-
468 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 2, NO. 4,
OCTOBER 2011
Combined Operations of Renewable Energy Systemsand Responsive
Demand in a Smart GridCarlo Cecati, Fellow, IEEE, Costantino Citro,
and Pierluigi Siano, Member, IEEE
AbstractThe integration of renewable energy systems (RESs)in
smart grids (SGs) is a challenging task, mainly due to the
in-termittent and unpredictable nature of the sources, typically
windor sun. Another issue concerns the way to support the
consumersparticipation in the electricity market aiming at
minimizing thecosts of the global energy consumption. This paper
proposes anenergy management system (EMS) aiming at optimizing the
SGsoperation. The EMS behaves as a sort of aggregator of
distributedenergy resources allowing the SG to participate in the
openmarket.By integrating demand side management (DSM) and active
man-agement schemes (AMS), it allows a better exploitation of
renew-able energy sources and a reduction of the customers energy
con-sumption costs with both economic and environmental benefits.It
can also improve the grid resilience and flexibility through
theactive participation of distribution system operators (DSOs)
andelectricity supply/demand that, according to their preferences
andcosts, respond to real-time price signals using market
processes.The efficiency of the proposed EMS is verified on a
23-bus 11-kVdistribution network.
Index TermsActive management, demand side management(DSM),
energy management systems (EMSs), smart grid (SG),wind
turbines.
I. INTRODUCTION
T ODAY, the integration of large amounts of renewableenergy
systems (RESs) with the grid [1][6] is widelystudied by many
researchers, but only few of them addressthese problems in
connection with a consumers potentialparticipation to the
electricity market [7][9], or analyze theadditional balancing costs
due to intermittent and partiallypredictable availability of RESs
[10][12]. On the other hand,continuous changes of power system
generation capacityimpose significant energy reserves, imported
energy, and theuse of efficient storage systems [13][15], thus
higher costs.Usually, stabilization of the available power is based
on au-tomatic resources such as primary and secondary frequency
Manuscript received November 19, 2010; revised May 24, 2011;
acceptedJune 26, 2011. Date of publication July 14, 2011; date of
current versionSeptember 21, 2011.C. Cecati is with the Department
of Industrial and Information Engineering,
and Economics, University of LAquila, and DigiPower Ltd.,
LAquila 67100,Italy (e-mail: [email protected]).C. Citro is
with the Department of Electrical Engineering, Poly-
technic University of Catalunya (UPC), Barcelona, 08222, Spain
(e-mail:[email protected]).P. Siano is with the Department
of Industrial Engineering, University of
Salerno, Fisciano (SA), 84084, Italy (e-mail:
[email protected]).Color versions of one or more of the figures in
this paper are available online
at http://ieeexplore.ieee.org.Digital Object Identifier
10.1109/TSTE.2011.2161624
control devices, reacting within seconds up to minutes, and
fastmanual resources (spinning and nonspinning reserves),
usuallyprovided by diesel generators, responding within 1015
min.Generation and load forecast systems can provide adequate
solutions to face these problems even if they usually are
affectedby errors requiring suitable regulation capabilities.
Predictionerrors can be strongly reduced if wind-forecasting errors
are in-dependent of those on the demand forecasting [16], and
shortforecast lead time can generally ease the need for standby
bal-ancing resources [17].One further element that could reduce
balancing require-
ments is the flexibility of load demand which can be obtained
byissuing price-based signals, and allowing customers to
decreasethe energy demand according to their real-time
availability[17], [18].Demand side management (DSM) includes
mechanisms of
both price responsive demand and demand response programs[19].
The first one refers to those changes applied by consumersto their
electric load profile in response to energy market pricesignals for
improving the economic efficiency of their energyconsumption. This
mechanism increases the economic effec-tiveness of electricity
markets by encouraging the energy loaddemand when the real-time
price is low and discouraging itwhen the price is high. As a
consequence, the peak demandcan be decreased and the additional
generation and transmis-sion infrastructures may be avoided or
reduced [20] and neweco-friendly standard of living encouraged
[21].Demand response, instead, is defined as the customers
ability
to alter their own electricity demand in response to signals
fore-casted by the system when reliability is put at risk.
Essentially,it refers to curtailment service programs actualized by
payingend-users to take their electrical load off the grid when it
is defi-cient in capacity or operating reserves. There are many
differentpotential balancing resources, for instance the management
ofspace heatings, air-conditioners, refrigerators,
washing/dryingmachines, electric vehicles, etc. [17]. Thousands of
such poten-tial balancing loads can quickly provide (within seconds
up toone minute) stable and predictable response without any
earlywarning of curtailment. However, a common characteristic
ofsuch a kind of load storage is that it is limited in duration as
cus-tomers may not accept a sustained outage period of
discomfort,considering that the value of lost load is always a very
impor-tant issue [22]. Whereas real-time pricing options are
alreadyavailable for large industrial and commercial consumers
[23],such schemes have limited implementations for domestic
cus-tomers [24][26], where not all the types of loads are able
toparticipate in responsive demand programs. Some researchespointed
out that active control of consumer loads could enable
1949-3029/$26.00 2011 IEEE
-
CECATI et al.: COMBINED OPERATIONS OF RESs AND RESPONSIVE DEMAND
IN AN SG 469
additional on-shore wind farms [27]. In [28], it has been
demon-strated that fast/emergency reserve can be provided by
respon-sive loads such as residential and small commercial
air-condi-tioners; in [29] the control of residential heaters and
pumps havebeen applied for managing daily peak demands. In [17], it
hasbeen reported that the value of the implementation of
real-timepricing in the U.K. would be at least 2.6 to 3.6 billion,
dueto peak loads reductions during low wind speed, thus
justifyingthe expense of installing and operating smart
meters.These new mechanisms require active management schemes
(AMS) as well as end-user-level complex communicationsystems,
necessary for making available information onreal-time-pricing and
availability of the electrical energy. Dueto the previous
considerations, this paper proposes an energymanagement system
(EMS) for smart grid (SG) managementthrough DSM and AMS [30]. In
the following, Section IIdescribes the EMS and a scheme for the
active control of anSG, and Sections III and IV present and analyze
the proposedmethod and different case studies, respectively.
