1
Abstract— The LINEAR project (Local Intelligent Networks
and Energy Active Regions) focuses on the introduction and
implementation of innovative smart-grid technologies in the
Flanders region, and aims at a breakthrough in the further
development and deployment of these solutions. It consists of a
research component and a large-scale residential pilot, both
focusing on active demand-side management of domestic loads.
This paper describes the unique approach, the main objectives
and the current status of this project. Selected business cases and
target applications are discussed in detail.
Index Terms—smart grid, demand side management, demand
response
I. INTRODUCTION
LINEAR is a large-scale research and demonstration
project focused on the introduction of smart grids, and more
specifically on active demand strategies, at residential
premises in the Flanders region in Belgium. This region is of
special interest due to its high standard of living and high
population density. Since the region has few economically
exploitable natural resources, energy prices are high. These
conditions, combined with the European 20-20-20 goals and
the Strategic Energy Technology (SET) plan put forward by
the European Commission [1], fostered a strong incentive with
the local government, industry, and domestic users to
investigate new technologies. It is expected that in a few years
time, the increasing integration of distributed energy resources
(DER) will severely impact the operation of the distribution
grid, unless active demand strategies are implemented by then.
Therefore, the Flanders region, which can be thought of as one
large city, is an ideal test-bed to implement a smart-grid trial.
The LINEAR project aims at a technological as well as an
implementation breakthrough of active demand techniques. Its
focus is twofold. On the one hand, the project deals with
The work is supported by the Ministry of Science (Minister I. Lieten) via the
LINEAR project organized by the agency for Innovation through Science and
Technology (IWT).
B. Dupont, P. Vingerhoets, P. Tant, T. De Rybel and R. Belmans are with the
Department of Electrical Engineering (ESAT), Research Group Electrical
Energy and Computer Architectures (ELECTA), KU Leuven, Kasteelpark
Arenberg 10, 3001 Leuven, Belgium (benjamin.dupont@esat.kuleuven.be).
K. Vanthournout, W. Cardinaels and E. Peeters are with the Flemish Institute
for Technological Research (VITO), Boeretang 200, B-2400 Mol, Belgium.
Both organizations are cooperating in EnergyVille, Dennenstraat 7, 3600
Genk, Belgium.
research and development efforts required to deploy active
demand technologies. On the other hand, it also aims at
implementing these technologies in a field test, by setting up a
residential pilot.
The project started in May 2009 and receives a partial
funding from the Flemish government for the academia and
research institutes (ESAT/ELECTA-KU Leuven, VITO, IBBT
and IMEC). Furthermore, several industrial partners, including
Belgacom, Eandis, EDF-Luminus, Fifthplay, Infrax,
Laborelec, Miele, Siemens, Telenet and Viessman, invest and
actively participate in the project and also review the work
plan, activities and output for relevance and quality. Finally,
the Flemish regulator for the electricity and gas market
(VREG), as well as industry and government interest groups
(Agoria, EWI, VOKA) either partake in, or provide their input
to, the steering board. Especially the multidisciplinary
collaboration and input of all parties, academia, research
institutes, government, industrial partners coming from the
telecom, energy, household devices and ICT sector, make this
project very challenging, but promising. Compared with other
European initiatives, such as ADDRESS [2] and EU-DEEP
[3], LINEAR is unique in a number of ways:
1) The main focus is on the implementation of
automated active demand technologies in the
distribution grid, and a field trial with around 250 end
users involved.
2) The project combines a customer centered approach
with a focus on technological research, where every
aspect is well supported by an extensive economical
framework. The knowledge exchange and
cooperation between these three viewpoints provides
the necessary means to come to a successful
implementation of active demand in a real field trial.
3) The collaboration with and cooperation of relevant
stakeholders allows the clustering of necessary
competences and experience in the field of smart
grids. Since the distribution system operators (DSOs)
are also involved, the link with their smart metering
pilots provides a detailed insight in user profiles of
over 1000 end users.
