This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON POWER DELIVERY 1

Modeling of LCC-HVDC Systems

Using Dynamic Phasors
Working Group on Modeling and Analysis of System Transients Using Digital Programs, General Systems
Subcommittee, T&D Committee, IEEE Power and Energy Society
M. Daryabak, S. Filizadeh, J. Jatskevich, A. Davoudi, M. Saeedifard, V. K. Sood, J. A. Martinez, D. Aliprantis,
J. Cano, and A. Mehrizi-Sani

Abstract—This paper presents an average-value model of a line- Electromagnetic-transient (EMT) simulation programs, in
commutated converter-based HVDC system using dynamic pha- which high-fidelity models of transmission systems and power
sors. The model represents the low-frequency dynamics of the con- electronic converters are available, have been extensively used
verter and its ac and dc systems, and has lower computational re-
quirements than a conventional electromagnetic-transient (EMT) for the analysis and design of HVDC schemes [4]–[6]. Despite
switching model. The developed dynamic-phasor model is verified its accuracy in representing the fast switching dynamics of
against an EMT model of the CIGRE HVDC Benchmark. Sim- converters, an EMT simulator may indeed to be excessively
ulation results confirm the validity and accuracy of the average detailed for the study of interconnected networks with em-
value model in predicting the low-frequency dynamics of the ac- bedded HVDC, particularly when low-frequency dynamics
and dc-side quantities. Merits and applicability limitations of the
average model are highlighted. of the network are of interest. For such cases, simulation of
the switching transients, which are often small in magnitude,
Index Terms—Dynamic average modeling, dynamic phasors, adds unnecessary computational complexity. Dynamic average
electromagnetic-transients (EMT) simulation, line-commutated
converter (LCC) HVDC. modeling [7] aims to develop low-intensity models that repre-
sent the slow dynamics of power-electronic-based systems by
neglecting the switching transients.
I. INTRODUCTION This paper presents an average-value model for an
LCC-HVDC system using dynamic phasors. Dynamic phasors
use the quasi-periodic switching nature of a power-electronic

H IGH-VOLTAGE direct-current (HVDC) transmission

systems play an important role in modern power net-
works. HVDC is considered to be a suitable option for such
converter, for which voltages and currents can be represented
using time-varying Fourier coefficients [8]–[10]. Dynamic
phasor modeling retains the low-frequency dynamic character-
applications as transmission via underground/submarine ca- istics of a power system without having to model the high-fre-
bles; linking ac systems with unequal frequencies; integration quency transients caused by the operation of power-electronic
of energy sources, such as hydropower; and long-distance switches and converters. In addition, a dynamic phasor model
transmission [1], [2]. Although a new breed of HVDC systems can be easily augmented to account for harmonic components,
using voltage-source converters (VSC) is also developed, the if so needed.
majority of existing HVDC systems are based on the con- Dynamic phasors have been successfully used in modeling
ventional line-commutated converters (LCCs). LCC-HVDC and analysis of electrical machines [11]–[13]; power system
schemes offer the benefits of a mature technology and are dynamics and faults [14], [15]; flexible ac transmission sys-
available in much larger ratings than the currently available tems (FACTS) devices [16], [17]; subsynchronous resonance
VSC-HVDC systems [3]. (SSR) [18]; dc-dc converters [19]; multiconverter dc systems
[20]; and multiconverter ac systems [21]. Special studies, in-
Manuscript received July 02, 2013; revised January 17, 2014; accepted Feb- cluding real-time and repetitive simulations [5], [22], [23], can
ruary 21, 2014. Paper no. TPWRD-00753-2013.
benefit from the reduction in computational intensity offered
The authors are contributing members of the IEEE Task Force on Dynamic
Average Modeling. by dynamic phasor modeling. This paper extends the use of
Task Force on Dynamic Average Modeling is with the Working Group on dynamic phasors to modeling of large-excursion transients of
Modeling and Analysis of System Transients Using Digital Programs, General
LCC-HVDC systems, caused by control actions or faults. The
Systems Subcommittee, T&D Committee, IEEE Power and Energy Society.
Task Force Members: S. A. Abdulsalam, D. Aliprantis, U.Annakkage, H. reduced simulation intensity of the model resulting from its ne-
Atighechi, J. Belanger, J. Cano, S. Chiniforoosh, A. Davoudi, V. Dinavahi, glecting switching transients makes it particularly useful in the
O. Faruque, S. Filizadeh, D. Goldsworthy, A. Gole, R. Iravani, J. Jatskevich
study of large-signal, low-frequency transients, or in repetitive
(Chair), R. Jayasinghe, H. Karimi, M. Kuschke, A. St. Leger, J. Mahseredjian,
J. Martinez, A. Mehrizi-Sani, N. Nair, L. Naredo, T. Noda, J. N. Paquih, J. Per- simulations when time saving over EMT simulations become
alta, A. Ramirez, A. Rezaei-Zare, M. Rioual, M. Saeedifard, K. Schoder, V. significant.
Sood, K. Strunz, A. VanDerMeer, X. Wang, and A. Yazdani.
This paper will proceed by presenting a synopsis of a typ-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. ical LCC–HVDC system, followed by a description of the
Digital Object Identifier 10.1109/TPWRD.2014.2308431 mathematical foundations of modeling using dynamic phasors.