Conclusionsare drawn in Section V.
II. EMS FOR SGsAs known, the term SG refers to a fully automated
electric
power system controlling and optimizing the operation of allits
interconnected elements, in order to guarantee safe and effi-cient
operations of energy generation, transmission, and distri-bution
[31], [32]. Today, many interesting examples of SGs areavailable in
many countries, including, for instance, the U.S.,Canada, Germany,
Japan, India, and Australia [33], [34]. Micro-grids (MGs) are
small-scenario versions of the centralized elec-tricity systems
that locally generate, distribute, and regulate theflow of electric
energy to consumers. They are connected to thebulk power grid and
allow higher reliability and energy cost re-ductions by encouraging
the end consumers to locally purchasegenerated electric power with
privileged tariffs [32].Further initiatives towards the future SGs
are concerned with
the so-called virtual power plants (VPPs), i.e., aggregationsof
interconnected distributed generations (DGs) located indifferent
places but managed in order to work as an uniquevirtual power plant
managing a well defined amount of energy.This solution allows even
the smallest DGs (aggregated inthe VPP) to access the electricity
market and contribute to theenergy cost reduction process [35].
Examples of VPPs can befound in Germany, Australia, and the U.S.
[36], [37].Regardless of the possible different implementations,
inno-
vative EMSs are required to achieve a dynamic control of
thedifferent interconnected elements. A possible scenario for
theimplementation of this infrastructure is shown in Fig. 1.
Themain elements of this system are: Energy management system
(EMS); Supervisory control and data acquisition (SCADA); Remote
terminal units (RTUs); Advanced metering infrastructure (AMI);
State estimation algorithms (SEAs); Generation and load forecast
system (GLFS).Optimization, monitoring, and control of the SG
per-
formances are entrusted to a suite of
hardware/softwareapplications constituting the EMS [38]. The SCADA
system
Fig. 1. EMS in the SG infrastructure.
transmits the measurement data, provided by an AMI and by aset
of remote collecting data devices (RTUs) placed in
strategicpositions along the SG, to the EMS. The latter determines
theactions required for the optimum state of the SG by using
SEAsand a GLFS.
III. METHOD DESCRIPTION
A. EMS PolicyAccording to the EUDirective EC 2006/32 on energy
end-use
efficiency and energy services [39], a mechanism of
real-timepricing (RTP) tariff should be offered to the market. In
thisstudy, the hourly spot market price is assumed as the
real-timeelectricity price for consumers available one day in
advance,as adopted by Denmark and Ireland in Europe [26], [40].
Evenif fluctuations between predicted and actual prices occur,
thiserror usually goes to zero [41]. In order to reduce the
elec-tricity costs, those consumers with demand regulation
capabilitycan reschedule their bids according to the real-time
electricityprice. The scenario of Ireland Single Electricity Market
(SEM)[41] demonstrated that DSM, optimized on one-day-ahead
pre-dicted electricity prices may promote the use of wind
generatedelectricity. Moreover, variable service subscription
(VSS)-typeprograms are assumed for customers that, under demand
lim-iting and demand subscription service, subscribe to a
demandthreshold. The solution is a centrally controlled limiting
loaddevice: when the generation capacity is insufficient or due
toreliability requirements, the EMS can limit the demand to
thetotal subscribed capacity and responsive loads are paid by
thedistribution system operator (DSO) according to the VSS [42].The
EMS behaves as a sort of aggregator of distributed energy
-
470 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 2, NO. 4,
OCTOBER 2011
resources [43], that allows the SG (or MG) participating in
theopen market, buying and selling active and reactive power tothe
bulk grid and optimizing the local (renewable)
productioncapabilities. It takes into account the bids received by
energyproducers and consumers.When buying active and reactive power
from the grid, the
EMS tries to maximize the benefit function of demand
whileminimizing the costs of energy and the costs paid to
consumersfor demand limiting. When selling active and reactive
powerto the bulk power grid, due to an excess of low price
renew-able generation, the EMS also tries to maximize revenues
byexchanging power with the grid. The SG (or MG) can also re-lieve
possible network congestions by transferring energy to thenearby
feeders of the distribution network [44].In other words, the
complementary operations executed by
the EMS are:1) A one day-ahead schedule of distributed
generators andresponsive loads according to the market prices, with
eachtrade day comprising 48 half hourly trading periods.
Alldispatchable generators and responsive loads bid the
oneday-ahead active and reactive power generation or loaddemand by
providing price and quantity information foreach trading period one
day ahead. For each trading periodthe dispatch schedules are
determined [45].
2) A real-time intraday optimization operation that
everyminutes, e.g., 5 min, modifies the scheduling in order
toconsider the operation and economic requirements.
As both price and reliability demand response (e.g.,
ancillaryservice) are considered, the scheduling is modified
according toboth the real-time electricity price and the support
offered bydistributed generators and responsive loads to the active
net-work operation.
