Front.

Energy
DOI 10.1007/s11708-017-0457-7

REVIEW ARTICLE

S. Surender REDDY, Vuddanti SANDEEP, Chan-Mook JUNG

Review of stochastic optimization methods for smart grid

© Higher Education Press and Springer-Verlag Berlin Heidelberg 2017

Abstract This paper presents various approaches used fuels are not enough to meet the increasing demand for a
by researchers for handling the uncertainties involved in long time and their use is not advisable in environmental
renewable energy sources, load demands, etc. It gives an terms. Renewable energy resources (RERs) are the only
idea about stochastic programming (SP) and discusses the solution to global warming and higher cost of fossil fuels
formulations given by different researchers for objective issues [1]. They produce almost no waste products such as
functions such as cost, loss, generation expansion, and carbon dioxide or any other harmful chemicals. Since
voltage/V control with various conventional and advanced renewable energies have some technical and economic
methods. Besides, it gives a brief idea about SP and its challenges such as reliability, as their generation depends
applications and discusses different variants of SP such as on the availability of different RERs such as wind and
recourse model, chance constrained programming, sample solar, they involve unpredictable and random behavior.
average approximation, and risk aversion. Moreover, it This makes the whole power system unpredictable. In this
includes the application of these variants in various power paper, different OPF methods, i.e., linear, nonlinear and
systems. Furthermore, it also includes the general advanced technologies are evaluated and compared.
mathematical form of expression for these variants and RERs can also be used as optimized source of
discusses the mathematical description of the problem and distributed generation [2]. As it is a decentralized
modeling of the system. This review of different generation, it can be generated near the customer point
optimization techniques will be helpful for smart grid and the losses can be extensively minimized. The
development including renewable energy resources investment in transmission and distribution equipment
(RERs). will also be reduced due to the distributed generation
approach. Moreover, the control action will also be fast and
Keywords renewable energy sources, stochastic optimi- more reliable compared to the conventional generation
zation, smart grid, uncertainty, optimal power flow (OPF) approach. The distributed generation approach has bene-
fited the deregulation of energy. Consumers are also
getting benefits from the low cost and better quality of
1 Introduction supply [3]. The distributed generation also helps in peak
shaving. It has some major issues with safety and security,
World population is increasing drastically and these huge protection and control, efficiency and connection, etc. It
mass of human also has a great consumption of energy. can be used as standalone or grid connected source of
Therefore, to meet this requirement of increasing demand, energy. Renewable energy suffers from the reliability
renewable energy can be used along with fossil fuel. Fossil problem. Therefore, with storage into the system, the extra
power generated can be stored and can be utilized when
there is wind or solar power available. Researchers have
Received June 2, 2016; accepted August 25, 2016
tried to perform optimal power flow (OPF) operation with
S. Surender REDDY renewable energy and storage integrated into the system
Department of Railroad and Electrical Engineering, Woosong Uni-
versity, Republic of Korea [4]. Even though renewable energy has been discussed by
many researchers for various OPF algorithms [5,6], a study
Vuddanti SANDEEP which reviews the stochastic programming (SP) with
School of Engineering, Central University of Karnataka, India
RERs in the smart grid context is required.
Stochastic nonlinear programming (S-NLP) is proposed
Chan-Mook JUNG ( ) ✉ by many researchers as it uses randomness as part of
Department of Railroad and Civil Engineering, Woosong University,
Republic of Korea optimization processes [7–9]. The basic idea for solving
E-mail: [email protected] the stochastic problem is to first convert the problem into
2 Front. Energy

an equivalent deterministic form so that the conventional related to the forecasted values for load demand, available
optimization techniques can be applied to it [10]. output power of wind and photovoltaic units and market
Reference [10] presented the hybrid of two variants- price are modeled by scenario-based SP. Reference [22]
recourse model and chance constrained programming formulates a two-stage SP, where the first-stage is
(CCP) to deal with the complex problem of available associated with the electricity market and its rules and
transfer capability (ATC) evaluation. Reference [11] constraints, and the second-stage is related to the actual
discussed the stochastic OPF using constraint relaxation, operation of the power system and its physical limitations
and the challenges attached to it. The algorithm presented in each scenario. In Ref. [23], a novel stochastic
solves this problem through potential benefits such as probabilistic energy and a reserve market clearing scheme
restricted attention to the small set of violated constraints are proposed in the presence of plug-in vehicles (PEV) and
through constraint relaxation. Besides, for each different wind power introducing a new model for PEV aggregators.
contingency, it tests the newly violated line flow In Ref. [24], the problem of modeling and stochastic
constraints without exclusively computing the distribution optimization for home energy management is considered.
matrix for each contingency by using rank-one updates to Reference [25] proposes a probabilistic model using the
calculate post contingency power flows. Stochastic OPF point estimate method to reduce the probability of
was first developed by Yong and Lasseter as described in congestions and voltage violations in a smart grid located
Ref. [12]. The optimization was developed as a two stage in a radial distribution system. A detailed review of
SP problem. Reference [13] presents both linear and literature using agent based modeling and simulation
nonlinear formulations for the stochastic OPF. Reference techniques for analyzing smart grids from the perspective
[14] uses the SP approach for the vulnerability constrained of the system is presented in Ref. [26]. A comprehensive
transmission planning problem. The equivalent determi- review on mathematical modeling methods of photovoltaic
nistic model using mixed integer nonlinear programming is (PV) solar cell/module/array which can be used for power
utilized to find the optimal value. system dynamic modeling purpose is given in Ref. [27].
The SP problem with deterministic constraints can be Reference [28] proposes a robust optimization model for
formulated as an expected value function subject to optimal self scheduling of a hydro-thermal generating
deterministic constraints [15]. Reference [15] discusses company. A stochastic framework based on the cloud
stopping rules and a validation analysis for sample average theory to handle the uncertainty effects in the optimal
approximation (SAA) optimization procedures using some operation of microgrids is proposed in Ref. [29]. A review
illustration examples. The recourse model is constructed of various mathematical models proposed by different
through actions followed by observations and reaction. researchers is conducted in Ref. [30]. These models are
Another model, named as the chance constrained method, developed based on objective functions, economics and
is developed by Cooper and Charnes using the statistical reliability studies involving design parameters. Reference
decision theory [16]. The size of stochastic program [31] proposes a unit commitment formulation for the
depends upon the number of realizations of random microgrid using the two stage scenario based SP method.
parameters presented in the problem. According to theory The day-ahead scenarios and hour-ahead scenarios provid-
of “curse of dimensionality”, as the number of possible ing the information about the location of electric vehicles,
realizations increases, the time periods also increases and the historical data are presented in Ref. [32]. In Ref.
exponentially [4]. [33], a co-optimization based OPF solver is developed to
Reference [17] formulates a stochastic problem for solve over contingencies and renewable uncertainties. A
microgrid energy scheduling. It minimizes the expected stochastic model with high number of scenarios and the
operational cost of the microgrid and power losses while modeling of different strategies of EV integration are
accommodating the intermittent nature of RERs. A hybrid proposed in Ref. [34]. Reference [35] proposes the effects
stochastic/robust optimization model is proposed in Ref. of uncertain renewable energy and loads on optimizing
[18] to minimize the expected net cost, i.e., expected total profit and cost in a microgrid power market. A cloud
cost of operation minus total benefit of demand. This computing framework in a smart grid environment by
formulation can be solved by mixed-integer linear creating small integrated energy hub supporting real time
programming. A probabilistic model for small scale energy computing for handling huge storage of data is proposed in
resources and load demand which is used to determine the Ref. [36].
optimal scheduling of microgrids with minimum operating In Ref. [37], an optimal operation planning method
cost is presented in Ref. [19]. In Ref. [20], the key features taking into consideration the uncertainties of renewable
of microgrids and a comprehensive literature survey on the power generations and load demand is proposed. In Ref.
stochastic modeling and optimization tools for a microgrid [38], the performance of four different scheduling algo-
are presented. Reference [21] proposes a stochastic model rithms is compared to integrate the electric vehicles into
for optimal energy management with the goal of cost and smart grid architectures by optimizing their charging and
emission minimization. In this model, the uncertainties feeding periods. A dynamic optimization framework to
 S. Surender REDDY et al. Review of stochastic optimization methods for smart grid 3

