Renewable and Sustainable Energy Reviews 13 (2009) 2169–2174

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: [Link]/locate/rser

A review on the economic dispatch and risk management considering wind power
in the power market
Ren Boqiang, Jiang Chuanwen *
Department of Electrical Engineering, Shanghai Jiaotong University, Huashan Road 1954, Shanghai 200030, PR China

A R T I C L E I N F O A B S T R A C T

Article history: With the rapid development of world economy, wind power has been given more and more
Received 10 November 2008 consideration owing to its energy saving and environmental protection. But due to intermittency and
Accepted 20 January 2009 unpredictability nature of wind power generation, many new problems come into being when infusing
wind power into power network with conventional generators. Aiming at these difficulties, this paper
Keywords: presents a review on the historical research production of this theme. The models of economic dispatch
Wind power
schedule of wind power considering dissimilar actual condition, different optimized algorithms and risk
Economic dispatch model
management in the electric market are discussed and the future trend is prospected in this paper.
Risk management
Optimized algorithm ! 2009 Elsevier Ltd. All rights reserved.
Electric market

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2170
2. Long-term economic dispatch model with wind power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2170
2.1. The objective function involving cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2170
2.2. The bi-objective function incorporating risk level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2171
2.3. The objective function combining different types of wind farms in delegated dispatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2171
2.4. The objective function considering environmental damage cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2171
3. Short-term economic dispatch model with wind power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2171
3.1. The objective function incorporating stochastic security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2171
3.2. The objective function combining energy storage in balancing market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2171
3.3. The objective function considering spinning reserves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2172
3.4. The objective function including start-up/shut-down cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2172
4. Optimized algorithm for dispatch model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2172
4.1. Direct search method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2172
4.2. Particle swarm optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2172
4.3. Simulated annealing approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2172
4.4. Genetic algorithm approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2172
4.5. Scenario construction method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2173
5. Risk dispatch management of wind power in power market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2173
5.1. Management method in power market. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2173
6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2173
Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2174
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2174

* Corresponding author. Tel.: +86 21 34205930; fax: +86 21 54747501.

E-mail address: jiangcw@[Link] (J. Chuanwen).
Abbreviations: HTG, hydro-thermal generation; WTG, wind turbine generation; ED, economic dispatch; CDED, constrained dynamic economic dispatch; WECS, wind energy
conversion system; EES, electrical energy storage; AV, added value; DSM, direct search method; EP, evolutionary programming; SA, simulated annealing; PSO, particle swarm
optimization; MOPSO, multi-objective particle swarm optimization; GA, genetic algorithm; IPP, independent power producer.

1364-0321/$ – see front matter ! 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/[Link].2009.01.013
2170 R. Boqiang, J. Chuanwen / Renewable and Sustainable Energy Reviews 13 (2009) 2169–2174

