IEEE Power & Energy Magazine - May/June 2018 - 42

maximize the values of "small" data and thus avoid "expensive" data. The crucial question is whether there is a simple
and cost-effective method that can provide the visibility of
flows in the LV distribution network.
in the united Kingdom, a series of projects have been
conducted by dnos to answer this question. The LV network Templates project was one of these flagship projects.
it provides remote monitoring of current and voltage at over
800 distribution substations. in addition, voltage monitors
have been installed in approximately 3,500 premises at the
end of LV feeders. a set of LV network substation templates
was then developed to provide benchmark load and voltage
profiles that could be applied across the united Kingdom.
Traditional load profiling methods are based on supervised learning techniques and customers mapped into
predefined domestic, commercial, and industrial classes.
Because the industry has little prior knowledge of individual LV loads and network states, the project proposes a
novel three-stage semisupervised learning algorithm. The
first step is an unsupervised clustering to search for typical
substation templates. The next step is to assign an unknown
substation to the most similar template. The final step is to
estimate the loading levels of the LV substations by the use
of clusterwise regression.
This has produced two significant outcomes. First, the
templates developed were validated by five out of seven dnos
in the united Kingdom and, overall, achieved an 87% accuracy. Second, an even greater and longer-lasting impact of
the project has been the learning associated with the voltage
profiles. This, in turn, has led to a further project for voltage
reduction analysis.
Western power distribution (Wpd) asked the university of Bath to undertake a study regarding the effects of a
voltage reduction scheme for those voltage profiles close to
the upper bound of the statutory limits. university of Bath
researchers concluded that such a scheme could significantly reduce demand, customer bills, and carbon emission.
adopting the method for customers in South Wales, Wpd
reduced voltage by 1%. This has saved customers in South
Wales approximately £14 million per year. plans are in place
to deploy this approach across Wpd's four licensed areas,
where possible. if rolled out nationally in conjunction with
the adoption of voltage tolerances (±10%), it is estimated that
the policy could save great Britain £315 million and 1.98
million tons of carbon dioxide each year.

Conclusions
managing flexibility in generation, demand, and storage is a
requisite for achieving greater efficiency in electricity supply. exploiting this flexibility to achieve a higher utilization
of system assets is essential for enabling the united Kingdom
to transition to a low-carbon energy economy. Shifting to a
sharing economy is one of the most promising approaches
for exploiting flexibility through the use of high-speed iCT,

ieee power & energy magazine

exploration of disruptive business models, and creation of
innovative horizontal markets.
a key enabler will be big data analytics that can inform
users and market participants, establish prices, and match
flexibility in demand with intermittent generation through
extensive analyses of metered data together with socioeconomic and weather information. in this article, we discussed p2p energy markets and Sna to illustrate the value of
embracing a sharing economy in the energy sector. Through
these developments, major returns can be delivered to lowcarbon developers and flexible energy customers by extracting more value from existing generation, network, and
customer assets.

Acknowledgment
This work was supported by the eu2020 p2p-SmarTest
project and the epSrC peer-to-peer energy Trading and
Sharing-3m project.

For Further Reading
F. Li, "incentives for local production and demand flexibility
to support community energy," in Proc. New Market Models
for Community Energy Conf., Jan. 2015.
r. Li, C. gu, F. Li, g. Shaddick, and m. dale, "development of low voltage network templates-part i: Substation clustering and classification," IEEE Trans. Power Systems, vol. 30, no. 6, pp. 3036-3044, 2014. doi 10.1109/tpwrs
.2014.2371474.
T. hong and S. Fan, "probabilistic electric load forecasting: a tutorial review," Int. J. Forecasting, vol. 32, no. 3, pp.
914-938, July-Sept. 2016.
r. Li, Z. Zhang, F. Li, and a. petri, "a novel business
model for distribution networks: multiple network operators
with shared network access," IEEE Power Eng. Lett., 2017.
h. Shi, m. Xu, r. Li, and F. Li, "deep learning for household load forecasting: a novel pooling deep rnn," IEEE
Trans. Smart Grid, vol. pp, no. 99, pp. 1-1, 2017.
r. Li, Z. Zhang, F. Li and p. ahokangas, "a shared
network access business model for distribution networks,"
IEEE Trans. Power Systems, vol. 33, no. 1, pp. 1082-1084,
Jan. 2018.

Biographies
Furong Li is with the university of Bath, united Kingdom.
Ran Li is with the university of Bath, united Kingdom.
Zhipeng Zhang is with the university of Bath, united
Kingdom.
Mark Dale is with Western power distribution, united
Kingdom.
David Tolley is with the university of Bath, united
Kingdom.
Petri Ahokangas is with the university of oulu, Finland.
p&e

may/june 2018

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - May/June 2018