IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 33, NO.

6, NOVEMBER 2018 6037

Investigations of Large-Scale Voltage-Dependent

Loads for Damping Inter-Area Oscillations:
Mechanism and Robust Decentralized Control
Chen Zhang , Student Member, IEEE, Deping Ke , Yuanzhang Sun, Senior Member, IEEE,
C. Y. Chung , Fellow, IEEE, and Jian Xu , Senior Member, IEEE

Abstract—To uncover the mechanism by which voltage- conditions of the system and the types of excitation systems
dependent loads (VDLs) influence electromechanical dynamics, employed.
this paper mathematically deduces the damping torque induced
by VDLs based on a single machine infinite bus system. Moreover,
by continuously regulating the terminal voltage of a typical large- A. Load Control for Damping
scale VDL, i.e., the aluminum electrolysis load, this paper proposes
a novel load damping control architecture to dynamically modu- Although VDLs have been widely studied for primary fre-
late the power consumption of controlled loads so as to eliminate quency control and voltage stability [3]–[6], their architecture
interarea oscillations in power systems. Specifically, the decentral- and techniques concerning their control to alleviate inter-area
ized load damping controllers (LDCs) used in this architecture are
designed by a proposed multistage mixed H2 /H∞ control ap-
oscillations are rarely reported. Reference [7] gave the earliest
proach to guarantee robustness against uncertainties (e.g., tie-line outline of the general framework to improve angle stability by
outages and wind generation fluctuations) as well as to ensure rea- modulating the power of some conceptually controllable loads.
sonable control efforts of controlled VDLs. Simulation results on Furthermore, Kamwa et al. [8], [9] used multi-band stabilizers
the modified New England and New York interconnected system to conduct constructive and inspiring work on shedding the ac-
validate the method for analysis of VDLs’ effects on interarea oscil-
lations and show the proposed load damping control strategy can
tive power of loads to enhance stability of the frequency and
satisfactorily damp interarea oscillations over multiple operating electromechanical oscillations. Besides, an anticipated control
points. strategy was proposed in [10] to dispatch the steady-state power
consumption of loads, rather than dynamically modulate the
Index Terms—Large-scale voltage-dependent load, inter-area
oscillations, load damping control, robust decentralized control. loads’ power, so that the operating point is regulated to damp
inter-area oscillations. However, these studies utilize loads with-
out considering the necessary models and constraints thereof,
I. INTRODUCTION possibly leading to undesirable control effects in practical appli-
cations. For example, [9] pointed out that certain loads might be
NTER-AREA oscillations are longstanding and challeng-
I ing security issues in power systems due to their associ-
ation with groups of components that are usually distributed
ineffectively modulated in the most unstable cases due to strong
oscillations of the bus voltage.
The authors’ team has concentrated for many years on aca-
and up to hundreds of miles away. The voltage-dependent load
demic studies and industrial applications of direct load control
(VDL) significantly influences inter-area oscillations. Based on
techniques, specifically those based on aluminum electrolysis
the modal analysis and nonlinear time-domain simulations, [1]
loads (AELs) that are quite sensitive to voltage [4], [5], [11].
and [2] explored the influence of VDLs on electromechanical
The industrial AEL, as a typical representative of large-scale
oscillations and observed they are highly related to the loaded
controllable VDLs in power systems, has two major merits that
benefit power system dynamics: 1) the power consumption of
the AEL can be rapidly and smoothly modulated by regulating
Manuscript received July 7, 2017; revised November 15, 2017 and March 18,
2018; accepted July 3, 2018. Date of publication July 10, 2018; date of current its terminal voltage when series magnetically controlled reac-
version October 18, 2018. This work was supported by the National Natural tors (MCRs) are installed; 2) the normal AEL operates with a
Science Foundation of China under Grant 51777143. Paper no. TPWRS-01017- fairly large and steady daily power consumption (usually up to
2017. (Corresponding Author: Deping Ke.)
C. Zhang, D. Ke, Y. Sun, and J. Xu are with the School of Elec- hundreds of MWs) but it can endure terminal voltage deviations
trical Engineering, Wuhan University, Wuhan 430072, China (e-mail:, of ±10% for hours, which means it can offer considerable mod-
eezhangchen@whu.edu.cn; kedeping@whu.edu.cn; yzsun@mail.tsinghua. ulable active power to engage in dynamics control. Therefore,
edu.cn; xujianedu.cn).
C. Y. Chung is with the Department of Electrical and Computer Engineer- it is readily inferred that the control bandwidth based on the
ing, University of Saskatchewan, Saskatoon, SK S7N 5A9 Canada (e-mail:, AEL can not only cover the widely studied primary frequency
c.y.chung@usask.ca). control issues [5], [11] but also the control problems associated
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. with faster system dynamics, i.e., inter-area power oscillations
Digital Object Identifier 10.1109/TPWRS.2018.2854648 in this paper.
0885-8950 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 6038 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 33, NO. 6, NOVEMBER 2018