formulate the optimal charging problem is proposed in Ref. sequences of x. Two different set of y’s can be chosen for
[39]. A methodology for modeling and controlling of the each possible outcome w. The recourse problem formula-
load demand in a residential distribution grid due to plug- tion is expressed as [44]
in hybrid electric vehicle (PHEV) battery charging and
discharging is proposed in Ref. [40]. A statistical model Minimize, f 1 ðxÞ þ E½f 2 ðyðwÞÞ,w, (1)
based optimization for developing dynamic, stochastic,
computationally efficient, and scalable platforms is s:t: g 1 ðxÞ£0, g m ðxÞ£0, (2)
proposed in Ref. [41]. Reference [42] proposes an
innovative technology to handle the growing complexity h1 ðx,yðwÞÞ£0, f or all w 2 W , (3)
of the smart grid and stochastic bidirectional OPFs to
maximize the penetration of renewable energy and to hK ðx,yðwÞÞ£0, f or all w 2 W , (4)
provide maximum utilization of available energy storage,
especially plug-in electric vehicles. Reference [43] pro- where x2X, y(w)2Y. The set of constraints h1, h2,..., hk
poses a stochastic energy scheduling model for a local area represent the links between the first stage decisions x, and
smart grid system with a single energy source and multiple the second stage decisions y(w). Recourse models can be
energy consumers. extended in a number of ways. The most common way is to
In this paper, various OPF methods i.e., linear, nonlinear include more stages. In the multi-stage problem, the
and advance techniques are evaluated and compared for decision is made now, waiting for some uncertainty to be
different objective functions such as cost, loss, generation resolved, and then another decision is made based on what
expansion, voltage/V, etc. aiming at developing an OPF for has happened. The objective is to minimize the expected
RERs using the best technology available considering the costs of all decisions made.
uncertainty and randomness added to the system due to Solving a recourse problem is difficult. Especially, the
RERs. The SP scheme and its variants are studied in detail hardest part is to solve the expected value expect the first
to deal with the uncertainty and randomness involved in stage [45]. In this model, the random constraints are
the system. The objective of this paper is to provide an modeled as ‘soft constraint’. The possible violations are
initial framework in the direction of smart grid with the accepted but the cost of these violations should be added. If
help of SP to optimize the system with renewable energy the recourse model does not change with the scenario, it is
sources. a fixed recourse problem [44].

2.2 CCP
2 Evaluation of stochastic optimization
methods for smart grid CCP is an approach in which the probability distribution of
random parameters are fixed and already known. This
This section describes various SP techniques. method is used where high uncertainty is involved and
reliability has a higher priority than other issues. In such
2.1 Recourse method application, one would like to settle up with an optimal
value that guarantees feasibility ‘as much as possible’ [46].
The recourse method is composed of more than one stage. The CCP is expressed as [45]
By solving the first stage, some uncertainties will be
P fAi ðωÞx³hi ðωÞg³i , (5)
resolved and other decision can be made based on the
values obtained from the first stage and this will be where 0 <  i < 1, i = 1,2,..., which is an index of
continued till the last stage is solved. The most difficult constraints. These constraints can be modeled in a general
part of this recourse problem is the evaluation of the expectation form as Eω (f i(ω, x(ω)))≥ i where f i is an
expected value at each stage except the first stage [44]. In indicator of {ω/Ai(ω)x≥hi(ω)}. The objectives for CCP
the recourse problem, one decision can be made now and are in one of the following forms: (a) It can be an
the expected costs of consequences of the decision made in expectation function (E-model); (b) It can be the variance
the first stage can be minimized. This method uses of some value (V-model); (c) It can be the probability of
randomly generated observations from random variables some occurrence (P-model); (d) It can be the quintile of a
to create statistical estimates. Stochastic decomposition random function. In CCP, the generalized concavity theory
methods can be further subdivided into the decomposition plays a vital role, as it is inclusive of powerful tools for
based method and the stochastic approximations based convex analysis [4].
method. The following is a step-by-step procedure to solve
Let x be a vector of decisions that must be taken, and y stochastic CCP.
(w) be a vector of decisions that describes the con- Step 1: Create stochastic model
4 Front. Energy

