optimal power flow methods: a comprehensive survey 7... · the optimal power flow problem is...
TRANSCRIPT
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2228 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
Abstract - Power flow study (load-flow study) is a steady-state
analysis whose target is to determine the voltages, currents,
and real and reactive power flows in a system under a given
load conditions. The objective of an Optimal Power Flow
(OPF) study is to find steady state operation point which
minimizes generation cost, loss, emission etc. Over the past
half-century, OPF has become one of the most important and
widely studied nonlinear optimization problems while
maintaining an acceptable system performance in terms of
limits on generators’ real and reactive powers, line flow
limits, output of various compensating devices etc.
Traditionally, classical optimization methods were used to
effectively solve OPF. But more recently due to incorporation
of FACTS devices and deregulation of a power sector, the
traditional concepts and practices of power systems are
superimposed by an economic market management.
Therefore, OPF has become complex. In recent years,
Artificial Intelligence (AI) methods have been emerged which
can solve highly complex OPF problems. The purpose of this
paper is to present a comprehensive survey of various
optimization methods like traditional and AI methods used to
solve OPF problems.
Index Terms
Artificial Intelligence, Optimal Power Flow
I. INTRODUCTION
A first comprehensive survey regarding optimal power
dispatch was given by H.H. Happ [1] and then an IEEE
working group [2] presented bibliography survey of
major economic-security functions in 1981. Then in
1985, Carpentair [3] presented a survey and classified
the OPF algorithms based on their solution
methodology. In 1990, Chowdhury [4] presented a
survey on economic dispatch methods. In 1999, J.A.
Momoh et al. [5] presented a review of some selected
OPF techniques. In 2008, K.S. Pandya [6] made a survey
on traditional and Artificial Intelligence optimization
methods. In 2009, Qiu et al. [7] presented a survey of the
key algorithms used to solve OPF problems. In 2010,
Zhang [8] summarized the common deterministic OPF
methods. In 2012 and 2013, Stephen Frank, Ingrida
Steponavice, and Steffen Rebennack [9, 10] provided an
introduction and survey of the deterministic and
stochastic optimization methods.
II. PROBLEM FORMULATION
The optimal power flow problem is concerned with
optimization of steady state power system performance
with respect to an objective function while subjected to
various equality and inequality constraints. For optimal
active power dispatch, the objective function F (total
generation cost) is given by the following equation
- Minimize
Where,
Fт: Total quadratic cost function; it could be also a
cubic function
Pᵢ: Real power generated
Ng: Number of generation busses
aᵢ, bᵢ, cᵢ: Fuel cost coefficients for ith unit
- Subject to various constraints.
(a) Active power balance in the network
Pi (V, δ) - Pgi + Pdi = 0 (i=1, 2, ……, NB)
(b) Reactive power balance in the network
Qi (V, δ) - Qgi + Qdi = 0
(i= NV +1, NV + 2, ……, NB)
(c) Security-related constraints called soft constraints.
Limits on real power generations
Pgimin Pgi Pgimax (i=1, 2, NG)
Limits on voltage generation
ViminViVimax
(i= NV +1, NV + 2, ……, NB)
Limits on voltage angles
Optimal Power Flow Methods:
A Comprehensive Survey
Amr Khaled Khamees*, N. M. Badra*, Almoataz Y. Abdelaziz**
*Department of Engineering Physics and Mathematics, Faculty of Engineering, Ain Shams University, Cairo, Egypt
**Department of Electrical Power & Machines, Faculty of Engineering, Ain Shams University, Cairo, Egypt
Ng
i
iiiiiT PcPbaF1
2 $/hr
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2229 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
δimin δ i δ imax (i=1, 2,……, NB)
- Real power flow equations are:
Pi (V,δ) = Vi
NB
j 1
Vj(Gijcos(δ i - δ j ) + Bijsin(δ i - δ j )
- Reactive power flow equations are
Qi (V,δ) = Vi
NB
j 1
Vj(Gijsin(δ i - δ j ) - Bijcos(δ i - δ j )
where NG is the number of generator buses, NB is
number of buses, NV is number of voltage controlled
buses, Pi is active power injection into bus i, Qi is
reactive power injection into bus i, Pdi is active load on
bus i, Qdi is reactive load on bus i, Pgi is active power
generation on bus i, Qgi is reactive power generation on
bus i, Vi is magnitude of the voltage at ith bus, δi is
voltage phase angle of bus i, and Yij = Gij + jBij are the
elements of admittance matrix.
III. OPTIMAL POWER FLOW METHODS
The optimal power flow solution methods are classified
into two categories: Traditional methods and Artificial
Intelligence based methods as shown in Fig. 1.
Fig. 1 – Optimal Power Flow Solution Methods
IV. TRADITIONAL METHODS
Traditional methods are called deterministic or
conventional optimization methods.
The application of these methods had been an area of
active research in the recent past. The conventional
methods are based on mathematical programming
approaches and used to solve different size of OPF
problems. To meet the requirements of different
objective functions, types of application and nature of
constraints, the popular conventional methods are further
subdivided into the following:
1. Linear Programming
2. Gradient methods
3. Quadratic Programming
4. Newton-Raphson
5. Nonlinear Programming
6. Interior Point
Even though, excellent advancements have been made in
classical methods, they suffer from the following
disadvantages:
1. Some required linearization
2. Some required differentiability
3. May get stuck at local optimum
4. Poor convergence
5. Weak in handling qualitative constraints
6. Become too slow if number of variables are large
7. Can find only a single optimized solution in a single
simulation run
a) LINEAR PROGRAMMING
Linear programming formulation requires linearization
of objective function as well as constraints with
nonnegative variables.
