You are on page 1of 9

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
1

Integration of Renewables into the Wide Area


Control Scheme for Damping Power Oscillations
Lazaros Zacharia, Student Member, IEEE, Lenos Hadjidemetriou, Member, IEEE,
and Elias Kyriakides, Senior Member, IEEE 1
Abstract—This paper presents a Wide Area Control (WAC) generators’ local controllers. These control signals intend to
method to coordinate the local controllers of generators and overcome the shortcomings of the local controllers and more
Renewable Energy Sources (RESs), for robust power oscillation specifically their lack of global observability. The main
damping. The WAC signals are synthesized by a two-level
benefit of having WAC signals in the power system is the
hierarchical controller, which utilizes wide area measurements
from all the installed synchronous generators and RESs for effective and simultaneous compensation of the system’s local
maximizing the performance of their corresponding local and inter-area oscillations. The latter are highly undesirable
controllers. In the resulting WAC scheme, the generators are and they can only be detected through the utilization of wide
“aware” of the RES oscillations and vice versa, the RESs are area measurements. Inter-area oscillations are found in large
“aware” of the generator oscillations. Further, the WAC signals power systems, where groups of generators are connected via
intended for the RESs are adjusted in order to require only slight
long distance transmission lines [3]. These oscillations are
modifications on the RES local controllers. This study illustrates
that the correct coordination of generators and RESs, through characterized by low frequency (0.2-1 Hz) and occur when
the WAC, improves the damping performance of the entire generators of one group swing against generators of another
system even under high penetration of renewables. The group [4], [5].
performance of the proposed scheme has been tested and Several design methods have been proposed for the
validated in the IEEE 9-bus and 39-bus test systems. development of a wide area controller. These can be separated
mainly into methodologies which are based on a model of the
Index Terms— Local controller, power oscillations, Renewable
system or on predefined typical operating conditions. More
Energy, Wide Area Control, wide area measurements.
specifically, in [6] a wide area controller has been
I. INTRODUCTION implemented according to an adaptive measurement-based
identification of the model. Furthermore, [7] proposes the
utilization of a network control system model for the WAC
T HE increasing energy demand and the necessity to meet
the consumer’s needs while satisfying the environmental
concerns, led to the introduction of the smart grid concept
design, according to linear matrix inequality techniques. The
authors in [8] use an expectation model to derive the WAC
[1]. This concept has recently emerged to propose the signals, which considers time delay distribution. In [9], the
development of an intelligent power system, which will WAC development is based on the Takagi-Sugeno fuzzy
support the increasing penetration of Renewable Energy approach which takes into account various operating
Sources (RESs) in the system, alongside with its improved conditions of the system. Finally, a control design method,
reliability and stability. The first step towards the realization which does not consider any simplified model or typical
of the smart grid, is to be able to observe the dynamic operating conditions of the system, is proposed in [10]. It
behavior of the system and to provide the correct control suggests the implementation of a hierarchical WAC, based on
signals, so that the system will be able to operate safely even wide area signals from all the generators and the reformulation
when close to its limits [2]. The deployment of Phasor of the power system into a suitable, closed form.
Measurement Units (PMUs) in the transmission level and their The increasing penetration of renewables in the power
capability in providing synchronized, quasi real time system may deteriorate the stability of the entire grid mainly
measurements from remote locations, has made this goal due to the reduction of the system’s inertia (e.g., displacement
achievable. The exploitation of the wide area measurements, of synchronous generators) and the change in the system’s
provided by the PMUs, for monitoring and controlling the
dynamic characteristics, as illustrated in [11]. The majority of
power system led to the introduction of the Wide Area
WAC design methodologies do not take into account the effect
Monitoring and Control (WAMC) systems [2].
of the installed RESs on the performance of the wide area
Wide Area Control (WAC) aims to utilize the available controller. However, the consideration of the effect of RESs
wide area measurements for the development of suitable while utilizing WAC or their integration in the WAC design,
control signals in order to enhance the performance of the is vital for the implementation of the smart grid, since the
smart grid of the future is expected to have a high penetration
This work has been supported by the European Union's Horizon 2020 of renewables. The current practice by the system operator
research and innovation programme under grant agreement No 739551 (KIOS
CoE) and from the Government of the Republic of Cyprus through the (especially at low penetration levels), is to utilize the RESs as
Directorate General for European Programmes, Coordination and negative loads and not as components able to contribute to the
Development. overall stability [12]. Therefore, the WAC concept provides
The authors are with the Department of Electrical and Computer
Engineering and the KIOS Research and Innovation Center of Excellence,
the opportunity and capability of including the renewables
University of Cyprus, Nicosia, Cyprus (email: zacharia.lazaros@ucy.ac.cy; (along with the synchronous generators) in a coordination
hadjidemetriou.lenos@ucy.ac.cy; elias@ucy.ac.cy).

0885-8950 (c) 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
2

11 validate the performance of the proposed control scheme in


Exciter Exciter compensating local and inter-area oscillations. The

PMU
RES
efd2 egfd2 egfd3 efd3 performance of the proposed WAC scheme has been evaluated
2 7 8 9 3
Vt2 Vt3 through the use of Prony analysis method. The main

PMU
G G

PMU
Pm2 Pm3 contributions of this paper are the quantification of the
10
ω2 PgGV2 12 P GV3 ω3
g
interactions between generators and RESs and the development
RES

PMU
Valve Valve
RES of coordination signals to explicitly compensate them.
PMU
Governor 5 Governor
6 The rest of the paper is structured as follows. Section II
overviews the components needed for the WAC design. A
vd, vq, δ, f, from each detailed description of the RESs which are used in this study is
generator 4
WAC provided in Section III. Section IV shows the design
vd-res, vq-res, δres, f, from PMU
methodology for the formulation of the WAC signals. The
each RES 1 evaluation of the proposed scheme is presented and discussed
egfd , PgGV Vt1 PgGV1
e g Pm1 Governor in Section V. Section VI concludes the paper.
fd1
Exciter

