.2016

Management Ari Pouttu UOULU 23.12.2016
Committee Jianzhong Wu CU
Furong Li BU
Document history
Revision Date Modification Authors

V1.0 5/12/2015 Table of Content was proposed by CU in the P2P- CU:
SMARTEST template. Dr. Chao Long,
KUL provided their contributions to Section 1 and Dr. Meng Cheng,
Section 3 (P2P based control paradigm). Mr. Chenghua Zhang,
BU added contributions to Section 1 and Section Prof. Jianzhong Wu,
4 (demand response in P2P network). KUL:
CU added their work on Section 1, Section 2, and Mr. Hamada Almasalma,
Section 5 (network control using SOP and Mr. Jonas Engels,
reconfiguration and considering multi-objectives).Mr. Sander Claeys,
Prof. Geert Deconinck,
BU: Mr. Heng Shi,
Prof. Furong Li.
P2P-SMARTEST 646469 QUALITY ASSURANCE MANUAL
10/12/2015 CU resembled all parts of work together and Dr. Chao Long,
V2.0 created links between different sections and Dr. Meng Cheng,
summarized the overall contributions. Mr. Chenghua Zhang,
Prof. Jianzhong Wu
18/12/2015 Comments received from KUL and BU. Mr. Hamada Almasalma
V3.0
Mr. Sander Claeys
Prof. Geert Deconinck
Mr. Heng Shi
Prof. Furong Li
V4.0 23/12/2015 CU revised the report, according to these Dr. Chao Long,
comments. Prof. Jianzhong Wu
3/84
P2P-SMARTEST 646469 D5.2 P2P-based probabilistic and predictive control functions
EXECUTIVE SUMMARY
This is the second report (D5.2) of the Work Package 5 (WP5) of the EU Horizon 2020 project
‘Peer to Peer Smart Energy Distribution Networks (P2P-SmartTest)’. This report introduces the
research output of Task 2 ‘P2P-based probabilistic and predictive control functions’.
P2P-SmartTest project aims to investigate and demonstrate a smarter electricity distribution system
based on the regional markets and innovative business models enabled by advanced information
and communication technology (ICT). The approach ‘Peer-to-Peer (P2P)’ will be employed to
facilitate the integration of distributed energy resources (DER) and other resources and the demand
side flexibility within the network while maintaining the energy balance, second-by-second power
balance and the quality and security of energy supply.
Smart Grid involves modernizing electric power networks, changing the way they operate and are
planned, facilitating changes in the behaviour of energy consumers, providing new services, and
supporting the transition to a sustainable low carbon economy. Although the driving forces and the
objectives of Smart Grid development are different in different countries, smart distribution
networks are always one of the key research topics.
Distribution networks are very large and topologically complicated systems which connect high
voltage transmission systems to end users. In today’s distribution networks, real-time monitoring
and control below the primary substations are very limited due to the lack of sensors,
communication systems and control equipment. Consequently the distribution network is a
bottleneck of the electricity supply chain. Conventional distribution network planning is not
responsive to changes in a network. It usually has a 5 - 20 years’ time horizon, and “worst case
scenario” and deterministic methods are widely used. Normally there is no continuous link between
network operation and planning.
The widespread use of DERs may lead to operational difficulties on a distribution network. Loads
are likely to become more volatile and unpredictable as customers gain greater knowledge of their
load profile, and new home energy management and EV charging mechanisms are introduced.
More intermittent distributed generation will be connected. Sophisticated measurements, control
elements and ICT infrastructure will be widely deployed. New business models and market
arrangements will be employed. Consequently the behaviour of distribution networks will become
increasingly uncertain and the boundary between operation and planning will become increasingly
unclear, as new Smart Grid and Smart Energy interventions are introduced with various extents
and time-lines.
There are well established distribution network operation and planning tools available. However
they are limited by: i) their passive features; ii) intolerance of information gaps and uncertainties;
iii) slow calculation speed for an extensive network, and iv) difficulties in generalizing the results
obtained from a particular network.
Alternative P2P based control philosophy of distribution networks was developed, and technical
requirements on the associated ICT infrastructure was analysed, based on our extensive previous
and current work on decentralized control of distribution networks. This part was presented in D5.1.
3/84
Novel probabilistic and predictive control functions to enable and facilitate the P2P based energy
trading and better network operation under extremely dynamic and uncertain conditions were
developed, and presented in this report (D5.2). These control functions mainly involved in
employing novel active network control, such as power electronic device soft open point,
coordinated network control among different active control devices, demand side integration, and
coordinated control considering multiple objectives.
The mainly findings include but not limit to the flowing aspects:
(1) Using a novel active network control device, i.e. soft open point was proved to be able to
significantly increase the network’s distributed generation (DG) hosting capacity. The
control scheme using the objective for voltage profile improvement increased the headroom
of the voltage limits by the largest margin. This control scheme dispatched increased
reactive power, and, hence, was at the expense of increased energy losses. The control
schemes using the objectives to achieve line utilization balance and energy losses
minimization showed the most improvement in circuit utilization and in limiting energy
losses, mainly relying on the real power exchange between feeders. According to network
characteristics, network operators are able to devise the control scheme using one or multi-
objective functions. The proposed network control using soft open point provides a
potential framework/solution for electricity network operators, allowing them to choose
appropriate control schemes more effectively. This selection requires the network operators
to attribute value to the increase in hosting capacity, mitigation in voltage issues, the
reduction in the maximum line utilization and the reduction in energy losses.
(2) The novel voltage coordinated control algorithms were proved to be able to operate
completely distributed, thereby keeping all control local and eliminating any single point of
failure. The algorithms use a gossip-based push-sum protocol to dynamically disseminate
data about the state of the grid without the need for one central point of information. The
algorithms calculate at various time steps and update of the required active and reactive
power to maintain voltages within accepted limits. Both algorithms treat the problem as an
optimization problem that controls reactive and active power flows. The first algorithm is
based on a gradient descent method, while the second algorithm is based on dual
decomposition. The calculation of voltage sensitivity to active and reactive power change
is based on a linearization of the actual distribution grid, which leads to constant
sensitivities that only need to be calculated once, thereby limiting the computational
requirements of the participating agents. Two realistic case studies were presented to assess
the performance of the methods. It is shown that both algorithms are able to follow quickly
variable load profiles while being able to maintain voltages within limits.
(3) The concept of Demand Side Integration (DSI) emerges when the conventional concept
demand side management is no longer appropriate for newly deregulated and competitive
market in future power system. These new types of demand side interventions include more
low carbon resources as well as demand response programs with engaged customers.
Through investigating demand side integration considering domestic households with
DC/AC system integrated together, and distributed PV generations, and performing optimal
operation, energy storage enabling energy management system (EMS) by their flexibilities,
and triggered by tariffs, this work explored the impact of demand side integration
interventions both of technical aspects and commercial aspects, differing energy storages
4/84
and pricing methods are compared. Regarding the technical intervention, electric vehicles
(EV) and battery storage are investigated accordingly. The result demonstrates positive
effects by using two types of energy storage to enable the EMS, however, with largely
different impact on demand. For differing pricing methods, the local price purely designed
to incentivise battery to absorb photovoltaic (PV), can bring fairly demand and uncertainty
reduction. However, it is not very capable of committing demand shift actions. The central
price designed according to wholesale price, is not only capable of reducing demand and
digest uncertainty, but also shifting peak demand to valley periods.
(4) Through formulating the network reconfiguration and the optimal soft open point operation
problem within a multi-objective framework using the Pareto optimality, the work
demonstrated the effectiveness of using distribution network reconfiguration and soft open
point to improve the distribution network operation, focusing on power loss reduction, load
balance and DG penetration level increase. The obtained Pareto frontiers present great
diversity, high quality and proper distribution of the non-dominated solutions among all
feasible solutions, which allows the Distribution Network Operators to choose based on
their priorities and necessities.
Test results showed that these alternative methodologies and tools for scalable distribution network
operation and planning enable P2P energy trading with varying degrees of input information gap
and uncertainties caused by decentralized P2P control philosophies.
5/84
TABLE OF CONTENTS
Executive Summary........................................................................................................................................3
Table of Contents ...........................................................................................................................................6
Abbreviations and Glossary ...........................................................................................................................8
1 Introduction ............................................................................................................................................9
1.1 Uncertainties and the requirement for distribution network control ............................................10
1.2 Potential control methods .............................................................................................................13
1.2.1 Control through network reconfiguration and SOP ..............................................................13
1.2.2 Coordinated voltage control .................................................................................................15
1.2.3 Demand side integration .......................................................................................................20
1.2.4 Coordinated control with multiple objectives.......................................................................22
1.3 Deliverable 5.2 .............................................................................................................................23
2 Network Control through SOP .............................................................................................................24
2.1 Problem formulation .....................................................................................................................24
2.1.1 Modelling of soft open point (SOP) .....................................................................................24
2.1.2 Jacobian matrix based sensitivity analysis ...........................................................................25
2.2 Optimization formulation .............................................................................................................27
2.2.1 Voltage profile improvement (VPI) .....................................................................................27
2.2.2 Line utilization balancing (LUB) .........................................................................................27
2.2.3 Energy loss minimization (ELM) .........................................................................................28
2.2.4 Constraints ............................................................................................................................28
2.3 Visualization of the SOP operating region ...................................................................................28
2.4 Case Study ....................................................................................................................................32
2.4.1 MV distribution network model ...........................................................................................32
2.4.2 DG penetration level.............................................................................................................33
2.4.3 Two-day performance with a 90% DG penetration – full observability of the network ......33
2.4.4 Overall performance – full observability of the network .....................................................37
2.4.5 Overall performance – limited & no observability of the network.......................................38
2.5 Discussion.....................................................................................................................................38
2.6 Conclusions ..................................................................................................................................39
6/84
3 Coordinated Voltage Control................................................................................................................40

3.2 Active and reactive power control of DERs .................................................................................42
3.3 Monitoring and measurements .....................................................................................................43
3.4 Proposed P2P voltage control algorithms .....................................................................................43
3.4.1 A basic P2P voltage control algorithm .................................................................................44
3.4.2 Case Study 1 .........................................................................................................................47
3.4.3 A P2P voltage control algorithm with minimum resources ..................................................50
3.4.4 Case Study 2 .........................................................................................................................53
3.5 Conclusion ....................................................................................................................................56
4 Demand Side Integration ......................................................................................................................58
4.2 Type I DSI - Energy management system ....................................................................................60
4.3 Type II DSI – pricing method: central and local energy market signal ........................................61
4.4 Results ..........................................................................................................................................62
4.5 Conclusions ..................................................................................................................................65
5 Coordinated Control with Multiple Objectives ....................................................................................67
5.1.1 Multi-objective functions .....................................................................................................67
5.1.2 Constraints ............................................................................................................................68
5.2 Solution methodology ..................................................................................................................68
5.2.1 DNR optimization using the ant colony optimization (ACO) method .................................68
5.2.2 SOP output optimization using the Taxi-Cab method ..........................................................68
5.2.3 Visualization of multi-objective optimization solutions using the Pareto optimality...........69
5.3 Test system and results .................................................................................................................69
5.3.1 Test system ...........................................................................................................................69
5.3.2 Multi-objective DNR results ................................................................................................70
5.3.3 Multi-objective SOP output optimization results .................................................................71
5.4 Conclusion ....................................................................................................................................73
6 Conclusions ..........................................................................................................................................74
7 References ............................................................................................................................................77
7/84
ABBREVIATIONS AND GLOSSARY
ACO Ant Colony Optimization

DEM Dynamic Energy Management
DER Distributed Energy Resource
DSI Demand Side Integration
DSO Distribution System Operator
DSM Demand Side Management
DNR Distribution Network Reconfiguration
ELM Energy Loss Minimization
EMS Energy Management System
ESCO Energy Service Company
EV Electric Vehicle
ICT Information Communication Technology
LBI Load Balance Index
LC Local Control
LCNF Low Carbon Networks Fund
LS Load Shifting
LV Low Voltage
LUB Line Utilisation Balance
MMC Modular Multi-level Converter
MNO Mobile Network Operator
MV Medium Voltage
NOP Normal Open Point
OLTC On-Load Tap Changer
PC Peak Clipping
PCC Point of Common Coupling
PID Proportional-integral-derivative
PMU Phasor Measurement Units
PV Photovoltaic
P2P Peer-to-Peer
SCADA Supervisory Control and Data Acquisition
STATCOM Static Synchronous Compensator
SVC Secondary Voltage Control
TCR Thyristor Controlled Reactor
TOU Time of Use
VPI Voltage Profile Improvement
VSC Voltage Source Converter
VF Valley Filling
WP Work Package
8/84
1 INTRODUCTION
To ensure Europe produces world-class science, removes barriers to innovation and makes it easier
for the public and private sectors to work together in delivering innovation, Horizon 2020 (The EU
Framework Programme for Research and Innovation) is implemented and backed by Europe’s
leaders and the Member of the European Parliament. By coupling research and innovation, Horizon
2020 is helping to achieve the goal with its emphasis on excellent science, industrial leadership and
tackling societal challenges. P2P-SmartTest project is one of the Horizon 2020 projects.
P2P-SmartTest project investigates and demonstrates a smarter electricity distribution system

based on the regional markets and innovative business models enabled by advanced information
and communication technology (ICT). It employs the Peer-to-Peer (P2P) approaches to ensure the
integration of demand side flexibility and the optimum operation of distribution energy resources
(DERs) and other resources within the network while maintaining the energy balance, second-by-
second power balance and the quality and security of the supply.
The objectives of this project are:

(1) To investigate and develop alternative business models for DSOs, MNOs, Aggregators, ESCOs,
Suppliers and Consumers for P2P energy trading to capture the whole supply chain value while
maintaining second-by-second power balance, maximizing Demand Response and DER utilization
and ensuring supply security. The magnitude of benefits from introducing P2P energy trading is
quantified and the required changes in technical, commercial and regulatory arrangements will be
identified. (This corresponds to WP2.)
(2) To evaluate existing ICT technologies and new ones for P2P energy trading. The focus is on
investigating the last-mile technologies, which support inter- and intra-Microgrids operation, also
the backbone telecom infrastructure is considered, which is critical for intra CELLs operation and
data exchange with transmission network operators. (This corresponds to WP3.)
(3) To develop P2P advanced optimization techniques to provide efficient P2P energy market
trading, while considering the new business models and ICT technologies. In order to fulfil a real
integration of the flexibility of demand and DER management using P2P, the whole market domain
will be explored including products/services to be traded and certification mechanisms to be
implemented. (This corresponds to WP4.)
(4) To develop alternative P2P based control paradigm of distribution networks, integrate
probabilistic and predictive control functions to enable and facilitate the P2P based energy trading
and better network operation under extremely dynamic and uncertain conditions, and model of
dynamic demand for operational functions of P2P smart distribution networks. (This corresponds
to WP5.)
This deliverable is part of WP5.
9/84
1.1 Uncertainties and the requirement for distribution network control
supporting the transition to a sustainable low carbon economy. Although the driving forces and the
objectives of Smart Grid development are different in different countries, smart distribution
networks are always one of the key research topics.
Distribution networks are very large and topologically complicated systems which connect high
voltage transmission systems to end users. In today’s distribution networks, real-time monitoring
and control below the primary substations are very limited due to the lack of sensors,
communication systems and control equipment. Consequently, the distribution network is a
bottleneck of the electricity supply chain. Conventional distribution network planning is not
responsive to changes in a network. It usually has a 5 - 20 years’ time horizon, and “worst case
scenario” and deterministic methods are widely used. Normally there is no continuous link between
network operation and planning.
arrangements will be employed. Consequently, the behaviour of distribution networks will become
and time-lines.
obtained from a particular network. There is an urgent need to develop alternative methodologies
and tools for scalable distribution network operation and planning, which can facilitate P2P energy
trading with varying degrees of input information gap and uncertainties caused by decentralized
P2P control philosophies.
Network controls are hierarchical in nature, becoming increasingly absent as the end-users are
approached. Voltage control is one of the most import control features in distribution networks.
The voltage variation problem in distribution networks can be solved by either network, generators,
load operational changes or network asset upgrades.
The technical solutions for solving the voltage problem can be classified according to the following
three categories [1]:
I. DSO (or microgrid operator): for solutions that are implemented within the grid operator
infrastructure and require no communication with the consumers (or prosumers).
10/84
II. Prosumer: for solutions which are implemented within the prosumers infrastructure and
require no communication between the prosumers, nor with the grid operator.
III. Interactive: for solutions that are implemented within both the grid operator and the
prosumers infrastructure and where the different components react based on signals
exchanged via a communication infrastructure.
The different technical solutions are summarized in the following Table 1.1.
Table 1.1 List of technical solutions

Category Technical Solution
Network reinforcement
Network reconfiguration
On load tap changer
DSO
DSO storage
Buck & Boost Transformer
Static VAR / Static Synchronous Compensation
Prosumer Active power control (drop control)
Reactive power control (drop control)
Interactive Coordinated voltage control
Network reinforcement is the more traditional action carried out in order to ensure compliance with
voltage and thermal requirements in case the connection of a new DERs may bring variations
outside the standards. In specific situations, this solution might include building a new feeder (and
eventually a new substations) instead of reinforcing an existing one (if it’s cheaper or more
effective). Its main disadvantage is the relatively high investment costs [2].
Voltage problems can also be solved by grid reconfiguration. Grid reconfiguration is defined as
altering the topological structure of distribution feeders by changing the open/closed states of
sectionalizers and tie switches so that a particular pre-defined objective function is minimized and
the constraints are met. A reconfiguration is limited to areas where at least ring or other meshed
structures exist which are usually operated with open switches. Therefore, this measure is mainly
relevant for MV grids and urban areas [3].
On load tap changer (OLTC) is one of the methods that can efficiently maintain the voltage within
the limits in the distribution grids. OLTC device is able to adjust the lower voltage value of an
energized transformer in order to return the system voltages inside the limits. Until recently, the
application of OLTC is conventionally limited to HV/HV or HV/MV transformers, while the small,
low-cost distribution transformers are normally equipped with off-circuit taps. But due to the
increasing needs of active voltage control in LV grids, OLTC for MV/LV transformers has become
a necessity for LV networks with high penetration of DERs [4], [5].
11/84
Buck-Boost Transformer is one of the methods that has been used in the past in long feeders to
compensate voltage drops exceeding standards. It can be imagined to use the same equipment for
mitigating negative impacts of DERs on voltage. A Buck-Booster Transformers are a transformers
of which one winding is intended to be connected in series with a feeder in order to alter its voltage
and the other winding is an energizing winding. They are MV-MV or LV-LV transformers that can
be used to stabilize the voltage along a feeder [6].
Depending on feeder characteristic (R/X ratio), voltage can be more or less controlled by reactive
power consumption. Possible solutions are: (1) Static VAR Compensator (SVC), Static
Synchronous Compensator (STATCOM) and (3) reactive power control by power inverters.
The SVC consists of a number of fixed or switched branches, of which at least one branch includes
thyristors, and the combination of branches can be varied a lot depending on requirements. An SVC
typically includes a combination of at least two of the following items: thyristor controlled reactor
(TCR), thyristor switched capacitor (TSC), harmonic filter (FC), mechanically switched capacitor
bank (MSC) or reactor bank (MSR). Most common topologies for SVCs are: TCR/FC or
TCR/TSC/FC [7].
STATCOM is an IGBT voltage-source converter based device which converts a DC input voltage
into an AC output voltage in order to compensate the reactive power of the system. Usually the
reactive output of a STATCOM is regulated to maintain the desired AC voltage at the bus to which
a STATCOM is connected [8].
STATCOM has better characteristics than an SVC. The major attributes of STATCOM are quick
response time, less space requirement, optimum voltage platform, higher operational flexibility and
excellent dynamic characteristics under various operating conditions [9].
Power inverters are capable to provide reactive power (Q) as well as active power (P). It is therefore
possible to require to an inverter to behave in a specific way in relation with reactive power. The
reactive power of the inverter can be a function of its active power production [Q=Q(P)] or a
function of local voltage measurements [Q=Q(U)] [10]. This solution is currently available from a
technical point of view. Inverters are considered as a cheap solution to provide reactive power as
the cost of inverters is reducing at three or four times the rate of costs for traditional var
compensation devices, like STATCOMs.
The effectiveness of reactive power on managing voltage ultimately depends on the impedance of
the feeder and is lower in case of high R/X ratio. This technical solution is therefore more effective
in MV networks than in LV ones. Therefore, in LV networks, regulating voltage via reactive power
can be adopted as a “first step” of a more complex control strategy and should be enhanced with
other techniques.
Voltage problems can be solved by curtailing the DER feed-in active power. Controlling the
voltage level via active power curtailment is efficient for feeders with a predominantly resistive
12/84
impedance, as it is the case in low voltage grids. For economic reasons, active power reduction
should be only used when all other less expensive solutions have been applied [11].
Although still very expensive, small decentralized storage systems can actively contribute to
reducing local voltage deviation in networks with high resistive impedance via injecting or storing
active power [12]. Storage systems are classified into mechanical, electrochemical, chemical,
electrical and thermal energy storage systems, more details can be found in [13].
For more reliable control system and optimal use of voltage regulating devices, active management
strategies, such as coordinated voltage control, that require communication between different
prosumers and the DSO (or microgrid operator) are frequently suggested for voltage regulation in
a distribution system with a high level of DERs [14].
1.2 Potential control methods
The P2P control paradigm will introduce more uncertainties. Probabilistic methods can be used to
deal with uncertainties and the predictive methods can be used to solve constrained finite horizon
optimal control problems using forecast future network information (based on historical network
operation data or scenario information of future networks). The potential control methods include:
control through network reconfiguration and power electronic devices, such as soft open point
(SOP), coordinated voltage control, demand side integration, coordinated network
reconfiguration/SOP control with multiple objectives, etc. Each of these active network control
methods is presented in the following subsections.
1.2.1 Control through network reconfiguration and SOP
Distribution networks are normally designed as meshed networks but are operated in a radial
manner with normally open points [15]. The network configuration can be changed during the
operation by changing the open/close status of the switchgear, manually or automatically [15].
Figure 1.1 gives an example of network reconfiguration, which is based on a three-feeder network
with three NOPs (dash lines). The configuration of the network is changed by closing NOPs 15 and
26 while opening closed switches 12 and 13.
13/84
(a) (b)
Figure 1.1 Three-feeder example network: (a) before network reconfiguration; (b) after network
reconfiguration [15]
The main objectives of network reconfiguration include [16], [17]:
1. Supply restoration：This optimally restores de-energized customers through alternative