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X
n
X uX
maximize z ¼
n u n
cj xj , (6) faij gxj þ Kαi t varfaij gx2j £bi : (16)
j¼1
j¼1 j¼1
( )
Xn
such that p aij xj £bi ³1 – αi , i ¼ 1,2,:::,m, 0<αi This constraint can now be put in the separable
i¼1 programming form by using the substitution
<1. aij is normally distributed with a mean of E{aij}, a vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uX
variance of var{aij}, and a covariance of cov{aij ,ai# j# }. u n
Xn yi ¼ t varfaij gx2j , (17)
Step 2: Define hi ¼ aij xj , and hi is normally j¼1
j¼1
Xn ( )
X
n
distributed with Efhi g ¼ Efaij gxj . P bi ³ aij xj ³αi , (18)
j¼1
j¼1

varfhi g ¼ X T Di X , (7) 8 9
> Xn
>
>
> aij xj – Efbi g >
>
>
< b – Efb g >
=
X ¼ fx1 ,x2 ,x3 ,:::xn gT , (8) j¼1
P pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ³ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i i
³αi : (19)
>
> varfbi g varfbi g > >
Covariance matrix >
> >
>
2 3 : ;
varfai1 g ::: covfai1 ,ain g
6 ::: ::: ::: 7 The necessary condition for this is
Di ¼ 4 5, (9)
covfain ,ai1 g ::: varfain g X
n
aij xj – Efbi g
( ) j¼1
h – Efhi g bi – Efhi g pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi £Kαi : (20)
fhi £bi g ¼ P pi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi £ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ³1 – αi , (10) varfbi g
varfhi g varfhi g
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Therefore, the stochastic constraint is now converted
where ðhi – Efhi gÞ= varfhi g is a standard normal into the deterministic linear constraint as
distribution with a mean of zero and a variance of one. X
n pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
This means that aij xj £Efbi g þ Kai varfbi g: (21)
! j¼1
hi – Efhi g
Pfhi £bi g ¼ Φ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , (11)
varfhi g
2.3 Monte Carlo simulation (MCS) / sample average
where F represents the CDF of the standard normal approximation (SAA)
distribution. Let Kai be the standard normal value such that
Stochastic programming incorporates deterministic opti-
ΦðK ai Þ ¼ 1 – ai : (12)
mization with random variables and probabilistic con-
P{hi£bi}≥1–αi is realized if and only if straints. Large size of these problems involves a very large
! number, and sometimes an infinite number of scenarios.
bi – Efhi g Eventually, this may lead to intractable models which
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ³Kαi : (13) require specially designed algorithm to solve them. The
varfhi g
Monte Carlo sampling based method is suggested for such
This yields the following nonlinear constraint kind of problem to limit the number of scenarios and still
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi obtain reasonable solutions [7,47]. The Monte Carlo
Xn
Efaij gxj þ Kαi X T Di X £bi , (14) simulation (MCS) may be the best solution to estimating
j¼1
the expectation function [8]. When there exists a very large
number of scenarios, the sample of N replications of a
which is equivalent to the original stochastic constraint. random variable d, i.e., d1, d2,..., dN can be generated. The
For the special case where the normal distributions are sample generated can also be stored in computer memory
independent, and alternatively, it can also be generated by using the
covfaij ,aij ,ai# j# g ¼ 0, (15) common random number generation method. This method
is explained clearly in Ref. [4]. However, the result
and the last constraint reduces to obtained using this method does not give the assurance of
 S. Surender REDDY et al. Review of stochastic optimization methods for smart grid 5

quality. d j ( j = 1, 2,..., N) has the identical probability

distribution as d. If d j is mutually exclusive, the sample
evaluated is IID (independently and identically distrib-
uted). For a sample evaluated, the expectation function p
(x) = E [P (x, d)] can be approximated by the average,
X
N
^p N ðxÞ ¼ N – 1 Pðx,δ j Þ: (22)
j¼1

This method is also known as sample average

approximation (SAA). By the law of large numbers
(LLN), under some conditions, ^pN ðxÞ converges point
wise with point 1 to p(x) as N ↕ ↓1. This is true if the
sample is IID according to LLN and the convergence is
uniform [4]. SAA is not an algorithm for solving any SP.
Therefore, a numerical procedure should be followed. The
recourse function of any stochastic problem can be
replaced by Monte Carlo estimate to solve it with the
SAA method [45]. This approach gives reasonably
accurate results if its variability is not too large and it has
a relatively complete recourse [4]. The sample average
functions ^pN ðxÞ on a probability space (W, P, P) can be
defined. The advantage of this method is that it does not
require any tuning parameter. However, this method is
computationally expensive and it requires nonlinear
optimization software.
MCS is a technique that uses random numbers and their
PDFs to solve problems. This method is often used when
the model is complex, nonlinear, or involves many
uncertain parameters. A comprehensive approach to
Fig. 1 Flowchart for handling the uncertainty using MCS in the
evaluating the system reliability concerning the stochastic smart grid context
modeling of plug-in-hybrid-electric vehicles (PHEVs),
renewable resources, availability of devices, and etc. is
proposed in Ref. [48]. In addition, a novel risk manage- E½f ðY 1 Þ<E½f ðY 2 Þ: (23)
ment method in order to reduce the negative PHEVs effects The function f is known as disutility function, and is
is introduced. Reference [49] deals with the issues assumed as non-decreasing and a convex function for the
concerning transaction costs and financial risks by minimization function, and vice versa for the maximization
introducing a MCS approach to risk analysis based on an function. Therefore, the objective function can be
entire life-cycle representation of renewable energy formulated as
technology investment projects. Reference [50] proposes
Min
a methodology that uses MCS to estimate the behavior of x2X E½f ðFðx,wÞÞ: (24)
economic parameters which may help decision-making,
considering the risk in project sustainability. The flowchart As function f is a convex function, by using Jensen’s
for handling the uncertainty using the MCS method in the inequality,
smart grid context is shown in Fig. 1. f ðE½Fðx,wÞÞ£E½f ðFðx,wÞÞ: (25)
2.4 Risk averse optimization So, the sure outcome of E[F(x,w)] is similar to random
outcome F(x,w).
Risk averse optimization is based on the economics theory
of expected utility [4]. The risk averse method does not
depend upon the law of large numbers. In this method, two 3 Comparison of SP techniques
random outcomes are compared by expected values of
some scalar transformation f: R ! R of the realizations of The use of SP application depends upon the parameters
these outcomes. A random outcome Y1 is preferred over Y2 such as application, accuracy, CPU time, robustness,
if convexity, continuity and stability properties. A compar-
6 Front. Energy

ison of parameters of SP techniques is given in Table 1. is used at each phase of power systems for better reliability
The structure of the stochastic problem expands propor- and control. Renewable energy is a kind of pollution-free,
tionally to the number of possible realizations of uncertain economic and easily available source of energy. The output
parameters [4], which eventually increases exponentially received from RERs flutters widely, rapidly and randomly
in the number of time periods and parameters. Besides, which results in an increased operational complexity.
some of the methods may be very efficient for small size Moreover, the availability of RERs is usually in remote
problems. However, they may not give similar results for locations. Therefore, transmission losses are very high as
large scale problems. The emphasis in this paper is to the power has to be transmitted over a long distance.
check the validity of selected methods for various Researchers have considered the storage technologies as a
applications from small scale to large scale. Table 2 solution to the above problems [1]. Some work has also
presents the comparison of various SP variants in terms of been done to use storage in the OPF problem either as a
initialization, necessary conditions and feasibility. constraint or in objective formulation with the integration
of renewable energy and micro grid [2–5].
In recent years, point estimate methods have been
4 Probabilistic OPF (P-OPF) widely used to solve P-OPF problems. The aim of any
point estimate method is to compute the moments of a
OPF is a tool to optimize a power system objective over random variable Z that is a function of m random input
power network variables under certain constraints. It is a variable pl, i.e., Z = F(p1,p2,...,pm). The first point estimate
gadget for better reliability and control of power systems. It method was developed by Rosenblueth in 1975 [51] for

