Page 1
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 1/10
572 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 23, NO. 2, MAY 2008
Assessment of Two Methods to Select Wide-AreaSignals for Power System Damping Control
Annissa Heniche , Member, IEEE , and Innocent Kamwa , Fellow, IEEE
Abstract—In this paper, two different approaches are applied tothe Hydro-Québec network in order to select the most effectivesignals to damp inter-area oscillations. The damping is obtainedby static var compensator (SVC) and synchronous condenser (SC)modulation. The robustness analysis, the simulations, and statis-tical results show, unambiguously, that in the case of wide-area sig-nals, the geometric approach is more reliable and useful than theresidues approach.In fact, this study shows that thebest robustnessand performances are always obtained with the stabilizer configu-ration using the signals recommended by the geometric approach.In addition, the results confirm that wide-area control is more ef-fective than local control for damping inter-area oscillations.
Index Terms—Compensator, control loop selection, geometricmeasures, inter-area oscillations, power system stabilizer, residues,
wide-area control.
I. INTRODUCTION
INTER-AREA oscillations have been observed in electrical
networks for many years [1]. Many power systems in the
world are affected by these oscillations [2]–[4] whose frequency
varies between 0.1 and 1 Hz. Currently, inter-area oscillation
damping is done with devices that use local signals. The basic
question we are asking here is: are these signals really the most
efficient?
In practice, the choice of measurement and control signals isa problem regularly faced by designers. In fact, to obtain the
desired performances and robustness, we have to select signals
that allow good observability and controllability of the system
modes. To quantify the observability and controllability of the
modes, measures have been defined in [5] and [6]. These mea-
sures, which are deduced from the Popov Belevich Hautus test
[7] and from residues, respectively, indicate how the th mode
is observable from available measurements and how it is con-
trollable from the system inputs. Thus, it is possible to select,
for each mode, the most efficient control loop.
By scientific curiosity, we wanted to know if the two methods
always lead at the same conclusion. Rapidly, we noted that it wasnot the case. The results of a first work were published [21], but
those associated with the 9 areas–23 generators test system [14]
were not. As Hydro-Québec is currently considering a project on
wide-area control, we thought that it was important to test the
two approaches on its network rigorously. In addition, even if
Manuscript received May 11, 2007; revised November 9, 2007. Paper no.TPWRS-00346-2007.
The authors are with IREQ, Hydro-Québec, Varennes, QC J3X 1S1, Canada(e-mail: [email protected] ; [email protected] ).
Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TPWRS.2008.919240
the results concern only the Hydro-Québec network, it is impor-
tant to notice that a statistical analysis was realized. This anal-
ysis allowed the test of the two approaches using 1140 different
configurations of the network.
The aims of this paper are on one hand to show that the two
measures do not provide the same conclusion in terms of con-
trol loop selection and on the other hand to demonstrate the effi-
ciency and reliability of one measure in comparison to the other.
To do that, the two measures were applied in order to select the
most effective control loops for damping the 0.6-Hz inter-area
mode of Hydro-Québec network. Local and global angle shiftswere considered. The inter-area damping is obtained by com-
pensators modulation. The modulation signal is produced by a
multi-band power system stabilizer (MBPSS) which uses only
intermediate frequency band [8]. The description and the pa-
rameters of this stabilizer are given in the Appendix.
This paper is organized as follows. Section II is devoted to
system modeling, while Section III presents a brief review of the
controllability-observability measures used in this work. Sec-
tion IV describes the application. Section V contains the re-
sults. Sections VI is devoted to the discussion of the results, and
Section VII is the conclusion.
II. SYSTEM MODELING
An electrical network is a nonlinear system which can be de-
scribed by the following nonlinear state equation:
(1)
where , and are the state,
input and output vectors, respectively. n is the dimension of
the system, m is the number of inputs, and p is the number of
outputs.
f: and g: are functions.
