chapter 2 power system state estimation strategies...
TRANSCRIPT
15
CHAPTER 2
POWER SYSTEM STATE ESTIMATION
STRATEGIES AND PERSPECTIVES
2.1 INTRODUCTION
Energy Management is the process of monitoring, coordinating and
controlling the generation, transmission and distribution of electrical energy.
An energy control centre utilizes the computer aided tools to monitor, control
and optimize the generation, transmission and distribution of electrical
energy. The functions of a typical control centre can be categorized into three
subsystems as shown in Figure 2.1 namely the data acquisition and processing
subsystem, the energy management / automatic generation control subsystem
and the security monitoring and control subsystem.
SCADA (Supervisory Control and Data Acquisition System) forms
the front end for Energy Management Systems (EMS). A simple SCADA
provides the raw data of the operating condition of the system to the control
centre operators. State Estimation forms the backbone for Energy
Management System. Although reliability remains a central issue, the need
for the real time network models becomes more important than before due to
new energy market related functions are to be added to the existing EMS.
These models are based on the results yielded by state estimation and are used
in network applications such as security monitoring, contingency analysis,
optimal power flow, economic dispatch, unit commitment, automatic
generation control and economic interchange evaluation.
16
Figure 2.1 Functional Diagram of Modern Energy Management System
Important aspect of a system’s operation is obtaining a clear picture
of the state of the system. The possible states of the power system are
Normal, Alert, Emergency, InExtremis and Restorative (Kundur 2010).
During a particular time stamp, the power system would be in any one of the
states. In the ‘Normal’ state, all load and operating constraints are satisfied.
The system is stable for any foreseeable and probable contingency. In the
‘Alert’ state, all the load and operating constraints are satisfied for the system,
but not for one or more of the possible contingencies from the list of pre-
defined contingencies. Preventive control actions are taken to bring the
system from vulnerable operating state to a normal secure operating state. If
these preventive actions fail, then the system moves to the ‘Emergency’ state.
In case of the Emergency state, all the load constraints are satisfied, but one or
more operating constraints are violated. By taking proper corrective control
actions, the system state moves from emergency operating state to the normal
Optimal
Power Flow
Security
Dispatch
EnvironmentalDispatch
Security Monitoring
And Control Subsystem
Security
Monitoring
RestorativeControls
VAR
Dispatch
PreventiveControls
Normal State
Alert State
Emergency
State InExtremis
StateEmergency
Controls
Data Acquisition and
Processing Subsystem
Parameter
Estimation
SCADA
measurements
State
Estimation
Network
Topology
Displays
ExternalEquivalents
Energy / Economy
Functions Subsystem
Load Forecast
Unit Commitment
Economic
InterchangeEvaluation
EconomicDispatch
Automatic
Generation
Control
ContingencyAnalysis
17
or alert operating state. If the corrective control actions fail, the system
moves to the ‘InExtremis’ state wherein one or more load constraints and one
or more operating constraints are violated. Emergency control actions are
taken which will bring the emergency operating state to a ‘Restorative’ state.
In order to bring back the system from the ‘Restorative’ state to the normal or
alert state, restorative control actions are taken such that all operating
constraints are satisfied, but one or more loads are disconnected. All these
controls are generally referred to as the ‘security controls’.
A state estimator is capable of filtering the information to provide a
more accurate picture of the status of the system. The state estimation can be
defined as a process which determines the operating state of the power system
to allow the system operator to make decisions aimed at maintaining the
security of the power system. Weighted Least Square (WLS) algorithm is
normally used for estimating the state of the system. The traditional objective
of the state estimation is to reduce measurement errors by utilizing the
redundancy available in the most measurement systems. In particular, the
objectives are to reduce the variance of the estimate and to improve the
overall efficiency. The other major objectives of traditional state estimation
are (Alvarado 2001):
Detection of erroneous measurements and bad data
Detection of erroneous assumptions about the system,
particularly the status of switches and breakers.
Ability to provide information for unmetered or unmonitored
parts of the system.
Use of redundancy in order to improve the parameters for the
electrical models of the system.
18
The various roles and functions of state estimation in Energy
Management System are shown in Figure 2.2 (Zhu et al 2009).
Figure 2.2 Roles of State Estimation in EMS
From its limited use during 1980s to its expanded but not central
role in the operation of the system in 1990s, state estimation has now become
nothing less than the cornerstone upon which a modern control centre for a
power system is built. State estimation stands in between the real time
information and power system control and monitor applications, playing a
very crucial role in the real time power system control and operation (Zhu
2008). The SCADA data, phasor measurement data, network model and the
pseudo measurements form the input for the power system state estimation
algorithm. The applications such as contingency analysis, security analysis,
optimal power flow etc., are carried out based on the estimates provided by
the state estimator.
Old Estimates,
Scheduled Values
NetworkModel
Conventional
Measurements
SCADA
MEASUREMENTS
PHASOR
MEASUREMENTS
PSEUDO
MEASUREMENTS
TOPOLOGY
PROCESSOR
CONTINGENCY
ANALYSIS
OPTIMAL
POWER FLOW
OTHER
APPLICATIONS
SECURITY
ANALYSIS
STATE
ESTIMATION
19
2.2 STATE ESTIMATION CONCEPTS
State estimation is a digital processing scheme, which provides a
real time data for many of the central control and dispatch functions in a
power system. Its purpose is to improve the dispatch of energy, system
reliability and planning capabilities by understanding the operating state of
the power system. In general the state variables in power system are the
voltage magnitudes and phase angles at all the buses except the slack bus. In
order to ensure secure and economical operation of the power systems, the
operator must be aware of the exact state of the power system at regular
intervals.
Today’s complex large scale power systems require highly
sophisticated techniques for monitoring and control to maintain the system in
a secure and reliable state. There is constant need to update information about
the system to be used for security assessment, load frequency control and a
host of other purposes. In this context, two aspects of the problem stand out
prominently. Firstly, it is uneconomical and in many cases not feasible to
monitor all possible information about the power system. Secondly, the
measuring and equipments that are used are subjected to random errors, which
make the data highly suspicious from the point of view of reliability.