Mukherjee, S.K. [11] presented a linear
programming based optimization method for optimal
power flow. Results show that a 220 kV five-bus system
convergence within only one iteration was achieved with
a considerable reduction in CPU time. The technique has
been recommended to Florida Power and Light
Company's Power System Control Division and has
proved successful for a 15 bus system using real-time
data.
T.S. Chung et al. [12] presented a linear
programming method for minimizing power loss and
finding the optimal location for capacitor in a
distribution system. Calculation is carried at 14-bus
system and this method does not require any matrix
inversion, thus saves computational time and memory
space.
E. Lobato et al. [13] presented a linear
programming based method for OPF for minimization of
OPF methods
Traditional methods
Artificial
Intelligence (AI) methods
1. Linear Programming
2. Gradient methods
3. Quadratic Programming
4. Newton-Raphson
5. Nonlinear Programming
6. Interior Point
1. Genetic algorithm
2. Particle swarm
3. Artificial neural network
4. Bee colony optimization
5. Differential evolution
6. Grey wolf optimizer
7. Shuffled frog-leaping
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2230 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
power loss and generator reactive margins for Spanish
power system. Shunt reactors and capacitors are discrete,
modeled by integer variables. Both the objective
function and the constraints are linearized in each
iteration.
Rau, N. [14, 15] presented DC power flow
formulation which involves simplifying assumptions and
linearization with a linear (or linearized) objective
function. Since the DC power flow constraint set is fully
linear, no further constraint linearization is needed. In
contrast to other linearization techniques, DC-OPF is
non-iterative; only a single solve is required to yield the
optimal solution. Because of its simplicity, speed, and
robust nature, DC-OPF is widely used in industry.
b) GRADIENT METHOD
Gradient methods were among the first attempts to solve
practical OPF problems at the end of the 1960s. Gradient
methods use the 1st order derivative vector of the
objective function of a non-linear optimization (that is,
the gradient) to determine improving directions for the
solution in iterative steps. Gradient methods are reliable,
easy to implement, and guaranteed to converge for well-
behaved functions. However, gradient methods are slow
compared to higher-order methods.
Dommel and Tinney [16] presented a gradient
method to OPF problem, using penalty techniques to
enforce the limits on the dependent variables and the
functional constraints. This work relied on 1st order
information of the (nonlinear) objective and the
constraints derived from the Jacobian matrix computed
from a conventional power flow.
Alsac and Stott [17] extended the work of
Dommel and Tinney to security-constrained OPF
(SCOPF) by adding predetermined contingency cases to
the power flow equations and penalizing security
violations of these cases in the objective function.
Peschon et al. [18] presented an application of
the Generalized Reduced Gradient (GRC) method to
OPF, presenting the benefits of the GRC method. These
benefits include the avoidance of penalty terms and the
straightforward computation of sensitivities.
c) QUADRATIC PROGRAMMING
Quadratic Programming (QP) is a special form of non-
linear programming whose objective function is
quadratic and constraints are linear.
J.A. Momoh [19] presented a generalized
Quadratic-based model for OPF. The construction of the
OPF algorithm includes the conditions for feasibility,
convergence and optimality. It is also capable of using
hierarchical structures to include multiple objective
functions. The algorithm using sensitivity of objective
functions with optimal adjustments in the constraints
yields a global optimal solution. Computational memory
and execution time required have been reduced.
N. Grudinin [20] presented a reactive power
optimization model based on Successive Quadratic
Programming (SQP) methods. Six other optimization
methods were used to test the IEEE 30-bus and 278-bus
systems. It is found that the SQP methods provide more
fast and reliable optimization in comparison with other
classical methods.
Reid and Hasdorff [21] presented QP OPF
using the Lagrange multiplier method and Taylor
expansion. Economic dispatch problem is formulated as
a quadratic programming problem and solved using
Wolfe's algorithm. The method is capable of handling
both equality and inequality constraints on p, q, and v
and can solve the load flow as well as the economic
dispatch problem. Convergence was obtained in three
iterations for all test systems considered and solution
time is small enough to allow the method to be used for
on-line dispatching at practical time intervals.
d) NEWTON – RAPHSON MEHTOD
Newton method is a second order method for
unconstrained optimization based on the application of a
2nd order Taylor series expansion about the current
candidate solution.
S. Chen et al. [22] presented Newton-Raphson
(NR) method to solve emission dispatch in real-time
with sensitivity factors incorporated. The Jacobian
matrix and the coefficients have been developed in terms
of the generalized generation shift distribution factor.
Penalty factor and the incremental losses are easily
obtained. Run time is lesser than that of the conventional
one.
K.L. Lo et al. [23] presented two Newton-like
load flow methods, the Fixed Newton method and the
modification of the right-hand-side vector method for
line outage simulation that is a part of contingency
analysis. These two methods are compared with the
Newton-based full AC load flow method and the fast
decoupled load flow to show their better convergence
characteristics.
H. Ambriz-Pérez [24] presented unified power
flow controller (UPFC) modeling within the context of
optimal power flow (OPF) by Newton Method. The
networks are modified to include several UPFCs are
solved with equal reliability. The UPFC model itself is
very flexible; it allows the control of active and reactive
powers and voltage magnitude simultaneously. It can
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2231 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
also be set to control one or more of these parameters in
any combination or to control none of them.
e) NON LINEAR PROGRAMMING
Nonlinear programming (NLP) deals with problems
involving nonlinear objective and/or constraint
functions. The main advantage of NLP formulations for
OPF is that they accurately capture power system
behavior. However, the computational and theoretical
challenges associated with NLP motivated the
development of simplified formulations, as discussed
below. In addition, certain NLP formulations redefine
the problem variables to reduce the degree of
nonlinearity.