For measurement and control signals selection, a linear model
of the network is used. The latter is obtained using the modal
analysis tool developed at Hydro-Québec’s Research Institute
(IREQ) [9]. The linear state representation (A,B,C,D) of the net-
work is obtained using the identification eigensystem realization
algorithm (ERA) which was originally introduced in [10]. In the
context of electrical power systems, this approach was first ap-
plied in [2], [11] and then in [12], [13], [14], and [15]. The first
stage consists in exciting the nonlinear system by means of a
pulse of duration 0.4 s and amplitude of 1%. Thereafter, the ex-
citation u and associated outputs y are used by the ERA identi-
0885-8950/$25.00 © 2008 IEEE
Page 2
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 2/10
HENICHE AND KAMWA: ASSESSMENT OF TWO METHODS TO SELECT WIDE-AREA SIGNALS 573
fication algorithm which provides the linear state representation
of the system given by
(2)
where and are the state, inputand output vectors, respectively. and
are state, input, and output matrices, respectively.
III. SIGNALS SELECTION
For measurement and control signals selection, two dif-
ferent approaches were used: the geometric and the residues
approaches.
Let us consider the identified linear model of the network
given by (2). An eigenanalysis of matrix A produces the eigen-
values (assumed distinct for ) and corresponding
matrices of the right and left eigenvectors ] and, respectively. The eigenvectors and cor-
responding to are orthogonal and normalized, which implies
that ( is the complex conjugate transpose
of G, and I is the identity matrix of size n).
A. Geometric Approach
The geometric measures of controllability and observ-
ability associated with mode ‘i’ are defined as follows [5]:
(3)
(4)
In (3) and (4), is the th column of is the th row of
is the acute angle between the input vector and the left
eigenvector is the acute angle between the output
vector and the right eigenvector and are, respec-
tively, the modulus and the Euclidien norm of z. Using (3) and
(4), the joint controllability/observability measure is expressed
by
(5)
B. Residues Approach
The interconnected system transfer function associated with
the state (2) is expressed by
(6)
where is the residues matrix associated with mode
(7)
As shown in (7), the residues matrix depends on matrices
B and C and the right and left eigenvectors. Note that is
the complex conjugate transpose of . For and
, the elements ) of matrix are given by
(8)
Using the residues matrix, the joint controllability/observ-
ability measure is given by [6]
(9)
In [6], the controllability measure and observability mea-
sure are deduced from (9) by setting and ,
respectively
(10)
(11)
From (3), (4), (10), and (11), we can say that if is orthog-
onal to , then pole is uncontrollable from input . If is
orthogonal to , then pole is unobservable from output .
The signals and for which and are maximum are
the most ef ficient for damping mode ‘i’. Equation (9) shows that
the joint measure associated with mode ‘i’ is proportional to the
norm of the associated residues matrix. According to [16], this
means that, if the maximal value of the residues associated with
mode ‘i’ is obtained with input k and output l, then andare the most ef ficient signals to damp mode ‘i’.
C. Remarks
The two approaches show that the modal observability and
controllability is related to the orthogonality between the eigen-
vectors and the output and input vectors, respectively. Contrary
to the geometric approach, residues are independent of any
scaling of , and . Indeed, as shown in (9) –(11), for
observability and controllability analysis, the residues use the
magnitudes of and as measures without scaling. That
means that we are free to scale the left and right eigenvectors
arbitrarily. For example, we can increase the observability of
mode “i” in output l by increasing the magnitude of as
one wants by multiplying the right eigenvector with a positive
constant. In this case, the magnitude of decreases since
is required. In addition, using the geometric approach,
the magnitudes of and can be written as
(12)
(13)
In (12), represents the amount of information in the output
l. A higher norm means that the mode “i” is clearly present in
output l, which means more observability. On the other hand, in(13), represents the power injected into the system by the
Page 3
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 3/10
574 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 23, NO. 2, MAY 2008
Fig. 1. The 2001 Hydro-Québec Network.
input k. A higher norm means more power injected by the input
and then more controllability.
On the basis of these considerations, we can say that the re-
sults of modal analysis for input and output selection depend
critically on the scaling of the variables. This is the way to ex-press the relative contribution of each signal and then to rigor-
ously select and classify control and measurement signals.
IV. APPLICATION
A. Power System Description
The power system used in this study is represented in Fig. 1.
This is the 2001 Hydro-Québec peak load network, which is
characterized by its very good robustness and well-damped
inter-area modes. In order to obtain more oscillations, the line
between Micoua and Saguenay was removed. The network was
divided into nine electrically coherent areas [17].In Fig. 1, the numbers inside the little circles and squares
represent the number of SCs and static var compensators (SVCs)
connected at the associated substation. In addition, the numbers
and the associated words located in the lower left corner refer
to the substations located in the southern part of the system and
identified, in Fig. 1, by 1-2-3.