B. Contributions
Besides the inadequate investigation on characteristics and
models of the controlled loads, previous studies (e.g., [7]–[9])
of relevance also have little consideration towards the damping
controllers’ robustness in nowadays power systems. For exam-
ple, the influences of operating point drifts (e.g., caused by wind
power generation) on the damping control design have been ad- Fig. 1. The SMIB integrated with a voltage-dependent load.
dressed in [12]–[14]. In addition, those studies have not pro-
vided effective approach to constrain the damping controllers’
control efforts although it is well known that excessive control II. EFFECTS OF THE VDL CHARACTERISTICS
efforts may remarkably deteriorate the dynamics of controlled
VDLs. A. Modified SMIB System
By using the AEL as a specific example, this paper pro- To explore the general mechanism of the influence of VDLs
poses to supplementarily control the large-scale VDL to sup- on electromechanical oscillations, the SMIB system is employed
press inter-area oscillations in power systems. Similar to other as shown in Fig. 1. A third-order synchronous generator and
damping control implemented based on, e.g., HVDC [15], a high gain static exciter (AVR) are adopted in this system.
FACTS devices [16], and energy storage systems (ESSs) The local load VL at the terminal bus is voltage dependent. In
[17], supplementary load damping controllers (LDCs) using addition, the power output of the generator PG is divided into
wide-area feedback signals as the inputs are installed in the two parts: PL consumed by the local load VL and PT transmitted
VDLs (AELs). Furthermore, compared to the above-mentioned to the infinite bus through a tie-line.
studies associated with load control, this paper is the first To simplify the analysis, it is assumed that the load VL only
to systematically address the use of large-scale VDLs to consumes active power and its voltage-dependent characteristic
suppress inter-area oscillations in detail with respect to as- can be represented by a general exponential form [18]:
pects of the mechanism and damping control system. The  α
main three contributions of this paper are summarized as Vt
PL (Vt ) = P0 (1)
follows: V0
1) Based on the classical single machine infinite bus (SMIB)
system, the transfer function between the damping torque where P0 and V0 denote the nominal values of the load’s active
and the VDL is first deduced so as to mathematically ex- power and terminal voltage, respectively; Vt is the actual ter-
plain the phenomenon that the loaded condition of the minal voltage; and α is the exponent. This mathematical model
power system and the types of excitation systems em- can be employed to depict constant power (CP), constant current
ployed can differentially affect the influence of the VDL (CC), and constant impedance (CI) load types by setting α to
on the electromechanical oscillation, as observed in [1] be 0, 1, and 2, respectively.
and [2]. The nonlinear dynamic model of the SMIB system is lin-
2) A novel wide-area load damping control architecture earized around the nominal operating point, and the deviations
is proposed. By establishing the concrete model of of the electromagnetic torque ΔMe , the transient voltage ΔEq ,
AELs, its terminal voltage can be regulated by the se- the terminal voltage ΔVt , and the load power ΔPL can be
ries MCR. Thus, by virtue of wide-area measurement sys- computed as follows:
tems (WAMS), the proposed LDCs can synthesize the ⎧ 
⎪
⎪ ΔMe = K1 Δδ + K2 ΔEq
wide-area closed-loop control system and continuously ⎪
⎪
⎨ ΔE  = G3 (ΔEf d − K4 Δδ)
modulate the power consumption of the AELs to suppress q
(2a)
inter-area oscillations. ⎪
⎪ ΔVt = K5 Δδ + K6 ΔE  q
⎪
⎪
3) Because the AELs are generally distributed in distant ar- ⎩
ΔPL = KL ΔVt
eas, this paper proposes a multi-stage mixed H2 /H∞ ro-
bust control approach to design the LDCs with a decentral- K3
G3 = , ΔEf d = −Ge ΔVt
ized architecture. The coordinated LDCs can effectively 1 + K3 T  d0 s
and robustly suppress inter-area oscillations against uncer- KA P0
tainties in practical power systems (e.g., tie-line outages, =− ΔVt , KL = α (2b)
1 + Te s V0
wind generation fluctuations).
The remainder of this paper is organized as follows. Detailed where K1 ∼ K6 are the commonly used coefficients in the
analysis and discussion of the effects of VDLs are illustrated in small-signal stability analysis, and their detailed calculations are
Section II. Sections III and IV respectively introduce the wide- presented in Appendix A [18]–[20]; ΔEf d is the deviation of

area load damping control strategy and the robust LDC design the exciter’s output voltage; T d0 is the d-axis open-circuit tran-
procedure. Simulations of the modified New England and New sient time constant; KA and Te are the gain and time constant
York interconnected system are conducted in Section V. Finally, of the exciter, respectively; and KL represents the sensitivity of
Section VI concludes the paper. the load power with respect to the terminal voltage.

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 ZHANG et al.: INVESTIGATIONS OF LARGE-SCALE VOLTAGE-DEPENDENT LOADS FOR DAMPING INTER-AREA OSCILLATIONS 6039

Fig. 2. Electromagnetic torque induced by the VDL.

Substituting (2b) into (2a) yields the open-loop transfer func-

tion from the deviation of the power angle Δδ to ΔVt :
ΔVt K5 − K4 K6 G3
GV (s) = = (3) Fig. 3. A phasor diagram of torque ΔM L with positive and negative K 5.
Δδ 1 + K6 Ge G3
Generally, the gain KA of an exciter in modern power systems
is fairly large and its time constant Te is very small. Thus, the
negative number, and its sign depends on the operating point and
transfer function (3) can be further simplified:
the system’s structure and parameters. ΔML D will be positive
K6 (K4 + K5 KA ) with a positive K5 , indicating that the load induces a positive
GV (s) ≈ K5 − (4)
K6 KA + T  d0 s damping torque and is beneficial to the electromechanical os-
Similarly, based on (2), the open-loop transfer function Δδ → cillation dynamics (Fig. 3). In contrast, if K5 is negative and
ΔPL can be deduced: K4 + K5 KA < 0, the load will have adverse effects on damping
  power oscillations compared to the previous case. Moreover, the
ΔPL K6 (K4 + K5 KA ) adverse effects will be seriously amplified by KA and KL in (6).
GL (s) = = KL K5 − (5)
Δδ K6 KA + T  d0 s As mentioned in [18] and [20], negative K5 easily emerges in
Using the complex torque method, [18] investigated the stressed operating conditions such as transmitting heavy power
mechanism of the excitation system in detail. This paper de- over long-distance tie-lines. Under such stressed conditions, the
pends on its work and additionally introduces a feedback loop VDL would therefore deteriorate damping ratios of the oscil-
(Δδ → ΔML ) to investigate the electromagnetic torque ΔML lation modes and lead to insufficient damping performance in
induced by the voltage-dependent load VL, as shown in Fig. 2. power systems.
Routinely, ΔML is orthogonally decomposed into the synchro- Although the stressed operating conditions will force the VDL
nizing torque ΔML S and the damping torque ΔML D in the to be adverse to small-signal stability, it is inferred that the phase
complex plane, as follows: of the torque induced by the VDL can be rotated by reasonably
regulating its terminal voltage so that positive damping effects
ΔML = ΔPL = ΔML S Δδ + ΔML D Δω (6a) are produced. Specifically, this auxiliary terminal voltage con-
⎧ 2 trol for VDLs can induce a supplementary positive damping
⎨ ΔML S = K L (K 5 T2 d 0 2ω −K2 4 K2 6 K A )
2 2

K K +T ω torque ΔMS D (green arrow in Fig. 3) and thus be beneficial for

6 A d0
(6b)
⎩ ΔM K L K 6 T  d 0 (K 4 +K 5 K A ) suppressing electromechanical oscillations in power systems.
LD = K 62 K A2 +T d20 ω 2
The detailed control architecture is introduced in the following
section.
Therefore, the phase shift from Δδ to ΔML (caused by
In particular, the main purpose of utilizing SMIB system is to
GV (s)) is critical for the load to produce damping torque.
clearly illustrate the way on which VDLs impact the electrome-
According to (5) and (6), the system component parameters
chanical oscillations’ damping. Indeed, there will be no theoret-
and the operating point primarily determine the phase shift and
ical difficulty to perform the induced damping torque analysis in
thus the load’s damping effect, which will be discussed in detail
a multi-machine system. For example, [21] deduces the damping
in the following.
torque which are induced by FACTS devices in a multi-machine
system. The investigation results based on the SMIB system or
B. Discussion
multi-machines will make no essential difference. Thus, this pa-
Generally, the synchronizing torque impacting the transient per will subsequently employ a multi-machine system to further
stability is adequate in modern power systems due to the high confirm the correctness of the above investigation.
gain and sensitive AVR [18], and is not of special concern in
this paper. However, the situation becomes complicated for the
damping torque and requires detailed investigations. III. PROPOSED VDL-BASED LOAD DAMPING
CONTROL FRAMEWORK
It is found from (6) that the sign of ΔML D is mainly decided
by the parameter K5 because the other parameters are always Compared to previous works [7]–[10], this paper is the first
positive [18]. Appendix A shows that K5 can be a positive or to propose in detail a novel load damping control architecture