J.A. Momoh et al. [25] presented a new
nonlinear convex network flow programming
(NLCNFP) model for solving the multi-area security
constrained economic dispatch (MAED) problem. A
combined method of quadratic programming and
network flow programming used to solve MAED
problem. The tie-line security and transfer constraints
are considered in each area. Method is tested in four
interconnected power systems.
D. Pudjianto et al. [26] presented LP and NLP
for reactive OPF for allocating reactive power among
generators in a deregulated environment. It was found
that the overall cost associated with the system reactive
requirement calculated by LP method was reasonably
accurate, but the generator’s individual commitment
may vary considerably. Whereas, NLP offers a faster
computation speed and accuracy for the solution but the
convergence could not be guaranteed for every
condition.
H. Habibollahzadeh [27] presented
hydrothermal OPF based on a combined LP and NLP
methods. Hydraulic modeling of systems with
considerable share of hydraulic generation is also
considered. It used the LP technique to help the
Generalized Reduced Gradient technique in a feasibility
adjustment step and feasible directions method for
solving nonlinear programming problems has been used.
f) INTERIOR POINT METHOD
Interior Point Methods (IPMs) are a family of projective
scaling algorithms for solving linear and nonlinear
optimization problems that constrain the search to the
feasible region by introducing barrier terms to the
objective function. In general, IPMs attempt to
determine and follow a central path through the feasible
region to the optimal solution.
Sergio Granville [28] presented IPM to the
optimal reactive power dispatch problem. It is based on
the primal dual logarithmic. This method was applied on
large power systems and it converged in 40 iterations at
CPU time 398.9 seconds. IPM advantages are: number
of iterations is not very sensitive to network size or
number of control variables, numerical robustness, hot
starting capability, no active set identification difficulties
and effectiveness in dealing with optimal reactive
allocation.
Torres, G.L. [29] presented the solution of an
optimal power flow (OPF) problem in rectangular form
by an IPM for nonlinear programming. OPF problem is
solved via a primal-dual IPM for NLP. Objective
function and constraints are quadratic. Desirable
properties of a quadratic function are: (a) its Hessian is
constant, (b) its Taylor expansion terminates at the
second-order without error, and (c) the higher-order
term is easily evaluated. Such quadratic features allow
for ease of matrix setup and inexpensive incorporation of
higher order information in a predictor-corrector
procedure that reduces the number of IPM iterations to
convergence.
Edgardo D. Castronuovo [30] presented new
versions of interior point methods applied to the optimal
power flow problem. Search for the optimum is based on
the combination of two directions: the affine-scaling and
the centralization, it is shown that the suitable
combination of these directions can increase the
potential of the optimization algorithm in terms of speed
and reliability. To solve optimization problems through
the IPM for Nonlinear Programming (NLP) a
perturbation parameter is introduced in the
complementarily Karush-Kuhn-Tucker (KKT) condition.
Hua Wei et al. [31] presented a new interior
point nonlinear programming algorithm for optimal
power flow problems (OPF) based on the perturbed
KKT conditions of the primal problem. Through the
concept of the centering direction, authors extend this
algorithm to classical power flow (PF) and approximate
OPF problems. Compared with the conventional data
structure of Newton OPF, the number of fill-ins of the
proposed scheme is roughly halved and CPU time is
reduced by about 15% for large scale systems. The
proposed algorithm includes four kinds of objective
functions and two different data structures. Extensive
numerical simulations on test systems that range in size
from 14 to 1047 buses have shown that the proposed
method is very promising for large scale application due
to its robustness and fast execution time.
V. ARTIFICIAL INTELLIGENCE METHODS
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2232 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
The intelligent search has become a very important
technique in searching the global or near-global optimal
solution. It also called Non-deterministic or stochastic
method, major advantages of these methods are:
1. Able to handle various qualitative constraints
2. Can find multiple optimal solutions in single
simulation run
3. Suitable in solving multi-objective optimization
problems
4. Can find the global optimum solution
Number of non-deterministic optimization methods have
been developed and applied to global optimization
problems to overcome the weak global search
capabilities of many conventional deterministic
optimization algorithms. Many of these techniques have
been applied to OPF problems like:
1. Genetic algorithm
2. Particle swarm
3. Artificial neural network
4. Bee colony optimization
5. Differential evolution
6. Grey wolf optimizer
7. Shuffled frog-leaping
a) GENETIC ALGORITHM
Genetic Algorithm (GA) operates on the encoded binary
string of the problem parameters rather than the actual
parameters of the system. Each string can be thought of
as a chromosome that completely describes one
candidate solution to the problem. A simple Genetic
Algorithm is an iterative procedure.
During each iteration step (generation) three genetic
operators (Selection, crossover, and mutation) are
performing to generate new populations (offspring).
Bakritzs, V. [32] presented economic dispatch problems
using genetic algorithm method. Its merits are: the non-
restriction of any convexity limitations on the generator
cost function and effective coding of GAs to work on
parallel machines. GA is superior to dynamic
programming, as per the performance observed in
Economic dispatch problem. The run time of the second
GA solution (EGA method) proportionately increases
with size of the system.
Po-Hung Chen [33] presented a large scale
economic dispatch problem by GA. He designed new
encoding technique where in, the chromosome has only
an encoding normalized incremental cost. There is no
correlation between total number of bits in the
chromosome and number of units. The unique
characteristic of Genetic Approach is significant in big
and intricate systems which other approaches fails to
accomplish. Dispatch is made more practical by
flexibility in GA, due to consideration of network losses,
ramp rate limits and prohibited zone’s avoidance. This
method takes lesser time compared to Lambda –iteration
method in big systems.