The main objective of state estimation in power systems is
therefore to build a complete and reliable database. Such a database is
obtained by feeding the measured data to a central real time computer, which
on the basis of a prewritten mathematical program, filters the data and extends
it to cover all information regarding the system. In short, state estimation
guarantees reliable information even if some of the measurements are
inaccurate. Thus, the central task of the state estimator is to validate the
information supplied to the system operator. The major ingredients of
20
state estimation are: measurement devices located at strategic points on the
system, high speed data transfer system to convey the measured information
to the control centre, a real time computer with interfacing equipment to
accept and display information and efficient estimation algorithm.
The state of a system may be defined as the minimal amount of
information that one has to know about the system in order to predict its
future evaluation. From this view point, the complex voltages in all buses in a
power system are qualified to be assigned as state variables. Specifically, for
an N bus system, taking a particular bus (preferably the swing bus) as
reference, we may assign N voltage magnitudes and (N – 1) phase angles of
voltages, which are to be called as state variables. Thus, for an N bus system,
the dimension of the state vector is (2N – 1).
The rationale behind this choice is that, knowing these variables
along with the active and reactive power injections at the N buses (real Pi and
reactive Qi at all buses except Pi at the swing bus) and system parameters it is
possible to compute all measurements pertaining to the system. When
observation errors are present the success of state estimation depends on the
redundancy of observed data. Thus, if the state variables are ‘n’ (equal 2N – 1) in
number and if ‘n’ load injections at the buses are given then the problem
reduces to a load flow calculation.
State estimation is different from load flow studies in that the
number of input variables ‘m’ should be greater than (2N – 1), the dimension
of the state vector. It is this redundant information (number of unknown
variables being less than the number of defining equations) which is to be
effectively used in some form of averaging process to filter the data. The
relationships between the different variables involved in the state estimation
are explicitly given in Figure 2.3.
21
X - True System State Vector
Z - Measurement Vector
u - Input Vector
v - Observation Vector
X̂ - Estimated State Vector
Figure 2.3 General Block diagram - State Estimation
The basic relation between Z, X and v is given as below:
Z = h(X) + v (2.1)
where h(X) is the non-linear function of the state X.
Depending on the number of measurements made available to the
control centres, the dimension ‘m’ of the measurement vector ‘Z’ may vary.
Different measurement schemes are identified with respect to state estimation
for an N bus system with M lines.
Case i: Z1 = h1(X) + v1 (2.2)
Z1 consists of (2N - 1) load injection measurements (active and
reactive). The dimension of Z1 is (2N - 1).
Case ii: Z2 = h2(X) + v2 (2.3)
Z2 consists of load injection measurements plus voltage magnitude
measurements at N buses. The dimension of is Z2 is (3N - 1).
verror
u
Physical SystemMeasurement
SystemState Estimator
X ZX̂
22
Case iii: Z3 = h3(X) + v3 (2.4)
Z 3 consists of measurement of active and reactive line flows at both
ends of each line. Here the dimension of Z3 is 4M, where M is the
number of lines.
Case iv: Z4 = h4(X) + v4 (2.5)
Z4 consists of measurements as in Z3 plus voltage magnitude
measurements at N buses. Dimension of Z4 is (4M+N).
Case v: Z5 = h5(X) + v5 (2.6)
Z5 consists of maximum possible measurements. It comprises of
(2N-1) load injections, N voltage magnitudes, (N-1) voltage phase
angles at N buses and 4M line flows. The dimension of Z5 is
(4N+4M-1).
A reliable state estimation is essential to guarantee a reliable
operation of the power system. The reliability of the estimation depends on
the number, type and location of the measurements. The first requirement to
obtain a state estimation is the observability of the system, i.e., the available
measurement set must contain enough information to obtain an estimate of all
states of the system. Also, in order to be reliable, the state estimator must be
robust to the presence of gross errors in the measurements and must be able to
cope with the loss of some of them.
London et al (2000) have proposed a method to identify the
redundancy level of each measurement associated to an observable power
system. The proposed method identifies the critical measurements and sets of
measurements that removed from the measurement set make the power
system unobservable. The redundancy level is very important to operators in
23
order to guide the search for adequate reinforcement of the available
measurement set. The robustness of state estimation can be guaranteed only if
the level of redundancy of the available measurements is high enough and
properly distributed throughout the system.
In other words, the information of each state can be extracted from
different measurements in such a way that the loss of some measurement does
not affect the observability of the system or the reliability of the estimation.
A measure of the redundancy may be denoted by the redundancy factor ,
which is defined as:
n
m
XofDimension
ZofDimension (2.7)
In practice the range of the redundancy factor , has been found
useful if its value is in between 1.5 and 2.8. i.e., 1.5 2.8. If too low
value of is chosen then the measurement errors are inadequately filtered. If
too high is chosen it leads to high investment cost in data acquisition.
2.3 RECENT TRENDS IN POWER SYSTEM STATE
ESTIMATION
Many researchers have analyzed the importance of state estimation
in real time monitoring of large scale power systems. Power system state
estimation provides an estimate for all metered and unmetered quantities. The
main aim of state estimation is to filter out small errors due to model
approximations and measurement inaccuracies and to detect and identify
discordant measurements called bad data. A state estimator is designed to
process the real time meter readings and handle all the uncertainties,
producing a real time reliable database, which is a true representation of the
actual system. The different perspectives with regard to the state estimation
24
problem can be widely classified as various solution methodologies, state
estimation in presence of FACTS (Flexible AC Transmission System)
devices, state estimation using different optimization techniques and presently
state estimation with PMU data.
Kurzyn (1983) has proposed an efficient two level hierarchical state
estimation (HSE) algorithm, suitable for real time monitoring of large scale
power systems. As a first step, state estimation is carried out simultaneously
and independently for all subsystems. In the second step, the subsystem
estimators are coordinated to find the state estimation solution. The mean
error of the proposed hierarchical state estimation algorithm is close to the
error of the weighted least square algorithm and the error of the HSE
algorithm does not necessarily increase with the increasing number of
subsystems. The method is very flexible, allowing fast state estimation.