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 6040 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 33, NO. 6, NOVEMBER 2018

to LDCs and then damping output signals Δu are added to

a series thyristor-controlled rectifier for MCRs to change the
series impedance ΔX between a public bus PB and a load
terminal bus TB (the transfer function GM (s) for the MCRs is
seen in [22]). Thus, the terminal voltage VA will be regulated
by ΔVD and, consequently, the controlled AEL is continuously
modulated to provide damping efforts to inter-area oscillations.
After inter-area oscillations subside, LDCs will not be active
and controlled AELs would return to the original steady state.

B. Available Active Power Capacity of the AEL for

Fig. 4. The proposed wide-area closed-loop control structure.
Damping Control
For practical considerations, the LDC output magnitude
should be limited to prevent excessively exerting the AEL and
ensure the quality production of aluminum. With this limitation
(shown in Fig. 5), it is therefore necessary to assess the available
active power capacity provided by AELs for damping control.
Generally, the active power PA consumed by the AEL can be
calculated as follows:

PA = aVA2 + bVA (7)

where a and b are coefficients depending on the resistance and

Fig. 5. The basic framework of proposed control. electromotive force of the AEL model, and VA is the terminal
voltage of the AEL. For example, [11] uses actual measure-
ments to identify values of a and b of 4.96 and −1.66, respec-
for the VDL and specifically assess the damping capabilities of tively. Moreover, the nominal power of the AEL is 330 MW at
controlled VDLs. The strategy can not only mitigate the adverse VA = 1.0 p.u. Thus, as VA is permitted to vary between 0.9 and
effects from the VDL but also effectively suppress inter-area 1.1 p.u., the power consumption of the AEL has a minimum of
oscillations in power systems. 252 MW and maximum of 417 MW; this indicates that up to
±20% of the AEL’s active power capacity is available for power
modulation to serve in the damping control. Hence, considering
A. Basic Principle and Architecture
that the nominal capacities of AELs are usually quite large, they
In modern power systems, many types of VDLs can com- can definitely spare adequate dynamically controllable capac-
monly tolerate wide and frequent variations of supplied voltage ity to improve power system stability. In other words, similar
[3]. In this paper, a typical example of such a large-scale VDL, to the characteristic of energy storage systems, the controlled
the AEL (powered by DC voltage) [4], is specifically stud- AELs also have the potential to offer active power modulation
ied. Furthermore, industrial AELs in China have been gradually to eliminate inter-area oscillations. Furthermore, the location of
subjected to technical improvements to equip the MCRs, which an AEL (or other types of VDLs) is another important factor
are in series with the AELs, to regulate their terminal voltage. to influence its capability for controlling electromechanical dy-
Specifically, MCRs have the advantages of good reliability, large namics. In fact, where the industrial VDLs with large capacities
capacity, and high voltage [22], [23]. Moreover, an AEL’s termi- are placed primarily depends on the practical needs and require-
nal voltage can be smoothly and quickly adjusted by the MCR. ments. For example, AELs are generally sited close to aluminum
In this paper, the series MCRs provide a convenient way to en- mines. For simplicity, this paper however concentrates on the
gage AELs in dynamics control because they can also rapidly design of LDCs which modulate the power of VDLs, and it is
respond to supplementary damping control signals. assumed that the studied VDLs have the locations which will
Based on the feasibility of implementing supplementary con- not affect their capabilities in controlling the electromechanical
trol of the AELs accompanied by MCRs, this paper further oscillations.
proposes to reshape the natural damping characteristics of the It should be noted that this paper mainly aims to investigate
AEL for suppressing inter-area oscillations of power systems. the mechanism by which VDLs influence electromechanical
Fig. 4 shows the proposed wide-area damping control archi- dynamics and further to propose a novel load damping con-
tecture based on the AELs. Briefly, the LDCs receive remote trol strategy to dynamically modulate the power consumptions
feedback signals from the WAMS center and generate auxiliary of controlled VDLs so as to suppress inter-area oscillations in
input signals to control the rectifier of the MCRs. power systems. Specifically, both the mechanism investigation
Moreover, Fig. 5 depicts the detailed series framework and and subsequent robust damping control design are presented
control blocks of controlled AELs. When inter-area oscillations based on general VDL models which can be flexibly specialized
occur in power systems, wide-area input signals will be delivered to a variety of VDLs in practice, such as electrolysis loads [4],

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 ZHANG et al.: INVESTIGATIONS OF LARGE-SCALE VOLTAGE-DEPENDENT LOADS FOR DAMPING INTER-AREA OSCILLATIONS 6041

Fig. 6. Configuration of the robust control strategy.

Fig. 7. Illustration of the multi-stage D-contour.

LED lighting [6], and electric water heating [24]. Namely, the
proposed control strategy can be compatibly applied to these y can be formulated to measure the impact of a disturbance,
VDLs if providing detailed load models. Therefore, the AEL as follows:
which is a specific yet typical VDL is employed in this pa-
per to complete the exemplary demonstration of the proposed S(s) = (I − G(s)D(s)K(s))−1 (9)
investigation and control design method.
Therefore, in the H2 channel, the transfer matrix between w and
IV. MULTI-STAGE ROBUST DESIGN STRATEGY z2 is given by
FOR DECENTRALIZED LDCS 
W1 (s)S(s)
A. Multi-Stage Robust Design Strategy Tw z 2 (s) = (10)
W2 (s)K(s)S(s)
A robust design strategy mainly aims to enhance the robust-
ness of LDCs against external disturbances and model uncer- By minimizing the H2 norm Tw z 2 (s)2 , the resultant LDCs will
tainties, such as tie-line outages and wind generation fluctua- optimize the control effort as well as achieve output disturbance
tions. In this paper, a mixed H2 /H∞ output-feedback control rejection. Accordingly, as for the H∞ channel, the transfer func-
approach is used to achieve the multi-objective robust control tion between w and z∞ is given by
purpose [25]. Moreover, H∞ control can guarantee stability of
a closed-loop system against uncertainties while H2 control Tw z ∞ (s) = W3 (s)K(s)S(s) (11)
can optimize transient responses of the system, e.g., the control
efforts of controllers [26]. Meanwhile, the required damping ef- Similarly, the closed-loop system can enhance the immunity of
fects are fulfilled by implementing constraints on the placement its stability by reducing the H∞ norm Tw z ∞ (s)∞ in the presence
of closed-loop poles in the control design. In addition, linear of system uncertainties.
matrix inequality (LMI) optimization is a powerful and mature Considering the wide distribution of the AELs and the imple-
tool currently used to solve linear control problems. mentation reliability of the designed controllers [27], a multi-
The detailed configuration of the proposed design control stage mixed H2 /H∞ robust control strategy is proposed in this
strategy is depicted in Fig. 6. G(s) is the transfer function of paper to design the decentralized LDCs in a sequential manner.
the open-loop power system and K(s) represents the transfer Meanwhile, to effectively suppress the inter-area oscillations,
function of the LDCs to be designed. Due to the utilization the relevant inter-area modes should be located in the proper
of wide-area feedback signals, the transmission time delay (τ ) region of the complex plane, such as the D-contour shown in
should be taken into account during the control design; thus, it Fig. 7. Therefore, a multi-stage region pole placement (RPP)
is approximated in this paper by the second-order Pade formula is proposed to step-by-step shift inter-area modes into the fi-
(D(s)) which is simple yet more accurate than the first-order nal desirable region. Specifically, each stage of the optimization
Pade formula for approximating time delays associated with the updates the desirable region by gradually narrowing it, as illus-
wide-area damping control, as follow: trated in Fig. 7, so that the damping burden can be allocated for
different LDCs. The shadow region confined by the D-contour
τ 2 s2 − 6τ s + 12
e−τ s ≈ (8) in Fig. 7 can be described as an LMI and expressed as follows:
τ 2 s2 + 6τ s + 12


