Applied Energy 87 (2010) 25812588

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

The role of demand-side management in the grid integration of wind power

Pedro S. Moura *, Anbal T. de Almeida
ISR University of Coimbra, Dep. Electrical and Computer Eng., University of Coimbra, Polo II, 3030-290 Coimbra, Portugal

a r t i c l e

i n f o

a b s t r a c t

Article history:
Received 9 December 2009
Received in revised form 10 February 2010
Accepted 23 March 2010
Available online 18 April 2010
Keywords:
Intermittent power
Wind power
Grid integration
Demand-side management
Demand response
System reliability

2010 Elsevier Ltd. All rights reserved.

1. Introduction

2. Wind intermittence

The 2001/77/EC European Commission Directive sets the target

of 22% of gross electricity generation from renewables for the Europe, by 2010. In the case of Portugal the target is more ambitious at
39%. In order to satisfy this target, the installation of additional
renewable generation capacity is planned, in which the largest part
of the increase it is in wind power, whose target is to reach
5700 MW by 2012, and 8000 MW by 2020.
In a scenario of large scale penetration of renewable production
from wind and other intermittent resources, it is fundamental that
the electric system has appropriate means to compensate the effects of the variability and randomness of the wind power availability. This concern was traditionally addressed by the
promotion of wind resource studies and in the identification of
solutions based on reversible hydropower dams [1] and other options related to the energy supply (e.g. back-up generation) and energy storage technologies. The use of technologies of demand-side
management and demand response is an option that must be considered in the system planning, due to the cost-effective possibility
to adjust the consumption in response to the variations in the wind
power production.

Wind energy has characteristics that differ from conventional

energy sources. If the energy contribution of this production
source is not a cause of concern, the power balance, and therefore
the impact in the supply security, needs attention due to the
intermittent and random character of this production option.
Wind capacity is installed to generate energy with negligible
CO2 emissions, but its contribution to meet peak load growth
requirements is limited.
Wind power cannot fully replace the need for a variety of
capacity resources, which are dispatchable generators that are
available to be used when needed to meet peak load. Wind
power must be considered an energy resource, but not a peak
capacity resource, because only a small fraction of total wind
capacity has a high probability of running consistently. Wind is
used when it is available and if it has some capacity value for
reliability operation planning purposes that should be viewed
as a bonus [2].
The output of wind power is driven by environmental conditions outside the control of the generators or the system operators.
Since the wind is determined by random meteorological processes
it is inherently variable. Supply of power from wind turbines is stochastic in nature and the actual power is more or less proportional
to the third power of the wind velocity. The wind output varies
seasonally between summer and winter [3] and the variations
are also present on shorter time scales, namely on hourly basis

* Corresponding author. Tel.: +351 239 796 304; fax: +351 239 406 672.
E-mail address: pmoura@isr.uc.pt (P.S. Moura).
0306-2619/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2010.03.019

2582

P.S. Moura, A.T. de Almeida / Applied Energy 87 (2010) 25812588

Nomenclature
P(r, h, y)BAU, P(i, h, y)BAU, P(t, h, y)BAU power in the residential (r),
industrial (i) and tertiary (t)
sector (BAU scenario) to the
year y (20102030) and
hour h (024), MW
P(r, h, 2008), P(i, h, 2008), P(t, h, 2008) power in the residential
(r), industrial (i) and tertiary (t) sector to the
year of 2008 and hour
h (024), MW
E(r, y), E(i, y), E(t, y) annual energy consumption in the residential
(r), industrial (i) and tertiary (t) sector to the
year y, GWh
E(r, 2008), E(i, 2008), E(t, 2008) annual energy consumption in
the residential (r), industrial
(i) and tertiary (t) sector to
the year y, GWh
P(r, h, y)DSM, P(i, h, y)DSM, P(t, h, y)DSM power in the residential (r),
industrial (i) and tertiary
(t) sector (DSM scenario)
to the year y (20102030)
and hour h (024), MW

