This paper is a postprint of a paper submitted to and accepted for publication in

IET Generation, Transmission & Distribution and is subject to Institution of Engineering

and Technology Copyright. The copy of record is available at IET Digital Library

Transient stability of power systems with ISSN 1751-8644

doi: 0000000000
www.ietdl.org
embedded VSC-HVDC links: Stability
margins analysis and Control
Juan Carlos Gonzalez-Torres1 , Gilney Damm2 , Valentin Costan3 , Abdelkrim Benchaib1 , Françoise
Lamnabhi-Lagarrigue2
1
SuperGrid Institute, 69100 Villeurbane, France
2
Laboratoire des signaux et systèmes (L2S), 91190 Gif-surYvette, France
3
EDF Lab Paris-Saclay, 91120 Palaiseau, France
* E-mail: [email protected]

Abstract: This paper investigates the impact of embedded Voltage Source Converter-based High Voltage Direct Current (VSC-
HVDC) links on AC grids transient stability. Firstly, using Transient Energy Functions (TEF), it is demonstrated that VSC-HVDC
links controlled to track constant power references, do not inherently improve transient stability of the surrounding AC grid as an
AC line naturally does. Then, a control law using the feedback linearization technique on a simple but representative power system
is derived. The control law highlights and combines the three main actions the VSC-HVDC link can offer to enhance rotor angle
stability: fast power reallocation, injection of synchronising power and injection of damping power. The control law is implemented
and validated in EMT simulation. It is then shown that an HVDC link can assure the synchronisation of different AC areas even if no
AC transmission lines interconnect them. Through another case study, it is shown how the HVDC link can help to share dynamic
frequency reserves in order to not jeopardise the stability of the system. A last example investigates the effect of a DC fault on AC
transient stability and how the control can help improving the system response.

1 Introduction • Providing transient stability limits: approximations of the Region

Of Attraction (ROA) of the power system, i.e. how far the system can
The structure and operation of electric transmission systems is evolv- move away from its equilibrium point without losing synchronism.
ing faster than ever. The introduction of renewable energies in • Providing transient stability margins: indicators of how close a
centralised facilities often located far from end-users, as well as system is to the transient stability limit following a given distur-
the disconnection of conventional generation units, pose new chal- bance.
lenges for system operators. Because of these trends, the transport
of bulk-power over long distances has become a challenging neces-
sity. Indeed, if large amounts of power are transported over large In power system engineering, the most common practice is to
AC corridors, the security of the system can be at risk [1]. A pos- quantify these limits and margins in terms of the Critical Clearing
sible solution to assure the system security under these conditions Time (CCT) for a given disturbance. The CCT can be computed by
is the reinforcement of the transmission system. To this end, the iterative numerical solution of the systems’ equations. Alternatively,
High Voltage Direct Current (HVDC) technology has raised not only the so-called Direct Methods, discussed in the classical literature of
as a cost-effective solution for long distance transmission but also power systems [4–7], give the possibility to analytically approximate
presents other technical advantages over AC lines. When a system the region of attraction of a system. These methods are based on the
is reinforced with an HVDC line based on Voltage Source Convert- use of the Transient Energy Functions (TEF), and they aim to pro-
ers (VSC-HVDC), the fast and independent control of active and vide information about the stability limits and margins in terms of
reactive power gives the possibility to use the VSC-HVDC link to energy. Although conservative, these methods can assess stability
enhance the stability of the surrounding AC grid. In fact, dynamic limits independently from the experienced disturbance and without
stability studies have been performed over many decades showing solution of the equations (which is not the case for the CCT). There-
that power modulation of HVDC links can enhance rotor angle sta- fore, to have more general conclusions, Direct Methods are used
bility. The injection of active and reactive power by HVDC systems in this paper to compare the stability margins of an AC transmis-
allows to damp power oscillations as well as to counteract power sion system against its equivalent AC/DC system. The comparison
imbalances, which can be the source of transient stability problems of transient stability margins is used to highlight the need of HVDC
in a multi-machine system when a disturbance occurs. Actually, as power modulation through supplementary controls in AC/DC grids.
defined in [2], it is already expected for the HVDC systems to partic- The study of supplementary controllers allowing to modulate
ipate on the stability enhancement of the surrounding system. This the control variables of HVDC links for AC stability enhancement
can be done by means of supplementary controllers such as the one started almost at the same time as the HVDC technology itself.
proposed in this paper. A survey on controllers for the particular use of power oscillation
Power system transient stability stands for the ability of the sys- damping can be found in [8]. Supplementary controllers of HVDC
tem to keep synchronism after being subjected to a large disturbance links found in literature mainly differ depending on:
such as a three-phase fault [3]. Its assessment requires the analysis of
the systems described in their non-linear nature, which can be com-
plex, especially when there is a large number of dynamic elements. • HVDC technology (LCC or VSC),
The main tasks of the Transient Stability Assessment (TSA) consist • measured feedback signals (phase, voltages, etc...),
in: • control objectives (damp oscillations, security improvement, volt-
age stabilisation, etc.. ),

IET Research Journals, pp. 1–12

c The Institution of Engineering and Technology 2020 1
• control input (active or reactive power, DC voltage, AC or DC studied. The proposed control law is based on three actions: damp-
current, etc..), ing power injection, synchronising power injection and feed-forward
• applied control technique, actions. In section 5, EMT simulations using detailed models are
• the use (or not) of Wide Area Measurement Systems (WAMS), performed. The advantages of the proposed control are empha-
• coordination of the HVDC link with other FACTS devices, sised, through the analysis of different contingencies, such as three
• topology of the AC/DC grid (embedded HVDC links or links phase-faults, uncoordinated fast frequency response and DC faults.
connecting two asynchronous areas)