This paper provides an overview of the approach, main
objectives and current status of the LINEAR project. First, the
general project structure and high-level goals are briefly
explained in Section 2, followed by a detailed discussion of
LINEAR Breakthrough Project: Large-Scale
Implementation of Smart Grid Technologies in
Distribution Grids
B. Dupont, Student Member, IEEE, P. Vingerhoets, P. Tant, K. Vanthournout, W. Cardinaels,
T. De Rybel, E. Peeters, R. Belmans, Fellow, IEEE
2
the targeted business cases and their economical feasibility in
Section 3. Section 4 describes the practical implementation of
the residential field test, including an overview of the control
strategies deployed in this context. The subsequent section
highlights some of the recent research achievements of the
project. Finally, conclusions are given in Section 6.
II. PROJECT STRUCTURE AND HIGH-LEVEL GOALS
One of the high-level goals of LINEAR is to define the
development needs to enable the introduction of active
demand techniques, and subsequently to initiate this
technology innovation in additional projects. Furthermore, it is
the intention to create an implementation breakthrough of
active demand by means of a field trial, and to build up a
unique research infrastructure in Flanders, based on a
residential test site and a laboratory infrastructure to back up
the field tests.
The methodology used in LINEAR is shown in Fig 1.
Basically, there are two main components: one focusing on
research and development — which aims at technological
breakthroughs — and one primarily focusing on an
implementation breakthrough by setting-up a residential pilot.
The research part has to lead to the definition of concrete
technological concepts for smart energy supply. As Fig. 1
shows, this is done in several work packages, each focusing on
a specific aspect. First, user and device profiles are gathered
by a full-scale monitoring and sub-metering campaign. These
profiles are linked to different types of users, based on results
of enquiries with end users to study user acceptance for active
demand and smart grids. Second, profiles of devices are linked
to the energy use of the building and the flexibility of the
different users and devices is studied. Another research topic
is the combination of distributed generation units, such as
photovoltaic systems, and storage, taking into account thermal
as well as electrical storage. Furthermore, the impact of
increasing DER on related networks (electricity,
communication, gas) is studied, including the possibility of
grid reconfiguration in Flanders. And finally, the application
of plug in hybrid electric vehicles as either a load, a storage
means or a producer is also studied in detail.
The knowledge about these different aspects is combined in
order to define optimal combinations and control strategies of
loads, storage and production units and to integrate these in an
estimation of the potential for the measuring sites and to
formulate general advices for other projects. This process is
based on the development of algorithms and simulation tools
to support the field test.
To stimulate the implementation breakthrough, a
demonstration of the smart grids concept in an existing typical
Flemish residential area is foreseen. This pilot-site will be
fully operational in 2013. Active demand management of
around 250 buildings will be realized within the project,
including non-predictable energy resources. Both the research
and implementation part of the project are constantly in close
cooperation with the Energy Markets work package. This
work package assesses market aspects such as current and
future market structures, regulation and business cases for
active demand. The four most viable business cases are
selected and evaluated in the LINEAR field test. This is
explained in greater detail in the next sections.
III. BUSINESS CASES AND ECONOMICAL FEASIBILITY
The current market structure faces several challenges,
which will only become more stringent in the near future.
Residential electric energy consumption continuously
increases due to electrification by the introduction of heat
pumps and plug-in hybrid electric vehicles (PHEVs). The rise
of distributed generation and renewable energy production
leads to a generation paradigm shift [4]. Finally, the ageing of
power systems induces a replacement wave of electrical
infrastructure in the coming decades [5], [6]. Regardless of
these challenges, the power system needs to operate in an
economic and reliable way. Active Demand (AD) for
residential consumers is a means to accomplish this, as a
Fig. 1. General overview of the project structure, and the two main components aiming at a technological as well as an implementation breakthrough.
