Design of Damping Controller Based on the

Wide-Area Signal Strength

1 Abhineet Prakash (Student member, IEEE), 2 Kundan Kumar and 3 S. K. Parida (Senior member, IEEE)
Department of Electrical Engineering, Indian Institute of Technology, Patna, India-801103
1
abhineetprakash1@[Link], 2 kkundan2893@[Link] and 3 skparida@[Link]
2020 21st National Power Systems Conference (NPSC) | 978-1-7281-8552-1/20/$31.00 ©2020 IEEE | DOI: 10.1109/NPSC49263.2020.9331866

Abstract—Due to enduring nature, low frequency oscillation Installation of power system stabilizers (PSSs) can sur-
(LFO) modes are most critical modes in a power system. pass inter-area modes through exciter of generators. Local
Therefore, these modes should be identified to design proper signals, which are conveniently available may not provide
control mechanism. In this article, wide-area power system
stabilizer (WAPSS) is designed to dampout the low frequency sufficient damping due to lack of observability. Moreover,
inter-area mode. First using detailed model of power system, rapid advancement in WAMS, communication networks and
linear state-space model is obtained at a stable equilibrium information technology provides flexibility to utilize wide-area
point. Then critical inter-area mode is identified amongst all low remote signals. Accessibility of such signals can overcome
frequency oscillation modes using small-signal stability analysis. the prior deficiency of obeservability problem. Furthermore,
Furthermore, based on highest geometric measures of joint
controllability and observability, control loop is selected. Phase it also gives the ease of access to damp the low-frequency
compensation and gain of WAPSS are determined by classical oscillations i.e. inter-area modes [4]. Several works have been
residue method and eigenvalue sensitivity method respectively. reported in the past regarding the design and synthesis of
The proposed method of WAPSS design is tested on 4-machine 11- control mechanisms using wide-area signals. Some of them
machine power system and comparatively complex New England- are residue based control [3], flexible AC transmission system
New York 16-machine 68-bus power system. Results obtained on
MATLAB/SIMULINK® platform shows the satisfactory damping (FACTS) based supplementary control [4], LQR/LQG control
performances with WAPSS for small disturbances. [5] and H2/H∞ control [6]. Because of simpler structure and
Index Terms—Power system stabilizer, Damping control, Wide- wide adaptibility, we have chosen residue based SISO WAPSS
area, Residue method, Small-signal stability, Inter-area mode, for damping of the critical inter-area mode.
Controllability and observability. Selection of proper signal for a single input single output
(SISO) WAPSS is also very important. To obtain desired
I. I NTRODUCTION damping performance, the selected control loop should have
better controllability and observability regarding a concerned
Due to interconnectivity of power systems, several modes mode. A. Heniche et al. have proposed two methods to
of oscillations takes place between different coherent areas. choose wide-area signals based on their strength. First method
Two modes i.e. local and inter-area mode are the main cause uses geometric measures whereas second method uses residue
of poorly damped oscillations in power sysem. In a frequency approach to choose input and output of a SISO WAPSS [7].
range of 1 Hz to 2 Hz, when oscillation of one generator In this paper geometric measure of joint controllability and
against rest of the system takes place, that mode is said to observability is used to chose control loop which provides
be local mode. Contrarily, when generators of a coherent area better damping performance with comparatively small gain.
swing against generators of another coherent area below 1 Hz, This paper is further organized into four sections. Section II
that mode is said to be inter-area mode. Impact of a local mode deals with the design of WAPSS which further includes modal
is localized while inter-area phenomenon is widely spread analysis, selection of control loop, determination of WAPSS
over the network. As the local power system stabilizers can gain, residue method to design the lead-lag parameters and
damp the local modes but due to growth of interconnectivity overall procedure of the proposed method. Proposed method to
among the coherent areas, control of inter-area mode became design WAPSS is demonstrated on 2-area 4-machine and large
a challenging task. Use of wide-area monitoring systems New England-New York 16-machine 68-bus power system in
(WAMS) can find a better possibility to dampout the low- Section III and IV respectively. Finally, conclusion has been
frequency inter-area modes [1], [2]. emphasized in Section V.
In a power system, various significant components like
generators and exciters are non-linear in nature. Generally, II. D ESIGN OF WAPSS
eigenvalue method is mainly used for modal analysis by lin- Overall schematic architecture of power system with wide-
earizing power system around a stable operating point. Many area damping control (WADC) is demonstrated in Fig. 1(a).
researchers have used reduced-order power system model Local PSSs are not capable to properly damp inter-area modes
for sake of simplicity. The unpredicted response of damping due to low observability. Therefore, WADC is adopted due to
conrol may be obtained due to simplified interaction between advancement of phasor measurement unit (PMU) technology
different modes [2], [3]. and high observability of feedback signals. Fig. 1(b) shows