2 AC/DC Power systems modelling for transient

For decades, the transient stability problem has been treated via stability studies
the control of the excitation of generators using the Automatic Volt-
age Regulators (AVR) and Power System Stabilisers (PSS). For this In this section, the Internal Node Model (INM) of a multi-
reason, in early development of point to point HVDC links, a natural machine/multi-converter power system is described. The INM
choice was to equip them with PSSs aiming the stability of the sur- allows to represent the system by a set of ordinary differential
rounding AC grid [9]. Another intuitive control scheme consisted in equations through reduction of the admittance matrix of the system
modulating the HVDC power proportionally to the speed deviation [27, 28]. For the derivation of the model, consider power system
of a nearby machine [10]. This early development led to the imple- composed of n generators and m HVDC stations interconnected via
mentation of controllers on existing projects based on measurements a transmission network described by the nodal admittance matrix Y.
done near one HVDC station [9, 11].
The introduction of WAMS allowed the implementation of con-
trol schemes using feedback signals measured at different locations
of the power grid. For example, in [12, 13] using Lyapunov’s the-
ory, it was proven that transient stability is enhanced by modulating
active power proportionally to the difference of frequency measured
at the PCC of both converters (also analysed in [14, 15]). This control
scheme was recently implemented in the western North Ameri-
can Power System on a real time simulation platform with actual
measurements from PMUs [16]. This results have been upgraded
and extrapolated for the proposition of supplementary controllers of
Multi-Terminal DC grids in [17]. Regarding the use of voltage phase
angles, two industrial applications have been reported to modulate
the power as a proportional function of the angle difference between
inverter and rectifier sides [18, 19]. Although this kind of control has
been tested on real applications, no dynamic analysis has been done.
Also, regarding the use of angles as feedback signal, recently a PSS- Fig. 1: Multi-machine/multi-converter power system
like control using angle difference as feedback signal was proposed
in [20] for designing an oscillation damping control.
The implementation of real WAMS in different power systems Synchronous generators are represented by the association of
[21] also gave value to the research done regarding advanced con- a complex voltage source representing the internal voltage of the
trollers, with more feedback signals and more control inputs. For 0 0
generator Ē 0 d,i = Ed,i ∠δi and the transient reactance Xd,i . The
instance, as for the coordination of different HVDC links, a cen- voltage V̄g,i denotes the voltage at the terminals of the i − th gener-
tralised scheme using a model predictive control was presented ator and V̄hvdc,k = Vhvdc,k ∠θhvdc,k is the voltage at the Point of
in [22]. There, it is proposed to use speed measurements of dif- Common Coupling (PCC) of the k − th converter. The equations of
ferent machines. A compilation of other results on supplementary the multi-machine system are given by:
controllers of HVDC links using WAMS can be found in [23].
Recently, promising tracks using the inner loops of the converter
for improving transient stability have been proposed. In [24], the δ̇i = ωi
authors show that controlling the inverter (rectifier) station in order (1)
Mi ω̇i = −KD,i ωi + Pm,i − Pe,i
to emulate the behaviour of a synchronous motor (generator) has
a favourable impact on transient stability. Furthermore, in [25, 26]
the authors design robust inner controllers of the converters taking where for the i − th generator (∀i = 1, ... , n), δi represents the
into account the dynamics of the surrounding AC grid. Such control rotor angle deviation with respect to a synchronous reference, ωi
techniques can be achieved only based on local measurements. is the rotor speed deviation with respect to the synchronous speed
In the cited works above, the advantage of synchronising power ωs , Mi is the inertia coefficient, Pm,i is the mechanical input power,
injections (proportional to phase angle deviations) is not highlighted. Pe,i is the electrical output power. In the classical model, the angle of
0
To the authors’ best knowledge, it has not been shown how the syn- the internal voltage Ed,i ∠δi represents also the rotor angle deviation
chronisation of different regions is possible through the control of δi of a synchronous generator with one pair of poles.
HVDC links even when no AC links are connecting those grids. The Regarding the converters, their dynamics are much faster when
present work brings answers to these points, and in addition, analy- compared with the electromechanical dynamics of the generators.
ses two different events poorly reported in the literature. The first one For the dynamics of interest, converters can be considered as
is the uncoordinated poor fast frequency response, and the second is complex current sources (I¯hvdc,k ) with independent active power
the study of the impact of DC faults on the overall system stability. (Phvdc,k ) and reactive power (Qhvdc,k ) injections. This assumption
The present paper is organised as follows. In section 2, a mod- is particularly valid when inner controllers designed for decoupling
elling technique allowing to represent mixed AC/DC systems as a the AC side and the DC side are used [29–33]. The controlled active
set of ordinary differential equations is presented. In section 3, a and reactive power injections can be divided in two components as
comparison between an AC system and an equivalent AC/DC sys- follows:
tem is performed using Direct Methods. This comparison shows that Phvdc,k = Phvdc0,k + ∆Phvdc,k
HVDC systems with constant power references do not intrinsically (2)
Qhvdc,k = Qhvdc0,k + ∆Qhvdc,k
increase transient stability margins as an AC line does. Therefore, the
need of a supplementary control loop for the HVDC link to enhance where Phvdc0,k and Qhvdc0,k are the set-points determined by
the AC stability is highlighted. In section 4, feedback linearization the operator of the system and ∆Phvdc,k and ∆Qhvdc,k are the
technique is used to derive a control law, and its physical meaning is supplementary control references of the converters.

IET Research Journals, pp. 1–12

2
c The Institution of Engineering and Technology 2020
 The impact of the active and reactive power injections on the gen- where ahvdc
ik and bhvdc
ik are called "distribution factors" and they are
erators’ dynamics is given by the development of the electrical out-
computed as follows:
put power term (Pei ) in Equation (1). Using Kirchoff’s law, we have
that the relation between currents and voltages through the admit- 0
Ed,i
tance matrix including the transient reactances of the synchronous ahvdc =
 hvdc
Fik hvdc
cos(δi − θhvdc,k ) + Dik sin(δi − θhvdc,k )

ik
generators Ybus is given by: Vhvdc,k
0
Ed,i
bhvdc hvdc hvdc
 
Ybus ik = Dik cos(δi − θhvdc,k ) − Fik sin(δi − θhvdc,k )
Vhvdc,k
z }| {  (10)
E0 d
  
Ig YA YB
= (3)
Ihvdc YC YD Vhvdc The same procedure can be performed in order to include loads
with constant power characteristic [4] as further done in this paper.
where YA ,YB , YC and YD are of dimensions (n × n), (n × m), Finally, Equations (1) and 9 can be rewritten to obtain the state space
(m × n) and (m × m) respectively. Ig and E0 d respectively contain model of the AC power system with HVDC power injections used in
the injected currents and internal voltages of generators. Ihvdc and this paper for the derivation of the control law, which is given by:
Vhvdc are the injected currents and voltages at AC bus-bars of the  
converters [10]. We can solve for the generator’s currents as follows: ẋ = δ̇ = f (δ, ω) + [BP BQ ] u (11)
ω̇
Ig = YSN E0 d + Khvdc Ihvdc (4) where

−1 −1 δ = [δ1 . . . δn ]T , ω = [ω1 . . . ωn ]T
with YSN = (YA − YB YD YC ) and Khvdc = YB YD . Matrix
YSN is the matrix relating all generators voltages and currents, u = [∆Phvdc,1 . . . ∆Phvdc,k ∆Qhvdc,1 . . . ∆Qhvdc,k ]T
Khvdc describes the effect of HVDC injected currents on the
generator. Elements of YSN and Khvdc are:
 ω1 
 .
.. 
 
hvdc hvdc 
ωn

ȲSNij = Gij + jBij K̄hvdc,ik = Dik
and + jFik  
(5) f (δ, ω) =  1 P

− K ω − P (δ)


M
 1 m,1 D,1 i gg,1 
Gij and Bij are the equivalent conductance and susceptance

 .. 
between generators i and j. Dik hvdc hvdc
and Fik are the real and
 . 
1


imaginary parts of the element Khvdc,ik which denote the effect Mn Pm,n − KD,n ωn − P gg,n (δ)
of the injected current of the k − th converter on the i − th genera-    
0n×k 0n×k
tor. Then, the expression of the active power supplied by the i − th  ahvdc ahvdc   bhvdc bhvdc 
machine Pe,i in Equation (1), is developed as:  11
M1
··· 1k
M1
  11
M1
··· 1k
M1

Bp =  , B =
   
.. .. ..  Q  .. .. .. 
   . . .   . . . 
n m
   
∗ ahvdc ahvdc bhvdc bhvdc
0 ∗ 0 ∗
I¯∗
n1 nk n1 nk
··· ···
X X
Pei = < Ēdi ȲSNij
Ē 0 dj − Ēdi K̄hvdc ik hvdck
 (6) Mn Mn Mn Mn
j=1 k=1

3 Impact of embedded HVDC on AC grid stability

Substituting (5) into (6), the electric power supplied by the i − th
generator is derived as: In this section, Direct Methods are used to assess the transient sta-
bility of the Single Machine Infinite Bus (SMIB) system in Figure 2.
n The case corresponds to an AC system that has been reinforced by
0 0
X
Pe,i = Ed,i Ed,j (Gij cos δij + Bij sin δij ) two different means: the system in figure 2a has been reinforced with
j=1 an AC line, while in the case represented in Figure 2b, an HVDC link
0
has been chosen for the reinforcement.
m
" #
X Ēd,i hvdc hvdc
+< (Dik + jFik )(Phvdc,k + jQhvdc,k )
V̄hvdc,k
k=1
(7)
0
with δij = δi − δj . The complex ratio of voltages (Ēdi /V̄hvdck )
can be approximated to constant values [10]. Following this assump-
tion, and applying Equations (2) into Equation (7), the electrical
power can be expressed as follows:
(a) AC SMIB (b) AC/DC SMIB
Pe,i = Pgg,i + ∆Pgv,i (8) Fig. 2: Comparison between: (a) AC SMIB and (b) AC/DC SMIB
with