G
dP, dQ Valve
efd1 ω1 II. MODELS NECESSARY FOR THE WAC DESIGN

Fig. 1. Proposed integration of WAC and RES A. Phasor Measurement Units


As it was mentioned in the previous section, the WAC
scheme in order to damp effectively all the power oscillations. formulation requires wide area measurements. It is a common
In [13], a hierarchical control scheme is proposed, which practice to use PMUs for WAMC applications, to measure
consists of centralized and local Power Oscillation Dampers voltage phasors, current phasors and frequency in a
(PODs) and Power System Stabilizers (PSSs) in order to synchronous way. Thus, PMUs are utilized in this study in
coordinate all the wind farms and generators of the system. order to provide the necessary signals to the WAC scheme.
Furthermore, [14] presents a WAC design, based on particle For this purpose, it is assumed that PMUs are installed on all
swarm optimization, for improving the performance of the generator and RES buses (Fig. 1). Furthermore, the
power system through the control of wind farms. In [15], an measurement signals are constructed in such a way as to
observer-driven reduced copy is utilized along with a linear emulate signals provided by PMUs. This means that their
quadratic regulator to damp inter-area oscillations, when two sampling rate and their measurement error are based on
generators are replaced by wind farms. All the aforementioned realistic PMU measurements as discussed in [17].
methods consider only Doubly-Fed Induction Generator
(DFIG) wind farms. A work which utilizes an average model B. Synchronous Generator
of the wind farms’ voltage-source converter for control The model of the generator is vital for the WAC
(instead of a DFIG model) is presented in [16], where the implementation, since the wide area controller is essentially
coordination of multiple wind farms with other power system controlling the generators through their local controllers.
devices is illustrated. Amongst the various models of the generators, the fourth
In this paper, the implementation of a two-level order model, expressed in the local d-q frame, is found
hierarchical controller is proposed as shown in Fig. 1, which sufficient for WAC design [18]. For this study, the selected
integrates both synchronous generators and RESs in its design. fourth order model depends only on internal voltages (ed and
The proposed WAC scheme is developed by advancing the eq), stator currents (id and iq) and the field voltage (efd), as
methodology presented in [10], in order to derive WAC shown in (1)- (2). Furthermore, the swing equation of the ith
signals not only for the generators but for the RESs as well. generator (as presented in (3)-(4)), is required as well:
This method has the advantage that by reformulating the '
Tdoi eqi' = − eqi' − ( xdi − xdi' )idi + e fdi (1)
power system into a suitable and closed form, it leads to the
Tqoi edi =
' '
− edi + ( xqi − xqi )iqi
' '
(2)
implementation of a reliable and effective wide area
controller. In this WAC scheme, wide area measurements 2 H iω pui d 2δ i Diω pui d δ i
+ =
Pmi − Pei (3)
from all the generators and RESs can be utilized to develop ωs dt 2 ωs dt
control signals for enhancing the performance of the = Pei vdi idi + vqi iqi (4)
corresponding local controllers. The investigation considers
the effect of the high penetration of large scale fully converted T’do and T’qo stand for the open circuit constants, while xd, xq,
renewables, the local controller of which has been slightly x’d and x’q are the generator’s synchronous and transient
modified in order to allow its coordination by the wide area reactances, all expressed on the local d-q frame. Pm and Pe
controller during disturbances. It is to be noted that in this represent the mechanical and electrical power. ω is the rotor
paper, a detailed model of the power system is combined with a speed, while ωs is the synchronous speed. Lastly, H, D and δ
very accurate model of the RES local controller for more represent the generator’s inertia constant, its damping
realistic operation. Electro-Magnetic Transient (EMT) coefficient and power angle, respectively. All the quantities
simulations on the IEEE 9-bus and 39-bus dynamic test are in per unit, except the time constants which are in seconds
systems with integrated large-scale RESs have been used to and the power angle in radians.

0885-8950 (c) 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
3

In addition, it is important to have equations which connect Generator Three-Phase Inverter (GSC)
the terminal voltage and stator currents of the generator to its Converter
+ LC Filter PCC Power
Grid
AC Lgrid
internal voltages, since the PMU can provide measurements eres r x