wind power be used in power network in a large scale. So how to

Nomenclature make reasonable long-term and short-term schedules and manage
the uncertainty risk of wind power in the market including wind
M number of conventional power generators turbine generators is the main thesis of this paper. The economic
N number of wind-powered generators dispatch (ED) of wind power should consider many factors, e.g.
k scenario k environmental damage cost and spinning reserves. Different
FT the expected social cost factors determine different dispatch models and solution. Risk
management of wind power in this paper does not include
Pi active power from the ith conventional generator
volatility of imbalance charge rates and market price which can be
m system security level
obtained or solved beforehand, but the stochastic characteristic
L load value cannot be eliminated at present.
Ci cost function for the ith conventional generator This paper is organized as follows. Sections 2 and 3 present
Ci,w cost function for the ith wind-powered generator multifarious kinds of wind power models based on long-term
Cp,w,i penalty cost function schedule and short-term schedule, respectively. Section 4 intro-
Cr,w,i required reserve cost function duces correlative optimized solution. The dispatch management of
C g( ) fixed running and startup costs of hydro-thermal wind power risk is proposed in detail in Section 5.
generation
Cr( ) scheduling costs of tertiary reserve services during 2. Long-term economic dispatch model with wind power
period
When the ED including wind power considers the uncertainty
wi;av available wind power from the ith wind-powered
nature of the wind power, the cost and reliability will be the main
generator problem. Majority of objectives of optimization is to maximize the
wi scheduled wind power from the ith wind-powered expected benefits or minimize the cost of system satisfying the
generator system constraints at the same time, especially on long-term
wi;r rated wind power from the ith wind-powered schedule.
generator
rtu p generation and voluntary demand-side up spin- 2.1. The objective function involving cost
ning reserves
rtdn generation and voluntary demand-side down For keeping the reliability of the system, the factors for
overestimation and underestimation of operator must be con-
spinning reserves
sidered when solving the ED problem. When available wind energy
gi( ) vector of HTG outputs
is more than what is generated, power could be wasted. So it is
pi( ) probability value
necessary to pay the economic loss of the wind producer because of
Bd( ) demand benefits the surplus energy. In the other condition, the system operator
di( ) forecast sequences underestimate the available wind energy, enough reverses are
li( ) involuntary load shedding needed to make up the scarcity of power. As a result, the cost of
vTi vector of values of lost load reverses should be added to the objective function [1].
ut historical mean wind speed during period In [2], Hetzer et al. creates a new model, which adds wind
st the simulated wind speed during period energy conversion system (WECS) to the ED problem.
vt the simulated wind speed during period
yt the time series value at time t Objective function
Pi,min minimum active power from the ith conventional X
M X
N X
N
FT ¼ C i ðPi Þ þ C w j ðwi Þ þ C p;w j ðwi;av % wi Þ
generator i i i
Pi,max maximum active power from the ith conventional X
N
generator þ C r;w;i ðwi % wi;av Þ (1)
i

Constraint
1. Introduction P i;min & P i & Pi;max (2)

With the rapid development of world economy, people have 0 & wi & wi;r (3)
paid more attention to the consumption of fuel and protection of
environment. As one of the most promising non-pollution X
M X
N
renewable energy sources, the wind power, which is very different pi þ wi ¼ L (4)
from the conventional energy sources, has been given more i i

consideration by policies. Comparing with conventional generator,

the wind turbine generator has the advantage of reducing the
dependence on fossil fuels and the transmission losses, enhancing Where the Weibull probability density function of the
the independence and flexibility of large power grid. stochastic wind speed characteristic is used to transform wind
Because the output of wind power depends on the wind speed distribution to the similar wind power distribution and compute
which is related with climatological and micrometeorological the reverse cost.
parameters, wind behavior is quite distinct from conventional Chen et al. considers the special contractual agreements for
energy operation at their rated capacities. Due to the intermittency buying or selling energy between each independent power
and unpredictability nature of wind power generation, which can producer (IPP) and the public utility [3]. If the system producer
influence the schedule of generation and control of frequency, the owned the wind generators, the cost of wind power and penalty
more scientific and reasonable theory should be researched to help cost should be added in the function.
 R. Boqiang, J. Chuanwen / Renewable and Sustainable Energy Reviews 13 (2009) 2169–2174 2171

A trapeziform subjected function of objective function is displacing capacity will not. The advantage of this valuation model
contributed to calculate the pleasure degree of system operator is to require a relatively small amount of data allowing for
[4]. In this way, the objective of the optimization is transformed considerable flexibility integrating technologies.
from the minimization of operational cost to maximization of
pleasure degree. 3. Short-term economic dispatch model with wind power