As shown in Fig. 1, the network includes six SVCs and four
synchronous condensers (SC). Table I describes the nomencla-
ture and the characteristics of the SVCs and SCs used in this
work.
Even if the network presents several inter-area modes [15],
we will devote our study to the 0.6-Hz mode which is considered
as the most dominant mode during major events. The two zonesaffected by this mode are James Bay and Churchill Falls.
TABLE I
SVC AND SC DESCRIPTION
Fig. 2. Control system configuration.
B. Power System Linear Model
The identified linear model of the Hydro-Québec network
represented in Fig. 1 is described by the linear state (2) with
the input vector u and output vector y given by the following
relations:
(14)
(15)
In (14), is the voltage reference of the compensator
) as represented in Fig. 2.
The output vector y corresponding to the 49 available mea-
surement signals includes 29 wide-area signals and 20 local sig-
nals. In (15), the subscripts G and L indicate wide-area refer-
enced and local measurements, respectively. Consequently, the
measurements coi, , and are global referenced
angle shifts whereas angle shifts and are local mea-
surements calculated by a line model which uses the voltage and
current phasors at bus k [18]. The definition of measurementsignals is given in Table II. For the global angle shift , the
Page 4
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 4/10
HENICHE AND KAMWA: ASSESSMENT OF TWO METHODS TO SELECT WIDE-AREA SIGNALS 575
TABLE II
AVAILABLE MEASUREMENTS DEFINITION
reference bus is equal to the Churchill bus, except if refers
to the Churchill plant. In this case, the reference bus is equal to
the Robert–Bourassa (LG2) bus.
The tools presented in Section III were applied to select the
most effective measurement and control signals to damp the
0.6-Hz inter-area mode of the Hydro-Québec network. As rep-
resented in Fig. 2, the damping is obtained by modulating the
reference voltage of the compensator with the stabiliza-
tion signal VPSS. The latter is produced by an MBPSS stabilizer
which uses a global or a local angle shift. Indeed, as shown in
Fig. 2, if the control system is in global mode, the input of the
PSS is , which represents the angle shift between two remote
buses. On the other hand, if the control system is in local mode,
the input of the PSS is equal to the angle shift between the com-
pensator bus (bus k) and a neighboring bus (bus nk or bus sk in
our case).
V. RESULTS
A. Residues Approach
For all compensators, the joint measures obtained using the
residues are given in Table IV. Taking into account the large
number of available measurements, for each control signal, we
chose to select only the measurements which exhibit the larger
residues. More precisely, for each compensator and for each cat-
egory of measurement signals as defined by lines of Table II, we
choose those which present the larger residues. The physical in-
terpretation of these measurements is given in Table III.
If we consider only the global measurement signals, the re-sults of Table IV show for each column and each row decreasing
value from top to bottom and from left to right, respectively.
That means that if we consider only one global measurement
signal as input for the PSS (one column in Table IV), then the
damping of the 0.6-Hz inter-area mode obtained in closed loop
depends on the modulation of the input of a given compensator.
More precisely, in closed loop, the damping obtained is larger if
the control site is SVC LVD or in other words if the modulation
concerns the voltage reference of SVC LVD. On the other hand,
the damping decreases if another compensator than SVC LVD
is selected. The diminution of the damping is observed from
top to the button, i.e., according to compensator controllability
classification given in the first column of Table IV. The samereasoning can be used to explain the decreasing value of each
TABLE III
SELECTED MEASUREMENTS SIGNALS
TABLE IV
JOINT MEASURES OBTAINED WITH RESIDUES APPROACH
row. In this case, if we consider one compensator, the damping
of the 0.6-Hz mode in closed loop depends on how observable
it is from the measurements.
Table IV shows that for damping the 0.6-Hz inter-area mode,
the best control strategy is to modulate the input of the SVC
LVD with a stabilization signal obtained by using the angle shift
between LA1 and Churchill. The results show that the
second choice corresponds to a joint measure equal to 0.8. In
this case, the control loops (y-u) are ( -SVC LVD) and
( -SC MAN). The third choice corresponds to a joint mea-sure equal to 0.7. It can be performed using the measurement
and by voltage reference modulation of compensators
NEM, ALB, and CHI.