DC-
link
a b
CD c
concerning only the former two. Equations which contain the Cf vgrid
C
DC - vres
terminal voltage of the generator expressed on the d-q axis (vd Wind Turbine VDC ires
Sampling Pulse Width Modulation Sampling
and vq), are the following stator dynamic equations: GSC vref
vdi =edi' − rsi idi + xqi' iqi (5) Controller
iL v, θPLL
Synchronization Method (PLL)
v θPLL |Vgrid|
vqi =eqi' − rsi iqi − xdi' idi (6) Current Controller PQ Controller (with FRT)
iref
where rs stands for the stator resistance. VDC* Vdc Pref P* Q* Qref Q
Pset-point controller ++ ++ profile
C. Local Controllers for the Generators dP dQ
From WAC scheme
Generators utilize local controllers (i.e., exciter and
Fig. 2. A fully converted RES with the generator converter, the GSC and
governor controllers) to compensate local disturbances, but its controller, and the slight modifications for considering WAC signals.
when inter-area oscillations occur, their lack of global
observability makes them unable to react effectively. WAC electronics technology as shown in Fig. 2. The generator side
aims to enhance their operation by providing suitable converter is responsible for the optimal power extraction from
coordination signals to their local control inputs. Thus, it is the wind turbine according to the wind speed, while the GSC
important to know the types of the local controllers used in the properly injects the produced power into the grid.
system, in order to identify correctly their local control inputs. The GSC controller developed for the purpose of this work
This study considers the exciter DC2A and the general (Fig. 2) is mainly based on: a robust synchronization method
purpose governor [19], [20]. Their respective local control [21] for estimating the phase angle (θPLL) of the grid voltage
inputs, which the wide area controller will interact with, are vres=[va vb vc]T at the Point of Common Coupling (PCC), an
the excitation signal (efd) and the steam valve output (PGV). accurate current controller [22] for regulating the injected
Furthermore, the following turbine model is required as well currents ires=[ia ib ic]T, and a PQ controller [23] for generating
to implement the WAC signals, since it connects the the reference currents iref=[ia* ib* ic*]T. The PQ controller needs
governor’s PGV and the generator’s mechanical power (Pm). to be enhanced with Fault Ride Through (FRT) capabilities to
1 1 + sTRH FHP ensure that the renewable source will meet the modern grid
Pm = PGV (7) regulations for interconnecting RES [24]. In the usual case, the
1 + sTCH 1 + sTRH
reference active and reactive power (Pref and Qref), which are
where TCH represents the delay due to the steam chest/inlet the inputs to the PQ controller, are generated by the Vdc
piping, TRH stands for the delay of the re-heaters and FHP is the controller and the Q profile scheme. The Vdc controller
total turbine power developed in the high-pressure cylinder. generates the Pref in order to inject the produced power by the
The delays are expressed in seconds and the power in per unit. RES into the grid by maintaining a constant DC link voltage
D. Power System Network (equal to VDC*). The Q profile determines the required Qref that
The interconnection of the transmission system can be needs to be injected by the RES according to the amplitude of
represented by two current equations, depending on whether the grid voltage at the PCC as defined by the grid regulations.
the system’s nodes are generator or generator-free buses ((8) It should be noted that the only requirement for integrating
and (9) respectively). Index m refers to the total number of the renewables into the proposed WAC scheme is to slightly
system’s buses. G’, B’ are the bus admittance matrix elements. adjust the GSC controller as shown in Fig. 2. The GSC needs
to be able to receive the remote signals dP and dQ in order to
−1 ' j (δ −π )
'
(edi + ( xqi' − xdi' )iqi + jeqi' ) e i 2 adjust the injected active and reactive power by the RES
jxdi according to the proposed WAC scheme. The signals dP and
m
(8)
+ ∑ (Gik + jBik )vk e =
' ' jθ k dQ are derived by the wide area controller, as will be
0
k =1
described in Section IV.B. These signals aim to decouple the
m oscillations caused by other generators and RESs and as a
∑ (G
k =1
'
ik + jBik' )vk e jθk =
0 (9) result to properly damp local and inter-area oscillations under
an intense power disturbance.

III. LARGE SCALE FULLY CONVERTED RES IV. WIDE AREA CONTROL DESIGN
The main contribution of this paper is the integration of the The methodology proposed in [10] for the formulation of
installed RESs into the design of the WAC scheme. For WAC signals is briefly presented in this section, along with
achieving this purpose, large scale fully converted RESs have the necessary modifications in order to integrate the installed
been considered in the design of the WAC scheme due to their RESs into the WAC design.
increased flexibility regarding the power controllability.
A fully converted RES consists of a wind turbine with any A. Conventional Design of Wide Area Control signals
type of a three-phase electric generator, a generator side The methodology presented in [10] essentially reformulates
converter, and a Grid Side Converter (GSC) based on power the power system model into a closed form, by utilizing (1)-

0885-8950 (c) 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
4

(6) and (8)-(9). More specifically, a change of variables is Start


performed, having as an objective to obtain a state space
representation of the system, which will have as state variables Load the topology of
the terminal voltage of the generators (expressed on the d-q the power system
axis). The main goal of the methodology is to develop
Identify the n generator buses and the l
coordination signals which explicitly cancel out all the RES buses
interactions (perturbation terms) between the generators of the
system, increasing that way the damping performance of the Reformulate the power system in terms of vd, vq
local controllers. Furthermore, the measurement signals and vres_d, vres_q based on (8), (9) and (19)
required by this methodology are the generator’s terminal
voltage, its frequency and its power angle. The latter is t=0
estimated according to PMU measurements [25].
The following three steps summarize the method’s procedure: Y
t=Tfinal+tstep? End
1) The voltages of the generator-free buses are expressed in
egfdi(t), PgGVi(t)
terms of the generators’ terminal voltages (vd and vq). N
dPi(t), dQi(t)
2) The internal voltage (ed and eq) and stator current (id and iq) Bus i=0
of each generator are expressed in terms of its terminal voltage Develop Utilize (24)-(33) to
generator i WAC derive RES i WAC
(vd and vq). signals through t=t+tstep signals
3) The derived expressions of the internal voltage and stator (14) and (15)
i=n+l ?
current are replaced in the generator model and swing equation. Y Develop deᵍres_d &
N deᵍres_q through (22)
The result of applying these steps are the ith generator’s and (23)
dynamic equations for the terminal voltage and rotor speed: i=i+1