0885-8977 © 2014 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 inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON POWER DELIVERY

Fig. 1. The 12-pulse LCC–HVDC system.

Dynamic phasors are then applied to develop a low-frequency

model of the CIGRE HVDC benchmark system as a case
study. Simulation results are given to investigate the accuracy
and applicability of the developed model in predicting the
large-signal transient behavior of the system.

II. LINE-COMMUTATED CONVERTER-BASED HVDC

Fig. 1 shows a schematic diagram of a monopolar Fig. 2. HVDC control scheme.
LCC-HVDC system, similar to the first CIGRE HVDC
benchmark system [24] used in the case studies in this paper.
loops (PLLs) are employed at both ends. Modeling of the PLL
A. High-Power Circuitry is presented in Section IV.
Filter banks and specially connected transformers interface
the HVDC converter blocks to the ac system on the inverter III. DYNAMIC PHASOR PRINCIPLES
and rectifier sides. The ac systems are modeled by Thevenin cir-
cuits, and are characterized by their effective short-circuit ratio A firm understanding of the concept of dynamic phasors and
(ESCR). In the system shown in Fig. 1, tuned (11th and 13th har- the way they are applied to dynamical systems is essential in de-
monics) and high-pass filters remove higher frequency compo- veloping a model using them. The following section, therefore,
nents and supply the required reactive power of the converters, presents a brief overview of dynamic phasors.
which exhibit a lagging power factor at both ends. The concept of dynamic phasors is based on the generalized
averaging method proposed in [9]. Here, a waveform is
B. Rectifier and Inverter Control Systems viewed over the interval and is described using
its Fourier series expansion coefficients , which are ob-
Fig. 2 shows typical control blocks of a conventional tained as follows:
LCC–HVDC system. The control variables used are also
marked in Fig. 1. Normally, the rectifier controls the dc line
(1)
current, whereas the inverter operates in constant extinc-
tion-angle control. Although additional control functions are
deployed for special modes of operation, during normal opera- where
tion, the rectifier and inverter control functions can be described
using proportional-integral (PI) controllers acting upon the dc
(2)
current and the inverter extinction angle, respectively. The
voltage-dependent current order limit (VDCOL) reduces the dc
current order during ac undervoltage conditions to protect the and ; in modeling power-electronic converters, is
valves [1], [2], [22]. The internal components and connections selected to be the switching period. The waveform may
used in the control system of Fig. 2 are available in detail in contain a variety of harmonic components. For example, the
[27], [28] and are used identically in this paper. The main con- dc-side voltage waveform of a six-pulse LCC converter contains
troller gains and time constants are listed in Table II. To ensure a large dc (i.e., 0-order) component as well as harmonics.
proper timing of firing pulses to the converters, phase-locked The describes the th-order harmonic component.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

DARYABAK et al.: MODELING OF LCC-HVDC SYSTEMS USING DYNAMIC PHASORS 3

Although the waveform may attain periodicity in

steady state, it may not generally repeat the same form in each
switching period during transients. As such, its Fourier coef-
ficients over the sliding interval will not be constant and will
be functions of time and, hence, the notation . These
coefficients are referred to as dynamic phasors.
The benefit of generalized state-space averaging enables its
users to select a desired number of Fourier coefficients to con-
struct an approximation of the original waveform. For example,
if the waveform comprises low- and high-frequency com-
Fig. 3. Six-pulse converter.
ponents, the user may choose to consider only its low-frequency
contents in a simplified representation. Normally, only the first
or first few dominant harmonic components are selected for this lows pertains to the six-pulse bridge in Fig. 3 and is directly
purpose. Inclusion of additional components increases the accu- applicable to the bottom six-pulse converter of Fig. 1 (i.e., the
racy of estimation at the expense of a higher order model. converter connected to the - transformer), whose leakage in-
When applying dynamic phasors to power-electronic circuits, ductance is shown as in Fig. 3.
the following properties prove useful: The phase voltages of the ac system at the ac system side of
a six-pulse converter (see Fig. 3) are as follows:
(3)