D = z ∈ C : fD (z) = L + zM + z̄ M̄ < 0
In the H2 channel, W1 (s) is a low-pass filter used to reject
⎡ ⎤ ⎡ ⎤
output disturbances while W2 (s) is a high-pass filter or simply −2α 0 0 −2α 0 0
a small constant to prohibit excessive control efforts of con- ⎢ ⎥ ⎢ ⎥
L=⎣ 0 0 0⎦ M = ⎣ 0 sin θ − cos θ ⎦ (12)
trollers. W3 (s) in the H∞ channel is also a high-pass filter to
guarantee robustness against system uncertainties. These filters 0 0 0 0 cos θ sin θ
are responsible for picking out the system dynamics within fre-
quency range of interest. Moreover, the sensitivity transfer func- At each stage of the design, one LDC is derived by the pro-
tion between the disturbance input w and the measured output posed mixed H2 /H∞ robust control strategy, which can be

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 6042 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 33, NO. 6, NOVEMBER 2018

Fig. 8. Modified New England and New York interconnected system.

Fig. 9. Root loci of four inter-area modes in two cases.

formulated as follows: select the most effective wide-area feedback signals for
damping control;
min ω∞ Tw z ∞ (s)2∞ + ω2 Tw z 2 (s)22 2) Shaping of the closed-loop system: A reduced-order model
K (s) (13)
s.t. λ∈D G(s) of the open-loop power system is obtained by the
Schur method [28] to synthesize the low-order LDC.
where ω∞ and ω2 are the weights for the H∞ and H2 perfor- Three weighted filters in Fig. 6 are also chosen and the
mance, respectively. Here, these two weights make clear senses closed-loops are shaped;
to impact final control results. For example, large ω∞ will tend to 3) Solving the optimization model: By adjusting the weights
enhance the robustness of the resultant LDCs against uncertain- of ω∞ and ω2 in (13), robustness and control efforts of the
ties but it possibly leads to relatively poor transient responses closed-loop system can be flexibly and feasibly assigned.
of the system and also excessive damping efforts from the con- Then, via the robust control toolbox of MATLAB, this
trolled AELs; ω2 however plays just opposite role in affecting optimization model (13) can be solved and the resultant
the control results. Therefore, by using these general properties robust LDC obtained.
as the guideline, the trial and error method can also efficiently 4) Iterative progress through nonlinear time domain simu-
adjust ω∞ and ω2 to fulfill the desired control objectives [26]. λ lations: The control efforts of the resultant LDCs should
is the set consisting of the closed-loop system’s poles of concern. be checked to see whether the controlled VDLs are ex-
cessively exerted. If so, the weights ω∞ and ω2 should
B. Design Procedure of the Decentralized LDCs be properly adjusted and Step 4 repeated until this issue
is resolved. Similarly, the required damping effects and
1) Small signal analysis of the open-loop power system: Lin- robustness can be obtained through the time domain sim-
earize the open-loop power system model at the nominal ulations and iterative tuning of the weights.
operating point, identify the critical inter-area modes, and

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 ZHANG et al.: INVESTIGATIONS OF LARGE-SCALE VOLTAGE-DEPENDENT LOADS FOR DAMPING INTER-AREA OSCILLATIONS 6043

Due to the sequential design, once a robust LDC is derived

via Steps 1–4, the open-loop system model is updated and the
RPP is also narrowed (Fig. 7). Those preparations precede the
design of the next LDC, which follows the same steps.

V. SIMULATION RESULTS AND ANALYSIS

The classic 5-area 16-machine test system (known as the
New England and New York interconnected system [29]), as
shown in Fig. 8, is used to exhibit the load damping effects
presented in Section II and validate the proposed robust load
damping control design. Three DFIG-based wind farms W1,
W2, and W3 with capacities of 800, 800, and 1200 MW, respec-
tively, are additionally connected at Bus-68, -39, and -18 in this
benchmark system. All these wind farms are comprised of a
same type of wind turbines with the active power capacity of 2 Fig. 10. Oscillation curves for Case 1.
MW (the parameters of the wind turbine are shown in Appendix
B). In particular, each wind farm is aggregately represented by
a single wind turbine during the simulation, according to the
aggregation method introduced in [30]. Moreover, the active
power loads of Bus-68, -39, and -18 are increased by 800, 800,
and 1200 MW, respectively [31]. Computing the eigenvalues
of this modified system shows four poorly damped inter-area
modes of oscillations (identified by M1, M2, M3, and M4).