B. Geometric Approach
Table V shows the joint measures obtained using the geo-
metric approach. As for the residues approach, only the signal
associated to the larger measures are listed. One notes that
the results obtained are different from those obtained with
the residues. Indeed, contrary to the residues, the geometric
approach recommends the angles shifts between CHI and
Churchill , ALB and Churchill , and LG2 and
Churchill . In addition, except for SC MAN and SVCLTD, the geometric approach leads to the conclusion that the
Page 5
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 5/10
576 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 23, NO. 2, MAY 2008
TABLE V
JOINT MEASURES OBTAINED WITH GEOMETRIC APPROACH
Fig. 3. Wide-area signal r coi.
best measurement is the angle shift between LG1 and Churchill.
At this stage, the question that remains to be answered is:
which is the most effective method? The answer to this question
is given by the comparative analysis.
C. Comparative Study
To compare the two approaches, we chose to consider the
control and measurement signals recommended by the residues.
Consequently, in what follows, the global measurement signals
are those which represent the columns of Table IV. Concerning
the control sites, only those with a controllability greater than
or equal to 0.7 (SVC LVD, SC MAN, SVC NEM, SVC ALB,and SVC CHI) will be considered.
1) Identification Results: Fig. 3 compares the output signal
coi of the network with that of the identified system. In this
case, the excitation is applied to the voltage reference of SVC
LVD.
As expected, the results obtained show that the system
remains stable even if the transmission line from Micoua to
Saguenay were removed. In addition, one observes a very good
superposition between the identified signal and the real signal
which is the output of the nonlinear simulation model obtained
using ST600 software. Note that the same results were observed
for all compensators and all measurements, which implies that
the identified linear models used for the signals selection andanalysis are good and reliable.
Fig. 4. Joint measures.
2) Control Loops Selection Results: Fig. 4 represents the
joint measures obtained with the two approaches. For each com-
pensator and for each method, the joint measures were divided
by the associated maximum measure.
The results show that the joint measures obtained with the
residues and the geometric approaches are different. As global
measurements are different from those given in Table V and
knowing that the geometric measures are scaling dependent, in
Fig. 4, it is normal that the conclusions concerning coi are
different from those in Table V.
For all compensators, as showed in Fig. 4, the results obtained
with the residues reveal that the observability of the 0.6-Hz
mode is maximum with the global angle shift between LA1 andChurchill . On the other hand, with the geometric ap-
proach, maximum observability is obtained with global angle
shift between NEM and Churchill buses in case of SVC
LVD and with global angle shift between LG1 and
Churchill for the other compensators. For all compensators, the
residues, contrary to the geometric approach, allow us to con-
clude that the worst wide-area measurement is coi. In ad-
dition, the results show that the conclusions concerning local
signals are the same with the two methods. Indeed, for the local
measurements, even if the values of the geometric measures are
higher than those obtained with the residues, for each compen-
sator, the two approaches recommend the same signals. For thefive selected compensators, in case of local measurement, ex-
cept for compensator MAN, the two methods reveal that the
best observability is obtained with local measurement .
On the other hand, in case of compensator MAN, the two ap-
proaches recommend the local measurement . Finally, the
results show that the joint measure associated to wide-area mea-
surements are greater than those related to local signals what
implies that global control is more effective than local control
to damp the 0.6-Hz inter-area mode.
3) Small-Signal Study Results: In this section, we compare
the performances and robustness of three PSSs: PSS RES,
which uses the global angle shift ( - )
recommended by the residues approach; PSS GEO, which,depending on the case, uses the measurement (
Page 6
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 6/10
HENICHE AND KAMWA: ASSESSMENT OF TWO METHODS TO SELECT WIDE-AREA SIGNALS 577
Fig. 5. Global measurement in open loop.
Fig. 6. Step at LG2, angle shift between Churchill plant, and LG2 plant buses.
Fig. 7. Step at LG2, angle shift between Churchill plant, and LG2 plant buseswith optimized PSS.