sources in the case of planned and unplanned power outages.
2. Load balancing among substation transformers or different feeders and equalising the
voltages.
3. Active power loss minimization at a given time or energy loss minimization during a period.
The methods used to solve the above network reconfiguration problems include those based on
practical experience and optimization techniques. The methods based on optimization techniques
determine the optimal network configuration by solving a complicated combinatorial, non-
differentiable constrained optimization problem [18]. Possible solutions include mathematical
algorithms [15], [19], computational intelligent-based algorithms (for example, generic algorithm,
simulated-annealing, fuzzy logic) [20]-[22], and hybrid algorithms which combine two or more
above algorithms [23], [24].
Network reconfiguration and its significant benefits have been reported in the past decades.
However, practical applications of automatic network reconfiguration are presently very limited
due to excessive costs of remotely-controlled switchgear, associated ICT infrastructures and
maintenance of hardware/software (e.g., excessive wear and tear of switches) [25], [26]. The
protection coordination requirement for the reconfigurable network is also a hurdle [27].
It is possible to use Soft Open Points (SOPs) to achieve the same aforementioned benefits
(objectives) offered by network reconfiguration. In the meantime, changes of network topology
and protection coordination are no longer needed when using the SOP. The potential and benefits
of using SOP, instead of traditional network reconfiguration, was discussed in further detail.
14/84
A soft open point (SOP) was used to replace the mechanical switch at a previously normally open
point of an MV network. An SOP is a power electronic device, usually using back-to-back voltage
source converters (VSCs). Such device has also been called “SIPLINK” [28], “DC-link” [29], [30],
[31], and “SNOP” [32] in the literature.
Compared with network reconfiguration, the use of SOPs has the following advantages of: 1)
regulating power flows in a continuous manner; 2) flexible and accurate controllability of active
and reactive power. In particular, the control of reactive power at each terminal is independent; 3)
short-circuit currents are not increased when using SOPs, due to the almost instantaneous control
of current; and 4) SOPs can be used to connect any group of feeders, e.g., supplied from different
substations or at unequal rated voltages [30].
SOP devices have been made commercially available [28], but the control strategies and their
impact on power networks have not been thoroughly investigated. The benefits of using SOPs in
power networks were analyzed in [30], [33], [34]. These studies, however, were limited to only a
few snapshots, rather than considering SOP operation under different load and generation
conditions and over a period of time. A few initiative pilot projects have been trialled using SOPs
in MV distribution networks in the UK, such as [35], [36], but they are in their early stage of
development. As an Ofgem Low Carbon Networks Fund (LCNF) project, Flexible Urban Networks
Low Voltage (FUN-LV) (initiated by UK Power Networks Ltd) has explored the use of SOP in LV
networks [37]. Dual- or multi-terminal SOPs have been trialled across 36 networks. Different
control modes, i.e. transformer equalization, voltage support, power factor support, and unbalance
support, were applied to different networks adopting a hysteresis method. They did not consider
the sensitivity of these control features, i.e. transformer loading, voltage, power factor, and network
unbalance, to the SOP’s active and reactive power injection.
Moreover, there were no detailed models investigating the different effects of active and reactive
power from an SOP on the networks. A VSC interconnected with an AC grid, is a nonlinear coupled
double-input double-output control object [38]. In [39] a PID (proportional-integral-derivative)
controller was adopted to design the transfer function where the active and reactive power were de-
coupled and active power was to manage voltage angle and reactive power was to regulate the
magnitude. Therefore, to obtain a model and subsequent design of controllers, a mathematical
analysis which simultaneously considers the active and reactive power injection from VSCs is
required. On the other hand, the model needs to be general to be applicable to networks with
different topologies. In [40] and [41], an optimization framework, so-called Intervals of Secure
Power Injection method, was developed to maximise admissible sets of power injections for secure
network operation under marginal changes in network topology.
1.2.2 Coordinated voltage control
Several active voltage control methods of different complexity and data transfer needs have been
proposed in publications. The simplest active voltage control methods are based only on local
measurements and do not require additional data transfer between network nodes. One can argue
that it is simpler and cheaper to regulate the voltage using only local control. However, this
15/84
individual local control results in operating scenarios that may be good for the local unit but not for
the network. For example, changing bus voltage locally has consequences on other units connected
to the same feeder and must be carefully studied to understand how to properly regulate the voltage
across the whole feeder. Besides, optimal operation of the grid can never be reached based on local
control. Therefore, coordinated voltage control is needed for more reliable and optimized power
system.
Coordinated voltage control can be organised according to many different control architectures.
They can range from fully centralized control where all decisions are made by a single central
controller, to completely distributed control where all decisions are made by the local DERs. The
required communication architecture changes accordingly. This report identifies four different
approaches, shown in Figure 1.2: (a) centralized control, (b) hierarchical control, (c) distributed
control and (d) fully decentralized P2P control [42].
It is impossible to impose these architectures 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 smaller microgrids, containing only a limited amount of DERs, can be a solution
[43].
In a fully centralized design shown in Figure 1.2(a), all available measurements of the considered
microgrid are gathered in a central controller that determines the control actions for all units. When
looking at a single microgrid, the centralized controller is often referred to as a Microgrid Central
Controller (MGCC). In [44], a control strategy based on the concept of Wide-area Coordinated
Control of Energy Storage System (WCCESS) has been proposed. The strategy can be used to
implement a centralized coordinated voltage control.
In [45], two centralized coordinated voltage control algorithms have been developed to mitigate
the voltage rise problem. The first algorithm is a relatively simple rule based algorithm and the
second one utilizes optimization. The algorithms use substation voltage and reactive and active
power of DERs as control variables. The proposed method consists of four steps: at the beginning
the algorithm tries to regulate the voltages using OLTC. If substation voltage control is unable to
restore network voltages between acceptable limits, reactive power control is activated. If
substation voltage control and reactive power control are not able to restore the network voltages
between acceptable limits, real power control is activated. Finally, if the maximum voltage has
decreased enough, the algorithm tries to restore real and reactive power of DERs, and substation
voltage to their original values.
In [46], an algorithm based on the sensitivity analysis has been designed for voltage regulation of
MV grids. It optimally manages active and reactive powers of DERs in order to return the system
voltages inside the limits. The proposed algorithm distinguishes between cheap and expensive
control actions using defined constants in its optimization part.
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
16/84
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 [47]. To
overcome these issues, more distributed control architectures are developed.
A first step towards a more decentralized control architecture is the introduction of a hierarchical
system, as shown in Figure 1.2(b). In this case, there exists some kind of aggregation of the local
DERs towards the central controller. 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. An example of such a system can be found in [48], a hierarchical
voltage control strategy has been designed to perform the voltage control by activating the
flexibilities from various DERs.
17/84
MGCC
Industrial Prosumer Commercial Prosumer
(a) Centralized Control Residential Prosumer
MGCC
Aggregator
Aggregator
(b) Hierarchical Control

Central Central
agent agent
Local agent
Microgrid
(c) Distributed Control
P2P
Local agent
P2P
P2P
P2P
P2P
P2P
Microgrid
(d) P2P Control
Figure 1.2 Architectures of a coordinated control
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. However, as we discussed before, 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.
18/84
The idea behind distributed control is to divide the centralized problem into a certain number of
local controllers or agents. Therefore, each agent does not have a global vision of the problem [49],
but by means of correct coordination 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 dual
decomposition methods or the alternating direction method of multipliers (ADMM) [50]. 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 consumption 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. In [51], ADMM based reactive power control algorithm has been proposed, the algorithm
aims to minimize power losses subject to finite inverter capacity and upper and lower voltage limits
at all nodes in the circuit.
Figure 1.2(c) represents such a 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 very scalable method. As the local DERs perform an optimization by
themselves, they do not need to hand over private information to a third party that controls their
resources. However, a central agent still exists, inherently resulting in a single point of failure.
To eliminate the problems that a more centralized control method possesses, having a single point
of failure, the idea of peer-to-peer microgrids has been developed. This type of architecture,
inspired by P2P computer networking [52], is characterized by the complete absence of a central
controller. All local DERs or agents, are equally important and can communicate to other agents
[53], in a peer-to-peer fashion, as shown in Figure 1.2(d).
The absence of a central controller leads to the term of autonomous control. Peer-to-peer
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 communication are gossiping
[54] and consensus algorithms [55]. A design of multi-agent based voltage control system for peer-
to-peer LV networks has been introduced in [56].
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 make this architecture a robust way of controlling a microgrid. Besides, all
information is kept local, eliminating possible privacy concerns.
19/84
Section 3 of this report presents two novel voltage control algorithms, based on peer-to-peer control
and gossiping communication. The proposed algorithms are able to operate in a distributed manner,
with no central coordinator, thereby keeping all control local and eliminating any single point of
failure.
1.2.3 Demand side integration
Modern power system are facing increasing stress due to the largely growing demand that are
approaching generation system capacity and network infrastructure limits [57]. The potential
threats of the demand increase are higher electricity costs and reliability risks [58].
Traditional approach to tackle this upcoming challenge is to deploy more generation resources to
balance demand/supply and to reinforce the network infrastructures to accommodate the growing
demand and generation. However, the financial risks and permitting difficulties make the system
operators exploiting a more efficient and economical way rather than simply conduct system
expansions and infrastructure reinforcements [59].
Recently, vast of researches have been focusing on how to accommodate the demand growth within
the existing infrastructure by encouraging the effective and efficient use of electricity. The adoption
of these demand side activities can achieve higher system utilization rate and reduced network
pressure by incentivising the change of load shape, energy efficiency, etc., towards more efficient
and economical ways [57]-[59]. Therefore, Demand Side Integration (DSI) has been proposed to
refer to all demand side activities focused on advancing the efficiency of electricity utilization [60],
including demand response and energy efficiency.
1.2.3.1 From Demand Side Management (DSM) to Demand Side Integration (DSI)
In 2006, CIGRE has established working group C6.9 to propose Demand Side Integration (DSI).
In restructuring power system with competitive market arrangements, DSI refers to all activities
focused on advancing the efficiency of electricity utilization [60], including demand response and
energy efficiency.
20/84
Figure 1.3 Concept and Objectives of Demand Side Integration [59]
The terminology DSI is a novel concept evolved from the term Demand Side Management (DSM).
In traditional power system, especially in vertically integrated utility environment with centralized
control paradigm. In modern restructured system with competitive market arrangements, more
demand side activities other than DSM programs can achieve similar target with the power of
competitive market. Therefore, the concept of DSI is thus emerged to represent the collection of all
relevant demand side activities that lead to more efficient and effective use of electricity. The range
of DSI is not only focusing on technical driven methods in DSM, but also include customer oriented
methods, such as alternative price methods, financial incentives, demand response [59].
As shown in Figure. 1.3(1), DSI can be classified into three general groups according to its
objectives: 1) Demand Response (DR) programs, 2) Energy Efficiency (EE) programs, and 3)
Strategic Load Growth (SLG) programs. In details:
1) DR program: DR programs refers to DSI activities that designed for short-term operational
targets in means of load shape objectives. These load shape objectives includes: peak
clipping (PC), valley filling (VF), load shifting (LS) and dynamic energy management
(DEM).
2) EE program: Unlike DR programs, EE programs focuses on methods that can take positive
effect on energy efficiency. For instance, building codes, appliance standards and
regulations that can further influence energy efficiency. The benefits are assessed in terms
of savings in generation, transmission and distribution infrastructure, and emissions
reductions [61].
3) SLG programs: SLG programs have the objective of promoting the load level through
electrification. This load rise may motivated by reduction in the use of fossil fuels and raw
materials [62]. However, this target is less concerned since the major purpose of DSI at this
stage is to accommodate the growth of demand side resources within the existing system.
21/84
1.2.3.2 Implementation methods of Demand Side Integration (DSI)
According to the report of CIGRE Working Group C6.9, a variety of implementation methods are
available for DSI programs, including:
1) Alternative Pricing: Due to the power of deregulated energy market, the visibility and
availability of differing supply service choices are almost equivalent to customers. Therefore, vast
of pricing methods are available to DSI targets. These methods lying on differing time horizons,
such as real time pricing aim to next day demand targets, pricing with seasonal rate takes effect
periodically in monthly basis. In conclude, alternative pricing methods are customer oriented, and
empowered by the competitive market arrangements.
2) Financial Incentives: Many incentive-based DR programs have been realized, such as cash
grant, discount rate, and so on. These methods are not only to incentivize customer engagement,
but also in response to DSI targets by changing the customer behaviors under the financial
incentives.
3) Technical-driven Methods: Unlike the pricing and financial incentive methods, technical-
driven methods are based on customers’ engagement. It relies on controllable and flexible load as
well as corresponding management system. For example, EMS, direct load control, curtailable load
are not customer oriented methods, with quick response speed (from minutes to hours).
4) Standards, Regulations, Information program: Standards and Regulations often take slow
effect but with great influence in long-run targets. Such as industry efficiency requirements,
building codes, renewable portfolio standards. While information programs attempts to affect the
customer behaviors through customer education, TV & internet advertisements, to make customers
aware of how to use electricity efficient and effective, hence make a solid influence to the power
system.
1.2.4 Coordinated control with multiple objectives
Distribution Network Reconfiguration (DNR) is the process of altering the topology of an electrical
distribution network by changing switch status (open/close), thus redirects power flows within the
network, in order to achieve certain objectives while satisfying network operational constraints.
DNR is a mixed-integer, non-linear optimization problem [63]. Although in theory a global optimal
configuration can be obtained by enumerating all feasible solutions and choosing the one which
meets the objective best, simple exhaustive searches are rarely sufficient for the complicated real
DNR applications.
There are solutions using heuristic algorithms to eliminate choices that are unlikely to lead to good
configurations. In [64], a branch and bound type heuristic method is presented. Although this
method is rapid in determining a configuration which reduces the power losses, it considers only
one pair of switching operations at a time. The result highly depends on initial switch states. A
22/84
power flow method-based heuristic algorithm is proposed in [65]. However, this method is only
suitable for small systems and becomes prohibitive when handling large distribution networks.
There are also DNR approaches based on artificial intelligence, which often incorporates with
metaheuristics. In [66], a particle swarm optimization method is proposed to maximize the power
system reliability while minimizing the power losses. A harmony search algorithm for DNR
problems with the objectives of improving voltage profiles and minimizing real power losses is
proposed in [67]. Compared to heuristic algorithms, metaheuristics are able to handle large-system
and multi-objective optimization problems. Furthermore, they are problem-independent so that can
be applied to different networks.
A number of literatures formulate the reconfiguration problem as multi-objective. In [68], a

simultaneous reconfiguration optimization method is proposed by aggregating all the objectives
into one and considering different weighting factors. In [69], a fuzzy satisfaction method is used.
It considers the desired objective value as a priori, which is hardly feasible in practice. In this work,
the Pareto optimality, which provides a set of diverse solutions representing trade-offs between
different objectives is used. The objectives focused on are power loss reduction, load balance and
DG penetration level increase.
An alternative solution to redirect power flows without changing the network topology is the
implementation of Soft Open Point (SOP). It is a power electronic device installed in place of a
Normally Open Points (NOP) between adjacent feeders of radial distribution networks, having the
capability to transfer real power and control reactive power between its connecting points [70]. The
optimal SOP outputs are determined based on the result of DNR to provide further improvements
along each of the aforementioned objectives.
1.3 Deliverable 5.2
This deliverable D5.2 is the second report of the Work Package 5 (WP5) of the P2P-SmartTest
project, corresponding to “P2P-based probabilistic and predictive control functions”. This
deliverable will develop alternative P2P control paradigm of distribution networks and investigate
the technical requirements on the associated ICT infrastructure (i.e. communication techniques,
information exchange and state estimation techniques.
This report is structured as follows. First, in Section 1, the uncertainty of distribution network when
employing P2P based energy trading and the requirement of network control function was
introduced, and then four potential control functions and state-of-the-art of these control functions
were described. The following sections detail these four control functions: In Section 2, network
control through new active control device, i.e. SOP, was presented. In Section 3, P2P based
coordinated voltage control for distribution networks was depicted in detail. In Section 4, active
network control function using demand side integration was presented. In Section 5, a coordinated
control considering multiple objectives was described. Finally, Section 6 concluded the main
findings of the work carried out for exploring the network control functions for the P2P energy
trading.
23/84
2 NETWORK CONTROL THROUGH SOP
In distribution network, there are only very limited information and communication infrastructure.
In today’s distribution networks, real-time monitoring and control below the primary substations
is very limited due to the lack of sensors, communication systems and control equipment.
Consequently, the distribution network is a bottleneck of the electricity supply chain. The increased
penetration of distributed generation requires more innovative control, from active network control
devices to coordinated control methods.
This section presents detailed work of network control through power electronic devices, soft open
point (SOP).
2.1 Problem formulation

2.1.1 Modelling of soft open point (SOP)
Feeder # 1
33/11kV
P1, Q1
Feeder # 2 P2, Q2
VSC2 VSC1
Figure 2.1 An MV network with an SOP at the remote ends of two feeders
Q
II I
VSC2
. Q2
Q1 . VSC1
P
P2 P1
S2 S1
III IV
Figure 2.2 An example of an SOP’s operating point
Figure. 2.1 shows a one line diagram of an MV distribution network with an SOP connected at the
remote ends of two feeders. The two VSCs (i.e.VSC1 and VSC2) are connected via a common DC
bus. 𝑃1 and 𝑄1 represent the active and reactive power that VSC1 provides to Feeder 1, and 𝑃2 and
𝑄2 are the active and reactive power that VSC2 provides to Feeder 2.
24/84
In a general case, n (n ≥ 2) feeders can be connected through an SOP composed of n VSCs sharing
the same DC bus. The AC terminal of each VSC is normally connected to an AC network via a
coupling transformer. An SOP with two- or multi- VSCs introduces additional degrees of flexibility
for network operation, and the power flow through the SOP can be adjusted within operating limits.
In this work, an SOP with two VSCs (i.e. two AC terminals) is considered. The power provided by
an SOP can be modulated in the four quadrants of the power chart, and each VSC can operate in
any region of the four quadrants. Figure. 2.2 shows an example of an SOP’s operating point where
two VSCs operate in region I and II, respectively. The two axes in Figure. 2.2 are for the active and
reactive power. Positive values represent the VSC providing power and negative values represent
the VSC absorbing power. The circles represent the size (i.e. maximum apparent power, 𝑆1, 𝑆2 ) of
the corresponding VSC. The power provided by the VSCs cannot exceed their ratings, as shown in
√(P1 )2 + (Q1 )2 ≤ S1; (2-1)
√(P2 )2 + (Q2 )2 ≤ S2 . (2-2)
With appropriate control, both VSCs produce their individual voltage waveforms with the desired
amplitude and phase angle. This provides full (four-quadrant) control of the active and reactive
power at both AC terminals. The reactive powers provided, or absorbed, by the two terminals, i.e.
𝑄1, 𝑄2 , are independent; whilst the active powers, i.e. 𝑃1 and 𝑃2 , are not independent variables, as
the sum of the active powers should be equal to zero, as shown in
∑ (PJ + P LOSS ) = 0 (2-3)