vdi =a1i vdi + a2i vqi + p1i e fdi +ψ di (10) Estimate generator i Estimate RES i
vqi =b1i vdi + b2i vqi + p2i e fdi +ψ qi (11) perturbation terms ψdi, Y Is bus i N perturbation terms
ψqi, and ψdωi generator ψres_d and ψres_q
2 H i ω pui d ωi Diω pui d δ i bus ?
+ =Pmi − Gii f (vdi2 + vqi2 ) −ψ ωi
ωs dt ωs dt (12) vd-gen(t), vq-gen(t), vd-gen(t), vq-gen(t),
δgen(t), fgen(t), vd-res(t), δgen(t), fgen(t), vd-res(t),
vq-res(t), δres(t), fres(t) vq-res(t), δres(t), fres(t)
where a1, a2, b1, b2, p1, p2 are parameters which mainly depend
on the network topology, resulting from the reformulation of Fig. 3. Flow diagram of proposed methodology
the power system into a closed form. ψd, ψq, ψω are the
coupling perturbation terms (or inter-generator interactions) −(ψ dvi d dvdi + ψ dvi q dvqi )
from other generators on the ith generator, which mainly e gfdi = (14)
depend on the terminal voltages of the former. Detailed p1i dvdi + p2i dvqi
expressions of all these terms are provided in [10]. TCHψ ωg + ψ ωg
Furthermore, the formulation of the wide area control
g
pGVi = − (15)
FHP
signals derives from the replacement of vd, vq and ω in the
dynamic equations by their error compared to their steady B. Integration of RESs in the WAC scheme
state values (dvd, dvq and deω). The main idea is to In the previous subsection, the methodology presented
differentiate the resulting dynamic equations until the local considers only the system’s generators to derive the WAC
control inputs (efd and PGV) appear in the equations. Then, signals. Therefore, all the necessary modifications of the
these control inputs are substituted by their local and global aforementioned methodology are discussed in detail here, in
signals (since the WAC interacts with them), where the global order to also include the renewable sources of the system into
signal expressions are chosen to explicitly cancel out all the the WAC design. More specifically, the formulation of WAC
perturbation terms of the dynamic equations. signals intended for the RES local controller is shown.
Looking at (10) and (11), it is evident that the exciter’s Furthermore, the ultimate goal is to illustrate that the correct
local control input appears in both dynamic equations. Thus, coordination of generators and RESs, through the WAC,
their combination is utilized to find a common global control improves the damping performance of the entire system
input, which cancels both perturbation terms (ψd and ψq). This instead of decreasing it, even under a high penetration of
is achieved through (13), which is essentially an expression renewables. The flowchart of Fig. 3 summarizes the overall
for the terminal voltage deviation (dvt). procedure of the proposed scheme for the development of
=
dvt
2
dvd2 + dvq2 (13) coordination signals for generators and RESs. Note that Tfinal
The following set of equations shows the resulting wide represents the total time of the simulation and tstep the
area control signals for the exciter and governor. An analytical reporting rate of the PMU (20 ms).
description of their development, along with full expressions The first step towards the integration of RESs into the
of all the reported parameters is presented in [10]. WAC scheme, is to obtain wide area measurements from the

0885-8950 (c) 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
5

renewables, similar to the ones collected from the generators. step of the proposed methodology.
Thus, PMUs are assumed to be connected at the PCC of each ires − di =
−Giif vres − di + Biif vres − qi
RES in order to provide their terminal voltage (vres) and n
frequency (Fig. 1). The problem that arises is that the power −∑ (Gikf vdk − Bikf vqk ) cos δ ki − (Gikf vqk + Bikf vdk ) sin δ ki
angle is also required when applying the methodology of [10]. k =1

However, in the case of RESs there is no power angle (in l

contrast to the generators). To overcome this obstacle and −∑ (Gijf vres − dj − Bijf vres − qj ) cos δ ji − (Gijf vres − qj + Bijf vres − dj ) sin δ ji
j =1
make the methodology applicable for renewables as well, a j ≠i
virtual power angle for RES (δres) is introduced that represents (20)
the angle between the terminal voltage vres and the virtual
ires − qi =
−Giif vres − qi − Biif vres − di
voltage eres. The latter voltage is the fundamental voltage n
phasor at the output of the GSC as shown in Fig. 2. It contains + ∑ (Gikf vdk − Bikf vqk ) sin δ ki + (Gikf vqk + Bikf vdk ) cos δ ki
high-frequency switching noise (this noise is eliminated by the k =1
l
LC filter) and cannot be measured directly. The computation + ∑ (Gijf vres − dj − Bijf vres − qj ) sin δ ji + (Gijf vres − qj + Bijf vres − dj ) cos δ ji
of δres is based on vres=[vres-d vres-q]T and the injected current j =1
phasor ires=[ires-d ires-q]T , expressed into the local dq-frame, j ≠i

according to (16)-(18). More specifically, (16) and (17) are (21)


used to calculate the corresponding components of the Note that respective expressions (which depend on terminal
eres=[eres-d eres-q]T and as a result the virtual δres can be obtained voltages of generators and renewables), will result when the
by (18). formulation of the generator WAC signals is considered.
eres − d = vres − d + rires − d − xires − q (16) The next step is to replace (20) and (21) into the model of
the RES. For this reason, the RES virtual model (which has
eres − q = vres − q + rires − q + xires − d (17) been defined in (16) and (17)) is utilized. Furthermore, all the
 eres − d  terminal voltages of generators and renewables existing in the
δ res = arctan   (18)
e resulting expressions are substituted by their error compared
 res − q  to their steady state values (corresponding dvd, dvq and dvres-d,
It is to be noted that r and x represent the series resistance dvres-q), deriving expressions of the deres-d and deres-q. This
and reactance of the LC filter between the inverter and the change of variables is required in order to cancel out the WAC
PCC, as shown in Fig. 2. The capacitor (Cf) of the LC filter contribution when the system is in steady state.
can be neglected since it damps only the high order harmonics. Similar to the generators’ case, in order to formulate the
By obtaining the appropriate wide area measurements from RES WAC laws, it is important to identify the suitable local
the RESs, the methodology of [10] can now be modified signals, which the wide area controller will utilize to interact
accordingly in order to become applicable for the renewables with the RES local controller. Based on the RES virtual model,
as well. Thus, the first step of the proposed procedure is that these are the eres-d and eres-q, presented in (16) and (17). Thus,
the generator-free and renewable-free buses are expressed in by decomposing deres-d and deres-q, into local and global control
terms of the generators and RESs terminal voltages (vd, vq and signals, the RES WAC signals are selected to explicitly cancel
vres-d, vres-q, respectively). This step is important in order to out the perturbation terms ψres_d and ψres_q. These terms are
make each generator “aware” of the oscillations caused by the equal to the superposition terms of (20) and (21), respectively,
rest of the generators and RESs and vice versa, each RES to be where dvd, dvq and dvres-d, dvres-q are considered. This leads to
“aware” of the oscillations caused by the rest RESs and the following expressions for the RES WAC signals:
generators. The result in the case of generators is to also g
deres −rψ res
_d =
i
_ d + xψ res _ q
i
(22)
include in the perturbation terms of (14) and (15) the terminal
voltages of the renewables. The rest of the steps remain as
g
deres −rψ res
_q =
i
_ q − xψ res _ d
i
(23)
they were described in the previous subsection, to derive the The result of the aforementioned analysis is that the wide
generators’ WAC signals, while for the RESs these steps are area controller can now provide the degres-d and degres-q control
different. signals of (22)-(23) to the renewables’ controller in order to
More specifically, in the case of renewables, (19) needs to regulate their operation for damping the power system
be added to the power system network model ((8)-(9)), which oscillations. However, the GSC controller of the RES can
actually represents the current equation for the ith RES bus. easily be regulated by active (dP) and reactive (dQ) power
j (δ i −π
m deviation signals. Thus, for compatibility issues, the degres-d and
(ires − di + jires − qi ) e + ∑ (Gik' + jBik' )vk e jθk =
)
2
0 (19) degres-q signals (according to the virtual RES model) should be
k =1