[4]. Unlike conventional capacity, wind-generated electricity cannot be reliably dispatched or perfectly forecasted, and exhibits significant temporal variability.
Several extreme ramp rates were recorded during storms [5]:
 Denmark 2000 MW (83% of capacity) decrease in 6 h or
12 MW (0.5% of capacity) in a minute on 8th January, 2005.
 North Germany over 4000 MW (58% of capacity) decrease
within 10 h, extreme negative ramp rate of 16 MW/min (0.2%
of capacity) on 24th December, 2004.
 Ireland 63 MW in 15 min (approx. 12% of capacity at the
time), 144 MW in 1 h (approx. 29% of capacity) and 338 MW
in 12 h (approx. 68% of capacity).
 Portugal 540 MW within 1 h (27% of the total wind installed
capacity) and 700 MW in 3 h on August 15, 2007.
 Spain 800 MW (7%) increase in 45 min (ramp rate of
1067 MW/h, 9% of capacity), and 1000 MW (9%) decrease in
1 h and 45 min (ramp rate 570 MW/h, 5% of capacity). Generated wind power between 25 MW and 8375 MW have occurred
(0.2% and 72% of capacity) in a single year.
 Texas, USA loss of 1550 MW of wind capacity at the rate of
approximately 600 MW/h over a 2 hour period on February
24, 2007.
Beyond the variability, a lot of wind generation occurs in hours
when energy use is low. The uncontrollable nature of wind makes
it less valuable to system operators than dispatchable power. The
variability and uncertainty of wind energy production require that
power system operators take measures to manage its delivery,
increasing the cost incurred to balance the system and maintain
reliability [6].

3. Options for managing intermittency

The intermittency of wind energy can be reduced by the use of
grid integration and technical and geographic distribution of the
generators. These techniques can be grouped as aggregation and
distribution methods and have as aim the substantial reduction
of the global wind power variations. Also the improved forecasting

Ps(r, h, y)DSM, Ps(i, h, y)DSM, Ps(t, h, y)DSM power saving due to

DSM in the residential (r), industrial (i)
and tertiary (t) sector
to the year y (2010
2030) and hour h
(024), MW
P(h, y)BAU global power (BAU scenario) to the year y (20102030)
and hour h (024), MW
P(h, y)DSM global power (DSM scenario) to the year y (20102030)
and hour h (024), MW
P(h, y)DR global power (DR scenario) to the year y (20102030)
and hour h (024), MW
Ps(h, y)DR power saving due to DR to the year y (20102030) and
hour h (024), MW

techniques can increase the predictability of the wind power production and therefore minimize the impact in the system.
However, although those improvements bring benefits, several
periods of low wind production and substantial variations will remain. Thus, tools to respond to short- to medium-term and longterm variability will be necessary, managing the operational and
capacity reserve, respectively. For large scale integration of wind
power the provision of flexible capacity reserve will be of crucial
importance. To achieve that aim several options are possible [7]:







Power plants providing operational and capacity reserve.

Interconnection with other grid systems.
Curtailment of intermittent generation.
Distributed generation.
Complementarily between renewable sources.
Energy storage.

All the above options have as aim the influence and control of
the supply. However, to minimize the intermittence impacts, also
the demand can be influenced, having a major role in the intermittence compensation, using demand-side management and demand-side response technologies.
4. - Demand-side management and wind power intermittence
As far as security of supply is concerned, the most severe problems due to the wind power intermittence occur in the peak load
hours. With higher energy consumption, the largest part of the
available system resources to deal with the intermittence is already used and a sudden reduction of the wind power production
can have critical consequences on the system reliability. Even in
situations where the wind power variation is not high but the wind
power production is low, if the wind power has a high share in the
total installed power, the system may not have enough resources
to face a lengthy absence of wind power. In both situations, notwithstanding the system reliability, the simultaneous occurrence
of a higher consumption and a low wind power production will
lead to higher operating costs.
Due to the higher peak load and wind availability, the sudden
reductions of the wind power production can be more severe in

P.S. Moura, A.T. de Almeida / Applied Energy 87 (2010) 25812588

2583

Fig. 1. Wind power production and forecasted power (July 21, 2008).

Fig. 2. Wind power production vs. temperature in a hot day in Portugal.

Fig. 3. Wind power production vs. energy consumption a hot day in Portugal.

winter days. However, the probability of a large period of time with

low wind availability is larger in the summer months, when the
hot days without wind are usual. Due to the increasing installed
capacity of air conditioners, the peak power consumption in the
summer days is growing fast. Fig. 1 shows a summer day in Portugal (July 21, 2008) with very low wind power production, during
all the day. On such a day the energy consumption was high.
On hot days the wind power production varies almost inversely
relatively to the average temperature (Fig. 2), whereas the energy
consumption varies almost directly with the temperature (Fig. 3).
Such situation will lead to reliability problems in the systems with
high wind power penetration.