2.2. The bi-objective function incorporating risk level For coping with discrete, large-scale, non-linear, and non-
convex problem of short-term of economic dispatch with wind
In power market, minimizing the operational cost of differing power, many accurate models which can conform to the theory of
generators and the risk level are two vital objectives. Because symmetrical fuzzy programming [10] are constituted to better
integrating the unpredictable and uncertainty characteristics of dispose the characteristic of wind generation while maintaining
wind power into the traditional thermal generation systems will or even enhancing the current stability and economic perfor-
bring the problem of system security, which the operator concerns. mance of power systems. Hong and Li explain the reason why the
In [5–7], a bi-objective economic dispatch problem related with fuzzy wind models are effective and make simulation to prove
wind power penetration is described. In this model, it considers it. Comparing with long-term schedule, short-term schedule
operational cost and security factors as opposite objectives which should be taken into account of more precise constraints, e.g.
should be minimized simultaneously. They define a fuzzy time limits.
membership function m as the system security level, which can
be described as two ways. In one side, the relationship between 3.1. The objective function incorporating stochastic security
system security level and wind power penetration in ED can be
linearization when the available wind power penetration is in the The stochastic security which can accounts for the expected
limit. In another side, a quadratic membership function is defined costs of preventive security actions and post-disturbance correc-
to reflect dispatcher’s different attitude, which is a corporate tive security actions is proposed as a way to further improve the
tactical or strategic plan that views wind power penetration with a systematic scheduling of reserves. Comparing with conservative
pessimistic or optimistic attitude. That is to say that the security deterministic approaches, stochastic security approach is accord-
level m will alter with wind power penetration. At the same time, ing more with actual condition.
Wang et al. added the quadratic functions with sine components, In [11], Bouffard and Galiana modeled the uncertain wind
which represent the rippling effects produced by the steam turbine generation (WTG) and load based on scenario approach
admission valve openings to the cost function of different considering scenario-specific unit commitment and ramping
generators, and utilize the function of the security level m to constraints. The aim of the electricity market-clearing problem
reflect the running cost of wind power. is also to minimize a measure social cost, the model is as
follows:
2.3. The objective function combining different types of wind farms in
delegated dispatch X
M X
FT ¼ ½C i ðPi Þ þ C r ðriu p ; ridn Þ( % pi ðkÞ½Bd ðdi ðkÞÞ
i¼1 ðk;IÞ 2 T
The delegated dispatch is a control centre, which monitors and
controls the wind generation producers to transmit the require- % C i ðg i ðkÞÞ % vTi li ðkÞ( (5)
ments of the system operator in instant situations. It includes
producers with different production controllability of the wind This model utilizes the probability function of the expected
park, which must be analysed in function of their possibility to social cost function to change the uncertainty of power system into
control its active and reactive outputs. But some older wind farms mathematics formula. Only in this way can their associated sets of
are less controllable. So the wind farms can be classified for three decisions variables be taken before the revelation of the
types based on their controllability [8], which are fully-control, uncertainty. So the concept of stochastic security is transformed
partial-control and non-control. into the style of probability.
According to this theory, Castronuovo et al. constitutes the
dispatch model, which can be divided into three parts. Because of 3.2. The objective function combining energy storage in balancing
the existence of different types of wind farms, the goal of the model market
is not only to maximize the profit of wind farms, but also to
decrease the quadratic difference between the active power and Bathurst and Strbac introduces a simple algorithm [12]
the maximum available instantaneous active power and to reduce comparing with more advanced methods [13], which can compute
the quadratic deviation between the power production angle of the the optimal dispatch of an electrical energy storage (EES)
wind parks and specified angles for this class of productions. combining the short-term power exchange and the expected
imbalance penalties of a wind farm. Based on the trait of the effect
2.4. The objective function considering environmental damage cost of daily price variation, imbalance price spread, market closure
lead-times, and wind contracting errors, writer institutes the
With the development of power technique, environmental optimal EES dispatch algorithm and model with wind farm to
problem has been attracting more and more people’s attention. increase the added value (AV) in tradeoff.
The most important environmental influence results from the Garcia and Weisser create two models to decide the size of grid
wasting of the primary fuel and its conversion into electricity for units and dispatch in a wind–diesel power system with hydrogen
the majority of fossil-fuel power generation. Kennedy presents a storage for reducing the cost [14]. One is linear programming, the
new method and a valuation model which includes energy, other is heuristic policies. Through comparison, linear program-
capacity, and environmental costs, and generation mix before ming is viewed as a benchmark to improve heuristic dispatch rules.
installing wind power capacity to estimate the social benefit of Though heuristic rule can deal with diversification of the wind
large-scale wind power production in long schedule [9]. This paper seasonality, it cannot be accounted for wind stochastic. The result
mainly discusses the impacts related to plant operations, so output shows the complex relationship between unit sizing decisions and
of displacing energy will produce an environmental benefit while dispatch policies in the presence of storage.
2172 R. Boqiang, J. Chuanwen / Renewable and Sustainable Energy Reviews 13 (2009) 2169–2174