M. Younes and M. Rahl [34] presented hybrid
Genetic Algorithm (combination of GA and Mat power)
that was used to solve OPF including active and reactive
power dispatches. The method uses GA to get a close to
global solution and the package of MATLAB – m files
for solving power flow and optimal power flow problem
(mat power) to decide the optimal global solution. Mat
power is employed to adjust the control variables to
attain the global solution. The method was validated on
the modified IEEE 57–bus system and the results show
that the hybrid approach provides a good solution as
compared to GA or Mat power alone.
M. Sailaja Kumari [35] presented OPF problem
formulated as a multi-objective optimization problem,
where optimal control settings for simultaneous
minimization of fuel cost and loss, loss and voltage
stability index, fuel cost and voltage stability index and
finally fuel cost, loss and voltage stability index are
obtained, it combines a new Decoupled Quadratic Load
Flow (DQLF) solution with Enhanced Genetic
Algorithm (EGA) to solve the OPF problem. A Strength
Pareto Evolutionary Algorithm (SPEA) based approach
with strongly dominated set of solutions is used to form
the pareto-optimal set.
b) PARTICLE SWARM OPTIMIZATION
Particle Swarm Optimization (PSO) is a population-
based stochastic optimization Technique. PSO is based
on processes arising naturally in socially organized
colonies such as flocks of birds and schools of fish. PSO
exploits a population of individuals to explore promising
regions within the search space. In the search procedure,
each individual (particle) moves within the decision
space over time and changes its position in accordance
with its own best experience and the current best particle
according to the model shown in Fig. 2.
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2233 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
Fig. 2 - Particle Swarm Model [36]
El-Gallad et al. [37] presented PSO to solve the
traditional economic dispatch problem. The objective
function was formulated as a combination of piecewise
quadratic cost functions with non-differential regions,
instead of adopting a single convex function for each
generating unit. This innovation in problem formulation
is due to the incorporation of practical operating
conditions, like valve-point effects and fuel types where
system demand and the balance of power, with network
losses incorporated and the generating capacity limits.
Djillani Ben Attous [38] presented economic
dispatch problem with a valve point effects. A valve
point effects is the rippling effects added to the
generating unit curve when each steam admission valve
in a turbine starts to open. Proposed method using PSO
has been examined and tested for standard IEEE 30-bus
system, it has been found that PSO method is highly
competitive for its better general convergence
performance.
Sanjeev Kumar [39] presented PSO and
integration of GA-fuzzy and PSO-fuzzy optimization to
calculate optimal power flow. The four algorithms GA,
GA-fuzzy, PSO, and FPSO are tested for solving
modified IEEE 30 bus test system. It is found that
integration of fuzzy with Genetic Algorithm and particle
swarm optimization techniques improve the performance
of average fitness obtained with the help of optimal
power flow method. The results of integrated GA-Fuzzy
and PSO-fuzzy are compared with simple GA and PSO
and found synergetic approaches are better.
T. Saravanan [40] presented the application of
PSO technique to solve OPF with inequality constraints
on line flow. Algorithm is implemented on a six-bus
three-unit system and the results are compared with
linear programming method.
J. Praveen and B. Srinivasa Rao [41] presented
optimization problem solved by PSO with power
injection model of the FACTS device. The proposed
methodology is tested on standard IEEE 30-bus test
system and the results are compared for single objective
optimization with and without FACTS device.
c) ARTIFICIAL NEURAL NETWORK
Artificial neural network (ANN) is an interconnected
group of artificial neurons that use a mathematical model
or computational model for information processing
based on a connection list approach to computation.
Mat Syai’in [42] presented a novel algorithm of an
optimal power flow (OPF). The proposed algorithm uses
neural networks (NNs) to model the generator capability
curves and set them as the output power constraints of
the generators. In addition, it also uses NNs to replace an
OPF based on the particle swarm optimization (PSO)
method so as to run in real time. Also, in order for the
proposed algorithm to be able to account for various
load conditions, a 500 kV Java-Bali power system
consisting of 23 buses is used as a benchmark system to
validate the proposed NN-based OPF. The simulation
results show that that the values obtained from the
proposed algorithm are in great agreement with those
calculated from the PSO-OPF. Method procedure is
shown in Fig. 3.
Nakawiro and Erlich [43] presented a speedup
strategy for OPF that uses a dedicated ANN to perform
the function of a power flow program. The ANN is
combined with a GA, which performs the optimization.
Tests show that their method significantly speeds up the
solution process compared to GAs alone, while
providing solutions of similar quality.
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2234 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
Fig. 3 - Flowchart design of ANN-OPF [41]
Chowdhury [44] presented Integrated Security
Constrained Optimal Dispatch (ISCOD) which could
solve the OPF problem when it was constrained by both
static and dynamic security. ISCOD utilized the
diagnostic and decision making capabilities of
Knowledge Base System (KBS), massive parallellisms
and learning features of an ANN along with
conventional power system network solution
methodologies to provide real-time control and
optimization. The KBS and the ANN are used
indifferent configuration for adding the dispatch or in
making control decisions.
d) BEE COLONY OPTIMIZATION
Artificial Bee colony (ABC) simulates the intelligent
foraging behavior of a honeybee swarm. ABC is simple
in concept, easy to implement, and has fewer control
parameters. Foraging artificial bees are divided into
three groups:
1. Employed bee
It stays on a food source and provides the neighborhood
of the source in its memory.
2. Onlooker bee
It gets the information of food sources from the
employed bees in the hive and selects one of the food.
3. Scout bee
It is responsible for finding new food, the new nectar
sources. The process is shown in Fig. 4.