- ) or ( - ) recommended by
the geometric approach; and PSS LOC, based on local
measurements.
a) Performances: Figs. 6 and 7 show the responses of the
closed-loop system in the case of a step applied to the voltage
reference of plant LG2 (Robert Bourassa). For each compen-
sator, the angles in per unit in Figs. 5 and 6 are the angles shiftsbetween Churchill plant bus and LG2 plant bus.
Fig. 8. Step at LG2, delay of 200 ms.
Fig. 6 reveals that the performances obtained with the
PSS GEO are better than those obtained with PSS RES. As
shown in Fig. 5, this is due to the fact that, contrary to the
geometric approach, the signal classification obtained with
residues is based only on signals amplitude. In this case, if
the gain of the stabilizer is too large, the risk of closed-loop
instability is higher.
As the closed-loop performances depend on the PSS parame-
ters and the measurements used, the stabilizers were optimized.
The optimized parameters are given in the Appendix.
As shown in Fig. 7, the optimization was performed in order
to obtain similar small-signal performances for the three PSSs.
This approach was adopted to allow a fair comparison between
the two control-loop selection methods and between globaland local control. This methodology is used to compare the
three PSSs in terms of a compromise between robustness and
performances.
b) Robustness: The gain MG, phase MP, modulus MM,
and delay MR margins as well as the sensitivity S and the com-
plementary sensitivity T functions [19], [20] were used to eval-
uate the robustness of the closed-loop system.
The results of Table VI show that the highest gain margin is
obtained with stabilizer PSS GEO. For compensators LVD and
CHI, compared to PSS RES, the gain margins obtained with
the local PSS are higher. The opposite effect is observed with
compensators MAN, NEM, and ALB. In addition, except forcompensator MAN with local control, the modulus margins are
satisfactory. For all compensators, the highest modulus margin
is obtained with PSS GEO. Except for compensator MAN, the
phase margin obtained with all compensators and all PSS is in-
finite. As shown in Table VI, for SC MAN, the highest phase
MP and delay MR margins are obtained with PSS GEO, which
means that this stabilizer tolerates the most important delay.
This result is confirmed by the simulation results illustrated in
Fig. 8.
For compensators ALB and LVD, Figs. 9 and 10 illustrate
the sensitivity functions associated with the three PSSs. The
sensitivity function T is always less than 1, which means that
the stabilizers have a good robustness with respect to modeluncertainties. Between 0.1 and 10 Hz, global control guarantees
Page 7
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 7/10
578 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 23, NO. 2, MAY 2008
TABLE VI
ROBUSTNESS MARGINS
Fig. 9. Sensitivity functions S and T, SVC ALB.
a better robustness than local control and stabilizer PSS GEO is
more robust than PSS RES. Note that, except for compensator
MAN with local control, these results have been observed with
all PSSs and compensators. In addition, between 0.1 and 10
Hz, the sensitivity function S associated with PSS GEO does
not present a peak. This result, observed with all compensators,
implies that PSS GEO provides better performances than
PSS RES and PSS LOC.
4) Large-Signal Results: The behavior of the three PSSs was
tested on the 2003 Hydro-Québec network using the transient
stability program ST600. Three contingencies were considered.
The first, 3SVC19c, is a fault at Micoua with a the loss of the linebetween Micoua and Saguenay. The second, 3SVCnem, and the
Fig. 10. Sensitivity functions S and T, SVC LVD.
Fig. 11. Contingency m0100-3SVC19c.
third, 3SVCalb, are successive line losses in Robert Bourassa
(LG2) and LG4 corridors, respectively.
The results, illustrated in Fig. 11, show that the two global
PSSs have similar performances, both better than those obtained
with the local PSS. The oscillation frequency is around 0.32 and
0.27 Hz with the global and the local PSS, respectively. This
result is in accordance with the associated sensitivity function S
given in Fig. 9.As shown in Fig. 12, in the case of successive line losses
in corridor LG4, the best performances are obtained with the
global stabilizer PSS GEO. With the stabilizer based on the
residues approach, oscillations around 0.84 Hz are observed.
This result was foreseeable because, as shown in Fig. 10, the
sensitivity function S associated with SVC LVD and PSS RES
has a peak at this frequency.