J=1, 2
where 𝑃𝑙𝑜𝑠𝑠 is the power losses of the SOP, including losses in the converters (conduction and
switching losses), the DC link capacitor, the filter and the coupling transformers.
With the use of the modular multi-level converter (MMC) technology, the operating loss of a VSC
is relatively low, approximately 1% per converter [71]. Therefore, for simplicity, the losses of the
SOP are neglected, and Equation (3) reduces to
∑ PJ = 0 (2-4)
J=1, 2
The SOP adopted is based on the MMC VSC technology at a commercially available size [28]. The
SOP is capable of providing the required power within its operational constraint. The detailed
control principle of the VSCs can be found in [72].
2.1.2 Jacobian matrix based sensitivity analysis
25/84
The sensitivity of voltages and currents in a network to the SOP’s active and reactive power
injections was analyzed using the Jacobian matrix [73]. For all the PQ buses, the voltage sensitivity
to the bus injection of active and reactive power is calculated from
∂P ∂P −1
|V|
∆Δ ∆P ∂Δ ∂|V| ∆P
[ ] = J−1 [ ] = ×[ ] (2-5)
∆|V|⁄|V| ∆Q ∂Q ∂Q ∆Q
|V|
[∂Δ ∂|V|]
where 𝑉 is the vector of the nodal voltage, and 𝛿 is the voltage angle; 𝑃 and 𝑄 are the injection of
active and reactive power at a node; 𝐽 is the Jacobian matrix.
The Jacobian matrix changes over time as the network configuration and load and generation
conditions vary. For a given time instant, t, the Jacobian matrix is considered constant, and the
corresponding inverse matrix is expressed by
C Δ2 P 2 ⋯ C Δ2 P N C Δ2 Q2 ⋯ C Δ2 QN
T T T T
⋮ ⋯ ⋮ ⋮ ⋯ ⋮
C ΔN P 2 … C ΔN P N C ΔN Q2 … CΔN QN
T T T T
J−1 T = (2-6)
CV2 P2 ⋯ CV2PN T CV2 Q2 ⋯ CV2 QN
T T T
⋮ ⋯ ⋮ ⋮ ⋯ ⋮
[CVN P2 T … CVN PN T CVN Q2
T
… CVN QN ]
T
where 𝑁 is the total number of nodes. 𝐶𝛿𝑖 𝑃𝑗 is a constant coefficient for the sensitivity between
𝑡
voltage angle, 𝛿𝑖 , and the active power injection at node 𝑗. 𝐶𝛿𝑖 𝑄𝑗 is a constant coefficient for the
𝑡
sensitivity between 𝛿𝑖 and the reactive power injection at node 𝑗. 𝐶𝑉𝑖 𝑃𝑗 is a constant coefficient
𝑡
for the sensitivity between voltage magnitude at node 𝑗, 𝑉𝑖 and 𝑃𝑗 . 𝐶𝑉𝑖 𝑄𝑗 is a constant coefficient
𝑡
for the sensitivity between 𝑉𝑖 and 𝑄𝑗 . 𝑖 = 2, 3, 4, … … 𝑁, j=2, 3, 4, … … 𝑁. Note that node 1 is
a “slack” or “infinite” bus, where the voltage magnitude is specified and phase angle is assumed to
be zero. Therefore it is not included in the Jacobian matrix.
At time instant t, the sensitivity of voltage angle and magnitude at node i are calculated by
∆𝛿𝑖 𝑡 = ∑ (𝐶𝛿𝑖 𝑃𝑗 ∗ ∆𝑃𝑗 𝑡 + 𝐶𝛿𝑖 𝑄𝑗 ∗ ∆𝑄𝑗 𝑡 ) (2-7)

𝑡 𝑡
𝑗=2
𝑁
∆|𝑉𝑖 | 𝑡 = |𝑉𝑖 | 𝑡 ∗ ∑ (𝐶𝑉𝑖 𝑃𝑗 ∗ ∆𝑃𝑗 𝑡 + 𝐶𝑉𝑖 𝑄𝑗 ∗ ∆𝑄𝑗 𝑡 )

𝑡 𝑡 (2-8)
𝑗=2
(𝑖 = 2, 3, 4, … … 𝑁)
It is assumed that, at time instant t, apart from the SOP’s active and reactive power injections, the
change in active and reactive power at all nodes is zero. Despite the variations in demand and
generation, the control of the SOP is able to be made as fast as milliseconds when using the power
electronic devices [72]. There might be delays in data measurement and communications which
26/84
will result in delay of the SOP control and require a sophisticated control algorithm design, but this
is out of the scope of this work. The assumption made here allows the impact of the SOP on the
network to be analysed, and more rigorous studies on real-time control of the SOP will be carried
out in the future. Hence, the voltage angle and magnitude at node i are presented as
𝛿𝑖 𝑆𝑂𝑃 = 𝛿𝑖 𝑡 + ∆𝛿𝑖 𝑡 (2-9)

𝑡
|𝑉𝑖 |𝑆𝑂𝑃 𝑡 = |𝑉𝑖 | 𝑡 + ∆|𝑉𝑖 | 𝑡

(2-10)
(𝑖 = 2, 3, 4, … … 𝑁)
where ∆𝛿𝑖 𝑡 and ∆𝑉𝑖 𝑡 only consider SOP’s active and reactive power injections.
2.2 Optimization formulation
Three optimization formulations were considered, each with a different objective. For each
optimization formulation, the voltage angle and magnitude at node i (i.e. 𝛿𝑖 𝑆𝑂𝑃 , |𝑉𝑖 |𝑆𝑂𝑃 𝑡 ) were
𝑡
calculated by Equations (2-7), (2-8), (2-9), and (2-10), and the active and reactive power provided
by the two VSCs were the decision variables.
2.2.1 Voltage profile improvement (VPI)
When improving the voltage profile of the network is desired, the objective function is
𝑁
2
𝐦𝐢𝐧 ∑(|𝑉𝑖 |𝑆𝑂𝑃 𝑡 − |𝑉𝑛𝑜𝑚 |) (2-11)
𝑖=1
where |𝑉𝑛𝑜𝑚 | is a target voltage for all nodes of the network.
This objective function leads to an optimal dispatch of the SOP’s active and reactive power values
to bring all nodal voltages as close as possible to the target value. The nominal voltage, i.e. 1 p.u.,
was taken as the target voltage, because this is considered as a mid-point of the future scenarios,
given that the integration of DG results in voltage rise and the electrification of transport and
heating leads to low voltages.
2.2.2 Line utilization balancing (LUB)
When the line utilization of the network is to be balanced, the objective function is
𝑛𝑙
2
𝐼𝑘
𝐦𝐢𝐧 ∑ ( ) (2-12)
𝐼𝑘 rate
𝑘=1
27/84
where 𝑛𝑙 is the total number of branches, and 𝐼𝑘 and 𝐼𝑘 rate are the actual and rated current of
branch k. Assuming that the node numbers of the two terminals of the branch k are i and j, and 𝑌𝑖𝑗
is the admittance between nodes i and j, the actual current in the branch k can be expressed by
𝐼𝑘 = 𝑌𝑖𝑗 ∗ (|𝑉𝑖 |𝑆𝑂𝑃 ∠𝛿𝑖 𝑆𝑂𝑃 − |𝑉𝑗 |𝑆𝑂𝑃 ∠𝛿𝑗 𝑆𝑂𝑃 ) (2-13)
to achieve balancing of line utilization.
2.2.3 Energy loss minimization (ELM)
When the energy losses of the network are to be minimized, the objective function is
𝑇
𝑛𝑙
𝐦𝐢𝐧 ∑( ∑ (𝐼𝑘 )2 ∗ 𝑟𝑘 + 𝑃𝑇 𝑙𝑜𝑠𝑠 ) ∗ 𝑡 (2-14)
𝑘=1
𝑡=0
where 𝑃𝑇 𝑙𝑜𝑠𝑠 is the power losses of the transformers, 𝑟𝑘 is the resistance of the network branch k,
and T is the time span of the period of interest.
to achieve the lowest line and transformer energy losses.
2.2.4 Constraints
Together with the constraints shown in (2-1), (2-2) and (2-4), the operation of the network cannot
breach the voltage and thermal limits, as shown in
𝑉𝑚𝑖𝑛 ≤ |𝑉𝑖 |𝑆𝑂𝑃 𝑡 ≤ 𝑉𝑚𝑎𝑥 (2-15)
|𝐼𝑘 | ≤ 𝐼𝑘 rate (2-16)
2.3 Visualization of the SOP operating region
Considering the voltage constraints, the SOP’s active and reactive power operating region is
visualized in the four quadrants of the power chart. For illustration purposes, the charts for four
different scenarios are presented.
a) Undervoltage in Feeder 1, undervoltage in Feeder 2 (Figure. 2.3);

b) Undervoltage in Feeder 1, voltage within limit in Feeder 2 (Figure. 2.4);
c) Undervoltage in Feeder 1, overvoltage in Feeder 2 (Figure. 2.5);
d) Overvoltage in Feeder 1, overvoltage in Feeder 2 (Figure. 2.6).
28/84
“Undervoltage”, “overvoltage”, and “voltage within limit” are used to define the feeder voltage
status without considering the SOP’s power injection. “Undervoltage” in a feeder means that the
feeder is relatively heavily loaded, and undervoltage occurs when there is no power injection from
the VSC. “Overvoltage” means that the feeder has more distributed generation, and overvoltage
occurs when there is no power injection from the VSC. “Voltage within limit” means that the
voltages in the feeder are within the limits when there is no power injection from the VSC.”
Figure. 2.3 shows two overlaid circles representing the active and reactive power limits for the two
VSCs of the SOP, assuming the two VSCs are the same size. The circle for VSC2 (connected to
Feeder 2) was mirrored in the y-axis so that the allowable operating region can be visualised. This
is done because the active power of the two SOP terminals must be symmetric (see Equation (4)).
As a consequence, for any operating point in Figure. 2.1, Equation (4) is met.
In Figure. 2.3, the two solid (and dashed) lines represent the lower and upper voltage limits of
Feeder 1 (and Feeder 2). These lines are constant voltage loci, i.e. the active and reactive power
operating points on each line bring the voltage to the same value. The boundary of the voltage
limits was obtained from Equation (2-15), and the |𝑉𝑖 |𝑆𝑂𝑃 𝑡 in Equation (2-15) was calculated from
Equations (2-10) and (2-8). The linear relation of the nodal voltages to the active and reactive power
from the SOP shown in Equation (2-8) resulted in linear boundaries when considering the voltage
limits. Therefore, to keep voltages in Feeder 1 (and Feeder 2) within the limits, the SOP’s active
and reactive power values must be within the two solid (and dashed) lines. The common active and
reactive power operating points on the circle, where VSC1 and VSC2 share the same active power
values (i.e. on a vertical line), are the operating regions for each VSC (i.e. the voltage constraint
for each feeder). Therefore, the ultimate voltage constraints for the two feeders are shown in the
red and blue blocks.
Similarly, the voltage constraints for scenarios b), c) and d) are presented in Figure. 2.4, Figure.
2.5 and Figure. 2.6. These Figures present the active and reactive power regions where the SOP
should operate under various scenarios.
In a general case, the SOP’s active and reactive power operating region varies depending on the
actual voltages and the sensitivity of the voltages to the active and reactive power injections of the
SOP (i.e. slope of the constant voltage loci) of the two feeders. “Undervoltage”, “overvoltage”, and
“voltage within limit” were in a way to reflect a feeder’s load and generation conditions. The
graphical method provides a general idea of active and reactive power regions at which an SOP
operates when the two feeders are under various load and generation scenarios, see Table 2.1. A
qualitative analysis with a graphical visualization not only helps network operators to understand
an SOP’s operating status, but also provides high level operational decision supports, such as
choosing control schemes, restraining operational boundaries, etc.
29/84
II I
Feeder 2 voltage constraints
Feeder 2: V2 = Vmax Feeder 1 voltage constraints
Feeder 2: V2 = Vmin Feeder 1: V1 = Vmax
III IV Feeder 1: V1 = Vmin

Figure 2.3 VSC operating regions with voltage constraints: Undervoltage in Feeder 1 and
undervoltage in Feeder 2 (VSC2 mirrored in the y-axis)
II I
Feeder 2: V2 = Vmax
Feeder 1: V1 = Vmax
Feeder 1: V1 = Vmin
Feeder 2: V2 = Vmin III IV Feeder 2 voltage constraints
within limit in Feeder 2 (VSC2 mirrored in the y-axis)
Q
Feeder 1: V1 = Vmax
Feeder 1: V1 = Vmin
II I
P
Feeder 2: V2 = Vmax Feeder 2: V2 = Vmin

III IV
overvoltage in Feeder 2 (VSC2 mirrored in the y-axis)
30/84
II Q I
Feeder 1: V1 = Vmax Feeder 2: V2 = Vmax
Feeder 1: V1 = Vmin Feeder 2: V2 = Vmin

III IV
Figure 2.6 VSC operating regions with voltage constraints: Overvoltage in Feeder 1 and
overvoltage in Feeder 2 (VSC2 mirrored in the y-axis)
Table 2.1 Operating region in a four quadrant chart of an SOP

Feeder 1
Undervoltage Voltage within limit Overvoltage
Undervoltage I, II I, II, III II, III
Voltage within
Feeder 2 I, II, IV - II, III, IV
limit
Overvoltage I, IV I, III, IV III, IV
1 2 3 4 5 6 7 8 9
Feeder # 1
Busbar
10 11 12 13 14 15 16 17 18 19
Feeder # 2
P1, Q1
20 21 22 23 24 25 P2, Q2
33/11kV
Feeder # 3
26 27 28 29 30 31 32 33 34
Feeder # 4
Figure 2.7 One line diagram of a radial MV network (The cable diameters of the first half of the
feeders, i.e. closer to the substation, are bigger than the second half, therefore they are shown in
thicker lines)
31/84
1.0
residential
Normalized load profile

industrial
0.8 commercial
0.6
0.4
0.2
0
12am 6am 12pm 6pm 12am 6am 12pm 6pm 12am
Time of day, half-hourly
Figure 2.8 Load profiles of the network (a weekend day and a weekday, data obtained from [29])
1.0
Normalized DG profile
0.8
0.6
0.4
0.2
Wind
PV
0
12am 6am 12pm 6pm 12am 6am 12pm 6pm 12am
Time of day, half-hourly
Figure 2.9 Wind and PV profiles of the network (data obtained from [28])
Table 2.1 Load and DG data

Load Distributed generation
Feeder #.
Type Total peak (MVA) Type Locations Ratings (MW)
1 Com. 3 Wind 5, 9 2, 0.5
2 Res. 2.5 PV 11, 14, 17 1, 2, 2.5
3 Ind. 3 Wind 22, 25 0.5, 2.5
4 Res. 2.5 PV, wind 30, 33 0.5, 0.5
2.4 Case Study

2.4.1 MV distribution network model
An example distribution network obtained from [77] was used with some modifications. As shown
in Figure. 2.7, the 11 kV network consists of four radial feeders (three-phase underground cables)
32/84
with different lengths (each segment is 1 km) and load types. The rated capacity of the 33/11 kV
transformer is 20 MVA. An SOP is connected at the remote ends of Feeder 2 and Feeder 3, with a
rated capacity of 3 MVA for each VSC. Various DGs are installed at different locations to represent
the expected load/generation distribution between feeders. Table 2.2 shows a summary of the load
and generation data. The impedance of the first half of the feeder (close to the substation) is
0.164+j0.08 Ω / km, and the rated current is 335 A per phase, and the second half is
0.320+j0.087 Ω/ km, with a rated current of 230 A per phase.
Figure. 2.8 shows the profiles for residential, commercial and industrial loads for two days (a
weekend day and a weekday), with power factors of 0.98, 0.95, and 0.90. Figure. 2.9 shows the
generation profiles for wind and PV systems for the two days under study, and all wind and PV
generations are considered to operate at unity power factor. These load data was obtained from [75]
and wind and PV generation data was obtained from [74]. These load and generation profiles are
all normalized to their own peak values, and the corresponding peak load of a feeder and the peak
of a DG unit are shown in Table 2.2.
Power flow calculations of the network were carried out in MATLAB using Newton-Raphson
method, and the tolerance of iteration was considered as 0.001 per unit. Jacobian matrix was
obtained when the power flow solution reached convergence. Then the non-linear programming
optimisation with non-linear constraints was also carried out in MATLAB, where an Interior-Point
algorithm and Hessian matrix were used to find the optimal solutions. The power flow calculations
and optimizations were run at each time step, i.e. every 30 minutes, for the proposed three objective
functions. The relevant computation experiments were performed on a desktop machine, Intel (R)
Core (TM) i7-4790 CPU @ 3.6 GHz, 16GB RAM and MATLAB version R2014a, and the
optimization process for each time step is approximately 450 milliseconds.
2.4.2 DG penetration level
DG penetration level is defined as the total of the rating of each individual DG in the network in
relation to the 33/11 kV transformer rated capacity, as shown in
𝑆𝑢𝑚 𝑜𝑓 𝑎𝑙𝑙 𝐷𝐺 𝑟𝑎𝑡𝑖𝑛𝑔𝑠

𝑃𝑒𝑛𝑒𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝐿𝑒𝑣𝑒𝑙 = (2-17)
33/11𝑘𝑉 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚𝑒𝑟 𝑟𝑎𝑡𝑖𝑛𝑔
Various penetration scenarios were considered by scaling up/down the rating of each DG. For
instance, the DG rating (MW) values shown in Table 2.2 represent a DG penetration level of 60%.
With these DG rating (MW) values halved, the penetration level is 30%. With 1.5 times of these
DG rating (MW) values, the penetration level is 90%.
2.4.3 Two-day performance with a 90% DG penetration – full observability of the network
Figure. 2.10 shows the performance of the network with the SOP and a 90% DG penetration. ±3%
of nominal was considered as the voltage limit [33], [76]. Figure. 2.10(a) presents the active and
33/84
reactive power consumption of the network. Part (b) of the figure is the voltage profiles at the
busbar and remote ends of the four feeders without an SOP. Part (c) and part (d) are the dispatched
active and reactive power values of the SOP and voltage profiles when optimizing the voltage
profile using the VPI objective. The dispatched active and reactive power values and the resultant
voltage profiles when using the line balance LUB and energy loss ELM objectives are shown in
part (e) and (f), and part (g) and (h).
Note that, without an SOP, overvoltages occurred in Feeder 2 and Feeder 3. These overvoltages
did not occur when using an SOP, irrespective of the objective function used. When using the VPI
objective, the voltage profiles were better (i.e. closer to the 1 pu target value) than when using the
LUB or ELM objective. In terms of dispatched active and reactive power values from the SOP, the
active values were similar when adopting the three objective functions. However, more reactive
power from SOP was dispatched when using the VPI objective than when using the LUB or ELM
objective.
The dispatched active and reactive power values (i.e. set-points) of the SOP are shown in the four
quadrant power chart in Figure. 2.11. As shown, more operating points were close to, or on, the
edge of the circle using the VPI objective than using the LUB or the ELM objective, and this also
illustrates more reactive power from SOP was dispatched using the VPI objective. The objectives
to achieve line balancing LUB and energy loss minimization ELM mainly relied on real power
exchange.
8 1.07
No SOP Busbar
4 VPI 1.06 Feeder 1 Bus 9
FLB Feeder 2 Bus 19
P, MW
0 ELM 1.05 Feeder 3 Bus 25

Feeder 4 Bus 34
-4 1.04
Upper limit
-8
Voltage, pu
1.03
-12 1.02
8
6 1.01
Q, Mvar
4 1.00
2 0.99
0
-2 0.98
-4 Lower limit
0.97
12am 6am 12pm 6pm 12am 6am 12pm 6pm 12am 12am 6am 12pm 6pm 12am 6am 12pm 6pm 12am
(a) Network power consumption (b) Voltage profiles without an SOP

3 1.05
Pm Busbar
Qm 1.04 Feeder 1 Bus 9
2 Pn Feeder 2 Bus 19
Qn Upper limit Feeder 3 Bus 25
1.03 Feeder 4 Bus 34
1
P/Q, MW/MVar
Voltage, pu
1.02
0 1.01
1.00
-1
0.99
-2
0.98
-3 0.97 Lower limit
(c) P and Q set-points for SOP - VPI (d) Voltage profiles using an SOP - VPI
34/84
3 1.05
Busbar
Feeder 1 Bus 9
2
Upper limit Feeder 3 Bus 25
P/Q, MW/MVar
Voltage, pu
1.02
0 1.01
1.00
-1
Pm 0.99
-2 Qm
Pn 0.98
-3 Qn 0.97 Lower limit
(e) P and Q set-points for SOP - LUB (f) Voltage profiles using an SOP - LUB
3 1.05 Busbar
Feeder 1 Bus 9
2 Feeder 3 Bus 25
Upper limit
1
P/Q, MW/MVar
Voltage, pu
1.02
0 1.01
1.00
-1
0.99
Pm
-2 Qm
Pn
0.98
-3 Qn 0.97 Lower limit
Time of a day, half-hourly Time of a day, half-hourly
(g) P and Q set-points for SOP - ELM (h) Voltage profiles using an SOP - ELM
Figure 2.10 Two-day performance of SOP for a 90% DG penetration
4 Q, MVar
II I
3
0 P, MW
-1 VPI - VSC1
VPI - VSC2
-2 LUB - VSC1
LUB - VSC2
-3 III IV ELM - VSC1
ELM - VSC2
-4
-4 -3 -2 -1 0 1 2 3 4
Figure 2.11 P and Q set-points of the SOP for different optimization objectives and a 90%
DG penetration
35/84
Without SOP - min Without SOP - max