978-1-7281-8552-1/20/$31.00 ©2020 IEEE

Authorized licensed use limited to: Indian Institute of Technology Patna. Downloaded on November 18,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 where, φ = [φ1 φ2 .....φn ] and ψ = [ψ1 ψ2 .....ψn ]T are the right
and left eigenvectors such that

Aφ = φΛ
(4)
ψφ = I
Λ is a modal matrix with [λ1 λ2 .....λn ] as diagonal elements
and other entries are zero whereas I is identity matrix.
A new state vector z is defined in such a way that x = φz.
Therefore equation (2) becomes

ż = φ−1 Aφz + φ−1 Bu = Λz + ψBu

(5)
y = Cφz + Du
The D matrix in equation in (2) and (5) are usually zero.
Therefore for a SISO system, input Ui to output Yo transfer
function can be written as,
Yo (s)
= Co φ(sI − Λ)−1 ψBi (6)
Ui (s)
where, Co is oth row and Bi is ith column of output matrix
Fig. 1. (a) Architecture of power system including WADC (b) Transfer C and input matrix B respectively. Now (6) can be further
function model of WAPSS simplified into,
n n
Yo (s)  Obsjo · Conji  Rjoi
the structure of WADC which is nothing but a classical lead- = = (7)
Ui (s) j=1 s − λj j=1
s − λj
lag type PSS. Now the transfer function of WAPSS can be
expressed as where, Obsjo is observability of Yo , Conji is controllability
 nb of Ui and Rjoi is the residue of closed-loop system Yo − Ui
1 + T1 s Tw s with respect to j th mode.
HW AP SS (s) = K (1)
1 + T2 s 1 + Tw s
B. Selection of control loop
where, K is WAPSS gain, T1 and T2 are lead-lag time
After modal analysis of the linearized system, it is important
constants, nb is number of lead-lag blocks and Tw is washout
to choose the control loop corresponding to critical inter-area
time constant.
mode. As geometric measures can assess the signal strength
A. System modelling and modal analysis of a control loop [7]. With respect to j th mode, geometric
measures of observability (gmojo ) and controllability (gmcji )
Modelling of a power system is usually combinations of is given as,
differential equations and algebraic equations. Differential  
equations generally represent synchronous machine dynamics    C o φ j 
gmojo = cos θ C o , φj =   (8)
on the other hand, algebraic equations represent network and C o  φj 
stator current equations [2]. The initial stable operating points
 
are obtained by proper load-flow solution and then system is    ψ j B i 
linearized, which is given by gmcji = cos β ψ j , B i =  
ψ j  B i  (9)

ẋ = Ax + Bu here, |x| and ||x|| are modulus and euclidian norm-2 of x

(2)
y = Cx + Du respectively. On the other hand, θ is the geometric angle
between output vector and right modal vector while β is the
where, x ∈ Rn is state variable, u ∈ Rp is input variable geometric angle between input vector and left modal vector
and y ∈ Rr is output variable of the system. On the other respectively with respect to j th mode.
hand, A ∈ Rn×n is state matrix, B ∈ Rn×p is input matrix, Now, for a SISO closed-loop control system Yo − Ui , ge-
C ∈ Rn×q is output matrix and D ∈ Rp×q is feedforward ometric measure of joint controllability/observability (JCOM)
matrix respectively. can be written as,
Suppose, λ is the eigenvalue of the state matrix A. Then,
they are also the non-trivial solution of equations gmcoj (i, o) = gmcji × gmojo (10)
Aφ = λφ control loop with comparatively large JCOM should be se-
(3)
ψA = λψ lected to design the controller.