All the AC lines are assumed to be loss-less lines of reactance

n X12 . The internal reactance of the generator is assumed negligi-
X 0 0 0
Pgg,i = Ed,i Ed,j (Gij cos δij + Bij sin δij ) ble against the reactance of the AC transmission line (i.e. Xd1 <<
j=1
X12 ). Using the methodology proposed in Section 2, the dynamics
of both systems are described by:
m m
ahvdc bhvdc
X X
+ ik Phvdc0,k + ik Qhvdc0,k (9) • a) SMIB AC system:
k=1 k=1
m m Pe,1 (δ1 )
ahvdc bhvdc
X X
∆Pgv,i = ik ∆Phvdc,k + ik ∆Qhvdc,k z }| {
k=1 k=1 M1 ω̇1 + kD,1 ω1 = Pm,1 − (2Pmax sin δ1 ) (12)

IET Research Journals, pp. 1–12

c The Institution of Engineering and Technology 2020 3
• b) SMIB AC/DC system: 3.2 Comparison via Energy functions

Pe,1 (δ1 ) For more general conclusions, the transient stability of both systems
z }|
{ can be compared by means of the Transient Energy Functions (TEF).
M1 ω̇1 + kD,1 ω1 = Pm,1 − Pmax sin δ1 + Phvdc,1 (13) These functions represent the sum of the kinetic and potential energy
of the system [35]. For a SMIB system represented by Equation (11),
0 the TEF can be found by computing the first integral of motion, as
with Pmax = Ed,1 V2 /X12 .
follows:

WKE WP E
3.1 The Equal Area Criterion z }| { z }| {
1 δ1
2
W (ω1 , δ1 ) = M1 ω1 + ∫ − [Pm,1 − Pe,1 (δ1 )] dδ1 (14)
The power-angle (P-δ) curve of both systems is depicted in Figure 3. 2 δs
This curve represents the power delivered by the generator as func-
tion of the rotor angle. By means of the Equal Area Criterion (EAC), where the total energy W is the sum of the kinetic and potential
the stability margins of both cases can be analysed [34]. energy noted WKE and WP E respectively. From Equation (14),
it must be noted that there is a relation between the Equal Area
Criterion and the potential energy function WP E . Indeed, WP E
represents the surface 1-6-7 on the equal area criterion in Figure 3.
The transient energy functions of the described systems under
comparison are the following:

• a) AC SMIB system:
1
W ac = M1 ω12
2
WPacE (δ1 ) (15)
z }| {
− Pm,1 (δ1 − δ s ) − 2Pmax (cos δ1 − cos δ s )
Fig. 3: Equal area criterion comparison • b) AC/DC SMIB system:
1
In this comparison, every transmission line (HVAC or HVDC) W dc = M1 ω12
2
in both systems transmits the same amount of power in the pre-fault (16)
steady-state situation. Therefore, the constant active power reference WPdcE (δ1 )
z }| {
for the HVDC link is set to be equal to the active power for one − (Pm,1 − Phvdc,1 ) (δ1 − δ s ) − Pmax (cos δ1 − cos δ s )
AC line in steady state (i.e. Phvdc1 = Pmax sin δ s ). This assump-
tion allows to have the same equilibrium points in both systems (the
same angle in steady state). In the plotted region systems have two where WPacE and WPdcE stand for the potential energy of the AC
equilibrium points, one being stable and the other unstable [1]. It SMIB and the AC/DC SMIB respectively. The evaluation of the time
is possible to note that, the stable equilibrium point (SEP) of both derivative of the total energy (W ac and W dc ) gives an estimation
systems is the same (δ1 = δ s ), as well as their unstable equilibrium on how the total energy decays when the system is subjected to a
point (UEP) (δ1 = δ u ). disturbance. The computation of the time derivative of Equations
Let us assume that a temporary three-phase fault occurs in bus 1. (15) and (16) yields (see [12] for more details):
When the fault occurs, the power delivered by the generator suddenly
changes from point 1 to 2 in Figure 3 (it goes to zero). During the Ẇ ac = Ẇ dc = −kD,1 ω12 (17)
fault, both systems accelerate following the same trajectory as the
machine’s delivered power is zero until the fault disappears. Due to This equation shows that the energy decays at a rate proportional
this acceleration, the angle increases from δs to δcl . During the time- to the damping coefficient (kD,1 > 0 ) and the square of the speed
span between fault occurrence and fault disappearance, the system deviation. It is also found that the rate of change of energy in both
acquires a kinetic energy proportional to surface 1-2-3-4. For the systems is the same. Therefore, for a more convenient comparison it
described fault, the acquired energy of both systems is the same (i.e. is proposed to analyse the potential energy on both systems. For this
AAC1 = ADC 1 ).
purpose, the potential energy functions WPacE and WPdcE are plotted
When the fault is cleared, the systems get back to their pre-fault alongside the power-angle curve of each system in Figure 4.
power-angle curve (trajectories go from points 3 to 5 in Figure 3).
The angle continues to deviate until the kinetic energy absorbed
during the fault is expended by transferring it to the grid [1]. The
systems move from 5 to 6, such that area AAC 2 (ADC
2 ) is equal
to area AAC1 (A DC
1 ). For the studied fault, it can be observed that
the maximum angle reached during the first swing in the AC/DC
DC AC
SMIB is larger than the one in the AC SMIB (i.e. δmax < δmax ).
This shows that the AC/DC SMIB is closer to instability for the
same fault. Furthermore, after fault clearing, the available energy for
decelerating the system is proportional to the surface 4-5-8, it can
be easily observed that in the AC/DC SMIB, this surface is smaller
than the surface in the AC SMIB case. This means that transient sta-
bility margins are larger in the AC SMIB than in the AC/DC SMIB.
This phenomenon is due to the nature of the AC lines, their synchro- Fig. 4: Potential energy and power-angle curve comparison.
nizing power injections (AC lines transmit power as function of the
transport angle) contribute to decrease the separation of areas. In the
AC/DC SMIB case, as the HVDC link is set in constant power mode, The evolution of the potential energy as function of the angle
its contribution against angle deviations is nonexistent. deviation shows that there is a local minimum in the SEP. Besides,

IET Research Journals, pp. 1–12

4
c The Institution of Engineering and Technology 2020
as described in [6], the Potential Energy Boundary Method states The vector field and the region of attraction of both systems are
that there is a local maximum in the UEP (δ1 = δ u , ω1 = 0) as obtained by numerical integration and are depicted in the phase plane
it is confirmed in Figure 4. According to the the Closest Unstable in Figure 5. The SEP is (δ1 = δ s =30o , ω1 =0), and the region of
Equilibrium Point method in [35], an approximation of the stability attraction is delimited by the thick line. Some trajectories starting
boundary can be made by finding the equi-potential contour crossing at different initial points (or points just after fault clearing) are also
the closest UEP. The potential energy value at this point is defined depicted. It can be observed that, when the post-fault trajectory starts
as the critical energy V cr . In both systems under study, V cr is equal inside the region of attraction, its energy is lower than the critical
to the potential energy at the UEP. The comparison of the local energy (W < V cr ) and the system remains stable, otherwise the
maximum values of potential energy of both systems shows that system will diverge. It can be also seen that, as damping has been
the critical energy is bigger in the AC SMIB than in the AC/DC neglected (KD,1 ), those trajectories continue to oscillate. As sug-
cr cr
SMIB (Vac > Vdc ). This means that for the systems under study gested by Equation (17) if the damping is neglected, the total energy
the transient stability margins are bigger in the AC SMIB case. (W ac and W dc ) is constant after fault clearing (equiscalar contour).
The same comparison can be done for all stable equilibrium On the contrary, if the damping coefficient is positive, the energy
points. In order to do so, the critical energy V cr as function of a dissipation will be positive leading the system to reach back the SEP.
given equilibrium point (δ1 = δ s , ω1 = 0) is expressed as follows:
3.3 Eigenvalue analysis
• a) Critical energy for the AC SMIB system:
If after the first swing, the described SMIB system remain stable,
cr s
Vac (δ ) = W ac (π − δ s , 0) = Pm,1 (2δ s − π) + 4Pmax cos δ s and the damping coefficient is positive, the systems undergo oscil-
(18) lations until they reach their equilibrium point. These oscillations
• b) Critical energy for the AC/DC SMIB (the power reference of in power systems are generally analysed using the Lyapunov’s first
the HVDC is constant and it is equal to Phvdc,1 = Pmax sin δ s ): method. Linearization of the system around its equilibrium point can
supply information about the frequency and damping factor of such
oscillations. Linearizing Equations (12) and (13) around an stable
cr s
Vdc (δ ) =W dc (π − δ s , 0) = (Pm,1 − Pmax sin δ s ) (2δ s − π) equilibrium point, the eigenvalues of the SMIB system can be found.
The damping factor ξ and the oscillation frequency f are found to
+ 2Pmax cos δ s
(19) be: s
KD,1 2
 