Thus, instead of acting in the supply side, to avoid the most severe intermittent situations, the demand-side must be influenced
in the direction of achieve consumption reductions. Rather than
attempting to match power generation to consumer demand, the
philosophy of load management takes action to vary the load to
match the power available. Through the proper application of demand-side management (DSM) technologies it is possible to reduce the consumption to match the reductions of the wind
power production and increasing the reliability [8].
As defined by EIA, DSM is the planning, implementation, and
monitoring of utility activities designed to encourage consumers
to modify patterns of electricity usage, including the timing and le-

2584

P.S. Moura, A.T. de Almeida / Applied Energy 87 (2010) 25812588

vel of electricity demand. It refers to only energy and load-shape

modifying activities that are undertaken in response to utilityadministered programs. It does not refer to energy and loadshaped changes arising from the normal operation of the marketplace or from government-mandated energy-efficiency standards.
Demand-Side Management covers the complete range of loadshape objectives, including strategic conservation and load management, as well as strategic load growth [10].
Generically, as far as security of supply is concerned any
consumption reduction will contribute to mitigate the intermittence. With lower energy consumption, the installed power in
renewable intermittent resources needed to meet the minimum
renewable targets, will be lower. Additionally, the most severe
intermittence problems occurs during the peak load hours and
thus the DSM measures with greater impact in such hours are of
large importance.
However, in cases of high wind power penetration, the energy
consumption reduction during the peak hours may not be enough,
because due to the big production variations, such impact in the
long term will not solve the supplydemand balance. In such situations it will be very important to have demand response (DR)
technologies to force consumption reductions at near real time,
in the precise moment in which the critical situations occur.
With the DR technologies it is possible to direct or indirectly
force a consumption reduction in critical situations, in a short time
[9]. In the past, the electric system has been planned and operated
under the supposition that the supply system must meet all customers energy use, and that is not possible to control the demand.
However that supposition is starting to change due to the creation
of opportunities for customers to manage their energy use in response to signals (prices or load contracts).
The idea behind DR is that if the marginal peak load price is
higher than the value that a consumer gets out of the services derived from the electricity, he would be willing to modify the demand, if paid the peak price or slightly less instead. A grid
operator can obtain an economic benefit paying to a customer to
reduce the consumption instead to paying a power producer to
supply more output, because in peak periods, the production cost
can be very high.
Traditionally the DR technologies were typically used to attend
to economical concerns. However, nowadays they can be used to
improve the system reliability, reducing instantaneously the energy consumption to prevent the most unbalanced situations, like
the problems that result from the large space conditioning con-

sumption on days with reduced wind velocity. As more customers

practice automated price-responsive demand or automatically receive and respond to directions to increase or decrease their electricity use, system loads will be able to respond to, or manage,
variability from wind power production.
5. - Demand-side management impact
Taking as reference the electric energy consumption in Portugal,
in the year of 2008, an evaluation of the consumption evolution
was made, considering a business-as-usual scenario (BAU) and a
scenario with DSM measures application. In such evaluation, a consumption increase of 3%/year to the BAU scenario (average forecasted growth) and the application of DSM measures
corresponding to 1% of the year consumption to the DSM scenario
(2006/32/EC European Commission Directive on energy end-use
efficiency) were considered (Fig. 4).
The application of demand-side management measures in Portugal with a consumption reduction of 1% by year can represent, in
2020, a consumption reduction of 10.3% relatively to the BAU scenario. For a renewable share, for instance of 50% of the generated
energy, the DSM can reduce the needs of renewable power in
8.8%, to accomplish that target, and thus the needs of intermittent
power.
The consumption reduction was divided between the three
activity sectors, keeping the actual distribution share (Fig. 5). To
enable the impact analysis of the measures, the consumption
reduction was divided between several measures for each sector,
having as selection criterions: the cost-effectiveness, the application potential and the consumption reduction during the peak
hours. The total consumption reduction for each measure was
determined taking into consideration the yearly potential, determined based on the market data for each technology.
The selected measures were the following:
 Inefficient refrigerators and freezers replaced by equipments
with a high efficiency class (in Europe this corresponds to A+
or A++ labels).
 Incandescent lamps replaced by CFLs.
 Inefficient air conditioning equipments replaced by heat pumps
with high efficiency class (in Europe this corresponds to A
labels).
 Fluorescent lamps T8 with magnetic ballasts replaced by fluorescent lamps T5 with electronic ballasts.

