Ain Shams Engineering Journal 12 (2021) 2763–2779

Contents lists available at ScienceDirect

Ain Shams Engineering Journal

journal homepage: www.sciencedirect.com

Electrical Engineering

Study and analysis of voltage source converter control stability for HVDC
system using different control techniques
Dalia Rabie a,⇑, Tomonobu Senjyu b, Salem Alkhalaf c, Yahia S. Mohamed a, E.G. Shehata a
a
Department of Electrical Engineering, Minia University, El-Minia 61517, Egypt
b
Faculty of Engineering, University of the Ryukyus, 1 Senbaru, Nishihara-cho, Nakagami, Okinawa 903-70213, Japan
c
Department of Computer, College of Science and Arts in Ar-Rass, Qassim University, Ar Rass, Saudi Arabia

a r t i c l e i n f o a b s t r a c t

Article history: A Stable and highly reliable DC link voltage represents an important factor for efficient power transfer in
Received 8 October 2020 high voltage direct current (HVDC) networks. In this framework, this paper investigates the control and
Revised 15 December 2020 stability analysis of voltage source converter (VSC) for DC link voltage regulation. To separately achieve
Accepted 26 December 2020
the independent active and reactive power control, the system voltages and currents are represented in
Available online 27 March 2021
the synchronous reference frame. In order to optimally design the parameters of proportional-integral
(PI) controller, the inner and outer loops’ transfer functions are thoroughly derived/developed. In addi-
Keywords:
tion,/moreover, in order to attain satisfactory/certain system performance, modulus optimum, symmet-
Control techniques
HVDC
rical optimum pole placement control approaches are studied and implemented for the purpose of tuning
High impedance grid the voltage/current controller gain parameters. In particular for the symmetrical optimum control
Manual tuning method, the gain parameters of the DC-bus voltage are determined under different values of network
Modulus optimum impedance. Furthermore, the impact/influence of changing gain parameters on the DC-link voltage and
Symmetrical optimum poles/zeros movement are investigated. MATLAB/Simulink model is built and simulation studies are car-
PI controller ried out to verify the introduced concepts.
VSC Ó 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams Uni-
versity. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

1. Introduction power transmission system is required to has a high degree of reli-

ability and efficiently but these requirements do not available in
In the recent years, renewable energy sources (RES) including the AC transmission system, which is also not feasible for transmis-
hydropower, onshore/offshore wind farms and solar photovoltaic sion distances larger than 400 km for AC overhead lines due to high
(PV) power plants have grown rapidly. These resources are widely reactive currents and undesirable wave reflections [2]. High-
used and represent a reliable, economical and environmentally voltage direct current transmission (HVDC) is an efficient alterna-
friendly alternatives to the traditional means of electricity genera- tive to overcome these limitations.
tion (e.g. burning of uranium and fossil fuels. . .etc.). For example, The HVDC system is regarded as the best choice for long-
in addition to other environmental goals, the European Union has distance power transmission that integrates RES in the public net-
set a target of expanding its share of renewable energy to 27% by work and enables the promotes the use of RES in cost-effective and
2030 [1]. Therefore, the trend of RES is constantly increasing in efficient manner. In addition, by employing such modern technol-
the future, and wind energy is widely regarded as an major source ogy, the interconnection between the countries can be established
of large-scale renewable energy systems. In order to integrate and the electricity market can be supported and competed. A clas-
these renewable energy sources in the utility grid, the electrical sical technology of HVDC transmission system is represented in
line-commutated converter (LCC) based HVDC (LCC-HVDC) sys-
tems, which has been implemented using thyristor-based convert-
⇑ Corresponding author.
ers. In 1972, the first project with thyristor valves has been
E-mail address: Dalia.rabie5@gmail.com (D. Rabie).
established as a back-to-back connection, working at 80 kV with
Peer review under responsibility of Ain Shams University.
a power rating of 320 MW. This project has been known as Eel
River project in New Brunswick, Canada [3]. The LCC converter
has been modified into other configurations in some applications
such as capacitor-commutated converters (CCCs), in which series
Production and hosting by Elsevier
capacitors are attached to the converters between valves and

https://doi.org/10.1016/j.asej.2020.12.013
2090-4479/Ó 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Nomenclature