1.06 VPI - min VPI - max
LUB- min LUB - max
Max and min voltage, pu

1.04 ELM - min ELM - max
Upper limit
1.02
1.00
0.98
Lower limit
0.96
0 10 20 30 40 50 60 70 80 90
Penetration level, %
Figure 2.12 Maximum and minimum voltage – full network observability
100% Without SOP

VPI
90% LUB
Maximum line utilization
ELM
80%
70%
60%
50%
40%
0 10 20 30 40 50 60 70 80 90
Figure 2.13 Maximum loading (line utilization) – full network observability
16 Without SOP
14 VPI
LUB
ELM
Energy losses, MWh
12
10
0
0 10 20 30 40 50 60 70 80 90
Figure 2.14 Total energy losses for the two days – full network observability
36/84
Table 2.3 Overall performance of the network with limited network observation
Without an SOP Voltage profile improvement Line utilization balancing Energy loss minimization
Penetrat
Voltage, pu Max Max Max Max
ion Losses, Voltage, pu Losses, Voltage, pu Losses, Voltage, pu Losses,
level, % Min Max line use, line use, line use, line use,
MWh Min Max MWh Min Max MWh Min Max MWh
% % % %
0 0.973 1.000 53.0 2.55 0.978 1.000 49.9 9.4 0.978 1.000 41.5 1.48 0.978 1.000 41.5 1.46
10 0.974 1.004 51.2 2.16 0.979 1.004 48.9 7.7 0.979 1.003 39.9 1.14 0.979 1.002 39.9 1.12
20 0.974 1.009 49.6 2.01 0.979 1.007 49.2 6.8 0.979 1.007 39.5 1.01 0.979 1.007 39.5 0.98
30 0.974 1.014 48.2 2.10 0.979 1.007 56.8 7.5 0.979 1.012 39.5 1.08 0.979 1.013 39.5 1.05
40 0.974 1.020 47.0 2.42 0.979 1.009 61.2 8.7 0.979 1.017 39.5 1.34 0.979 1.019 39.5 1.31
50 0.974 1.028 47.7 2.96 0.979 1.012 65.4 9.7 0.979 1.023 42.3 1.79 0.979 1.025 45.2 1.76
60 0.974 1.036 60.8 3.72 0.979 1.016 70.0 11.1 0.979 1.029 53.7 2.44 0.979 1.029 54.4 2.41
70 0.974 1.044 73.8 4.70 0.979 1.024 75.4 12.7 0.979 1.032 65.9 3.31 0.979 1.032 66.9 3.28
80 0.974 1.052 86.7 5.89 0.979 1.028 86.5 14.0 0.979 1.032 80.5 4.57 0.979 1.032 81.8 4.54
90 0.974 1.059 99.4 7.28 0.979 1.031 98.5 15.1 0.979 1.032 98.3 6.26 0.979 1.032 98.4 6.23
2.4.4 Overall performance – full observability of the network

The network performance was examined by investigating the maximum and minimum voltages,
the maximum line utilization and the energy losses of the network with DG penetration from 0 to
90% (10% per step), and they are presented in Figure. 2.12 – Figure. 2.14.
Without using an SOP, overvoltages were present from 60% DG penetration, see Figure. 2.12.
When using an SOP and disregarding the objective function used, the network reached 90% DG
penetration without violating the voltage limits. The maximum voltages were kept lower when
optimizing the voltage profile using the VPI objective than when using the line balance LUB or
energy loss ELM objective, along all DG penetrations.
As shown in Figure. 2.13, for DG penetrations from 10 to 60%, the maximum line utilization was
always larger when adopting the VPI objective than the case without an SOP. This was because the
increased reactive power injection from the SOP resulted in an increase in the currents of some
circuits. However, a slight decrease was shown when the penetration was more than 70%. This was
because with higher penetrations of DG, the VSCs of the SOP began to consume reactive power
reducing currents of some parts of the network. When using the LUB or the ELM objective, the
maximum line utilization was always smaller than the case without an SOP, for all DG penetrations.
In Figure. 2.14, it is shown that, by using the VPI objective, the total energy losses were
approximately twice those without an SOP. Considering only the loads (i.e. 0% DG penetration),
the total energy losses for the two days were 2.55 MWh, which corresponds to approximately 1%
of the total energy consumption. With a 90% DG penetration, and without an SOP, the total energy
losses were 7.28 MWh (i.e. ~2.9% of the total energy consumption). Also, due to the increased
dispatch of reactive power, adopting the VPI objective resulted in a significant increase in losses,
where the total energy losses were 15.1 MWh (i.e. ~5.9% of the total energy consumption). In
contrast, a reduction of energy losses was shown throughout all penetration levels when adopting
LUB or ELM as the objective. For a 90% DG penetration, the energy losses when using the LUB
were 6.13 MWh, and using the ELM the energy losses were 6.12 MWh, and both showed a slight
reduction compared to the case without an SOP.
37/84
2.4.5 Overall performance – limited & no observability of the network
In reality, the control system will not have full network observability. A case in which loads were
20% smaller than the optimization input data was considered, assuming all DGs were correctly
measured. The reduction rate for each load was randomly selected from a range of 10% to 30%,
and the overall load reduction was 20%. This was a simple assumption made here to represent a
global measurement error, when a network has limited measurements. The overall performance of
the network is shown in Table 2.3. In this table, voltages above the limit are marked in red. It is
seen that, when using the LUB and ELM objectives, overvoltages were present from 70% DG
penetration; whilst, when using the VPI objective, the voltages were within the limits until the DG
penetration reached 90%. For the network without an SOP, as the DG penetration increased,
voltage excursions were encountered before thermal overloading. Therefore, the method
optimizing voltage profiles, VPI, performed better than the line utilization balancing or loss
minimization method (e.g. LUB or ELM), in terms of mitigating the network voltage and thermal
constraints.
At minimal network observability, only the voltages at the SOP terminals are known by the SOP
control system. In this case, the optimization formation with the voltage improvement VPI as the
objective was used, and Equation (2-11) only considered the voltages at the two terminals rather
than all the nodes. It is found that, by using an SOP, the network’s DG hosting capacity was
increased from 50 to 80%.
2.5 Discussion
An SOP with a given rating and location was considered. This research can be a framework for
higher level studies, including finding the optimal number and size of SOPs with different network
topologies and configurations. When a network is equipped with multiple SOPs, Equations (2-9)
and (2-10) should include all the terminals of all SOPs. When the two feeders, to which an SOP is
connected, are supplied by different substations, one Jacobian matrix is calculated for each
substation, and two sets of Equations similar to (2-9) and (2-10) are created in order to include both
terminals of the SOP.
Balanced three-phase load and generation were considered. Through adequate control of VSCs, the
SOP is able to provide three-phase unbalanced power injections. The method used in this research
to quantify the benefit of using an SOP could be applied to a three-phase unbalanced system.
In this work, a data set of 30-minute granularity with different load and generation conditions was
taken. This is because this work is mainly for planning purposes, to provide distribution network
operators with high level decision supports, e.g. selecting control schemes and restraining SOP
operation boundaries. However, given that the optimisation calculation for each run took
approximately 450 milliseconds, the optimisation is able to be run in real time, and the methodology
is able to be used for real-time operation purposes.
This work did not consider active control devices, such as battery storage, OLTC, capacitor banks,
etc. The control time frame of these active control devices is normally minutes or hours. Although
38/84
some electronic interfaced battery storage is able to change control settings on a millisecond
timeframe, due to the battery life time concern, battery banks normally operate in a steady state
time frame. In the contrast, SOPs are not constrained by mechanical wear or life time concern
therefore are able to change operating points more frequently. On the other hand, this work focuses
on the performances of SOP control schemes, and provides decision supports of selecting control
schemes. Hence, these active control devices were not considered. Future work can be undertaken
investigating the real-time operation of SOP with battery storage systems, OLTC, and capacitor
banks.
2.6 Conclusions
A non-linear programming optimization, to set the real and reactive power operating set-points for
an SOP on an 11 kV network, was developed. Through a Jacobian matrix based sensitivity analysis,
the SOP’s operating region was defined within its voltage-limit bounds, and visualized in a
graphical manner for different load and generation conditions at the grids or feeders at the two
terminals of the SOP. The exact operating point was determined using three optimization
objectives: voltage profile improvement, line utilization balancing and energy loss minimization.
Results showed that the use of an SOP significantly increases the network’s DG hosting capacity.
The control scheme using the objective for voltage profile improvement increased the headroom of
the voltage limits by the largest margin. This control scheme dispatched increased reactive power,
and, hence, was at the expense of increased energy losses.
The control schemes using the objectives to achieve line utilization balance and energy losses
minimization showed the most improvement in circuit utilization and in limiting energy losses,
mainly relying on the real power exchange between feeders.
This work does not make a suggestion on which optimisation objective is better than others, but
presents the performance of each and leaving the selecting options to network operators based on
their needs. Defining a unifying cost function is difficult, because the cost of breaching voltage and
thermal limits, and cost of energy losses may vary from network to network. According to a
network’s characteristics, network operators are able to devise the control scheme using one or
multi-objective functions. The proposed methodology provides a potential framework/solution for
electricity network operators, allowing them to choose appropriate control schemes more
effectively. This selection requires the network operators to attribute value to the increase in hosting
capacity, mitigation in voltage issues, the reduction in the maximum line utilization and the
reduction in energy losses.
39/84
3 COORDINATED VOLTAGE CONTROL

As more distributed energy resources (DERs) such as PV generation, batteries and electrical
vehicles are installed in the distribution grid, controlling the voltage to be within allowed limits is
becoming a major challenge. A possible solution is to design a control system to optimally
coordinate local generation, on-load tap-changers (OLTC) and other equipment used to control
voltage in distribution networks and microgrids. For this purpose, new control algorithms have to
be developed that are able to coordinate the DERs and other equipment and allow them to
participate in voltage control.
Different solutions for solving the voltage problem have been discussed in [77]. The developed
algorithms in this report are only used for the coordination of DERs, however, they can be extended
to include more equipment, i.e. OLTC.
The voltage on all nodes of the distribution grid has to be within the predefined limits. These limits
are typically depending upon the voltage level on the grid itself (e.g. tighter in medium voltage than
in low voltage grids). In Europe, for low voltage distribution networks, these limits are determined
by the EN50160 norm [78]:
- 90% 𝑉𝑛𝑜𝑚 ≤ 𝑉 ≤ 110% 𝑉𝑛𝑜𝑚 , for 95% of all 10 minutes average RMS values,
- 85% 𝑉𝑛𝑜𝑚 ≤ 𝑉 ≤ 110% 𝑉𝑛𝑜𝑚 , for all 10 minutes average RMS values,
with 𝑉𝑛𝑜𝑚 = 230 𝑉.
Up till now, voltage quality in the distribution grid is achieved based on the layout 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 anymore 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. Also the
reverse can occur: e.g. connecting a large amount of electric vehicles to charge in the evening, a
moment with already a high consumption peak, could lead to voltages going below allowed limits
[79], [80].
In the high voltage transmission network, the voltage problem is normally addressed by regulating
the reactive power output of the connected generators. This is because the transmission network
40/84
has a very low R/X ratio, linking the injected reactive power to the voltage drop over a line in the
following way:
𝑋𝑄𝐴
Δ𝑉 = 𝑉𝐴 − 𝑉𝐵 ≅
𝑉𝐴
However, in distribution grids, the R/X ratio is a lot higher (in the order of 5 to 10), thereby making
the reactive power less suited for performing voltage control. Reactive power can be used though,
however, large amounts will be needed, thereby introducing a considerable amount of additional
losses. Besides, the thermal limits of the lines should be respected at all times. A high R/X ratio
makes active power a good candidate for controlling voltage, however, as active power has an
economic value, modifying its consumption or production will have a considerable cost.
It is thus important that the voltage control method will maintain voltages within limits in an
optimal way. This can be translated in a well-known optimal power flow problem:
minimize ∑ 𝑐𝑗𝑃 (Δ𝑃𝑗2 ) + ∑ 𝑐𝑗𝑄 (Δ𝑄𝑗2 ) + 𝑐 𝑙𝑜𝑠𝑠 (Δ𝑃𝑙𝑜𝑠𝑠 ) (3-1a)

Δ𝑃,Δ𝑄 𝑗 𝑗
𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝑉𝑚𝑖𝑛 ≤ |𝑉𝑖 | ≤ 𝑉𝑚𝑎𝑥 , ∀𝑗𝑖 = 0, … , 𝑁 (3-1b)
Δ𝑃𝑚𝑖𝑛,𝑗 ≤ Δ𝑃𝑗 ≤ Δ𝑃𝑚𝑎𝑥,𝑗 , ∀𝑗 = 0, … , 𝑛 (3-1c)
Δ𝑄𝑚𝑖𝑛,𝑗 ≤ ΔQj ≤ Δ𝑄𝑚𝑎𝑥,𝑗 , ∀𝑗 = 0, … , 𝑛 (3-1d)
𝑆𝑖𝑗 ≤ Smax (3-1e)
|𝐼𝑖𝑗 | ≤ 𝐼𝑚𝑎𝑥 (3-1f)
Here, 𝑐𝑗𝑃 (Δ𝑃𝑗 ) represents the cost of a change in active power at node 𝑖 with an amount ΔP𝑗 ,
𝑐𝑗𝑄 (Δ𝑄𝑗 ) the cost of a change in reactive power at node 𝑖 with an amount Δ𝑄j, and 𝑐 𝑙𝑜𝑠𝑠 (Δ𝑃𝑙𝑜𝑠𝑠 ),
the cost of the additional losses created in the system with the control actions. These could also be
incorporated to the costs 𝑐𝑗𝑃 (Δ𝑃𝑗 ) and 𝑐𝑗𝑄 (Δ𝑄𝑗 ). In that case the last term can be ignored. The
change in active and reactive powers Δ𝑃𝑗 and Δ𝑄𝑗 will have to be within limits at each node 𝑖 =
0, … , 𝑛, with 𝑛 the number nodes participating in the voltage control. The absolute value of the
voltage |𝑉𝑖 | has to be within limits at all nodes 𝑖 = 0, … , 𝑁 of the grid. The apparent power 𝑆𝑖𝑗
flowing through each line connecting node 𝑖 and 𝑗 has to be maintained below the maximum value
Smax , as well as the current |𝐼𝑖𝑗 | through that line.
An additional problem is the fact that the current distribution grid is almost not monitored. This
means that potential issues and voltage problems will occur unnoticed. The grid operator currently
does not have any view on the real actual powers that are flowing in the grid, neither on the real
voltages and currents. Besides, the distribution grid consists of a geographically large area, with a
large amount of connection points and a complex topology. At the same time, the control system
has to be robust against any failures and ensure the stability of the grid at any time. This poses some
additional hard constraints on any type of communication and controlling infrastructure the grid
operator might impose.
41/84
3.2 Active and reactive power control of DERs
The effect of reactive power and/or a change in active power consumption on the voltage depends
on the type of the grid (the R/X ratio or the type of cable used) and the topology of the grid. The
effect of active and reactive power on the voltage can be calculated as follows. Let us examine a
single piece of line with power flowing through, from point A to point B, as depicted in Figure 3.1.
We transfer a complex power 𝑆𝐴 = 𝑃𝐴 + 𝑗 ∙ 𝑄𝐴 from point A. If we take the voltage at point A as
reference, the angle of this voltage is zero, and thus has only a real component. The impedance of
the line is 𝑍 = 𝑅 + 𝑗 ∙ 𝑋. The complex voltage drop over the line and the magnitude of the resulting
voltage at point B is described by following equations:
Figure 3.1: Power flow through a line
𝑅 ∙ 𝑃𝐴 + 𝑋 ∙ 𝑄𝐴 + 𝑗 ∙ (𝑋 ∙ 𝑃𝐴 − 𝑅 ∙ 𝑄𝐴 )
Δ𝑉 = 𝑉𝐴 − 𝑉𝐵 = (3-2a)
𝑉𝐴∗
(𝑅 2 + 𝑋 2 )(𝑃𝐴2 + 𝑄𝐴2 )
|𝑉𝐵 | = √|𝑉𝐴 |2 − 2(𝑅 ∙ 𝑃𝐴 + 𝑋 ∙ 𝑄𝐴 ) + , (3-2b)
|𝑉𝐴 |2
As can be seen, the influence of active and reactive power transported over a power line on the
voltage depends on the values of R and X of the power line. The bigger the resistance R, the bigger
the influence of the active power and the smaller the influence of the reactive power, as is the case
in distribution grids. However, even with rather large values of R/X ratios, reactive power can still
contribute to control the voltage, however, higher amounts of reactive power will be needed and
therefore, active power losses will increase as well.
Different DER technologies are able to contribute to this way of voltage control. The most common
DER is a classical PV inverter. As these consist of an active front end connected to the grid, they
have the capability to inject positive or negative reactive power into the grid. Different strategies
can be deployed for this, such as a local V-Q droop control that varies the reactive power injection
depending on the local voltage measured [81]. This is a very robust type of control, with the absence
of any form of communication. However, if a voltage problem occurs on a different position in the
grid than at the inverter is located, it will pass unnoticed, and there will be no control action.
Besides, PV inverters, all DERs with an active front end, such as batteries or chargers for electric
vehicles, can participate in reactive power control.
42/84
Another option is the adjustment of active power. A battery or a demand response application seem
to be the most appropriate DER devices for performing voltage control with active power, as the
active power can be stored (in the battery), or consumed at a different time (demand response) [82],
[83]. In this case, the active power consumption can typically be both increased and decreased.
Electric vehicles are also a suitable candidate for this, however, they might have stricter constraints
than a regular battery, as they have to be charged by a certain time. PV inverters can also perform
active power curtailment, thereby contributing to voltage stability. Without a battery however, this
energy will be lost, meaning that there is a considerable economic loss.
3.3 Monitoring and measurements

Besides the actual control of the voltage, another problem is the fact that the present-day
distribution grid is almost not monitored at all. It is therefore almost impossible to have an idea
about the state of the grid and to know where the problems occur.
An appropriate smart meter can be a solution to this problem. If this smart meter is able to measure
the voltage and has some form of communication, installing these smart meters at all connection
points will allow for monitoring the voltages over the grid. The same can be done for active power
consumption, however privacy might be an issue, as it is possible to determine the specific
appliances that are running from power consumption profiles with a high resolution.
Many voltage control methods require and accurate description of the grid’s topology. However,
distribution grids have often grown organically throughout time, with varying construction rules
and methods. Therefore, a correct topology of the distribution grid or even the considered feeder is
not always available. Smart grids can also be a solution here. The combination of power profiles
and voltage measurements at the connection points of the grid can be used to estimate the grid
topology. Regression or other machine learning techniques can be used here, that fit the measured
data to a grid model, resulting in sensitivity matrices or impedances between nodes. From this one
can easily obtain the real grid topology. For more information, see for instance [84], [85].
3.4 Proposed P2P voltage control algorithms