KD,1 Ks
Dividing Equations (18) and 19 by Pmax and eliminating the ξ= √ , f= − (21)
cr cr
2 Ks M1 M1 2M1
common terms, it can be shown that Vac > Vdc when
where Ks is known in the literature as the synchronising power con-
s s s stant and, in this case, represents the slope of the power-angle curve.
2 cos δ > sin δ · (π − 2 δ ) (20) For both systems it can be calculated as follows:

which is true for all stable equilibrium points of interest (i.e. 0 < • a) SMIB AC system
δ s < π/2). Consequently it can be stated that, in all the stable equi-
librium points (0 < δ s < π/2), the critical energy of the AC SMIB Ks = KsAC = 2Pmax cos δ s (22)
system is higher than the critical energy of the AC/DC SMIB. Fur-
ther analysis can show that, for a given equilibrium point (0, δ s ), • b) SMIB AC/DC system (the power reference of the HVDC is
all positive values of Phvdc1 will result in lower values of critical constant and equal to Phvdc,1 = Pmax sin δ s )
energy than its equivalent AC SMIB system.
Ks = KsDC = Pmax cos δ s (23)

The synchronising power constant from case (a) KsAC is bigger than
the one in case (b) KsDC for all the stable values of δ s . Conse-
quently, the damping coefficient is higher for the hybrid AC/DC case
than in the pure AC system.
It is possible to conclude, that HVDC links with constant ref-
erences can damp power oscillations in a mixed AC/DC system,
with respect to their equivalent AC system; as it has been actually
observed by power system operators (e.g. [36]). However it doesn’t
mean that transient stability is enhanced, as it was previously shown
in this Section.
(a) AC SMIB
4 Linearizing feedback control
In order to enhance the stability margins of the AC network, the
active power references of the VSC-HVDC are modulated based on
feedback linearization techniques, such as to actively support the
AC grid. The idea behind the feedback linearisation technique is
to compensate the nonlinearities of the system by a control signal,
produced by feedback of measured variables and knowledge of the
system [37]. The obtained system, with reduced nonlinearities is fur-
ther controlled applying linear control techniques for deciding the
dynamic behaviour of the system. The proposed control is used to
(b) AC/DC SMIB derive a supplementary control scheme which computes the neces-
sary active power modulation for the power-mode converter allowing
Fig. 5: Comparison of the regions of attraction. Characteristics
to improve the rotor angle stability of the surrounding AC grid, damp
of the system: δ s = 30, Pm,1 = 1pu , Pmax = 1pu, Phvdc,1 =
the electro-mechanical oscillations and to ensure the synchronisation
0.5pu, kD,1 = 0, M = 0.12MW s2 /MVA rad (i.e. H1 = 4s ). of two AC interconnected networks.

IET Research Journals, pp. 1–12

c The Institution of Engineering and Technology 2020 5
 where ulf is the linearizing feedback part of the control law and v
Phvdc,1 + jQhvdc,1 −Phvdc,1 +jQhvdc,2
introduces the linear dynamics that are to be imposed to the system.
G1 3 4 G2 For the system represented in Equation (25) the linearizing control
0 0
Ed,1 MMC1 MMC2 Ed,2 law is given by:
2Xac
0 0
Xd,1 T1 T2 Xd,2
1 2
M1 +M2 E 0 d1 E 0 d2
h
M1 M2
u = −Phvdc0,1 − ahvdc M hvdc M2 M1 M2 Xeq sin δ12 )
1 1 +a2

2Xac
L L L L
a a a a
+ PMm1 − PMm2

1 2
− ( M111 − M212 )PL1 + ( M121 − M222 )PL2 − v
(28)
Pload,1 + jQload,1 Pload,2 + jQload,2 Note that if the internal reactance of the machines and the trans-
0
former are small compared to the AC line (Xd1 << Xac ), then
1
Fig. 6: 2-Area 2-Machines AC/DC power system L L hvdc hvdc
a21 = a12 = 0 and a2 = a2 = 1, and the system is found
to be described by:
Pm1 −PL1
Following the procedure described in Section 2, and neglect- δ̈12 = M1 − Pm2M−P
2
L1

ing damping effects, the dynamic equations of the power system M1 +M2 E 0 d1 E 0 d2
depicted in Figure 6 are given by: − M1 M2 (Phvdc0,1 + Xac sin(δ12 ) + ∆Phvdc1 )
(29)
  0
E 0 d2
 Then the linearizing control law is:
1
M1
Pm1 − E d1 Xeq
sin(δ1 − δ2 )
E 0 d1 E 0 d2
  
 L aL hvdc
+ ahvdc u = −Phvdc0,1 − sin(δ12 )
  −a11 PL1 − 12 PL2 − (a11
12 )Phvdc0,1

 Xeq (30)
ω̇1   M1 M2 Pm1 −PL1
+M ( M1 − Pm2 −PL1
)− M M1 M2
v
0
E 0 d2 1 +M2 M2 1 +M2
  
ω̇2   1
Pm2 − E d1 sin(δ2 − δ1 ) 
  = M2 Xeq 
 δ̇1   L   The first part of the control law compensates the nonlinearities of
−a21 PL1 − aL hvdc
22 PL2 + (a21 + ahvdc
22 )Phvdc0,1

δ̇2 


 the original system, while the term v is used to define the desired

 ω1 
 closed loop dynamics. Then using linear control techniques, v can
be designed to impose the electromechanical mode between two
ω2 regions [39]:
 
hvdc  δ̄12 − δ12
1
+ ahvdc
 
M1 (a11 12 )