Authorized licensed use limited to: Indian Institute of Technology Patna. Downloaded on November 18,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 Fig. 2. Residue compensation for j th mode

Fig. 3. A typical 4-machine 2-area power system

C. Determination of lead-lag time constants and gains
Due to inclusion of WAPSS, change in eigenvalue with TABLE I
respect to j th mode is given as [3] E LECTROMECHANICAL MODES WITH HIGHEST PARTICIPATION FACTORS

Mode Damping Frequency

Mode type Eigenvalue
Δλj = Rjoi KH (λj ) (11) no. ratio (%) (Hz)
1 Inter-area 0.02-3.76i -0.504 0.599
where, K is gain and H(λj ) is lead-lag compensation. 2 local -0.64-6.99i 9.133 1.113
Placement of the critical pole to the left half of complex 3 local -0.65-7.05i 9.204 1.123
plane can damp the critical mode. Residue method is adopted
to evaluate WAPSS time constants in this paper. If Rjoi
is the residue of SISO closed-loop control system Yo − Ui all the generators are represented by sub transient model along
with respect to j th mode. As shown in Fig. 2, the required with IEEE static exciter (ST1A) [8].
phase compensation φjoi for the shift of eigenvalue to the left
without change in frequency is given as, A. Modal analysis and control loop selection
After linearization of system about equilibrium stable point,
φjoi = π − arg Rjoi (12) modal analysis is conducted. All low frequency oscillation
modes are tabulated in Table I. From Table I, it can be
further, determination of lead-lag time constants can be ob-
identified that Mode 1 is the inter-area mode. Because of
tained as
T2 1 − sin (φjoi /nb) low frequency 0.599 Hz and relatively less damping ratio of -
α= = (13) 0.504%, Mode 1 has been identified as critical inter-area mode.
T1 1 + sin (φjoi /nb)
Mode shape of Mode 1 is presented in Fig. 4, from which it
1
T1 = √ (14) can be said that the generator 1 and 2 are rotating against
2πfj α generator 3 and 4.
T2 = αT1 (15)
th
here, fj is the frequency of j mode. As each block of
lead-lag compensator can compensate up to 60° properly.
Therefore, nb ≤ 3.
D. Procedure of the proposed approach to design WAPSS
• First, the system is linearized about equilibrium operating
point and state-space model is obtained.
• Modal analysis is done to find the critical inter-area mode
amongst all LFO modes.
• Depending on the joint controllability/observability mea-
sure (JCOM), control loop is determined.
• Lead-lag parameters and gain of WAPSS are obtained
using residue method.
• Lastly, the designed WAPSS is verified on original non-
linear models.

III. C ASE STUDY I

In this case, proposed methodology is carried on 4-machine
2-area power system [1] as demonstrated in Fig 3. The two
area power system is connected by a 220 KM long line. In
equilibrium operating condition, area-1 is delivering almost
440 MW power to area-2. Each area has two generators and Fig. 4. Mode shape for Mode no. 1

Authorized licensed use limited to: Indian Institute of Technology Patna. Downloaded on November 18,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 Fig. 5. Responses of 4-machine 2-area power system