B. Mathematical Problem Formulation
During each time interval, the objective function to be
maxi-mized is the sum of the total demand benefits, minus the sum
ofthe total generation costs and the costs paid for load
curtailingunder VSS [46]
(1)
where is the vector of dependent variables, containing the
am-plitudes and angles of the buses voltages; is the vector of
con-trol variables, including the secondary voltage of the
on-loadtap-changers (OLTCs) and the active and reactive power
in-jected or absorbed by generators and loads; is the set of
poolload buses; is the set of pool generator buses; is theset of
responsive loads; is the cost for curtailing 1 MWh ofthe th
responsive load under VSS; is the curtailed en-ergy for the th
responsive load; is the demandvector; is thebenefit of consumer ;
is the supply vector; and
is the cost of supplier. Subscript and subscript specify a
relationship with activeor reactive power, respectively. In the
pool model, productioncosts and benefit functions are quadratic
functions of active andreactive power of pool loads and generators,
as follows:
(2)(3)(4)(5)
The price-dependent load is modeled with a consumer
benefitfunction , concave and increasing, with including boththe
real and reactive power demand [46].In order to integrate the
simulation of reactive power ex-
change, price-dependent reactive loads are considered. Since
re-active power acts more as a service enabling the consumption
ofreal power, a benefit function different from the real power
ben-efit equation is determined. Accordingly, the benefit of the
re-active power is considered as the avoidance of its shifting
froma given desired level for a specified active power
consumption.Desired reactive power demand is that required by the
load atthe given load level and can be defined as a function of the
realpower demand [46] . Assuming themagnitudeof the function
increasing with as and con-sidering a concave function for as(5) is
obtained.In order to maximize the objective function, the
nonlinear
programming formulation of the OPF, described in [47][51],is
modified including the AMS and DSM.
C. Discrete Variables Handling
The OPF can be approached as a mixed
discrete-continuousoptimization nonlinear problem with a single
integer variable:the OLTC transformer tap. The solution of this
problem is im-plemented by a two-stage approach [52], [53]. First,
a solutionover the full range of variables is generated while
assuming thatall variables are continuous; then, the discrete
variable is movedto the nearest discrete setting, and treated as
constant during asecond-stage solution.
D. Constraints
The equality constraints represent the static loadflow equations
such as Kirchhoff current law , whereis the set of busses (indexed
by ) and Kirchhoff voltage law
, where is the set of lines (indexed by ) [51], [52].
Theinequality constraints are listed in the following:1) Active and
reactive power constraints for the interconnec-tion to external
network (slack bus)
(6)
where is the set of external sources (indexed by ),and are the
active and reactive power outputs of ,respectively, and and are the
min/maxvalues they can assume.
-
CECATI et al.: COMBINED OPERATIONS OF RESs AND RESPONSIVE DEMAND
IN AN SG 471
2) Active and reactive power constraints for generators:
(7)
where and are the active and reactive power out-puts of ,
respectively, and and are themin/max values they can assume.
3) Active and reactive power constraints for consumers load,
(8)
where and are the active and reactive powerabsorbed by consumer
, respectively, and and
are the min/max values it can assume.4) Voltage level
constraints
(9)
where is the voltage at , and are the max/minvalues it can
assume.
5) Flow constraints for lines and transformers
(10)
where and represent the active and reactive powerinjections onto
, respectively, and the maximum powerflow on .
The additional constraints derived from the AMS are relatedto
the coordinated OLTC voltage, theWTs and diesel generators(DGens)
power factor angles.1) Coordinated OLTC voltage constraint
(11)
where is the secondary voltage of the OLTC,are the max/min
values it can assume.
2) Coordinated generator reactive power constraints,
(12)
where is the power factor angle of , are themax/min values it
can assume.
IV. CASE STUDIES
The proposed technique is applied to a 23-bus 11-kV
radialdistribution system, shown in Fig. 2. The three feeders are
sup-plied by a 6-MVA 33/11-kV transformer; the tap position
allowsnine different voltages with a step p.u. Voltagelimits are
taken to be 10% of the nominal value and feederthermal limits are
1.5 MVA (81 A/phase). The phasor dynamic
Fig. 2. Test network.
TABLE INETWORK LOADING
TABLE IIWTS GENERATED ACTIVE AND REACTIVE POWER
models for the WTs, the DGens, the OLTC and the other
dis-tribution system elements are implemented using Matlab
Sim-PowerSystems.The load at each bus is assumed to track a load
curve [30]:
discrete load bands across one year are considered:
maximum,normal, medium, and minimum load. The load levels for
eachband are summarized in Table I.In the test network, two wind
turbines (WT1 and WT2) are
connected at nodes 7 and 16, respectively. Each WT
generatesabout 1.05 MW at a wind speed of 12 m/s, operating within
apower factor varying between 0.85 leading and lagging [54].The
power extracted from a WT is a function of the availablewind power,
the power curve of the machine and the ability ofthe machine to
react to wind variations. The WTs generated ac-tive and reactive
power dependence on the wind speed is givenin Table II.A high cost
DGen generating a maximum active power of
600 kW is connected at bus 9. The cost curve used for theDGen is
approximated by a second-order polynomial function[55], considering
the diesel generator starting cost assessed at15 euros [56]. The
values of the cost coefficients are calculatedconsidering the fuel
consumption curve of a real diesel gener-ator obtained from the
data provided in [57], setting a fuel priceof 1 euro per liter
[58]. The diesel unit can operate between 25%and 100% of its rate
capacity. As regards with the consumers, it
-
472 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 2, NO. 4,
OCTOBER 2011
TABLE IIISUPPLIERS CHARACTERISTICS
TABLE IVCUSTOMER CHARACTERISTICS SCENARIO 1
TABLE VCUSTOMER CHARACTERISTICS SCENARIO 2
has been assumed that each consumer has both fixed and
priceresponsive load.Operation of the considered EMS endowed with
AMS is first
evaluated considering discrete load and wind speed states
[30],varying in the range between minimum and maximum load
andbetween 0 and 12 m/s, respectively.The following analysis
considers different DSMmechanisms
such as Price Responsive Demand and Demand Response Pro-grams
[19] and aims at evaluating the benefits of real-time elec-tricity
price. Two different scenarios are analyzed as follows:1) consumers
are involved in a demand response program;2) consumers are involved
in a demand response program andalso participate as price
responsive demand.