Suitability of the method and the algorithm are examined using two 220 kV
networks. Several comparisons are made with the classical and centralized
state estimation methods to illustrate the practicality of the hierarchical
method.
Power system state estimation is usually formulated as a weighted
least-squares problem and solved iteratively by the normal equations method.
Gu et al (1983) refers to the power system state estimator as the heart of the
data processing activities in the modern electric utility energy control centre.
The normal equation solution methods for finding the state variables are well
known to exhibit a tendency to be numerically unstable on some networks.
As a result, long precision arithmetic is usually employed in solving the
normal equations. In extremely ill-conditioned cases, the state estimator may
fail to converge. A suitable numerical measure of matrix conditioning has
been defined and a linear analysis of the condition of some simple measured
networks is performed. It provides an insight into the cause of ill-conditioning
25
in realistic networks. The technique for arriving at a compromise between the
conflicting requirements of numerical stability (good conditioning) and
sparsity has been described. Thus an analysis on the sources of
ill-conditioning in the power system state estimation problem and an
alternative solution by Peters and Wilkinson state estimator algorithm that
overcomes the ill-conditioning without losing matrix sparsity has been
presented.
Allemong (2005) has commented on the various requirements for
the successful implementation of state estimation in a utility Energy
Management System. The following requirements of three basic categories of
information had been enforced.
A redundant, reliable and accurate measurement set
Accurate network topology, constructed from the real time
states of the switching elements
Accurate parameters of the network elements.
Pajic (2007) has proposed improvements in Power System State
Estimation and Contingency Constrained Optimal Power Flow (CCOPF) in
stochastic multiple contingencies framework. The existing Newton
Orthogonal factorization algorithms for state estimation are too slow and too
fragile numerically. A new and more robust method that is based on Trust
Region Method (TRM) has been proposed. TRM is based on a globalization
of Newton’s method which is very often the key to the success (finding a
global minimum) of the algorithm. For the first time, TRM has been tested on
the power system state estimation problem.
Li et al (2011) have reviewed the algorithms for power system state
estimation namely the least square algorithm, fast decoupled method,
26
orthogonal transformation algorithm, state estimation algorithm based on
measurement transformation, district coordinated algorithm, etc. The
advantages and disadvantages of each method are also reviewed.
Researchers have also focused on the usage of various linear and
evolutionary programming algorithms for solving the power system state
estimation problem. Hybrid algorithms have also been proposed for
estimating the state of the system. The meta-heuristic methods are iterative
techniques that can search not only local optimal solutions but also a global
optimal solution depending on the problem domain and time limit. In the
meta-heuristic methods, the techniques frequently applied to the state
estimation problem are Genetic Algorithms (GA), Tabu Search (TS),
Evolutionary Programming (EP), Simulated Annealing (SA), Particle Swarm
Optimization (PSO), etc. They are general purpose search techniques based
on the principles inspired from the chromosomes and particles observed in
natural systems and populations of living beings. These methods have the
advantage of searching the solution space more thoroughly.
Gremling and Passino (2000) describes Genetic Algorithm (GA)
that can perform on-line adaptive state estimation for linear and non-linear
systems. The construction of a genetic adaptive state estimator and the way in
which GA evolves the model in a state estimator in real time are discussed.
The operation and performance of the genetic adaptive state estimator has
been illustrated. The genetic adaptive state estimator has the potential to offer
higher performance for non-linear systems compared with the other methods.
Hybrid Particle Swarm Optimization based distribution state estimation have
been proposed by Naka and Fukuyama (2001). This method considers both
non-linear characteristics of the practical equipment and actual limited
measurements in distribution systems and estimates load and distributed
generation output values at each node by minimizing difference between
27
measured and calculated voltages and currents. The number of calculations
involved in the PSO and the HPSO, is found to vary minimally whereas the
results indicates that HPSO generates high quality solutions compared with
the PSO.
In sequential state estimation, each measurement is processed
sequentially, by avoiding matrix procedures. They work for very small
networks but not for medium to large networks. In transformation methods,
the measurements are transformed into new ‘measurements’ that are functions
of the state and of the original measurements. The functional relationships are
via the network structure of the system. The WLS formulation may be
decoupled by separating the measurement set into real and reactive power
groups and by using the same simplifying assumptions as used in the fast
decoupled load flow. The use of state estimation techniques for real world
process applications of significant size is believed to be ground breaking and
the developments described allow a new generation of applications to be
considered.
2.4 STATE ESTIMATION WITH PHASOR MEASUREMENT
UNITS
Phasor measurement units are devices which by employing widely
used satellite technology offer new opportunities in power system monitoring,
protection, analysis and control. Post-disturbance analyses are much improved
due to precise snapshots of the system states, which are obtained through
Global Positioning Satellite (GPS) synchronization. Advanced protection
could be implemented based upon synchronized phasor measurements with
options for improving overall system response to catastrophic events. The
estimate obtained from the phasor measurement unit provides the current
operating state of the power system which primarily helps in maintaining the
28
security of the power system and for several other applications. Synchronized
Phasor Measurement Unit is a monitoring device, which was first introduced
in early 1980s. It gives the real time status of the power system operating
conditions, which is required for power system analysis and control. Real
time monitoring of power systems has become possible with the advent of
phasor measurement units.
Recent developments in time synchronizing techniques coupled
with the computer based measurement technique have been explained by
Phadke (1993). This provides a novel opportunity to measure the phasor and
phase angle differences in real time. Measuring systems using digital
computers are introduced in the power industry. The author gives an insight
into the measurement process, its limitations and its potentialities after the
advent of computer relaying. The importance of the phase angle in electric
power engineering has been emphasized.