TABLE II 20 % without change in frequency so the respective eigenvalue

JCOM ASSOCIATED WITH M ODE NO . 1 shift to the left is obtained as 0.3971 and 0.7870. Using (11),
Output/Input Vs1 Vs2 Vs3 Vs4 value of K is obtained 0.0049 and 0.0097.
I8−9 0.72 0.98 0.74 1.00 Based on the above discussion and using detailed model,
I7−8 0.72 0.97 0.73 0.99 simulation is performed on MATLAB/SIMULINK® platform.
P7−8 0.67 0.91 0.68 0.93
P8−9 0.67 0.9 0.68 0.92 At t=1 sec, reference voltage to exciter of generator 1 is
I6−7 0.52 0.71 0.53 0.73 increased by 5%. System responses of speed difference, angle
I9−10 0.50 0.68 0.51 0.69 difference and power flow of line (6-7 and 9-10) are shown
P9−10 0.33 0.45 0.34 0.46 in Fig. 5. From Table I, Mode 1 (critical inter-area mode)
P6−7 0.23 0.32 0.24 0.33
is unstable without WAPSS. From Fig.5, it can be noted that
the responses are improved sufficiently with proposed WAPSS
For the two area system, voltage references (Vs1 to Vs4 ) with K (0.0049 and 0.0097). Values of K are generally high
to the excitation system are considered for potential inputs when local signals are used to design power system stabilizer.
whereas power and current flow in lines are considered for One interesting thing can be noted that the value of K is very
potential outputs. To select the control loop, JCOM associated small when wide-area signal is used.
with Mode 1 is calculated and tabulated in table 2. JCOM is IV. C ASE STUDY II
normalized and the control loop with highest JCOM is selected
In this case, proposed WAPSS is designed for comparatively
to design WAPSS. From Table 2, it can be observed that the
complex IEEE 16-machine 68-bus power system. The overall
loop I8−9 − Vs4 has heighest JCOM. Therefore this loop is
system is divided into 5 area. New England test system
selected as control loop to design WAPSS.
(NETS) and New York power system (NYPS) are represented
by group of generators while other 3 areas are reprsented by
B. Design of WAPSS and simlation evaluation
equivalent 3 generators as shown in Fig. 6. Generators 1 to 13
Once the control loop is selected, residue corresponds to are equipped with IEEE DC exciter (DC4B) and generators
critical inter-area mode is obtained. Using (7) residue value 14 to 16 are equipped with IEEE static exciter (ST1A). To
is obtained as −4.04 − 18.15i. Now, according to (12), (13), damp local mode of oscillations, local PSSs are included with
(14) and (15) parameters of phase compensation are given as generators 1 to 6. All the generators are represented by sub
nb = 2, Tw = 10, T1 = 0.5538 and T2 = 0.1275. The value transient model [8].
of K plays a very important role, which is decided based on
increasing the damping ratio of critical mode. As it is shown A. Modal analysis and control loop selection
that the damping ratio of corresponding mode no. 1 is -0.504 System is linearized about stable equilibrium point then
%. Let us suppose the damping ratio is increased to 10 % and modal analysis is conducted to find inter-area mode. From

Authorized licensed use limited to: Indian Institute of Technology Patna. Downloaded on November 18,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 Fig. 6. A typical 16-machine 68-bus power system

TABLE IV
TABLE III JCOM ASSOCIATED WITH M ODE NO . 2
E LECTROMECHANICAL MODES WITH HIGHEST PARTICIPATION FACTORS
Output/Input Vs13 (×10−4 ) Vs14 Vs15 Vs16
Mode Damping Frequency I18−49 0.2108 0.2431 0.0481 1.0000
Mode type Eigen value
no. ratio (%) (Hz)
P49−46 0.2056 0.2371 0.0470 0.9755
1 Inter-area -0.42 - 2.48i 16.77 0.39
I46−38 0.2054 0.2369 0.0469 0.9745
2 Inter-area -0.12 - 3.37i 3.67 0.54
P46−38 0.1971 0.2273 0.0450 0.9351
3 Inter-area -0.34 - 3.86i 8.66 0.61
P18−49 0.1970 0.2271 0.0450 0.9343
4 Inter-area -0.17 - 4.91i 3.47 0.78
I49−46 0.1931 0.2226 0.0441 0.9159
I45−35 0.1825 0.2104 0.0417 0.8656
I38−31 0.1742 0.2009 0.0398 0.8264