(4)

only if is a real function (5) (6)

where denotes the complex conjugate operator.
Equation (3) holds for situations where the fundamental fre- where is the nominal system frequency (50 or 60 Hz) and the
quency is constant. It is shown that the same formulation will phase shift of phase-a voltage is introduced to simplify the
remain highly accurate when the system frequency changes, in representation of the switching functions later.
particular, when the rate of change of frequency is slow [9]. Fig. 4 shows the dc-side voltage, as well as the voltage and
The fundamental operations presented in this section are next current switching functions of the phase- of the converter. A
applied to the building blocks of an LCC-HVDC system and are switching function serves to relate the ac- and dc-side quantities,
then interfaced to form its average model. and is obtained by inspecting the waveforms it interrelates [34]
with. For example, the line current in Fig. 3 can be obtained
by multiplying its switching function [Fig. 4(b)] by the dc cur-
IV. MODELING OF LCC-HVDC SYSTEMS
rent . Note that the switching functions do include the effect
In this section, the dynamic phasor model of an HVDC of the transformer leakage inductance in creating an overlap pe-
system is developed. It will include the first dominant Fourier riod denoted by . Transitions in the current switching function
component for the ac and the dc sides of the system; that is, are assumed to be linear. While this is an approximation, it ap-
the zero-order term for the dc side and the first-order term for pears to only have negligible impacts on the results. Nonlinear
the ac-side quantities. This model will represent the dynamics approximations for the transitions have also been proposed [25].
of the dc component of the dc quantities and the fundamental Switching functions for the quantities of the other phases are ob-
frequency component of the ac quantities. tained by shifting phase-a switching functions by .
The dc-side voltage waveform of the six pulse converter can
A. Line-Commutated Converter Model be expressed in terms of the ac-side phase voltages and their
Fig. 3 shows a six-pulse bridge, which is the building block of switching functions as follows:
an LCC-HVDC system. It consists of six thyristor valves, fired
consecutively every 60 with an overall delay of known as (7)
the firing angle. In derivation of an average model, a lossless
The , , and are switching functions of the voltages
converter is assumed. It is also assumed that the smoothing re-
of phases , , and , respectively. The function , is shown
actor at the dc side is adequately large so that the ripple on the
in Fig. 4(e). The other two functions are phase-shifted versions
dc current is negligible, that is, . Normally,
of this switching function. To determine the dc value (0th-order
two six-pulse bridges are connected in series in what is called
component) of the dc-side voltage, one can apply (2) with
a 12-pulse configuration, using - and - transformers as
to (7); therefore
shown in Fig. 1. The 12-pulse converter arrangement offers im-
proved harmonic spectrum and achieves larger dc voltage. (8)
It is instructive and convenient to first develop an average
model of the six-pulse converter. Once it is developed, it can The three terms on the right-hand side of (8) include the aver-
be easily scaled to the 12-pulse case. The derivation that fol- ages of the products of the switching functions and phase volt-
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON POWER DELIVERY

where is the dc component (i.e., the average) of the

dc-side voltage, and

(13)

(14)

The is often called the equivalent commutation resistance,

is called ideal no-load dc voltage, and is the average
of the dc current at the dc terminal of the six-pulse converter
[26], [27] and may vary with time. The amplitude of the PCC
voltage ( ) may also vary with time due to, for example, the
varying amount of current through the ac line.
Note that the generalized averaging formula used here allows
inclusion of harmonics that are ignored in conventional aver-
aging techniques. For instance, it would have been straightfor-
ward to retain higher harmonic components in (9) if a more ac-
curate expression for the dc-side voltage would have been de-
sired. This flexibility in deciding the level of detail in a model
is a direct benefit of the dynamic phasor approach.
Similarly, it is noted that for the ac-side current, the following
expression holds:

(15)
Fig. 4. Waveforms and switching functions: (a) actual line and dc-side voltages
of the six-pulse converter; (b) current switching function for the - 6-pulse In order to determine the fundamental-frequency component
bridge; (c) current switching function for the - 6-pulse bridge; (d) current
switching function for the 12-pulse converter; and (e) voltage switching func- of ac current , (2) is applied
tion. to (15) with 1, which yields (16) and (17) for the magnitude
and phase of the fundamental Fourier component of the phase-a
current
ages. These terms can be expanded into products of averages
using (4), as follows:
(16)

(17)
(9)
Note that the term is due to the assumed phase angle of
For a low-frequency average, the terms involving harmonics
phase-a voltage [see (6)]. In general, this must be set to the ac-
other than the fundamental, that is, terms with subscripts
tual phase angle of the phase-a voltage.
are ignored. The index-1 averages of the phase-a voltage and the
The expressions given in (12), (16), and (17) describe the dy-
switching function for phase-a are obtained using (1) as follows.
namics of the dc voltage and ac current of the six-pulse con-
The corresponding terms for phases b and c are simply phase-
verter. It is noted that the dc voltage of the 12-pulse converter
shifted versions of these expressions
( or in Fig. 1) results from the series connection of two
six-pulse converters with equal average voltages. Therefore, the
(10) average of the dc voltage of the 12-pulse converter is simply ob-
tained as follows:
(11)