A. Investigation of Damping Effects from VDLs

Two operating conditions of the multi-machine system are
employed in this investigation:
Case 1–Stressed Condition: High gain (KA = 200) and in-
sensitive excitation systems (Te = 0.36 s) [18] are used for all
of the generators; one tie-line between Bus-53 and -54 and one
tie-line between Bus-60 and -61 are switched off.
Case 2–Normal Condition: All of the generators are equipped Fig. 11. Oscillation curves for Case 2.
with excitation systems of the IEEE-DC1A type (KA = 20 and
Te = 0.05 s) [18] and all of the loads are decreased by 10%.
Compared to CP loads, which consume constant power irre- the CI load at Bus-17 induces a negative damping torque on the
spective of terminal voltage deviations, CI loads are sensitive to shaft of G13 and thus weakens the damping of inter-area modes.
terminal voltage. Therefore, it is assumed that all loads in the However, Fig. 10 shows the phase difference between ω13 and
test system are composed of CI and CP loads with proportions P17 is less than 90° when 100% CI loads are also employed in
β and (1 − β), respectively. As β increases from 0 to 100%, the Case 2. Therefore, the damping torque of G13 is additionally
root loci of the four inter-area modes (eigenvalues) in the two reinforced by the loads.
cases move as depicted in Fig. 9, where arrows show the direc-
tions of the modes. The four inter-area modes in Case 1 move
B. Application of Robust Decentralized LDCs
to the right as β increases. Specifically, M1 and M2 even move
to the unstable region. Such results demonstrate that, under Reference [29] specifically addresses the infeasibility of in-
stressed conditions, VDLs impair the damping of inter-area os- stalling PSSs in the four largest machines (G13–G16) to damp
cillations in power systems. In contrast, the leftward movements the inter-area oscillations because these machines are actually
of the inter-area modes in Case 2 indicate that VDLs in normal aggregated models. Thus, the alternative solution proposed in
operating conditions have positive impacts on the inter-area this paper to modulate large-capacity VDLs (as an replacement
modes. of PSSs installed in G13–G16) for damping control is an exact
Nonlinear time domain simulations are conducted to confirm fit for this test system.
the above modal analysis. After a small disturbance, Fig. 10 It is supposed that three AELs (AL1, AL2, and AL3), se-
illustrates the oscillation curves for the angular speed (ω13 ) lected as typical controllable VDLs, are integrated into the test
of G13, voltage amplitude (V17 ) of Bus-17, and active power system (Fig. 8). Specifically, three AELs hypothetically plac-
consumption (P17 ) of the CI loads at Bus-17. Clearly, ω13 and ing at Bus-17, -27 and -42, respectively, are just for validating
P17 are almost anti-phase when 100% CI loads are used in the effectiveness of the proposed control design. In Area 4, the
Case 1. According to Section II and Fig. 2, this indicates that load replaced at Bus-42 by AL1 is 970 MW (out of the original

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 6044 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 33, NO. 6, NOVEMBER 2018

TABLE I TABLE II
REPRESENTATIVE OPERATING POINTS RESULTS OF EIGENVALUE ANALYSIS OVER MULTIPLE OPERATING POINTS

in the optimization model (13) are set to 0.55 and 0.45, respec-
tively, to simultaneously account for the robustness and required
damping performance. The three LDCs are sequentially synthe-
1850 MW); in Area 2, a portion (730 MW) of the original load sized by three stages, and the damping ratio used for constructing
(6000 MW) at Bus-17 is replaced by AL2; and, in Area 1, the the gradually compressed LMI region (Fig. 7) is 0.1, 0.125, and
original load connected at Bus-27 is replaced by AL3 with the 0.15, respectively, for these stages. By solving the optimization
same capacity of 640 MW. The models and parameters of the model (13), the transfer functions of three LDCs are obtained,
AELs are based on studies of an isolated power system in Inner as follows:
Mongolia, China [11]. Moreover, each AEL is assumed to be HLDC1 (s)
accompanied by an MCR so that its terminal voltage can be
rapidly and smoothly regulated. At the nominal operating point, 0.095s4 + 10.12s3 − 69.68s2 + 25.87 s + 0.004
= 0.76
power outputs of three wind turbines are supposed to be their s4 + 1.701s3 + 22.79s2 + 20.82 s + 0.0005
expected values (EVs): 400, 400, and 600 MW, respectively. (15)
Three damping controllers LDC1, LDC2, and LDC3, de- HLDC2 (s)
signed by the proposed strategy in Section IV, are locally in-
stalled with AL1–AL3, respectively. The voltage-phase angle 0.039s4 + 5.659s3 + 64.82s2 + 14.57 s + 0.005
= 0.56
δ51 of Bus-51 is selected as the feedback signal to LDC1 while s4 + 3.084s3 + 6.574s2 + 0.032 s
the voltage-phase angle δ17 of Bus-17 is the effective feed- (16)
back signal for LDC2 and LDC3, based on the residue analysis HLDC3 (s)
at the nominal operating point [28]. Thus, it is noted that the
implemented damping control has employed a typical quasi- 3.726s4 +354.5s3 − 1826s2 + 4189 s − 795.8
= 0.52 (17)
decentralized control architecture which is characterized by uti- s4 + 3.294s3 + 75.81s2 + 5.981 s + 2.402
lizing remote signals to enhance the effectiveness of controllers As a comparison to the proposed design, the case where the
installed locally with controlled devices, such as VDLs [32]. generators G1-G12 are equipped with local PSSs [29] for the
The time delays of the feedback loops are commonly assumed damping control is simulated. Moreover, the local PSSs which
to be 150 ms [33]. Additionally, the three weighted filters in use the generators’ rotor speeds as the feedbacks signals are also
Fig. 6 are given by optimized based on a modal decomposition method proposed
30 10s in [34] where detailed information about this designed method
W1 (s) = W2 (s) = 0.5 W3 (s) = can be found. At the nominal operating point, eigenvalues of the
s + 30 s + 90
(14) closed-loop power system are computed and shown in Table II
The parameters (i.e., time constants) of these filters are se- (operating point 1). With the installation of the local PSSs, M1
lected according to their cut-off frequencies which should be and M3 are slightly shifted left and still poorly damped while the
decided by considering the frequencies of critical modes of damping ratio of M4 is only 0.077 (less than the required value
interest [28]. For example, frequencies of inter-area power os- 0.15). However, the robust LDCs can make the damping ratios
cillations are generally (not strictly) lower than 1.5 Hz (e.g., the of all four inter-area modes sufficient (more than 0.15), which
four concerned inter-area modes in the test system). According indicates the proposed design for VDLs is highly effective at
to the loop shape (Fig. 6), the cut-off frequencies of the low- enhancing damping ratios of inter-area modes.
pass and high-pass filters (W1 (s) and W3 (s)) should be larger Time domain simulations are conducted to further demon-
than 1.5 Hz. Therefore, as these filters are already selected to strate the effectiveness of the proposed control strategy. An
be first-order filters in order to reduce the computational com- instantaneous three-phase short-circuit fault occurs at Bus 60 at
plexity, their parameters are then readily calculated according 1.0 s that self-clears 100 ms later. In particular, the main pur-
to the cut-off frequencies. Moreover, the weights ω∞ and ω2 posed of using this fault is to drive the system states deviating

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 ZHANG et al.: INVESTIGATIONS OF LARGE-SCALE VOLTAGE-DEPENDENT LOADS FOR DAMPING INTER-AREA OSCILLATIONS 6045

Fig. 12. Power angle oscillations (solid line: proposed robust LDCs; dotted Fig. 14. Power angle oscillations at different operating points (solid line:
line: local PSSs; dashed line: no controller). proposed robust LDCs; dotted line: local PSSs; dashed line: no controller).

boundaries, with the exception of the short fault period (100 ms).
Compared to the case without any damping controller, the active
power of AL1 to AL3 becomes more fluctuant due to the LDCs.
Therefore, the controlled AELs must sacrifice their transient dy-
namic performance to meet the demands of the damping efforts.
However, from a long-term viewpoint, quickly damped oscil-
lations and stable operation of the system are the tradeoff for
such sacrifices because the local PSS-controlled system (or the
open-loop system) has unacceptable (even unstable) dynamics.