xb Base Angular Frequency (rad/s) T iv Integral Time Constant for the DC-Link Voltage Con-
Z dc;base Base DC Impedance (X) troller (s)
C dc
link DC-Link Capacitor (F) s Time Constant
Rdc
link DC-Link Resistance (X) pu Per Unit
V dc DC-Link Voltage (V) Ki Integral Gain
Gc;CL Closed-Loop Transfer Function for Inner Loop Controller K iv Integral Gain for the DC-Link Voltage Controller
Gv ;CL Closed-Loop Transfer Function for Outer Loop Controller Kp Proportional Gain
Ta PWM Time Delay (s) K pv Proportional Gain for the DC-Link Voltage Controller
Ti Integral Time Constant (s) f Damping Factor

transformers. This configuration has been served as a section of the called linear time-invariant model, in which the model parameters
commutation voltage to enhance voltage stability and provide depend on normal operating conditions. Therefore, the conven-
some of the converter reactive power compensation. Another con- tional feedback control strategies can provide adequate perfor-
figuration called controlled series capacitor converter (CSCC). mance, if the system is only mildly nonlinear or remain in
HVDC technology has a transition from conventional thyristor- vicinity of normal steady state, since the nonlinearity effect may
based HVDC that suffers from the commutation failure of thyristor be not severe. But if the system has a highly nonlinear behavior,
valves to voltage source converter-based HVDC (VSC-HVDC) due to and shifted to another operating point, conventional controllers
the development in power electronics technologies. In 1994, a may cause instability under certain circumstances, since its control
small scale HVDC system with two VSCs, based on IGBTs, has been parameters are identified through the one-point linearization of
put into operation [4]. VSC-HVDC makes the use of pulse width the original nonlinear system. Therefore, conventional control can’t
modulation (PWM) techniques easy. Such techniques enable the provide globally consistent control performance [11]. Conse-
generation of modulated voltage signals with predefined voltage quently, many advanced nonlinear controller design techniques
amplitude and phase angle instantaneously away from the AC have been proposed for two-terminal VSC-HVDC structure, which
system. result in linearized model independent of the operating point.
VSC-based HVDC system represents a worthy and growing These control techniques comprise perturbation observer-based
technology for remote integration of RES due to its key features sliding-mode control (POSMC) strategy which aggregates a combi-
in contrast to its traditional LCC-based HVDC counterpart. These natorial impact of nonlinearities, parameter uncertainties, unmod-
benefits include black start capability, management of active and elled dynamics and time-varying external disturbances. This
reactive power independently, and power reversal capability with- technique is implemented in [12] which confirm that POSMC can
out the need to change the polarity of DC voltage. In addition, a preserve a steady state control performance with less power over-
connection can be established with weak AC grids and passive net- shoot. input/output linearization [11], feedback linearization con-
works. One of the key factors that contributes to the grid weakness trol (FLC) literatures [13], and [9] investigated the feedback
is the high impedance characterization due to the remotely located linearization techniques based on sliding mode control (FLSMC),
RES [5], 2018 [6–8]. Such factor increases the challenges of stability in which the input–output linearization cancels the nonlinearities
and optimal operation of the power networks. Therefore, the con- to transform the converter system to a linear one and then a robust
trol and operation of converters connected to the grid are seen to control strategy can be utilized while considering the parametric
be a striking research direction. uncertainties. Moreover nonlinear adaptive control approach, valid
The successful design and operation of system controllers to solve uncertainties from system parameters, has been proposed
depend on two tasks, in which a tradeoff between the transient in a considerable amount of literature such as nonlinear adaptive
response and robustness must be considered. One of the two tasks robust control method [14], adaptive control design based on back-
is to determine the controller’s structure design, and the other one stepping method [15], and power-synchronization control method
is to adjust the controllers’ parameters that called ‘‘parameter tun- [16].
ing control” in order to provide an optimal system performance. Other controller techniques have also been introduced for con-
Vector current control is a widely used method in VSC-based HVDC trolling the multi-terminal VSC-HVDC system. These techniques
systems. The major objectives of the VSC vector current control are include simple Fuzzy self-tuning PI controller [17], passive control
to realize the independent active and reactive power control and design via energy shaping [18], and finally adaptive droop con-
also to preciously adjust the DC link voltage V dc to preserve a bal- troller approach [18]. Reference introduces an effectiveness com-
ance among the DC link power and power supplied to the network. parison of three different control under various operation
The converter stations of the VSC-HVDC transmission link have conditions. However, these undesirable issues that appears in con-
different inputs parameters, which their interaction creates indeed ventional adaptive control can be easily avoided by a proper tuning
nonlinear multiple input multiple output control system [9]. How- of control parameters. Therefore, when a highly nonlinear process
ever, this nonlinearity doesn’t refer to that the linear theory is use- is controlled by conventional PI controllers, the controllers must be
less. Many nonlinear systems can be described well by linear tuned very conservatively for stable behavior over entire range of
theory under certain conditions, which called a linearization [10]. operating condition. The tuning procedure is realized in a compro-
In conventional adaptive control strategies, in order to perform mising approach among speed and stability for undesirable distur-
modal analysis and control design, the system with nonlinear bances. In [19], three tuning techniques for PI control have been
dynamics is linearized by one-point linearization methods such discussed. In [20], modeling approach while achieving optimal PI
as Taylor series. The one-point linearization is done around an control of VSC-HVDC transmission system have been considered
operating point by keeping only the linear term in order to produce for system stability improvement. References [21–23] introduced
a linear system similar to the addressed nonlinear system behavior traditionally PI control utilizing the dq vector control approach.
within a limited operating range. This kind of systems is often Reference [24] applied the conventional vector control approach