Suppliers and customers coefficients are given in Tables III,IV,
and V, respectively [59], [60].In both minimum and maximum load
scenarios, the suppliers
and customer characteristics are equal, thus the supplied
loadand the total cost paid for energy delivering change only
fornormal and maximum loads.
A. Scenario 1: Consumers Involved in a Demand ResponseProgramIn
this scenario, consumers can be limited only for relia-
bility requirements (i.e., for avoiding constraints violations)
by
TABLE VITOTAL ACTIVE POWER ABSORBED BY DEMAND AT BUSES 3, 12,
AND 17 [kW]
TABLE VIISUM OF THE TOTAL NETWORK DEMAND [kW]
means of a centrally controlled limiting load device.
When,during maximum and normal load, demand is limited to thetotal
subscribed capacity, consumers are paid by the DSO at200 euros/MWh
according to a VSS [43].During normal operation each load is
supplied at its max-
imum value for a wind speed equal to 12 m/s or when the loadis
minimum. When wind speed varies between 0 and 10 m/s,all consumers
are supplied at their desired demand level, ex-cept those connected
at buses 3, 12, and 17, which, are, instead,limited in order to
satisfy reliability requirements, as shown inTable VI. Since
variable loads operate at fixed power factor,the absorbed reactive
power exhibits a similar trend. Due to theoverall load increase,
the active power absorbed by consumersconnected at buses 3, 12, and
17 is limited by the thermal con-straints on the wires 01 and 012.
The sum of total networkdemand is shown in Table VII.The percentage
peak demand reduction is shown in Fig. 3 for
wind speeds below or equal to 6 m/s. During maximum load,the
total demand at buses 3, 12, and 17 decreases from 1706 to1032 kW
(or by 39%) and to 829 kW (or by 51%), when thewind speed is 6 and
0 m/s, respectively.For instance, in case of maximum load and wind
speed
varying from 0 to 8 m/s, it is worth noting that: the total
curtailed power decreases proportionally to thewind speed from 878
to 139 kW;
the DGen always generates its maximum active power of600 kW,
except for a wind speed of 8 m/s, when it gener-ates 320 kW.
The DGen supplies active power only during maximum andnormal
load states and wind speeds varying from 0 up to 8 m/s.
-
CECATI et al.: COMBINED OPERATIONS OF RESs AND RESPONSIVE DEMAND
IN AN SG 473
Fig. 3. Percentage peak demand reduction in Scenario 1.
Its output power varies from a minimum value around 293 kW,in
case of normal load and wind speed of 6 m/s, to a maximumvalue of
600 kW, in coincidence with maximum load and windspeeds lower than
8 m/s.The active and reactive power imported from the grid tend
to
increase while decreasing the wind speed and with an
increasingload value. While this trend is always verified in case
of reactivepower, the relationship between the tendencies of the
imported/exported active power in relationshipwith the load value
is morecomplex as it depends on the active power generated by
theWTsand the DGen and on the active power absorbed by
variableloads. When the wind speed is equal to 12 m/s and the load
isminimum, about 285 kW of active power are exported to thebulk
power grid. Due to the implemented AMS and, in orderto relieve the
voltage constraints, the WTs always supply lead(capacitive)
power.
B. Scenario 2: Consumers Involved in a Demand ResponseProgram
also Participating as Price Responsive DemandAs in the previous
scenario, consumers can be limited due
to reliability constraints, moreover price responsive
consumersunder RTP tariff (at buses 3, 12, and 17) can modify their
de-mand in response to high real-time electricity prices
occurringduring normal and maximum load.Real-time electricity price
signals, available to both con-
sumers and producers, represent an effective
coordinationmechanisms suitable to drive both to change their
bids/offersin both constrained and unconstrained feeder
conditions.While for wind speeds higher or equal to 8 m/s, the
impact
of price responsive demand and RTP tariff is negligible dueto
the low-price wind energy, consumers under RTP tariff ad-just their
demand bids in response to high real-time electricityprice for wind
speeds below 8 m/s. In the case of maximumload, RTP tariff induces
consumers to move consumption awayfrom costly peak hours and the
total demand decreases fromabout 1706 kW to about 678 kW ( 60%) and
to about 341 kW( 80%) in correspondence of a wind speed of 6 and 0
m/s, re-spectively, as shown in Fig. 4.RTP also allows reducing the
power generated by the DGen,
that is equal to about 240 kW during maximum load.
Fig. 4. Percentage peak demand reduction in Scenario 2.
