to take into account capacity constraints at the
distribution network level.
• Virtual power plants with distribution system
operator. This model involves three levels of
control.
First,
VPPs
provide
the
transmission-level system operator with offer
curves for generation or demand reduction for
the resources they operate. Second, the
transmission-level system operator carries out
constrained dispatch of the transmission system. Third, DSOs modify the operation of
distributed resources to take into account
capacity constraints at the distribution network level.
• Distribution system operator. This model
involves two levels of control. First, DSOs act
as sole distribution-level VPPs, controlling all
system resources at the distribution level, and
carry out constrained dispatch of the whole
distribution system. DSOs provide the
transmission-level system operator with bid
or offer curves for the generation and demand
resources they operate. Second, the
transmission-level system operator carries out
constrained dispatch of the transmission
system.
Key criteria in this decision are the computational requirements, institutional characteristics
and operational vulnerability of each model. As
explained above, systems with high computational requirements, or small improvements in
computing technology, will require more levels
of resource coordination, while systems with low
computational requirements, or large improvements in computing technology, will require
fewer levels of resource coordination, and—with
sufficiently developed computing technology—
potentially a whole-system operator. Preferences
for particular institutional characteristics are also
relevant. For example, the whole-system operator
and distribution system operator models involve
coordination of resources by a single operator.
Consumers may or may not have concerns over
price, quality of service and privacy. If they have
concerns, they might prefer a model involving
control of resources by VPPs, which compete to
meet customer requirements. Finally, the models
may have different degrees of vulnerability to
failures in information and communications
technologies (ICT), such as those caused by
cyberattacks.
Hierarchies of resource control may be needed
until ICT capabilities are sufficiently developed.
While in the near term a single operator may be
able to optimise the electricity system with relatively low volumes of distributed resources, once
sufficient volumes of distributed resources are
deployed, the task of optimisation may be too
great for a single operator. Hierarchies of control
may be established early to ensure that increasing
volumes of distributed resources can be accommodated if improvements in computing technology fail to keep pace with increases in
computational requirements. Even if computing
technology improves to the point where a fully
distributed system can be optimised by a single
operator, an increase in the temporal resolution
of system control (for example, from half-hourly
settlement towards real time) would result in
significant further increases in computational
requirements. A shift from hierarchies of control
to a whole-system operator model would only be
viable if improvements in computing technology
are sufficient to meet the requirements of optimising a fully distributed system in real time.
(4) Coordinated investment in decentralised
resources
Electricity systems with largely centralised
resources provide adequate information to
investors on system needs. Electricity systems
periodically require investment in new resources,
such as new generation plants. In theory, developers invest in new resources in response to a
price signal in the wholesale market. These new
investments are typically large. In principle, this
leads to a coordination problem, whereby either
several investors might plan to develop a similar
resource (overinvestment) or investors might not
invest in required resources due to the risk that
other investors might do so (underinvestment). In
practice, these risks are minimised because the
transmission system operator knows which
resources are under development and awaiting
574
S. Yaodong and A. Gillespie
distribution network level.
• Virtual power plants with distribution system
operator. This model involves three levels of
control.
First,
VPPs
provide
the
transmission-level system operator with offer
curves for generation or demand reduction for
the resources they operate. Second, the
transmission-level system operator carries out
constrained dispatch of the transmission system. Third, DSOs modify the operation of
distributed resources to take into account
capacity constraints at the distribution network level.
• Distribution system operator. This model
involves two levels of control. First, DSOs act
as sole distribution-level VPPs, controlling all
system resources at the distribution level, and
carry out constrained dispatch of the whole
distribution system. DSOs provide the
transmission-level system operator with bid
or offer curves for the generation and demand
resources they operate. Second, the
transmission-level system operator carries out
constrained dispatch of the transmission
system.
Key criteria in this decision are the computational requirements, institutional characteristics
and operational vulnerability of each model. As
explained above, systems with high computational requirements, or small improvements in
computing technology, will require more levels
of resource coordination, while systems with low
computational requirements, or large improvements in computing technology, will require
fewer levels of resource coordination, and—with
sufficiently developed computing technology—
potentially a whole-system operator. Preferences
for particular institutional characteristics are also
relevant. For example, the whole-system operator
and distribution system operator models involve
coordination of resources by a single operator.
Consumers may or may not have concerns over
price, quality of service and privacy. If they have
concerns, they might prefer a model involving
control of resources by VPPs, which compete to
meet customer requirements. Finally, the models
may have different degrees of vulnerability to
failures in information and communications
technologies (ICT), such as those caused by
cyberattacks.
Hierarchies of resource control may be needed
until ICT capabilities are sufficiently developed.
While in the near term a single operator may be
able to optimise the electricity system with relatively low volumes of distributed resources, once
sufficient volumes of distributed resources are
deployed, the task of optimisation may be too
great for a single operator. Hierarchies of control
may be established early to ensure that increasing
volumes of distributed resources can be accommodated if improvements in computing technology fail to keep pace with increases in
computational requirements. Even if computing
technology improves to the point where a fully
distributed system can be optimised by a single
operator, an increase in the temporal resolution
of system control (for example, from half-hourly
settlement towards real time) would result in
significant further increases in computational
requirements. A shift from hierarchies of control
to a whole-system operator model would only be
viable if improvements in computing technology
are sufficient to meet the requirements of optimising a fully distributed system in real time.
(4) Coordinated investment in decentralised
resources
Electricity systems with largely centralised
resources provide adequate information to
investors on system needs. Electricity systems
periodically require investment in new resources,
such as new generation plants. In theory, developers invest in new resources in response to a
price signal in the wholesale market. These new
investments are typically large. In principle, this
leads to a coordination problem, whereby either
several investors might plan to develop a similar
resource (overinvestment) or investors might not
invest in required resources due to the risk that
other investors might do so (underinvestment). In
practice, these risks are minimised because the
transmission system operator knows which
resources are under development and awaiting
574
S. Yaodong and A. Gillespie