2764
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

to different configurations of HVDC systems. In addition to that, The relationship between the input AC and output DC powers
authors in [19] used the symmetrical optimum method for tuning can be expressed by:
gain parameters of the DC-link voltage controller and considered
the performance of this method in the transient response only. 3

p¼ v d :id þ v q :iq ¼ v dc :idc ð3Þ
while, this current paper has considered the performance of this 2
method in both transient and steady state responses by comparing
the transient response of the symmetrical optimum method with where v dc represents the DC output voltage and idc represents the DC
that of pole placement method which its gain parameters have output current, that can be given by:
been tuned according to the desired response. While the steady
state response this method has been compared with the steady dV dc
idc ¼ C: þ IL ð4Þ
state of pole placement and the manual tuning method. dt
When the adjustment of the controllers’ gains has been the The q-axis component of the grid voltage is defined as v q ¼ 0
main factor for effective stable operation of the system under dif- due to the alignment of its d-axis component along the d-axis
ferent operating conditions [13], the current paper has been direction. Thus, the instantaneous real and reactive power are rep-
focused on studying the effect of gain parameters’ change on the resented by:
output DC voltage response and poles’/zeroes’ movement as well
as analyzing the VSC output system stability by different control p ¼ 32 :v d :id
ð5Þ
methods. The current paper is organized as follows: Section 2 pre-
q ¼ 32 :v d :iq
sents the system model, while the entire VSC control scheme is
considered in Section 3 including the current and DC-link voltage Therefore, by employing the transformation into rotating dq
control loops. The modulus optimum method is implemented for corrdinate the active component of current, which supplies the
tuning the current control loop, while the DC-voltage control loop essential power feed to the DC bus and reactive components of cur-
is tuned by symmetrical optimum criterion. Moreover, pole- rent, are utilized. Therefore, independent active and reactive power
placement method as well as manual tuning method are intro- control can be accomplished.
duced Section 4. Extensive simulations along with discussion are
presented in Section 5.
3. Schematic control system for VSC
2. System model analysis
The three-phase VSC is controlled by two control loops: inner
The under study HVDC system, illustrated in Fig. 1 consists of current control loop and outer voltage control loop as shown in
two three-level neutral-point clamped (NPC)-VSCs connected as a Fig. 3. In such control, the inner current control loop exhibits a fas-
back-to-back connection through DC-link cable. Depending on vec- ter response compared to that obtained from the outer DC-link
tor current control technique [25], the VSC analysis and control can voltage control loop due to the inherent slow nature of capacitor
be achieved by operating the system in dq axis which permits an voltage. Both of the inner and outer control loop are achieved by
independent active /reactive power control. The VSC control per- using the PI controller.
mits a regulating of the DC-link voltage V dc and maintaining it
fixed in order to keep the balance between the DC-link power
3.1. Inner control loop
and AC power. Fig. 2 shows a typical example of PWM VSC con-
nected to the AC grid.
As stated in (2), the currents in d-/q-axes become perfect con-
According to [26–27] and by applying standard Kirchhoff’s volt-
stant synchronized vectors with cross coupling terms of xL:iq
age law, the basic system behavior has been expressed as follows:
and xL:id . Hence, feed-forward is utilized to compensate the
d cross-coupling effects and providing an independent current con-
v abc ¼ R:iabc þ L iabc þ v abc;conv ð1Þ
trol. The PI regulators with gain parameters K p and T i are utilized
dt
where the measured three phase voltages and currents of the AC to regulate the resulted d and q independent current components.
grid are defined as v abc ; iabc ; v abc;conv referred to converter input volt- Then, the PI regulators output signals are introduced to the
ages; R and Lrepresent the filter resistance and inductance respec- PWM converter which have a time delay T a , defined by 1=ð2F sw Þ
tively between the AC grid and the VSC. According to [19,28–29]. and caused by VSC switches. Based on [30] the inner current con-
The converter three-phase voltages and currents are converted into troller has an open loop transfer function defined by:
constant vectors in the steady state with synchronous rotating    
1 þ T i :s 1 1 1
speed (x) at a specific fundamental frequency using the Clark and GC;OL ¼ K P ð6Þ
Park abc to dq transformations as follows: T i :s 1 þ T a :s R 1 þ s:s