Fig. 4 - Foraging behavior of honeybees to find food
sources [45]
Sumpavakup et al. [46] presented ABC to solve
the optimal power flow problems. The effectiveness of
the Bees Algorithm was verified by testing with IEEE-
14 and IEEE-30 bus system. The technique was also
compared with other swarm intelligence techniques
namely, Genetic Algorithms and Particle Swarm
Optimization and its simulation results show that Bees
Algorithm can converge towards the better solution
slightly faster than the rest methods.
Khorsandi [47] presented a fuzzy based
modified artificial bee colony (MABC) algorithm to
solve discrete optimal power flow (OPF) problem that
has both discrete and continuous variables considering
valve point effects. The OPF problem is formulated as a
multi-objective mixed-integer nonlinear problem, where
optimal settings of the OPF control variables for
simultaneous minimization of total fuel cost of thermal
units, total emission, total real power losses, and voltage
deviation are obtained. The proposed approach is applied
to the OPF problem on IEEE 30-bus and IEEE 118-bus
test systems. The performance and operation of the
proposed approach is compared with the conventional
methods. The results confirm that the MABC algorithm
is more effective in global search exploration and faster
than the other algorithms.
A. N. Afandi and H. Miyauchi [48] developed a
Harvest Season Artificial Bee Colony algorithm to
improve performance in terms of the search mechanism
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2235 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
and convergence speed. The effectiveness of the
proposed method was tested on IEEE-62 bus system and
compared to those obtained by ABC, SFABC, SBABC,
MOABC and IABC. The numerical results showed that
the proposed method reduced the number of iterations.
e) DIFFERENTIAL EVOLUTION
Differential Evolution (DE) is a population-based, direct
stochastic search algorithm. DE combines simple
arithmetic operators with the classical evolutionary
operators of crossover, mutation and selection to evolve
from a randomly generated starting population to a final
solution. DE uses a greedy, rather than stochastic,
approach to solve the problem. DE procedure is shown
in Fig. 5.
Fig. 5 - DE flow chart [48]
Coelho and Mariani [49] presented improved
DE algorithms to solve economic dispatch problems of
electric energy that takes into account nonlinear
generator features such as ramp rate limits and
prohibited operating zones. Two economic dispatch
problems with 6 and 15 thermal units with ramp rate
limits and prohibited operating zones in the power
system operation and transmission loss are employed to
demonstrate the performance of this method. DE
algorithms offer potential advantages: they find the true
global minimum regardless of the initial parameter
values, and they display fast convergence and use few
control parameters.
M. A. Abido [50] presented a multi-objective
differential-evolution-based approach to solve the
optimal power flow (OPF) problem where different
objective functions and operational constraints have
been considered and also clustering algorithm is applied
to manage the size of the Pareto. An algorithm based on
fuzzy set theory is used to extract the best compromise
solution. Results on IEEE 30-bus and IEEE 118-bus
standard test systems show the effectiveness of the
proposed approach in solving true multi-objective OPF
and also finding well-distributed Pareto-optimal
solutions.
Sayah and Zehar [51] presented DE for solving
OPF with non-smooth and non-convex generator fuel
cost curves. They consider effective modifications in the
mutation, enhancing the convergence rate while
improving the solution quality. They also showed that
their modified DE algorithm outperforms the classical
DE algorithms in global convergence speed and obtains
similar results compared to Evolutionary Programming
and Tabu Search methods.
f) GREY WOLF OPTIMIZER
Grey wolf optimizer (GWO) is a population based
heuristics algorithm simulates the leadership hierarchy
and hunting mechanism of gray wolves in nature.
Privileged wolves are alpha wolves – decision makers,
then beta wolves – alpha assistant, and then delta wolves
– lowest ranking and omega are inferior wolves
preceded by scoffed kappa and lambda as in Fig. 6.
Fig. 6 - The hierarchy of the grey wolves
El-Fergany and Hasanien [52] presented the
grey wolf optimizer to solve the optimal power flow
(OPF) problem. The indicator of the static line stability
index is incorporated into the OPF problem. The fuzzy-
based Pareto front method is tested to find the best
compromise point of multi-objective functions.
Algorithm used to determine the optimal values of the
continuous and discrete control variables. These
algorithms are applied to the standard IEEE 30-bus and
118-bus systems with different scenarios.
Sulaiman et al. [53] presented GWO in solving the
optimal reactive power dispatch (ORPD) problem. Two
case studies of IEEE 30-bus system and IEEE 118-bus
system are used to show the effectiveness of GWO
technique compared to other techniques. The results of
this research show that GWO is able to achieve less
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2236 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
power loss and voltage deviation than those determined
by other techniques.
g) SHUFFLED FROG-LEAPING
Shuffled frog Leaping algorithm (SFLA) is memetic
heuristic that is designed to seek a global optimal
solution by performing a heuristic search based on the
evolution of memes carried by individuals and a global
exchange of information among the population. It
combines the benefits of the local search tool of the
particle swarm optimization and the idea of mixing
information from parallel local searches to move toward
a global solution. Shuffling of frogs in memeplexes is
shown in Fig. 7.
Fig. 7 - Shuffled behavior of leaping frogs in search of
food [54]
R. Jahani [55] presented a new method based
on shuffled frog leaping algorithm for distributing the
power optimized distribution to minimize power
production costs and to optimize allocation of every
powerhouse share in addition to providing required
power for a network. An inferior purpose function is
expressed via the basis of units’ productive power and
restrictions are modeled in the form of linear equalities
and inequality equation. The presented method is
simulated on a standard IEEE 30-bus system. Results
demonstrate that the algorithm has great potential in
functions optimization.