WP1 Data collection and user
acceptance
WP2 User profiles and
flexibility
WP3 Production
and storage
WP4 Impact on
electricity, gas and
communication
networks
WP5 Combination
building function
and mobility
WP6 Integration of all aspects and control strategies
Technological breakthrough
WP9 Market structure and
business cases
WP8 Residential
pilot
Implementation
breakthrough
Definition of development needs
Technology-innovation via
additional projects
3
significant amount of flexible resources is available on the
demand side [7]. Although AD offers a solution for the
upcoming challenges, it impacts the roles and responsibilities
of the current market actors. Moreover, new actors, like an
aggregator aiming at optimizing energy supply and
consumption, can arise [8]. To seize this changing market
structure, the LINEAR project explores four Business Cases
(BCs) based on demand response programs (Fig. 2): portfolio
management, wind balancing, load pattern control of low
voltage distribution transformers, and voltage control in low
voltage lines:
A. Portfolio management
The portfolio management BC is induced by the generator
and the retailer. Generators offer their electricity output in a
wholesale market, in which retailers buy electricity. This
results in wholesale prices which typically fluctuate during the
year and even within a day. Peak periods are characterized by
higher generation costs, because expensive peaking plants are
ramped up to cover demand. During off-peak periods, demand
is usually covered by cheaper base load plants, facing must-
run requirements. Although wholesale prices vary, residential
tariffs are typically kept constant for months, reflecting
average costs during that period. Therefore, residential
consumers get no economic incentive to shift consumption
away from peak periods. The incapability for residential
consumers to react, results from the lack of metering and real-
time billing [9]. Moreover, policy makers believed that the
responsiveness of electricity consumers to dynamic tariffs was
too low in comparison with the implementation costs [10].
If more renewable electricity generation from wind and
solar energy is integrated, the inconsistency is only
strengthened, as the impact of predicted wind power
penetration on wholesale prices is considerable [11], [12].
The European Commission’s strategy for competitive,
sustainable and secure energy towards 2020 tackles these
shortcomings and creates a framework for smart metering and
AD [13]. This makes the consumer capable of participating in
the electricity market by shifting consumption to lower price
periods.
In the portfolio management BC, residential consumers
help operating the power system in a more efficient way by
responding to dynamic electricity prices. This leads to several
short-term and long-term benefits. In the short term, energy
cost are reduced by prioritizing cheaper renewable electricity
generation and avoiding the ramping up of more expensive
peaking plants [14]. In the long term, utilities avoid capacity,
transmission and distribution costs [15], because the system
can be tuned on a lower peak demand due to sustained demand
response. Moreover, the large scale integration of renewable
energy in the long term is facilitated by AD, leading to
environmental benefits [16].
As Fig. 2 shows, the BC calculation elaborates these
benefits in two parts. First, the business model is developed,
modeling all market actors influenced by the business case
together with the interactions between them. Second, the
economic feasibility of the business model is assessed based
on an investment analysis. Although the costs and benefits
defer depending on the business case, the same methodology
is used throughout each business case.
B. Wind balancing
The wind balancing BC is induced by the Balancing
Responsible Party (BRP) in charge of controlling the balance
of injection and off-take in its perimeter on a quarter-hourly
basis. If a BRP is not able to control the balance due to
unexpected events, the instantaneous imbalance is managed by
the Transmission System Operator (TSO). In return, the TSO
transfers the costs involved for making use of power reserves
to the BRP. This is stated as the imbalance costs.
Renewable energy sources are characterized by their high
variability, low predictability and low controllability. This
implies that the imbalance risk of a BRP with renewable
energy sources in its portfolio is higher. A BRP can counter
this risk by increasing the predictability and controllability of
its portfolio. One way to accomplish this is by making use of
AD. In the wind balancing BC, a BRP faces real-time
imbalances in his portfolio caused by deviations from the wind
power prediction. The imbalance can be minimized by
controlling real-time flexible loads at residential premises in
its control area.
C. Low-voltage distribution transformer load
The BC which controls the load pattern of low-voltage
distribution transformers is induced by the DSO, who is
responsible for a reliable operation of the distribution grid and
a reliable supply of electrical energy to end users. Within this
business case, the DSO is the actor who manages the load on
the low voltage transformer. Although low-voltage
transformers are designed to continuously operate at nominal
power, the load profile of the transformer typically varies in a
cyclic way, with different consumption peaks during the day.
