IEEE Power & Energy Magazine - January/February 2015 - 89

as weighted information to minimize the deviation between
measured and calculated power flow values.
one of the big challenges of smart grids is adding real
intelligence-for example, implementing automatic reconfiguration of distribution grids. at grid4eu, two initiatives deal with this issue. The spanish one is working on to
automatic service recovery after an unforeseen incident in
the grid, and the german one is focusing on optimization of
electric flows of energy so as to increase hosting capacity for
distributed renewable energy sources.
The spanish system is a rule-based, centralized approach
that analyzes the topology of the grid and identifies the fault
location so as to implement the necessary reconfigurations of
the grid and resupply as many consumers as possible in the
shortest possible time.
The german solution was conceived as a decentralized
approach using a multiagent system. There is a control center agent (cc-agent) placed at the primary substation that
plays the supervisory role and provides the gateway to the
supervisory control and data acquisition system. switchable
agents (s-agents) and measuring agents (M-agents), placed
at secondary substations (sss), are responsible for monitoring and switching at specific points in the grid and perform
cooperative negotiation and coordination actions. The agents
observe the grid and estimate an optimal load flow situation.
at the beginning, a completely decentralized approach
was proposed in which the system's intelligence was
divided among the agents. This approach proved to require
a huge effort, however. a two-step approach was therefore
chosen. In the first step, the system intelligence will be
concentrated in the cc-agent, steering the s-agents. The
second step will be an implementation of the decentralized approach. both systems constitute a first step toward
dynamic topology reconfiguration, which represents a new
concept of operation.
The French demonstration links production and demand
forecasting. a pV production forecast tool would enhance
the energy management system in two ways:
✔✔ It would identify possible voltage violations in cases of
large penetrations of distributed pV.
✔✔ It would facilitate the islanding of the network in the
case of a planned outage, balancing pV production and
consumption on the islanded network and using batteries as an energy buffer.
predictions at the closest grid points from the solar plants
are available at a three-hour time resolution and will be
interpolated to a half-hourly resolution. The european center for Medium-range Weather Forecasts, Meteo France,
and the National oceanic and atmospheric administration
are the sources of information.
The French demonstration is trying to implement a
demand-forecasting tool based on historical load curves and
information from smart meters, corrected by forecasts made
the day before and complemented by primary substation real
consumption data. readjustments via half-hourly readings
january/february 2015

received from the sss are made. The main demonstration
place in France is carros. The center in Figure 4 is used
to develop and test the forecasting algorithms that are used
in grid4eu.
The forecast for customers without smart meters is estimated through profiling. The sum of the forecasts for consumers with smart meters and without them should be close
to the forecast of total output of the ss. The algorithm tries to
maintain the coherence of the individual historical measurements with real-time ones.
Network planning and asset management for the new
working conditions remain open issues.

Innovative Power Management
of Low-Voltage Grids
as noted earlier, clusters four and five of the eegI road map
deal with MV and LV issues, but LV functionalities are less
developed in actual networks. Traditionally, LV grids have been
neither monitored nor automated. The reason was that distribution grids could be operated with partial automation of the MV
level. Nowadays, distributed generation is making it necessary
to monitor LV grids as well and develop new functionalities.
ceZ distribuce, in the czech republic, is implementing a demonstration of LV grid automatic reconfiguration.
switching and protection components have been installed at
LV panels, replacing the fuses, at six LV street cabinets. The
new components are being remotely controlled by an operator in hradec Králové. Local staff members are also able to
operate them manually.
The final objective is the fully automated control and
protection of an LV network. components will self-actuate
when current parameters are exceeded (as in an overload or
short circuit). The purpose is parameterization in the area
of full automation, so that selectivity will be assured along
with the ability to verify potential fault situations in normal
as well as islanded operation. Field outcomes are expected
next year.
LV grids are so vast and they suffer so many incidents
caused by external agents that it is really difficult to maintain

figure 4. A PV generation forecasting center. (Used with
permission.)
ieee power & energy magazine

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - January/February 2015