IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO.

4, NOVEMBER 2013 4885

Foreword for the Special Section on Analysis and

Simulation of Very Large Power Systems

E LECTRIC power systems across the world are facing sig-

nificant challenges at several levels due to the increasing
penetration of distributed energy resources (DER), changes
in types of loads at customer facilities, opening of electricity
market for cross border trading and more and more reliance
on HVDC technology for bulk power transfer and wide area
monitoring systems (WAMS) for network state estimation and
control. These changes are affecting the way electrical power
systems are managed and operated ranging from technical is-
sues related with network planning and operation to economic
and regulatory aspects. The temporal and spatial uncertainties Fig. 1. Number of papers in the area of “large system” analysis published over
associated with the large-scale integration of stochastic and the years in the IEEE TRANSACTIONS ON POWER SYSTEMS (and its predecessor,
IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS).
intermittent DER and opening of electricity market bring
significant operational challenges. These may lead to violation The new generation technologies, new types of loads and new
of system technical constraints such as inadmissible voltage power electronics interfaces for both, load and generation, as
profiles or transmission line congestion and overload. well as increased uncertainties in both network modeling and
The drive towards interconnection of the national networks operation require new models of the network and its compo-
in the past had the major role of sharing generation reserve in nents and faster and more accurate computational algorithms to
case of tripping of a generating unit. The savings in genera- fully harness possibilities offered by this new technology.
tion capacity and the possibility to install larger, more efficient In short, we are quickly shifting from a paradigm in which
units in the smaller utilities paid for the investment of the in- physically interconnected systems are planned and operated es-
terconnection lines. The consequences of the existing rules that sentially as if they were made up of isolated subsystems or
govern interconnected power systems at the operation level in- areas, to a more complex multilateral world where tie-lines will
clude: 1) The flows on the interconnectors were normally low play an important role in electricity trading. Duly coordinating
or zero; 2) In case of major disturbances, the interconnections those remote transactions, so that security of operation remains
were intentionally tripped in order to avoid the extension of a within current standards, requires that faster and more sophis-
possible blackout; 3) The N-1 network security was managed ticated software tools be developed, specifically designed for
at national level; 4) Very few data exchange between countries large-scale multi-area systems.
was needed. Indeed, researchers in the power community have shown a
The liberalization of the electricity market and the imple- sustained interest in computational issues arising in the analysis
mentation of the Internal Electricity Market (IEM) in Europe and simulation of large-scale power systems, since the very be-
make use of the total available cross border transfer capacity and ginning of the digital computer era (see for instance the special
call for reinforcement of the transmission network. The integra- issue [1], with outstanding contributions from pioneers). This
tion of a large amount of DER, either dispersed in the distribu- involves a number of numerical techniques of common interest
tion system or remotely located far away from the load centers, to both engineers and mathematicians, such as repeated solution
requires network reinforcement, cross border load balancing, of large, sparse systems of equations, eigensystem analysis, nu-
much more sophisticated grid operation and operation planning, merical integration of differential equations, parallel computa-
and more actively controlled distribution networks. The trans- tion, clustering and tearing, etc. Interested readers, particularly
mission networks are therefore required to operate much closer the junior ones, are referred to the proceedings of the SIAM-
to their physical limits. In order to facilitate such network oper- sponsored conference [2], where a nice overview of the state-of-
ation and to ensure system integrity at all times, the network op- the-art was jointly provided by both research communities.
erator needs not only to be fully aware of the state of the system An analysis of the IEEE Xplore data base reveals that
but also to extend this familiarity to system wide phenomena over 500 papers have been published so far in the IEEE
that may endanger secure operation of the network. TRANSACTIONS ON POWER SYSTEMS (and its predecessor,
Proliferation of new technologies, in particular information IEEE TRANSACTIONS ON POWER APPARATUS AND SYSTEMS)
and communication technology (ICT) and power electronics, containing “large scale” in their metadata (title, abstract, and
facilitates greater flexibility of interconnected power network. keywords). As shown in Fig. 1, there clearly was a break-
through in the 1980s, probably explained by the advent of
Digital Object Identifier 10.1109/TPWRS.2013.2282839 microcomputer technology.