TABLE VIIITOTAL COSTS [Euro/h]
A reduction of the peak demand can be evidenced in Scenario2
when compared to Scenario 1: for wind speeds below 8 m/s,the
percentage decrease of the peak demand is within 21%37%and 34%60%,
during normal and maximum load, respectively.This reduction leads
to significant economic benefits for bothconsumers and DSO, that
avoids paying consumers according toa VSS at 200 euros/MWh for
demand limiting, as in Scenario 1.A maximum cost reduction, varying
from 301 to 28 euros/h,
can be achieved during maximum load in Scenario 2 when com-pared
with Scenario 1. The total cost reductions, if compared tothe
previous Scenario, are shown in Table VIII.Moreover, RTP encourage
consumers shifting consumptions
during periods of high wind energy production and supportingthe
use of renewable energy resources.
C. Base-Year AnalysisThe benefits of the AMS and price
responsive consumers
under RTP tariff during one year are assessed following
theapproach used in [61]. Based on their joint probability
ofoccurrence, defining the number of coincident hours over theyear,
wind availability and demand have been aggregated intoa number of
wind/demand scenarios. Actual data for bothdemand and wind
production have been taken from [61].The set of scenarios obtained
by combining wind availability
and load demand real data for one year are shown in Fig. 5.
-
474 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 2, NO. 4,
OCTOBER 2011
Fig. 5. Coincident hours for demand/generation scenarios.
Fig. 6. Total costs without AMS.
Each scenario represents the combination between wind speedand
load demand values, indicated in percentage terms (x-axis),and is
characterized by a defined number of hours over theyear (y-axis).
Such a number represents the time (number ofhours) during which
each combination wind/demand occurs inthe course of the year.As
shown in Figs. 6 and 7, by summing the costs of each
scenario, the proposed EMS endowed with AMS allows a totalannual
cost reduction around 383 keuros, if compared to thescenario
without AMS. The annual curtailed energy is equalto about 634
MWh/year without considering the AMS, while itdecreases down to
about 573 MWh/year when using the AMS,with a reduction of about 9%
of the curtailed energy.AMS, such as the coordinated voltage
regulation of OLTC
and the power factor control ofWTs, are able, in fact, to
increasethe total energy absorbed by loads. For instance, when the
loaddemand is within 70% and 100%, the power factor control ofWTs
can increase the energy absorbed by loads up to 10% ifcompared with
the scenario in which only a regulation of the
Fig. 7. Total costs with AMS.
OLTC is applied, and up to 20% if compared with the
scenariowithout AMS.Combining AMS with price responsive consumers
under
RTP tariff (at buses 3, 12, and 17) an additional 93
keurosannual cost reduction over the scenario with the sole AMS
isachieved.It is worth noting that, by mitigating network
constraints, the
sole use of AMS may increase the energy absorbed by loads upto
20%.Moreover, real-time electricity price signals drive
consumers
to find a different time schedule of their consumptions, thus
re-ducing expensive peak power demands further contributing
tonetwork constraints reduction.Hence, a combination of both
mechanisms through the active
participation of producers and consumers, represents a good
op-tion for improving both resilience and flexibility of SGs and
forsupporting the use of renewable energy resources.
D. Computational Performances EvaluationSimulation results
demonstrate that the proposed method is
fast enough to be executed in real-time: for the considered
net-work, a personal computer with an Intel CoreTM i7
processorrunning at 2.67 GHz and with 8-GByte RAM requires less
than3 min for the solution of a single OPF.The proposed
optimization approach (i.e., Sequential
Quadratic Programming), requires low computational re-sources
while providing very good results, comparable withperformances
obtained using interior point method solution ofOPF relaxation
[53]. It is worth noting that this method, alsocoded in the AIMMS
optimization modeling environment [30],is scalable, i.e., it can be
used with a larger number of controlvariables.
V. DISCUSSION AND CONCLUSIONIn this paper, an EMS for the
optimization of SGs has been
proposed.The EMS behaves as a sort of aggregator of distributed
energy
resources allowing the SG participating in the open market
in
-
CECATI et al.: COMBINED OPERATIONS OF RESs AND RESPONSIVE DEMAND
IN AN SG 475
order to optimize the local production capabilities as well as
tominimize the cost of bought energy.The proposed system integrates
AMS with DSM without re-
quiring significant additional hardware.Simulation results
evidenced that the combined operations
of RES and Price Responsive Demand mitigate network con-straints
while satisfying higher demand levels and reducing theenergy
costs.AMS offer technical benefits: they allow a better
coordina-
tion between DSOs and electricity supply and demand that,
sat-isfying their preferences at minimum costs, can respond to
real-time price signals using market processes.It is worth pointing
out that each active or DSM solution, or
the combination of them, should be evaluated on a
case-by-casebasis as the implementation and cost-effectiveness of
each solu-tion depends on network characteristics. A combination of
bothmechanisms will, however, represent in most cases, the best
op-tion to improve the SGs resilience and flexibility through
re-source use optimization and peak loads reduction.The
implementation of AMS and DSM requires both a hard-
ware as well as a software infrastructure that are expected to
be-come standard in SGs. Conversely, the actual implementationof
AMS and DSM also requires a new regulatory frameworkbased on
economic signals and providing incentives for bothconsumers and
generator owners and special bilateral contractsbetween them and
the DNOs.
REFERENCES[1] R. Sioshansi andW. Short, Evaluating the impacts
of real-time pricing
on the usage of wind generation, IEEE Trans. Power Syst., vol.
24, no.2, pp. 516524, May 2009.
[2] R. Sioshansi, Evaluating the impacts of real-time pricing on
the costand value of wind generation, IEEE Trans. Power Syst., vol.
25, no.2, pp. 741748, May 2010.
[3] V. Vittal, The impact of renewable resources on the
performance andreliability of the electricity grid, Bridge, vol.