Table 1 Comparison of SP variants [9–13]

Chance constrained/ probability Sample average approximation/ Monte
Parameters Recourse model Risk aversion
constraint programming Carlo simulation
Handling of It can efficiently handle It can efficiently handle Most of the approximation methods It can handle the nonlinear-
nonlinearity nonlinearities nonlinearities involve some discretization of an ity through the linearization
underlying probability function which may approximate
nonlinear utilities
Complexity The fixed recourse model is It involves complexity. Specially, Choosing approximations and models It involves less complexity
simple and easily applicable for it involves more complexity in is difficult and deriving solutions for compared to other methods
any stochastic problem joint CCPs compared to recourse models based on the discrete
problems approximations are complex
Convex It is efficient for convex problems It can handle convex problems It can handle convex problems It can handle convex
problems (with discrete random variables, problems
and piecewise linear)
Large scale It can be applied to large scale It can be applied to large scale It can either be applied to large sample It is not very efficient for
problems problems problems approximation, where asymptotic large scale problems
statistical characteristics may apply or
be implemented with small sample
batches attached to an algorithm that
converges asymptotically
Iterations For fixed recourse, the number It depends upon the number of It depends upon the number of After each stage, the size of
of iterations is less compared probabilistic constraints samples selected the problem grows, by
to other methods applying multi cut strategy,
and the number of iterations
can be minimized
Speed It leads to a considerable size It is a very less time consuming It is inversely proportional to the number High
expansion of the problem and method of samples
eventually increases computational
burden, if the variables are
continuous
Accuracy It is a highly accurate method The accuracy is lower than the It depends upon the size of sampling High
recourse model but higher than
other models
Robustness It is a robust method. It involves Good Robustness depends upon the sample size. Good
different scenarios and it is suitable It is very good for a large number of
for all conditions samples
 S. Surender REDDY et al. Review of stochastic optimization methods for smart grid 7

symmetric variables and was later revisited in 1981 [52] to power system problem, and the optimal solution leads to
consider asymmetric variables. In Ref. [53], four different the ideal operation of the power system. For any typical
Hong’s point estimate schemes are presented and tested on OPF problem, first, all inputs need to be modeled so that it
the probabilistic power flow problem. Reference [54] can fit to the mathematical form. According to the desired
presents an application of a two-point estimate method form of output, objective functions and constraints should
(2PEM) to account for uncertainties in the OPF problem in be defined. The problem formulated can either be solved
the context of competitive electricity markets. The 2PEM by using classical methods such as linear programming,
needs 2n runs of the deterministic OPF for n uncertain nonlinear programming, quadratic programming, integer
variables to get the result in terms of the first three programming, dynamic programming or any of the
moments of the corresponding PDFs. More details about advanced methods of optimization such as adaptive
the MCS and the point estimate methods can be found in dynamic programming, evolutionary programming, artifi-
Refs. [55–57]. cial intelligence methods and heuristic programming, etc.
OPF problems involve different power system objec- Table 3 is generated for comparison of the above methods
tives such as cost minimization, environmental dispatch, for different objective functions. It gives the guidelines for
maximum power transfer, reactive power objectives — selecting a method for a given optimization problem and its
minimization of MW and MVA losses, general objec- relative application.
tives — minimum deviation from a target schedule, mini- One needs to be aware of the optimization formulations
mum control shifts, least absolute shift approximation of relative to each objective function. Table 4 is prepared to
control shift, and constraints such as limits on control satisfy the same need and it can be customized to fit into
variables — generator output in MW, transformer tap the system requirement.
limits, shunt capacitor limits, operating limits on line and OPF formulations should be tested for various con-
transformer flows — MVA, amps, MW and MVA, MW and tingencies [47,70] in order to evaluate the robustness of
MVA reserve margins, voltage and angle, control para- solution for different situations. As the power system
meters — control effectiveness, limit priorities through should be designed to work satisfactorily under con-
engineering rules and operating limit enforcement, voltage tingencies such as line outage, generation outage,
stability limits, local and non optimized controls — gen- transformer problems, and relay malfunctions, etc. For a
erator voltage, general real power, transformer output typical OPF problem, the base case cost is mostly higher
voltage, MVA and shunt/SVC controls, equipment gang- when contingencies are imposed as the system behaves
ing and sharing — tap changing, generator MVA sharing, conservatively. As a matter of fact, it is not possible to
and control ordering [6]. It is a mathematical form of any model each possible contingency. Therefore, usually

Table 2 Comparison of SP variants in terms of initialization, necessary and feasible conditions

Chance constrained
SP variant Recourse model Sample average approximation Risk averse optimization
programming
Initialization Z = X + EqY with x T=(x 1,x 2,...) with Min zP(x, x r, x k)= cTx + prqTy Min z = cTx + q Min cTx + qTy: Ax + By = z(w),
probability pr=(p1,p2,...) (x r)+(1–pr)qTy(x k) such that constraints with x x2X, y2Y
uniformly distributed
Necessary P
K SPEV³pr z*r Jensen lower bound and Edmun- decide x–observe w–decide
condition Solve equations Esþ1 ¼ pk :ðπ vk ÞTk X k * don-Madansky upper bound y = y(x, w)
k¼1
þ p zk ¼ WS
P
K
£Lj ðxv Þ
and esþ1 ¼ pk :ðπ vk Þ k≠r
k¼1
¼ !Ω QLj ðxv ,ÞPjL dω
j ðx Þ
£U v

¼ !Ω QU
j ðx ,ÞPj dω
v U

Feasibility v ³wv ? –0 v ³£Lj ðxv Þ Risk measures satisfied?

check £EVRS–EPEV and
£VSS v ³£Uj ðx Þ
v
£EVRS–SPEV
£EVRS–WS
Update s = s + 1, add feasibility cut x r, x k s = s + 1, add feasibility cut Fix x for non-anticipatively
Optimization x* is optimal solution x* x* is optimal solution ^x* is the optimal solution
8 Front. Energy

Table 3 Comparison of OPF methods [15,16], [45,46], [58–61]