3.3. The objective function considering spinning reserves advantage of the DSM algorithm is beginning with an initial
feasible solution and looking for the optimal solution along a
For solving the problem of system load forecast uncertainties, trajectory, which can maintain a feasible solution that is the most
wind speed forecast uncertainties and balancing the system load, notable advantage of the DSM algorithm.
Siider proposes a new methodology [15] to establish the operation
planning emphasizing on keeping of reserves which can be 4.2. Particle swarm optimization
trisected for instantaneous, fast and slow reserves and establishes
two models which are the wind power model and the conventional Particle swarm optimization is one of the most recent
hydro-thermal model. One includes the forecast of total wind developments in the category of evolutionary computational
power generation and the uncertainty of the forecast, and the other metaheuristic optimizations. This method has been developed
extends to take into account load forecast uncertainty and reserve based on the natural phenomenon of a bird flock or fish school. PSO
margins of the generation units. So it is more accurate and all-sided is stochastic optimization technique. The individual in a popula-
to make short-term dispatch plans of wind-hydro-thermal power tion represents a particle. The position of each particle is called a
systems through reckoning available capacities of the correspond- potential solution to the target problem. The velocity of each
ing reserve type. particle represents the distance to be moved at the next step. The
fitness of a particle is computed by objective function with former
3.4. The objective function including start-up/shut-down cost position. So the new velocity and the new position of each particle
is updated after each iteration to adjust the variable for achieving
The economic dispatch including wind power is to find an the global best position These courses make a PSO system combine
optimal assignment style of different generators, so the cost of the local search methods with global search methods.
start-up and shut-down of generators should be a significant The classical PSO algorithm does not have the ability to solve
component of the total cost. Reducing the frequency of start-up multi-objective optimization problems because there is no
and shut-down of generators is efficiency method to cut down the absolute global optimum existing there. So Wang and Singh in
operational cost. [5] modify the PSO algorithm combing the nondominated pareto-
In [16], Chen proposed a linear model which consider additional optimal solutions in terms of the specified multiple design
up/down spinning reserve capacity and total actual wind power objectives, and utilize the multi-objective particle swarm optimi-
generation. Because wind power needs no fuel, the total cost of zation (MOPSO) to settle the problem of wind power dispatch with
WTG from the public utility is the cheapest. Based on this theory, respect to risk and cost. Comparing with classic particle swarm
the writer decomposes operation cost into two parts: One is the optimization, i.e. MOPSO which has the ability to deal with
system generation cost of the fuel expense for balancing the constraints through a constraint checking procedure called
system load demand and reserves. Another is the start-up cost, rejecting strategy through creating archive and a candidate
which the unit operation trajectory controlled. A method for solution. Numerical simulations prove the validity and applic-
solving short-term scheduling considering start-up/shut-down ability of this approach.
cost is also proposed in [17,18]. On the side, the combination of the
mixed integer linear objective function including operational costs, 4.3. Simulated annealing approach
the turbine (on/off) status, start-up and shut-down of each turbine
and the non-linear function of minimization of the mismatch Simulated annealing approach shows the process of annealing
between the total wind park generation output (active and in physics in which a crystalline solid is heated and then very
reactive) and wind park dispatch center requests, are also brought slowly cooled until it achieves its regular crystal lattice shape. The
forward in [19]. lowest energy state is to reach the aim of this process. If the cooling
process is sufficiently slow, physical substances usually move from
4. Optimized algorithm for dispatch model higher energy states to lower ones in this process. SA is widely
utilized in structural optimization problems because of its inherent
With the development of traditional calculus and stochastic simplicity and ability to find the global optimum even if there are
searching techniques consisting of direct search method (DSM), many design variables.
evolutionary programming (EP), genetic algorithm (GA), particle Chen writes the paper combining an effective constrained
swarm optimization (PSO) and simulated annealing (SA), economic dynamic economic dispatch (CDED) method with the improved SA
dispatch problem of power network with wind generators is [16]. The SA algorithm with special random-perturbation scheme
further optimized in computing precision and convergence speed. which can satisfy the most difficult minimum uptime and
downtime constraints is used to optimize the generating unit
4.1. Direct search method schedule. And then the ramp-rate CDED is calculated by a direct
search method. Through this improvement, the time-dependent
Direct search method is a local optimization heuristic method constraints, such as ramp-rate constraints and wind generation
that purposes to maximize (or minimize) an unconstrained fluctuation constraints, can be computed in the economic dispatch
function using values only. Direct search methods do not claim with wind power at the same time.
global or local optimality of the computed solution but in many
cases they can prove useful. 4.4. Genetic algorithm approach
But this complex economic dispatch problem cannot apply DSM
directly, so the writer decompounds the confused problem into Genetic algorithm approach is stochastic search arithmetic,
wind and thermal sub-problems. To estimate the wind generation which is based on natural selection in biology. Through
solution for initialization, a brief method [3] based on an equal successive selection, crossover, mutation and reproduction,
incremental cost rule is used. After deciding total actual wind GA can approach the global optimum and have the capability
power generation, system spinning reserve capacity and the total of dealing with problem including the inequality constraints
operation cost of the WTG can also be reckoned. The sequential efficiently.
thermal sub-problem can be solved using the DSM algorithm in the Hong and Li consider the power generation with uncertainties
standard reserve constrained economic dispatch. The most notable which are modeled by a fuzzy set, and propose a genetic algorithm
 R. Boqiang, J. Chuanwen / Renewable and Sustainable Energy Reviews 13 (2009) 2169–2174 2173