Based on (19), the injected current phasor (ires-d and ires-q) of converted into the compatible signals, dP and dQ, by the wide
each RES is expressed in terms of its terminal voltage (vres-d area controller and then be sent into the RES local controller as
and vres-q) respectively. This leads to (20)-(21), where indices n shown in Figs. 2 and 3.
and l refer to the total number of generator and RES buses, To overcome this compatibility issue, firstly the wide area
respectively. Gf and Bf are terms based on the elements of the controller needs to compute the new expressions for the virtual
admittance matrix, which result from the application of the first internal RES voltage according to (24) and (25), as resulting
from the application of the WAC signals of (22)-(23).

0885-8950 (c) 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
6

− d = vres − d + rires − d − xires − q + deres _ d


new
eres g
(24) discrete time (using a sampling rate of 10 kHz). The
development of the test systems, WAC, and RES is performed
e = vres − q + rires − q + xires − d + de
new
res − q
g
(25)
res _ q in MATLAB/Simulink using discrete time EMT simulations.
Thus, the resulting active and reactive power flow between The objective of this section is to validate the performance
the grid tied inverter and the LC filter can be calculated: of the proposed control scheme. To achieve this, various
3 new simulation case studies were performed, based on the two
= Pdq (eres − d ires − d + eres
new
− q ires − q )
(26)
2 dynamic test systems. To verify the performance of the
3 new proposed WAC-RES integration (WAC-RES), a comparison
= Qdq (eres − q ires − d − eres
new
− q ires − q )
(27) was made with the normal operation scenario where no WAC
2 exists in the system, (No WAC) and the scenario where WAC
Furthermore, the new resulting active and reactive power of
exists in the system, but it considers only the generators (not
RES at the PCC (Pnew and Qnew) according to the WAC signals
the RESs) in its design (WAC only). This comparison was
are given in (30)-(31), by subtracting the LC filter losses of
carried out by utilizing the Prony analysis, to evaluate the
(28)-(29) from the power of (26)-(27).
damping capability of all the above scenarios. This analysis
3 2
= Plosses (ires − d + ires
2
− q )r
(28) tool has the important advantage of requiring only the
2 measured data to provide all the information regarding the
3 2 stability of the system. Therefore, the Prony method is applied
=Qlosses (ires − d + ires
2
−q ) x
(29)
2 to extract the damping ratio (ζ) and frequency (f) of local and
= Pdq − Plosses
Pnew (30) inter-area modes for all the examined scenarios and case
studies (Table I and Table II). Note that only the dominant
= Qdq − Qlosses
Qnew (31) modes are provided, which are estimated through their energy
Finally, the required coordination signals (dP and dQ) are content [28] and their constant appearance [29]. A higher
derived according to (32)-(33), where Pres and Qres represent damping ratio, means better WAC performance.
the current active and reactive power output of RES.
B. IEEE 9-Bus Dynamic Test System
=
dP Pnew − Pres (32) The IEEE 9-bus dynamic test system includes three
=
dQ Qnew − Qres (33) machines and three loads. The initial total load of the system is
315 MW and 115 MVAr. To illustrate the performance of the
It is to be noted that the lower bound of dP, can be as
proposed methodology, a significant penetration of large-scale
much as the WAC needs it to be. However, its upper bound is
renewables is considered. More specifically, three 20 MVA
limited by the available active power, which the RES can
RESs are integrated at buses 5, 6 and 8 through step-up
provide. The resulting dP and dQ signals derived by the WAC
transformers, as shown in Fig. 1.
scheme can now be sent to the respective renewables in order
to ensure that both generators and RESs can properly 1) Case Study 1: Constant Loads
contribute to the power oscillation damping. To present the increased performance of the proposed
WAC, in this first case study a symmetric fault occurs. More
V. CASE STUDY specifically, a 5-cycle three-phase fault (grounded) is
A. Simulation Description considered in the simulation (which occurs on the line
connecting bus 4 and bus 6 at t=8 s) along with constant loads.
The proposed integration of WAC with RESs has been
Table I lists the Prony analysis results, for the dominant local
implemented and applied to the IEEE 9-bus and 39-bus
and inter-area modes of the system. From Table I, it can be
dynamic test systems, both developed according to [26]. The
noted that in the case of both local and inter-area oscillations,
generators included in the simulations are equipped with their
the proposed WAC-RES scheme provides significant
own local controllers (exciter and governor) in order to achieve
additional damping compared to the WAC only and No WAC
dynamic responses [26]. The renewable is implemented
scenarios. Furthermore, Fig. 4a presents the frequency
according to a full detailed model of a central grid tied inverter
oscillation of generator 2 (G2), where it is evident that the
[24], (a typical three-phase two-level voltage source converter
proposed scheme surpasses all the other scenarios.
based on six switching IGBTs). The controller of the RES (Fig.
2) is based on advanced synchronization [21], current control 2) Case Study 2: Dynamic Loads
[22], and PQ control techniques [23], in order to achieve a In order to further verify the stability improvement of the
proper operation under grid disturbances. The parameters of proposed method, the aforementioned symmetric fault is
the LC filter of the GSC are crucial for the design of the WAC utilized again for the case where dynamic non-linear loads are
scheme. For the IEEE 9-bus system, the LC filter parameters of included in the test system. Specifically, the constant loads
all renewables are set to r=0.01 Ω, x=0.086 Ω, and Cf =292 μF located at buses 6 and 8 (Fig. 1) are substituted by
according to [27]. In the case of the IEEE 39-bus test system corresponding dynamic loads (considering the same initial
RESs, r=0.01 Ω, x=0.297 Ω, and Cf =0.132 mF. All the local active and reactive power). The model of the dynamic load
controllers utilized in the study, such as the RES controllers adopted in this case study is the exponential dynamic load
and generator local controllers, have been digitally designed in model, which is described by the following equations:

