INTERNATIONAL TRANSACTIONS ON ELECTRICAL ENERGY SYSTEMS

Int. Trans. Electr. Energ. Syst. (2014)

Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etep.1966

Ranking of phasor measurement units based on control strategy for

small-signal stability
Charu Sharma*, and Barjeev Tyagi
Electrical Engineering Department, Indian Institute of Technology, Roorkee, India

SUMMARY
This paper presents a methodology to rank phasor measurement unit (PMU) locations based on small-signal
stability. In the proposed scheme, integer linear programming is utilized to identify the optimal locations of
PMUs. These locations are ranked using analytical hierarchy process. Modal analysis is conducted to
identify inter-area modes leading to instability. A control scheme is proposed, to nullify this instability.
Based on the control scheme, few critical buses are determined, which are given the highest priorities while
ranking. Because a critical bus can be a generator, or tie line or load bus, therefore, three different ranking
criteria are utilized. Based on ranking, PMUs are placed in multiple stages such that in the initial stage itself,
all critical buses are observed. To check the effectiveness of the proposed scheme, fuzzy and proportionalintegral-derivative controllers are employed. Proposed methodology is successfully tested on 16-machine,
68-bus system. Copyright 2014 John Wiley & Sons, Ltd.
key words:

phasor measurement units; wide-area network; small-signal stability; Eigen value analysis;
observability; analytical hierarchy process

1. INTRODUCTION
Small perturbations in load or small oscillations within a given power system can have signicant
effects on damping characteristics of the system. These disturbances result in synchronization issues
such as rotor angle displacement or oscillation problem due to insufcient damping. Monitoring of
these oscillations is of immense importance for secure operation of power systems [1]. Recent developments in wide-area monitoring provide real-time and accurate information, which was not possible
with earlier energy management systems [1]. However, this wide-area monitoring starts with allocation
of phasor measurement units in given system [1,2].
Signicant work has been carried out in the past to deploy phasor measurement units in power
systems. Authors in [3] determine phasor measurement unit (PMU) locations based on observability
concept. In References [4,5], integer linear programming (ILP) has been used to formulate the topological observability. Concept of pseudo-observability and its importance is nicely summarized in [6,7].
Considerable work has been carried out in [8,9] to show the effect of conventional measurements on
observability. In [10], an integer quadratic approach has been introduced to minimize the PMU
locations, which ensures observability during normal and during loss of single line or PMU. The
aforementioned literature shows that PMU allocations are generally carried out according to topological observability. However, method for complete topological and numerical observability of power
system has been considered in Reference [11]. It is reported that economic constraints associated with
this technology compels utilities to place the limited number of PMUs in the system. As a solution, few
researchers have considered PMU placement in stages/phases [6,12,13]. Authors in [6] have used the
concept of depth of observability to place PMUs in stages. However, in [12], PMUs are placed in such
a manner that placement results after nal stage are identical to single-stage PMU placement using ILP
*Correspondence to: Charu Sharma, Electrical Engineering Department, Indian Institute of Technology, Roorkee, India.

E-mail: charudee@iitr.ernet.in
Copyright 2014 John Wiley & Sons, Ltd.

C. SHARMA AND B. TYAGI

approach. Author in [13] has used multi-criteria decision-making approach to place PMUs considering
various criteria such as bus, tie line, and voltage observability.
Recently, few researchers have utilized probabilistic approaches to locate signicant PMU locations
in a given system. In [14,15], authors have utilized analytical reliability methods to improve probability of observability for given system under random power outages. However, for placement of
PMU, authors have considered topological observability. In another reference [16], Gaussian Markov
random eld model of power system is proposed. Authors utilized greedy algorithm to nd optimal
PMU sites, which provides maximum information gain.
Real-time control of small-signal oscillations is also as important as real-time monitoring of
small-signal oscillations. In literature, deterministic and probabilistic approaches for small-signal
analysis were reported [1724]. Robust damping controller using h-innity, linear matrix
inequality techniques have been widely used [17,18] for damping out inter-area oscillations.
Various intelligent controllers are also reported [19,20]. Apart from controller design, selection
of controller location and feedback signals is a recent topic of concern [21,22]. In Reference
[21], the author described wide-area fuzzy controller (WAFC) for various operating points. In
the aforementioned literature [1722], the authors have assumed that PMUs are placed on all
generator terminals, which is practically not possible. Few researchers have also applied probabilistic approach for small-signal stability studies [23,24]. In these references, authors have
explored Monte Carlo-based probabilistic Eigen analysis technique for various operating points
and uncertainties of the power system. However, authors have not considered topological
observability of the system. Therefore, the major contribution of this paper is to select, rank,
and place PMUs on those buses, which are essential for monitoring and control of small-signal
stability analysis of a large network.
For secure operation of power system, monitoring and control of small-signal stability (SSS) is
of immense importance. This paper focuses on ranking of PMU locations in a given network for
SSS studies. Based on these rankings, PMUs are placed in stages to completely observe the system. The proposed ranking scheme is divided into three steps. The optimal PMU number and
locations are determined in the rst step using ILP, such that each bus is observable at least by
two PMUs. In the second step, Eigen value analysis is carried out to compute inter-area modes
responsible for small-signal instability, and control scheme is proposed for small-signal instability.
For each inter-area mode, location of controllers and feedback signals are determined through
participation factors and controllability indices. Buses associated with controller locations or feedback input signals are identied as critical buses. All optimal PMU locations obtained through rst
step are ranked in the third step. Three criterias, namely, generator bus observability index
(GBOI), tie-line bus observability index (TLBOI), and bus observability index (BOI) are utilized
in the analytical hierarchy process (AHP) to compute priority-based nal performance indices
(FPIs). Based on these FPIs, PMUs are ranked. All critical buses identied in Step 2 are given
the highest priorities in ranking the PMU locations. Further, important tie lines of the system are
also determined and given the next higher priority. To test the effectiveness of the proposed
approach, fuzzy logic and proportional-integral-derivative (PID) controller has been considered at
the selected locations to control the perturbations.