40, no. 1, pp. 512, 2010.
[4] S. M. Amin and B. F. Wollenberg, Toward a smart grid: Power
de-livery for the 21st century, IEEE Power Energy Mag., vol. 3, no.
5,pp. 3441, Sep./Oct. 2005.
[5] A. Vojdani, Smart integration, IEEE Power Energy Mag., vol.
6, no.6, pp. 7179, Nov./Dec. 2008.
[6] A. Ipakchi and F. Albuyeh, Grid of the future, IEEE Power
EnergyMag., vol. 7, no. 2, pp. 5262, Mar./Apr. 2009.
[7] M. Carrin, A. B. Philpott, A. J. Conejo, and J. M. Arroyo, A
sto-chastic programming approach to electric energy procurement for
largeconsumers, IEEE Trans. Power Syst., vol. 22, no. 2, pp.
744754,May2007.
[8] E. Gmez-Villalva and A. Ramos, Optimal energy management of
anindustrial consumer in liberalized markets, IEEE Trans. Power
Syst.,vol. 18, no. 2, pp. 716723, May 2003.
[9] A. B. Philpott and E. Pettersen, Optimizing demand-side bids
in day-ahead electricity markets, IEEE Trans. Power Syst., vol. 21,
no. 2, pp.488498, May 2006.
[10] T. Senjyu, H. Takara, K. Uezato, and T. Funabashi,
One-hour-aheadload forecasting using neural network, IEEE Trans.
Power Syst., vol.17, no. 1, pp. 113118, Feb. 2002.
[11] H. Bludszuweit, J. A. Domnguez-Navarro, and A. Llombart,
Statis-tical analysis of wind power forecast error, IEEE Trans.
Power Syst.,vol. 23, no. 3, pp. 983991, Aug. 2008.
[12] A. Fabbri, T. G. S. Romn, J. R. Abbad, and V. H. M.
Quezada, As-sessment of the cost associated with wind generation
prediction errorsin a liberalized electricity market, IEEE Trans.
Power Syst., vol. 20,no. 3, pp. 14401446, Aug. 2005.
[13] G. Strbac, A. Shakoor,M. Black, D. Pudjianto, and T. Bopp,
Impact ofwind generation on the operation and development of theUK
electricitysystems, Elect. Power Syst. Res., vol. 77, no. 9, pp.
12141227, 2007.
[14] M. Black and G. Strbac, Value of bulk energy storage for
managingwind power fluctuations, IEEE Trans. Energy Convers., vol.
22, no.1, pp. 197205, Mar. 2007.
[15] H. H. Zeineldin, T. H. M. El Fouly, E. F. El Saadany, and
M. M. A.Salama, Impact of wind farm integration on electricity
market prices,IET Renew. Power Gener., vol. 3, no. 1, pp. 8495,
2009.
[16] A. Papavasiliou and S. S. Oren, Coupling wind generators
with de-ferrable loads, in Proc. IEEE Energy 2030 Conf., Atlanta,
GA, 2008,pp. 17.
[17] A. J. Roscoe and G. Ault, Supporting high penetrations of
renewablegeneration via implementation of real-time electricity
pricing and de-mand response, IET Renew. Power Gener., vol. 4, no.
4, pp. 369382,2010.
[18] G. Strbac, Demand side management: Benefits and challenges,
En-ergy Policy, vol. 36, no. 12, pp. 44194426, 2008.
[19] V. Thornley, R. Kemsley, C. Barbier, and G. Nicholson, User
per-ception of demand side management, in Proc. CIRED Seminar
2008:SmartGrids for Distribution, Jun. 2008, pp. 14.
[20] H. Chao, Price responsive demand management for a smart
gridworld, Elsevier, The Electricity J., vol. 23, no. 1, pp. 720,
2010.
[21] H. Nilsson, The many faces of demand-side
management,IETPower Eng. J., vol. 8, no. 5, pp. 207210, 1994.
[22] J. N. Sheen, C. S. Chen, and T. Y. Wang, Response of large
industrialcustomers to electricity pricing by voluntary time of use
in Taiwan,IEE Proc. Gener., Transmiss. Distrib., vol. 142, no. 2,
pp. 157166,1994.
[23] C. Aubin, D. Fougere, E. Husson, and M. Ivaldi, Real-time
pricing ofelectricity for residential customers: Econometric
analysis of an exper-iment, J. Appl. Econ., vol. 10, pp. 171191,
1995.
[24] M. Filippini, Swiss residential demand for electricity by
time-of-use,Resour. Energy Econ., vol. 17, no. 3, pp. 281290,
1995.
[25] S. M. Baladi, J. A. Herriges, and T. J. Sweeney,
Residential responseto voluntary time-of-use electricity rates,
Resour. Energy Econ., vol.20, pp. 225244, 1998.
[26] P. Finn, C. Fitzpatrick, M. Leahy, and L. Relihan,
Promotion of windgenerated electricity using price responsive
demand side management:Price prediction analysis for imperfect
energy storage, in Proc. 2010IEEE Int. Symp. Sustainable Systems
and Technology (ISSST), 2010,pp. 15.
[27] V. Hamidi, F. Li, and F. Robinson, Responsive demand in
networkswith high penetration of wind power, in Proc. Transmission
and Dis-tribution Conf. Exposition, T&D (IEEE/PES 2008), 2008,
pp. 17.
[28] B. Kirby, Spinning Reserve From Responsive Loads Oak Ridge
Na-tional Laboratory, ORNL/TM-2003/19, 2003.