0885-8950 (c) 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
7

TABLE I TABLE II
Prony Analysis for the IEEE 9-bus test system Prony Analysis for the IEEE 39-bus test system
Constant Loads Load Change
Type Type
No WAC WAC only WAC-RES No WAC WAC only WAC-RES
ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz)
Local mode 3.85 1.9 4.68 1.9 5.27 1.9 Local mode 8.49 1.4 9.82 1.5 11.3 1.4
Inter-area Inter-area
mode 4.62 0.93 5.69 0.95 6.29 0.96 mode 20.8 0.66 22.3 0.66 27.4 0.67

Type Dynamic Loads Type Symmetric Fault


No WAC WAC only WAC-RES No WAC WAC only WAC-RES
ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz)
Local mode Instability Instability 4.35 1.9 Local mode 3.29 1.4 5.22 1.4 6.23 1.3
Inter-area Inter-area
mode Instability Instability 6.35 0.95 mode 12 0.79 10.6 0.85 12.3 0.83
DFIG and Dynamic Loads Asymmetric Fault
Type Type
No WAC WAC only WAC-RES No WAC WAC only WAC-RES
ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz) ζ(%) f(Hz)
Local mode 5.27 1.9 5.85 1.9 6.1 1.9 Local mode 5.45 1.4 5.72 1.5 6.56 1.5
Inter-area Inter-area
mode 5.3 0.93 5.57 0.94 6.08 0.94 mode 15.1 0.77 16.3 0.77 19.2 0.75
1.01 (a)
(a) 1.06

V7 (pu)
f3 (pu)

1.005
1.05
1
1.04
6 8 10 12 14 16 18 20 22 2 3 4 5 6 7 8
(b) 1.05
2
(b)
V27 (pu)
Vt1 (pu)

1
1

0 0.95
2 3 4 5 6 7 8
10 15 20 25 30
(c)
1.004 (c)
Vt3(pu)

f10 (pu)

1.002
1 1
0.998
0.996
0.9
2 3 4 5 6 7 8
6 8 10 12 14 16 18 20 22
WAC-RES WAC only No WAC Time (s) WAC-RES WAC only No WAC Time (s)
Fig. 4. Comparison of WAC-RES, WAC only and No WAC performances on Fig. 5. Comparison of WAC-RES, WAC only and No WAC performances
the IEEE 9-bus test system considering: (a) the frequency oscillation of G3 for on the IEEE 39-test system considering: (a) the terminal voltage oscillation
constant loads, (b) the voltage oscillation of G1 for dynamic loads, and (c) the of G7 during load change, (b) the voltage oscillation of bus 27 during a
voltage oscillation of G3 for DFIG and dynamic loads. symmetric fault, and (c) the frequency oscillation of G10 during an
asymmetric fault.
np
V  1 WAC-RES scenario damps effectively all the local and inter-
P( s) = P0   area oscillations. Consequently, in this case study, only the
 V0  1 + Tp s
(34)
proposed scheme is capable of restoring the system’s stability,
nq
V  1 outperforming the other two scenarios.
Q( s) = Q0  
 V0  1 + Tq s
(35) 3) Case Study 3: DFIG and Dynamic Loads
The objective of the third test is mainly to study the impact
where V, P and Q are the positive sequence voltage, the active of considering two dynamic loads and a DFIG wind turbine
power and the reactive power of the load respectively, while (in addition to the fully converted RES) in the test system,
V0, P0 and Q0 are their initial values. np and nq represent the when a symmetric fault takes place at bus 2. The dynamic
exponents which control the nature of the load. Lastly, Tp and loads are placed once again on buses 6 and 8, but in this case
Tq are the recovery time constants for the active and reactive study [np nq]=[1.01 1.04] and [Tp Tq]=[0.1s 0.2s]. The DFIG
power, respectively. For this case study, [np nq]=[0.39 1.72] wind turbine has substituted the fully converted renewable of
and [Tp Tq]=[363s 0.1s] in order to achieve an intense dynamic bus 12. The methodology, as it is presented in Section IV.B,
behavior of the two loads. As illustrated by Fig. 4b and Table can be applied directly to the GSC of the DFIG wind turbine
I, the No WAC and the WAC only scenarios lead the system regarding the reactive power coordination (dQ). The active
to instability when the fault occurs. On the contrary, the power coordination (dP) cannot be used directly in this type of
wind turbine due to the strong coupling between the active and