With a cyclic load, the ageing of the transformer accelerates
during the peak period due to higher temperatures in the
transformer. As described in [17] and [18], the relation
Fig. 2. Business cases induced by market actors.
4
between temperature and life time is non-linear. This implies
that the ageing is higher when the temperature varies more,
given that the average temperature is the same.
The introduction of distributed renewable energy sources
and DER, including PHEVs, influences the altitude and
variability of the load profile at the transformer. Photovoltaic
generation in the distribution network reduces the total load
delivery, although the peak loads remain almost unchanged.
PHEVs add peak loads, contributing to the ageing of the
transformer [19]. One way to cope with this problem is by
making use of AD. In this business case, the DSO decreases
the peak load of a given distribution transformer by
controlling flexible energy resources at residential premises in
its network in real-time, to defer or decrease network
investments.
D. Voltage profile
The BC which controls the voltage level in low-voltage
feeders is also induced by the DSO. Traditionally, voltage
levels are set at the beginning of the feeder, using tap
changing transformers. However, line voltages tend to
decrease when the feeder is heavily loaded, or increase due to
the impact of distributed generation units. Standards limit the
lower and upper boundaries of the voltage level [20].
The introduction of distributed renewable energy sources
and DER add complexity to this reasoning. Due to the boom
of photovoltaic panels, voltage rises can occur going above the
upper boundary. On the other hand, the integration of PHEV
leads to excessive voltage drops unless the charging is
coordinated. One way to cope with these fluctuations is AD.
In this business case, a DSO optimizes the voltage profile
on a LV feeder by controlling real-time flexible energy
resources at residential premises in its grid network to defer or
decrease network investments.
IV. LARGE-SCALE RESIDENTIAL PILOT PROJECT
The LINEAR project includes a large-scale field trial in
which residential active demand technology is validated in
real- life conditions. Next to technical validation, this includes
evaluating the social acceptance of the various scenarios that
are executed. Therefore, a large-scale enquiry was conducted,
involving a total of almost 2000 users. The results of this
social investigation were linked to the consumption profiles of
the customers. Customer sampling techniques were used to
guarantee the representativeness of the consumption profiles
[21]. This yielded a set of representative electricity and gas
demand profiles, which is used in the simulations of the
LINEAR project (see section V). At the end of the field trial
an enquiry towards user acceptance is planned with all
participants.
The field trial consists of three phases, i.e., a first phase
consisting of 33 so-called friendly users – colleagues and
family of LINEAR co-workers – where technology is
validated after the lab test phase and before it is rolled out full
scale in the other phases. Phase 1 started beginning 2011 and
remains active during the full length of the LINEAR project.
The second phase targets at 100 users, geographically
dispersed over the Flanders region and focuses on the
retailer/BRP business cases. The third phase is geographically
concentrated in two city regions, where the LINEAR field trial
will install active demand equipment in an additional 100
houses and make use of the DSO’s local smart metering
infrastructure. This implies that phase 3 will enable LINEAR
to measure and react in real-time to stimuli from the local
distribution grid. Phase 2 is active from April 2013 till April
2014, phase 3 from July 2013 till July 2014, each preceded by
a year of reference measurements at the participant’s premises.
Fig. 3 gives a general overview of the active demand
infrastructure that is deployed for the phase 1 users in the field
test. On top of the system resides the LINEAR Smart Grid
server, which is a combination of standard application
software and an in-house developed data analysis tool [22].
Fig. 3. The LINEAR field trial ICT infrastructure.
Server
Fifthplay
Touch
Display
Energy
meas.
Plugs
Network
Server
Fifthplay
Linear
Server
VITO DHW
buffer
INTERNET
HOME
GW GW GW GWGW
Miele
dish
washer
BSH
dish
washer
Miele
wash.
machine
Miele
tumble
dryer
BSH
wash
mach
VITO
controller
Miele
GW
Smart
Meter
Transformer
measurement
H&L
Historic
data
5
The LINEAR server is communicating with the servers of the
industrial partners that take the role of gateway (GW)
provider. Each GW provider maintains a platform consisting
of a server which accepts the active demand control
primitives, and gateways installed at the participant’s premises
which control the local smart devices and the user interfaces.