0885-8950 © 2013 IEEE

4886 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 28, NO. 4, NOVEMBER 2013

The 25 most-cited papers in this category, according to IEEE taking into account uncertainties about the next day genera-
Xplore, deal with a diversity of topics, as follows: tion/load scenario. By ranking contingencies into different clus-
Unit commitment ters according to the type of control actions needed to cover the
Models and analysis tools for transient stability worst uncertainty pattern, and using a fixed point algorithm that
Expansion planning loops over a discrete bi-level program (BLV) and a customized
Optimal power flow security constrained optimal power flow (SCOPF), they solve
Economic dispatch the SCOPF using a MILP solver while the BLV is decomposed
Network-constrained market clearing algorithms into a series of MILPs.
Stochastic and risk-based security analysis Efficient handling of large power flow problems by
Reactive power dispatch Newton-Krylov methods is demonstrated in the paper by
Idema et al. The combination of incomplete factorizations
SVC optimal location for oscillation damping
as pre-conditioner, inner-outer Krylov methods and Al-
Determination of maximum loadability
gebraic Multigrid adopted in this work demonstrates that
Power system restoration
Newton-Krylov methods can lead to power flow solvers that
State estimation
scale linearly in the problem size and are many times faster
This special section, aimed at gathering the most recent works
than traditional methods for large power flow problems.
on analysis and simulation of large-scale systems, is composed
of 10 papers, submitted from North and South America, Eu- B. Computational Algorithms for Dynamic Analysis of Large
rope, and Asia. The topics include novel computational algo- Power Networks
rithms and hardware for efficient handling of steady state and
The efficient numerical algorithm for time domain simula-
dynamic simulations of large networks, clustering methodolo-
tions proposed by Fu et al. is capable of handling online as-
gies to split a large network into several smaller and easier to
sessment of system dynamic performance. It focuses on con-
handle sections that can be analyzed more efficiently and new
struction strategies of differential and algebraic sets of equa-
tools for electricity market modeling and analysis in large net-
tions of system model, different integration methods, nonlinear
works. Each of the above areas is discussed in a separate section and linear solvers. The performance of the algorithm, i.e., speed
below. up of time-domain simulations while preserving acceptable nu-
merical accuracy, is demonstrated on multi-thousand bus power
A. Computational Algorithms for Steady State Analysis of systems.
Large Power Networks Similarly, Fabozzi et al. develop novel algorithms for faster
dynamic simulations by applying staged decomposition of
The paper by Pinto et al. presents an efficient optimized par-
sparse linear system model. The improvement in computa-
allel algorithm based on stochastic dual dynamic programming
tional time without compromising the accuracy of the solution
applied to the problem of large-scale hydrothermal system op-
is brought forward by first decomposing the system model
eration planning. The planning horizon is broken down into
according to its bordered block diagonal structure in order to
several stages and the hydrothermal operation problem is mod-
handle only the relevant parts (solve partial problem or update
eled as a linear programming (LP) problem while the dual vari-
relevant submatrices of the Jacobian). Following this, the
ables associated with the solution are used to construct the Ben-
Jacobian structure is further exploited to localize the system
ders cuts. The problems distribution is performed by a dynamic
response, i.e., involve only the components identified as active,
scheduler that communicates with the other processes informing
with an acceptable and controllable decrease in accuracy.
which LP problem should be solved. Several optimizations were
The work of Mantzaris et al. applies a reduced order, ex-
incorporated in the algorithm resulting in an efficient parallel
tended quasi-steady-state model of long-term dynamics to the
implementation. The algorithm speeds up the computation for
analysis of low-frequency interarea oscillations in the South
a planning horizon of 5 years for the Brazilian power system
Eastern part of the European system, with the emphasis on
about 40 times on 64 cores and about 50 times on 128 cores.
the contribution of hydro and steam-turbine governors on
The improvement in solving unit commitment problem (UC)
oscillation damping and in particular on the pronounced effect
in large systems is presented in the work of Morales-España
of steam-turbine governor dead-band. Indicative power system
et al. Morales-España et al. apply improved, reformulated
stabilizer (PSS) design for the Hellenic System generators
“classical”, mixed-integer linear programming (MILP) algo-
to damp the low-frequency European interarea mode is also
rithm that is simultaneously tight and compact. The proposed
presented relying on sparsity for better handling of large
algorithm significantly reduces the computational burden and
system matrices and selective calculation of eigenvalues and
provides overall speedup above 10 times compared with tradi-
eigenvectors.
tional unit commitment solvers.
The application of MILP to a very large power system is fur-
C. Electricity Market Integration and Analysis
ther demonstrated in the paper of Fliscounakis et al. on a very
large European system model with 9241 buses and 5126 con- In the paper of Biskas et al., a centralized market-splitting
tingencies. They address the issue of day-ahead security man- algorithm is implemented comprising both power pools and
agement with respect to a postulated set of contingencies, while power exchanges, with each local/national market respecting
FOREWORD FOR THE SPECIAL SECTION ON ANALYSIS AND SIMULATION OF VERY LARGE POWER SYSTEMS 4887