Fig. 13 confirms that the best performances are obtained with
the stabilizer based on the geometric approach. In addition, the
frequency of the oscillations obtained with the PSS RES corre-
sponds to the resonance peaks observed on the associated sen-
sitivity functions.
5) Statistical Analysis Results: A total of 1140 contingen-cies were simulated for the statistical analysis. Among them,
Page 8
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 8/10
HENICHE AND KAMWA: ASSESSMENT OF TWO METHODS TO SELECT WIDE-AREA SIGNALS 579
Fig. 12. Contingency p0100-3SVCalb.
Fig. 13. Contingency p0100-3SVCnem.
780 associated with 60 load flows and 13 faults were applied to
the peak load network. The others, associated with 30 load flows
and 22 faults, were applied to the summer network, which cor-
responds to a load equal to 17 500 MW.
In the statistical analysis, to evaluate the impact of SVC andSC control on the damping of the 0.6-Hz inter-are mode, the
results in closed loop were compared with those obtained in
open loop, i.e., without PSS connected to the compensators. In
this paper, the open loop is considered as the reference case. In
addition, to compare local and global controls, two scenarios
were considered. In the first, only SVC NEM, SVC LVD, and
SVC ALB were equipped with local and global stabilizers. In
the second, in addition to these SVCs, the SC MAN was also
equipped with a PSS. The behavior of the stabilizers was com-
pared using four indicators. The first is the number of stable
cases obtained. The second is the number of voltage and/or
frequency criteria-violation cases. The third is the number of
loss-of-synchronism cases observed, and the last is the numberof cases of unstable before 2 s obtained.
Fig. 14. Statistical analysis in the case of three SVCs.
Fig. 15. Statistical analysis in the case of three SVCs and one SC.
In Figs. 14–16, for the open loop and the closed loop associ-
ated to each stabilizer, the bars represent the value of the four in-
dicators. The numbers from which these figures were obtained
are given in the Appendix. As shown in Figs. 14 and 15, the
two global PSSs are more ef ficient than the local PSS. Com-
pared to local PSS, the global stabilizers allow an increase in the
number of stable cases and a larger reduction of the number of
voltage and frequency criteria violations, loss-of-synchronism
cases, and unstable before 2 s cases. To be more precise, the
global stabilizers allow an increase around 3.5% of the numberof stable cases with respect to the reference case. Concerning the
number of criterion violation cases, with local PSS, the results
reveal an augmentation of 15% and 4% in the case of three and
for compensators, respectively, whereas with the global PSS,
one observes a diminution of 22% and 34%. Concerning the
increase in stable cases and the reduction of criteria-violation
cases, Figs. 14 and 15 show that the two global stabilizers are
equivalent. On the other hand, the results show that the reduc-
tion in the number of loss-of-synchronism and unstable before
2 s cases is more significant with the PSSbased on the geometric
approach.
Fig. 16 shows the statistical results obtained when a 200-ms
delay is added in the global control loops. In this case, only theSC MAN, SVC LVD, and SVC ALB are equipped with global
Page 9
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 9/10
580 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 23, NO. 2, MAY 2008
Fig. 16. Statistical analysis: Global PSS with 200-ms delay.
stabilizers. The results obtained show that the stabilizer based
on the geometric approach is more robust with respect to thedelay than that based on the residues approach.
In fact, with respect to the reference case, the number
of losses of synchronism increases by 15% with PSS RES,
whereas it increases by only 5% with PSS GEO. In addition,
the number of stable cases is lower with PSS RES and the per-
centage of reduction of criteria-violation and unstable before
2 s cases is the same with the two stabilizers.
VI. DISCUSSION OF THE RESULTS
In the statistical study, the time delay was increased from 0 to
200 ms. The results highlighted similar behavior of the global
stabilizers until 100 ms. Even if the technology today allows toobtain delays of around 50 ms, we chose to consider a longer
delay for two reasons. The first was, to evaluate the robustness
of the closed-loop with respect to the delay and neglected dy-
namics, while the second relates directly to the implementa-
tion solution cost. Indeed, as a future work, Hydro-Québec is
currently considering a project to test wide-area control on its
network.