2. FORMULATION FOR PMU PLACEMENT

The PMU can make installed bus and its neighboring buses observable [1]. The objective of placing
PMUs in power systems is to decide a minimal set of PMUs such that the whole system is observable.
The problem to determine optimal PMU location for complete observability can be formulated in
following manner [4].

Minimize

N
X


Ck yk ;

k1

Copyright 2014 John Wiley & Sons, Ltd.

yk

1; if PMU is placed over bus

0; otherwise

(1)

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

RANKING OF PHASOR MEASUREMENT UNITS


Subject to GY 2; gij

1; if i j or if i and j are connected

0; otherwise

(2)

Where N is the total number of PMUs, Ck is the cost of a kth PMU, yk is the kth binary placement
variable, G is the connectivity matrix, and Y is a vector of binary variable y. In proposed work, cost of
each PMU is taken equal. To make formulation simple, Ck is generally taken as 1 per unit. Connectivity matrix G is formed using topological connectivity of buses as given in Equation 2. Details
for formation of G are nicely summarized in [11]. The aforementioned formulation is an ILP problem.
Equation 1 computes the minimal subset of PMUs and 2 is the basic constraint equation. In 2, righthand side of inequality ensures that each bus in the system is observed by two PMUs, such that any
loss of the line or a line outage will not affect complete observability of the system [13]. Presence
of conventional measurements reduces the number of PMUs required for complete observability
[12,25]; therefore, zero injection measurements are also incorporated in the aforementioned formulation. The aforementioned ILP problem is solved using the algorithm described in [12] to obtain optimal
allocations for PMUs.

3. CONTROL AND IDENTIFICATION OF CRITICAL BUSES

In multi-machine power system, small-signal oscillations are the result of group of the generators in
one area swinging against a group of generators in another area. To prevent system from these
oscillations, wide-area damping controllers are generally provided [17]. In this paper, a control scheme
is proposed, which comprises of selection of controller location and selection of controller input
signals. However, design of wide-area damping and PID controllers is described in [26,27]. The
following sections give a detailed description of proposed schemes for control, selection of input
signals, and controller locations.
3.1. Selection of input signals
In the control scheme of a wide-area network, proper selection of feedback signals is of utmost importance. In this work, generator speed signals are considered as input signals for damping controllers.
Therefore, selection of the input signal is the selection of generator speed signal. Signals that are
selected as input signals for wide-area controllers should affect the inter-area modes that are responsible for small-signal oscillations. Therefore, Eigen value analysis [28] is carried out to determine interarea modes. According to Reference [29], to modify a mode of oscillation by the feedback, the chosen
input must excite the mode, and it must also be visible in chosen output. The measures of these two
properties are controllability and observability [30,31]. Consider the state space model of a system
as follows:
x Ax Bu

(3)

yC x

(4)

Where x is the state vector, y is the output vector, u is the control vector, A is the state matrix, B is the
input matrix, and C is the output matrix. The controllability [30] of ith mode of 3 from the jth input is
proportional to the cosine of the angle between bj and qi. This relationship is called as modal controllability index (CI) and is given as
 T 
q b j 
 

(5)
CI cos qi ; bj i  
kqi k bj
Where the symbol | | means that the absolute value of a scalar and || || means the standard two-norm
of a vector. qi is the left Eigen vector of A (state matrix) satisfying equation ATqi = iqi, and bj is the jth
column of B (input matrix). The selections of appropriate wide-area feedback signals for proposed
controllers are carried out through the modal CI given by Equation 5. Thus, a higher value of CI for
a particular mode indicates that the chosen input signal will be effective in controlling the mode.
Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

C. SHARMA AND B. TYAGI

3.2. Selection of controller locations

In wide-area network, controllers are needed with machines that have the largest effect over oscillating
modes of interest. Reference [21] utilized joint controllability and observability modal indices for
selecting the location of wide-area controller. Reference [31] relates participation factors of selective
modal analysis with modal controllability and observability index. Therefore, participation factors
are utilized in the proposed work to determine suitable controller location.
3.3. Control scheme for small-signal stability
In the proposed control scheme, speed signals of the machines that are having the highest CI are treated
as wide-area signals. These remote signals are communicated via PMUs to other areas, if the generator
bus is observed by PMUs directly or in directly [32]. The controller is placed with the machine that has
highest participation factor, and the speed feedback is taken from the machine having highest CI.
Differences between speed signals of generators that are having highest CI and participation factor
for a given mode are calculated. These speed deviations and derivative of speed deviations are given
as inputs to WAFCs. Output signals of controllers are fed back to the generator, through the exciter.
From the aforementioned control scheme, it is observed that there are certain buses that are critically
important, such as generator buses to which controller is attached or generator buses from which feedback signals are taken or tie-line buses to monitor inter-tie performance. Therefore, all such buses are
termed as critical buses of the given network. In such cases, PMUs are placed at these preferred buses
in initial stages. Rests of buses are made observable in later stages.

4. RANKING OF OPTIMAL PMU LOCATION

The main focus of this paper is to provide a methodology to rank the optimal PMU locations based on
the control scheme described in the previous section. Based on this ranking, PMUs are placed in
multiple stages to observe the given system. In this work, three criteria, namely, generator bus, tie line,
and bus observability criteria, are proposed to monitor and control SSS. For SSS monitoring, tie-line
observability criterion is considered, and for control, GBOI criterion is considered. However, bus
observability criterion depends on topological connectivity of the bus. These criteria are utilized in
analytic hierarchy process [33] to rank PMU locations.
Few important considerations:
It is observed that every optimal location (obtained in Step 1) is either a generator or tie line or
simple load bus and may be or may not be a signicant bus.
It is also noted that all generator, tie line, and load buses present in optimal locations are not
critical buses. Therefore, to distinguish between critical and noncritical buses, each criterion
(i.e., generator, tie line, and bus observability) is further divided into two criteria, one for
critical buses and another for noncritical buses.
It may happen that a critical bus does not belong to the optimal locations. In such case, nearest
optimal bus is considered as critical bus, and highest priority is assigned to that bus.