C. Verification of Robustness
To verify the robustness of the LDCs, five representative op-
Fig. 13. Dynamics of active power consumed by AL1 to AL3 (solid line: erating points in terms of different wind generations and tie-line
proposed robust LDCs; dotted line: no controller).
outages are simulated in this paper, as shown in Table I.
Eigenvalue analysis of the closed-loop power system under
the five operating points is conducted, with the damping ratios
from the equilibrium point so that the post-fault system dynam- of the inter-area modes shown in Table II. At all of the operating
ics can be utilized for the verification of control effects [29]. points, the damping performance of the robust LDCs is better
The relative power angles of different generators can be used to than that of the local PSSs. Although damping ratios of some
clearly observe the dynamics of the four inter-area modes, as inter-area modes slightly decrease as the operating point devi-
shown in Fig. 12. In the case with the local PSSs, the oscillations ates from the nominal one, they are still greater than 0.1 when
of M1 and M2 barely subside and the oscillations of M3 and M4 the LDCs are installed. Nonlinear time domain simulations are
are slightly suppressed but do not disappear within 10 s. In the also conducted, with the most inferior power angle dynamics of
case with the proposed LDCs, however, the oscillations of four the inter-area modes at each operating point indicated in Fig. 14.
inter-area modes, especially M1 and M2, are all satisfactorily Obviously, even under the emergent operating conditions (i.e.,
damped within 10 s. operating points 2-5), the LDCs outperform the local PSSs to
Moreover, as mentioned in Section III-B, the control efforts provide quite satisfactory damping effects on the inter-area os-
of the robust LDCs should be evaluated to prevent excessive cillations. In particular, due to the large energy of the disturbance
damping modulation and maintain safe operation of the con- (without reclosing fault tie-line 40–41) at operating point 5, the
trolled AELs. The dynamics of the active power consumed by output of LDC1 is saturated and the active power of AL1 si-
AL1 to AL3 are plotted in Fig. 13, where the dashed horizontal multaneously reaches its floor (lower boundary), as shown in
lines denote their upper and lower boundaries specified for the Fig. 15. Therefore, the above linear (eigenvalue) analysis and
power modulation. At the nominal operating point, the active nonlinear simulation results are solid evidence of the excellent
power curves of AL1 to AL3 almost consistently stay within the robustness of the proposed LDCs.

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 6046 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 33, NO. 6, NOVEMBER 2018

Fig. 15. Power output of AL1 and output of LDC1 at operating point 5.

Fig. 16. Power angle oscillations dominated by M1/M3 under different

TABLE III time delays.
EFFECTS OF DIFFERENT TIME DELAYS ON THE FOUR INTER-AREA MODES

Fig. 17. The oscillation of active power in the key tie-line 53–54 with time-
D. Effects of Time Delays varying delays.
To investigate the impacts of time delays on control effects
of inter-area oscillation modes in power systems, simulations
with different time delays are performed. As pointed out by 225 ms to 300 ms the power oscillation will quiet down with
[35] and [36], the communicating time delay is generally less relatively slow speed. However, as the time delay decreases
than 300 ms for the wide-area control in practical power systems to 150 ms, the power oscillation can be quickly damped. Par-
(e.g., China Southern Power Grid). Accordingly, five time de- ticularly, missing communication data is also a common yet
lays are selected when the test system is at the nominal operating inevitable issue which may obviously deteriorate the control
point, as shown in Table III. It can be found that the proposed effects. Thus design of damping controllers which are robust
LDCs show favorable robustness and have sufficient damping against the feedback data missing is an important direction to
control performance subjected to different time delays. Mean- improve the studies of this paper in the future.
while, because the time delay of 150 ms is considered during
the design of LDCs (Subsection V-B), the damping ratios of the
four inter-area modes are overall better than those in the other VI. CONCLUSION
cases of time delays. Besides, the impacts of time delay on these This paper mathematically investigates the mechanism of the
modes are also dependent on their frequencies. As for M1, due influence of VDLs on inter-area oscillations in power systems by
to its lower frequency, the time delay can result in less phase lag complex torque analysis. It concludes that, under stressed sys-
of the wide-area feedback signal and consequently less adverse tem operating conditions, the VDL is more likely to deteriorate
impacts. In contrast, the frequency of M3 is higher so that the inter-area modes. To address this issue, a novel damping con-
time delay can cause more phase lag which obviously brings trol architecture based on typical large-scale VDLs, i.e., AELs,
more impacts on the control results of this mode. Furthermore, is proposed to improve the damping of inter-area oscillations.
the nonlinear time domain simulations also verify the above Specifically, such architecture uses the decentralized wide-area
analysis, as shown in Fig. 16. Compared to the small distinc- LDCs to supplementarily control the AELs. Moreover, a multi-
tions among dynamic curves of δ15 − δ13 which are dominated stage robust design method is applied to synthesize the LDCs.
by M1, the larger time delay induces more apparent fluctuant Simulation results conducted on the modified New England and
dynamics of δ9 − δ13 because the damping ratio of M3 is more New York interconnected system validate the correctness of the
obviously deteriorated. torque analysis and verify the effectiveness and robustness of
The robustness of proposed LDCs against time-varying de- the proposed load damping control strategy. Hence, the pro-
lays is also simulated. The dynamic of active power carried by posed damping control for large-scale VDLs can be regarded as
the key tie-line 53–54 is typically selected and illustrated in a potential solution to the challenging issue of poorly-damped
Fig. 17. It is seen that when the time delay is changed from inter-area oscillations in modern power systems.