0885-8950 (c) 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
8

reactive power of the induction generator that does not VI. CONCLUSION
straightforwardly allow the decoupled coordination of the A hierarchical WAC scheme that coordinates the generator
DFIG. More specifically, by sending the dP signal to the GSC and renewable local controllers is proposed to improve the
will have as a result to also affect the rotor side converter power system dynamic stability under a high penetration of
which in turn will change the active power of the rotor. As a renewable energy sources. It utilizes wide area measurements
consequence, the active power of the stator of the induction from the grid to provide suitable control actions to all the local
machine will change as well. Due to the coupling nature of the controllers for maximizing their performance. This control
induction machine, the reactive power of the DFIG will also scheme became feasible by making the generators “aware” of
be affected. Therefore, only the dQ signal is sent to the GSC the oscillations of the renewable sources and vice versa,
which allows the decoupled coordination of the reactive power through the WAC signals. In addition, adjustments of the
without affecting the active power. Nevertheless, the Prony derived RES WAC signals are presented, so that minor
results of Table I illustrate once again that the transient modifications will be required on the existing RES local
performance of the system is improved with the proposed controllers. Various case studies under different contingencies
WAC-RES scheme. More specifically, in Fig. 4c one can note based on the IEEE 9-bus and 39-bus dynamic test systems, are
that the WAC-RES compensation of local oscillations excels used to evaluate the performance of the proposed control
all the other scenarios. scheme. In all cases, the simulation results indicate that the
C. IEEE 39-Bus Test System proposed WAC coordination of generators and RESs improves
the damping performance of the entire system under a high
This test system consists of ten machines and nineteen
loads. The initial total load of the system is 6097 MW and penetration of renewables. The results are obtained in
comparison to the normal operation scenario where no WAC
1409 MVAr [26]. To apply a high penetration of renewables,
exists in the system and the scenario where WAC exists in the
five 250 MVA RESs are considered to be installed on buses 8,
system, but it does not consider the RESs in its design. To
14, 24, 25 and 27 through step-up transformers. Table II
verify the results, the Prony analysis has been utilized.
presents the Prony analysis results for the dominant local and
inter-area modes of all the case studies and scenarios.
VII. REFERENCES
1) Case Study 1: Load Change
To evaluate the performance of the proposed scheme in this [1] D. Roberson and J. F. O’Brien, "Loop Shaping of a Wide-Area Damping
system as well, firstly a total load increase of 640 MW and Controller Using HVDC," IEEE Trans. on Power Systems, vol. 32, no. 3,
pp. 2354-2361, 2017.
200 MVAr (11% increase of the initial load) occurs on the
[2] B. P. Padhy, S. C. Srivastava, and N. K. Verma, "A Wide-Area Damping
loads of buses 3 and 16 at t=3 s. As shown in Table II and Fig. Controller Considering Network Input and Output Delays and Packet
5a, the proposed method surpasses the other two scenarios Drop," IEEE Trans. on Power Systems, vol. 32, no. 1, pp. 166-176, 2017.
regarding the effective damping of both local and inter-area [3] F. Bai, Y. Zhu, X. Wang, K. Sun, Y. Ma, M. Patel, E. Farantatos, and N.
oscillations. Bhatt, "Design and implementation of a measurement-based adaptive
wide area damping controller considering time delays," Electric Power
2) Case Study 2: Symmetric Fault System Research, vol. 130, pp. 1-9, Jan. 2016.
[4] A. R. Messina, "Signal Processing Methods for Small-Signal Dynamic
Next, a 5-cycle three-phase grounded fault has been applied Properties," in Inter-area oscillations in power systems: A nonlinear and
on the line connecting bus 3 and bus 4, at t=3 s. Table II nonstationary perspective (power electronics and power systems),
presents the Prony results of this case study, where it can be Guadalajara, Springer, 2009, pp. 1-36.
seen that the WAC-RES scenario provides (once again) better [5] X. Wu, F. Dörfler, and M. R. Jovanović, "Input-Output Analysis and
damping of the system’s oscillations (Fig. 5b). An interesting Decentralized Optimal Control of Inter-Area Oscillations in Power
Systems," IEEE Trans. on Power Systems, vol. 31, no. 3, pp. 2434-2444,
point here is that although the No WAC scenario has a local 2016.
mode damping that is approximately half of the proposed [6] L. Zhu, H. Liu, Z. Pan, Y. Liu, E. Farantatos, M. Patel, S. McGuinness,
method’s performance, it has a comparable damping ratio to and N. Bhatt, "Adaptive wide-area damping control using measurement-
the latter regarding the compensation of the inter-area modes. driven model considering random time delay and data packet loss," Proc.
IEEE PES General Meeting, pp. 1-5, 2016.
3) Case Study 3: Asymmetric Fault [7] S. Wang, X. Meng, and T. Chen, "Wide-area control of power systems
through delayed network communication," IEEE Trans. Control Systems
Finally, the application of an asymmetric fault has been Technology, vol. 20, no. 2, pp. 495-503, Mar. 2012.
investigated. More specifically, a two-phase ungrounded fault [8] X. Zhang, C. Lu, X. Xie, and Z. Y. Dong, "Stability analysis and
takes place on the line connecting bus 3 and bus 4, at t=3 s. controller design of a wide-area time-delay system based on the
Table II illustrates clearly that in the case of an asymmetric expectation model method," IEEE Trans. Smart Grid, vol. 7, no. 1, pp.
520-529, July 2016.
fault, the proposed scheme still has a better damping of both,
local and inter-area oscillations, compared to the WAC only [9] B. Padhy, S. Srivastava, and N. Verma, "A network delay compensation
technique for wide area and SVC damping controller in power systems,"
and No WAC scenarios. An example of the local oscillation Proc. IEEE PES T&D Conference and Exposition, pp. 1-6, 2014.
damping is illustrated in Fig. 5c, which presents the frequency [10] F. Okou, L.-A. Dessaint, and O. Akhrif, "Power system stability
response of G10. enhancement using a wide area signals based hierarchical controller,"
IEEE Trans. Power Systems, vol. 20, no. 3, pp. 1465-1477, Aug. 2005.
[11] D. Gautam, V. Vittal, and T. Harbour, "Impact of increased penetration
of DFIG-based wind turbine generators on transient and small signal