Fig. 4. Electric energy consumption evolution with BAU and DSM scenarios.

P.S. Moura, A.T. de Almeida / Applied Energy 87 (2010) 25812588

2585

Fig. 5. Reduction of energy consumption by activity sector.






Conventional traffic lights replaced by LEDs.

Conventional public lightning lamps replaced by LEDs.
Replacement of conventional motors by high efficiency motors.
Use of variable speed drives.

The impact of the DSM measures in the residential sector load

diagram was analyzed, considering the consumption evolution
forecasted for the sector. The load diagrams supplied by the transmission operator were used and updated for each year, by the relation between the consumptions (Eq. (1)).

Pr; h; yBAU Pr; h; 2008  Er; y=Er; 2008

Taking into account the load diagrams of the residential sector

disaggregated by loads, estimated by the electric utility EDP, the
variation for each load was transformed in a percentage variation
(Fig. 6) and used to distribute the consumption reductions due to
DSM measures along the load diagrams. The technologies with larger consumption during the peak load hours are lighting, air conditioning and washing machines. Therefore the related measures
have a larger potential impact. However, high-efficiency washing
machines present a low cost-effectiveness. To the load diagram
to each year, the diagram of the DSM measures was subtracted,

obtaining the BAU and DSM scenarios (Eq. (2)). As example, the
Fig. 7 shows the load diagram, in a work day, to the residential sector, in January, 2020.

Pr; h; yDSM Pr; h; yBAU Psr; h; yDSM

The impact of the DSM measures in the industrial sector load

diagram was analyzed, considering the consumption evolution
forecasted for the sector (Eq. (3)). In the absence of reliable estimative for the disaggregated industrial sector load diagram, and due
to the fact that the largest potential of applications of the DSM
measures are in industries of continuous process, the consumption
reductions were considered homothetic. To the load diagram of
each year, the diagram of the DSM measures was subtracted,
obtaining the BAU and DSM scenarios (Eq. (4)).

Pi; h; yBAU Pi; h; 2008  Ei; y=Ei; 2008

Pi; h; yDSM Pi; h; yBAU Psi; h; yDSM

The impact of the DSM measures in the tertiary sector load diagram was analyzed, considering the consumption evolution forecasted for the sector. The load diagrams determined by the
transmission operator were used and updated for each year, by
the relation between the consumptions (Eq. (5)).

Fig. 6. Load profile to the residential sector.

2586

P.S. Moura, A.T. de Almeida / Applied Energy 87 (2010) 25812588

Fig. 7. Residential sector load diagrams for the BAU and DSM scenarios, on a work day, in January, 2020.

Fig. 8. Load profile to the tertiary sector.

Fig. 9. Tertiary sector load diagrams for the BAU and DSM scenarios, on a work day, in January, 2020.

P.S. Moura, A.T. de Almeida / Applied Energy 87 (2010) 25812588

Pt; h; yBAU Pt; h; 2008  Et; y=Et; 2008

Taking into account the load diagrams of the tertiary sector

disaggregated by loads, estimated by the electric utility EDP, the
variation for each load was transformed in percentage variation
(Fig. 8) and used to distribute the consumption reductions due to
the DSM measures along the load diagrams. The technologies with
larger consumption on the peak load hours are the lighting and air
conditioning and therefore the related measures have a larger impact. In the BAU load diagram of each year, the diagram of the DSM
measures was subtracted, obtaining the BAU and DSM scenarios
(Eq. (6)). As example, the Fig. 9 shows the load diagram, in a work
day, for the tertiary sector, in January, 2020.