Taher Niknam [56] presented a modified
shuffle frog leaping algorithm for multi-objective
optimal power flow by considering the emission issue. A
modified SLFA called MSLFA algorithm which profits
from a mutation in order to reduce the processing time
and improve the quality of solutions, particularly to
avoid being trapped in local optima. MSLFA algorithm
is applied to the 30-bus IEEE test system. The results
showed that MSFLA method is capable of obtaining
accurate and acceptable solutions
K. Lenin [57] presented an algorithm based on
Modified Shuffled Frog-Leaping Optimization for
solving the multi-objective reactive power dispatch
problem considering various generator constraints in a
power system. This method formulates reactive power
dispatch problem as a mixed integer non-linear
optimization problem and determines control strategy
with continuous and discrete control variables. This
paper introduces a new search-acceleration parameter
into the formulation of the original shuffled frog leaping
algorithm to create a modified form of the shuffled frog
algorithm.
VI. CONCLUSION
This paper presents a review for various optimization
methods used to solve OPF problems. Even though
excellent advancements have been made in classical
methods they suffer from the following disadvantages:
some required linearization and differentiability, may get
stuck at local optimum, poor convergence, weak in
handling qualitative constraints and become too slow if
number of variables are large whereas, the major
advantage of the AI methods is that they are relatively
versatile for handling various qualitative constraints. AI
methods can find multiple optimal solutions in single
simulation run. So they are quite suitable in solving
multi-objective optimization problems. In most cases,
they can find the global optimum solution. Table 1
shows advantages and disadvantages of AI methods
presented in this survey.
Authors strongly believe that this survey article will be
very much useful to the researchers for finding out the
relevant references as well as the previous work done in
the field of OPF methods, so that further research work
can be carried out.
REFERENCES
[1] H. H. Happ, “Optimal power dispatch-A comprehensive survey”, IEEE Trans. Power Apparat. Syst.,vol. PAS-90, pp. 841-854, 1977.
[2] IEEE working group, “Description and bibliography of major economic-security functions part-II and III, IEEE Trans. Power Apparat. Syst., vol. PAS-100, pp. 215-235, 1981.
[3] J. Carpentier, “Optimal power flow, uses, methods and development”, Planning and operation of electrical energy system Proc. of IFAC symposium, Brazil, 1985, pp. 11-21.
[4] B. H. Chowdhury and Rahman, “Recent advances in economic dispatch”, IEEE Trans. Power Syst., no. 5, pp. 1248-1259, 1990.
[5] J. A. Momoh, M. E. El-Harwary and Ramababu Adapa, “A review of selected optimal power flow literature to 1993, part- I
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2237 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
and II”, IEEE Trans. Power Syst., vol. 14, no. 1, pp. 96-111, Feb. 1999.
[6] Pandya, K., Joshi, S.: A survey of optimal power flow methods. J. Theor. Appl. Inf. Technol. 4(5), 450–458 (2008)
[7] . Qiu, Z., Deconinck, G., Belmans, R.: A literature survey of optimal power flow problems in the electricity market context. In: IEEE/PES Power Systems Conference and Exposition. PSCE’09, Seattle, WA, pp. 1–6 (2009)
[8] Zhang, X.P.: In: Fundamentals of Electric Power Systems, pp. 1–52. Wiley, New York (2010). doi:10.1002/9780470608555.ch1
[9] Stephen Frank, Ingrida Steponavice, and Steffen Rebennack. Optimal power flow: a bibliographic survey, I: formulations and deterministic methods. Energy Systems, 3:221–258, September 2012.
[10] Stephen Frank, Ingrida Steponavice, and Steffen Rebennack. Optimal power flow: a bibliographic survey, II: nondeterministic and hybrid methods. Energy Systems, 3:259–289, September 2013.
[11] Mukherjee, S.K. Optimal power flow by linear programming based optimization, Southeastcon '92, Proceedings., IEEE , 527 - 529 vol.2, 1992
[12] T. S. Chung and Ge Shaoyun, “A recursive LP-based approach for optimal capacitor allocation with cost-benefit consideration”, Electric Power Syst. Research, vol. 39, pp. 129-136, 1997.
[13] [1] E. Lobato, L. Rouco, M. I. Navarrete, R. Casanova and G. Lopez, “An LP-based optimal power flow for transmission losses and generator reactive margins minimization”, in Proc. of IEEE porto power tech conference, Portugal, Sept. 2001
[14] Rau, N.: Issues in the path toward an RTO and standard markets. IEEE Trans. Power Syst. 18(2), 435–443 (2003)
[15] Rau, N.: Optimization Principles: Practical Applications to the Operation and Markets of the Electric Power Industry. Wiley/IEEE Press, Hoboken (2003)
[16] [2] Dommel, H., Tinney, W.: Optimal power flow solutions. IEEE Trans. Power Appar. Syst. 87(10), 1866–1876 (1968)
[17] Alsac, O., Stott, B.: Optimal load flow with steady-state security. IEEE Trans. Power Appar. Syst. BPAS-93(3), 745–751 (1974)
[18] Peschon, J., Bree, D., Hajdu, L.: Optimal power-flow solutions for power system planning. Proc. IEEE 6(1), 64–70 (1972)
[19] J. A. Momoh, “A generalized quadratic-based model for optimal power flow”, CH2809- 2/89/0000-0261 $1.00 © 1989 IEEE, pp. 261- 267.
[20] N. Grudinin, “Reactive power optimization using successive quadratic programming method”, IEEE Trans. Power Syst., vol. 13, no. 4, pp. 1219-1225, Nov. 1998.
[21] G. F. Reid and L. Hasdorff, “Economic dispatch using quadratic programming,” IEEE Transactions on Apparatus and Power Systems, Vol. 92, Issue: 6, Nov. 1973, pp. 2015-2023.