Substituting these terms into (9) yields the following expression (18)
for the average dc voltage of the six-pulse bridge:
where is the average of the dc current of the 12-pulse
converter.
Similarly, the ac-side current of the 12-pulse converter can be
(12) obtained by noting that the switching functions for ac currents
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

DARYABAK et al.: MODELING OF LCC-HVDC SYSTEMS USING DYNAMIC PHASORS 5

In a converter system comprising six-pulse bridges, the

average equivalent inductance of the combined ac and dc sys-
tems for each of the receiving and sending ends will therefore
be as follows:

Fig. 5. DC transmission-line model. (22)

The for the rectifier and the inverter sides may be different
of the - and - -connected six-pulse bridges have equal due to the leakage inductance of the transformer at the respec-
fundamental components and are in phase; i.e., tive side. The dynamic equations of the dc transmission line are
as follows:

(19)

therefore

(20)
(23)
For converter banks with higher pulse numbers (e.g., 24 or
48 pulses), it is merely enough to multiply the averaged values The dynamic phasor equivalents of these equations are ob-
of the constituent six-pulse converters by the respective number tained using (2) with (to denote the dc component). The
of blocks used to create the bank. resulting equations will have a form similar to the ones in (23)
except that all voltages and currents will appear with the
notation.
B. DC System C. AC Network
Modeling of the ac network involves development of dy-
The dc system of an HVDC transmission system consists of namic phasor representations for components, such as trans-
a dc transmission line/cable and the smoothing reactors at the formers, filters, loads, and transmission lines [25], [29]. It was
rectifier and inverter ends. Here, a T-model is used for the dc shown in Section IV-A that if the dc current of the converter is
transmission line as shown in Fig. 5. This configuration with assumed to be purely dc, that is, comprising only one component
constant (frequency independent) lumped circuit elements is with a 0 Hz frequency, the fundamental component of the ac line
used throughout this paper for all of the developed models. The current will be a sine wave with a fixed amplitude [as shown in
smoothing reactors are included in the equivalent series induc- (16)] and with a frequency of . During transients, the dc cur-
tance of the line. rent may undergo oscillations and, hence, will not be constant. If
The dynamic behavior of the line is described using three the dc current during its transient variations has a low-frequency
state variables—its two inductor currents and one capacitor oscillatory component at , it can be shown that the amplitude
voltage. It can be shown that the equivalent leakage inductance of the fundamental component of the ac line current will have
of the ac transformer, seen from dc side of the converter, is oscillations at . This amplitude-modulated fundamental com-
different during the commutation and noncommutation periods ponent results in two sidebands of and [31]. If
[27]. During noncommutation intervals of a six-pulse bridge, is relatively small, both sidebands are adequately close to
two ac transformers each with an inductance of are in series and the ac system may be represented using a constant admit-
with the dc-side inductance ( ), thereby presenting an ac tance representation. However, if is large, both sidebands are
equivalent inductance of . During the commutation period, significantly different from and the constant-admittance rep-
the paralleled ac inductances of the two phases involved in com- resentation of the ac network is no longer acceptable, and the
mutation and the single ac-side inductance of the return path dynamic phasor model of the ac network must be used to prop-
are in series, creating an ac equivalent inductance of . erly represent the ac network dynamics during transients [30],
In each interval of , the commutation and noncommutation [31].
periods are and radians long, respectively. There- Consider, for example, a series RL circuit. The time-domain
fore, the average inductance of the ac system seen from the dc differential equation for such a circuit will be as follows:
terminals of the six-pulse converter will be as follows.
(24)

By applying (2) and (3) for (fundamental component)

to (24), the following dynamic phasor relationship is obtained:

(21) (25)
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON POWER DELIVERY

Fig. 7. Phasor-based model of the PLL shown in Fig. 6.

