IEEE Power & Energy Magazine - May/June 2017 - 54

almost entirely on existing infrastructure (the OLTC is a
common asset in distribution networks around the world)
thus reducing complexity and cost, and 2) its scalability,
as it can unlock flexibility from thousands or millions of
customers by controlling only a limited number of points
(primary substations).
The voltage-led LM scheme proposed in CLASS was
successfully demonstrated with field trials involving 60
primary substations, i.e., more than 350,000 customers.
This project, arguably the largest voltage-led LM trial in
the world, considered not only the technical aspects but
also the social, environmental, and economic implications,
resulting in comprehensive learning useful for DNOs,
TSOs, and regulators willing to explore this source of flexibility. Table 1 presents a fact sheet detailing key aspects
of the project.

Reducing Voltages To Reduce Demand
When the supplied voltage is reduced, load undergoes a
reduction in demand-although to different extents depending on its type (e.g., resistor, induction motor, or fluorescent
lamp); in other words, there is a positive correlation between
the two. For instance, if the supplied voltage of an electric
kettle (which can be considered a resistor) is reduced by 2%,
the corresponding demand will reduce by 4%; for a compact
fluorescent lamp, which draws the same current constantly,
this demand reduction would be 2%.
However, in practice, customers use different types and
number of appliances throughout the day, resulting in a timevarying composition (in terms of load type). Consequently,
although a reduction in the supplied voltage is expected to
trigger a reduction in demand, its size will depend largely
on the time period in which this takes place. The concept is
illustrated in Figure 1, which, for simplicity's sake, shows the
supplied voltage of one house [Figure 1(a): U.K. values] and
the corresponding power consumption [Figure 1(b)] when
a voltage reduction of 10 V is applied during two half-hour
periods: one early in the day (at 2:00 a.m.) and one later in the
day (at 9:00 p.m.).
In Figure 1(b), we see that during the early part of the
day a reduction of lower than 10 W was achieved, in contrast to 50 W later in the day. This is not only because
of different demand volumes (more demand at 9:00 p.m.
than at 2:00 a.m.) but also because of significantly different load compositions. As can be seen from the example in
Figure 1(c), more responsive appliances (e.g., lighting) dominate the demand at 9:00 p.m. compared to less responsive
ones (e.g., freezers and refrigerators), which are the main
component at 2:00 a.m.
Hence, in the context of a voltage-led LM scheme aimed
at providing a demand reduction exclusively when and to
the extent required by the TSO, the assessment of the timevarying (daily and seasonal) demand composition is a major
challenge that must first be addressed. This requires developing time-varying load models that define the mathematical
54

ieee power & energy magazine

relationships between voltages and demand of any load, including the aggregation of thousands of customers downstream
from a given point in a network.
In contrast to a CVR scheme, commonly aimed at introducing at all times the largest voltage reduction possible to
achieve energy savings, a flexibility-driven, voltage-led LM
scheme needs to quantify the potential demand reduction
throughout the day so that it can provide estimates of the
available flexibility to the TSO. The second challenge, therefore, is to estimate how much demand reduction could be
achieved at different periods while always maintaining end
customer voltages within statutory limits.
It is worth highlighting that loads (appliances) are designed to operate within a statutory range of voltages. This
is because supplied voltages, in practice, are subject to
daily and seasonal fluctuations as a result of changing
demand and generation, as well as the actions of voltage
regulation devices-but they are always kept within the
statutory limits. Figure 1(a) illustrates this variability and
also that the supplied voltage can be higher than the nominal value (up to 18 V in this case). This practice, common
in most countries, ensures that remote customer voltages
are above the lower limit (according to national standards)
during peak hours.
Considering the acceptable range of operating voltages
of modern appliances and the positive correlation between
voltage and demand, reducing the supplied voltage of a
house (within statutory limits) reduces the demand of appliances without any detrimental effect and, ultimately, without directly involving the customer. This raises the questions
investigated by CLASS: what if the same principle is applied
to millions of houses? Would the resulting demand reduction
be meaningful for the provision of flexibility to the TSO?
A key aspect for the success of the CLASS voltageled LM scheme is the ability to induce a voltage reduction
throughout a designated short period and to a large number
of customers and so achieve the desired demand reduction
(the period and volume of the reduction to be determined
by the needs of the TSO). This idea, although promising,
requires understanding how it can be deployed in a practical
manner, i.e., leveraging the existing infrastructure, as discussed in the following section.

Implementing CLASS: Today's
Infrastructure for Tomorrow's Flexibility
DNOs around the world use a variety of devices to continuously maintain voltages within statutory limits. CLASS
leverages this existing infrastructure, in particular the OLTC
of primary substation transformers, a common network asset
in the United Kingdom as well as in many countries around
the world. The OLTC, in normal operation, changes the
transformer tap ratio to regulate the busbar voltage and, as
a consequence, the voltage of all downstream customers. In
European-style networks, this is also the closest voltage control point to customers.
may/june 2017

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - May/June 2017