[22] S. D. Chen and J. F. Chen, “A new algorithm based on the Newton-Raphson approach for real-time emission dispatch”, Electric Power Syst. Research, vol. 40, pp. 137-141, 1997.
[23] K. L. Lo and Z. J. Meng, “Newton-like method for line outage simulation”, IEE proc.- Gener. Transm. Distrib., vol. 151, no. 2, pp. 225-231, March 2004.
[24] H. Ambriz-Pérez ; E. Acha ; C.R. Fuerte-Esquivel ; A. De la Torre , Incorporation of a UPFC model in an optimal power flow using Newton's method, IEE Proceedings - Generation, Transmission and Distribution, Volume 145, Issue 3, May 1998,
[25] J. A. Momoh and J. Zhu, “Multi-area power systems economic dispatch using nonlinear convex network flow programming”, Electric Power Syst. Research, vol. 59, pp. 13-20, 2001.
[26] D. Pudjianto, S. Ahmed and G. Strbac, “Allocation of VAR support using LP and NLP based optimal power flows”, IEE proc.- Gener. Transm. Distrib., vol. 149, no. 4, pp. 377-383, July 2002.
[27] Habibollahzadeh, Hooshang, Guang-Xiang Luo, and Adam Semlyen. "Hydrothermal optimal power flow based on a combined linear and nonlinear programming methodology." Power Systems, IEEE Transactions on 4.2 (1989): 530-537.
[28] S. Granville, “Optimal reactive dispatch through interior point methods”, IEEE Trans. Power Syst., vol. 9, no. 1, pp. 136-146, Feb. 1994.
[29] Torres, G.L , An interior-point method for nonlinear optimal power flow using voltage rectangular coordinates Power Systems, IEEE Transaction , 13 , 1211 – 1218, Nov 1998
[30] Castronuovo, Edgardo D., Jorge M. Campagnolo, and Roberto Salgado. "New versions of interior point methods applied to the optimal power flow problem." IEEE Transactions on Power Systems 16 (2001): 3.
[31] Hua Wei, An interior point nonlinear programming for optimal power flow problems with a novel data structure , Power Systems, IEEE Transactions on , vol 13 , 870 – 877, 1998.
[32] A. Bakritzs, V. Perirtridis and S. Kazarlis , Genetic Algorithm Solution to the Economic Dispatch Problem , IEE Proc,-Generation Transmission Distribution, vol. 141, no. 4, pp. 377-382, July 1994
[33] Po-Hung Chen and Hong-Chan Chang , Large-Scale Economic Dispatch by Genetic Algorithm Discover , IEEE Transactions on Power Systems, Vol. 10, no. 4, pp. 1919 – 1926, Nov. 1995.
[34] M.Younes, M. Rahli and L. Abdelhakeem-Koridak , Optimal Power based on Hybrid Genetic Algorithm Journal of Information Science and Engineering, vol. 23, pp.1801-1816, Jan 2007.
[35] M. Sailaja Kumari and Sydulu Maheswarapu, Enhanced Genetic Algorithm based computation technique for multi-objective Optimal Power Flow solution, Electrical Power and Energy Systems 32 (2010) 736–742
[36] Hosseini, H. , Shahbazian, M. , & Takassi, M. A. (2014). The Design of Robust Soft Sensor Using ANFIS Network. Journal of Instrumentation Technology, 2(1), 9-16.
[37] A. I. El-Gallad, M. El-Hawary, A. A. Sallam, and A. Kalas, “Swarm intelligence for hybrid cost dispatch problem,” in Proc. Canadian Conf. Elect. Comput. Eng., 2001, vol. 2, pp. 753–757
[38] Djillani BEN ATTOUS , Particle swarm optimization based optimal power flow for units with non-smooth fuel cost functions, Electrical and Electronics Engineering, 2009, IEEE Xplore
[39] Sanjeev Kumar and D.K. Chaturvedi, Optimal power flow solution using fuzzy evolutionary and swarm optimization. Electrical Power and Energy Systems 47 (2013) 416–423,2013
[40] T. Saravanan and G. Saritha and V. Srinivasan , Optimal Power Flow Using Particle Swarm Optimization , Middle-East Journal of Scientific Research 20 (11): 1554-1560, 2014
[41] J. Praveen and B. Srinivasa Rao. (2014). Single objective optimization using PSO with Interline Power Flow Controller. International Electrical Engineering Journal, Vol. 5, No. 12, pp. 1659-1664
[42] Mat Syai’in and Adi Soeprijanto, New Algorithm for Neural Network Optimal Power Flow (NN-OPF) including Generator Capability Curve Constraint and Statistic-fuzzy Load Clustering, International Journal of Computer Applications (0975 – 8887), Vol. 36, 2011.
[43] Nakawiro,W., Erlich, I.: A combined GA-ANN strategy for solving optimal power flow with voltage security constraint. In:
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2238 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
Asia-Pacific Power and Energy Engineering Conference (APPEEC) (2009).
[44] Badrul H. Chowdhury, “Towards the concept of integrated security: optimal dispatch under static and dynamic security constraints”, Electric Power Syst. Research, vol. 25, pp. 213-225, 1992.
[45] Sharma, T. K. (2012). Improved Local Search in Artificial Bee Colony using Golden Section Search. Journal of Engineering, 1(1), 14-19
[46] Sumpavakup C, Srikun I, Chusanapiputt S (2010) A solution to the optimal power flow using artificial bee colony algorithm. In: 2010 international conference on power system technology (POWERCON), pp 1–5
[47] Khorsandi A, Hosseinian SH, Ghazanfari A. Modified artificial bee colony algorithm based on fuzzy multi-objective technique for optimal power flow problem. Electr Power Syst Res 2013; 95:206–13.