v = −Kδ −Kω (31)
− 1 (ahvdc + ahvdc ) ω12
+
 M2
21 22  ∆Phvdc
1
0 
where Kδ and Kω are control gains. Injecting the linear control law
0 (31) and the feedback linearizing control law (28) into the system
(24) (25), the dynamics of the closed-loop are found to be:
where ∆Phvdc1 stands for the control input and it represents the
supplementary power injection of the VSC-HVDC link and coef- δ̈12 = −Kω δ̇12 − Kδ (δ̄12 − δ12 ) (32)
ficients aL
il are the distribution factors expressing the impact of
the l − th load on the i − th generator. The equivalent reactance with δ̇12 = ω12 . The control gains can be designed using the pole
is the sum of the reactances between generator G1 and G2 (i.e. placement technique. For this case, the gains can be chosen as func-
0 0
Xeq = Xd1 + XT 1 + Xac + XT 2 + Xd2 ). Generators are said to tion of the desired natural frequency (ωn ) and damping coefficient
be in synchronous equilibrium if all angular distances δi − δj are (ξ) of the controlled system, as follows:
bounded and all frequencies are identical ωi = ωj , therefore the
objective of the control is to make ω1 − ω2 = 0. The error between 2
Kδ = ωn and Kω = 2ξωn (33)
the two regions can be modelled following the change of variables
ω1 − ω2 = ω12 and δ1 − δ2 = δ12 as: Since the system has been linearized by the control action, it can
be concluded that the region of attraction of the controlled system is
M1 + M2 E 0 d1 E 0 d2 theoretically determined only by the power limits (rated power) of
δ̈12 = − sin(δ12 ) the VSC-HVDC link.
M1 M2 Xeq
1 1
+ Pm1 − Pm2 4.1 Physical interpretation of the control actions
M1 M2
aL
11 aL aL aL The control law in (30) is composed of three main actions:
−( − 21 )PL1 + ( 12 − 22 )PL2 (25)
M1 M2 M1 M2
• The first action (noted ulf in Equation (27)) allows to compensate
ahvdc
1 ahvdc for nonlinearities of the system. This action can also work as a feed-
+( + 2 )Phvdc0,1
M1 M2 forward action that compensates disturbances, especially if some of
ahvdc ahvdc those values are measured (e.g. the power flowing through the par-
1
+( + 2 )∆Phvdc0,1 allel AC transmission line). Indeed, from the control law (27) it can
M1 M2 be remarked that the term f (δ12 ) corresponds to the actual power
with δ̇12 = ω12 . It can be observed that using Equation (25), the flowing through the AC corridor, therefore the term c corresponds to
system is now written in strict-feedback form (see [38]), where: the scheduled power to flow through such AC corridor. This prop-
erty can highly simplify the practical implementation of this control
action.
δ̈12 = f (δ12 ) + c + b u (26)
• An action injecting power proportionally to the speed deviation of
both regions (see Equation (31)), contributing to the damping torque
with u the control input, f (δ12 ) is the nonlinear function of the angle of the system.
difference, and c is a constant making its derivative equal to zero in • An action injecting power proportionally to the transport angle
steady state (i.e. δ̈12 = δ̇12 = 0). The linearizing feedback control between both regions (see Equation (31)), contributing to the syn-
law can be computed as follows: chronising torque. As it will be shown in the examples, this action
allows to synchronise both regions even if the parallel AC lines are
u = ulf + v = −b−1 (f (δ12 ) + c + v) (27) tripped.

IET Research Journals, pp. 1–12

6
c The Institution of Engineering and Technology 2020
5 Test of the proposed control under different
disturbances
In this section, the proposed control scheme is applied to two differ-
ent systems using the electromagnetic transient software EMTP-RV.
The interest of using an EMT software is to take into account the
dynamics of the Phasor Measurement Units (PMU), detailed con-
trols of the converters, as well as studying the behaviour of the
control in case of a DC fault. The HVDC stations in the link are (a) Frequency difference between both areas
two 500MVA Half-bridge Modular Multilevel Converters linked by
a 200km wide-band DC cable model, the implemented EMT models
are described in [40, 41].
In the first two examples of this section, the simulated power
system corresponds to a two-area power system connected by two
200km AC lines in parallel with a VSC-HVDC link as shown in
Figure 6. In the simulated system, two 900MVA generators equipped
with AVR and PSS are implemented. Their parameters are the ones
described in [1]. The load in the first area is 300MW, while in the
second area the load is 800MW, so the power flowing through the (b) Transport angle between both areas
AC/DC corridor is 500MW. The implemented supplementary con-
trol is depicted in Figure 7. By means of the PMUs, phase angle
(δi ) and electrical pulsation (ωi ) at the PCC of both converters are
measured. The communication system between the PMUs and the
controller is represented by a delay. In this article it was considered
a delay of 60ms [42]. The reference angle between both stations
δ̄12 and the scheduled power exchanged are signals received from
a higher level controller.
(c) Power transferred through the HVDC link

Fig. 8: Comparison of different control schemes after a 250ms three-

phase fault on one AC line followed by the line tripping

the power flowing through the tripped AC line before the fault, is
reallocated on the remaining AC line. This is reflected in the incre-
ment of the angle in the post-fault steady-state compared with the
pre-fault value (35o before the fault against 58o after the fault). The
NFL control, by taking into account the disturbance in the system,
will rapidly compensate the power imbalance due to the fault. This
rapid action will limit the angle deviation between regions during
Fig. 7: Control implementation the first swing. After the fault, if power is available, the power of
the tripped AC line is reallocated to the HVDC link. This action will
take the transport angle near its initial condition. As a consequence,
stability margins of the post-fault configuration are similar as in the
initial configuration.
5.1 3-phase fault and trip of the faulted line
In order to quantify the transient stability margins for each strat-
The first case study corresponds to a 250ms 3-phase fault occur- egy, the Critical Clearing Time for two different fault locations are
ring in the middle of the AC line at t = 1s. Three different control computed by iterative simulations and are summarised in Table 1.
strategies are compared: In every case, the fault is cleared by the tripping of the faulty line. It
can be observed that both supplementary controllers (POD and NLF)
• CR (Constant references): The power reference of the converter is increase the stability limits with respect to the CR strategy. Besides,
maintained constant. it is also shown that the stability margins are higher when the NLF
• POD (Power oscillation damping): The control proposed in [13] strategy is implemented.
is implemented. This controller modulates the power reference as a
function of the frequency deviation in both PCCs of the HVDC link Table 1 Critical clearing times for different fault locations
(∆Phvdc = K(ω1 − ω2 )).
• NLF (Nonlinear Feedback): The proposed control is implemented CCT(ms)
Case
and tuned to obtain an inter-area oscillation with a natural frequency CR POD NLF
of ωn = 0.5Hz and a damping factor of ξ = 0.5. The gains are Kδ
= 1.7MW/deg and Kω = 197MW/Hz Fault in the Middle of the line 3-4 290 355 460
Fault near bus 3 120 220 335
In Figure 8, it is seen that in the CR case, angle deviations dur-
ing the first swing are higher compared with the other strategies.
This means that the AC system is closer to lose transient stability (or
power system split). With the CR strategy, it is also clearly observed 5.2 Synchronisation in case of the disconnection of the AC
that inter-area oscillations are poorly damped. When the POD strat- parallel lines
egy is implemented, it can be seen that the VSC-HVDC link is able
to damp power oscillations via the modulation of the active power. In this scenario, the disconnection of the whole group of AC trans-
It is also observed that the angle deviation during the first swing has mission lines connecting both areas after a fault is studied. Let us
been decreased. In the post-fault equilibrium point, the power refer- consider that after a fault on the AC transmission lines, both of them
ence of the HVDC goes back to its initial reference. In other words, are tripped, then no AC lines are interconnecting both areas. In order

IET Research Journals, pp. 1–12

c The Institution of Engineering and Technology 2020 7
to preserve the power exchanged between areas, a standard solution are connected in the first region in order to provide fast frequency
is to reallocate the power transmitted by the AC lines into the VSC- response (FFR).
HVDC link (similar to the fast ramp strategy studied in [43]). Power
reallocation would assure the right power exchange between both
regions, however the split of the system won’t be avoided. If both 7 8
700MW 700MW
regions are not completely synchronised, the reconnection of the AC
line cannot be done. 5

1 3
6
10
9
10km 11 10km
25km 25km

20/230kV

2 4
P=967MW P=2400MW
Q=-287MW Q=-437MW

Slack 700MW

Fig. 10: Modified version of the Klein-Rogers-Kundur’s two-area

(a) Speed deviation of both areas power system with power electronics interfaced energy sources

The procedure described in Section 4 is applied to this power sys-

tem, and it is chosen to derive a control law that linearizes the input
with respect to the error of frequency and angle between two regions
given by:
 