approach to solve this problem which can computed as symme- and wind generation operator data to reduce the uncertainty of the
trical fuzzy program [10]. Because feasible and infeasible regions original problem. Through this measure, the system dispatcher and
compose the solution space for GA, the writers add the penalty independent generator can evaluate the risk of contracts not-
functions in GA to reinforce the corresponding interferential accomplishment even though wind power is quite small compar-
constraints from the infeasible region into the feasible region to ing with the total load.
increase the computation speed. Through comparison with Due to the wind speed trait mainly keeps to a Weibull
simulated annealing, this measure shows the steady convergence distribution along with time changing at a given place [23]. Hetzer
of iterations in a reasonable time. et al. utilize the Weibull distribution function, which has two
parameters to compute the wind power probability function [1],
4.5. Scenario construction method because Weibull distribution function is more common than the
single parameter Rayleigh distribution, and less complex than the
Bouffard et al. constructed the scenario-based approach [11] to five-parameter bivariate distribution. In [24], the evaluation of
compute the unit commitment, reserve levels, HTG dispatches, wind power uncertainty and its standard error are disposed with
negatively correlated WTG and load more explicitly on short-term Weibull distribution in detail.
schedule, and then utilize generation methods and the scenario Salles et al. propose two methodologies which are Monte Carlo
modeling to approximate the value of WTG and load over the Simulation (MCS) and Bar and Jenkins (B&J) approach for wind
scheduling limit. In addition, because the stochastic approach power economic analysis considering uncertainties related with
includes the extra flexibility provided by coordinated involuntary wind speed [25]. Integrated sequences of wind speeds and the
load shedding and wind spillage actions, it permits the expansion wished value of financial indicators which can compute empirical
of the feasible space of the security-constrained market-clearing probability distributions for financial risk analysis can be obtained
problem. So scenario construction method can solve the issue of by these two methodologies from a wind speed observed series.
wind power dispatch satisfying stochastic security on short-term Considering the wind speed forecast imprecision and weather
schedule. instability, a generic method to evaluate short-term wind power
forecast risk on-line is introduced [26,27]. For avoiding a restrained
5. Risk dispatch management of wind power in power market hypothesis on the distribution of the errors, confidence intervals
with a confidence level including the prediction horizon, the cut-
Risk dispatch management, which attempts to obtain the total off risk and the power class are presented based on the resampling
maximal of revenue or reduce the cost and risk of system aims to approach. This method is quite universal and can be applied to all
solve the uncertainty of wind speed in this paper. For security and kinds of wind forecasts to solve the problem of wind power
stabilization of system, short-term forecast which is up to 48 h uncertainties.
ahead of schedule are necessary. But because the information The uncertainties of wind power may break the normal
related with climatological and micrometeorological parameters transmission to load, so the special capacity of reverses is