Classical methods— Classical Intelligent systems Intelligent systems— Intelligent systems Intelligent systems
linear programming methods—nonlinear —artificial neural genetic algorithm —ant colony —article swarm
programming network optimization
Objective Generation cost, Generation fuel cost Reactive power and Transmission loss Loss reduction, Economic emission
functions power generation minimization, voltage control, minimization, optimal load dispatch, reactive
dispatch, power minimization of power generation scheduling, reactive power shedding, voltage power optimization,
generation losses, minimization of generation cost, loss control, unit stability, voltage generation cost
scheduling, power production cost, minimization, reactive commitment, control, economic minimization,
transmission minimization of power dispatch, cost economic dispatch load dispatch, optimal minimize fuel cost,
planning, economic generation cost, LMP, minimization and load allocation, capacitor placement, voltage stability
dispatch (minimum economic dispatch voltage control, VAR planning enhancement,
cost allocation of optimal allocation of minimization of
demand), locational static VAR polluted gas emission,
marginal price (LMP) compensators reactive loss
minimization
Merits They can efficiently They give solutions It requires less training It does not require any It has inherent It has very less
solve local involving nonlinearity in compared to other external rules because parallelism; computational burden;
constraints; the objective function as intelligent systems; it works on internal it is useful for it is a simple method
they are reliable for well as constraints; it can detect nonlinear rules traveling salesman and easy to implement;
most engineering they are better fitted for relationships between problem and similar it is a robust method
applications highly constrained dependent and applications; it can compared to other
methods independent variables efficiently handle methods
dynamic problems
Demerits They are less accurate They have slow It has a “black box” kind It does not give Theoretical analysis is It is less accurate;
compared to other convergence near of structure, so the exact solution; difficult; it is not efficient for
methods; optimal solution; results obtained are it is a slow method probability local search space
they cannot handle they are not suitable for highly questionable; distribution changes
the randomness and multi objective lots of computations are by iteration
stochastic nature of optimization problems involved
the problem;
there is no correlation
among variables
Significant They make best The introduction of It is used to model It is useful when the Convergence is It has flexibility in
contribution possible use of time, Krush Kuhn Tucker complex relationships search space is large guaranteed but the global and local control
machines, labor, etc. conditions and penalty between input and and complex; it is time required for exploration of the
function gives the output and discover useful when convergence is search space
optimal solution patterns in data mathematical analysis uncertain
is not possible
Nonlinear The objective They can solve It can solve It can solve It can solve It can solve
problems function and nonlinear problems nonlinear problems nonlinear problems nonlinear problems nonlinear problems
constraints must be
linear
Uncertainty and They cannot handle They cannot handle the It can work for problems It can work for It can work for It can work for
randomness the uncertainty and uncertainty and the involving uncertainty problems involving problems involving problems involving
the randomness in the randomness in the and randomness uncertainty and uncertainty and uncertainty and
problem problem randomness randomness randomness
Speed The solution can be They have slow It depends on the method It is a slow method It requires very high It is a fast method as
achieved very fast for convergence compared used and nature of the computational time low computation is
some of the to other methods problem; it has slow required and it has a
applications convergence when the fast convergence speed
optimal solution is far
from local maxima/
minima
 S. Surender REDDY et al. Review of stochastic optimization methods for smart grid 9

(Continued)
Classical methods— Classical Intelligent systems Intelligent systems— Intelligent systems Intelligent systems
linear programming methods—nonlinear —artificial neural genetic algorithm —ant colony —article swarm
programming network optimization
Working They work on They are mathematical It mimics the behavior It mimics the It is a meta heuristic It is a stochastic
principal mathematical methods that can solve of biological neurons; it Darwinian selection optimization method optimization method
principal; they work linear and nonlinear is basically an adaptive process; it gives based on ant behavior mimicking the social
for linear objective objective functions and network that accepts solutions to seeking a path behavior of bird
and linear constraint; constraints information internally optimization process through their colony flocking and fish
the constraints may be and externally; it uses through simulation of in search of food schooling
an equality or an weights to change the natural evaluation
inequality function. network parameter phases such as
some variables may inheritance, mutation,
be restrictive to selection and cross
non-negative over
Methods Graphical method, Branch and bound, Clustering, back –– –– ––
duality, integer Langrangian, mixed propagation, feed-
programming, integer programming forward
Langrangian, simplex etc.
method, etc.
Applicatio-ns Power system Optimal control, power Online-load-frequency Loss minimization, Real time applications Economic dispatch,
(power systems) planning, operation dispatch, unit control, security power system along with energy power system
and economics commitment, and power assessment, voltage planning, operation, management system, operation, and voltage
problems generation planning stability, minimization of maintenance and power loss control
problems power losses, fault control applications minimization, power
analysis, and etc. system planning and
design applications
Real world large They are not able to They are not able to The convergence is not It is useful for most It is useful for real It is capable of solving
scale power solve these problems solve these problems as guaranteed when of the real life world applications; it large problems
system problems as they require they require complex optimal solution is very applications is not useful for large
complex mathematical work and far from local minima; size applications
mathematical work they are unable to solve so, it cannot be used
and they are unable multi objective for real life applications
to solve multi problems
objective problems

researchers account for most probable contingencies only 5 Conclusions

depending upon particular power system operation.
The OPF methods presently under research require the This paper presented the review of stochastic optimization
accommodation of load flow control and monitoring techniques for use in the smart grid. It aimed to provide the
devices such as load tap-changing/phase-shifting transfor- initial work in this direction with the help of SP to optimize
mers, the modern flexible AC transmission system a system with renewable energy involvement. It reviewed
(FACTS), the phasor measurement unit (PMU), renewable OPF techniques and formulations given by various
energy, smart meters and storage and many other advanced researchers, the relative pros and cons, the applications,
communication devices. It is the advanced distribution the adaption of these techniques, and formulations for large
management system (DMS) and the SCADA technology and complex systems, etc. Though the review of SP has
that support the smart grid. The advance DMS extends its been presented in literature, a detailed review and
work from transmission to distribution and addresses formulations in presence of variability and uncertainty of
control functions such as reactive dispatch, voltage renewable energy sources are not presented. Therefore, this
regulation, contingency analysis and ranking, capability paper made an attempt at bridging this gap. It presented the
maximization or line switching. Improved and advanced development of SP variants and their application, algo-
OPF tools also address the requirements of smart rithm and comparison. It also discussed different variants
distribution network which can also be attached to the of SP in detail. It compared these variants stage by stage
advanced DMS. from the initialization point to the optimization point.
10 Front. Energy