δCOI1 − δCOI2
x= (34)
ωCOI1 − ωCOI2
(b) Transport angle between both areas
where δCOI1 and δCOI2 are the angles of the centre of inertia
(COI) of Area 1 and Area 2 respectively, ωCOI1 and ωCOI2 are
the frequencies of the COI of Areas 1 and 2.

p p
1 X 1 X
δCOIq = Mi δi , ωCOIq = Mi ω i
MT MT
i=1 i=1
p
(35)
X
(c) Power transferred through the HVDC link MT = Mi
i=1
Fig. 9: Comparison of different control schemes when the AC
transmission line is disconnected. where p is the number of generators of synchronous generators in
the Area q of interest.
5.3.1 Poorly coordinated frequency response: A result of
In this case study, two strategies are compared: POD and NFL. the increase in the penetration of renewable energies in a grid is the
Figure 9 shows that, after a few seconds, the POD control is able to loss of inertia. The lack of inertia in the grid will lead to a more
ensure the right power exchange between regions and takes the value volatile power system. Controllable devices such as HVDC trans-
of the speed difference ω12 to a stable value different from zero. This mission and Power electronics interface sources must take part in the
means that synchronism of both areas is not achieved. If the system stabilisation of the grid. As reported in [44], one possibility to do so,
are not synchronised it is not possible to reconnect the AC lines (a is to make the renewable sources participate into frequency control
reconnection attempt is shown at t=10s). by supplying fast frequency response. While the primary response of
The implementation of the NFL control shows that synchroni- synchronous machines acts some seconds after the disturbance, the
sation of both areas can be achieved even if no physical AC line power electronics interfaced sources can supply a faster response.
is connected. The proposed control scheme allows to automatically However, as mentioned in [44], this fast power injections can jeop-
replace the disconnected group of lines while damping power oscil- ardise system transient stability. In the case presented in this section,
lations, assuring power exchange and maintaining synchronism by it is shown how those fast power injections may lead to system loss
emulating the behaviour of a modifiable transmission line connect- of synchronism.
ing both areas. Besides, it can be seen that if the AC group of lines The evaluated contingency in this case study is the loss of gen-
is reconnected later, at t=10s, the system remains stable and goes erator G5 in Fig. 10. In order to describe the phenomena, two cases
back to its initial equilibrium point. Synchronisation is achieved are compared. In both cases, the point to point VSC-HVDC system
when ω12 = 0. If both areas are synchronised, the AC transmission is controlled to track constant power references.
line can be reconnected, even if the frequency of each region is not Stable case: No fast frequency response is injected by the
stabilised at 1pu. renewable sources after generator tripping, only the primary reserve
of the generators participates on frequency control. In Figure 11a, the
5.3 Split of the system due to uncoordinated Frequency frequency of the COI of each region as well as the COI of the whole
Response system are represented. It can be seen that after generator tripping
the global frequency drops, and frequency of each areas oscillates
The system used in the last two cases correspond to a modified ver- around the global frequency. The system remains in synchronism
sion of the Klein-Rogers-Kundur’s two-area power system [1]. In and some seconds after the contingency, frequencies reach the same
this modified version the two areas are interconnected by a group steady state value.
of AC transmission lines and a VSC-HVDC link in bipolar con- Unstable case: Fast frequency response is injected. While gen-
figuration. In addition, power electronics interfaced power sources erator tripping occurs in zone two the fast frequency response is

IET Research Journals, pp. 1–12

8
c The Institution of Engineering and Technology 2020
 (a) Frequency of the COI of the system and both areas. (a) Frequency of the COI of the system and both areas.

(b) Injected dynamic frequency reserve (b) Frequency Containment Reserve.

(c) Angle difference (c) Angle difference between both regions

Fig. 11: System split caused by a poorly coordinated fast frequency

response.

(d) Injected active power by the HVDC link

injected in zone one, so the system will naturally transport the Fig. 12: Comparison of controllers when generator 5 is tripped and
injected power from zone 1 to zone 2 by increasing the transport dynamic frequency response is applied
angle between regions. FFR power injections will limit the drop of
the frequency on Area 1, while Area 2 has to wait the transport angle
to become larger in order to receive that increase of power. As shown
in Figure 11a, during the first hundred milliseconds, the FFR limits
5.4 Transient stability in case of DC fault
the frequency drop of Area 1 but not in Area 2, this will lead to
higher frequency difference between regions, thus a transport angle The impact of DC faults on transient stability as well as strategies to
beyond the stable limit and consequently to system split. minimise this impact, have been scarcely addressed in the literature
The HVDC link is a suitable mechanism to share the fast reserves [46]. Hence, this study case aims to show how transient stability can
while maintaining the synchronism of the system and damping be jeopardise in case a fault occurs in the DC side of the transmission
power oscillations. As done before, two controllers are studied, the line. In addition, we aim to show how the modulation of active and
POD and the NFL control. Figure 12 shows the system response reactive powers can be used to enhance transient stability in such
with different controllers. It can be observed that with both strate- cases. To do so, we can consider that a temporary pole-to-ground
gies, the synchronisation is assured and frequency oscillations are fault occurs on DC side of the negative pole of the first station.
well damped while the FFR response is correctly delivered by the All stations in this case are half-bridge Modular Multilevel Con-
power electronic interfaced source. verters. The protection strategy used for treating the DC fault is the
Comparing those results with the system response without sup- Converter-Breaker strategy presented in [45]. The different steps of
plementary controllers, it is observed that the stable frequency in the protection strategy shown in Figures 14 and 13 are summarised
post-fault operation is higher than in the case where no FFR is in the following table.
supplied (0.996 p.u. vs 0.994 p.u.). When no FFR is injected, the
minimum value of the global frequency is around 0.992 p.u. during
the first swing. This minimal value is improved when controllers are
implemented. Since the proposed control takes the angle difference Table 2 Description of the protection sequence.
as a feedback signal, in the equilibrium point after the disturbance,
the VSC-HVDC link transmits more power than in the initial equi- tn t [s] Description
librium point (250 MW vs. 200 MW). This means that the angle t0 1 Temporary Pole-to-ground fault on negative pole
difference between regions in the post fault equilibrium point is
t1 1.1 MMC deblocking and start of STATCOM operation
smaller when the NFL control is implemented, so stability mar-
gins of the post-fault situation are bigger when the NFL control is t2 1.35 DC Line reconnection and DC voltage recovery
implemented. t3 1.6 Beginning of active power restoration

IET Research Journals, pp. 1–12

c The Institution of Engineering and Technology 2020 9
 (a) Pos. pole active power (b) Neg. pole active power

(c) Pos. pole reactive power (d) Neg. pole reactive power

Fig. 13: Active and reactive power injections during execution of the converter breaker strategy triggered by a DC fault[45]

When the fault occurs (t0 ), the current flowing through the sta-
tion increases until the fault is detected and the DC circuit breakers
(DCCB) disconnect the faulted cable, and hence suppress the cur-
rent contribution to the fault. It should be noted that after fault and
before DCCB breaker opening, MMC converters are rapidly blocked
and the AC contribution to the fault passes through the freewheeling
diodes [47]. When the converter is blocked it exhibits the behaviour
of an uncontrolled rectifier. When the faulted cable is isolated, the
MMC can be deblocked (t1 ), therefore the faulted pole can operate
in STATCOM mode. As it can be seen in Figure 13d, while the MMC
is blocked (between t0 and t1 ), reactive power is zero, while after
deblocking the reactive power control is recovered. Before reconnec-
tion, the overhead line requires deionization, which in this simulation
(a) Speed deviation between zones
lasts 350ms (t2 ) after DC circuit breaker opening [46]. After this
time, the cable can be reconnected and the control of the DC voltage
is recovered. Then, the system is ready to return to operation (t3 )
and power is smoothly restored to its initial value.
For this example power references are set to 300 MW for the
VSC-HVDC link and the total exchanged power between 2 regions
is 500 MW. Figures 13 and 14 show the response of the DC and AC
systems respectively, in case of a DC fault. Three control strategies
are compare:

• Constant references: Active and reactive power references are

maintained constant (b) Angle difference between zones
• NFL- Qref constant: Active power references are modulated
using the proposed non-linear feedback control Fig. 14: AC System response after a DC fault and implementation
• NFL- ∆Qhvdc : Reactive power injections are taken into account of the converter breaker strategy [45]
in the linearizing feedback law

When no supplementary control is implemented, the temporary action of the controller includes the information about the power
power interruption in one pole leads to the loss of transient stabil- references (Phvdc0 ), thus it is possible to update the value of the
ity. In the case where NFL is implemented, the linearizing feedback static references including the loss of power of the faulted pole with

IET Research Journals, pp. 1–12

10
c The Institution of Engineering and Technology 2020
a negligible communication delay. In the second and third case, 5 Machowski, J., Kacejko, P., Nogal, Ł., Wancerz, M.: ‘Power system stability
this action compensates the lost pole with the power available in enhancement by WAMS-based supplementary control of multi-terminal HVDC
networks’, Control Engineering Practice, 2013, 21, (5), pp. 583–592
the healthy pole as it can be seen in Figure 13a. The healthy pole 6 Padiyar, K.: ‘Structure Preserving Energy Functions in Power Systems: Theory
goes from its initial value (150MW) to its maximum power (around and Applications’. (Boca Raton, Fla, US: CRC Press, 2016)
220MW). From the AC grid point-of-view, the power lost is only of 7 Anderson, P.M., Fouad, A.A.: ‘Power system control and stability’. (John Wiley
around 80 MW. This power compensation will reduce the effect of & Sons, 2008)
8 Elizondo, M.A., Fan, R., Kirkham, H., Ghosal, M., Wilches.Bernal, F., Schoen-
the contingency on transient stability. wald, D., et al.: ‘Interarea oscillation damping control using high-voltage dc
When the MMC station is deblocked after fault clearing, it is transmission: A survey’, IEEE Transactions on Power Systems, 2018, 33, (6),
possible to work in STATCOM mode before recovering the active pp. 6915–6923
power injections. As shown in Equations (11) reactive power injec- 9 Cresap, R.L., Mittelstadt, W.A.: ‘Small-signal modulation of the Pacific HVDC
intertie’, IEEE Transactions on Power Apparatus and Systems, 1976, 95, (2),
tions of the MMC stations have an impact on the active power of pp. 536–541
the synchronous machines. Therefore when the STATCOM opera- 10 Pai, M.A., Padiyar, K.R., Radhakrishna, C.: ‘Transient stability analysis of multi-
tion begins, then reactive power injections can be used as a control machine AC/DC power systems via energy-function method’, IEEE Transactions
input for enhancing transient stability in case of DC fault. In Figures on Power Apparatus and Systems, 1981, , (12), pp. 5027–5035
11 Hao, C. ‘Design of modulation controller to damp power oscillations of parallel AC
14 and 13c reactive power injections of the negative and positive line in the Tianshenqiao to Guangdong HVDC transmission’. In: Proc. TENCON
poles are shown. While the healthy pole compensates the power loss ’93. IEEE Region 10 International Conference on Computers, Communications
(between t1 and t2 in Figure 13a), the inner current control gives and Automation. vol. 5. (Beijing, China), 1993. pp. 244–247
priority to the active power injections, this is why reactive power 12 Latorre, H.F., Ghandhari, M., Soder, L. ‘Application of Control Lyapunov Func-
tions to Voltage Source Converters-based High Voltage Direct Current for Improv-
goes to zero (Figure 13c). However, after deblocking, reactive power ing Transient Stability’. In: Proc. 2007 IEEE Lausanne Power Tech. (Lausanne,
injections of the faulted pole can be modulated for reducing the Switzerland, . pp. 244–249
impact of the active power loss. Figure 14b shows that angle dif- 13 Latorre, H.F., Ghandhari, M., Söder, L.: ‘Active and reactive power control of a
ference between regions are slightly reduced when reactive power is VSC-HVDC’, Electric Power Systems Research, 2008, 78, (10), pp. 1756–1763
14 Sigrist, L., Echavarren, F., Rouco, L., Panciatici, P. ‘A fundamental study on the
modulated. The impact of reactive power injections depends on the impact of HVDC lines on transient stability of power systems’. In: Proc. 2015
distribution factors determined by the topological situation of the IEEE Eindhoven PowerTec. (Eindhoven, Netherlands), 2015. pp. 1–6
VSC-HVDC link. 15 Gonzalez-Torres, J.C., Mermet-Guyennet, J., Silvant, S., Benchaib, A. ‘Power sys-
tem stability enhancement via vsc-hvdc control using remote signals: Application
on the nordic 44-bus test system’. In: 15th IET International Conference on AC
and DC Power Transmission ACDC 2019. (Coventry, UK), 2019. pp. 1–6
6 Conclusions 16 Schoenwald, D.A., Pierre, B.J., Wilches.Bernal, F., Trudnowski, D.J. ‘Design and
implementation of a wide-area damping controller using high voltage DC modula-
In this article, the impact of embedded VSC-HVDC links on the tion and synchrophasor feedback’. In: Proc. 20th IFAC World Congress. (Toulouse,
France), 2017. pp. 67–72
surrounding AC grid is discussed and a supplementary control 17 Renedo, J., Garcia.Cerrada, A., Rouco, L.: ‘Active power control strategies for
for stability purposes has been developed. The proposed control transient stability enhancement of AC/DC grids with VSC-HVDC multi-terminal
has been tested for different contingencies. Important conclusions systems’, IEEE Trans Power Syst, 2016, 31, (6), pp. 4595–4604
should be taken from this article: 18 Henry, S., Despouys, O., Adapa, R., Barthold, L., Bayfield, C., Bell, K., et al.
‘Influence of embedded HVDC transmission on system security and AC network
performance’. (Paris, France): Cigré, 2013.
• Transient stability margins are lower when a system is reinforced 19 Francos, P.L., Verdugo, S.S., Álvarez, H.F., Guyomarch, S., Loncle, J.
with VSC-HVDC lines instead of AC lines. Indeed, VSC-HVDC ‘INELFE—Europe’s first integrated onshore HVDC interconnection’. In: Proc.
links controlled to track constant power references, do not inherently 2012 IEEE Power & Energy Society General Meeting. (San Diego, CA, USA), .
pp. 1–8
improve transient stability of the surrounding AC grid as an AC line 20 Xing, Y., Marinescu, B., Belhocine, M., Xavier, F. ‘Power Oscillations Damp-
naturally does. Hence, supplementary controllers on VSC-HVDC ing Controller for HVDC Inserted in Meshed AC Grids’. In: 2018 IEEE PES
links are needed for supporting AC transient stability. Innovative Smart Grid Technologies Conference Europe. (Sarajevo, Bosnia and
• Through the proposition of the control, three main actions for sup- Herzegovina), 2018. pp. 1–6
21 Aminifar, F., Fotuhi-Firuzabad, M., Safdarian, A., Davoudi, A., Shahidehpour, M.:
porting transient stability were identified: injection of synchronising ‘Synchrophasor measurement technology in power systems: Panorama and state-
power, injection of damping power and fast compensation of power of-the-art’, IEEE Access, 2014, 2, pp. 1607–1628
disturbances. 22 Fuchs, A., Imhof, M., Demiray, T., Morari, M.: ‘Stabilization of large power sys-
• Synchronisation of two power grids can be achieved through the tems using VSC-HVDC and model predictive control’, IEEE Transactions on
Power Delivery, 2014, 29, (1), pp. 480–488
proposed control even when no AC link is interconnecting both 23 Li, Y., Yang, D., Liu, F., Cao, Y., Rehtanz, C.: ‘Interconnected Power Systems:
grids. Wide-Area Dynamic Monitoring and Control Applications’. (Springer Berlin
• Poor Fast Frequency Response can destabilise the grid. Through Heidelberg, 2015)
the proposed control, the VSC-HVDC link can help the system to 24 Aouini, R., Marinescu, B., Kilani, K.B., Elleuch, M.: ‘Synchronverter-based emu-
lation and control of HVDC transmission’, IEEE Transactions on Power Systems,
properly share those reserves. 2016, 31, (1), pp. 278–286
• The effect of a DC fault on transient stability has been analysed. 25 Arioua, L., Marinescu, B.: ‘Robust grid-oriented control of high voltage DC links
The importance of fast power reallocation as well as reactive power embedded in an AC transmission system’, International Journal of Robust and
injections in case of a fault on the DC side of the converter has been Nonlinear Control, 2016, 26, (9), pp. 1944–1961
26 Arioua, L., Marinescu, B.: ‘Multivariable control with grid objectives of an HVDC
highlighted. link embedded in a large-scale AC grid’, International Journal of Electrical Power
& Energy Systems, 2015, 72, pp. 99–108
27 Ghandhari, M., Andersson, G., Hiskens, I.A.: ‘Control Lyapunov functions for
Acknowledgement controllable series devices’, IEEE Transactions on Power Systems, 2001, 16, (4),
pp. 689–694
28 Fan, X., Shu, J., Zhang, B.: ‘Coordinated control of dc grid and offshore wind
This work is supported by the French Government under the pro- farms to improve rotor-angle stability’, IEEE Transactions on Power Systems,
gram Investissements d’Avenir (ANE-ITE-002-01). 2018, 33, (4), pp. 4625–4633
29 Zama, A., Benchaib, A., Bacha, S., Frey, D., Silvant, S.: ‘High dynamics control
for MMC based on exact discrete-time model with experimental validation’, IEEE
Transactions on Power Delivery, 2018, 33, (1)
7 References 30 Chen, Y., Damm, G., Benchaib, A., Netto, M., Lamnabhi-Lagarrigue, F. ‘Control
induced explicit time-scale separation to attain DC voltage stability for a VSC-
1 Kundur, P., Balu, N.J., Lauby, M.G.: ‘Power system stability and control’. HVDC terminal’. In: Proc. 19th IFAC World Congress. vol. 19. (CapeTown, South
(McGraw-hill New York, 1994) Africa), 2014. pp. 540–545
2 ENTSO.E. ‘ENTSO-E Draft Network Code on High Voltage Direct Current 31 Chen, Y., Jiménez Carrizosa, M., Damm, G., Lamnabhi-Lagarrigue, F., Li, M., Li,
Connections and DC connected Power Park Modules’. (ENTSO-E, 2014). Y.: ‘Control-induced time-scale separation for multiterminal high-voltage direct
3 Kundur, P., Klein, M., Rogers, G.J., Zywno, M.S.: ‘Application of power sys- current systems using droop control’, IEEE Transactions on Control Systems
tem stabilizers for enhancement of overall system stability’, IEEE Transactions Technology, 2020, 28, (3), pp. 967–983
on Power Systems, 1989, 4, (2), pp. 614–626 32 González-Torres, I., Miranda-Vidales, H., Espinoza, J., Méndez-Barrios, C.F.,
4 Pai, M.A.: ‘Energy function analysis for power system stability’. (Boston, MA, González, M.: ‘State feedback control assisted by a gain scheduling scheme
US: Springer Science & Business Media, 2012)