[29] G. C. Heffner, C. A. Goldman, and A. M. M. Moezzi,
Innovative ap-proaches to verifying demand response of water heater
load control,IEEE Trans. Power Del., vol. 21, no. 1, pp. 388397,
2006.
[30] P. Siano, P. Chen, Z. Chen, and A. Piccolo, Evaluating
maximumwind energy exploitation in active distribution networks,
IET Gener.Transm. Distrib., vol. 4, no. 5, pp. 598608, 2010.
[31] Smart Grids European Technology Platform SmartGrids
[Online].Available: http://www.smartgrids.eu
[32] S. Ergeg, Smart Metering With a Focus on Electricity
RegulationRef_R07-RMF-04-03, 2007.
[33] S. Morozumi, Micro-grid demonstration projects in Japan, in
Proc.Power Conversion Conf. (Nagoya, PCC07), 2007, pp. 635642.
[34] S. Balijepalli, V. S. K. M. Khaparde, and S. A. Dobariya,
DeploymentofMicroGrids in India, in Proc. IEEE Power and Energy
Society Gen-eral Meeting, 2010, pp. 17.
[35] Siemens and RWE Energy, First Virtual Power Plant Oper-ated
by Siemens and RWE Energy 2008 [Online].
Available:http://w1.siemens.com/press/en/pressrelease
[36] FENIX [Online]. Available: http://www.fenix-project.org[37]
LichtBlick [Online]. Available: http://www.lichtblick.de[38] H.
Muller, A. Rudolf, and G. Aumayr, Studies of distributed energy
supply systems using an innovative energy management system,
inProc. Power Industry Computer Applications, 22nd IEEE Power
En-gineering Society Int. Conf. Innovative Computing for
PowerElectricEnergy Meets the Market (PICA 2001), 2001, pp.
8790.
[39] Directive 2006/32/EC of the European Parliament and of the
Councilon Energy End-Use Efficiency and Energy Services and
RepealingCouncil Directive 93/76/EEC Apr. 5th, 2006 [Online].
Available:http://www.energy.eu/directives/l_11420060427en00640085.pdf
[40] H. Weihao, Z. Chen, and B. Bak-Jensen, Impact of optimal
load re-sponse to real-time electricity price on power system
constraints inDenmark, in Proc. 45th UPEC, 2010, pp. 16.
-
476 IEEE TRANSACTIONS ON SUSTAINABLE ENERGY, VOL. 2, NO. 4,
OCTOBER 2011
[41] A. P. Finn, C. Fitzpatrick, and M. Leahy, Increased
penetration ofwind generated electricity using real time pricing
& demand side man-agement, in Proc. IEEE Int. Symp. Sustainable
Systems and Tech-nology (ISSST09), 2009, pp. 16.
[42] G. L. Doorman, Capacity subscription: Solving the peak
demand chal-lenge in electricity markets, IEEE Trans. Power Syst.,
vol. 20, no. 1,pp. 239245, Feb. 2005.
[43] F. Pilo, G. Pisano, and G. G. Soma, Advanced DMS to manage
activedistribution networks, in Proc. IEEE Power Tech 2009,
Bucharest,Romania, 2009, pp. 18.
[44] A. G. Tsikalakis and N. D. Hatziargyriou, Centralized
control for op-timizing microgrids operation, IEEE Trans. Energy
Convers., vol. 23,no. 1, pp. 241248, Mar. 2008.
[45] E. Kee, Margadh Aibhleise na hEireann: A new electricity
market forIreland, Electricity J., vol. 17, no. 1, pp. 5160,
2004.
[46] J. D. Webera, T. J. Overbye, and C. L. De Marco, Modeling
the con-sumer benefit in the optimal power flow, Decision Support
Syst., vol.24, no. 34, pp. 279296, 1999.
[47] G. P. Harrison, A. Piccolo, P. Siano, and A. R. Wallace,
Exploring thetradeoffs between incentives for distributed
generation developers andDNOs, IEEE Trans. Power Syst., vol. 22,
no. 2, pp. 821828, May2007.
[48] G. P. Harrison, A. Piccolo, P. Siano, and A. R. Wallace,
Hybrid GAand OPF evaluation of network capacity for distributed
generation con-nections, Electrical Power Syst. Res., vol. 78, no.
3, pp. 392398,2008.
[49] A. Piccolo and P. Siano, Evaluating the impact of network
invest-ment deferral on distributed generation expansion,
IEEETrans. PowerSyst., vol. 24, no. 3, pp. 15591567, Aug. 2009.
[50] P. Siano, L. F. Ochoa, G. P. Harrison, and A. Piccolo,
Assessing thestrategic benefits of distributed generation ownership
for DNOs, IETGener. Trans. Distrib., vol. 3, no. 3, pp. 225236,
2009.
[51] H. Liu, Y. Shen, and Z. B. Zabinsky et al., Social welfare
maximiza-tion in transmission enhancement considering network
congestion,IEEE Trans. Power Syst., vol. 23, no. 3, pp. 11051114,
Aug. 2008.
[52] W. F. Tinney, J. M. Bright, K. D. Demaree, and B. A.
Hughes, Somedeficiencies in optimal power flow, IEEE Trans. Power
Syst., vol. 3,no. 2, pp. 676683, May 1988.
[53] F. Capitanescu and L. Wehenkel, A new heuristic approach to
dealwith discrete variables in optimal power flow computations, in
Proc.IEEE Bucharest Power Tech Conf., Bucharest, Romania, Jun.
28Jul.2 2009, pp. 16.