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 ZHANG et al.: INVESTIGATIONS OF LARGE-SCALE VOLTAGE-DEPENDENT LOADS FOR DAMPING INTER-AREA OSCILLATIONS 6047

APPENDIX [9] I. Kamwa, R. Grondin, and D. Asber, “Large-scale active-load modulation

for angle stability improvement,” IEEE Trans. Power Syst., vol. 14, no. 2,
A. Definitions of K1 − K6 Coefficients pp. 582–590, May 1992.
[10] B. Ramanathan and V. Vittal, “Small-disturbance angle stability enhance-
According to the linearization process shown in [18]–[20], ment through direct load control part I—Framework development,” IEEE
the SMIB model in Fig. 1 can be linearized based on following Trans. Power Syst., vol. 21, no. 2, pp. 773–781, May 2006.
[11] J. Xu, S. Liao, and Y. Sun, “An isolated industrial power system driven
coefficients: by wind-coal power for aluminum productions: A case study of frequency
X q −X d V cos δ 0 control,” IEEE Trans. Power Syst., vol. 30, no. 1, pp. 471–483, Jan. 2015.
K1 = X d +X t iq 0 V sin δ0 + X q +X t Eq 0 [12] M. Beiraghi and A. M. Ranjbar, “Additive model decision tree-based
X q +X t 
X d +X t adaptive wide-area damping controller design,” IEEE Syst. J., vol. 12, no.
K2 = X d +X t iq 0 K3 = X d +X t 1, pp. 328–339, Mar. 2018.
[13] T. Wang, Z. Wang, J. Liu, J. S. Thorp, and Y. Yang, “Classification and
X d −X d
K4 = X d +X t V sin δ0 (18) regression tree-based adaptive damping control of inter-area oscillations
  using wide-area signals,” IET Gen. Transmiss. Distrib., vol. 8, no. 9,
Xq X d
K5 = V
Vt 0 X q +X t Vtd0 cos δ0 − X d +X t Vtq 0 sin δ0 pp. 4297–4309, Nov. 2016.
[14] D. Ke and C. Y. Chung, “Design of probabilistically-robust wide-area
Vt q 0 Xt power system stabilizers to suppress inter-area oscillations of wind in-
K6 = V t 0 X d +X t tegrated power systems,” IEEE Trans. Power Syst., vol. 31, no. 6,
pp. 471–483, Jan. 2015.
where Xd and Xq are the d-axis and q-axis synchronizing re- [15] X. M. Mao, Y. Zhang, and X. C. Wu, “Coordinated control of interarea
actances, respectively, of the generator and Xd is the transient oscillation in the China Southern power grid,” IEEE Trans. Power Syst.,
vol. 21, no. 2, pp. 845–852, May 2006.
reactance; Xt is the external reactance; δ is the rotor angle of [16] S. Elenius, K. Uhlen, and E. Lakervi, “Effects of controlled shunt and
the generator; iq is the q-axis component of stator current; Vtd series compensation on damping in the Nordel System,” IEEE Trans.
and Vtq are the d-axis and q-axis components, respectively, of Power Syst., vol. 20, no. 4, pp. 1946–1957, Nov. 2005.
[17] J. Fang, W. Yao, and Z. Chen, “Design of anti-windup compensator for en-
the terminal voltage Vt . The subscript ‘0’ means that the associ- ergy storage-based damping controller to enhance power system stability,”
ated variables are calculated as the system is at the steady state. IEEE Trans. Power Syst., vol. 29, no. 3, pp. 1175–1185, May 2014.
It can be found that the coefficient K5 can be either positive [18] P. Kundur, Power System Stability and Control. New York, NY, USA:
McGraw-Hill, 1994.
or negative, which is related to the operating condition of the [19] A. Kumar, “Power system stabilizers design for multimachine power sys-
system. For instance, when the system load is heavy, the rotor tems using local measurements,” IEEE Trans. Power Syst., vol. 31, no. 3,
angle δ will be large and consequently lead to negative K5 . pp. 2163–2171, May 2016.
[20] F. P. deMello and C. Concordia, “Concepts of synchronous machine sta-
bility as affected by excitation control,” IEEE Trans. Power App. Syst.,
B. Parameters of the Wind Turbine Generator vol. PAS-88, no. 4, pp. 316–329, Apr. 1969.
[21] P. Pourbeik and M. J. Gibbard, “Damping and synchronizing torques
This paper employs the DFIG-based wind turbine with the induced on generators by FACTS stabilizers in multimachine power sys-
structure given in [37] and the parameters on base of machine tems,” IEEE Trans. Power Syst., vol. 11, no. 4, pp. 1920–1925, Nov. 1996.
[22] M. Tian, “Design and simulation of a voltage-control system with a con-
ratings 2.24 MVA and 690 V as follows: cut-in wind speed trollable reactor of transformer type,” in Proc. IEEE Conf. Ind. Technol.
ωc -in = 4 m/s; rated wind speed ωr ate = 13 m/s; cut-out wind ICIT, Hong Kong, 2005, pp. 419–423.
speed ωc -ou t = 25 m/s; wind turbine radius R = 46.7 m; stator [23] J. Yuan, J. Zhou, and B. Chao, “A novel compact high-voltage test system
based on a magnetically controlled resonant transformer,” IEEE Trans.
winding resistance Rs = 0.0116 p.u.; stator inductance Ls = Magn., vol. 51, no. 11, Nov. 2015, Art. no. 1700204.
0.0229 p.u.; inertia time constant of wind turbine Ht = 3.9 s; [24] L. Liang, Y. Hou, and D. J. Hill, “Enhancing flexibility of an islanded
damping coefficient Dtw = 0.02 p.u. microgrid with electric springs,” IEEE Trans. Smart Grid, to be published.
[25] C. Scherer, P. Gained, and M. Chilali, “Multiobjective output-feedback
control via LMI optimization,” IEEE Trans. Autom. Control, vol. 42, no. 7,
REFERENCES pp. 896–911, Jul. 1997.
[26] Y. Zhang and Anjan Bose, “Design of wide-Area damping controllers for
[1] M. Klein, G. J. Rogers, and P. Kundur, “A fundamental study of inter-area interarea oscillations,” IEEE Trans. Power Syst., vol. 23, no. 3, pp. 1136–
oscillations in power systems,” IEEE Trans. Power Syst., vol. 6, no. 3, 1143, Jul. 2008.
pp. 914–921, Aug. 1991. [27] B. Chaudhuri, B. C. Pal, A. C. Zolotas, “Mixed-sensitivity approach to H ∞
[2] J. V. Milanovic and I. A. Hiskens, “Oscillatory interaction between syn- control of power system oscillations employing multiple FACTS devices,”
chronous generator and local voltage-dependent load,” IEE Proc. Gener- IEEE Trans. Power Syst., vol. 18, no. 3, pp. 1149–1156, Aug. 2003.
ation Transmiss. Distrib, vol. 142, no. 5, pp. 473–480, Sep. 1995. [28] B. Pal and B. Chaudhuri, Robust Control in Power Systems. New York,
[3] Z. Akhtar, B. Chaudhuri, and S. Y. R. Hui, “Primary frequency control NY, USA: Springer, 2005.
contribution from smart loads using reactive compensation,” IEEE Trans. [29] C. Canizares, T. Fernandes, and E. Geraldi, “Benchmark models for the
Smart Grid, vol. 6, no. 5, pp. 2356–2365, Sep. 2015. analysis and control of small-signal oscillatory dynamics in power sys-
[4] T. Cui, W. Lin, and Y. Sun, “Excitation voltage control for emergency fre- tems,” IEEE Trans. Power Syst., vol. 32, no. 1, pp. 715–722, Jan. 2017.
quency regulation of island power systems with voltage-dependent loads,” [30] E. Muljadi, C. P. Butterfield, and A. Ellis, “Equivalencing the collector
IEEE Trans. Power Syst., vol. 31, no. 2, pp. 1204–1217, Mar. 2016. system of a large wind power plant,” in Proc. IEEE Power Eng. Soc. Gen.
[5] H. Jiang, J. Lin, and Y. H. Song, “Demand side frequency controls scheme Meeting, Montreal, Canada, 2006, pp. 1–8.
in an isolated wind power system for industrial aluminum smelting pro- [31] W. Du, J. Bi, J. Cao, and H. F. Wang, “A method to examine the impact
duction,” IEEE Trans. Power Syst., vol. 29, no. 2, pp. 844–853, Mar. 2014. of grid connection of the DFIGs on power system electromechanical os-
[6] J. Liu, W. Zhang, and Y. Liu, “Primary frequency response from the control cillation modes,” IEEE Trans. Power Syst., vol. 31, no. 5, pp. 3775–3784,
of LED lighting loads in commercial buildings,” IEEE Trans. Smart Grid, Sep. 2016.
vol. 8, no. 6, pp. 2880–2889, Nov. 2017. [32] D. Dotta, A. S. Silva, and I. C. Decker, “Wide-area measurements-
[7] O. Samuelsson and B. Eliasson, “Damping of electro-mechanical oscilla- based two-level control design considering signal transmission de-
tions in multimachine system by direct load control,” IEEE Trans. Power lay,” IEEE Trans. Power Syst., vol. 24, no. 1, pp. 208–216, Feb.
Syst., vol. 12, no. 4, pp. 1604–1609, Nov. 1997. 2009.
[8] I. Kamwa, R. Grondin, and D. Asber, “Active-power stabilizers for mul- [33] M. Beiraghi and A. M. Ranjbar, “Adaptive delay compensator for the
timachine power systems challenges and prospects,” IEEE Trans. Power robust wide-area damping controller design,” IEEE Trans. Power Syst.,
Syst., vol. 13, no. 4, pp. 1352–1358, Nov. 1998. vol. 31, no. 6, pp. 4966–4976, Nov. 2016.