4.1. Generator bus observability index

In a power system, the angular separation between generators indicates that whether system will be
leading to angular instability or not. Also, to improve inter-tie performance, wide-area damping controllers are generally provided with generators [17,21]. According to Reference [21], generator speeds,
terminal bus frequency, and active power are the most commonly used control input signals, which can
be used in designing damping controllers. Therefore, it is important to monitor the generator buses
directly or indirectly using the PMUs. It is to be noted here that there may be few buses that are strategically important and are not present in optimal locations obtained in Step 1. In such a case, nearest
or second nearest optimal location to that critical bus is given priority.
Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

RANKING OF PHASOR MEASUREMENT UNITS

According to the control scheme described in the previous section, there may be situations when a generator bus is having controller or providing speed feedback signals. In such situation, these generator buses
are of prime importance and termed as critical buses. Therefore, all generator buses present in optimal
locations are divided into two groups, that is, critical generator buses and noncritical generator buses.
For critical generator bus, the generator observability index is represented as GBOIS and is given as
0
1
!
!
X k
X
X
k
k
k
wc3 
(6)
U A wc2 
U
U
GBOIS @wc1 
gen

gen

Ngen

Ngen

NNgen

NNgen

Where GBOISk is the GBOI for kth bus, which has more signicance for SSS analysis. U kgen represents number of generator buses connected to kth bus. Similarly, U kNgen and U kNNgen represent the
number of neighboring and neighbor to neighboring buses, respectively, connected to bus k. wc1,
wc2, and wc3 are constant weights related to GBOIS and assigned higher values.
In case of noncritical generator bus, the generator bus observability index is indicated as GBOI and
is given by the following expression:
!
!
!
X
X
X
(7)
U kgen w2 
U kNgen w3 
U kNNgen
GBOI k w1 
gen

Ngen

NNgen

Where GBOIk is the GBOI for any kth bus. w1, w2, and w3 are constant weights, which are having
lower values as compared with wc1, wc2, and wc3.
4.2. Tie-line bus observability index
In large power systems, performances of damping controllers are observed through dynamic changes
in inter-ties. Therefore, to monitor tie-line response, PMUs should be placed on these tie lines. To observe tie lines, seven different cases are considered throughout this paper. Six cases are directly taken
from Reference [13]. Details of these six cases are not included in this paper because of space limitations. However, rst and second cases are described here for understanding. For example, P and Q are
end buses of a tie line. R and S are the neighboring buses of P and Q, respectively. In the rst case,
both end buses of the tie line (P and Q) belong to optimal PMU locations, obtained in the rst stage.
Because PMUs are to be placed on both the end buses, both buses will have equal importance; therefore, TLBOI will be TLBOIP = TLBOIQ = 1. Likewise, in the second case, the rst end of the tie line
and the neighbor of the second end are present in PMU locations. To monitor tie-line performance
at least from one end, TLBOIP has been given higher priority than TLBOIS [13].
In a similar manner, PMU locations are prioritized in seventh case, which is additionally introduced
in this work. In this particular case, one end bus of tie line (P) and its adjacent bus (R) at same end both
are candidate of optimal PMU sites.
As stated before, tie-line bus is of great importance. Therefore, TLBOI of tie-line bus (P) is
TLBOIP = 1.5, and observability index of its adjacent bus (R) is TLBOIR = 0.25. All cases are summarized in Table I and Figure 1. For a given system, each tie line will satisfy any one condition, from
stated seven conditions. TLBOI for kth tie line or its neighboring bus is given by expression 8.

TLBOI
k

TL; if kth bus is tie line or its neighboring bus; TL value is taken from Table 1
0; otherwise

(8)

As mentioned before, not all tie lines have signicance in SSS analysis. Therefore, tie lines are also
divided into critical and noncritical tie lines. Tie-line observability index for critical and noncritical tieline buses are represented as TLBOIS and TLBOI, respectively.
Table I and Figure 1 give tie-line observability indices for noncritical tie lines. However, in
case of critical tie line, TLBOI of the higher priority end is assigned with value greater than
or equal to 2. For example, if an important tie line belongs to Case b, then its higher priority
end, TLBOI (i.e., TLBOIP) is modied as TLBOIS, and a value greater than or equal to 2 is
Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

C. SHARMA AND B. TYAGI

Table I. Various cases for tie-line bus observability index (TLBOI).

Case a

When both tie-line buses (P,Q) have PMUs

Case b

When one tie-line bus (P) and neighboring bus

of other end (S) have PMUs
When two different tie lines are neighbor ({P-Q}
and {P-Q})
When two different tie lines are having same
neighbor ({P-Q} and {P-Q})
None of tie-line buses (P,Q) have PMUs
When one tie-line bus (P) and adjacent bus of
the same end (R) have PMUs
When a tie line is not present in optimal PMU
locations (obtained through Step 1).