This section will present two decentralized voltage control algorithms, controlling the voltage
within limits by use of a change in active and/or reactive power of DERs installed in the distribution
grid itself. Both algorithms are based on an optimization problem, which is distributed over all
agents participating in the voltage control. The first has as objective the minimization of the voltage
deviations beyond limits and uses a gradient descent that is distributed over all agents. The second
algorithm has as objective the minimization of active and reactive power, with constraints the
voltages that have to be within limits. It uses a dual decomposition to distribute the optimization
and will actually control the dual variables linked to the violated constraints.
Other objective functions could be used, and by using a similar reasoning as the two algorithms
presented in this section, one can obtain a P2P voltage control algorithm.
43/84
3.4.1 A basic P2P voltage control algorithm
This section will describe a basic P2P voltage control algorithm, of which the only goal is to
minimize the voltage deviations beyond the allowed threshold, with all available resources, but
with no other objectives [86]. The algorithm is completely distributed, thereby keeping all control
local, using a simple, linearized grid model to calculate the required actions by the agents. The
algorithm is designed to work asynchronously, on different types of grid (both low voltage and
medium voltage with different impedances and R/X ratios) and is able to utilize both active and
reactive power of the DERs in the grid. At the controlled nodes, the voltages have to be measured
and communicated to all participating devices. So at each controlled node, an agent has to be
present that is able to measure voltages and participate in communication. The DER devices that
will actively react on any voltage deviations by modifying their active or reactive power, will be
called compensators. Note that not all nodes need to have a compensator assigned.
The algorithm defines three different layers that interact with each other:
 The grid layer: the part of the distribution grid that is included in the considered microgrid.
The voltage behaviour of the grid is linearized, in order to limit the computational
requirements of the participating agents. The R/X ratio of the grid and absolute value of the
impedance between different nodes of the grid is needed in the approach.
 The control layer: local intelligence at each agent that determines the action of the agent
based on the perceived state of the grid. In the presented algorithm, this is chosen to be an
adapted version of a least squares optimization to minimize the voltage deviations at all
nodes of the considered microgrid.
 The communication layer: determines how the locally measured grid voltages are
communicated to the required agents. This algorithm proposes a P2P communication
scheme, where every agent is able to talk to a random other agent in the microgrid. The
method used is a gossip-based push-sum algorithm.
Since both the control and the communication layer are based on converging algorithms, and the
former uses the output of the latter, they should act on different time scales. The gossiping
algorithm used in the communication layer should already be converged to a sufficient precision
before using these values as an input for the calculation of the correct actions in the control layer.
The time steps of the control layer should thus be larger than the time steps of the communication
layer.
A. Grid layer
As the voltage drop is a function of the transported power over a line, resulting from the traditional
power flow equations are nonlinear and depending on the 𝑅/𝑋 ratio of the considered line, the
proposed control algorithm will make use of a transformation and linearisation of the voltage
behaviour, as presented in [87]. The transformation allows for eliminating the dependency of the
𝑅/𝑋 ratio on the relation between the transported active and reactive power and the voltage drop
Δ|𝑉𝑖𝑗 | = |𝑉𝑖 | − |𝑉𝑗 | between buses 𝑖 and 𝑗. The active power P and reactive power Q transported
over a line are converted to a transformed active and reactive power P' and Q', as presented in
following equation:
44/84
𝑃′ 𝑋/𝑍 −𝑅/𝑍 𝑃
[ ]=[ ][ ]
𝑄′ 𝑅/𝑍 𝑋/𝑍 𝑄 (3-3)
Using this transformation in equation (3-2a) results in:

′
𝑍𝑖𝑗 𝑄𝑖𝑗 + 𝑗 ∙ 𝑍𝑖𝑗 𝑃𝑖𝑗′
Δ𝑉𝑖𝑗 =
𝑉𝐴∗
When assuming that the angle between the voltages is small, the imaginary part can be neglected,
and the absolute difference in voltages becomes:
𝑍𝑖𝑗 ′
Δ|𝑉𝑖𝑗 | ≅ 𝑄 , (3-4)
𝑉𝑛𝑜𝑚 𝑖𝑗
where 𝑉𝐴 is approximated to be equal to the nominal voltage.
B. Control layer
The objective of the considered algorithm is to minimize the sum of squares of all voltage
deviations beyond predefined limits Δ|𝑉𝑖 |, for all controlled nodes 𝑖 = 1, … , 𝑛 of the considered
grid:
𝑛
(3-5a)
minimize ∑ Δ|𝑉𝑖 |2
𝑄′
𝑖=1
′
𝑄𝑚𝑖𝑛,𝑗 ≤ 𝑄𝑗′ ≤ 𝑄𝑚𝑎𝑥,𝑗
′
, ∀ 𝑗 = 0, … , 𝑁 (3-5b)
subject to
where Δ|𝑉𝑖 | should be calculated as:
|𝑉𝑖 | − 𝑉𝑚𝑎𝑥 , 𝑉𝑚𝑎𝑥 < |𝑉𝑖 |
Δ|𝑉𝑖 | = { 0, 𝑉𝑚𝑖𝑛 ≤ |𝑉𝑖 | ≤ 𝑉𝑚𝑎𝑥 (3-6)
|𝑉𝑖 | − 𝑉𝑚𝑖𝑛 , |𝑉𝑖 | < 𝑉𝑚𝑖𝑛
The algorithm uses a gradient descent method to find the optimum and this in combination with a
backtracking line search together with Armijo’s condition [88]. A gradient descent method
decreases in each time step 𝑘 in the direction 𝑑 𝑘 opposite of the gradient, which can be calculated
for agent 𝑗 to be:
𝑛
𝜕|𝑉𝑖 |
Δ𝑑𝑗𝑘 = −2 ∑ Δ|𝑉𝑖 | ∙ (3-7)
𝜕𝑄𝑗′
𝑖=1
The derivative 𝜕|𝑉𝑖 |/𝜕𝑄𝑖′
can be approximated by the linear model presented by equation (4), if
one takes Δ|𝑉𝑖𝑗 | = 𝑉𝑖 − 𝑉𝑗 = 𝑉𝑛𝑜𝑚 − 𝑉𝑗 , where node 𝑖 represents the slack node of the grid that is
assumed to be always the nominal voltage.
At each time step 𝑘, each compensator 𝑗 updates its injected 𝑄𝑗′𝑘 as described in algorithm 1 below.
In operation 3 and 6, 𝑓 denotes the objective function of equation 4. In operation 3, 𝑓 𝑘−1 is
45/84
calculated with the measured Δ 𝑉𝑖𝑘−1, while in operation 6, the effect of 𝑄𝑗′𝑘 is incorporated into
the calculation of 𝑓 𝑘 by using the linear model of equation (3-7). In operation 5, the first step size
𝛼0 has to be chosen to be large enough for rapid convergence (as 𝑑𝑗𝑘 is small, values up to 𝛼0 =
108 can be used). The while loop in operations 7 to 10 implements the backtracking line search
with Armijo's condition and will ensure that the updates of 𝑄𝑗′𝑘 will reduce the objective function
by a sufficient amount. The values 𝜎 and 𝛾 are parameters that can be tuned to increase stability or
convergence speed. Since the line search algorithm is applied for every compensator and thus not
searching in the direction of the full gradient, it will overestimate its needed contribution.
Therefore, to ensure stability, in step 11, 𝑄𝑗′𝑘 is divided by the total number of compensators 𝑛𝑐,𝑡𝑜𝑡 .
Operations 12 to 16 finally make sure that the boundary conditions of the problem (4) are respected
and the compensators are not delivering more 𝑄′ than what they are capable of.
C. Communication layer
Finally, the communication layer is used for disseminating the locally measured voltages to all
compensators in the considered microgrid. As explained previously, this will be done in a peer-to-
peer fashion, thereby eliminating any single points of failure a more centralized method will
inherently contain. The dissemination algorithm should be dynamic, meaning that when new
measurements are available, the algorithm should disseminate the latest available voltage
measurements. Besides, it should be resistant to packet losses or delays. This work uses the gossip-
based push-sum algorithm, presented by Kempe et al. [89]. The algorithm is slightly adapted so
that it can be dynamically updated and used for dissemination of multiple values, as shown in
algorithm 2.
Each entry of the vector 𝑽𝒊 represents a sum, and 𝑾𝒊 the weight of this sum, at agent 𝑖. The estimate
of the voltages can then be calculated as 𝑽𝒊 /𝑾𝒊 , as in operation 7 of the algorithm. These values
46/84
are always transmitted in pairs, so if a packet gets lost, all the other pairs continue to have a correct
estimate of the voltages. In operation 1, 𝒆𝒊 denotes a unit vector with 1 on the 𝑖 th position and 0
𝑡
everywhere else. 𝑉𝑙𝑜𝑐𝑎𝑙,𝑖 is the locally measured voltage at agent 𝑖, at time step 𝑡. In operation 5,
the 𝑖 element of vector 𝑽𝒕𝒊 = 𝑉𝑖,𝑖𝑡 , is updated with the difference of 𝑉𝑙𝑜𝑐𝑎𝑙,𝑖 between the current and
th
previous time step, so the latest available measurements are always being disseminated.
3.4.2 Case Study 1
In this section, we will present a case study of the presented voltage control algorithm, simulated
on a benchmark low voltage microgrid of CIGRE [90], developed by the EU project Microgrids
and slightly adapted to suit our needs. The microgrid is shown in Figure 3.2. Realistic, fast changing
load profiles are used, that allow to assess the ability of the algorithm to follow such profiles. In
the case study we will only use reactive power, supplied by PV inverters, however the algorithm
can also be used to incorporate active power compensation.
Figure 3.2: Considered microgrid in the case study
47/84
A. Set up of the case study

The microgrid consists of five nodes, each representing an aggregated small neighbourhood of
around 40 households. All loads connected to the nodes are considered to be three phase balanced
loads and all PV generation three phase balanced injections, so that the grid is balanced at all times.
All PV generators 𝑗 are agents participating in the voltage control, leading to a total of 5
compensators. The PV generators are only allowed to inject or absorb reactive power, so the
computed transformed reactive power 𝑄𝑗′ from algorithm 1 is transformed back into reactive power
𝑄𝑗 . They are limited by their rated power 𝑆𝑟𝑎𝑡𝑒𝑑,𝑗 and the amount of active power 𝑃 they are
producing. The minimum and maximum reactive power can thus, at each time step 𝑡, be calculated
as:
𝑡 𝑡 2
𝑄𝑚𝑎𝑥,𝑗 = −𝑄𝑚𝑖𝑛,𝑗 2
= √𝑆𝑟𝑎𝑡𝑒𝑑,𝑗 − 𝑃𝑗𝑡 (3-8)
The active power generation profile for the PV generation is the same for all installations, as they
are assumed to be located close to each other. The profile itself is measured in 2009 at a fixed
rooftop PV installation at KU Leuven, with a time resolution of 5 minutes. Between every 5 minutes
time step, the profile is interpolated linearly to represent a continuously variable generation profile.
The load profiles of the residential loads are generated with a time resolution of 1 minute and are
based on the model of Richardson et al. [91], [92], and [93]. For each household connected to a
node, a separate load profile is generated. These are aggregated towards a single load profile per
node. The load profiles at nodes 2 and 5 are linearly interpolated to represent continuously varying
profiles, while the load profiles at nodes 1, 3 and 4 are implemented as discrete steps with a time
resolution of 1 minute, to represent the power steps that are found in distribution feeders. The
households are assumed to consume power with a constant power factor of 0.85.
The time resolution of the load and generation profiles should be as small as possible, in order to
correctly represent the fast varying loads in a distribution feeder. The time resolution of 1 minute
for the loads and 5 minutes for the PV generation and the combination of continuously varying
profiles with discrete power steps used here will show the ability of the algorithm to follow these
fast varying loads.
The gossiping algorithm 2 is executed with a time step of 100ms, meaning that each agent sends a
̂, 𝑾
{(𝑽 ̂)} pair to another random agent every 100ms. A communication latency of 100ms is
assumed, corresponding to the technical capabilities of a 3G wireless connection [94]. The
algorithm is executed asynchronously, so that no synchronisation is needed between the agents,
which would result in a demanding additional constraint.
The voltage control algorithm 1 is executed at each agent with a time step of one second. This is a
slower time scale than the gossiping algorithm, allowing the gossiping algorithm to converge to a
precision that is good enough before calculating the updated reactive power. The parameters 𝛼0 , 𝜎
and 𝛾 are determined empirically to be 𝛼0 = 108 , 𝜎 = 0.30 and 𝛾 = 0.8 , for fast but stable
48/84
convergence. The algorithm is executed asynchronously, eliminating all needs for synchronisation
between agents.
The voltage limits 𝑉𝑚𝑎𝑥 and 𝑉𝑚𝑖𝑛 are enforced on ± 2% of the nominal voltage, resulting in
𝑉𝑚𝑎𝑥 = 1.02 per unit (pu) and 𝑉𝑚𝑖𝑛 = 0.98pu. This is stricter than what is usual in distribution
networks, however, such narrow limits allow to show the performance of the algorithm, as
controlling the voltages within these strict limits is not straightforward when considering high loads
or high generation. The case study is executed for a summer day in July to be able to incorporate
the effect of high PV generation. The simulation is performed from 12h00 at noon to 22h00 in the
evening, to be sure to incorporate the PV generation peak during noon and the consumption peak
in the evening when there is little PV generation.
B. Results of the case study

The resulting load profiles at the five nodes are plotted in the upper graph of Figure 3.3 The negative
power, up to -80kW, results from the PV generation in the afternoon, and the positive power in the
evening shows the household consumption when PV generation is absent. One can clearly see the
fast changes in load and generation, for instance between 13h and 14h. The second graph of Figure
3.3 shows the voltage profiles at the five nodes, without any voltage control. One can see that the
profiles vary strongly, with peaks over 1.10pu in the afternoon and peak under 0.90pu in the
evening. The third graph of Figure 3.3 shows the reactive power compensation of the PV inverters,
resulting from the P2P voltage control algorithm. None of the inverters exceeds their minimum or
maximum reactive power as determined by equation 8.
The last graph of Figure 3.3 shows the resulting voltage profiles at the five nodes, which are
regulated to be between 𝑉𝑚𝑎𝑥 = 1.02𝑝𝑢 and 𝑉𝑚𝑖𝑛 = 0.98𝑝𝑢 by the P2P voltage control algorithm.
One can clearly see that the proposed algorithm succeeds in this. Some small peaks, resulting from
rapid load changes, fall beyond the limits, but are effectively regulated to be within limits. The
method succeeds in following the rapidly varying load without problems.
49/84
Figure 3.3: Results of the case study. Upper graph: resulting active load profiles with PV
generation. Second graph: voltage profiles per unit (pu), without any voltage control applied.
Third graph: reactive power compensation by the PV inverters following from the presented
voltage control algorithm. Lower graph: resulting voltage profiles per unit with the P2P voltage
control algorithm applied. One can clearly see that the proposed algorithm is able to keep
voltage between boundaries (±2%) at all times.
3.4.3 A P2P voltage control algorithm with minimum resources
This section will describe a more advanced P2P voltage control algorithm that regulates the voltage
within allowed limits, with minimum resources. The algorithms use a change of active and reactive
power consumption or injection of some participating DERs installed in the grid to regulate the
voltage. All controlled nodes should have some way to measure the voltage and are able to
communicate with other nodes.
50/84
The optimization problem can be written as follows:
minimize ∑ 𝑐𝑗𝑃 ∙ Δ𝑃𝑗2 + ∑ 𝑐𝑗𝑄 ∙ Δ𝑄𝑗2 (3-9a)

Δ𝑃,Δ𝑄 𝑗 𝑗
(3-9b)
𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 |𝑉𝑖 | ≤ 𝑉𝑚𝑎𝑥 , ∀𝑖 = 0, … , 𝑁
(3-9c)
𝑉𝑚𝑖𝑛 ≤ |𝑉𝑖 | , ∀𝑖 = 0, … , 𝑁
Δ𝑃𝑚𝑖𝑛,𝑗 ≤ Δ𝑃𝑗 ≤ Δ𝑃𝑚𝑎𝑥,𝑗 , ∀𝑗 = 0, … , 𝑛 (3-9d)
Δ𝑄𝑚𝑖𝑛,𝑗 ≤ ΔQj ≤ Δ𝑄𝑚𝑎𝑥,𝑗 , ∀𝑗 = 0, … , 𝑛 (3-9e)
On can see that the objective (3-9a) is to minimize a cost function of all changes in active and
reactive power needed for maintaining the voltage within limits. In this case, the cost functions
consists out of the sum of simple quadratic cost functions per node, however more complicated
cost function can be incorporated. The factors 𝑐𝑗𝑃 and 𝑐𝑗𝑄 can be calculated to incorporate losses on
the network (related to 𝑃2 and 𝑄 2 ) and other costs. The objective function is thus basically a sum
of separate cost functions, one for each participating agent.
The constraints (3-9d) and (3-9e) ensure that the voltage support DERs are always operating within
their limits. These constraints are local, meaning that they only influence the local decision variable
Δ𝑃𝑗 and Δ𝑄𝑗 , and therefore these constraints can be enforces locally.
The constraints (3-9b) and (3-9c) on the other hand are global constraints, as |𝑉𝑖 | =
𝑓(Δ𝑃1 , … , Δ𝑃𝑁 , Δ𝑄1 , … , Δ𝑄𝑁 ), meaning that it is not possible to distribute these easily amongst all
the participating agents. Besides, the function |𝑉𝑖 | = 𝑓(Δ𝑃, Δ𝑄), is nonlinear and quite complex,
as described by equations (3-2a) and (3-2b). However, when using a first order approximation, this
function becomes a separable sum of sensitivities of the voltage at the considered node towards a
change of active or reactive power:
𝑁 𝑁
𝜕|𝑉𝑖 | 𝜕|𝑉𝑖 |
|𝑉𝑖 | ≈ 𝑉𝑖,0 + ∑ ∙ Δ𝑃𝑗 + ∑ ∙ Δ𝑄𝑗 (3-10)
𝜕𝑃𝑗 𝜕Qj
𝑗=1 𝑗=1
As the partial derivatives are not easy to calculate and dependent on the system state, these are
often approximated by:
𝜕|𝑉𝑖 | 𝑅𝑖𝑗 𝜕|𝑉𝑖 | 𝑋𝑖𝑗
≅ 𝑎𝑛𝑑 ≅ (3-11)
𝜕𝑃𝑗 𝑉𝑛𝑜𝑚 𝜕𝑄𝑗 𝑉𝑛𝑜𝑚
which is a good approximation when the angle 𝛿𝑖𝑗 between the voltages at different nodes is small
[46]. The resistance 𝑅𝑖𝑗 and reactance 𝑋𝑖𝑗 between two points 𝑖 and 𝑗 can be calculated from the
topology of the grid and the knowledge of the impedances of the lines. In [95], a bus injection to
branch current BIBC and a bus current to branch voltage BCBV matrix are calculated from the line
51/84
incidence matrix from the grid, which allows for easy calculation of all approximated partial
derivatives as in equation (3-11) of the grid:
𝜕|𝑉𝑖 | 𝑟𝑒𝑎𝑙(𝐵𝐶𝐵𝑉 × 𝐵𝐼𝐵𝑉) 𝜕|𝑉𝑖 | 𝑖𝑚𝑎𝑔(𝐵𝐶𝐵𝑉 × 𝐵𝐼𝐵𝑉)

≅ 𝑎𝑛𝑑 ≅ (3-12)
𝜕𝑃𝑗 𝑈𝑛𝑜𝑚 𝜕𝑄𝑗 𝑈𝑛𝑜𝑚
The linearization of the function |𝑉𝑖 | = 𝑓(Δ𝑃1 , … , Δ𝑃𝑁 , Δ𝑄1 , … , Δ𝑄𝑁 ) into affine constraints,
makes it possible for the whole optimization problem (3-9) to be solved in a distributed way. A
straight forward way to do this, is by using a dual decomposition technique. This will use iterations
on the dual variables, trying to satisfy constraints (3-9b) and (3-9c), while each node will solve its
own sub problem, using the current value of the dual.
As the problem is solved in its dual form, let’s start with forming a separable partial Lagrangian:
𝐿(Δ𝑃, Δ𝑄, λmax , 𝜆𝑚𝑖𝑛 ) =

𝑁 𝑛 𝑛
(3-13)
∑(𝑐𝑗𝑃 Δ𝑃𝑗2 + 𝑐𝑗𝑄 ΔQ2𝑗 ) + ∑ 𝜆max
𝑖 (|𝑉𝑖 | − 𝑉max ) + ∑ 𝜆𝑚𝑖𝑛
𝑖 (𝑉𝑚𝑖𝑛 − |𝑉𝑖 |).
𝑗=1 𝑖=1 𝑖=1
Using (3-11), this can be written as:
𝑁 𝑛
𝑅𝑖𝑗 𝑋𝑖𝑗
𝐿= ∑ (𝑐𝑗𝑃 Δ𝑃 + 𝑐𝑗𝑄 ΔQ2𝑗 + ∑ 𝜆max
𝑖 (𝑉𝑖 + Δ𝑃𝑗 + Δ𝑄𝑗 − 𝑉max )
𝑉𝑛𝑜𝑚 𝑉𝑛𝑜𝑚
𝑗=1 𝑖=1
𝑛 (3-14)
𝑅𝑖𝑗 𝑋𝑖𝑗
+ ∑ 𝜆𝑚𝑖𝑛
𝑖 (𝑉𝑚𝑖𝑛 − 𝑉𝑖 − Δ𝑃𝑗 − Δ𝑄𝑗 )).
𝑉𝑛𝑜𝑚 𝑉𝑛𝑜𝑚
𝑖=1
When keeping 𝜆𝑚𝑖𝑛