Table 4 Various optimization problem formulations for specific objective using different methods
Optimization methods
Objective function
Nonlinear programming [62] SP-benders decomposition [63] Genetic algorithm [64]
XX
Cost n X
X n
Minimize fFj vj ðkÞ þ Aj vj ðkÞ þ Ej ðtj ðkÞÞg fi ðPi Þ ¼ ai Pi2 þ bi Pi þ ci ;
f ð$Þ ¼ – Gij fðfi – fj Þ2 þ ðei – ej Þ2 g, k 2K j2J X
i¼1 j¼1 Pi ¼ Pload þ Ploss
s:t: X Scheduled
i 2 TPP
n X
X n – Pk ;
– Bij fðfi – fj Þ2 þ ðei – ej Þ2 g, X X k 2 HPP
i¼1 j¼1 tj ðkÞ þ Bnp ðδp ðkÞ – δn ðkÞÞ Pimin £Pi £Pimax
j 2X p 2 ΩnX
X An

ða2i PGi
2
þ a1i PGi þ a0i Þ, þ Hi ui ðkÞ – Knp ð1 – cosðδp ðkÞ – δn ðkÞÞÞ,
i 2 SG i2x p 2 Ωn
¼ Dn ðkÞ : fn ðkÞ,
X
PGi , spinning reserve constraint per period is
i 2 SG X X
T j vj ðkÞ þ ∏i U i
power flow equations: j 2 JX i2I
X
n ³ Dn ðkÞ þ RðkÞ 8k 2 K,
PGi ¼ PDi þ ðei ðej Gij – fj Bij Þ þ fi ðfj Gij n2N
j¼1

T j vj ðkÞ£lj ðkÞ£T j vj ðkÞ 8j 2 J ,8k 2 K,

þ ej Bij Þ,
– Cnp £Bnp ðδp ðkÞ – δn ðkÞÞ£Cnp 8n 2 N ,8p 2 Ωn ,8k 2 K,
X
n
QRi ¼ QDi þ ðfi ðej Gij – fj Bij Þe – ei ðfj Gij hydraulic continuity equations
j¼1 X
xi ðkÞ ¼ xi ðk – 1Þ – ui ðkÞ þ Wi ðkÞ þ up 8k 2 K,8i 2 I
þ ej Bij Þ, p 2 Ωi
X i £xi ðkÞ£X i 8k 2 K,8i 2 I
system constraints:
yi ðkÞ³vj ðkÞ – vj ðk – 1Þ 8k 2 K,8j 2 J
Pij min£Pij £ Pij max,
vj ðkÞ,yj ðkÞ 2 f0,1g 8k 2 K,8j 2 J

Pij ¼ ðe2i þ fi2 – ei ej – fi fj ÞGij

þ ðei fj – ej fi ÞBij ,

2
Vimin £ðe2i þ fi2 Þ£Vimax
2
,
PGimin £PGi £PGimax ,
QGimin £QGi £QGimax ,
QRimin £QRi £QRimax ,

Optimization methods
Objective function
Interior point method [65] Extended conic quadratic programming [66] SP-benders decomposition [67]
Loss X
p
Minimize Ps Min1 T :εk
Min f ðxÞ – k ðlns1j þ lns2j Þ
j¼1
s.t. Pi= PGi–PDi,
X
p Qi= QGi–QDi+ QCi, s:t:gk ðx0k ,u0 þ εk Þ ¼ 0,
– k
ðlns3j þ lns4j Þ
PGi min£PGi £PGi max,
k £ðxk ,u0 þ εk Þ£hk ,
j¼1
hmin 0 max

such that g(x)= 0; QGi min£QGi £QGi max

1T :εk þ lðu*0 – u0 Þ£0
– s1 – s2 – h min
þh max
¼ 0;

– hðxÞ – s2 þ hmax ¼ 0;

– s3 – s4 – ^xmin þ ^xmax ¼ 0;

– ^I X – s4 þ ^xmax ¼ 0
 S. Surender REDDY et al. Review of stochastic optimization methods for smart grid 11

(Continued)
Optimization methods
Sequential quadratic programming (SQP) Nonlinear programming [69] Mixed integer nonlinear
[68] programming [70]
X X ( )
Generation capacity/ f ðPG , PT Þ ¼ Cg ðPg Þ þ CT ðPT Þ, X X X
NG
expansion g T min CGi ðPGi Þ þ IGGi ðPGi Þ F1 ¼ ai þ bi PGi þ ci PGi
2
ð$Þ,
PGi
i2G i 2 Gþ i¼1
X X
ðPtk þ jQtk Þ þ ðPgk þ jQgk Þ X X
s:t: ðPGi þ QGi Þ ¼ ðPDs þ QDs Þ X
Nl
tX g
i2G s2D
F2 ¼ PLoss ðx,uÞ ¼ Pl ,
þ ðPdk þ jQdk Þ ¼ 0 l¼1
d   F3 ¼ ðx,uÞ,
jIsf Z~ SFCL j£Ismax ,
C  S  1000
Vbmin <Vb <Vbmax , Vsmin £Vs £Vsmax , F4 ¼ ð$Þ,
8760  5
LBg < Pg < UBg, fsrk £fsrk
max
,
St <Stmax , min
PGi £PGi £PGi
max
, s.t. PGi – ρPDi= fpi(x,u),

Gi £QGi £QGi
Qmin max QGi – QDi ¼ fQi ðx,uÞ,
PT <PTmax ,
min
VGi £VGi £VGi
max
i 2 1,:::,NG,
Qmin max
T <QT <QT
min
PGi £PGi £PGi
max
i 2 1,:::,NG,

Gi £QGi £QGi i 2 1,:::,NG,

Qmin max

VLimin £VLi £VLimax i 2 1,::,Nd,

SLi £SLi
max
i ¼ 1,::::, Nl,

PST : min ££max ,

SPST £SPST
max
,
HFC : k min £k£k max ,
0£KC £KCmax ,
0£KL £KLmax ,
0£Km £Kmax
m ,