Case c
Case d
Case e
Case f
Case g

TLBOIP = TLBOIQ = 1
TLBOI P = 1.5 TLBOI S = 0.5
TLBOI Q = TLBOI P =1.5 TLBOI P = 1
TLBOI P = TLBOI Q = 1.5 TLBOI S = 1
TLBOI R = TLBOI S = 2
TLBOI P = 1.5 TLBOI R = 0.25
TLBOI P = TLBOI Q = 0

Figure 1. Various cases to determine TLBOI.

assigned to it. Likewise, TLBOI of every signicant tie line or its neighbor is modied and
renamed as TLBOIS and is represented as

TLBOISk

TLS; if kth bus is significant tie line or its neighboring bus; TLS > TL and TLS2
0; otherwise
(9)

4.3. Bus observability index

The BOI depends upon the topological connectivity of the bus. Like other two indices, this index is
also divided into two indices. Bus observability index for noncritical bus is represented as BOI. However, for critical bus, the BOI is abbreviated as BOIS. BOI for a PMU bus k is dened as the number of
buses directly connected to kth bus and mathematically expressed as

N
X
1; if j and k are connected
k
(10)
Gkj
BOI
0; otherwise
j1
jk

This criterion prioritizes those buses that have maximum connectivity in the given system. BOI for
critical buses are given as
BOISk 1BOI k ; if k-th bus itself is significant generator bus

(11)

However, if a signicant generator does not belong to optimal placement locations (obtained in
Stage 1), then its immediate neighbor bus BOI (BOInk) is modied as BOIS.
BOISnk 1BOI nk
Copyright 2014 John Wiley & Sons, Ltd.

(12)
Int. Trans. Electr. Energ. Syst. (2014)
DOI: 10.1002/etep

RANKING OF PHASOR MEASUREMENT UNITS

Where nk is the neighbor bus of the signicant generator bus. From the aforementioned expression,
it is clear that BOI of signicant buses are separated from the rest of BOIs, athough they have the same
formulation. This is carried out to distinguish between critical and noncritical buses. This distinction is
useful for analysis, particularly, while assigning priority in AHP. After computing, in all six criteria
and optimal PMU locations, AHP is applied, which is described in the following section.
4.4. Analytical hierarchical process
The AHP is a structured technique for analyzing complex decision problems. In AHP, the problem is
broken into various criteria and alternatives [33]. In PMU placement problem, optimal placement
locations obtained through rst step are treated as alternatives and various observability indices are
criteria. In AHP, rst pairwise matrix (PM) is constructed, which selects relative importance of
different criteria. According to SSS problem, GBOIS is given highest priority, TLBOIS is given
second highest, and BOIS is assigned third highest priority, whereas GBOI, TLBOI, and BOI are given
lower priorities, as shown in Figure 2.
All six criteria are prioritized using 15 priority rules given in Table II, and PM is constructed based
on these rules.
The PM considered in this paper is given in Table III. In this table, the second element of the rst row
depicts that GBOIS is six times more important than GBOI (as per rst priority rule). Likewise, all entries of the Table III are derived from Table II. PM is then utilized to compute weights for each criterion.
To calculate weights, Eigen values of PM are determined. Eigenvector associated with the largest
Eigen value of PM, gives W, that is, the weight vector W = [w1, w2, w3, w4, w5, w6]. Weights for all
criteria are normalized by utilizing expression 13.
wj
j 1; 2; 3; 4; 5; 6
(13)
nwj 6
wj
After computing weights for each criterion, all six criteria (i.e., GBOIS, TLBOIS, BOIS, GBOI,
TLBOI, and BOI) are further normalized using 14 for each PMU location.


I kj  I min;j
 j 1; 2; 3; 4; 5; 6
NI kj 
(14)
I max;j  I min;j
Where NIkj represents normalized value of jth criterion for kth optimal location. Imin,j and Imax,j are
the minimum and maximum value of jth criterion, respectively. However, Ikj is the actual value of jth
criterion for kth optimal location. Normalized values of weights and criteria are used to compute the
FPI for each optimal PMU location, which is given as
FPI k 6j1 nwj NI kj

k 1; 2; 3K

(15)

Figure 2. Priority Scale.

Table II. Priority matrix rules.

GBOIS is 6 times more important than GBOI
GBOIS is 2 times more important than TLBOIS
GBOIS is 4 times more important than BOIS
GBOIS is 7 times more important than TLBOI
GBOIS is 9 times more important than BOI
TLBOIS is 4 times more important than GBOI
GBOI is 2 times more important than TLBOI
BOIS is 5 times more important than GBOI
Copyright 2014 John Wiley & Sons, Ltd.

GBOI is 3 times more important than BOI

TLBOIS is 6 times more important as TLBOI
TLBOIS is 2 times more important than BOIS
TLBOIS is 7 times more important than BOI
BOIS is 5 times more important than TLBOI
TLBOI is 2 times more important than BOI
BOIS is 5 times more important as BOI

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

C. SHARMA AND B. TYAGI

Table III. Pairwise matrix for the 16-bus system.

GBOIS
GBOI
TLBOIS
TLBOI
BOIS
BOI

GBOIS

GBOI

TLBOIS

TLBOI

BOIS

BOI

1
1/6
1/2
1/7
1/4
1/9

6/1
1
4/1
1/2
5/1
1/3

2/1
1/4
1
1/6
1/2
1/7

7/1
2/1
6/1
1
5/1
1/2

4/1
1/5
2/1
1/5
1
1/5

9/1
3/1
7/1
2/1
5/1
1

Where NIkj are normalized value of all six indices; nwj is the normalized weight for each criterion;
FPI1, FPI2FPIk correspond to FPIs for each PMU location; and K represents the total number of
optimal locations. These FPIs prioritize the PMU locations. Now, for a xed number of phases,
priority placement is carried out to make all generators, tie line, and load buses observable.

5. SIMULATION RESULTS AND DISCUSSION

The ve-area 16-machine, 68-bus power system shown in Figure 3 represent New England and New
York Interconnection. Reference [29] shows that a given power system based on coherency can be
divided into ve groups. Generators 14, 15, and 16, each forms a single-generator group. However, group
4 consists of New England system (generator 19), and group 5 (generators 1013) consists of New York
system. Each generator is assumed to be provided with governors; IEEE ST1A type static exciter.
In present work, all simulation studies are carried out on Intel Core i7-2600 central processing
unit, with 3.40 GHz processing speed and 16 GB RAM. SSS analysis is conducted on the system
according to Reference [29]. Calculation of state matrices, Eigen value analysis and inter-area
mode identication are performed as described in [29] and implemented with Power System Toolbox [34] and MATLAB [35]. However, for designing fuzzy controllers, Fuzzy Logic Toolbox is utilized along with Power System Toolbox.