PMU measures voltage and current phasors in a power system,
which has higher accuracy than conventional measurements. Synchronism
among phasor measurements is achieved by sampling of voltage and current
waveforms using a common synchronizing signal from the GPS. A PMU
provides time-stamped measurements of active power, reactive power,
frequency, current, voltage magnitude, and phase angle. The time-stamped
characteristic of a PMU is one of its most innovative features which makes it
useful for many other applications such as system protection, control and
stability assessment, aid topology error identification, parameter error
detection and correction and improves the accuracy of state estimation.
With regard to the unpredictable changes in the size and
interconnections of the power system network, optimal location of the phasor
measurement units has also to be changed in order to maintain the complete
29
network observability. State estimation results become more accurate with
the PMU data received from the optimally located PMUs in a power system
network. By strategically locating PMUs, the effects of measurement errors
can be reduced. The principal developments in state estimation and related
areas are observability analysis, erroneous data processing, network topology
processing, topology estimation, and parameter estimation.
The first prototype of the PMU was developed and tested in
Virginia Tech in the early 1980s. The first commercial phasor measurement
unit, the Macrodyne 1690 was developed in 1991. In the late 1990s,
Bonneville Power Administration (BPA) developed a Wide Area
Measurement System (WAMS), which initiated the usage of PMUs for large
scale power systems. A PMU, when placed at a bus, can provide a highly
accurate measurement of the voltage phasor at that bus, as well as the current
phasors through the incident transmission lines (depending on the available
measurement channels). The major advantages of using Synchronized
Measurement Technology (SMT) are that the measurements from widely
dispersed locations can be synchronized with respect to a Global Positioning
System clock. The voltage phase angles can be measured directly which was
so far technically infeasible and the accuracy and speed of energy
management system applications (e.g., state estimation) increase manifold.
Bai et al (2006) have proposed the process-oriented state estimation
using innovation network graph based PMUs. Process-oriented state
estimation is being carried out using all the measurements within a period of
time, which can provide characteristic states. In order to develop the new
method, the operating process is divided into several processes and sub-
processes according to the topology change. In each process or sub-process, a
characteristic state is derived which can represent the average status of this
30
process or sub-process. In order to compute the expected states, an expected
innovation network graph is derived. The innovation network graph has good
robustness on the topology changes in the network. Instead of dealing with
the huge measurements within the process, an innovation network graph is
computed from where the expected states of the network can be derived.
IEEE 5-Bus system is used to illustrate the effectiveness of this method.
The key factor for widespread deployment of the PMUs is to
provide appropriate penetration and redundancy of synchronized
measurements. Such widespread deployment can be achieved when
integrating the PMU function with modern microprocessor based relays for
metering, fault recording and sequence of event recording capabilities
(Kasztenny 2007). Zhu and Abur (2007) revisited the state estimation
problem formulation by assuming availability of at least one phasor
measurement unit in the system. The author investigates on the requirements
to ensure robust state estimation in the presence of single PMU errors. The
requirements are verified by implementing a GPS referenced state estimator
using test systems containing one or more PMU measurements. One of the
issues faced by the state estimators is the choice of reference bus phase angle
when phase angle measurements are present. This issue is easily resolved by
eliminating the reference phase angle from the conventional formulation. This
revised formulation will yield consistent state estimation results even when
any one of the phasor measurements is in error, provided that certain
redundancy conditions are satisfied.
Kamireddy (2008) has proposed a technique in which various
sensors distributed across different parts of the electric power grid will be able
to provide measurements to the control centre operator for situational
awareness of the system. The voltage transformer, current transformer, relay
31
and phasor measurement units are the types of sensors for power system
monitoring. The utilities monitor the operating condition of the system by
processing the measurements received from these sensors using a state
estimator. The measurements are refined and compensated for any lost data
thus providing a snapshot of the power system. Further analysis can be done
based on the most recent data and required state of the system. The electric
power grid is vulnerable to blackouts caused by physical disturbances, human
errors and external disasters. These disturbances can also cause loss of data,
sensor failure or communication link failure. Focus is towards comparing
state estimation algorithms with loss of measurement data. Weighted Least
Square (WLS), Least Absolute Value (LAV) and Iteratively Reweighted
Least Squares (IRLS) implementation of Weighted Least Absolute Value
(WLAV) algorithms are compared for state estimation with clustered and
scattered loss of data.
Chakrabarti et al (2010) have proposed a comprehensive
formulation of the hybrid state estimator incorporating conventional, as well
as PMU measurements. The performance of the state estimators is compared
in terms of the convergence properties and the variance in the estimated
states. Modern PMUs have features like frequency measurement,
measurement of derived quantities (i.e., power components, power quality
related indicators, etc.,) and monitoring of the status of substation apparatus.
The properties of state estimation problem solution can be
essentially improved owing to the new phasor measurements provided by
PMU. The PMUs are the main measurement equipments of WAMS, that
allows the electric power system state to be controlled synchronously and
with high accuracy. As compared to a standard set of measurements received
from SCADA, a PMU installed at a bus can measure voltage phases at that
bus and current phases in some or all branches adjacent to this bus depending
32
on capacity of communication channels. The measurements obtained from
PMU can solve a number of problems concerning low redundancy of
measurements and considerably enhance the efficiency of state estimation
problem solution. The proposed methodology (Glazunova et al 2008) of joint
use of PMU and SCADA measurements for state estimation is checked on the
test network. The results show that PMU measurements allow one to
essentially enhance the efficiency of bad data detection in measurements and
increase the accuracy of the estimates obtained.
Valverde et al (2009) have proposed a multi area state estimator
based on wide area measurements where only boundary buses are considered
in the coordination level. The power injection measurements are not used
during the coordination level thus reducing the number of states and therefore
size of the problem. Instead a set of pseudo measurements are included
whenever a power injection measurement is available in boundary buses. The
proposed methodology can be used when information regarding the
surrounding boundary buses is unavailable at the coordination level and
delivers similar quality of results compared to those obtained when including
internal buses adjacent to boundary buses, but with reduced size, computation
time and complexity at the coordination level.