0885-8950 (c) 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRS.2018.2822315, IEEE
Transactions on Power Systems
9

stability of power systems," in IEEE Trans. on Power Systems, vol. 24, VIII. BIOGRAPHIES
no. 3, pp. 1426-1434, Aug. 2009. Lazaros Zacharia (S’15) obtained the BSc in Electrical
[12] D. A. Halamay, T. K. A. Brekken, A. Simmons, and S. McArthur, and Computer Engineering from the National Technical
"Reserve requirement impacts of large-scale integration of wind, solar, University of Athens, Greece, in 2013. In 2014, he
and ocean wave power generation," in IEEE Trans. on Sustainable obtained the MSc in Sustainable Energy Futures with
Energy, vol. 2, no. 3, pp. 321-328, July 2011. distinction from Imperial College, London. He is
currently a Ph.D. student in the Department of Electrical
[13] T. Surinkaew and I. Ngamroo, "Hierarchical co-ordinated wide area and
and Computer Engineering at the University of Cyprus
local controls of DFIG wind turbine and PSS for robust power oscillation and a Researcher in the KIOS Research and Innovation
damping," IEEE Trans. Sustainable Energy, vol. 7, no. 3, pp. 943-955, Center of Excellence. His research interests lie in the areas of Wide Area
Jan. 2016.
Monitoring and Control, power systems dynamics and stability, smart grids
[14] M. Mokhtari and F. Aminifar, "Toward wide-area oscillation control and renewables.
through doubly-fed induction generator wind farms," IEEE Trans. Power
Systems, vol. 29, no. 6, pp. 2985-2992, Mar. 2014.
Lenos Hadjidemetriou (S’11-M’16) received the
[15] A. Yogarathinam and N. R. Chaudhuri, "Wide-Area Damping Control Diploma in Electrical and Computer Engineering in
using Multiple DFIG-based Wind Farms Under Stochastic Data Packet 2010 from the National Technical University of Athens,
Dropouts," IEEE Trans. Smart Grid, no. 99, 2016. Athens, Greece, and his Ph.D. degree in Electrical
[16] A. E. Leon and J. A. Solsona, "Power Oscillation Damping Improvement Engineering in 2016 from the University of Cyprus.
by Adding Multiple Wind Farms to Wide-Area Coordinating Controls," Currently, he is a Postdoctoral Research Fellow at the
IEEE Trans. on Power Systems, vol. 29, no. 3, pp. 1356-1364, May KIOS Research and Innovation Center of Excellence,
2014. University of Cyprus. His research interests include
[17] L. Zacharia, M. Asprou, and E. Kyriakides, "Design of a data delay renewable energy systems, grid synchronization
compensation technique based on a linear predictor for wide-area methods, fault ride through control, power electronics, and micro-grids.
measurements," Proc. IEEE PES General Meeting, pp. 1-5, 2016.
[18] P. W. Sauer and M. A. Pai, Power system dynamics and stability, Elias Kyriakides (S’00, M’04, SM’09) received the
Prentice Hall, 1998. B.Sc. degree from the Illinois Institute of Technology
in Chicago, Illinois in 2000, and the M.Sc. and Ph.D.
[19] P. M. Anderson and A. A. Fouad, "Typical System Data," in Power
degrees from Arizona State University in Tempe,
system control and stability, 2nd Edition, New Jersey, Wiley-IEEE Press,
Arizona in 2001 and 2003 respectively, all in
2002, pp. 555-581.
Electrical Engineering. He is currently an Associate
[20] M. Eremia and M. Shahidehpour, Handbook of electrical power system Professor in the Department of Electrical and
dynamics, Wiley-IEEE Press, 2013. Computer Engineering at the University of Cyprus,
[21] L. Hadjidemetriou, E. Kyriakides, and F. Blaabjerg, "A robust and a founding member of the KIOS Research and Innovation Center of
synchronization to enhance the power quality of renewable energy Excellence. He served as the Action Chair of the ESF-COST Action IC0806
systems," IEEE Trans. Industrial Electronics, vol. 62, no. 8, pp. 4858- “Intelligent Monitoring, Control, and Security of Critical Infrastructure
4868, Feb. 2015. Systems” (IntelliCIS) (2009-2013). He is an Associate Editor of the IEEE
Systems Journal and an Editor of the IEEE Transactions on Sustainable
[22] L. Hadjidemetriou, E. Kyriakides, and F. Blaabjerg, "A grid side
Energy. His research interests include wide area monitoring and control of
converter current controller for accurate current injection under normal
power systems, the optimization of power system operation techniques, and
and fault ride through operation," Proc. Industrial Electronics Society,
the integration of renewable energy sources.
IECON, pp. 1454-1459, 2013.
[23] L. Hadjidemetriou, E. Kyriakides, and F. Blaabjerg, "An adaptive tuning
mechanism for phase-locked loop algorithms for faster time performance
of interconnected renewable energy sources," IEEE Trans. on Industry
Applications, vol. 51, no. 2, pp. 1792-1804, Aug. 2015.
[24] L. Hadjidemetriou, P. P. Demetriou, and E. Kyriakides, "Investigation of
different fault ride through strategies for renewable energy sources,"
Proc. IEEE PES PowerTech, pp. 1-6, 2015.
[25] M. Heidary and S. A. M. R. Alikhanlou, "Estimation of rotor angle based
on operating variables measured by PMU," Proc. IAJC-ISAM
International Conference, 2014.
[26] P. Demetriou, M. Asprou, J. Quiros-Tortos, and E. Kyriakides,
"Dynamic IEEE test systems for transient analysis," IEEE Systems
Journal, pp. 1-10, July 2016.
[27] A. Reznik, M. G. Simões, A. Al-Durra, and S. M. Muyeen, "LCL Filter
Design and Performance Analysis for Grid-Interconnected Systems,"
Trans. on Industry Applications, vol. 50, no. 2, pp. 1225-1232, Mar-Apr.
2014.
[28] V. S. Patel, F. S. Bhil, F. S. Kazi, and S. R. Wagh, "Energy-sorted Prony
analysis for identification of dominant low frequency oscillations," in
Australian Control Conference, Fremantle, 2013 .
[29] D. P. Wadduwage, U. D. Annakkage, and K. Narendra, "Identification of
dominant low-frequency modes in ring-down oscillations using multiple
Prony models," IET Gener. Transm. Distrib., vol. 15, no. 9, pp. 2206-
2214, 2015.

0885-8950 (c) 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.

You might also like