𝑖 and 𝜆𝑚𝑎𝑥
𝑖 fixed, this represents a sum of separate objective functions for each
participating agent. The solution of the optimization problem for one such agent is quite
straightforward:
𝑛 𝑛
1 𝑅𝑖𝑗 𝑅𝑖𝑗
Δ𝑃𝑗∗ (𝜆𝑚𝑎𝑥 , 𝜆𝑚𝑖𝑛 ) = 𝑃 (∑ 𝜆max
𝑖 − ∑ 𝜆min
𝑖 ), (3-15a)
2𝑐𝑗 𝑉𝑛𝑜𝑚 𝑉𝑛𝑜𝑚
𝑖=1 𝑖=1
𝑛 𝑛
1 max
𝑋𝑖𝑗 𝑋𝑖𝑗
Δ𝑄𝑗∗ (𝜆𝑚𝑎𝑥 , 𝜆𝑚𝑖𝑛 ) = 𝑄 (∑ 𝜆𝑖 − ∑ 𝜆min
𝑖 ). (3-15b)
2𝑐𝑗 𝑖=1 𝑉𝑛𝑜𝑚 𝑉𝑛𝑜𝑚
𝑖=1
When Δ𝑃𝑗∗ < Δ𝑃𝑚𝑖𝑛,𝑗 or Δ𝑃𝑚𝑎𝑥,𝑗 < Δ𝑃𝑗∗ , or equivalent for Δ𝑄𝑗∗, the values Δ𝑃𝑗 or Δ𝑄𝑗 have to be
limited to their boundaries, since it is not possible for the DER to go beyond these limits.
52/84
The dual variables 𝜆𝑚𝑖𝑛

𝑖 and 𝜆𝑚𝑎𝑥
𝑖 should only be greater than zero, when the voltage at node 𝑖 goes
beyond the allowed limits (𝑉𝑚𝑖𝑛 , 𝑉𝑚𝑎𝑥 ). Because of the KKT conditions, the dual variables cannot
be smaller than zero. The variables should be calculated in such a way, that they maximize the dual
function. This can be done with a subgradient descent iterative algorithm:
𝜆𝑚𝑎𝑥,𝑘
𝑖 = max{𝜆𝑚𝑎𝑥,𝑘−1
𝑖 + 𝛼 𝑘 (|𝑉𝑖 | − 𝑉max ), 0} , (3-16a)
𝜆𝑚𝑖𝑛,𝑘
𝑖 = max{𝜆𝑚𝑖𝑛,𝑘−1
𝑖 + 𝛼 𝑘 (𝑉𝑚𝑖𝑛 − |𝑉𝑖 |), 0} , (3-16b)
where 𝛼 𝑘 is a parameter that has to be an appropriately size in order for quick but stable
convergence.
Since each pair of dual variables (𝜆𝑚𝑖𝑛

𝑖 , 𝜆𝑚𝑎𝑥
𝑖 ) is directly linked to the voltage difference on one
node of the grid, it seems reasonable that each pair of dual variables is calculated at the node they
belong to. This can be done using the equations (3-16) above and a measurement of the voltage.
As the agents 𝑗 participating in voltage control have to react with an appropriate (Δ𝑃𝑗 , Δ𝑄𝑗 ),
according to equations (2-15), the agents measuring the voltages and calculating (𝜆𝑚𝑖𝑛𝑖 , 𝜆𝑚𝑎𝑥
𝑖 ) have
to send it to the voltage control agents and wait an appropriate amount of time for their reaction,
before updating (𝜆𝑚𝑖𝑛
𝑖 ) with a new iteration. The voltage control algorithm thus consists out
of two types of agents: voltage measurement agents 𝑖, connected to all controlled nodes (could be
part of a smart meter), that measure the voltage at their node, calculates (𝜆𝑚𝑖𝑛 𝑖 , 𝜆𝑚𝑎𝑥
𝑖 ) and
communicates these to other agents. Secondly, the voltage controlling agents 𝑗, or compensators,
that participate actively in voltage control, they generate the appropriate amount of (Δ𝑃𝑗 , Δ𝑄𝑗 ).
These can be batteries, PV inverters, or other DER devices. The whole control method can be
summarized as follows:
1. Each node 𝑖 measures the local voltage and calculates (𝜆𝑚𝑖𝑛

𝑖 ) according to equations
(3-16)
2. The calculated (𝜆𝑚𝑖𝑛
𝑖 ) are broadcasted to the other agents in the grid via an
appropriate communication algorithms (could be a gossiping algorithm).
3. When receiving a new pair (𝜆𝑚𝑖𝑛 𝑖 , 𝜆𝑚𝑎𝑥
𝑖 ) , each compensator 𝑗 updates its (Δ𝑃𝑗 , Δ𝑄𝑗 ) ,
according to equations (3-15), and puts this on the grid.
4. When a sufficient amount of compensators have updated their (Δ𝑃𝑗 , Δ𝑄𝑗 ), thus after an
appropriate amount of time, repeat from point 1.
3.4.4 Case Study 2
A. Set up of the case study

This section presents a case study of the algorithm presented above simulated on a real distribution
grid with realistic profiles. The grid used in a real low voltage residential semi-urban feeder in the
region of Flanders, Belgium, shown in Figure 3.4 [96], [97]. To overload the grid with PV, every
second house has a PV inverter. All PV inverters participate in the voltage control, but only with
53/84
reactive power compensation. This means no active power will be curtailed, corresponding to
setting the cost 𝑐(Δ𝑃) = ∞.
Figure 3.4: Schematic diagram of the feeder used in the case study. Cable lengths are drawn to
scale [96]
As all the houses are located on the same feeder, they are geographically close to each other.
Therefore all PV will have more or less the same amount of generation. In the case study, all PV
generators will therefore have the same PV profile. The profile itself is the same as used in the case
study above, measured in 2009 at a fixed rooftop PV installation at KU Leuven, with a time
resolution of 5 minutes. Between every 5 minutes time step, the profile is interpolated linearly to
represent a continuously variable generation profile. The reactive power compensation of the PV
inverters is limited by their power rating and momentarily active power injection P, defined by
equation (3-8) as presented in the previous case study. The rated power of all PV inverters is
dimensioned on 5kVA.
The load profiles are also generated in the same way as in the case study presented previously, with
a time resolution of 1 minute and are based on the model of Richardson et al. [91], [92], and [93].
For each household connected to a node, a separate load profile is generated. All households are
assumed to consume power with a constant power factor of 0.85.
The voltages at all nodes are controlled, also the nodes without a PV inverter. This means that each
node must be able to measure the voltage locally, calculate the dual variables and communicate
with other nodes. The dissemination of the dual variables is implemented with a simple gossiping
algorithm, where a vector of (𝜆𝑚𝑎𝑥1 , … 𝜆𝑚𝑎𝑥
𝑛 , 𝜆1𝑚𝑖𝑛 , … 𝜆𝑚𝑖𝑛
𝑛 ) is sent around to a random neighbour
and update when new values of 𝜆𝑖 are received, or a new local value of 𝜆𝑖 is calculated. The limits
of the voltage 𝑉𝑚𝑖𝑛 and 𝑉𝑚𝑎𝑥 are set at 0.95pu and 1.05pu. This is tighter than the normal ±10%
encountered in real life, however these tighter limits allow for better evaluation of the performance
of the algorithm in this case study.
The time step of the gossiping algorithm is set at 100ms, meaning that every 100ms every agent
sent its latest estimate of the dual vector (𝜆1𝑚𝑎𝑥 , … 𝜆𝑚𝑎𝑥
𝑛 , 𝜆1𝑚𝑖𝑛 , … 𝜆𝑚𝑖𝑛
𝑛 ) to a random neighbour. A
communication latency of 100ms is assumed, corresponding to the technical capabilities of a 3G
wireless connection [94]. The gossiping algorithm is executed asynchronously, so that no
synchronisation is needed between the agents, which would result in a demanding additional
constraint.
54/84
The local updates of (𝜆𝑚𝑖𝑛

𝑖 ) are calculated with a time step of 1s. This is enough to ensure
that the gossiping algorithm has disseminated the latest dual values to all compensators, so that
they can update their (Δ𝑃𝑗 , Δ𝑄𝑗 ) before a new local update of the dual is calculated.
The case study is executed for a summer day in July to be able to incorporate the effect of high PV
generation. The simulation is performed from 12h00 at noon to 22h00 in the evening, to be sure to
incorporate the PV generation peak during noon and the consumption peak in the evening when
there is little PV generation
B. Results and discussion

The results of the case study are shown in Figure 3.5. One can clearly see that the algorithm is able
to keep the voltages reasonably within limits, by using the minimum amount of reactive power
needed. When the voltage is not limited, the reactive power compensation by the PV inverters is
most of the time zero, as the constraints (3-9b, and 3-9c) of problem (3-9) are not active. One can
see from the results that the algortihm is clearly fast enough to follow the quickly varying load and
PV generation profiles. One can also see that the reactive power limits are always satisfied (black
lines, third graph in Figure 3.5).
Only around 13h00, the voltage profiles show a significant peak above the set limits. This is due to
the limit on reactive power at that time. As one can see, all inverters are injecting the maximum
amount of reactive energy possible, but the voltage cannot be reduced further. To keep the voltage
within limits at that time, other means will be necesairy, like active power curtailment. However,
that was not in scope of the case study. One can conclude that this peak is not a flaw of the control
algorithm, but rather a lack of resources.
55/84
Figure 3.5: Results of the case study. Upper graph: sum of the active load profiles with PV
generation of all loads. Second graph: voltage profiles per unit (pu), without any voltage control
applied. Third graph: reactive power compensation by the PV inverters following from the
presented voltage control algorithm. The black lines denote the upper and lower limits. Lower
graph: resulting voltage profiles per unit with the P2P voltage control algorithm applied. One can
clearly see that the proposed algorithm is able to keep voltage between boundaries (±5%) at all
times, when enough reactive power is available.
3.5 Conclusion
This section presented two novel voltage control algorithms suited for the operation of peer-to-peer
smart grids. The algorithms are able to operate completely distributed, thereby keeping all control
local. The algorithms use a gossip-based push-sum protocol to dynamically disseminate all voltages
(first algorithm) at the controlled nodes of the grid and the lagrangian multipliers (second
algorithm) to all participating agents, in a robust way. The algorithms calculate at various time
steps an update of the required active and reactive power to maintain voltages within accepted
limits.
Both algorithms treat the problem as an optimization problem that controls reactive and active
power flows. The first algorithm is based on a gradient descent method, while the second algorithm
is based on dual decomposition.
56/84
The calculation of voltage sensitivity to active and reactive power change (gradient and partial
derivatives) is based on a linearization of the actual distribution grid, which leads to constant
sensitivities that only need to be calculated once, thereby limiting the computational requirements
of the participating agents. The algorithms are designed to work asynchronously. Two realistic case
studies were presented to assess the performance of the methods. It is shown that both algorithms
are able to follow quickly variable load profiles while being able to maintain voltages within limits.
Future work consists of the incorporation of a more realistic representation of the communication
network and the use of machine learning algorithms for model-free control.
57/84
4 DEMAND SIDE INTEGRATION
In P2P environment, low carbon resources, such as distribution generations, electric vehicles and
energy storages are widely deployed. Therefore, this section investigates the validation of two types
of demand side integration (DSI) programs: 1) Energy Management System and 2) Alternative
Price, under typical P2P scenario.
This section presents detailed work of active network control function through demand side
integration under P2P scenarios
Figure 4.1 Peer-2-Peer scenario: household connected with PV, EV, AC/DC system and battery storage
The layout of testing system is presented in Figure. 4.1. In testing scenario, 4 types of low carbon
resources are involved:
1) Electric Vehicles
In this report, the testing EVs are equipped with a 24 kWh capacity battery storage, with 3 kW
maximum charging and discharging rate. According to National Travel Survey in 2015 [99], the
EV arrives at the plug-in charging station from 7:00 a.m. to 11:00 a.m. The probability distribution
of arrival time is nearly following a Gaussian distribution between this time periods. Regarding the
departure time, the survey suggests 17:00 p.m.
58/84
Without charging strategy, the typical profile of EV can be estimated as following Figure. 4.2
Figure 4.2 Typical demand profile of EV
2) Battery Storage
In the testing household, an in-home battery is connected to enable EMS system. Battery
parameters used in the system are shown in Table 4.1. Used summer TOU tariff is shown in Figure.
4.2 [100].
Table 4.1 Battery parameters

Battery parameters Unit
Capacity 35 kWh
Charging power limit 4.3 Ah
Discharging power limit 4.3 Ah
Max SOC 0.9
Min SOC 0.3
Round-trip efficiency 0.9
3) Photovoltaic Panel
The photovoltaic are connected to the DC side of the households. Therefore, any energy export or
import will result in energy loss due to the converter efficiency. The PV are simulated with typical
PV profiles measured from smart meters from pilot project in Bristol, UK [101] in the following
Figure. 4.3.
59/84
Figure 4.3 Typical generation profile of PV
4.2 Type I DSI - Energy management system
A home area energy management system is designed to simulate the customer demand change over
different energy price incentive. It is assumed there are PV and battery at customers’ homes as
shown in Figure 4.4.
PV B a tte ry
D C B us
B i-d ire c tio n a l

c o n v e rte r
A C B us
A C lo a d
M a in g rid S m art h o m e
Figure 4.4 Overview structure of homes in P2P trading
The operation of energy management system is guided by the optimal control strategy. The
optimization model is illustrated as follows and solved by mixed integer linear programming
(MILP).
 Objective
The objective of battery operation is to minimize the cost of purchasing electricity from the main
grid.
96
𝑀𝑖𝑛 ∑[𝐶(𝑡)𝑃(𝑡) + 𝑆(𝑡)𝑃𝑠(𝑡)]𝑇 (4-1)

𝑡=1
60/84
where, C(t) is TOU rate at time t, P(t) is electrical power required from the main grid at time t, S(t)
is electricity selling price at time t, Ps(t) is electrical power output to the main grid at time t. T is
the length of time settlement, which is a constant. In this model T=0.25h. In this initial study, it is
assumed electricity selling price is zero.
𝑃(𝑡) = 𝑃𝑙𝑜𝑎𝑑 (𝑡) + 𝑃𝑐𝑜𝑛𝑣 (𝑡)

𝜂𝐴/𝐷 𝑃𝐷𝐶 (𝑡) 𝑖𝑓 𝑃𝐷𝐶 (𝑡) > 0
𝑃𝑐𝑜𝑛𝑣 (𝑡) = { 0 𝑖𝑓 𝑃𝐷𝐶 (𝑡) = 0
(4-2)
𝜂𝐷/𝐴 𝑃𝐷𝐶 (𝑡) 𝑖𝑓 𝑃𝐷𝐶 (𝑡) < 0
𝑃𝐷𝐶 (𝑡) = 𝑃𝐵 (𝑡) − 𝑃𝑃𝑉 (𝑡)
where 𝑃𝑙𝑜𝑎𝑑 (𝑡)is the customer load, 𝑃𝑐𝑜𝑛𝑣 (𝑡)is power from DC bus to AC bus, 𝑃𝐷𝐶 (𝑡)is the DC
power, 𝜂𝐴/𝐷 , 𝜂𝐷/𝐴 are converter efficiencies, 𝑃𝐵 (𝑡) is the battery demand, when battery discharges,
𝑃𝐵 (𝑡) < 0; when charges, 𝑃𝐵 (𝑡) > 0. 𝑃𝑃𝑉 (𝑡) is the PV output power.
 Constrains
Constrains come from the battery physical properties and battery power balance of the day:
𝑃𝐷𝑚𝑎𝑥 ≤ 𝑃𝐵 (𝑡) ≤ 𝑃𝐶𝑚𝑎𝑥 (4-3)

where PCmax and PDmax are the maximum charging and discharging rate.
𝐸𝑚𝑖𝑛 ≤ 𝐸𝐵 (𝑡) ≤ 𝐸𝑚𝑎𝑥 (4-4)

where 𝐸𝐵 (𝑡)is energy stored in the battery at time t, 𝐸𝑚𝑖𝑛 and 𝐸𝑚𝑎𝑥 are allowed maximum and
minimum energy that the battery can store.
96
∑ 𝑃𝐵 (𝑡) = 0 (4-5)
𝑡=1
The battery is operated at daily basis, therefore the sum of charging and discharging power of the
day is zero.
4.3 Type II DSI – pricing method: central and local energy market signal
Central and local market signals are input into energy management system to assess demand
uncertainty changes. The central and local energy price are shown in figures below. Central price
has six price periods and three different tariffs. The local energy price is the “sunshine tariff”[4]
with three price periods and two different tariffs.
61/84
Figure 4.5 Central energy price
Figure 4.6 Local energy price
4.4 Results
1) Result of EMS integration
In this section, two scenarios are analysed: 1) EMS system enabled by in-home battery; 2) EMS
system enabled by EV. The impact of EMS system is investigated in each section to demonstrate
the validity of technical-driven DSI method under P2P scenario. Then two scenarios are compared
to indicate the impact of EV as a substitute of battery storage.
Figure. 4.7 demonstrate the household load profile change with battery enabled EMS integrated.
As shown in the figure, the system shift evening peak demand into early morning between 0:00
a.m. and 6:00 a.m., where the electricity tariff is the lowest. On the other hand, the system prevents
the energy export from household to the grids, since the testing tariff does not consider P2P trading
62/84
price. In general, EMS with battery can lower the demand peak, shift the demand from high price
periods to low price periods, hence reduce the electricity bills for customers.
Figure 4.7 Household load profile change with battery enabled EMS integration
Figure. 4.8 illustrates the load change result considering EVs. The blue line refers to the original
load without EV involved, which in consistent with the original load in Figure. 4.8. When taking
EVs into consideration but do not employs charging strategies, the load slightly increased during
day time between 8:00 a.m. and 16:00 p.m. When applying the EMS system, the system shift the
morning peak demand to afternoon period between 14:30 p.m. to 17:00 p.m., where the tariff are
secondly lower. The reason that EMS does not shift demand to early morning is due to the
flexibility of EVs. The result indicates the unique characteristics of EVs, i.e., its flexibility are
featured by differing time and space.
Figure 4.8 Household load profile change with EV enabled EMS integration
The performance of above result are assessed by three benchmarks: 1) bill reduction to evaluate
the financial benefits, 2) peak reduction and 3) uncertainty reduction measured with STD.
63/84
Table 4.2 EMS performance assessment with battery/EV involved

Electricity Peak Reduction Uncertainty
Network Architecture
Bills (p) (kW) Reduction (kW)
EMS (battery enabled) 16.45 0.78 0.22
EMS (EV enabled) 18.01 5.37 1.20
2) Result of Pricing Method
The load profile changed by energy management system are shown in the figure below. The central
price shifts the demand during afternoon and evening to early morning by PV and battery. Local
price encourages the battery absorb the energy from PV during the afternoon to reduce evening
peak demand.
Figure 4.9 Load profile comparison
The demand uncertainty comparison is shown in the figure below. One day is divided into six time
periods according to different central energy price steps as shown in Figure 4.5.
The demand uncertainty varies over time of the day as shown in the red bars. The demand
uncertainty is lower during early morning and morning. The demand uncertainty is the highest
during the period 3, which is between 10:00-13:29. One of the reasons for high demand uncertainty
during period 3, 4 (13:30-15:59) and 5 (16:00-17:59) is PV output uncertainty.
The energy management system could reduce the demand uncertainty as shown in the blue and
green bars, since the energy management system regularly changes the demand at certain time of
64/84
the day. The energy management system with central price incentive changes the demand more
significantly during the day. As a consequence, energy management with central price incentive
generally reduces more uncertainty than that with local price incentive.
Demand uncertainty can be shifted as demand. As shown in the green bars, with central price
incentive, the energy management system shifts demand uncertainty in daytime to early morning.
The accurate load information is essential to P2P control. Development of control strategy will
largely depend on load information since it determines the ability of load rescheduling, i.e. it can
influence the proper time and amount of load shifting and reduction. However, in reality, there is
inherent uncertainty in domestic load profiles, i.e., it varies vigorously from different days and
consumers, and witnesses many demand spikes within a single day. The inherent demand
uncertainty can lead to considerable misestimate of domestic consumers’ flexibility to take part in
P2P trading. On the contrary, the P2P demand side control will change the demand uncertainty.
Therefore, this calls for researches on demand uncertainty for customers.
Therefore, the case study indicates that the P2P control strategy not only changes the load profile
but also the demand uncertainty.
Figure 4.10 Demand uncertainty comparison
4.5 Conclusions
This section focuses on Demand Side Integration regarding its drivers, motivations, and methods.
The concept of Demand Side Integration emerges when the conventional concept demand side
management is no longer appropriate for newly deregulated and competitive market in future power
system. The concept of DSI has been proposed by CIGRE working group C6.9 in 2006, which
includes more forms of demand interventions other than traditional demand management. These
new types of demand side interventions include more low carbon resources as well as demand
response programs with engaged customers.
65/84
For P2P paradigm, this section then investigates a typical future scenario of with P2P trading
market for technical level, to indicate the impact of DSI programs on network demand and
uncertainties. The investigated scenario are considering domestic households with DC/AC system
integrated together, and distributed PV generations connected. To perform optimal operation,
energy storage is employed to enable EMS system by their flexibilities, and triggered by tariffs.
In order to exploit the impact of DSI interventions both of technical aspects and commercial
aspects, differing energy storages and pricing methods are compared. Regarding the technical
intervention, EV and battery storage are investigated accordingly. The result demonstrates positive
effects by using two types of energy storage to enable the EMS system, however, with largely
different impact on demand. For differing pricing methods, the local price purely designed to
incentivise battery to absorb PV, can bring fairly demand and uncertainty reduction. However, it
not very capable of committing demand shift actions. The central price designed according to
wholesale price, are not only capable of reducing demand and digest uncertainty, but also shift peak
demand to valley periods.
66/84
5 COORDINATED CONTROL WITH MULTIPLE OBJECTIVES
A steady state analysis framework to quantify the operational benefits of a MV distribution network
with SOPs was presented in Section 2, which considers feeder load balancing, energy loss
minimization and voltage profile improvement. The framework also considered traditional network
reconfiguration and the combination of both SOP control and network reconfiguration to quantify
the benefits. It was found that in the case study using only one SOP can achieve improvement in
network operation compared to the case of using network reconfiguration with all branches
equipped with remotely controlled switches. The combination of both SOP control and network
reconfiguration can achieve the optimal network operation.
This section presents detailed work of active network control function through coordinated control
of both network reconfiguration and SOP with multiple objectives.

