frequency. For flexible resources to be deployed
at volumes that bring the greatest benefits,
mechanisms exist to reward them for the system
services they provide.
A simple set of markets for each system
service allocates existing flexible resources and
signals investment need. Storage and demand
response are already actively engaged in balancing supply and demand in several wholesale
markets worldwide. However, markets for system stability are typically not developed enough
to allow flexible resources to participate to their
fullest extent. Procurement mechanisms for
system services typically specify these services
in terms of the properties of the thermal generators that historically provided them, such that
these mechanisms are often closed to new,
low-carbon flexible resources. Recent attempts
to create procurement mechanisms for flexible
resources have resulted in a patchwork of
complex and mutually inconsistent mechanisms.
For example, in the UK, electricity storage
facilities providing a short-term ancillary service called enhanced frequency response are not
allowed to participate in the capacity mechanism. A simple set of markets to reward all
system services provided by flexible resources
is needed to ensure adequate investment in
non-network alternatives.
(3) Model for controlling decentralised
resources
While distribution networks today are largely passive, they will need to become active to accommodate distributed resources. As demand on the
distribution network is typically inflexible, generators for the transmission system operate flexibly to
meet demand and maintain system security. As
decentralised electricity resources (distributed
generation, storage and demand response) are
deployed in the distribution network, these
resources will need to operate flexibly.
Distributed resources increase the complexity
of the electricity system. Optimising the operation of the electricity system involves finding the
optimal volume of output and consumption of
every resource in the power system. A conventional, centralised electricity system typically
includes a limited number of large
transmission-connected generators, suppliers and
large industrial consumers with flexible demand.
However, a decentralised electricity system will
include a very large number of small decentralised resources. The number of resources to be
optimised might increase by a factor of several
million.
A similar increase in computational demand
would occur if the temporal resolution at which
resources are controlled increases. For example,
shifting
from
hourly
to
real-time
(second-by-second) settlement of all dispatch
and consumption actions would increase the
computational demands of optimal system balancing by a factor of 3,600.
If the system is too complex for a single
system operator to balance, a hierarchy of
resource control will be needed. Complete optimisation of a decentralised electricity system
would have very significant computational
requirements. If computing technology is not
able to meet these requirements, then control of
the system will be distributed.
Virtual power plants (VPPs) and distribution
system operators have a role to play in a hierarchy of resource coordination. VPPs, also known
as aggregators, could coordinate (aggregate)
decentralised resources and coordinate them
individually to present the (transmission) system
operator with a level of net generation (or consumption). If the system is complex, more than
two levels of resource coordination might be
needed, for example, with some VPPs coordinating the activity of smaller VPPs further down
the hierarchy. Distribution system operators are
VPPs that coordinate all resources in a given
distribution system, either directly or via intermediate VPPs.
Hierarchies of resource control provide only
partial optimisation of the electricity system. In
order to optimise the whole electricity system, a
single optimising agent must know the demand
and supply curves for each system resource.
Where no single agent has this information, only
partial optimisation is possible, as groups of
resources for which information is available must
be optimised separately. Markets between groups
98
W. Xiaoming et al.
at volumes that bring the greatest benefits,
mechanisms exist to reward them for the system
services they provide.
A simple set of markets for each system
service allocates existing flexible resources and
signals investment need. Storage and demand
response are already actively engaged in balancing supply and demand in several wholesale
markets worldwide. However, markets for system stability are typically not developed enough
to allow flexible resources to participate to their
fullest extent. Procurement mechanisms for
system services typically specify these services
in terms of the properties of the thermal generators that historically provided them, such that
these mechanisms are often closed to new,
low-carbon flexible resources. Recent attempts
to create procurement mechanisms for flexible
resources have resulted in a patchwork of
complex and mutually inconsistent mechanisms.
For example, in the UK, electricity storage
facilities providing a short-term ancillary service called enhanced frequency response are not
allowed to participate in the capacity mechanism. A simple set of markets to reward all
system services provided by flexible resources
is needed to ensure adequate investment in
non-network alternatives.
(3) Model for controlling decentralised
resources
While distribution networks today are largely passive, they will need to become active to accommodate distributed resources. As demand on the
distribution network is typically inflexible, generators for the transmission system operate flexibly to
meet demand and maintain system security. As
decentralised electricity resources (distributed
generation, storage and demand response) are
deployed in the distribution network, these
resources will need to operate flexibly.
Distributed resources increase the complexity
of the electricity system. Optimising the operation of the electricity system involves finding the
optimal volume of output and consumption of
every resource in the power system. A conventional, centralised electricity system typically
includes a limited number of large
transmission-connected generators, suppliers and
large industrial consumers with flexible demand.
However, a decentralised electricity system will
include a very large number of small decentralised resources. The number of resources to be
optimised might increase by a factor of several
million.
A similar increase in computational demand
would occur if the temporal resolution at which
resources are controlled increases. For example,
shifting
from
hourly
to
real-time
(second-by-second) settlement of all dispatch
and consumption actions would increase the
computational demands of optimal system balancing by a factor of 3,600.
If the system is too complex for a single
system operator to balance, a hierarchy of
resource control will be needed. Complete optimisation of a decentralised electricity system
would have very significant computational
requirements. If computing technology is not
able to meet these requirements, then control of
the system will be distributed.
Virtual power plants (VPPs) and distribution
system operators have a role to play in a hierarchy of resource coordination. VPPs, also known
as aggregators, could coordinate (aggregate)
decentralised resources and coordinate them
individually to present the (transmission) system
operator with a level of net generation (or consumption). If the system is complex, more than
two levels of resource coordination might be
needed, for example, with some VPPs coordinating the activity of smaller VPPs further down
the hierarchy. Distribution system operators are
VPPs that coordinate all resources in a given
distribution system, either directly or via intermediate VPPs.
Hierarchies of resource control provide only
partial optimisation of the electricity system. In
order to optimise the whole electricity system, a
single optimising agent must know the demand
and supply curves for each system resource.
Where no single agent has this information, only
partial optimisation is possible, as groups of
resources for which information is available must
be optimised separately. Markets between groups
98
W. Xiaoming et al.