L didtd ¼
Rid þ xLiq
v dconv þ v d where s, is defined by L=R and called the time constant. Fig. 4 intro-
ð2Þ
¼
Riq
xLid
v qconv þ v q duces the block diagram of the inner current controller loop in a per
di
L dtq
unit system [19].

Fig. 1. The system studied in a simplified diagram [24].

2765
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Fig. 2. Three-phase VSC system connected to AC network [18].

Fig. 3. Entire system diagram including the power circuit and the VSC control system based on vector control strategy.

Fig. 4. A per unit representation of the current control loop.

3.2. Outer control loop generates Idref , the inner control loop gives the relationship
between Idref and Id . Hence, in order to properly design the outer
In this control loop, the PI regulators with gain parameters of control loop, the second order transfer function of the closed loop
K pv and T iv are used to regulate the DC-link voltage. Similar to that current controller can be replaced by an equivalent first order
happen in the current control loop, it also passes through the PWM approximation with time constant, T eq ¼ 2T a for a simplification
converter with time delay T a . Since the DC-link voltage controller as follows:

2766
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

1 1 parameters are also tuned by pole placement tuning method.

ffi ð7Þ
2T 2a S2 þ 2T a :S þ 1 T eq :S þ 1 Moreover, a brief study of the effect of K p ; and K i parameters on
the output response and poles’/zeroes’ movement is achieved by
Based on (3)–(5), the DC-link dynamics is described as: manual tuning method.
dV dc 3 V d
C ¼ :Id
IL ð8Þ 4.1. Modulus optimum criteria
dt 2 V dc
Therefore, in order to linearize the nonlinear DC-link dynamics According to (6), one of the two-time constans of the control
as inferred from (8), the one-point linearization is implemented system is dominant and the other one is minor. Hence, the modu-
around an operating point at a reference input V dref . Hence, the lus optimum method is implemented for a fast response. This
liearized DC-link dynamics is expressed by: method is realized by the cancellation of largest time constant
[31]. The transfer function of the inner loop in per unit system is
DV dc ðsÞ 3 V d;0 1
¼ : ð9Þ obtained from Fig. 4, as follows:
Did ðsÞ 2 V dc;ref S:C    
1 þ T i :s 1 1 1
Therefore, the open loop and closed loop transfer function of the GC;OL ¼ K p;p:u ð13Þ
T i :s 1 þ T a :s Rp:u 1 þ sp:u :s
outer controller are expressed by (10) and (11) respectively [30]:
     By applying T i ¼ sp:u , the non-dominant pole is canceled by the
1 þ T iv :s 1 3 Vd 1
GC;OL ¼ K Pv ð10Þ controller zero and the modulus optimum tuning criteria is
T iv :s T eq :s þ 1 2 V dc S:C achieved. Therefore, a simplified open loop transfer function of
inner controller is derived as:
3 Vd K Pv Vd
2 V dc
: C:T :s þ 32 V dc
: C:TKeqPvT
iv K p;pu 1
GC;cL ¼ eq
Vd K Pv Vd
ð11Þ GC;OL ðsÞ ¼ : ð14Þ
s3 þ T1eq :s2 þ 32 V dc
: C:T eq
:s þ 32 V dc
: C:TKeqPTv ivv spu :Rpu s:ð1 þ T a :sÞ
In cascade control system the outer control loop features a In order to calculate the current controller’s gain parameter,
slower response to changes than the inner control loop. Hence, this condition achieved by:
the feed-forward control is utilized for minimizing the slow    
 GC;OL ðjxÞ   K p;pu 