the angle of the ac waveform within a 20% window around the

Fig. 6. Three-phase PI-controlled PLL.
nominal frequency .
From the viewpoint of the fundamental ac components, the
PLL can be represented as a closed-loop control system as
shown in Fig. 7. Whenever there is a change (due to perturba-
where and are the dynamic phasors of voltage and cur-
tions in the system or changes in system setpoints) in the phase
rent, respectively. Note that (25) reduces to the conventional
angle of the voltage at the PLL point, its output undergoes
phasor equivalent of the series RL circuit if the rate of change
a transient period before it detects and locks onto the phase
of tends to zero. Neglecting the term in other sit-
angle. The closed-loop system of Fig. 7 is used to control the
uations changes the equations from differential to algebraic, al-
dynamics of this transient period.
beit at the expense of reduced accuracy. Therefore, the con-
It must be noted that the use of a PLL is not the only available
stant-admittance matrix formulation of ac systems is, in fact,
option to track an ac waveform. Methods, such as the virtual flux
a special subset of their dynamic phasor model. Separating (25)
concept [33], are also available for tracking.
into real and imaginary components yields the following differ-
ential equations for the circuit: V. SIMULATION RESULTS AND MODEL VERIFICATION
A dynamic phasor-based model for the CIGRE HVDC
(26)
benchmark model [24] is developed in MATLAB/SIMULINK
as per the procedures in Section IV. The benchmark model has
where a power circuitry and control layout similar to Figs. 1 and 2.
Other system specifications are summarized in the Appendix.
(27) The developed average dynamic phasor model is validated
against a detailed switching model that has been implemented
Similar equations can be derived for RC and other combi- in PSCAD/EMTDC. In order to investigate the accuracy of
nations as well. The ac system in the CIGRE HVDC model the dynamic phasor model in representing transients below
consists of an impedance for the terminating ac system and the switching frequency, a third model of the system is also
shunt-connected RLC branches for the filters and reactive com- used in which the bridge converters are modeled as shown in
ponents at the converter terminals [24]. Dynamic phasor models Section IV-A, but the ac system is represented using a constant
for these elements are obtained using a similar procedure and in- admittance matrix. In this model the passive elements of the ac
terfaced to form a complete ac-side representation. The ac quan- systems at the rectifier and the inverter sides are represented
tity dynamic phasors are complex numbers (as shown in (27)); using impedances at the fundamental frequency of 50 Hz. This
the dc quantity dynamic phasors are real numbers. The linkage constant-admittance matrix model ignores the time variations
between these two sets is provided through the converter equa- of the first Fourier component of the line current, as described
tions [such as (16)]. in Section IV-C.
Inclusion of higher order harmonics in the representation of A number of case studies involving all three models are pre-
ac quantities is simply achieved by developing dynamic equa- sented as follows:
tions similar to (25) with higher index terms (e.g., 11 and Case 1) rectifier-side dynamics;
13) to include 11th- and 13th-order harmonics. Appendix B Case 2) complete system dynamics;
shows the required formulation as well as simulation results of Case 3) fault analysis without commutation failure;
the rectifier-side ac current. Case 4) fault analysis with commutation failure.
D. Phase-Locked Loop (PLL) A. Rectifier-Side Dynamics (Case 1)
A PLL is a control system for tracking the phase angle of the For the simulation of Case 1, the inverter side is modeled as a
fundamental positive-sequence component of an ac waveform. constant dc source of 500 kV (replacing in Fig. 5). The recti-
Several PLL architectures are available; and the model devel- fier control system follows its dc-current order via adjusting the
oped here pertains to the PLL shown in Fig. 6 [32]. It can, how- converter’s firing angle. A PLL locked onto the rectifier PCC
ever, be readily extended to other configurations as well. maintains synchronism with the ac voltage and provides a ref-
In Fig. 6, the output is a ramp train between that erence for generation of the firing pulses.
is synchronized with the positive zero-crossing of the positive- Fig. 8 shows the response of the system to a step change in
sequence fundamental component of phase-a voltage. The limits the current order from 1.1 to 0.9 p.u. Three traces generated by
shown on the integrator input in Fig. 7 ensure that the PLL tracks the three models are shown. The dynamic phasor model follows
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

DARYABAK et al.: MODELING OF LCC-HVDC SYSTEMS USING DYNAMIC PHASORS 7

Fig. 8. Step response of the dc current (Case 1: rectifier-side dynamics).

the transient simulation model’s response excluding the high-

frequency switching contents. The constant-admittance matrix
model fails to show the transient oscillation of the actual re-
sponse, although it settles into the correct steady-state value. As
shown in Fig. 8, the system response has damped oscillations at
a frequency of approximately 62 Hz. The dynamic phasor model
successfully predicts these oscillations. These oscillations are,
however, too rapid for the constant admittance matrix represen-
tation, and hence are absent in its predicted response. It must be
noted that the 62-Hz oscillations have occurred in the first dom-
inant Fourier component of the dc current, that is, the 0-order
term, for which a dynamic phasor model is developed. Predic-
Fig. 9. Response of the system to a step change in the current order. top: rec-
tion of these oscillations does not require inclusion of additional tifier current, middle: inverter current, bottom: PCC ac voltage (Case 2).
Fourier components since the harmonic contents of the dc cur-
rent of a 12-pulse converter (with symmetrical firing) include
the considered dc component as well as harmonics (720
Hz, 1440 Hz, etc.), which are, by far, higher than 62 Hz.
Note that the enhanced accuracy of the complete dynamic
phasor model results from inclusion of the dynamics of both the
converters and the ac system. The developed dynamic phasor
model of the converter, despite being “low-frequency” in na-
ture, represents some higher frequency oscillations such as the
ones shown in Fig. 8. Once these oscillations are presented to
the ac network, they undermine the accuracy of its constant-ad-
mittance matrix model, and hence the requirement for the use
of a better ac system representation.