Figure 3. Five-area 16 bus system.

Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

RANKING OF PHASOR MEASUREMENT UNITS

5.1. Optimal PMU placements

As described in Section 2, the optimal placement locations of PMUs are found according to Equations
1 and 2. To achieve the complete observability of the given system, total 42 PMUs are required. The
optimal locations for these PMUs are 6, 8, 11, 12, 14, 16, 17, 19, 20, 22, 23, 25, 26, 29, 31, 32, 34, 36,
37, 41, 42, 44, 45, 46, 47, 48, 49, 51, 52, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, and 68. Placing
PMUs on these locations will make the system completely observable at least by two PMUs, even
when single-transmission line or PMU is lost.
5.2. Eigen value analysis and identication of critical buses
To prevent the system from small-signal instability, rst inter-area modes are determined from Eigen
value analysis, which are given in Table IV.
Table IV shows that three inter-area modes and one unstable mode are present. To implement the
control scheme for the aforementioned inter-area modes, speed participation factors and controllability
indices are computed. Participation factors determine the suitable location for placing the damping
Table IV. Eigen value analysis of 16-machine system.
Eigen values
Unstable mode
1

0.3264 6.8364i

Inter-area mode
1
2
3

-0.2251 2.393i
-0.2782 3.280i
-0.1781 4.219i

0.4

Frequency

Damping

1.0880

-0.0477

0.3809
0.5221
0.6715

0.0936
0.0845
0.0422

0.14
0.12

0.3

0.1
0.2

0.08
0.06

0.1

0.04
0
0.02
-0.1

10

12

14

16

(a)

10

12

14

16

(b)
0.2

0.3
0.25

0.15

0.2
0.15

0.1

0.1

0.05

0.05
0

0
-0.05

10

12

(c)

14

16

-0.05

10

12

14

16

(d)

Figure 4. (a) Participation factors for unstable mode, (b) participation factors for rst inter-area mode, (c)
participation factors for second inter-area mode, and (d) participation factors for third inter-area mode.
Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

UnstableMode
Inter-areaMode1
Inter-areaMode2
Inter-areaMode3

Copyright 2014 John Wiley & Sons, Ltd.

0.19
0.04
0.17
0.44

0.25
0.06
0.04
0.03

2
0.21
0.12
0.25
0.03

3
0.04
0.92
0.011
0.021

4
0.06
0.06
0.01
0.50

5
0.08
0.03
0.014
0.44

6
0.07
0.05
0.01
0.033

7
0.11
0.01
0.161
0.580

8
0.08
0.004
0.032
0.043

9
0.041
0.002
0.463
0.063

10

Table V. Controllability index table.

0.0118
0.008
0.124
2.3e-4

11

0.02
.001
0.24
0.02

12

0.0014
1.6e-4
0.037
4.5e-5

13

0.0012
2.8e-5
0.038
1.6e-4

14

0.0012
6.3e-5
0.003
7.5e-6

15

0.0021
2.8e-5
0.162
4.1e-4

16

C. SHARMA AND B. TYAGI

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

RANKING OF PHASOR MEASUREMENT UNITS

controller, and CI determines the feedback signal selection. Figure 4(ad) shows speed participation
factors corresponding to unstable and inter-area modes of Table IV.
Figure 4(a) corresponds to unstable mode and shows that unstable mode is coupled to ninth generator (G9). Therefore, placing damping controller with G9 will make the system more stable. Figure 4
(b) shows that for rst inter-area mode, damping controller can be placed with any generator because
each generator has a positive speed participation factor. However, the 13th generator (G13) has the
highest participation, so it is the best candidate for placing damping controller. Similarly, Figure 4
(c) and (d) shows that 16th and 13th generators have the highest participation for second and third
mode respectively. Therefore, 9th, 13th, and 16th generators are three possible locations, which are
selected for the wide-area damping controller.
To select the appropriate feedback signal for three proposed damping controllers, controllability
indices are computed corresponding to each mode. Results so obtained are shown in Table V.
For unstable mode, second generator (G2) has the highest CI. For rst and third inter-area modes, a
single damping controller is placed with G13. The CI table shows that generator 4 signal has highest
CI compared with generator 8 (G8). Therefore, generator 4 speed signals should be fed to damping
controller placed with 13th generator. Likewise, for the second mode, 10th generator (G10) speed
signal is selected as feedback signal, and damping controller is placed with 16th generator (G16).
From the aforementioned Eigen value analysis, it is clear that six generators 2nd, 4th, 9th, 10th,
13th, and 16th are of immense importance for SSS. Hence, it becomes essential to place PMUs on
these generator buses in the initial phases.
5.3. Phased PMU placement
In this work, it is considered that the PMUs are placed in the network in four phases. In rst three
phases, 11 PMUs are placed in each phase, while in fourth phase, 9 PMUs are placed. There are 16
generator buses and eight tie lines in the given system (as shown in Table VI). From the SSS view
point, only six generators from 16 generators and four tie lines from eight tie lines are of importance.
Third column of Table VI shows placement scheme according to various cases of TLBOI.
For each PMU location, values of GBOIS, TLBOIS, BOIS, GBOI, TLBOI, and BOI criteria are
computed and normalized according to normalization expressions given in Section 4.4.
The PM is formed to compare the relative importance of each criterion. Further, PM is utilized to
calculate the weight vector W, which is Eigenvector associated with largest Eigen values of PM. In this
paper, W, that is, weight vector is [0.7976, 0.1278, 0.4779, 0.0873, 0.3286, 0.0590], and FPIs are
computed according to 15.
Table VII shows normalized values of each criterion for every PMU location, and FPI are also given
in eighth column of Table VII. Ninth and tenth columns of the same table depict ranking and prioritized FPI, respectively.
Phased PMU placement for 16-bus system is given in Table VIII. For the given 16-machine bus
system, tie lines 127, 12, 89, and 5051 are most important tie lines [29]. Eigen value analysis
shows that generator buses 54, 56, 61, 62, 65, and 68 corresponding to generator G2, G4, G9, G10,
G13, and G16 are critical buses for controlling small-signal instability.
Table VI. Tie-line buses in the 16-bus system.
Tie-line data
From bus
1
1
1
8
42
52
46
50