Ghassemian et al (2009) have proposed new implementation and
testing strategies for phasor assisted state estimation of New York State
Transmission System. Phasor measurement units with GPS synchronization
was incorporated into the data acquisition subsystem of the energy
management system. The modified state estimator was subjected to pre-field
and post-field installation testing. The pre-field installation testing was done
to verify the correctness of the solution algorithm, to identify the impact of
phasor metering accuracy on the quality of estimator solution, to show the
relative effectiveness of phasor measurements with respect to other
33
measurements, to identify the effect of the reference bus selection, to detect
the impact of PMUs on the state estimator convergence and to establish the
minimum clock synchronization sampling accuracy required for the PMUs.
The post-field installation tests were conducted to identify the effectiveness of
the PMUs in the state estimator solution using the real time data as well as to
detect the effect of time skewing and measurement weights on the state
estimator solution.
Hoffman et al (2010) have proposed practical state estimation
techniques for primarily radial distribution networks. From their view, the
smart meter information at the customer side may not be readily usable in
state estimation, but they can be used to verify its power quality and its
voltage levels. Measurements those are inaccurate due to meter, telemetry or
other types of errors will deteriorate the state estimation if they are not
detected, identified and eliminated. Thus, bad data detection and identification
in state estimation will play a crucial role to ensure the quality of state
estimation results.
Chakrabarti et al (2010) have proposed a comprehensive
formulation of the hybrid state estimator in the presence of conventional and
PMU measurements and investigated three different methods of inclusion of
current measurements by PMUs in a power system state estimator. The three
possible ways of including PMU current measurements into the conventional
state estimator are given as follows:
Current phasor magnitude and phase angle measurement.
Real and imaginary part of the complex current measurement.
Pseudo-voltage measurement with the help of current phasor
measurement and known line parameters.
34
The performance of the state estimator in the presence of
conventional measurements and optimally placed PMUs is evaluated in terms
of convergence characteristics and estimator accuracy. Test results on the
IEEE 14-Bus and IEEE 300-Bus systems are analyzed to determine the best
possible method of inclusion of PMU current phasor measurements.
Valverde et al (2010) have proposed a constrained formulation for
hybrid state estimation of power systems. The conventional and
synchrophasor measurements are simultaneously incorporated in the
estimation problem without using any transformation of measurements. This
constrained formulation makes it possible to take advantage of information
from phasor measurement unit branch current and voltage measurements,
improving the accuracy of the estimator.
Hurtgen (2008) explains that observability is a crucial factor when
trying to solve the state estimation problem. A PMU placement method based
on meta-heuristics is proposed and compared to an integer programming
method. A given PMU placement can provide full observability or
redundancy. The PMU configuration can also take into account the zero
injection nodes which further reduce the number of PMUs needed to observe
the network. Finally, a method is proposed to determine the order of the
PMU placement to gradually extend the observable area.
Jaime De La Ree et al (2010) have explained about the uses of
phasor measurements for improved monitoring, protection and control of
power networks. In the early stages these measurements were used only for
the post-event monitoring. This was due to the difficulties faced with regard
to the communication channels required for real time monitoring, control and
protection application. With the occurrence of major blackouts in many
power systems around the world, the value of data provided by PMUs has
35
been recognized and installation of PMUs on power transmission networks of
most major power systems have become an important current activity.
After 1996 U.S West Coast blackouts and 2003 North Eastern U.S
blackouts, the PMU monitoring has become very essential for the post-fault
analysis of the events. One of the recommendations from the United States-
Canada Task force on the 14 August 2003 blackout is to “require use of time
synchronized data recorders” at all utilities. Hence the Eastern Interconnection
Phasor Project [EIPP], now known as North American Synchrophasor Project
Initiative (NASPI) was created. The EIPP performed the first real time wide
area monitoring in U.S to solve some interesting problems such as the
determination of a common phase for the whole eastern grid. With the PMU
installation cost ranging from 10 K to 70 K, (depending on the utility, location
and availability of communication channels) placing PMUs in the optimum
locations is one of first steps of a wide area monitoring system. At present,
phasor measurement units are the most widely used Synchronized
Measurement Technology based devices for power system applications.
2.4.1 Optimal Placement of Phasor Measurement Units
PMUs are increasingly being used in different parts of the world as
the major technology enabler of the Wide Area Monitoring, Protection and
Control system. The general objective of these PMU installation activities is
to eventually make a transition from the conventional supervisory control and
data acquisition based measurement system to a more advanced measurement
system that will utilize synchronized measurements from geographically
distant locations and increase the situational awareness by monitoring a wide
area of the power system in real time. The optimal placement of phasor
measurement units is an off-line problem to be solved during the planning
stage and the results obtained such as number of PMUs to be installed and
their locations are considered as planning data. Several researchers have
proposed algorithms for solving power system state estimation problem using
36
both SCADA and PMU data. From their analyses using different test cases, it
is clear that use of the data obtained from the optimally located PMUs in a
network shortens the computer time and increases the precision of the
estimate obtained.
The main purpose of optimal PMU placement problem is to
minimize the number of installed PMUs and for an n-bus system the
optimization problem is given as follows:
Minimizen
iiixw (2.8)
Subject to f(X) î
where X is a binary decision variable vector, whose entries are defined as:
xi = { 1, if a PMU is installed at bus i , 0 otherwise
wi is the installation cost of the PMU at bus i.
f(X) is a vector function representing the constraints, whose entries
are non-zero if the corresponding bus voltage is solvable using the
given measurement set and zero otherwise.
î is a vector whose entries are all equal to 1.
Several test cases are considered (6-Bus, Anderson and Fouad
9-Bus, IEEE 14, IEEE 30 and IEEE 118-Bus systems) to solve for the optimal
PMU placement using Binary Integer Linear Programming (BILP) technique.
Also real time State Electricity Board systems such as 110kV (North and
South), 230kV and 400kV sub networks are considered to find the Optimal
PMU placement solution.