SHFC £SHFC
max
,
UPFC : rmin £r£rmax ,
gmin £g£gmax ,
SUPFC £SUPFC
max

References 6. PSERC Executive Committee. “Challenges in integrating renewable

technologies into an electric power system” white paper, April
1. Pimentel D, Herz M, Glickstein M, Zimmerman M, Allen R, Becker 2010
K, Evans J, Hussain B, Sarsfeld R, Grosfeld A, Seidel T. Renewable 7. Amarnath R V, Ramana N V. State of art in optimal power flow
energy: current and potential issues. Bioscience, 2002, 52(12): solution methodologies. Journal of Theoretical and Applied
1111–1120 Information Technology, 2011, 30(2): 128–154
2. Ochoa L F, Harrison G P. Minimizing energy losses: optimal 8. Selvi V, Umarani R. Comparative analysis of ant colony and particle
accommodation and smart operation of renewable distributed swarm optimization techniques. International Journal of Computers
generation. IEEE Transactions on Power Systems, 2011, 26(1): and Applications, 2010, 5(4): 1–6
198–205 9. AlRashidi M R, El-Hawary M E. A survey of particle swarm
3. Pepermans G, Driesen J, Haeseldonckx D, D’haeseleer W, Belmans optimization applications in electric power systems. IEEE Transac-
R. Distributed generation: definition, benefits and issues. Energy tions on Evolutionary Computation, 2009, 13(4): 913–918
Policy, 2005, 33(6): 787–798 10. Momoh J A, Adapa R, El-Hawary M E. A review of selected
4. Rasmussen C N. Energy storage for improvement of wind power optimal power flow literature to 1993. I. Nonlinear and quadratic
characteristics. IEEE Trondheim Powertech, 2011: 1–8 programming approaches. IEEE Transactions on Power Systems,
5. Jabr R A, Pal B C. Intermittent wind generation in optimal power 1999, 14(1): 96–104
flow dispatching. IET Generation, Transmission & Distribution, 11. Monticelli A, Pereira M V F, Granville S. Security constrained
2009, 3(1): 66–74 optimal power flow with post-contingency corrective rescheduling.
12 Front. Energy

IEEE Transaction on Power System, 1987, pwrs-2:1(1): 175–180 30. Siddaiah R, Saini R P. A review on planning, configurations,
12. Stott B, Alsac O, Monticelli A. Security analysis and optimization. modeling and optimization techniques of hybrid renewable energy
Proceedings of the IEEE, 1987, 75(12): 1623–1644 systems for off grid applications. Renewable & Sustainable Energy
13. Birge J R, Louveaux F. Introduction to Stochastic Programming. Reviews, 2016, 58: 376–396
New York: Springer-Verlag, 1997 31. Mohammadi S, Mohammadi A. Stochastic scenario-based model
14. Allan R N, Da Silva A M L, Burchett R C. Evaluation methods and and investigating size of battery energy storage and thermal energy
accuracy in probabilistic load flow solutions. IEEE Transactions on storage for micro-grid. International Journal of Electrical Power &
Power Apparatus & Systems, 1981, PAS-100(5): 2539–2546 Energy System, 2014, 61: 531–546
15. Chandy K M, Low S H, Topcu U, Xu H. A simple optimal power 32. Soares J, Vale Z. Stochastic optimization of distributed energy
flow model with energy storage. 49th IEEE Conference on Decision resources in smart grids. 2015, http://sites.ieee.org/psace-mho/files/
and Control, 2011: 58(8) 1051–1057 2015/07/01_15PESGM2908.pdf
16. Yau T, Walker L, Graham H, Gupta A. Effects of battery storage 33. Chiang H D. Stochastic security-constrained AC optimal power
devices on power system dispatch. IEEE Transactions on Power flow solver for large power networks with renewable. 2015, http://
Apparatus & Systems, 1981, PAS-100(1): 375–383 sites.ieee.org/psace-mho/files/2015/07/05_15PESGM2914.pdf
17. Su W, Wang J, Roh J. Stochastic energy scheduling in microgrids 34. Miranda V. Electric vehicles in smart grids: a hybrid Benders/EPSO
with intermittent renewable energy resources. IEEE Transactions on solver for stochastic reservoir optimization. 2015, http://sites.ieee.
Smart Grid, 2013, 5(4): 1876–1883 org/psace-mho/files/2015/07/06_15PESGM2910.pdf
18. Liu G, Xu Y, Tomsovic K. Bidding strategy for microgrid in day- 35. Mohan V, Singh J G, Ongsakul W, Unni A C, Sasidharan N.
ahead market based on hybrid stochastic/robust optimization. IEEE Stochastic effects of renewable energy and loads on optimizing
Transactions on Smart Grid, 2016, 7(1): 227–237 microgrid market benefits. Procedia Technology, 2015, 21: 15–23
19. Nikmehr N, Najafi-Ravadanegh S. Probabilistic optimal power 36. Reka S S, Ramesh V. Demand side management scheme in smart
dispatch in multi-microgrids using heuristic algorithms. In: Smart grid with cloud computing approach using stochastic dynamic
Grid Conference, Tehran, 2014: 1–6 programming. Perspectives in Science, 2016
20. Liang H, Zhuang W. Stochastic modeling and optimization in a 37. Sobu A, Wu G. Optimal operation planning method for isolated
microgrid: a survey. Energies, 2014, 7(4): 2027–2050 micro grid considering uncertainties of renewable power genera-
21. Niknam T. Abarghooee R A, Narimani M R. An efficient scenario- tions and load demand. In: 2012 IEEE PES Innovative Smart Grid
based stochastic programming framework for multi-objective Technologies-Asia, Tianjin, 2012: 1–6
optimal micro-grid operation. Applied Energy, 2012, 99(6): 455– 38. Küster T, Lützenberger M, Voß M, Freund D, Albayrak S. Applying
470 heuristics and stochastic optimization for load-responsive charging
22. Falsafi H, Zakariazadeh A, Jadid S. The role of demand response in in a smart grid architecture. In: 2014 IEEE PES Innovative Smart
single and multi-objective wind-thermal generation scheduling: a Grid Technologies Conference Europe, Istanbul, 2014: 1–6
stochastic programming. Energy, 2014, 64(1): 853–867 39. Zhu Z, Lambotharan S, Chin W H, Fan Z. A stochastic optimization
23. Khazali A, Kalantar M. A stochastic–probabilistic energy and approach to aggregated electric vehicles charging in smart grids. In:
reserve market clearing scheme for smart power systems with plug- 2014 IEEE Innovative Smart Grid Technologies-Asia, Kuala
in electrical vehicles. Energy Conversion and Management, Lumpur, 2014: 51–56
2015,105: 1046–1058 40. Tan J, Wang L. A stochastic model for quantifying the impact of
24. Yu Z, Jia L, Murphy-Hoye M C, Pratt A, Tong L. Modeling and PHEVs on a residential distribution grid. In: IEEE International
stochastic control for home energy management. IEEE Transactions Conference on Cyber Technology in Automation, Control and
on Smart Grid, 2012, 4(4): 2244–2255 Intelligent Systems, Nanjing, 2013: 120–125
25. Romero-Ruiz J , Pérez-Ruiz J , Martin S , AguadoJ A, De la Torre S. 41. Haghi H V, Qu Z, Lotfifard S. Analytics-based optimization for
Probabilistic congestion management using EVs in a smart grid with smart grid operations. In: IEEE International Workshop on
intermittent renewable generation. Electric Power Systems Intelligent Energy Systems, San Diego, 2014: 58–63
Research, 2016, 137: 155–162 42. Venayagamoorthy G K. Dynamic, stochastic, computational, and
26. Ringler P, Keles D, Fichtner W. Agent-based modelling and scalable technologies for smart grids. IEEE Computational
simulation of smart electricity grids and markets: a literature review. Intelligence Magazine, 2011, 6(3): 22–35
Renewable & Sustainable Energy Reviews, 2016, 57: 205–215 43. Gong C, Wang X, Xu W, Tajer A. Distributed real-time energy
27. Koohi-KamaliS, RahimN A, MokhlisH, TyagiV V. Photovoltaic scheduling in smart grid: stochastic model and fast optimization.
electricity generator dynamic modeling methods for smart grid IEEE Transactions on Smart Grid, 2013, 4(3): 1476–1489
applications: a review. Renewable & Sustainable Energy Reviews, 44. Momoh J A. Adaptive Stochastic Optimization Techniques with
2016, 57(C): 131–172 Applications. Boca Raton: CRC Press, 2015
28. Soroudi A. Robust optimization based self scheduling of hydro- 45. Alguacil N, Conejo A J. Multiperiod optimal power flow using
thermal Genco in smart grids. Energy, 2013, 61(6): 262–271 benders decomposition. IEEE Transactions on Power Systems,
29. Parhoudeh S, Baziar A, Mazareie A, Kavousi-Fard A. A novel 2000, 15(1): 196–201
stochastic framework based on fuzzy cloud theory for modeling 46. Sortomme E, El-Sharkawi M A. Optimal power flow for a system of
uncertainty in the micro-grids. International Journal of Electrical microgrids with controllable loads and battery storage. Power
Power & Energy. 2016, 80: 73–80 Systems Conference & Exposition, 2009, 107(1): 1–5
 S. Surender REDDY et al. Review of stochastic optimization methods for smart grid 13