To bus

Placement scheme

2
27
47
9
41
42
49
51

No PMUs on tie line end buses

No PMUs on tie line end buses
Second end bus of tie line will have PMU
First end bus of tie line will have PMU
Both buses of tie line are with PMUs
Both buses of tie line are with PMUs
Both buses of tie line are with PMUs
Second end bus of tie line will have PMU

Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

6
8
11
12
14
16
17
19
20
22
23
25
26
29
31
32
34
36
37
41
42
44
45
46
47
48
49
51
52
56
57
58
59
60

OPP
vector

0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.333
0
1
1
0
0.166
0.666
0
0.166
0.666
0
1
0
0
0
0
0

TLBOI
(normalized)
0.375
0
0.625
0
0
0.312
0
0.687
0.687
0.687
0.687
0.687
0.625
0.375
0.375
0.375
0.312
0.687
0.687
0.687
1
0
0
0
0
0
0.312
0
0.687
0
0.781
0.781
0.781
0.781

GBOI
(normalized)
0
0.60
0.60
0.40
0.60
1
0.60
0.60
0.40
0.60
0.60
0.60
0.80
0.60
0.80
0.60
0.60
0.80
0.60
0.60
0.60
0.60
0.80
0.40
0.40
0.40
0.40
0
0.80
0
0.20
0.20
0.20
0.20

BOI
(normalized)
0
1
0
0
0
0
0.571
0
0
0
0
0.571
0.571
0
0.571
0
0
0
0
0
0
0
0
0
0.571
0
0
1
0
0
0
0
0
0

TLBOIS
(normalized)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0

GBOIS
(normalized)
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.50
0
0.25
0
0
0
0

BOIS
(normalized)
0.205
0.271
0.060
0.012
0.018
0.052
0.163
0.064
0.058
0.064
0.064
0.209
0.211
0.043
0.194
0.043
0.039
0.086
0.064
0.109
0.130
0.018
0.032
0.042
0.156
0.019
0.063
0.343
0.115
0.468
0.058
0.058
0.058
0.058

FPI

Table VII. Normalized values of criteria and nal placement index for 16-bus system.

10
7
25
42
40
33
12
20
32
21
22
9
8
34
11
35
37
17
23
16
14
41
38
36
13
39
24
6
15
1
26
27
28
29

Rank
0.468
0.468
0.468
0.468
0.468
0.343
0.271
0.211
0.209
0.205
0.194
0.163
0.156
0.130
0.115
0.109
0.086
0.073
0.073
0.064
0.064
0.064
0.064
0.063
0.060
0.058
0.058
0.058
0.058
0.058
0.058
0.058
0.052
0.043

Priority
FPI

Copyright 2014 John Wiley & Sons, Ltd.

(Continues)

56
61
62
65
68
51
8
26
25
6
31
17
47
42
52
41
36
66
67
19
22
23
37
49
11
57
58
59
60
63
64
20
16
29

Final priority
placement

C. SHARMA AND B. TYAGI

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

Copyright 2014 John Wiley & Sons, Ltd.

61
62
63
64
65
66
67
68

OPP
vector

0
0
0
0
0
0.333
0.333
0

TLBOI
(normalized)
0
0
0.781
0.781
0
0.781
0.781
0

GBOI
(normalized)
0
0
0.20
0.20
0
0.20
0.20
0

BOI
(normalized)
0
0
0
0
0
0
0
0

TLBOIS
(normalized)
1
1
0
0
1
0
0
1

GBOIS
(normalized)

Table VII. (Continued)

0.25
0.25
0
0
0.25
0
0
0.25

BOIS
(normalized)