5.1.1 Multi-objective functions
The multi-objectives of both DNR and SOP output optimizations were similar including
minimizing the power loss, the load balancing and maximizing the DG penetration level. The
objectives are presented as Ploss, LBI and DGPL as shown below:
A. Minimizing the power loss (Ploss)

𝑛𝑏𝑟𝑎𝑛𝑐ℎ
𝑜𝑏𝑗1 = 𝑚𝑖𝑛𝑃𝑙𝑜𝑠𝑠 = ∑ 𝐼𝑖2 × 𝑟𝑖 (5-1)

𝑖=1
𝐼𝑖 is the current passing through branch 𝑖 and 𝑟𝑖 is the resistance of the branch. 𝑛𝑏𝑟𝑎𝑛𝑐ℎ is the total
number of branches in a distribution network.
B. Load balancing
Load balancing is achieved by minimizing the load balance index (LBI), which is defined as:
𝑛𝑏𝑟𝑎𝑛𝑐ℎ
𝐼𝑖−𝑎𝑐𝑡𝑢𝑎𝑙 2
𝑜𝑏𝑗2 = 𝑚𝑖𝑛𝐿𝐵𝐼 = ∑ ( ) (5-2)
𝐼𝑖−𝑟𝑎𝑡𝑒𝑑
𝑖=1
where 𝐼𝑖−𝑎𝑐𝑡𝑢𝑎𝑙 and 𝐼𝑖−𝑟𝑎𝑡𝑒𝑑 are the real and rated current of branch 𝑖.
C. Maximizing the DG penetration level (DGPL)

67/84
∑ 𝑃𝐷𝐺
𝑜𝑏𝑗3 = 𝑚𝑎𝑥𝐷𝐺𝑃𝐿 = × 100% (5-3)
∑ 𝑃𝑙𝑜𝑎𝑑
∑ 𝑃𝐷𝐺 is the sum of active power injections from all connected DGs, and ∑ 𝑃𝑙𝑜𝑎𝑑 is the total active
power demand of the system. In this study, the upper boundary of DGPL was set to be 200%.
5.1.2 Constraints
The backward-forward sweep method was adopted to evaluate the network performances among
the candidate solutions. The following network constraints were considered including power flow
conservations; thermal limits of transformers and lines; voltage limits; radial configurations and
serving all loads.
5.2 Solution methodology

5.2.1 DNR optimization using the ant colony optimization (ACO) method
When solving the multi-objective DNR optimization problem, the search space of different network
configurations is explored by means of ACO, which is derived from the inspiration of foraging
behavior of natural ant colonies [103]. It is a metaheuristic solution based on the mutual interactions
among artificial agents named as ants. The discrete optimization problem is formulated as a graph,
and each ant generates a solution by leaving a pheromone trail on the path from nest to food. The
pheromone evaporates with time, and will be reconstructed more rapidly on a shorter path. The
amount of pheromone deposited on each path is directly proportional to the quality of the solution.
The transition of ants between nodes is determined by a probabilistic selection rule based on the
value of pheromone deposition which will cause more ants to choose the shorter path. Over a period
of time, the path corresponding to the optimal solution is the one that presents the highest
pheromone deposition.
5.2.2 SOP output optimization using the Taxi-Cab method
An SOP can be implemented with Back-to-Back Voltage Source Converters (B2B VSC)
connecting two adjacent feeders of the network as shown in Figure. 5.1. The operational constraints
of an SOP are:
𝑃𝐶1 + 𝑃𝐶2 = 0 (5-4)

2
√𝑃𝐶1 2 (5-5)
+ 𝑄𝐶1 ≤ 𝑆𝐶1
2
√𝑃𝐶2 2 (5-6)
+ 𝑄𝐶2 ≤ 𝑆𝐶2
68/84
where 𝑃𝐶1 and 𝑄𝐶1 , 𝑃𝐶2 and 𝑄𝐶2 are real and reactive power injections/absorptions of each VSC
to/from the connecting point. 𝑆𝐶1 and 𝑆𝐶2 are the rated capacity of each VSC. These device
operational boundaries are further considered in addition to the constraints illustrated in Section
5.2.2.
When optimizing the SOP output, a general optimization method named as Taxi-cab [104] is used.
The outputs of an SOP: [𝑃𝐶1 , 𝑄𝐶1 , 𝑄𝐶2 ] (𝑃𝐶2 is not included as it equals to 𝑃𝐶1 ) are specified as the
decision variables. Objectives Ploss and LBI (𝑜𝑏𝑗𝑛 (𝑛=1,2) in Section 2.1) are to be minimized, while
DGPL is incorporated with input variables and is considered as negative loads. In the optimization
procedure, decision variables [𝑃𝐶1 , 𝑄𝐶1 , 𝑄𝐶2 ] are presented as unit vectors in the searching space.
Along each unit vector, 𝑜𝑏𝑗𝑛 (𝑛=1,2) is a function of one decision variable. The minimization of
𝑜𝑏𝑗𝑛 along each vector is accomplished by applying the golden ratio search algorithm. The
procedure is repeated along each vector consecutively to generate a sequence of minimal values of
𝑜𝑏𝑗𝑛 until no further decrease can be achieved. This method is simple in implementation and does
not require calculating the derivatives of the objective functions.
5.2.3 Visualization of multi-objective optimization solutions using the Pareto optimality
In multi-objective optimization problems, since there rarely exists a single solution that can
simultaneously optimize all objectives, the Pareto optimality is used. It compares candidate
solutions which satisfy the imposed constraints in the concept of dominance, and provides a set of
trade-off solutions amongst different objectives. ‘A’ is said to be dominated over ‘B’, if and only
if Equation.5-7 and Equation.5-8 are satisfied concurrently.
∀𝑖 ∈ [1, 2, … , 𝑛𝑜𝑏𝑗 ]: 𝑓𝑖 (𝐴) ≤ 𝑓𝑖 (𝐵) (5-7)

∃𝑖 ∈ [1,2, … , 𝑛𝑜𝑏𝑗 ]: 𝑓𝑖 (𝐴) < 𝑓𝑖 (𝐵) (5-8)
The set of obtained non-dominated solutions is referred to as the Pareto set, and the image of the
Pareto set presented in the solution space is referred to as the Pareto frontier.
5.3 Test system and results

5.3.1 Test system
A program has been written in MATLAB software based on the proposed methodologies. The
widely used 33-bus distribution system [105] was adopted as the test system. It is rated at 12.66 kV
with a total demand of 3.7 MW and 2.3 MVar. The initial open switches are s33-s34-s35-s36-s37.
To analyze the impact of various DG penetration levels, the system was modified to consider four
sites for DG as shown in Figure. 5.2. Power injections from DGs were calculated according to
Equation 5-3, and all DGs were modeled as negative PQ loads with a power factor equal to one.
69/84
Figure 5.1 Single line diagram of a distribution network with a SOP connected
Figure 5.2 Single line diagram of the modified 33-bus distribution system
5.3.2 Multi-objective DNR results
Figure. 5.3 presents the obtained Pareto frontier in three dimensions. Table 5.1 presents the
solutions with optimal values along each objective and the sets of open switches in those solutions.
Results show that the three objectives cannot be optimized simultaneously. Each of them is
obtained with different network configurations and certain DG penetration levels, which verifies
the significance of using the Pareto optimality in solving multi-objective problems. Results also
show that DNR is able to improve DGPL without violating any network operational constraints. It
can be illustrated by the results that, with the increasing DGPL, Ploss and LBI will decrease first.
However, when DGPL increases to a high value, Ploss and LBI will start to increase. It reveals the
fact that DG has the capability to reduce power losses and balance the load within a certain
penetration level. However, large DG penetrations will change the direction of power flow and
increase the burden of power transfer of the network.
Among solutions in the Pareto set, the network configuration with s7-s9-s14-s28-s32 open is
selected as an optimal solution. This is because the topology is taken by the majority of solutions
in the Pareto set. The minimum P Ploss and the LBI obtained by this topology are 65.228 kW and
0.03743 which are close to the optimal values in Table 5.1, namely 64.676 kW and 0.0286. In
addition, the optimal value along obj3, i.e. DGPL equals to 200%, can be obtained. If DNOs are
interested in optimizing one of the three objectives, this topology can be changed to the preferred
one through simple switch operations. For instance, by closing s28 and open s37, a network with
70/84
minimum Ploss is obtained. Consequently, this topology is adopted for SOP output optimization in
the next section.
5.3.3 Multi-objective SOP output optimization results
After obtaining the optimal network configuration in Section 5.2, SOP is installed into the network
replacing each of the open switches (s7-s9-s14-s28-s32) at a time. The optimal outputs of SOP are
calculated using the method in Section 5.2. Figure. 5.4 presents the Pareto frontier of SOP output
optimization. The optimal solutions along each objective are presented in Table 5.2, with the
corresponding SOP locations and outputs. Results show that SOP is able to reduce Ploss and LBI
by further 58.21% and 52.45%. When the DG penetration is low, SOP transfers real power to
support the feeder with large power demand and injects reactive power to the network as
compensators. When the DG penetration is high, SOP transfers real power from the feeder with
large DG injections to the feeder with large power demand, and absorbs reactive power from the
network to mitigate the voltage rise issue.
Branch28 is the optimal site for SOP installation. As shown in Table 5.2, the optimal solutions
along each objective can be obtained with SOP installed in Branch28. The solutions obtained when
SOP operates in Branch28 take the majority in the Pareto set.
Figure 5.3 The obtained Pareto Frontier with DNR optimization
71/84
Figure 5.4 The obtained Pareto Frontier with the SOP output optimization
Table 5.1 Optimal solutions along each objective with the DNR optimization
72/84
Table 5.2 Optimal solutions along each objective with the SOP optimization
5.4 Conclusion
In this work, the reconfiguration problem and the optimal SOP operation problem are formulated
within a multi-objective framework using the Pareto optimality. The obtained results show the
effectiveness of using DNR and SOP to improve the distribution network operation, focusing on
power loss reduction, load balance and DG penetration level increase. The obtained Pareto frontiers
present great diversity, high quality and proper distribution of the non-dominated solutions among
all feasible solutions, which allows the Distribution Network Operators to choose based on their
priorities and necessities.
73/84
6 CONCLUSIONS
supporting the transition to a sustainable low carbon economy. Distribution networks are very large
and topologically complicated systems which connect high voltage transmission systems to end
users. In today’s distribution networks, real-time monitoring and control below the primary
substations are very limited due to the lack of sensors, communication systems and control
equipment.
arrangements will be employed. Consequently the behaviour of distribution networks will become
and time-lines.
obtained from a particular network. There is an urgent need to develop alternative methodologies
and tools for scalable distribution network operation and planning, which can facilitate P2P energy
trading with varying degrees of input information gap and uncertainties caused by decentralized
P2P control philosophies.
Alternative P2P based control philosophy of distribution networks was developed, and technical
requirements on the associated ICT infrastructure was analysed based on our extensive previous
and current work on decentralized control of distribution networks. Novel probabilistic and
predictive control functions were devised to enable and facilitate the P2P based energy trading and
better network operation under extremely dynamic and uncertain conditions.
These control functions mainly involved in employing novel active network control, such as power
electronic device soft open point, coordinated network control among different active control
devices, demand side integration, and coordinated control considering multiple objectives. The
mainly findings include but not limit to the flowing aspects:
(1) Using a novel active network control device, i.e. soft open point was proved to be able to
significantly increase the network’s distributed generation (DG) hosting capacity. The
control scheme using the objective for voltage profile improvement increased the headroom
of the voltage limits by the largest margin. This control scheme dispatched increased
reactive power, and, hence, was at the expense of increased energy losses. The control
74/84
schemes using the objectives to achieve line utilization balance and energy losses
minimization showed the most improvement in circuit utilization and in limiting energy
losses, mainly relying on the real power exchange between feeders. According to network
characteristics, network operators are able to devise the control scheme using one or multi-
objective functions. The proposed network control using soft open point provides a
potential framework/solution for electricity network operators, allowing them to choose
appropriate control schemes more effectively. This selection requires the network operators
to attribute value to the increase in hosting capacity, mitigation in voltage issues, the
reduction in the maximum line utilization and the reduction in energy losses.
(2) The novel voltage coordinated control algorithms were proved to be able to operate
completely distributed, thereby keeping all control local. The algorithms use a gossip-based
push-sum protocol to dynamically disseminate all voltages (one algorithm) at the controlled
nodes of the grid and the lagrangian multipliers (another algorithm) to all participating
agents, in a robust way. The algorithms calculate at various time steps and update of the
required active and reactive power to maintain voltages within accepted limits. Both
algorithms treat the problem as an optimization problem that controls reactive and active
power flows. The first algorithm is based on a gradient descent method, while the second
algorithm is based on dual decomposition. The calculation of voltage sensitivity to active
and reactive power change (gradient and partial derivatives) is based on a linearization of
the actual distribution grid, which leads to constant sensitivities that only need to be
calculated once, thereby limiting the computational requirements of the participating
agents. The algorithms are designed to work asynchronously. Two realistic case studies
were presented to assess the performance of the methods. It is shown that both algorithms
are able to follow quickly variable load profiles while being able to maintain voltages within
limits.
(3) The concept of Demand Side Integration emerges when the conventional concept demand
side management is no longer appropriate for newly deregulated and competitive market in
future power system. These new types of demand side interventions include more low
carbon resources as well as demand response programs with engaged customers. Through
investigating demand side integration considering domestic households with DC/AC
system integrated together, and distributed PV generations, and performing optimal
operation, energy storage enabling energy management system (EMS) system by their
flexibilities, and triggered by tariffs, this work explored the impact of demand side
integration (DSI) interventions both of technical aspects and commercial aspects, differing
energy storages and pricing methods are compared. Regarding the technical intervention,
electric vehicles (EV) and battery storage are investigated accordingly. The result
demonstrates positive effects by using two types of energy storage to enable the EMS
system, however, with largely different impact on demand. For differing pricing methods,
the local price purely designed to incentivise battery to absorb photovoltaic (PV), can bring
fairly demand and uncertainty reduction. However, it not very capable of committing
demand shift actions. The central price designed according to wholesale price, are not only
capable of reducing demand and digest uncertainty, but also shift peak demand to valley
periods.
(4) Throuth formulating the network reconfiguration and the optimal soft open point operation
problem within a multi-objective framework using the Pareto optimality, the work
demonstrated the effectiveness of using distribution ntwork reconfiguration and soft open
75/84
point to improve the distribution network operation, focusing on power loss reduction, load
balance and DG penetration level increase. The obtained Pareto frontiers present great
diversity, high quality and proper distribution of the non-dominated solutions among all
feasible solutions, which allows the Distribution Network Operators to choose based on
their priorities and necessities.
Test results showed that these alternative methodologies and tools for scalable distribution network
operation and planning enable P2P energy trading with varying degrees of input information gap
and uncertainties caused by decentralized P2P control philosophies.
76/84
7 REFERENCES
[1] M. Vandenbergh, D. Craciun, V. Helbrecht, R. Hermes, R. Lama, P. M. Sonvilla et al.,