dynamic response of a cascade control [30]. To maintain the power  ¼ ¼1 ð15Þ
1 þ G ðjxÞ 
balance and enhance the relative stability, the DC capacitor current C;OL spu :Rpu :T a :ðjxÞ þ spu :Rpu :ðjxÞ þ 1
2

ic is controlled to to be zero, in which Idc ¼ IL . Therefore, from (3)

Hence, the gain can be defined as:
the reference value of Id that represent the feed-forward term is
defined as follows: sp:u Rp:u
K p:u ¼ ð16Þ
2:T a
2 V dc

Id ¼ :IL ð12Þ
3 Vd According to the per unit system, sp:u ¼ Lp:u = xb Rp:u , where,Lp:u
is the per unit inductance that defined by, Lp:u ¼ xb :LðIb =V b Þ and
Fig. 5 shows the block diagram of the outer controller in per
Rp:u is the per unit resistance that defined by, Rp:u ¼ RðIb =V b Þ. There-
unit system [19]. Where xb is the the base frequency and C pu is
fore, the open loop and closed loop transfer functions of the current
the per unit capacitance of the DC-link.
control loop are [19].

4. Approaches for tunning of PI controller GC;OL ðsÞ ¼ 2:T1 a : s:ð1þT

1
a :sÞ
ð17Þ
GC;CL ðsÞ ¼ 2T 2 s2 þ2T :sþ1
1
a a
In this section different tuning methods are used for selecting
the controller parameters (i.e. gain and time constant) carefully where the natural frequency oscillation and damping factors of the
to accomplish a system stability and adequate DC voltage regula- considered system are xn ¼ T 1pffiffi2 and f ¼ p1ffiffi2, respectively.
a

tion, as well as enhance the speed response of inner control loop.

In this paper, the current controller’s gain parameters are tunned 4.2. Symmetrical optimum criteria
by modulus optimum condition due to its fast response and easy
implementation, while the outer control loop parameters are tuned The modulus optimal method is employed in the previous sec-
by applying the symmetrical optimum condition in order to tion as the system has one dominant pole and another non-
achieve an optimum voltage regulation and stability with respect dominant pole. However, in case of systems containing poles near
to disturbances. Furthermore, the voltage controller’s gain to the origin or at the origin itself, an alternative criterion called

Fig. 5. A per unit block diagram representation of the DC-voltage control loop [23].

2767
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
symmetrical optimum should criterion be applied [32–33]. As u
  K v :K u ðxd :T iv Þ2 þ 1
GV;OL ðjxÞ ¼ p ;pu
: t ¼1 ð27Þ
xd :T iv :T c
xd :T eq 2 þ 1
mentioned in (10), the open loop transfer function of the DC volt- 2
age control loop has two poles that are located at the origin. There-
fore, the symmetrical optimum method has been implemented.
That is,
The main feature of symmetrical optimum criterion is to permit
the system tolerate more delays since it maximize the phase mar- Tc
K pv ;pu ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð28Þ
gin at low frequencies. Hence, the open loop transfer function of K: T iv :T eq
the DC voltage controller in per unit system according to Fig. 5
and without considering the disturbance input and the feed- Consequently, the open loop and closed loop transfer functions
forward, can be obtained as: after applying the tuned PI controller parameters can be rewritten
     as [19]:
1 þ T iv :s 1 V d;p:u xb :C p:u  
GC;OL ¼ K Pv ;p:u ð18Þ 1 1 þ a2 :T eq :s
T iv :s T eq :s þ 1 V dc;p:u S GV;OL ðsÞ ¼ : ð29Þ
a3 :T eq 2 :s2 1 þ T eq :s
As observed from the above transfer function, if the modulus
optimum criteria is applied by setting T iv ¼ T eq , the system will
1 þ a2 :T eq :s
converts into unstable system with two poles at origin. Hence, this GV;CL ðsÞ ¼ ð30Þ
V 1 þ a2 :T eq :s þ a3 :T eq 2 :s2 þ a3 :T eq 3 :s3
method is impractical. By appling K ¼ V dpu and T c ¼ x 1C pu . Therefore,
dcpu b