VII. CONCLUSION
The geometric measures and the residues were used to
select the most effective control loops to damp the 0.6-Hzinter-area mode of the Hydro-Québec network. The damping is
obtained by compensators voltage reference modulation. Two
control scenarios were studied: global and local control. The
results showed that the two selection methods produce different
control loops. The analysis of open-loop signals reveals that,
contrary to the geometric approach, the residues recommend
high-amplitude signals independently of the phase. For all
compensators, the robustness analysis reveals that the PSS
based on the geometric approach is the more robust. On the
other hand, in terms of robustness, it is not possible to clearly
discriminate between the PSS based on residues and the local
PSS. All these observations were confirmed by small-signal
simulations which allowed the validation of the theoreticallycalculated robustness margins. In addition, in the case of severe
Fig. 17. MBPSS description.
TABLE VIICASE OF THREE SVCS EQUIPPED WITH PSS
TABLE VIIICASE OF THREE SVCS AND ONE SC E QUIPPED WITH PSS
TABLE IXDELAY OF 200 MS
TABLE XMBPSS PARAMETERS
TABLE XIMBPSS OPTIMIZED GAIN KI
contingencies produced with the transient stability program
ST600, for all compensators, the best performances were ob-
tained with the PSS based on the geometric approach. For all
contingencies, the link between the frequency oscillations and
the peaks observed on the sensitivity functions was established.
On the basis of the statistical study, we can conclude that global
control is more effective than local control to damp the 0.6-Hz
inter-area mode of the Hydro-Québec network. In addition,
concerning the control-loop selection, the results showed un-
ambiguously that the geometric approach is more reliable than
the residues approach, which is the one usually used in severalstudies. This last result confirms those presented in [21].
Page 10
7/30/2019 Assessment of Two Methods to Select Wide-Area
http://slidepdf.com/reader/full/assessment-of-two-methods-to-select-wide-area 10/10
HENICHE AND KAMWA: ASSESSMENT OF TWO METHODS TO SELECT WIDE-AREA SIGNALS 581
APPENDIX
Fig. 17 describes MBPSS. Table VII shows the case of three
SVCs equipped with PSS, Table VIII shows the case of three
SVCs and one SC equipped with PSS, Table IX shows a delay
of 200 ms, Table X shows the MBPSS parameters, and Table XI
shows the MBPSS optimized gain ki.
REFERENCES
[1] P. Kundur, “Investigation of low frequency inter-area oscillations prob-lems in large interconnected power systems,” in Canadian Electrical.
Association, Rep. 294T622, Ontario Hydro, Dec. 1993.
[2] I. Kamwa and L. Gérin Lajoie, “State-space system identification—to-ward MIMO models formodal analysis andoptimizationof bulk powersystems,” IEEE Trans. Power Syst., vol. 15, no. 1, pp. 326–335, Feb.
2000.[3] N. Martins, A. A. Barbosa, J. C. R. Ferraz, M. G. dos Santos, A. L. B.
Bergamo, C. S. Yung, V. R. Oliveira, and N. J. P. Macedo, “Retuningstabilizers for the north-south brazilian interconnection,” in Proc. IEEE Power Eng. Soc. Summer Meeting, July18–22, 1999, vol. 1, pp. 58–67.
[4] H. Breulman, E. Grebe, M. Losing, W. Winter, R. Witzman, P. Dupuis,M. P. Houry, T. Margotin, J. Zerenyi, J. Duzik, J. Machowski, L.Martin, J. M. Rodriguez, and E. Urretavizcaya, “Analysis and damping
of inter-area oscillations in the UCTE/CENTREL power system,”CIGRE 2000, Paris, Paper 38-113.[5] H. M. A. Hamdan and A. M. A. Hamdan, “On the coupling measures
between modes and state variables and subsynchronous resonance,” Elect. Power Syst. Res., vol. 13, pp. 165–171, 1987.
[6] M. Tarokh, “Measures for controllability, observability, and fixedmodes,” IEEE Trans. Autom. Control, vol. 37, no. 8, pp. 1268–1273,Aug. 1992.
[7] T. Kailath , Linear Systems. Englewood Cliffs, NJ: Prentice-Hall,1980.
[8] R. Grondin, I. Kamwa, G. Trudel, J. Taborda, R. Lenstroem, L. GérinLajoie, J. P. Gingras, M. Raine, and H. Baumberger, The Multi-BandPSS, A Flexible TechnologyDesignedto Meet OpeningMarkets,2000,CIGRE 39-201.