Pt; h; yDSM Pt; h; yBAU Pst; h; yDSM

Finally, the global impact of the DSM measures in the Portuguese load diagram was evaluated, adding the impact of the three
sectors (Eqs. (7) and (8)). As example, Fig. 10 shows the load dia-

2587

gram, on a work day, in January, to 2020. In such image the largest

impact of the DSM measures in the peak load hours can be observed. In this example, a reduction of 13% in the peak load was obtained, due to the applied DSM measures, satisfying the defined
objective, to minimize the most dangerous situations originated
by the intermittence of the renewable resources. It must be noted
that such impact, caused by the DSM measures, is obtained with an
average cost of 0.023 /kWh, that is much less than the production
cost of any renewable energy.

Ph; yBAU Pr; h; yBAU Pi; h; yBAU Pt; h; yBAU

Ph; yDSM Pr; h; yDSM Pi; h; yDSM Pt; h; yDSM

Additionally, controlling 5% of the peak load (nearly 490 MW)

with DR technologies it is possible to smooth the loads diagram.
In the Fig. 11 it is possible to see the impact of the DR measures

Fig. 10. Total load diagram for the BAU and DSM scenarios, on work day, in January, 2020.

Fig. 11. Load diagram DSM and DR to a work day, in January, 2020.

2588

P.S. Moura, A.T. de Almeida / Applied Energy 87 (2010) 25812588

Fig. 12. Evolution of the peak load, in January, to the BAU, DSM and DR scenarios.

added to the DSM scenario (Eq. (9)). With the aggregated action of
the measures it will be possible to achieve a peak load reduction, in
2020 of 17.4%.

Ph; yDR Ph; yDSM Psh; yDR

Taking into account the evaluated impacts to the various

months, the evolution of the peak load was estimated, for the scenarios BAU, DSM and DR. In the Fig. 12 it is possible to see the evolution of the peak load, in January. It can be noted the increasing
reduction of the peak load with obvious benefits in the attenuation
of the intermittence problems.
6. Conclusions
The most severe problems due to the wind power intermittence
occur during the peak load hours, when the wind power production can be low and the energy consumption very high. Thus, the
DSM measures can have a major role in the integration of the wind
power production, reducing the consumptions to avoid the most
dangerous reliability problems.
Generically, any consumption reduction will contribute to reduce the intermittence as far as security of supply is concerned, because with lower energy consumption, the installed power in
renewable intermittent resources needed to accomplish the minimum renewable targets, will be lower. With the application of
measures to reduce the consumption of 1% by year, it is possible
to reduce the needs of renewable intermittent power in 11.24%,
with a lowest cost (0.023 /kWh) than any source of renewable
power.

The DSM measures with greater impact in the peak load hours
will also mitigate the problems caused by the variations and reduced wind power productions in the peak hours. Additionally,
when abrupt variations of the wind power production occur, DR
technologies can in near real time adjust the demand to avoid
the most unbalanced situations. The aggregated action of the
DSM and DR measures, in Portugal, can reduce the peak load in
Portugal, in 2020, in 17.4% accomplishing such objective.
References
[1] de Almeida A, Moura P, Marques A, Almeida J. Multi-impact evaluation of new
medium and large hydropower plants in Portugal centre region. Renewable
Sustainable Energy Rev 2005;9(2):14967.
[2] DeCarolis J, Keith D. The costs of winds variability: is there a threshold? Electr
J 2005(January/February):6977.
[3] Hamacher T, Haase T, Weber H, Dweke J. Integration of large scale wind
power into the grid. Power-Gen Europe 2004; 2004.
[4] Holttinen H. Hourly wind power variations in the Nordic countries. Wind
Energy 2005;8(2):73195.
[5] Holttinen H. Design and operation of power systems with large amounts of
wind power. State-of-the-art report. International Energy Agency, IEA WIND
Task, 25; 2007.
[6] Lund H. Large-scale integration of wind power into different energy systems.
Energy 2005;30(13):240212.
[7] Moura P, de Almeida A. Methodologies and technologies for the integrations of
renewable resources in Portugal. Renewable Energy World Europe 2009,
Cologne, Germany; May 2009.
[8] Kushler M et al. Using energy efficiency to help address electric systems
reliability. Energy 2003;28(4):30317.
[9] IEA. Demand response in liberalised electricity markets. International Energy
Agency; 2003.
[10] EIA. Glossary, US Energy Information Administration, Independent Statistics
and Analysis; 2010. <http://www.eia.doe.gov/glossary/glossary_d.htm>