are not extremely completed, wind speed is a non-stationary and necessary. In [28], Doherty et al. point out another approach to
non-linear resulting in a main trouble which the prediction estimate the reserve of the system with uncertain wind power.
intervals of cut-off speed can change from zero to maximum. Considering the probability of load and generation being out, wind
[20,21]. So it is necessary to find an optimal risk management of power and load forecast error, it is possible to create relationship
the wind power production to enhance the robusticity of the whole between the reserve level on the system in each hour and the
system with wind power. reliability of the system over the year to reckon value of reverse. In
this way the capacity of reverses and security of system can be
5.1. Management method in power market connected to make the power network with wind generator more
robust.
As a result of the intrinsic stochastic characteristic of wind
speed, which may bring on the disturbance of system, several 6. Conclusion
measures in allusion to wind speed uncertainty are proposed in
this paper. The goal in this paper is giving a review on the economic
Kariniotakis et al. utilize the time series method to transform dispatch problem and risk management of wind power in
the uncertainty wind speed to output of wind power [26]. Because electricity market. Based on the different styles of schedule, the
stochastic nature and chronological changeability of the wind factors which the dispatch models consider are dissimilated.
velocity determine the power output, there exists a non-linear Except for the total fuel cost, models append factors of risk level,
relationship between the power generated by a wind turbine type of wind farms, spinning reserves, environmental damage cost,
generators and the available wind speed. Though the historical etc. For further improving constringency speed of computation,
mean wind speed ut and its standard deviation st, the simulated many optimized algorithms are combined with relevant model.
wind speed vt can be computed as followed: Aiming at intermittency and unpredictability characteristic of
vt ¼ ut þ s t yt (6) wind power generation, this paper introduces several manage-
ment methods to solve the wind power uncertainty problem to
In this way wind speed can be estimated ahead. And then the keep system stable and secure.
power output can be calculated through establishing quadratic From these studies, we can conclude from the following three
equation of wind speed based on the relationship with cut-in speed aspects:
and cut-out speed. This method is effective when the orderliness of
wind speed can be found. ) The wind power dispatch and risk management has much
A fuzzy approach [22] based on a multi-linearized fuzzy power depended on the precision of speed forecast which may be
flow is presented by Mangueira et al. to estimate uncertainties of meaningful if the historical wind speed has disciplinarian to
the wind generation systems. For improving the quality of the follow at present, so it is still a difficult problem for ruleless wind
linearization, the writers’ models the assessment of the influence speed.
of wind power on the conventional electric grid as fuzzy numbers ) The more factors of wind dispatch are considered, the more
which include the last meteorological data, forecast accuracy data complex the dispatch model is. So simplex optimized algorithms,
2174 R. Boqiang, J. Chuanwen / Renewable and Sustainable Energy Reviews 13 (2009) 2169–2174