0.468
0.468
0.058
0.058
0.468
0.073
0.073
0.468

FPI

2
3
30
31
4
18
19
5

Rank

0.043
0.042
0.039
0.032
0.019
0.018
0.018
0.012

Priority
FPI

32
46
34
45
48
14
44
12

Final priority
placement

RANKING OF PHASOR MEASUREMENT UNITS

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

C. SHARMA AND B. TYAGI

A close inspection of phase-1 of Table VIII reveals that in phase-1 itself, all important tie lines and
generators are having PMUs, to aid wide-area monitoring and control. First, ve locations correspond
to generators, which are important for SSS. Bus 51 aids in monitoring tie line between areas 1 and 4.
PMU placement at bus 8 will help in monitoring tie lines 89. Similarly, 25 and 26 PMU locations
help in monitoring tie lines connecting buses 12 and 127 respectively, which are between areas 4
and 5. Tenth location of phase-1 will provide measurements from generator-2 from depth rst. Last
location in phase-1, provides measurements for tie lines 4649. However, by the end of phase-1, all
critical tie-line buses and critical generator buses are observed, either by placing PMU directly on
bus or neighboring buses connected to critical buses. After second-, third-, and fourth-stage placement,
whole system is made observable. Proposed phased PMU placement approach makes all generators
and tie-line buses observable, which assists in small-signal monitoring in the initial phases.
Based on the previous text, it can be concluded that instead of placing PMUs on all generator buses
[1722], it is good to place few PMUs on signicant generator buses. From proposed method,
signicant PMU locations are easily identied based on control scheme designed for SSS.
Presented method is also compared with probabilistic technique of Reference [23]. In [23], authors
have only taken account of fast and slow dynamic observability of the system utilizing probabilistic
approach. Complete topological observability is not considered, in the work. Authors of [23] have
identied PMU locations for New England system, which only satisfy dynamic observability criteria.
However, proposed ranking and placement approach deal simultaneously with complete topological
observability and small-signal dynamics.
In addition to the previous text, both small-signal and transient stability analysis requires real-time
rotor measurements. Therefore, in proposed work, GBOI is introduced for small-signal analysis. This
GBOI is given highest priority, which ensures PMU placement on generator buses. Consequently,
proposed technique can be applied for transient stability studies also, without any signicant changes.
There may be situation, when a tie-line bus and its adjacent bus are suitable candidate for PMU
locations. TLBOI for this situation is not reported in [13]. In present work, TLBOI for this particular
case is also introduced. Proposed ranking technique is robust because it satises the robustness criteria
for different placement budget mentioned in [16]. Because PMUs are added sequentially in the system
and phase-1 is always subset of phase-2, -3, and -4.
Further, in [24], authors have used Monte Carlo-based probabilistic Eigen value analysis for various
operating conditions and uncertainties. In terms of performance, the proposed ranking and placement
will also work well for different operating scenarios. Ranking methodology presented in this paper
totally depends on control scheme employed. If there is any change in control algorithm, ranking of
PMUs will change. In such situation, problem of changing operating conditions can be solved by
employing a suitable adaptive control scheme, without affecting ranking. For example, variation in
loading conditions results into multiple operating conditions. At different operating conditions, generator speed signals will be different. According to proposed control scheme, this speed variation will
vary speed error and error derivative. This speed error will be reected as variation in universe of
discourse of fuzzy inputs. An adaptive controller such as fuzzy controller dealing with this variation
will work well for various operating scenarios, as described in Reference [21].
Table VIII. Phased PMU placement for 16-bus system.
No. of PMUS
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

Phase-1

Phase-2

Phase-3

Phase-4

56
61
62
65
68
51
8
26
25
6
31

17
47
42
52
41
36
66
67
19
22
23

37
49
11
57
58
59
60
63
64
20
16

29
32
46
34
45
48
14
44
12
-

Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

RANKING OF PHASOR MEASUREMENT UNITS

Suggested PMU placement technique ensures complete observability with high redundancy.
Therefore, once placement is accomplished, any changes in operating conditions will not affect system
performance as system is completely observable.

5.4. Wide-area damping controller

Change in Interarea Power (p.u)

After rst-stage placement, control scheme proposed in Section 3 is implemented, to check the effectiveness of the proposed placement scheme. WAFC and PID controller described in the Reference
[26,27] are utilized.

80
60
40
20
0
-20
-40
-60
-80
-100

10

12

14

16

18

20

Time(s)

8
WAFC
PID

6
4
2
0
-2
-4
-6
-8

10 12 14 16 18 20

Change in Inter area power (p.u)

Change in Inter area power (p.u)

Figure 5. Inter-tie power response without control.

6
WAFC
PID

4
2
0
-2
-4
-6

2.5
WAFC
PID

2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5

10 12 14 16 18 20

(b)
Change in Inter area power (p.u)

Change in Inter area power (p.u)

Time(s)
(a) Tie line between buses 1-2 (area4,5)

10 12 14 16 18 20

Time(s)
Tie line between buses 8-9 (area4,5)

0.8
WAFC
PID

0.6
0.4
0.2
0
-0.2
-0.4
-0.6

10 12 14 16 18 20

Time(s)

Time(s)

(c) Tie line between buses 50-51 (area1,4)

(d) Tie line between buses 1-27 (area4,5)

Figure 6. Inter-tie response with control.

Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

C. SHARMA AND B. TYAGI

To simulate the system for SSS, a disturbance of 5% increase in magnitude of reference voltage of
generator-1 has been considered between t = 2.0 and 2.5 s. The system is initially unstable even for
small disturbances as shown by Figure 5.
Figure 6(ad) shows the performance of fuzzy and PID controllers on four major tie lines of the
16-machine system. These tie lines are 12, 89, 5051, and 127. Figure 6 shows that fuzzy
controllers perform well, and control inter-tie oscillations satisfactorily.

6. CONCLUSION
In this paper, multi-staged PMU placement scheme has been proposed to monitor and control smallsignal instability in a large network. In the proposed approach, rst optimal placement sites are
computed, which make the system completely observable, even under loss of single PMU or transmission line. For the given control scheme, critical buses are determined based on modal controllability
and participation factors. Generator, tie line, and bus observability criteria are developed, and optimal
locations are ranked according to these criteria. AHP algorithm is utilized, and PMU placements are
carried out to make all generators and tie lines observable. From placement results, it is observed that
all critical generator and tie-line buses are observed in rst-stage itself. Therefore, control scheme can
be implemented after rst-stage placement, which will benet in wide-area SSS analysis. The proposed
scheme is robust for different stage budget and can be extended for transient and voltage stability
analysis. To check the effectiveness of control scheme, PMU measurements are used to damp interarea oscillations by wide-area fuzzy and PID controllers. Results show that the proposed placement
scheme provides a maximum advantage in terms of observability and stability of a large system.