B. Complete System (Case 2)

The complete system model with both converters and their re-
spective ac systems and with a current controller on the rectifier Fig. 10. Response of the system to a 5% reduction in the inverter-side ac
side and constant extinction-angle controller on the inverter side voltage magnitude. top: rectifier current, bottom: inverter current (Case 3).
with a VDCOL is considered. Fig. 9 shows the step response
of the system to a large current-order change from 1.05 to 0.5
p.u. The rectifier current controller is engaged to adjust its firing faults, which do not result in commutation failure, are consid-
angle. The inverter controller maintains its extinction angle at ered. Figs. 10 and 11 show the rectifier and inverter current when
15 . The figure also shows the dynamics of the ac voltage mag- the magnitude and phase angle of the inverter-side ac voltage
nitude at the rectifier PCC. are changed by 5% and 10 , respectively. Such changes may
The dynamic phasor model follows the system response in- occur due to remote faults and while they disturb the operation
cluding the sharp rise of the voltage at the inception of the step of the system, they are not severe enough to lead to commutation
change at 2.0 s. The traces in Fig. 9 show conformity be- failure. As shown, the EMT and the dynamic phasor model cor-
tween the EMT and the dynamic phasor models of the system, rectly represent the dynamics of the response. The constant-ad-
except for the high-frequency ripple, which is ignored in the mittance matrix model, however, has considerable error during
averaging process. The constant-admittance matrix model does the transient period.
not accurately follow the transients. Fig. 12 shows the response of the models to a severe fault in
the inverter-side ac system. The fault is modeled with a 20%
C. Fault Analysis (Cases 3 and 4) reduction in the ac voltage magnitude. The EMT model clearly
Performance of the models in representing faults is assessed indicates commutation failure. Both the dynamic phasor and the
in two stages. In the first stage (Case 3), adequately remote constant-admittance matrix models have large error during the
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE TRANSACTIONS ON POWER DELIVERY

TABLE I
CIGRE HVDC BENCHMARK SYSTEM DATA

The simulations shown in this paper are conducted with a time

step of 50 s for the EMT model and 800 s for the dynamic
phasor and the constant-admittance matrix models. Since the
Fig. 11. Response of the system to a change in the inverter-side ac EMT model and the two average models are not implemented
voltage phase angle. top: rectifier current, bottom: inverter current (Case 3). in the same simulation platform, it is not directly possible to
compare their simulation speed. It is, however, expected that
the reduced computational intensity of the average models, due
to their much larger time step, will reduce their simulation time.
Note that use of a larger time step in an EMT simulation may
speed up, to an extent, its run time. It, however, must also be
noted that the integration methods used in EMT simulation may
produce erroneous or divergent results for adequately large time
steps. Therefore, achievement of simulation acceleration while
maintaining accuracy by using large time steps in EMT simula-
tion is not generally possible. The dynamic phasor model, how-
ever, has intrinsically lower details, which then allows the use
of a large time step while ensuring accuracy of representation
of the contents it represents.

VI. CONCLUSION

In this paper, the concept of dynamic phasors was utilized

to derive an average-value model of an LCC-HVDC system.
This model maintains the low-frequency content of the system
response and removes the high-frequency switching transients.
The model was shown to be accurate in predicting the transient
and steady-state large-signal response of the system over a fre-
quency range that well exceeds the fundamental frequency of
the variables of interest (e.g., Fig. 8 showed correct prediction of
Fig. 12. Response of the system to a 20% reduction in the inverter-side ac a 62-Hz oscillatory transient on the dc current). This expanded
voltage magnitude top: rectifier current, middle: inverter current, bottom: cur-
rent order (Case 4).
range is unattainable by constant admittance matrix models. The
dynamic phasor model is, therefore, deemed an acceptably ac-
curate and computationally inexpensive option for the analysis
of the dynamic behavior of an LCC-HVDC system and for the
transient period, although they converge to the correct steady-
study of interactions between ac and dc systems. The particular
state value. The reason for the discrepancy in the transient pe-
model developed falls short in accuracy during commutation
riod is that these two models are developed without provisions
failure, due to its underlying assumptions of symmetrical and
for operation of the converter under abnormal conditions, which
equidistant firing, based on which the switching functions are
occur during commutation failure.
designed. Additional modifications become necessary to equip
It is also important to investigate the benefits of dynamic
the model for study of commutation failure.
phasor averaging in accelerating the simulation. EMT simula-
tions normally require a small time-step in the order of 5–50
s, to correctly predict the high-frequency switching transients. APPENDIX A
Since these high-frequency components are neglected in dy- SYSTEM PARAMETERS
namic phasor modeling, it is expected that larger time steps
can be used, which lead to speed-ups in the simulation time. See Tables I and II.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