The Spanning Tree for different test systems considered has been
obtained (Sodhi and Srivastava 2008). Multi partitioning algorithm is applied
37
to form various blocks from the respective spanning tree. For each block the
objective function and the constraint equations are formulated and solved
using Integer Linear Programming. The number and location of PMUs for
the IEEE test systems and the different subsystems of State Electricity
network are given in Table 2.1. It shows one of the optimum feasible
solutions for the PMU placement problem.
Table 2.1 Optimal Number and Location of PMUs
NetworkNumber
of PMUsLocation of PMUs
6-Bus system 2 3, 4
Anderson and Fouad 9-Bus system 3 4, 7, 9
IEEE 14-Bus system 4 2, 6, 7, 9
IEEE 30-Bus system 10 1, 2, 6, 9, 10, 12, 15, 19, 25, 27
IEEE 118-Bus system 32 2, 5, 9, 11, 12, 17, 21, 24, 25, 28,
34, 37, 40, 45, 49, 52, 56, 62, 63,
68, 73, 75, 77, 80, 85, 86, 90, 94,
101, 105, 110, 114
110 KV (North) 16 6, 8, 9, 11, 13, 15, 20, 24, 27, 29,
32, 40, 42, 45, 47, 48
110 KV (South) 14 2, 7, 14, 15, 16, 21, 27, 28, 32, 36,
43, 44, 48, 50
230 KV 3 3, 6, 8
400 KV 10 2, 6, 9, 17, 18, 20, 26, 29, 34, 36
Fang Chen et al (2008) have proposed a reduced state estimation
model including phasor measurement units. In the proposed model, each
PMU can supply two state variables, and hence the number of unknown state
variables is decreased. If a system of N buses is configured with NA number
of PMUs, the redundancy factor is raised as follows:
= m / (2N-1-2NA) (2.9)
Correspondingly, the capability of state estimation to detect bad
data is improved. If the state variables supplied by PMU are accurate, then
the redundancy level of the system will be higher. The ability to measure the
38
voltage phasors at buses directly by PMU has implications on the traditional
state estimation. It is of no doubt that direct measurement of state can
improve the observability and the accuracy of state estimation.
State estimator provides the optimal estimate of the system state
based on the received measurements and the knowledge of the network
model. Measurements may include power injections (real / reactive), power
flows (real / reactive), bus voltage magnitude, line current magnitude and
current injection magnitude. PMUs provide two other types of measurements
namely, the bus voltage phasor and branch current phasor. Depending on the
type of the PMUs, the number of channels used for measuring voltage and
current phasors will vary. Generally, it is assumed that each PMU has enough
channels to record the bus voltage phasor at its associated bus and current
phasors along all branches that are incident to that bus.
Considering the IEEE 14-Bus system and IEEE 30-Bus system, the
redundancy factors with SCADA measurements and with measured data from
PMUs are given in Table 2.2.
Table 2.2 Estimation of Redundancy Factor with SCADA and PMU
Measured Data
Parameters
SCADA Measured Data PMU Measured Data
IEEE 14
Bus system
IEEE 30
Bus system
IEEE 14
Bus system
(4 PMUs)
IEEE 30
Bus system
(10 PMUs)
Number of Buses (N) 14 30 14 30
Number of state variables
(n)27 59 27 59
Number of measurements
(Nm)42 81 38 100
Redundancy Factor ( ) 1.56 1.37 1.407 1.69
39
If the redundancy factor is within the range, 1.5 2.8, the
measurement obtained from the state estimator is more accurate with less
investment cost in data acquisition. Redundancy seeks more than
observability. Additional measurement locations are needed to increase the
redundancy and strengthen the network observability. Regardless which state
estimation formulation is used, the accuracy of state estimation solutions is
dependent on the quality of data as well as measurement configuration and
redundancy. The quality of data can be improved by using high quality
metering devices and communication systems. On the other hand, the
measurement configuration needs to be well designed to ensure robust and
accurate performance of the state estimator. The locations and types of
measurements should allow the state variables of the entire network to be
calculated uniquely, i.e. the network should be observable. There should be
enough redundancy to filter the inevitable random noise associated with the
data, and to detect and eliminate bad data in the observable areas. Low
redundancy causes the state estimators to be very sensitive to the noise and
also limits the bad data detection and identification capability.
2.4.2 Observability
Observability analysis is a fundamental component of real time
state estimation, which checks for enough available measurement in order to
estimate all the states of the electric power system. Two methods used to
determine the system observability are the numerical observability and
topological observability based methods. The topological observability based
approaches utilize the graph theoretical concept to find the optimal locations
and thus to make the system topologically observable. The topological
methods are based on whether a spanning tree of full rank can be constructed.
The numerical methods rely on whether the measurement information gain or
Jacobian matrix is of full rank. If the voltage of a node can be measured
40
directly or can be calculated by other voltage phasor and current phasor, the
node is observable. For making the system topologically observable using
PMUs, following rules are followed:
If voltage phasor and current phasor at one end of a branch are
known, voltage phasor at the other end of the branch can be
calculated using Ohm’s law
If voltage phasor at both the ends of a branch are known,
branch current can be calculated
If there is zero injection bus without a PMU, whose outgoing
currents are known except for one, then the unknown outgoing
current can be calculated using Kirchhoff’s Current Law
Complete observability refers to the PMU placement scenario when
the number and location of the PMUs are sufficient to determine the complete
set of state variables of the network being considered.
In the practical power system, the time required to send the
measurement from SCADA to the control centre is about 2 seconds whereas
the measurement from PMU needs only 40 milliseconds to reach the control
centre (Xue et al 2007). There will be 50 times measurements from PMU sent
to the control centre in the interval that two measurements from SCADA were
sent to the control centre. Thus the remedial actions in case of a contingency
can be carried out more faster and effectively.