Table 2, Mode 2 is found to be critical inter-area mode because

of low frequency 0.54 Hz and damping ratio 3.67%. Also,
mode shape of Mode no. 2 is demonstrated in Fig. 7, which
reveals that the generators of area-2 and area-5 are rotating
against the generators of area-1, area-3 and area-4.
There are several inputs (reference to excitation of generator
1 to generator 16 i.e. Vs1 to Vs16 ) and wide-area outputs (line
currents and powers) available from which a SISO control
loop can be formed. Out of which, control loop I18−49 −
Vs16 is selected to design WAPSS because this control loop
is associated with highest JCOM as given in Table IV.
B. Design of WAPSS and simlation evaluation
For Mode no. 2, residue value is obtained as -1.9887
+11.7434i with control loop I18−49 − Vs16 . Moreover, using
(12), (13), (14) and (15) parameters of phase compensation
are obtained as as nb = 2, Tw = 10, T1 = 0.6378 and
T2 = 0.1375. Value of K (equals to 0.0102) is selected based
on increasing the damping ratio of Mode no. 2 to 20% without
change in frequency.
Fig. 7. Mode shape for Mode no. 2
Simulation is conducted based on the above WAPSS design.
A 3-phase fault occurs in the middle of line 30-31 at 1 sec

Authorized licensed use limited to: Indian Institute of Technology Patna. Downloaded on November 18,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 Fig. 8. Responses of 16-machine 68-bus power system

and cleared at 1.5 sec. Responses (speed difference and power ACKNOWLEDGMENT
flow) of the benchmark 16-machine 68-bus system is shown Authors acknowledge the support from science and en-
in Fig. 8. From Fig. 8, it can be said that the responses are gineering research board (SERB) India under the scheme
improved when the proposed WAPSS is used with K equals CRG/2018/002915 to carry out this research work.
to 0.0102.
R EFERENCES
V. C ONCLUSION [1] P. Kundur, Power System Stability And Control. McGraw-Hill, 1994.
In this paper, a wide-area signal driven PSS is designed. [2] B. Pal and B. Chaudhuri, Robust Control in Power Systems. Springer
US, 2005.
Since power system is a multi-input multi-output system, a [3] J. Zhang, C. Y. Chung, and Y. Han, “A novel modal decomposition control
SISO loop is chosen to design residue method based power and its application to PSS design for damping interarea oscillations in
system stabilizer. To measure the signal strength, geometric power systems,” IEEE Transactions on Power Systems, vol. 27, no. 4,
pp. 2015–2025, nov 2012.
measures of controllability and observability are determined. [4] W. Yao, L. Jiang, J. Wen, Q. H. Wu, and S. Cheng, “Wide-area damping
Then based on highest joint controllability and observability controller of FACTS devices for inter-area oscillations considering com-
measure (JCOM), the appropritae control loop is selected. The munication time delays,” IEEE Transactions on Power Systems, vol. 29,
no. 1, pp. 318–329, jan 2014.
designed controller is then anticipated on 4-machine 2-area [5] A. K. Singh and B. C. Pal, “Decentralized control of oscillatory dynamics
power system. Furthermore, the WAPSS is also designed for in power systems using an extended lqr,” IEEE Transactions on Power
16-machine 68-bus power system to verify the robustness on Systems, vol. 31, no. 3, pp. 1715–1728, 2015.
[6] Y. Zhang and A. Bose, “Design of wide-area damping controllers for
a much complex and large system. From system responses, interarea oscillations,” IEEE Transactions on Power Systems, vol. 23,
as shown in Fig. 6 and Fig. 8, one can say that the proposed no. 3, pp. 1136–1143, aug 2008.
WAPSS can damp the concerned inter-area LFO mode. Due to [7] A. Heniche and I. Kamwa, “Assessment of two methods to select wide-
area signals for power system damping control,” IEEE Transactions on
highly controllable and observable loop, the gain of WAPSS Power Systems, vol. 23, no. 2, pp. 572–581, may 2008.
is reduced to a low value as explained in this paper. Finally, [8] A. K. Singh and B. C. Pal, “Ieee pes task force on benchmark systems for
we can conclude that the low frequency inter-area mode can stability controls report on the 68-bus 16-machine 5-area system,” IEEE
Power Energy Soc., vol. 3, 2013.
be damped using proposed WAPSS with a low gain.

Authorized licensed use limited to: Indian Institute of Technology Patna. Downloaded on November 18,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.