DARYABAK et al.: MODELING OF LCC-HVDC SYSTEMS USING DYNAMIC PHASORS 9

TABLE II REFERENCES
CONTROLLER PARAMETERS FOR THE CASE STUDY
[1] E. W. Kimbark, Direct Current Transmission. New York: Wiley,
1971, vol. 1.
[2] V. K. Sood, HVDC and FACTS Controllers – Applications of Static
Converters in Power Systems. Norwell, MA, USA: Kluwer, 2004.
[3] M. P. Bahrman and B. Johnson, “The ABCs of HVDC transmission
technologies,” IEEE Power Energy Mag., vol. 5, no. 2, pp. 32–44,
Mar./Apr. 2007.
[4] M. Faruque, Y. Zhang, and V. Dinavahi, “Detailed modeling of
CIGRE HVDC benchmark system using PSCAD/EMTDC and
PSB/SIMULINK,” IEEE Trans. Power Del., vol. 21, no. 1, pp.
378–387, Jan. 2006.
[5] A. Gole, S. Filizadeh, R. Menzies, and P. Wilson, “Optimization-en-
abled electromagnetic transient simulation,” IEEE Trans. Power Del.,
vol. 20, no. 1, pp. 512–518, Jan. 2005.
[6] V. K. Sood, V. Khatri, and H. Jin, “EMTP modeling of CIGRE bench-
mark based HVDC transmission system operating with weak ac sys-
tems,” presented at the IEEE Int. Conf. Power Electron., Drives Energy
Syst. Ind. Growth, New Delhi, India, 1996.
[7] S. Chiniforoosh, J. Jatskevich, A. Yazdani, V. Sood, V. Dinavahi, J.
Martinez, and A. Ramirez, “Definitions and applications of dynamic
average models for analysis of power systems,” IEEE Trans. Power
Del., vol. 25, no. 4, pp. 2655–2669, Oct. 2010.
[8] S. Henschel, “Analysis of Electromagnetic and Electromechanical
Power System Transients with Dynamic Phasors,” Ph.D., Dept. Elect.
Eng., Univ. British Columbia, , Vancouver, BC, Canada, 1999.
[9] S. R. Sanders, J. M. Noworolski, X. Z. Liu, and G. C. Verghese, “Gen-
eralized averaging method for power conversion circuits,” IEEE Trans.
Power Electron., vol. 6, no. 2, pp. 251–259, Apr. 1991.
[10] A. Yazdani and R. Iravani, “A generalized state-space averaged model
of the three-level NPC converter for systematic dc-voltage-balancer
and current-controller design,” IEEE Trans. Power Del., vol. 20, no. 1,
pt. 1, pp. 1105–1114, Apr. 2005.
[11] A. Stankovic, B. Lesieutre, and T. Aydin, “Modeling and analysis of
single-phase induction machines with dynamic phasors,” IEEE Trans.
Fig. 13. Current at one of the ac filter branches with (top) and without (bottom) Power Syst., vol. 14, no. 1, pp. 9–14, Feb. 1999.
considering the 11th- and 13th-order components of the dynamic phasor simu- [12] A. M. Stankovic, S. R. Sanders, and T. Aydin, “Dynamic phasors in
lation. modeling and analysis of unbalanced polyphase AC machines,” IEEE
Trans. Energy Convers., vol. 17, no. 1, pp. 107–113, Mar. 2002.
[13] A. M. Stankovic, B. C. Lesieutre, and T. Aydin, “Applications of gen-
eralized averaging to synchronous and induction machines,” in Proc.
APPENDIX B North Amer. Power Symp., 1996, pp. 277–282.
HIGH-ORDER HARMONICS [14] A. M. Stankovic and T. Aydin, “Analysis of asymmetrical faults in
power systems using dynamic phasors,” IEEE Trans. Power Syst., vol.
15, no. 3, pp. 1062–1068, Aug. 2000.
The modified equations for a simple RL circuit, including [15] T. H. Demiray, “Simulation of power system dynamics using dynamic
higher harmonics, can be obtained from (24) as follows: phasor models,” Ph.D. dissertation, Dept. Elect. Eng., Swiss Fed. Inst.
Technol.,, Zurich, Switzerland, 2008.
[16] P. Mattavelli, G. C. e. Verghese, and A. M. Stankovic, “Phasor dy-
namics of thyristor-controlled series capacitor systems,” IEEE Trans.
(B1) Power Syst., vol. 12, no. 3, pp. 1259–1267, Aug. 1997.
[17] P. C. Stefanov and A. M. Stankovic, “Modeling of UPFC operation
under unbalanced conditions with dynamic phasors,” IEEE Trans.
Expression (B2) shows the th-order component of the ac cur- Power Syst., vol. 17, no. 2, pp. 395–403, May 2002.
[18] P. Mattavelli, A. M. Stankovic, and G. C. Verghese, “SSR analysis with
rent of phase-a of a six-pulse converter which is obtained by dynamic phasor model of thyristor-controlled series capacitor,” IEEE
applying (2)–(15) Trans. Power Syst., vol. 14, no. 1, pp. 200–208, Feb. 1999.
[19] V. A. Caliskan, O. C. Verghese, and A. M. Stankovic, “Multifrequency
averaging of dc-dc converters,” IEEE Trans. Power Electron., vol. 14,
no. 1, pp. 124–133, Jan. 1999.
[20] A. Emadi, “Modeling and analysis of multiconverter DC power elec-
tronic systems using the generalized state-space averaging method,”
IEEE Trans. Ind. Electron., vol. 51, pp. 661–668, Jun. 2004.
(B2) [21] A. Emadi, “Modeling of power electronics loads in AC distribution
systems using the generalized state-space averaging method,” IEEE
Trans. Ind. Electron., vol. 51, no. 3, pp. 992–1000, Oct. 2004.
where and is the phase angle of phase-a [22] S. Filizadeh, A. M. Gole, D. A. Woodford, and G. D. Irwin, “An opti-
of the commutating voltage. mization-enabled electromagnetic transient simulation-based method-
Fig. 13 shows the current at one of the ac filter branches with ology for HVDC controller design,” IEEE Trans. Power Del., vol. 22,
no. 4, pp. 2559–2566, Oct. 2007.
and without considering the 11th- and 13th-order components [23] H. Duchen, M. Lagerkvist, R. Kuffel, and R. Wierckx, “HVDC sim-
of the dynamic phasor simulation. In this case, a step change is ulation and control system testing using a real-time digital simulator
applied to the dc current order of the rectifier from 1 to 0.8 p.u. (RTDS),” in Proc.1st Int. Conf. Digital Power Syst. Simulators, 1995,
pp. 213–218.
while the inverter is modeled as a constant dc source of 500 kV. [24] M. Szechtman, T. Wess, and C. V. Thio, “First benchmark model for
HVDC control studies,” Electra, vol. 135, pp. 55–75, Apr. 1991.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