[54] P. Chen, P. Siano, B. Bak-Jensen, and Z. Chen, Stochastic
optimiza-tion of wind turbine power factor using stochastic model
of windpower, IEEE Trans. Sustainable Energy, vol. 1, no. 1, pp.
1929,Apr. 2010.
[55] C.-L. Chen, S.-C. Hsieh, T.-Y. Lee, and C.-L. Lu, Optimal
integrationof wind farms to isolated wind-diesel energy system,
Energy Convers.Manage., vol. 49, no. 6, pp. 15061516, 2008.
[56] B. Asato, K. Uchida, T. Goya, T. Senjyu, A. Yona, and N.
Urasaki,Optimal operation for diesel generators in small isolated
island powersystem considering controllable load, in Proc. Int.
Conf. ElectricalMachines and Systems (ICEMS), 2010, pp.
18691872.
[57] C. Cecati, C. Citro, A. Piccolo, and P. Siano, Smart
operation of windturbines and diesel generators according to
economic criteria, IEEETrans. Ind. Electron., vol. 58, no. 10, pp.
45144525, 2011.
[58] R.Wies and R. Johnson, Effects of Village Power Quality on
Fuel Con-sumption and Operating Expenses, Final Report Alaska
Energy Au-thority, 2008, pp. 190.
[59] J. Kennedy, B. Fox, and D. Flynn, Use of electricity price
to matchheat load with wind power generation, in Proc. 1st Int.
Conf. Sustain-able Power Generation and Supply, Nanjing, China,
2009, pp. 16.
[60] L. Xie, J. Joo, and M. D. Ilic, Integration of intermittent
resourceswith price-responsive loads, in Proc. North American Power
Symp.(NAPS), 2009, pp. 16.
[61] L. F. Ochoa, C. J. Dent, and G. P. Harrison, Distribution
networkcapacity assessment: Variable DG and active networks, IEEE
Trans.Power Syst., vol. 25, no. 1, pp. 8795, Feb. 2010.
Carlo Cecati (M90SM03F06) received the Dr. Ing. degree in
elec-trotechnic engineering from the University of LAquila,
LAquila, Italy, in1983.He is currently with the Department of
Industrial and Information Engi-
neering, and Economics, where he is a Professor of Industrial
Electronics andDrives and he is a Rectors Delegate. He is the
Founder and Coordinator of thePh.D. courses on management of
renewable energies and sustainable buildingsat the University of
LAquila. In 2007, he was the Founder of DigiPowerLtd., a spin-off
company dealing with industrial electronics and renewableenergies.
His research and technical interests cover several aspects of
powerelectronics and electrical drives. In those fields he authored
more than 100papers published in international journals and on
conference proceedings.Dr. Cecati is a Coeditor-in-Chief of the
IEEE TRANSACTIONS ON
INDUSTRIAL ELECTRONICS; he has been a Technical Editor of the
IEEE/ASMETRANSACTIONS ON MECHATRONICS. He has been a General
Cochair ofthe IEEE International Symposium on Industrial
Electronics (ISIE) 2002,IEEE-ISIE2004, IEEE-ISIE2008, a Honorary
Cochair of the IEEE-ISIE2010,a Technical Program Cochair of the
IEEE Industrial Electronics Conference(IECON) 2007, and a Track
Cochair or Special Session Chair of IEEE-ISIEand IEEE-IECON
conferences. From 2000 to 2004, he was an AdCom memberof the IEEE
Industrial Electronics Society (IES), and from 2005 to 2006, anIES
Vice President. Since 2007, he has been an IES Senior AdCom
memberand, until 2008 IEEE, IES Region 8 Coordinator. He is a
member of IEEE IESCommittees on Power Electronics and on Renewable
Energy Systems and aCochair of IEEE IES Committee on Smart
Grids.
Costantino Citro received the M.Sc. degree with honors in
electronic engi-neering from the University of Salerno in 2010. In
the same year, he joined theDepartment of Information and
Electrical Engineering, University of Salerno,with a research
fellowship in the fields of power electronics and electrical
powersystems. Since March 2011, he has been working toward the
Ph.D. degree inelectrical engineering at Polytechnic University of
Catalunya (UPC), Spain.His research interests include power
electronic converters and integration of
renewable energy systems in smart grids.
Pierluigi Siano (M09) received the M.Sc. degree in electronic
engineering andthe Ph.D. degree in information and electrical
engineering from the Universityof Salerno, Fisciano, Italy, in 2001
and 2006, respectively.Currently, he is Assistant Professor with
the Department of Industrial Engi-
neering, University of Salerno, Fisciano, Italy. His research
activities are cen-tered on the integration of renewable
distributed generation into electricity net-works and Smart Grids
and on the application of soft computing methodologiesto analysis
and planning of power systems. In these fields, he has
publishedmore than 70 technical papers including 30 international
journal papers and 40international conference papers.Dr. Siano is
Associate Editor of IEEE TRANSACTIONS ON INDUSTRIAL
INFORMATICS, member of the editorial board of the International
Journal onPower System Optimization, Energy and Power Engineering,
Smart Grid andRenewable Energy. He served as reviewer and session
chairman for manyinternational conferences. He has been Special
Sessions Cochair of IEEE-ISIE2010 and Guest Editor of the Special
Sections of the IEEE TRANSACTIONSON INDUSTRIAL ELECTRONICS on
Methods and Systems for Smart GridsOptimization and on Smart
Devices for Renewable Energy Systems. He issecretary of the
Technical Committee on Smart Grids and member of theTechnical
Committee on Renewable Energy Systems of the IEEE IES.