[48] Afandi, A. N., & Miyauchi, H. (2014). Improved artificial bee colony algorithm considering harvest season for computing economic dispatch on power system. IEEJ Transactions on Electrical and Electronic Engineering, 9(3), 251-257.
[49] Coelho, L., Mariani, V.: Improved differential evolution algorithms for handling economic dispatch optimization with generator constraints. Energy Convers. Manag. 48, 1631–1639 (2007)
[50] [3] M. A. Abido and N. A. Al-Ali , Multi-Objective Optimal Power Flow Using Differential Evolution, Arab J Sci Eng (2012) 37:991–1005
[51] Sayah, S., Zehar, K.: Modified differential evolution algorithm for optimal power flow with nonsmooth cost functions. Energy Convers. Manag. 49, 3036–3042 (2008)
[52] Attia A. El-Fergany & Hany M. Hasanien Single and Multi-objective Optimal Power Flow Using Grey Wolf Optimizer and Differential Evolution Algorithms, Electric Power Components and Systems, 43:13, 1548-1559,2015
[53] Mohd Herwan Sulaiman, Zuriani Mustaffa, Mohd Rusllim Mohamed, Omar Aliman (2015) “Using the gray wolf optimizer for solving optimal reactive power dispatch problem” Applied Soft Computing, Vol. 32, pp: 286-292
[54] Afzalan, E., Taghikhani, M (2012). Optimal placement and sizing of dg in radial distribution networks using sfla.International Journal of Energy Engineering, 2(3), 73-77.
[55] Jahani, R., H. Chahkandi Nejad, and H. Ghiasi Zadeh. "Applying Shuffled Frog Leaping Algorithm for Optimal Power Distribution." Middle-East Journal of Scientific Research 8.1 (2011): 150-154
[56] Niknam, Taher, et al. "A modified shuffle frog leaping algorithm for multi-objective optimal power flow." Energy 36.11 (2011): 6420-6432.
[57] K. Lenin , B.Ravindranath Reddy ,M.Surya Kalavathi , A Modified Shuffled Frog-Leaping Optimization Algorithm for Solving Optimal Reactive Power Dispatch Problem, International Electrical Engineering Journal, Vol. 4 (2013) No. 2, pp. 1049-1058
[58] Wong, L. I., Sulaiman, M. H., and Mohamed, M. R. (2015, August). Solving Economic Dispatch Problems with Practical Constraints Utilizing Grey Wolf Optimizer. In Applied Mechanics and Materials (Vol. 785, pp. 511-515). Trans Tech Publications.
[59] Saber, A. Y. & Venayagamoorthy, G. K. (2008, September). Economic load dispatch using bacterial foraging technique with particle swarm optimization biased evolution. In Swarm Intelligence Symposium, 2008. SIS 2008. IEEE(pp. 1-8). IEEE.
[60] Gerhardt, E., & Gomes, H. M. (2012, July). Artificial bee colony (ABC) algorithm for engineering optimization problems. In International Conference on Engineering Optimization (pp. 1-11).
[61] Selvi, V., & Umarani, D. R. (2010). Comparative analysis of ant colony and particle swarm optimization techniques. International Journal of Computer Applications (0975–8887), 5(4).
[62] Tu, Jack V. "Advantages and disadvantages of using artificial neural networks versus logistic regression for predicting medical outcomes." Journal of clinical epidemiology 49.11 (1996): 1225-1231.
[63] Bakirtzis, Anastasios G., et al. "Optimal power flow by enhanced genetic algorithm." Power Systems, IEEE Transactions on 17.2 (2002): 229-236.
International Electrical Engineering Journal (IEEJ)
Vol. 7 (2016) No. 4, pp. 2228-2239
ISSN 2078-2365
http://www.ieejournal.com/
2239 Khamees et. al., Optimal Power Flow Methods: A Comprehensive Survey
AI method Advantages Disadvantages
Genetic algorithm
- Capability of solving any optimization
problem based on chromosome approach
- Capability to handle multiple solution
search spaces
- Less complex and more straightforward
compared to other algorithms
- Does not always come with global optimum
all the time.
- A tool is dependent on computers speed and
only real time application can produce quick
respond time.
- To a non-professional, the return encoded
result maybe not be able to understand
Particle swarm optimization
- Simple concept and easy implementation
- Robust in controlling the few parameters,
computationally efficient and requires less
memory
- It can be easily applied to nonlinear non-
continuous optimization problem
- It gets trapped in local optima when
handling heavily constraint problems due to
limited local/global searching capabilities
- Updating is performed without considering
quality of solutions and the distance
between solutions
Artificial neural network
- Less formal statistical training,
- Ability to detect all possible interactions
- Availability of multiple training
algorithms.
- Its “black box” nature
- Greater computational burden, proneness to
over-fitting
- The empirical nature of model development.
Bee colony optimization
- It has few parameters
- It is a global optimizer
- Flexible
- High computational time
Differential evolution
- Few control variables and fast
convergence
- Stable and simple
- The convergence to proper target is very
late
Grey wolf optimizer
- Easy to implement due to its simple
structure
- Faster convergence
- Few parameters to control
- Avoids local optima
- The algorithm is still under research and
development
Shuffled frog-leaping
- Robust, accurate, efficient and fast
- It combines the profits of the local search
tool of PSO and the idea of mixing
information from parallel local searches to
move toward a global solution
- It gets trapped in local optima
- The convergence to proper target is very
late
Table 1 - Advantages and disadvantages of AI methods [57]-[62]