especially the intelligent algorithm cannot satisfy the demand of [10] Hong Y-Y, Li C-T. Short-term real-power scheduling considering fuzzy factors
in an autonomous system using genetic algorithms. The Institution of Engi-
optimization, many studies are inclined to integrate the neering and Technology 2006;153(6):684–92.
advantages of every algorithm and adopt the mixed algorithm [11] Bouffard F, Galiana FD. Stochastic security for operations planning with sig-
to deal with different variables. nificant wind power generation. IEEE Transactions on Power System 2008;23(2):
306–16.
) The security and stabilization of the system are quite significant [12] Bathurst GN, Strbac G. Value of combining energy storage and wind in short-
when integrating the wind power into the network, so the factors term energy and balancing markets. Electric Power Systems Research
of risk management for wind power uncertainty should be 2003;1–8.
[13] ESD Ltd. Maximising the commercial value of wind energy through forecast-
further taken into account in dispatch model. ing. ETSU 2000.
[14] Garcia RS, Weisser D. A wind–diesel system with hydrogen storage: joint
Acknowledgement optimisation of design and dispatch. Renewable Energy 2006;2296–320.
[15] Siider L. Reserve margin planning in a wind-hydro-thermal power system.
IEEE Transactions on Power Systems 1993;8(2):564–71.
The authors gratefully acknowledge the support and assistance [16] Chen C-L. Simulated annealing-based optimal wind-thermal coordination
of The National High Technology Research and Development scheduling. The Institution of Engineering and Technology 2007;1(3):447–55.
Program of China (863) (no.: 2007AA05Z458). [17] Contaxis GC, Kabouris J. Short term scheduling in a wind/diesel autonomous
energy system. IEEE Transactions on Power Systems 1991;6(3):1161–7.
[18] Göransson L, Johnsson F. Dispatch modeling of a regional power generation
References system—integrating wind power. Renewable Energy 2008;1–10.
[19] Moyano CF, Lopes JAP. An optimization approach for wind turbine commitment
[1] Yuewen J, Chong C, Buying W. Wind power system economic dispatch based and dispatch in a wind park. Electric Power Systems Research 2008;2705–14.
on PSO with stochastic simulation. Advanced Technology of Electrical Engi- [20] Diniza SMC, Sadek F, Simiu E. Wind speed estimation uncertainties: effects of
neering and Energy 2007;26(3):37–41 [in Chinese]. climatological and micrometeorological parameters. Probabilistic Engineering
[2] Hetzer J, Yu DC, Bhattarai K. An economic dispatch model incorporating wind Mechanics 2004;361–71.
power. IEEE Transactions on Energy Conversion 2008;23(2):603–11. [21] Simiu E, Scanlan RH. Wind effects on structures. New York: Wiley; 1996.
[3] Chen C-L, Lee T-Y, Jan R-M. Optimal wind-thermal coordination dispatch in [22] Mangueira HHD, Saavedra OR, Pessanha JEO. Impact of wind generation on the
isolated power systems with large integration of wind capacity. Energy Con- dispatch of the system: a fuzzy approach. Electrical Power and Energy Systems
version and Management 2006;3456–72. 2008;67–72.
[4] Haiyan C, Jinfu C, Xianzhong D. Wind power system economic dispatch of [23] Patel MR. Wind and solar power systems. Boca Raton, FL: CRC Press; 1999.
fuzzy modeling and optimization. Automation of Electric Power System [24] Ramı́rez P, Carta JA. Influence of the data sampling interval in the estimation of
2006;30(2):22–6 [in Chinese]. the parameters of the Weibull wind speed probability density distribution: a
[5] Wang L, Singh C. Balancing risk and cost in fuzzy economic dispatch including case study. Energy Conversion and Management 2005;2419–38.
wind power penetration based on particle swarm optimization. Electric Power [25] Salles ACN, Melo ACG, Legey LFL. Risk analysis methodologies for financial
Systems Research 2008;2623–31. evaluation of wind energy power generation projects in the Brazilian system.
[6] Miranda V, Hang PS. Economic dispatch model with fuzzy wind constraints International Conference on Probabilistic Methods Applied to Power Systems
and attitudes of dispatchers. IEEE Transactions on power systems 2005;20(4): 2004;1–6.
2143–5. [26] Kariniotakis GN, Pierre P. Uncertainty of short-term wind power forecasts—a
[7] Wang L, Singh C. Tradeoff between risk and cost in economic dispatch methodology for on-line assessment. International Conference on Probabil-
including wind power penetration using particle swarm optimization. Inter- istic Methods Applied to Power System 2004;1–7.
national Conference on Power System Technology 2006;1–7. [27] Pinson P, Kariniotakis GN. Wind power forecasting using fuzzy neural net-
[8] Castronuovo ED, Martı́nez-Crespo J, Usaola J. Optimal controllability of works enhanced with on-line prediction risk assessment. IEEE Bologna Power-
wind generators in a delegated dispatch. Electric Power Systems Research Tech Conference 2003.
2007;1442–8. [28] Doherty R, O’Malley M. A new approach to quantify reserve demand in
[9] Kennedy S. Wind power planning: assessing long-term costs and benefits. systems with significant installed wind capacity. IEEE Transaction on Power
Energy Policy 2005;1661–75. systems 2005;20(2):587–95.