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.
 6048 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 33, NO. 6, NOVEMBER 2018

[34] J. Zhang, C. Y. Chung, and Y. Han, “A novel modal decomposition control Yuanzhang Sun (M’99–SM’01) received the B.S.
and its application to PSS design for damping interarea oscillations in degree in hydro and electrical engineering from
power systems,” IEEE Trans. Power Syst., vol. 27, no. 4, pp. 2015–2025, Wuhan University, Wuhan, China, in 1987, the M.S.
Nov. 2012. degree from the Electric Power Research Institute,
[35] C. Lu, X. Zhang, X. Wang, and Y. Han, “Mathematical expectation mod- Beijing, China, in 1982, and the Ph.D. degree in elec-
eling of wide-area controlled power systems with stochastic time delay,” trical engineering from Tsinghua University, Beijing,
IEEE Trans. Smart Grid, vol. 6, no. 3, pp. 1511–1519, May 2015. China, in 1988.
[36] K. Zhu, M. Chemome, L. Nordstrom, S. Holmstrom, and G. Ericsson, He is a currently a Professor with the School
“Design requirements of wide-area damping systems—Using empirical of Electrical Engineering, Wuhan University, and a
data from a utility IP network,” IEEE Trans. Smart Grid, vol. 5, no. 2, Chair Professor with the Department of Electrical
pp. 829–838, May 2014. Engineering and a Vice Director with the State Key
[37] C. Zhang, D. Ke, Y. Sun, C. Y. Chung, J. Xu, and F. Shen, “Coordinated Lab of Power System Control and Simulation, Tsinghua University, Beijing,
supplementary damping control of DFIG and PSS to suppress inter-area China. His main research interests are in the areas of power system dynamics
oscillations with optimally controlled plant dynamics,” IEEE Trans. Sus- and control, wind power, voltage stability and control, and reliability.
tain. Energy, vol. 9, no. 2, pp. 780–791, Apr. 2018.

C. Y. Chung (M’01–SM’07–F’16) received the

Ph.D. degree in electrical engineering from The Hong
Kong Polytechnic University, Hung Hom, Hong
Chen Zhang (S’16) received the B.S. degree in elec-
Kong, in 1999. He is currently a Professor and
trical engineering from Wuhan University, Wuhan,
the SaskPower Chair in Power Systems Engineer-
China, in 2014. He is currently working toward the
Ph.D. degree at the School of Electrical Engineer- ing with the Department of Electrical and Computer
Engineering, University of Saskatchewan, Saskatoon,
ing, Wuhan University. At present, he is also funded
SK, Canada.
by Chinese Scholarship Council as a joint Ph.D. de-
His research interests include power system sta-
gree at the Department of Electrical Engineering and
Computer Science, Case Western Reserve University, bility/control, smart grid technologies, renewable en-
ergy, computational intelligence applications, power
Cleveland, Ohio, USA.
markets, and electric vehicle charging.
His research interests are in power system stability
Dr. Chung is an Editor of the IEEE TRANSACTIONS ON POWER SYSTEMS and
and control, wind power, and direct load control.
IEEE TRANSACTIONS ON SUSTAINABLE ENERGY. He is also a member-at-Large
(Smart Grid) of IEEE PES Governing Board and the IEEE PES Region 10 North
Chapter Representative.

Deping Ke received the B.S. degree in electrical en- Jian Xu (M’08–SM’18) received the Ph.D. degree
gineering from the Huazhong University of Science in electrical engineering from the Wuhan University,
and Technology, Wuhan, China, in 2005 and the Ph.D. Wuhan, China, in 2007. He joined Wuhan Univer-
degree in electrical engineering from The Hong Kong sity as a Lecturer in 2007. Since 2015, he has been
Polytechnic University, Hung Hom, Hong Kong, in a Professor with the School of Electrical Engineer-
2012. He is currently an Associate Professor with the ing, Wuhan University. During 2013–2014, he was a
School of Electrical Engineering, Wuhan University, Visiting Scholar with the Energy Systems Innovation
Wuhan, China. Center, Washington State University, Pullman, WA,
His research interests are in power system dynam- USA.
ics and control, and economic operation of power His research interests include power system sta-
systems. bility and control, wind power, and voltage stability.

Authorized licensed use limited to: Indian Institute Of Technology (Banaras Hindu University) Varanasi. Downloaded on July 24,2022 at 13:04:06 UTC from IEEE Xplore. Restrictions apply.