10 IEEE TRANSACTIONS ON POWER DELIVERY

[25] C. Karawita and U. Annakkage, “Multi-in-feed HVDC interaction [30] C. Osauskas and A. Wood, “Small-signal dynamic modeling of HVDC
studies using small signal stability assessment,” IEEE Trans. Power systems,” IEEE Trans. Power Del., vol. 18, no. 1, pp. 220–225, Jan.
Del., vol. 24, no. 2, pp. 910–918, Apr. 2009. 2003.
[26] P. C. Krause, O. Wasynczuk, and S. D. Sudhoff, Analysis of Electric [31] K. R. Padiyar, HVDC Power Transmission Systems: Technology and
Machinery and Drive Systems. Piscataway, NJ, USA: IEEE Press/ System Interactions. New York, USA: Wiley, 1990.
Wiley, 2002. [32] “PSCAD/EMTDC User’s Manual,” Manitoba HVDC Research Ctr.,
[27] W. Hammer, “Dynamic Modeling of Line and Capacitor Commutated Winnipeg, MB, Canada, 2010.
Converters for HVDC Power Transmission,” Ph.D., Dept. Elect. Eng., [33] J. L. Durate, A. Van. Zwam, C. Wijnands, and A. Vandenput, “Ref-
Swiss Federal Inst. Technol., Zurich, Switzerland, 2003. erence frames fot for controlling PWM rectifiers,” IEEE Trans. Ind.
[28] V. K. Sood, V. Khatri, and H. Jin, “Performance assessment using Electron., vol. 46, no. 6, pp. 628–630, Jun. 1999.
EMTP of two gate firing units for HVDC converters operating with [34] L. Gyugyi and B. R. Pelly, Static Power Frequency Changers. –
weak AC systems,” in Proc. Int. Conf. Power System Transients, Theory, Performance, and Application. New York, USA: Wiley,
Lisbon, Portugal, Sep. 1995, pp. 517–522. 1976.
[29] M. Parniani and M. R. Iravani, “Computer analysis of small signal sta-
bility of power systems including network dynamics,” Proc. Inst. Elect.
Eng., Gen., Transm. Distrib., vol. 142, no. 6, pp. 613–617, Nov. 1995.