the standard constraints imposed by its own regulatory frame- It combines a conventional graph partitioning algorithm with
work. The approach is based on iterative process, iterating an evolutionary algorithm to partition a power network to opti-
between the overall optimization algorithm, and intra-zonal mize a multi-attribute objective function based on electrical dis-
power flows of the countries/regions. The iterative process ter- tances, cluster sizes, the number of clusters and cluster connect-
minates when all internal transmission constraints are satisfied. edness. It is demonstrated on the IEEE 118-bus and a 2383-bus
The proposed algorithm is tested in terms of computational case that clusters produced in this way can be used to identify
efficiency using the full UCTE network and examined against buses with dynamically coherent voltage angles, without the
a “myopic” network view of the physical markets’ internal need for dynamic simulation, which is particularly useful for
network. It is shown that the “myopic” network modeling power system applications where ensuring deliverability is im-
leads to infeasible solutions, which may cause significant portant, such as transmission planning or determination of syn-
chronous reserve zones.
congestions in real-time operation and that joint application
of agreed solutions between market operators and TSOs, and
harmonization of the organization of the electricity markets is Jovica V. Milanović
required. School of Electrical and Electronic Engineering
Apostolopoulou et al. in their contribution focus on the The University of Manchester
construction of an optimized financial transmission rights Manchester, U. K.
[email protected]
(FTR) or congestion revenue rights portfolio for an FTR
market participant given his assessment of the frequency and
Antonio Gómez-Expósito
economic impacts of binding constraints in the transmission
Department of Electrical Engineering
network. Their approach relies on the underlying product of
University of Seville
“binding constraints”, which are physically observable phe-
Seville, Spain
nomena, based on the mathematical insights into the structural [email protected]
characteristics of the model used for the clearing of the hourly
day-ahead markets. They further exploit the salient topological Ali Abur
characteristics of large-scale interconnections by the effective Department of Electrical and Computer Engineering
deployment of the orthogonal matching pursuit algorithm to Northeastern University
construct the optimized FTR portfolio characterized by the Boston, MA 02115-5000 USA
minimum number of node pairs for the specification of the [email protected]
FTR elements. The application of proposed methodology is
illustrated on PJM, the largest electricity market in North
America. REFERENCES
D. Efficient Partitioning of Large Power Networks [1] Proc. IEEE, Special Issue on Computers in the Power Industry, vol.
62, no. 7, Jul. 1974.
A hybrid method for identification of coherent sub-graphs in [2] “Electric power problems: The mathematical challenge,” in Proc.
large networks is presented in the paper of Cotilla-Sanchez et al. SIAM Conf., Seattle, WA, USA, Mar. 18–20, 1980.

Jovica V. Milanović (F’10) received the Dipl.Ing. and M.Sc. degrees from the University of Belgrade, Yugoslavia, the Ph.D.
degree from the University of Newcastle, Australia, and the Higher Doctorate (D.Sc. degree) from The University of Manchester,
U.K., all in electrical engineering.
He is currently a Professor of Electrical Power Engineering and Director of External Affairs in the School of Electrical and
Electronic Engineering of the University of Manchester, Manchester, U.K., Visiting Professor at the University of Novi Sad, Novi
Sad, Serbia, and Conjoint Professor at the University of Newcastle, Newcastle, Australia.

Antonio Gómez-Expósito (F’05) received the electrical engineering and doctor degrees from the University of Seville, Spain.
He is currently the Endesa Red Industrial Chair Professor at the University of Seville. His primary areas of interest are optimal
power system operation, state estimation, digital signal processing and utilization of FACTS for enhanced integration of renewable
sources.

Ali Abur (F’03) received the B.S. degree from Orta Doğu Teknik Üniversitesi, Ankara, Turkey, and the M.S. and Ph.D. degrees
from the Ohio State University, Columbus, OH, USA.
He is currently a Professor at the Department of Electrical and Computer Engineering at Northeastern University, Boston, MA,
USA. His research interests are in computational methods for the solution of power system monitoring, operation and control
problems.