The control systems and algorithms that convert high-level
active demand commands into smart device actions are
integrated into the GW provider platforms.
Next to a gateway, the ICT infrastructure installed at the
residential premises consists of a touch display to
communicate pricing and other information to the participants,
an energy meter at the mains connection to measure the total
energy consumption and production in the household, and
smart plugs, which are mainly used as a sub-metering system.
The smart appliances currently scheduled for deployment in
the LINEAR project are:
1) Miele Smart Grid Ready washing machines, tumble
dryers and dishwashers, using power line
communication (PLC) to communicate with a
dedicated Miele gateway, which in turn is controlled
via an XML interface over Ethernet by the main GW.
2) Converted electric domestic hot water (DHW) buffers
using a custom-developed smart DHW controller. An
in-house prototype was developed and demonstrated in
a lab setup [23]. The controller communicates to the
gateways by means of open LINEAR JSON interfaces
over Ethernet, Wi-Fi or PLC. The communication
medium is decided upon per residence in function of
the local structural limitations.
3) Regular ‘non-smart’ washing machines, tumble dryers
and dishwashers, which are enhanced with a remote
delayed start, using the custom-developed VITO smart
grid controller box.
4) Finally, also smart heat pump systems are investigated
on their feasibility within the LINEAR time frame.
It should be noted that the smart devices are designed and
controlled in such a way that the impact on the comfort level
of the participant is minimized. All devices have comfort
settings which always take precedence over the active demand
control commands.
As discussed in the previous section, a wide selection of
potential applications of residential flexibility or business
cases are investigated. Portfolio management tries to achieve a
better production/consumption balance, based on day-ahead
predictions and prices and focusing on renewables, such as
wind and sun. Real-time intra-day balancing of wind
production uses residential active demand to cover the delta
between predicted and actual wind production. Furthermore,
the load pattern on low-voltage transformers and voltage
deviations in distribution grids are also considered. In order to
achieve these technical goals, different control primitives are
evaluated. Fig. 4 gives an overview of which control strategies
are used for what business cases and in which field trial phase
the combinations are validated. The number on the graph
indicates the phase of the field trial. On the left axis, for every
business case the different control strategies are indicated:
Fig. 4. Business cases versus control strategies during the LINEAR field trial.
1) Manual Time of Use (ToU): variable prices for fixed
timeslots are communicated day-ahead to the
participants using web services and touch displays,
such that the participants can shift their electricity
consumption accordingly. The participants receive a
financial remuneration as an incentive, based on their
shifting and compared to reference consumption
profiles compiled during the reference measurements.
2) Automated ToU: identical to manual ToU, safe that
smart devices automatically optimize their energy
consumption in function of the ToU tariffs.
3) Automatic flexibility control: multiple houses are
combined into a cluster, with the total net consumption
or production of the houses combined as the prime
control parameter. Active demand commands
essentially increase or decrease this value and are
executed by controlling the smart devices in the
cluster.
At the time of publication, the roll out for the phase 2
participants is complete, and reference measurements are
ongoing. Meanwhile the selection of the phase 3 participants
is being conducted in the region of Hombeek-Leest. On the
technical side, algorithms are being developed and
implemented to evaluate each of the business cases and define
the interaction between the LINEAR server and the gateways.
V. SCIENTIFIC RESEARCH AND SIMULATION TOOLS
As discussed previously, the majority of the research in the
project serves as input for the different phases of the field test,
which in turn provides data to validate the theoretical
approach. The main research components of the LINEAR
project are the impact of DER on electricity and gas networks,
the integration of electric vehicles and electrical and thermal
storage.
One of the key issues in the upcoming decade is dealing
with an increasing penetration of photovoltaic (PV)
installations in the distribution grid. This can lead to
unpredictable production peaks, which can cause power
quality problems in the distribution grid. To investigate the
scale and magnitude of these problems, load flow simulations
have been performed on four typical Flemish feeders, from a
rural to a city environment [24]. Important input of the
simulations is the detailed structure of the feeders and the load
profiles, which were provided by the industrial partners and