47. Delson J K, Shahidehpour S M. Linear programming applications to 59. Momoh J A, Surender Reddy S. Review of optimization techniques
power system economics, planning and operations. IEEE Transac- for renewable energy resources. IEEE Symposium on Power
tions on Power Systems, 1992, 7(3): 1155–1163 Electronics and Machines for Wind and Water Applications,
48. Hashemi-Dezaki H, Hamzeh M, Askarian-Abyaneh H, Haeri- Milwaukee, WI, 2014
Khiavi H. Risk management of smart grids based on managed 60. Momoh J A, Reddy S S, Baxi Y. Stochastic Voltage/Var control with
charging of PHEVs and vehicle-to-grid strategy using Monte Carlo load variation. In: 2014 IEEEPES General Meeting | Conference &
simulation. Energy Conversion and Management, 2015, 100: 262– Exposition, National Harbor, MD (Washington, DC Metro Area),
276 2014
49. Arnold U, Yildiz Ö. Economic risk analysis of decentralized 61. Momoh J A, Baxi Y, Idubor A O. Frame work for real time optimal
renewable energy infrastructures—a Monte Carlo simulation power flow using real time measurement tools and techniques. In:
approach. Renewable Energy, 2015, 77: 227–239 North American Power Symposium, Boston, 2011,1–7
50. Pereira E J S, Pinho J T, Galhardo M A B, Macêdo W N. 62. Wei H, Sasaki H, Kubokawa J, Yokoyama R. An interior point
Methodology of risk analysis by Monte Carlo method applied to nonlinear programming for optimal power flow problems with a
power generation with renewable energy. Renewable Energy, 2014, novel data structure. IEEE Transactions on Power Systems Pwrs,
69(3): 347–355 1998, 47(3): 14–18
51. Rosenblueth E. Point estimates for probability moments. Proceed- 63. Alguacil N, Conejo A J. Multi period optimal power flow using
ings of the National Academy of Science of the United States of benders decomposition. IEEE Transactions on Power Systems,
America, 1975, 72(10): 3812–3814 2000, 15(1): 196–201
52. Rosenblueth E. Two-point estimates in probability. Applied 64. Todorovski M. Rajiˇcic D. An initialization procedure in solving
Mathematical Modelling, 1981, 5(5): 329–335 optimal power flow by genetic algorithm. IEEE Transactions on
53. Morales J M, Pérez-Ruiz J. Point estimate schemes to solve the Power Systems, 2006, 21(2): 480–487
probabilistic power flow. IEEE Transactions on Power Systems, 65. Torres G, Quintana V H. An interior point method for nonlinear
2007, 22(4): 1594–1601 optimal power flow using voltage rectangular coordinates. IEEE
54. Verbic G, Canizares C A. Probabilistic optimal power flow in Transactions on Power Systems, 1998, 13(4): 1211–1218
electricity markets based on a two-point estimate method. IEEE 66. Jabr R A. Optimal power flow using an extended conic quadratic
Transactions on Power Systems, 2006, 21(4): 1883–1893 formulation. IEEE Transactions on Power Systems Pwrs, 2008, 23
55. Surender Reddy S, Bijwe P R, Abhyankar A R. Optimal posturing in (3): 1000–1008
day-ahead market clearing for uncertainties considering anticipated 67. Li Y, McCalley J D. Decomposed SCOPF for improving efficiency.
real-time adjustment costs. IEEE Systems Journal, 2015, 9(1): 177– IEEE Transactions on Power Systems Pwrs, 2009, 24(1): 494–495
190 68. Vovos P N, Harrison G P, Wallace A R, Bialek J W. Optimal power
56. Reddy S S, Momoh J A. Realistic and transparent optimum flow as a tool for fault level-constrained network capacity analysis.
scheduling strategy for hybrid power system. IEEE Transactions on IEEE Transactions on Power Systems, 2005, 20(2): 734–741
Smart Grid, 2015, 6(6): 1–1 69. Moon G H, Wi Y M, Lee K, Joo S K. Fault current constrained
57. Reddy S S, Bijwe P R, Abhyankar A R. Joint energy and spinning decentralized optimal power flow incorporating superconducting
reserve market clearing incorporating wind power and load forecast fault current limiter (SFCL). IEEE Transactions on Applied
uncertainties. IEEE Systems Journal, 2015, 9(1): 152–164 Superconductivity, 2011, 21(3): 2157–2160
58. Geidl M, Andersson G. A modeling and optimization approach for 70. Verbiˇc G. Cañizares C A. Probabilistic optimal power flow in
multiple energy carrier power flow. Power Technology, IEEE electricity markets based on a two-point estimate method. IEEE
Russia, 2005, 38(16): 1–7 Transactions on Power Systems, 2006, 21(4): 1883–1893