[9] I. Kamwa and L. Gérin-Lajoie, “State-space identification-towardsMIMO models for modal analysis and optimization of bulk powersystems,” IEEE Trans. Power Syst., vol. 15, no. 1, pp. 326–335, Feb.
2000.[10] J. Juang and R. S. Pappa, “An eigensystem realization algorithm for
modalparameter identification and system reduction,” J. Guid. Control,vol. 8, pp. 620–627, 1985.
[11] I. Kamwa, R. Grondin, J. Dickinson, and S. Fortin, “A minimal real-ization approach to reduced-order modelling and modal analysis forpower system response signals,” IEEE Trans. Power Syst., vol. 8, no.3, pp. 1020–1029, Aug. 1993.
[12] I. Kamwa, G. Trudel, and L. Gérin Lajoie, “Low-order black-boxmodels for control system design in large power systems,” IEEE Trans.Power Syst., vol. 11, no. 1, pp. 303–311, Feb. 1996.
[13] J. J. Sanchez-Gasca and J. H. Chow, “Computation of power systemslow-order models from time domain simulations using a Hankel ma-
trix,” IEEE Trans. Power Syst., vol. 12, no. 4, pp. 1461 –1467, Nov.1997.
[14] A. Heniche and I. Kamwa, “Control loop selection to damp inter-areaoscillations of power systems,” IEEE Trans. Power Syst., vol. 17, no.2, pp. 378–384, May 2002.
[15] I. Kamwa, A. Heniche, G. Trudel, M. Dobrescu, R. Grondin, and D.
Lefebvre, “Assessing the technical value of FACTS-based wide-areadamping control loops,” in Proc. IEEE Power Eng. Soc. General Meeting, Jun. 2005, vol. 2, pp. 1734–1743.
[16] N. Martin and L. T. G. Lima, “Determination of suitable locations forpower systems stabilizers and static VAR compensators for damping
electromechanical oscillations in large scale power systems,” IEEE Trans. Power Syst., vol. 5, no. 4, pp. 1455–1469, Nov. 1990.
[17] I. Kamwa, R. Grondin, and Y. Hebert, “Wide-area measurement basedstabilizing control of large power systems—a decentralized/hierar-chical approach,” IEEE Trans. Power Syst., vol. 16, no. 1, pp. 136–153,Feb. 2001.
[18] P. M. Anderson and R. G. Farmer, Series Compensation of Power Sys-tems PBLSH Inc., 1996.
[19] S. Skogestad and I. Postlethwaite , Multivariable Feedback Control: Analysis and Design. New York: Wiley, 1997.
[20] H. Bourles, “Systémes linéaires de la Modélisation á la commande,”Hermés, 2006.
[21] A. Heniche and I. Kamwa, “Using measures of controllability and ob-servability for input and output selection,” in Proc. 2002 IEEE Int.Conf. Control Applications, 2002, pp. 1248–1251.
Annissa Heniche (M’02) received the B.Eng. degree in electrical engineeringfrom Ecole Nationale des Ingénieurs et Techniciens d’Algérie in 1985 and theMaster’s and Ph.D. degrees from Paris 11 University, Paris, France, 1992 and1995, respectively.
She joined the Hydro-Québec Research Institute, Varennes, QC, Canada, in2001, where she is involved as a researcher in the Power System Analysis, Op-eration, and Control Department.
Dr. Heniche is a member of the Québec Order of Engineers.
Innocent Kamwa (S’83–M’88–SM’98–F’05) received the graduating degreeand Ph.D. degree in electrical engineering from Laval University, Qu ébec City,
QC, Canada, in 1988 and 1984, respectively.Since then, he has been with the Hydro-Québec Research Institute, Varennes,
QC, where he is at present a Principal Researcher with interests broadly in bulk system dynamic performance. Since 1990, he has held an Associate Professor’sposition in electrical engineering at Laval University, where five students havecompleted their Ph.D. under his supervision.
Dr. Kamwa is a recipient of the 1998and 2003IEEE PES Prize Paper Awardsand is currently serving on the System Dynamic Performance Committee,AdCom. He is a member of CIGRé