“Prioritisation of technical solutions available for the integration of PV into the distribution
grid,” DERlab, Eur. Distrib. Energy Resources Lab. e. V., Kassel, Germany, Tech. Rep. D,
vol. 3, 2013.
[2] 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.
[3] P. C. Ramaswamy, P. Vingerhoets, and G. Deconinck, “Reconfiguring distribution grids for
more integration of distributed generation,” in Electricity Distribution (CIRED 2013), 22nd
International Conference and Exhibition on. IET, 2013, pp. 1–4.
[4] N. Efkarpidis, C. Gonzalez de Miguel, T. Wijnhoven, D. Van Dommelen, T. De Rybel, and
J. Driesen, “Technical assessment of on-load tap-changers in flemish LV distribution grids,”
2013.
[5] F. A. Viawan, A. Sannino, and J. Daalder, “Voltage control with on-load tap changers in
medium voltage feeders in presence of distributed generation,” Electric power systems research,
vol. 77, no. 10, pp. 1314–1322, 2007.
[6] J. H. Harlow, Electric power transformer engineering. CRC press, 2004, ch. 2.
[7] J. Dixon, L. Moran, J. Rodriguez, and R. Domke, “Reactive power compensation technologies:
State-of-the-art review,” Proceedings of the IEEE, vol. 93, no. 12, pp. 2144–2164, 2005.
[8] P. Rao, M. Crow, and Z. Yang, “Statcom control for power system voltage control applications,”
IEEE Transactions on Power delivery, vol. 15, no. 4, pp. 1311–1317, 2000.
[9] B. Singh, R. Saha, A. Chandra, and K. Al-Haddad, “Static synchronous compensators
(STATCOM): a review,” IET Power Electronics, vol. 2, no. 4, pp. 297–324, 2009.
[10] 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.
[11] R. Tonkoski, L. A. Lopes, and T. H. El-Fouly, “Coordinated active power curtailment of grid
connected PV inverters for overvoltage prevention,” IEEE Transactions on Sustainable Energy,
vol. 2, no. 2, pp. 139–147, 2011.
[12] F. Geth, D. Six, P. Tant, and J. Driesen, “Grid supporting battery energy storage systems in the
low voltage distribution grid,” 2011.
[13] H. Ibrahim, A. Ilinca, and J. Perron, “Energy storage systems—characteristics and
comparisons,” Renewable and sustainable energy reviews, vol. 12, no. 5, pp. 1221–1250, 2008.
[14] S. N. Salih and P. Chen, “On coordinated control of OLTC and reactive power compensation
for voltage regulation in distribution systems with wind power,” 2015.
[15] S. Civanlar, J. J. Grainger, H. Yin, and S. S. H. Lee, "Distribution feeder reconfiguration for
loss reduction," Power Delivery, IEEE Transactions on, vol. 3, pp. 1217-1223, 1988.
[16] J. Ekanayake, N. Jenkins, K. Liyanage, J. Wu, and A. Yokoyama, Smart Grid: Technology and
Applications, Wiley Press, 2012.
77/84
[17] I. Roytelman and V. Ganesan, "Coordinated local and centralized control in distribution
management systems," Power Delivery, IEEE Transactions on, vol. 15, pp. 718-724, 2000.
[18] R. S. Rao, K. Ravindra, K. Satish, and S. V. L. Narasimham, "Power Loss Minimization in
Distribution System Using Network Reconfiguration in the Presence of Distributed
Generation," Power Systems, IEEE Transactions on, vol. 28, pp. 317-325, 2013.
[19] A. Merlin and H. Back, "Search for a minimal-loss operating spanning tree configuration in an
urban power distribution system," presented at the Proc. 5th Power System Computation
Conforence, Cambridge, 1975.
[20] K. Nara, A. Shiose, M. Kitagawa, and T. Ishihara, "Implementation of genetic algorithm for
distribution systems loss minimum re-configuration," Power Systems, IEEE Transactions on,
vol. 7, pp. 1044-1051, 1992.
[21] D. Das, "A fuzzy multi-objective approach for network reconfiguration of distribution
systems," Power Delivery, IEEE Transactions on, vol. 21, pp. 202-209, 2006.
[22] J. Young-Jae, K. Jae-Chul, O. K. Jin, S. Joong-Rin, and K. Y. Lee, "An efficient simulated
annealing algorithm for network reconfiguration in large-scale distribution systems," Power
Delivery, IEEE Transactions on, vol. 17, pp. 1070-1078, 2002.
[23] Y. Yu and J. Wu, "Loads combination method based core schema genetic shortest-path
algorithm for distribution network reconfiguration," in Proceedings of Power System
Technology Conference, 2002, pp. 1729-1733, vol 3.
[24] A. Ahuja, S. Das, and A. Pahwa, "An AIS-ACO Hybrid Approach for Multi-Objective
Distribution System Reconfiguration," Power Systems, IEEE Transactions on, vol. 22, pp.
1101-1111, 2007.
[25] C. Chao-Shun, T. Cheng-Ta, L. Chia-Hung, H. Wei-Lin, and K. Te-Tien, "Loading Balance of
Distribution Feeders With Loop Power Controllers Considering Photovoltaic Generation,"
Power Systems, IEEE Transactions on, vol. 26, pp. 1762-1768, 2011.
[26] W. Cao, J. Wu, and N. Jenkins. "Feeder Load Balancing in MV Distribution Networks Using
Soft Normally-Open Points," in IEEE Innovative Smart Grid Technologies Conference Europe,
2014, PP 1-6.
[27] S. K. Bhattacharya and S. K. Goswami, "Distribution Network Reconfiguration Considering
Protection Coordination Constraints," Electric Power Components and Systems, vol. 36, pp.
1150-1165, 2008.
[28] J. Flottemesch and M. Rother, "Optimized energy exchange in primary distribution networks
with DC links," in Proceedings of Electric Utility DRPT 2004., Hong Kong, China, pp. 108-
116, vol. 1, Apr. 2004.
[29] C.Y. Tang, Y.F. Chen, Y.M. Chen and Y.R. Chang, "DC-link voltage control strategy for three-
phase back-to-back active power conditioners," IEEE Trans Industrial Electronics, vol. 62, no.
10, pp. 6306-6316, Oct. 2015.
[30] E. Romero-Ramos, A. Gómez-Expósito, A. Marano-Marcolini, J.M. Maza-Ortega and J.L.
Martinez-Ramos, "Assessing the loadability of active distribution networks in the presence of
DC controllable links," IET Generation, Transmission & Distribution, vol. 5, no. 11, pp. 1105-
1113, Nov. 2011.
[31] J.M. Maza-Ortega, A. Gómez-Expósito, M. Barragán-Villarejo, E. Romero-Ramos and A.
Marano-Marcolini, "Voltage source converter-based topologies to further integrate renewable
78/84
energy sources in distribution systems," IET Renewable Power Generation, vol. 6, no. 6, pp.
435-445, Nov. 2012.
[32] W. Cao, J. Wu and N. Jenkins, "Feeder load balancing in MV distribution networks using soft
normally-open points," in 2014 IEEE PES ISGT-Europe, Istanbul, Turkey, pp.1-6, Oct. 2014.
[33] J.M. Bloemink and T.C. Green, "Benefits of Distribution-Level Power Electronics for
Supporting Distributed Generation Growth," IEEE Trans Power Del., vol. 28, no. 2, pp. 911-
919, Apr. 2013.
[34] W. Cao, J. Wu, N. Jenkins, C. Wang and T. Green, "Benefits analysis of Soft Open Points for
electrical distribution network operation," Applied Energy, vol. 165, pp. 36-47, Mar. 2016.
[35] Western Power Distribution (2016, Sep 20) Low Carbon Networks Fund submission from
Western Power Distribution – Network Equilibrium [Online] Available:
https://www.ofgem.gov.uk/publications-and-updates/low-carbon-networks-fund-submission-
western-power-distribution-%E2%80%93-network-equilibrium
[36] SP Energy Networks (2016, Sep 20), Electricity NIC submission: SP Energy Networks –
ANGLE-DC, [Online] Available: https://www.ofgem.gov.uk/publications-and-
updates/electricity-nic-submission-sp-energy-networks-angle-dc
[37] UK Power Networks (2016, Sep 20). Flexible Urban Networks – Low Voltage (FUN LV)
[Online] Available: http://innovation.ukpowernetworks.co.uk/innovation/en/Projects/tier-2-
projects/Flexible-Urban-Networks-Low-Voltage/
[38] M. Kalantar, S.M. Mousavi. "Dynamic behavior of a stand-alone hybrid power generation
system of wind turbine, microturbine, solar array and battery storage," Applied Energy, vol. 87,
pp.3051–64, 2010.
[39] C.L. Trujillo, D. Velasco, J.G. Guarnizo, N. Díaz. "Design and implementation of a VSC for
interconnection with power grids, using the method of identification the system through state
space for the calculation of controllers," Applied Energy, vol 88, pp. 3169–75, 2011.
[40] D. Georgiev, et al. "Computing Intervals of Secure Power Injection", IFAC Proceedings, vol.
47, no. 3, pp: 2253-2259, 2014.
[41] P. Vor, et al. "Modular algorithms for computing Intervals of Secure Power Injection", 2016
IEEE International Energy Conference (ENERGYCON), 2016.
[42] H. Almasalma, J. Engels, and G. Deconinck, “Peer-to-peer control of microgrids,” in IEEE
Benelux PELS/PES/IAS Young Researchers Symposium. KU Leuven, 2016, pp. 1–6.
[43] R. H. Lasseter, “Smart distribution: Coupled microgrids,” Proceedings of the IEEE, vol. 99,
no. 6, pp. 1074–1082, 2011.
[44] Q. Lu, W. Hu, Y. Min, W. Ge, and Z. Wang, “Wide-area coordinated control of large scale
energy storage system,” in Power System Technology (POWERCON), 2012 IEEE International
Conference on. IEEE, 2012, pp. 1–5.
[45] A. Kulmala, S. Repo, and P. Järventausta, “Coordinated voltage control in distribution
networks including several distributed energy resources,” IEEE Transactions on Smart Grid,
vol. 5, no. 4, pp. 2010–2020, 2014.
[46] B. B. Zad, J. Lobry, and F. Vallyè, “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. IEEE, 2016, pp. 1–6.
79/84
[47] S. Sojoudi, J. Lavaei, and A. G. Aghdam, Structurally Constrained Controllers: Analysis and
Synthesis. Springer Science & Business Media, 2011.
[48] X. Han, A. M. Kosek, O. Gehrke, H. W. Bindner, and D. Kullmann, “Activate distributed
energy resources’ services: Hierarchical voltage controller as an application,” in 2014 IEEE
PES T&D Conference and Exposition. IEEE, 2014, pp. 1–5.
[49] M. Yazdanian and A. Mehrizi-Sani, “Distributed control techniques in microgrids,” IEEE
Transactions on Smart Grid, vol. 5, no. 6, pp. 2901–2909, 2014.
[50] 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.
[51] 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.
[52] R. Schollmeier, “A definition of peer-to-peer networking for the classification of peer-to-peer
architectures and applications,” in Peer-to-Peer Computing, 2001. Proceedings. First
International Conference on. IEEE, 2001, pp. 101–102.
[53] 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 1: Concepts,
approaches, and technical challenges,” IEEE Transactions on Power systems, vol. 22, no. 4, pp.
1743–1752, 2007.
[54] D. Kempe, A. Dobra, and J. Gehrke, “Gossip-based computation of aggregate information,” in
Foundations of Computer Science, 2003. Proceedings. 44th Annual IEEE Symposium on. IEEE,
2003, pp. 482–491.
[55] 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.
[56] G. Rohbogner, S. Fey, P. Benoit, C. Wittwer, and A. Christ, “Design of a multiagent-based
voltage control system in peer-to-peer networks for smart grids,” Energy Technology, vol. 2,
no. 1, pp. 107–120, 2014.
[57] Baitch A, Chuang A, Mauri G, et al. “International perspectives on demand-side integration’’,
Proceedings of the 19th International Conference on Electricity Distribution (CIRED), Vienna.
2007.
[58] Chuang A S. Demand-side integration for system reliability, Power Tech, 2007 IEEE Lausanne.
IEEE, 2007: 1617-1622.
[59] Chuang A S, Gellings C W. Demand-side integration in a restructured electric power industry
[J]. CIGRE, number Paper C6-105, Session, 2008.
[60] Advancing the Efficiency of Electricity Utilization “Prices to DevicesTM”: Background Paper,
2006 EPRI, Summer Seminar, EPRI, Palo Alto, C.A. Available at http://www.epri.com.
[61] Hungerford Z, Bruce A, MacGill I. “Review of demand side management modelling for
application to renewables integration in Australian power markets”, Power and Energy
Engineering Conference (APPEEC), 2015 IEEE PES Asia-Pacific. IEEE, 2015: 1-5.
[62] Energy Efficiency: Challenges and Trends for Electric Utilities
[63] D. Shirmohammadi and H. Hong, "Reconfiguration of electric distribution of networks for
80/84
resistive line losses reduction," IEEE Trans. Power Deliv., vol. 4, pp. 1492-1498, 1989.
[64] S. Civanlar, J. J. Grainger, H. Yin, and S. S. H. Lee, "Distribution feeder reconfiguration for
loss reduction," Power Delivery, IEEE Transactions on, vol. 3, pp. 1217-1223, 1988.
[65] S. K. Goswami and S. K. Basu, "A new algorithm for the reconfiguration of distribution feeders
for loss minimization," Power Delivery, IEEE Transactions on, vol. 7, pp. 1484-1491, 1992.
[66] B. Amanulla, S. Chakrabarti, and S. Singh, "Reconfiguration of Power Distribution Systems
Considering Reliability and Power Loss," IEEE Trans. Power Del., vol. 27, pp. 918-926, 2012.
[67] R. Rao, S. Narasimham, M. Raju, and A. Rao, "Optimal Network Reconfiguration of Large-
Scale Distribution System Using Harmony Search Algorithm," IEEE Trans. Power Syst., vol.
26, pp. 1080-1088, 2011.
[68] I. Roytelman, V. Melnik, S. S. H. Lee, and R. L. Lugtu, "Multi-objective feeder reconfiguration
by distribution management system," IEEE Transactions on Power Systems, vol. 11, pp. 661-
667, 1996.
[69] Y. T. Hsiao and C. Y. Chien, "Multiobjective optimal feeder reconfiguration," IEE Proceedings
- Generation, Transmission and Distribution, vol. 148, pp. 333-336, 2001.
[70] J. M. Bloemink and T. C. Green, "Increasing distributed generation penetration using soft
normally-open points," ed., 2010, pp. 1-8.
[71] P.S. Jones and C.C. Davidson, "Calculation of power losses for MMC-based VSC HVDC
stations," in 2013 15th European Conference on Power Electronics and Applications (EPE),
Lille, 2013, pp. 1-10.
[72] W. Cao, J. Wu, N. Jenkins, C. Wang and T. Green, "Operating principle of Soft Open Points
for electrical distribution network operation," Applied Energy, vol. 164, pp. 245-257, Feb. 2016.
[73] J. Mutale, G. Strbac, C. Curcic and N. Jenkins, "Allocation of Losses in Distribution Systems
with Embedded Generation," IEE Proc.-Gener. Transm. Distrib., vol. 147, no. 1, p. 7–14, Jan.
2000.
[74] H.E.Z. Farag, E.F. El-Saadany, "A novel cooperative protocol for distributed voltage control in
active distribution systems," IEEE Trans. Power Syst., vol. 28, no. 2, pp. 1645-1656, May 2013.
[75] DTI Centre for Distributed Generation and Sustainable Electrical Energy, United Kingdom
Generic Distribution System Phase One, Mar. 2006. [Online]. Available:
http://www.sedg.ac.uk/ukgds.htm.
[76] C. L. Masters, "Voltage rise: The big issue when connecting embedded generation to long 11
kV overhead lines," Power Eng. J., vol. 16, no.1, pp. 5–12, 2002.
[77] 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.
[78] “Nen-en 50160 voltage characteristics of electricity supplied by public electricity networks,”
Tech. Rep., Tech. Rep., 2010.
[79] M. Mahmud, M. Hossain, and H. R. Pota, “Analysis of voltage rise effect on distribution
network with distributed generation,” IFAC Proceedings Volumes, vol. 44, no. 1, pp. 14796–
14801, 2011.
[80] C. Gonzalez, J. Geuns, S. Weckx, T. Wijnhoven, P. Vingerhoets, T. De Rybel, and J. Driesen,
“LV distribution network feeders in Belgium and power quality issues due to increasing PV
81/84
penetration levels,” in 2012 3rd IEEE PES Innovative Smart Grid Technologies Europe (ISGT
Europe). IEEE, 2012, pp. 1–8.
[81] A. Bidram and A. Davoudi, “Hierarchical structure of microgrids control system,” IEEE
Transactions on Smart Grid, vol. 3, no. 4, pp. 1963–1976, 2012.
[82] J. von Appen, M. Braun, and T. Kneiske, “Voltage control using PV storage systems in
distribution systems,” in 22nd International Conference and Exhibition on Electricity
Distribution (CIRED 2013), 2013.
[83] 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, 2013.
[84] 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.
[85] D. Deka, S. Backhaus, and M. Chertkovy, “Estimating distribution grid topologies: A graphical
learning based approach,” arXiv preprint: 1602. 08509, 2016.
[86] Jonas Engles, Hamada Almasalma and Geert Deconinck, “A distributed gossip-based voltage
control algorithm for peer-to-peer microgrids,” in Innovative Smart Grid Technologies
Conference (Sydney), 2016, pp. 1–6.
[87] 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.
[88] D. P. Bertsekas, Nonlinear programming. Athena scientific Belmont, 1999.
[89] D. Kempe, A. Dobra, and J. Gehrke, “Gossip-based computation of aggregate information,” in
Foundations of Computer Science, 2003. Proceedings. 44th Annual IEEE Symposium on. IEEE,
2003, pp. 482–491.
[90] S. Papathanassiou, N. Hatziargyriou, K. Strunz et al., “A benchmark low voltage microgrid
network,” in Proceedings of the CIGRE symposium: power systems with dispersed generation,
2005, pp. 1–8.
[91] I. Richardson, M. Thomson, and D. Infield, “A high-resolution domestic building occupancy
model for energy demand simulations,” Energy and buildings, vol. 40, no. 8, pp. 1560–1566,
2008.
[92] I. Richardson, M. Thomson, D. Infield, and A. Delahunty, “Domestic lighting: A high-
resolution energy demand model,” Energy and Buildings, vol. 41, no. 7, pp. 781–789, 2009.
[93] 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.
[94] I. Grigorik, High Performance Browser Networking: What every web developer should know
about networking and web performance. " O’Reilly Media, Inc.", 2013.
[95] G. Dáz, J. Gómez-Aleixandre, and J. Coto, “Direct backward/forward sweep algorithm for
solving load power flows in ac droop-regulated microgrids,” 2015.
[96] J. Tant, F. Geth, D. Six, P. Tant, and J. Driesen, “Multi-objective battery storage to improve PV
integration in residential distribution grids,” IEEE Transactions on Sustainable Energy, vol. 4,
no. 1, pp. 182–191, 2013.
82/84
[97] 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 2012 3rd IEEE PES Innovative Smart Grid Technologies
Europe (ISGT Europe). IEEE, 2012, pp. 1–8.
[98] C.W. Gellings, et al., “Demand-Side Management, Volume 3: Technology Alternatives and
Market Implementation Methods”, Electric Power Research Institute, EA/EM-3597, Volume 3,
Research Project 2381-4, December 1984.
[99] National Statistics (2016, Dec 15) National Travel Survey: 2015 report. [Online] Available:
https://www.gov.uk/government/statistics/national-travel-survey-2015
[100] Demand Side Response Performance Assessment: An Impact Analysis of Load Profile
Accuracy on DSR Performances
[101] Western Power Distribution (2016, Nov 10) SoLa Bristol SDRC 9.8 Final Report, [Online]
Avalable: https://www.westernpowerinnovation.co.uk/Document-
library/2016/WPDT2003_SoLa-Bristol_SDRC9-8-resubmissionv2-0.aspx
[102] M. Dorigo, V. Maniezzo, and A. Colorni, "Ant system: optimization by a colony of cooperating
agents," Systems, Man, and Cybernetics, Part B: Cybernetics, IEEE Transactions on, vol. 26,
pp. 29-41, 1996.
[103] M. D. a. L. M. Gambardella, "Ant colonies for the traveling salesman problem," BioSystems,
vol. 43, pp. 73-81, 1997.
[104] J. H. M. a. K. Fink. (2004). Module for the Powell Search Method for a Minimum. [Online]
Available: http://mathfaculty.fullerton.edu/mathews/n2003/PowellMethodMod.html
[105] M. E. Baran and F. Wu, "Network reconfiguration in distribution systems for loss reduction and
load balancing," IEEE Trans. Power Del., vol. 4, pp. 1401-1407, 1989.
83/84

A-2 P2P Based Probabilistic

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

A-2 P2P Based Probabilistic

Uploaded by

Copyright:

Available Formats

Ref.

Peer to Peer Smart Energy Distribution Networks

D5.2 P2P-based probabilistic and predictive control functions

Document ID: P2P-SMARTEST-WP5-D5.2-V4.0

Deliverable Title: P2P-based Probabilistic and

Responsible beneficiary: Cardiff University (CU)

Start Date of the Project: 1 January 2015

PROPRIETARY RIGHTS STATEMENT

Document ID: P2P-SMARTEST- WP5-5.2-151227-V4

Name Organization Date Visa

Coordinator Ari Pouttu UOULU 31.12.2016

Revision Date Modification Authors

3 Coordinated Voltage Control................................................................................................................40

ABBREVIATIONS AND GLOSSARY

ACO Ant Colony Optimization

P2P-SmartTest project investigates and demonstrates a smarter electricity distribution system

The objectives of this project are:

This deliverable is part of WP5.

1.1 Uncertainties and the requirement for distribution network control

Table 1.1 List of technical solutions

1.2 Potential control methods

1.2.1 Control through network reconfiguration and SOP

The main objectives of network reconfiguration include [16], [17]:

1. Supply restoration：This optimally restores de-energized customers through alternative

1.2.2 Coordinated voltage control

It is impossible to impose these architectures on the whole distribution grid, as it incorporates

Industrial Prosumer Commercial Prosumer

(a) Centralized Control Residential Prosumer

(b) Hierarchical Control

(c) Distributed Control

(d) P2P Control

Figure 1.2 Architectures of a coordinated control

1.2.3 Demand side integration

Figure 1.3 Concept and Objectives of Demand Side Integration [59]

1.2.3.2 Implementation methods of Demand Side Integration (DSI)

1.2.4 Coordinated control with multiple objectives

A number of literatures formulate the reconfiguration problem as multi-objective. In [68], a

1.3 Deliverable 5.2

2 NETWORK CONTROL THROUGH SOP

2.1 Problem formulation

Figure 2.2 An example of an SOP’s operating point

√(P1 )2 + (Q1 )2 ≤ S1; (2-1)

√(P2 )2 + (Q2 )2 ≤ S2 . (2-2)

∑ (PJ + P LOSS ) = 0 (2-3)

2.1.2 Jacobian matrix based sensitivity analysis

∆𝛿𝑖 𝑡 = ∑ (𝐶𝛿𝑖 𝑃𝑗 ∗ ∆𝑃𝑗 𝑡 + 𝐶𝛿𝑖 𝑄𝑗 ∗ ∆𝑄𝑗 𝑡 ) (2-7)

∆|𝑉𝑖 | 𝑡 = |𝑉𝑖 | 𝑡 ∗ ∑ (𝐶𝑉𝑖 𝑃𝑗 ∗ ∆𝑃𝑗 𝑡 + 𝐶𝑉𝑖 𝑄𝑗 ∗ ∆𝑄𝑗 𝑡 )

𝛿𝑖 𝑆𝑂𝑃 = 𝛿𝑖 𝑡 + ∆𝛿𝑖 𝑡 (2-9)

|𝑉𝑖 |𝑆𝑂𝑃 𝑡 = |𝑉𝑖 | 𝑡 + ∆|𝑉𝑖 | 𝑡

2.2 Optimization formulation

2.2.1 Voltage profile improvement (VPI)

where |𝑉𝑛𝑜𝑚 | is a target voltage for all nodes of the network.

2.2.2 Line utilization balancing (LUB)

2.2.3 Energy loss minimization (ELM)

𝑉𝑚𝑖𝑛 ≤ |𝑉𝑖 |𝑆𝑂𝑃 𝑡 ≤ 𝑉𝑚𝑎𝑥 (2-15)

|𝐼𝑘 | ≤ 𝐼𝑘 rate (2-16)

2.3 Visualization of the SOP operating region

a) Undervoltage in Feeder 1, undervoltage in Feeder 2 (Figure. 2.3);

Feeder 2: V2 = Vmax Feeder 1 voltage constraints

Feeder 2: V2 = Vmin Feeder 1: V1 = Vmax

III IV Feeder 1: V1 = Vmin

Feeder 1 voltage constraints

Feeder 2: V2 = Vmax Feeder 2: V2 = Vmin

Feeder 1: V1 = Vmin Feeder 2: V2 = Vmin