Synchronized measurement devices are being deployed in certain
parts of the world and used in applications such as system monitoring, post
disturbance analysis, monitoring of inter-area oscillations and system
modelling. In North America more than two hundred PMUs have been
installed and more are in the pipeline under the North American
41
Synchrophasor Initiative (NASPI). A number of utilities in the United States,
Canada and Mexico are involved in this project. Central and Western
European Countries have started using PMUs extensively. Their major focus
is on developing systematic ways for monitoring and damping of inter-area
oscillations, such as the feedback control of High-Voltage DC (HVDC) links
or Static Var Compensators (SVCs) by using the PMU measurements.
In China, the State Grid Company and manufacturers have issued
the standard on PMUs and WAMS in 2005. More than 700 PMUs are already
in operation and according to the 11th five-year plan of the power grid, all
500kV substations and 300MW and above power plants in the Chinese power
grid will install PMUs within the next five years. Major applications that are
currently in use are the real time visualization of the system dynamics and
transmission capacity, wide area data recording and playback and monitoring
of inter-area low frequency oscillations. The other major objectives for which
the work is in progress include applications such as enhanced state estimation,
on-line security assessment, adaptive protection and emergency control.
The PowerGrid, an Indian central transmission utility is planning to
install 20 to 25 PMUs at critical buses in different regional grids. The
synchronized measurements from these PMUs will be used for model
validations and for the development of a common state estimator combining
the regional state estimators. Based on the success of this stage, more PMUs
are planned to be installed to explore different advantages of SMT and
develop remedial action schemes and System Integrity Protection Schemes
(SIPSs). Similarly Brazil, Russia and other countries are also in the
development of PMU related protection and control for their power networks
(Chakrabarti et al 2009).
The beneficial impacts of PMU data on state estimation depend on
PMU measurement accuracy and calibration, the number of PMUs, PMU
42
locations and related SCADA data accuracy (Wu and Giri 2006). The PMU
data can improve the estimation of inaccurate active power measurements
close to a PMU substation. The wrong measurement may come from the same
substation or neighbouring substations. PMU data with high accuracy is
especially effective when the measurement has a large error. PMU data has
no obvious effect on reactive power and voltage magnitude estimations if
only the phase angle measurement is used. Phasor measurement units should
be installed in substations evenly distributed in the whole system to achieve
the best performance. The location of the reference PMU has no impact on
state estimation. A single PMU can not necessarily improve state estimation
performance. PMU data from the external areas may help operators locate the
outside problem quickly and prevent cascading events. PMU data trend
analysis can detect circuit breaker or switch status changes in the network,
which may improve the topology estimation and error detection.
2.5 STATE ESTIMATION IN PRESENCE OF FACTS DEVICES
Flexible AC Transmission Systems (FACTS), based on either
voltage or current source converters (VSC / CSC), can be used to control
steady state as well as transient performance of the power systems. Interline
Power-Flow Controller (IPFC) is a voltage source converter based FACTS
controller used for series compensation with the unique capability of power
flow management among multi-lines of a substation. IPFC was first proposed
by Gyugyi in 1998 and has the capability to equalize both real and reactive
power flow between transmission lines, transfer power from overloaded to
under loaded line, compensate against reactive voltage drop and the
corresponding reactive line power. Due to these features there is an
increasing interest in the analysis of IPFC in power system state estimation.
Traditional state estimation methods without integrating FACTS devices will
not be suitable for power systems with FACTS devices embedded in the
network.
43
Xu and Abur (2003) presented an algorithm for State estimation of
networks embedded with FACTS devices. State estimation formulation is
modified in order to incorporate the detailed model of the Unified Power-
Flow Controller (UPFC). This necessitates the use of equality and inequality
constraints that account for the limits associated with the device operation and
ratings. The proposed algorithm by them is not only used for state estimation
but also can be used for determining the controller settings of FACTS devices
for a desired operating condition. Initially they introduced a steady state
model of the Unified Power-Flow Controller with operating and parameter
limits. The issues of network observability and bad data analysis have been
discussed using the proposed state estimation algorithm for networks
embedded with FACTS devices. Simulation results on IEEE 14-Bus and 30-
Bus systems are provided to illustrate the performance of the algorithm as a
state estimator in the presence of bad data and also as a solver for determining
UPFC settings for controlling power flows in a power system.
Qifeng et al (2000) have proposed an efficient method suitable for
state estimation embedded with FACTS devices and Multi-Terminal DC
(MTDC) systems, called as the improved sequential method. The proposed
approach is sequential in nature and exhibits good convergence characteristics
compared to conventional techniques. The variables and measurement
equations of the FACTS and MTDC systems related to the problem
formulation are discussed. FACTS devices and MTDC systems can be
included in the existing state estimation algorithms and hence the model
reduces the software development efforts and maintenance costs. Since the
method is developed from the WLS gain matrix, it maintains good
convergence property as the conventional WLS method. The effectiveness of
the algorithm has been demonstrated using test systems and the results are
compared with the other state estimators.
44
The dynamic behaviour of two different FACTS devices namely
the Interline Power-Flow Controller and the Unified Power-Flow Controller
have been discussed by Zhang and Yokoyama (2006). The small signal
model of the Interline Power-Flow Controller is developed and validated
using detailed electromagnetic transients simulation. Using this validated model,
the damping capabilities of the IPFC and the UPFC are compared and
rationalized. The IPFC’s two series branches in contrast to the UPFC's single
series branch permit more opportunities for network segmentation. Hence, the
IPFC is found to have greater potential for improving the system's dynamic
performance.
With the FACTS devices incorporated, the power flow in the
interconnected power systems can be controlled flexibly. A model for state
estimation with IPFC is introduced with power injections and the effect of
IPFC on the power flow is transferred to the lines which are connected to it.
The Interline Power-Flow Controller employs a number of dc to ac inverters
in order to offer series compensation for each line. As a new concept for the
compensation and effective power flow management, it addresses the target
of compensating a number of transmission lines at a given substation.