IET Research Journals, pp. 1–12

c The Institution of Engineering and Technology 2020 11
 for three-level NPC VSC-HVDC transmission systems’, Electric Power Systems
Research, 2018, 157, pp. 227–237
33 González-Torres, I., Miranda-Vidales, H., Méndez-Barrios, C.F., Espinoza, J., Cár-
denas, V.: ‘Long-length horizons dynamic matrix predictive control for a MMC
inverter’, Electric Power Systems Research, 2019, 168, pp. 137–145
34 Gonzalez-Torres, J.C., Costan, V., Damm, G., Benchaib, A., Bertinato, A., Poul-
lain, S., et al.: ‘HVDC protection criteria for transient stability of AC systems with
embedded HVDC links’, The Journal of Engineering, 2018,
35 Chiang, H.D.: ‘Direct Methods for Stability Analysis of Electric Power Systems:
Theoretical Foundation, BCU Methodologies, and Applications’. (Wiley, 2010)
36 Breuer, W., Hartmann, V., Povh, D., Retzmann, D., Teltsch, E. ‘Application of
HVDC for large power system interconnections’. In: Proc. CIGRE Session. (Paris,
France), 2004. pp. B4–106
37 Slotine, J.J.E., Li, W., et al.: ‘Applied nonlinear control’. vol. 199. (Prentice hall
Englewood Cliffs, NJ, 1991)
38 Khalil, H.K.: ‘Nonlinear Systems’. Pearson Education. (Prentice hall Upper Saddle
River, NJ, 2002)
39 Gonzalez-Torres, J.C., Costan, V., Damm, G., Benchaib, A., Lamnabhi-Lagarrigue,
F., Luscan, B. ‘Procédé de commande d’un lien de transmission électrique incluant
une ligne haute tension continu’. (French patent), 2018. WO2019174999A1
40 Shinoda, K., Benchaib, A., Dai, J., Guillaud, X.: ‘Virtual capacitor control:
mitigation of DC voltage fluctuations in MMC-Based HVDC Systems’, IEEE
Transactions on Power Delivery, 2018, 33, (1), pp. 455–465
41 Loume, D., Tuan, M.N., Bertinato, A., Raison, B. ‘DC cable modelling and
High Voltage Direct Current grid grounding system’. In: Proc. 9th International
Conference on Insulated Power Cables. (Versailles, Fr), 2015. pp. 1–6
42 Cui, Y., Kavasseri, R.G., Chaudhuri, N.R. ‘Modeling and simulation of dynamic
communication latency and data aggregation for wide-area applications’. In: Proc.
2016 Workshop on Modeling and Simulation of Cyber-Physical Energy Systems
(MSCPES). (Vienna, Austria), . pp. 1–6
43 IEEE Committee Report: ‘Dynamic performance characteristics of North Ameri-
can HVDC systems for transient and dynamic stability evaluations’, IEEE Trans-
actions on Power Apparatus and Systems, 1981, PAS-100, (7), pp. 3356–3364
44 Eriksson, R., Modig, N., Elkington, K.: ‘Synthetic inertia versus fast frequency
response: a definition’, IET Renewable Power Generation, 2017, 12, (5), pp. 507–
514
45 Loume, D.S., Bertinato, A., Raison, B., Luscan, B. ‘A multi-vendor protection
strategy for HVDC grids based on low-speed DC circuit breakers’. In: Proc. 13th
IET International Conference on AC and DC Power Transmission (ACDC 2017).
(Manchester, UK), 2017. pp. 1–6
46 Tang, G., Xu, Z., Zhou, Y.: ‘Impacts of three MMC-HVDC configurations on AC
system stability under DC line faults’, IEEE Transactions on Power Systems, 2014,
29, (6), pp. 3030–3040
47 Bertinato, A., Gonzalez-Torres, J.C., Loume, D., Creusot, C., Luscan, B. ‘Devel-
opment of a protection strategy for future DC networks based on low-speed DC
circuit breakers’. In: Proc. CIGRE Session 2018. (Paris, France), 2018.

IET Research Journals, pp. 1–12

12
c The Institution of Engineering and Technology 2020