7. LIST OF SYMBOLS AND ABBREVIATIONS

7.1. Symbols
x
y
u
q

state vector
output vector
control vector
left Eigen vector
Eigen value

7.2. Abbreviations
PMU
AHP
EMS
SSS
ILP
GBOI
TLBOI
BOI
CI
PM
W
FPI
OPP

phasor measurement unit

analytical hierarchy process
energy management systems
small-signal stability
integer linear programming
generator bus observability index
tie-line bus observability index
bus observability index
controllability index
pairwise matrix
weight vector
nal performance index
optimal PMU placement
REFERENCES

1. Ree JDL, Centeno V, Thorp JS, Phadke AG. Synchronized phasor measurement applications in power systems.
IEEE Transactions on Smart Grid 2010; 1:2027.
Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep

RANKING OF PHASOR MEASUREMENT UNITS

2. Zhang P, Li F, Bhatt N. Next-generation monitoring, analysis, and control for the future smart control center. IEEE
Transactions on Smart Grid 2010; 1:186192.
3. Phadke AG, Thorp JS, Karimi KJ. State estimation with phasor measurements. IEEE Transactions on Power
Systems 1986; 1:233241.
4. Xu B, Abur A. Observability analysis and measurement placement for systems with PMUs. Proceedings of IEEE
Power Systems Conference and Exposition 2004; 2:943946.
5. Abur A. Optimal Placement of Phasor Measurement Units for State Estimation. PSER publication, 2005.
6. Nuqui RF, Phadke AG. Phasor measurement unit placement techniques for complete and incomplete observability.
IEEE Transactions on Power Delivery 2005; 20:23812388.
7. Phadke AG, Thorp JS, Nuqui RF, Zhou M. Recent developments in state estimation with phasor measurements.
Proceedings of IEEE Power Systems Conference and Exposition 2009; 17.
8. Gou B. Optimal placement of PMUs by integer linear programming. IEEE Transactions on Power System 2008;
23:15251526.
9. Gou B Generalized integer programming formulation for optimal PMU placement. IEEE Transactions on Power
System 2008; 23:10991104.
10. Chakrabarti S, Kyriakides E. Optimal placement of phasor measurement units for power system observability. IEEE
Transactions on Power System 2008; 23:14331440.
11. Sodhi R, Srivastava SC, Singh SN. Optimal PMU placement method for complete topological and numerical
observability of power system. Electric Power System Research 2010; 80:11541159.
12. Dua D, Dambhare S, Gajbhiye RK, Soman SA. Optimal multistage scheduling of PMU placement: an ILP approach.
IEEE Transactions on Power Delivery 2008; 23:18121820.
13. Sodhi R, Srivastava SC, Singh SN. Multi-criteria decision-making approach for multistage optimal placement of
phasor measurement units. IET Generation Transmission Distribution 2011; 5:181190.
14. Aminifar F, Fotuhi F, Shahidehpour M, Khodaei A. Observability enhancement by optimal PMU placement considering random power system outages. Energy System 2011; 2:4565.
15. Aminifar F, Bagheri SS, Fotuhi FM, Shahidehpour M. Reliability modeling of PMUs using fuzzy sets. IEEE
Transactions on Power Delivery 2010; 25:23842391.
16. Qiao L, Tao C, Yang W, Rohit N, Franz F, Marija DI. An information-theoretic approach to PMU placement in
electric power systems. IEEE Transactions on Smart Grid 2013; 4:446456.
17. Chaudhuri B, Majumder R, Pal BC. Wide-area measurement-based stabilizing control of power system considering
signal transmission delay. IEEE Transactions on Power System 2004; 19:19711979.
18. Ni H, Heydt GT, Mili L. Power system stability agents using robust wide area control. IEEE Transactions on Power
System 2002; 17:11231131.
19. Yuan Y, Sun Y, Li G. Evaluation of delayed input effects to PSS inter-area damping control design. Proceedings of
IEEE Power Engineering Society General Meeting 2007; 15.
20. Ray S, Venayagamoorthy GK. Real-time implementation of a measurement-based adaptive wide-area control
system considering communication delays. IET Generation Transmission Distribution 2008; 2:6270.
21. Padhy BP, Srivastava SC, Verma NK. Robust wide area TS fuzzy output feedback controller for enhancement of
stability in multimachine power system. IEEE Systems Journal 2012; 6:426435.
22. Chompoobutrgool Y, Vanfretti L, Ghandhari M. Survey of power system stabilizers control and their prospective
applications for power system damping using synchrophasor-based wide area systems. European Transactions on
Electric Power 2011; 21:20982111.
23. Cepeda JC, Rueda JL, Erlich I, Colome DG. Probabilistic approach-based PMU placement for real-time power system vulnerability assessment. Proceedings of IEEE PES Innovative Smart Grid Technologies Europe 2012; 18.
24. Rueda JL, Erlich I. Probabilistic framework for risk analysis of power system small signal stability. Journal of Risk
and Reliability, Special Issue Paper 2012; 226:118133.
25. Sharma C, Tyagi B. An approach for optimal placement using binary particle swarm optimization with conventional
measurements, International Journal of Engineering, Science and Technology 2011; 3:5663.
26. Behbehani HM, Bialek J, Lubosny Z. Enhancement of power system stability using fuzzy logic based supervisory
power system stabilizer. Proceedings of 2nd IEEE International Conference on Power and Energy 2008; 479484.
27. Sharma C, Tyagi B. Wide area fuzzy controller design considering time latencies in communication networks.
Proceedings of 7th National Power System Conference, 2012.
28. Kundur P. Power System Stability and Control. Mc Graw-Hill: New York, 1994.
29. Rogers G. Power System Oscillations. Kluwer: Norwell MA, 2000.
30. Hamdan AMA, Elabdalla AM. On the coupling measures between modes and state variables and subsynchronous
resonance. Electric Power System Research 1987; 13:165171.
31. Hamdan AMA, Elabdalla AM. Geometric measures of modal controllability and observability of power system
models. Electric Power System Research 1988; 15:147155.
32. Hu D, Venkatasubramanian V. New wide-area algorithms for detection and mitigation of angle instability using
synchrophasors. Proceedings of IEEE Power Engineering Society General Meeting 2007; 18.
33. Saaty TL. Decision making with analytical process. International Journal Services Sciences 2008; 1:8398.
34. Power System Toolbox. Available online: http://www.ecse.rpi.edu/pst/PST.html
35. Matlab User Guide. The Math Works, Inc.: Natick MA, 1998.

Copyright 2014 John Wiley & Sons, Ltd.

Int. Trans. Electr. Energ. Syst. (2014)

DOI: 10.1002/etep