Generally, the Interline Power-Flow Controller is a combination of
two or more independently controllable static synchronous series
compensators (SSSC) which are solid-state voltage source converters which
inject an almost sinusoidal voltage at variable magnitude and couples via a
common DC link as shown in Figure 2.4. Conventionally, series capacitive
compensation fixed, thyristor controlled or SSSC based IPFC is employed to
increase the transmittable real power over a given line and to balance the
loading of a normally encountered multi-line transmission system. They are
controlled to provide a capability to directly transfer independent real power
45
between the compensated lines while maintaining the desired distribution of
reactive flow in the line (Zhang and Yokoyama 2006).
Figure 2.4 Simplified Schematic of the IPFC Model
In the simplified schematic of IPFC model, each compensating
inverters are linked together at their dc terminals. With this scheme, in
addition to providing series reactive compensation, any inverter can be
controlled to supply real power to the common dc link from its own
transmission line.
Thus, an overall surplus power can be transferred from the
underutilized lines which can be used by other lines for real power
compensation. Evidently, this arrangement maintains the overall power balance
at the common dc terminal by appropriate control action. The injection power
flow IPFC model is based on the representation of IPFC in steady-state
conditions by two voltage sources, each in series with a reactance. A
conventional Newton Raphson power flow program has been modified in
order to incorporate the power injection IPFC model. The simplest IPFC
consists of two back-to-back dc to ac converters, which in a substation are
connected in series with two transmission lines via transformers and the dc
terminals of the converters are connected together via a common dc link.
46
An IPFC can be represented in steady-state conditions by two
voltage sources representing fundamental components of output voltage
waveforms of the two converters and impedances being the leakage reactance
of the two coupling transformers. In the two voltage source model both the
voltage sources, Vser are controllable in both magnitudes and phase angles.
Vser is defined as:
Vser = r Vi e (2.10)
The values of r and are defined within specified limits given by
equation (2.11). The variable r represents certain percent of the voltage
magnitude Vi at bus i.
0 r rmax and 0 2 (2.11)
According to the operating principle of the IPFC, the operating
constraint representing the active power exchange (Pser) among the converters
via the common dc link is given by:
Pser2 = - Pser1 (2.12)
The above equality is valid when the losses are neglected. If the
IPFC is located between nodes i, j and k in a power system, the admittance
matrix is modified by adding a reactance equivalent to Xser between nodes i
and j and nodes i and k. The Jacobian matrix is modified by addition of
appropriate injection powers. The detailed solution steps of the proposed
algorithm can be summarized as:
Step 1: Input system data and telemeter measurements
Step 2: Set iteration count k = 0
47
Step 3: Calculate system measurements
Step 4: Initialize the state vector v(0)
, e(0)
Step 5: Compute Jacobin matrix H(x(k)
) with IPFC
Step 6: Obtain V(k+1)
and(k+1) .
V(k+1)
V(k)
V(k+1)
,(k+1) (k) (k+1)
Step7: Check for convergence.
If max {| V(k+1)| , | (k+1)|} > €, set k = k + 1, go to Step 4
else go to Step 8
Step 8: Print results.
Anderson and Fouad 9-Bus system shown in Figure 2.5 has been
considered to find the estimate of the state of the system with the IPFC
incorporated in the buses 4, 5, and 6.
Figure 2.5 Anderson Fouad 9-Bus System with IPFC
The estimates of the state of the system with and without IPFC are
given in Table 2.3. The solution is found to be more accurate, the computational
effort is reduced and there is an improvement in the voltage profile of the
system considered. The tolerance assumed for convergence is 104.
48
Table 2.3 State Estimation Results for 9-Bus System
Bus
No.
Without IPFC With IPFC
V/pu (°) V/pu (°)
1 1.0400 0 1.0400 0
2 1.0250 9.280 1.0256 8.817
3 1.0250 4.665 1.0250 4.043
4 1.0258 -2.217 1.0259 -2.217
5 0.9956 -3.989 0.9972 -4.306
6 1.0127 -3.687 1.0129 -4.464
7 1.0258 3.720 1.0254 3.254
8 1.0159 0.728 1.0155 0.195
9 1.0324 1.967 1.0321 1.345
The algorithm retains good convergence property as the traditional
WLS method and it possesses the main merit of extending the state estimation
algorithm including the effects of Interline Power-Flow Controller.
2.6 CONCLUSION
The significance of state estimation for proper monitoring and
control of power system operations is reviewed. This chapter is dealt with the
study and application of phasor measurements in power system state
estimation. The importance of phasor measurements in state estimation has
been envisaged. The measurements from PMU are proven to increase the
observability of power systems by strategic placement of minimal number of
phasor measurement units.
Due to the increase in the complex data to be handled in a power
system there is a need for flexible and expandable information integration
environment such that the interaction with different power utilities can be
49
done effectively and efficiently. The deregulation policies, ever increasing
load demand and changing conditions in the topological structure of a power
system have resulted in a requirement for integration of heterogeneous legacy
power system applications as well as new applications inside and outside an
electric utility organization. Thus the architectural model to be proposed
needs to allow pluging-in of new services or upgrading existing services in a
granular fashion to address the new requirements.
The utilities tend to adopt on-line based approach for power system
analysis. With this approach a real time estimation of the system state
variables are continuously updated by distributed data measurements and
adopted as reference for the solution of system state equations. This analysis
if integrated with advanced tool for dynamic loadability assessment of power
equipments, leads to an improvement of the infrastructures allowing system
operators to provide more realistic operational guidance in planning, preventive
and corrective actions aimed to mitigate the effect of critical contingencies.
Information integration and interoperability are two serious
problems in distributed systems, which mainly include communication
networks and communication protocols. For this purpose the International
Electro-technical Committee (IEC) standards are proposed. Some of these
standard protocols can be used over IP-based WANs. However, future power
systems, which contain many renewable energy sources in all voltage levels,
can employ the Internet / Intranet WAN for both control and telemeter. There
are several standards available which can be applied for the control and telemeter
over Internet / Intranet WAN. The standard protocols are to be compared and
the factors for choosing the right protocol for particular purpose are to be
analyzed. In the next chapter, a generalized service oriented model, which
will be customized exclusive for power system applications, is presented.