the transfer function can be simplified into: Based on (30), since the closed loop transfer function denomi-
    nator has a pole of s ¼
1=a:T eq , the system transfer function can
1 þ T iv :s K 1 be rearranged to:
GV;OL ðsÞ ¼ K pv ;pu : : : ð19Þ
T iv :s 1 þ T eq :s s:T c
1 þ a2 :T eq :s
The transfer function in the frequency domain can be expressed GV;CL ðsÞ ¼
 ð31Þ
a:T eq :s þ 1 : a2 :T eq 2 :s2 þ a:ða
1Þ:T eq :s þ 1
as:
1 þ T iv :ðjxÞ
K pv ;pu :K From the transfer function denominator given in (31), the of the
GV;OL ðjxÞ ¼ ð20Þ
T iv :T c :ðjxÞ 1 þ T eq :ðjxÞ
2 characteristic equations are:

1
The stability Nyquist criteria is applied in order to achieve the s1 ¼ ð32Þ
symmetrical optimum tuning criteria as follows [34]: a:T eq
 
GV;OL ðjxÞ ¼ 1and]GV;OL ðjxÞ ¼
180 þ UM ð21Þ   sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2ffi
a
1 a
1 1
s2 ; s3 ¼

ð33Þ
where UM represents the phase margin, and based on the trans- 2:T eq 2:a:T eq a:T eq
fer function of the open loop system:

 Hence, for a particular value of T eq , the roots’ position of transfer

]GV;OL ðjxÞ ¼
180 þ tan
1 ðxT iv Þ
tan
1 xT eq function on the root locus changes depending on the value of ‘‘a”
 parameter as follows:
¼
180 þ u ð22Þ
In order to achieve a stable system, the angle u need to have a  at a < 3, the system has conjugate pair dominant poles hence it
positive for all expected values of x. The angle u is differentiated has under damping response.
with respect to x to obtain the maximum value of phase margin  at a ¼ 3, the system has two equal real dominant poles hence it
and the condition for such case is expressed as: has a critical damping response.
 at a > 3, the system has two distinct real dominant poles hence
1
xd ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi ð23Þ it has an over damping response.
T iv T eq
Therefore, the phase margin can be expressed as: In symmetrical optimum tuning method, the designed fre-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi quency, xd ¼ 1=aT eq where the maximum phase margin is pro-
UM ¼ tan
1 T iv =T eq
tan
1 T eq =T iv ð24Þ vided, depends on the selected value of ‘‘a” parameter, as
illustrated in Fig. 6, and Fig. 7. In particular, the time domain anal-
Hence, the time constant of the outer controller can be defined ysis of the outer controller and the relative system stability at dif-
by: ferent values of a shown in Fig. 6 and Fig. 7, respectively. From
  Fig. 6, with increasing the value of a the overshoot is decreased
1 þ sin UM
T iv ¼ T eq : ð25Þ but the rise time and settling time is increased, Hence, increasing
1
sin UM
the value of a improves the overshoot but decrease the system
As a result, at the crossover frequency of xd that is located in speed. Fig. 7, indicates that the value a ¼ 3 provide better system
between 1=T i and 1=T eq , the frequency characteristic of the open relative stability since it produe the largest value of phase margin
loop has a maximum phase margin of UM . Consequently, by using at intermediate frequency as compared with the other two values
symmetric property: of a. Therefore, the value a ¼ 3 that provide a better time response
and phase margin is the value that is chosen. Thus
T iv ¼ a2 T eq ð26Þ
Tc
where the symmetrical distance between 1=T iv to xd and 1=T eq to T iv ¼ 9:T eq and K pv ;pu ¼ ð34Þ
3:K:T eq
xd is represented by a.
Similarly, the controller gain can be tuned by applying the mag- In case of steady state condition, V dpu ¼ V dcpu ¼ 1ðpuÞ, that is
nitude condition as follows: K ¼ 1, the gain parameter can be defined as:

2768
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Tc
K pv ;pu ¼ ð35Þ
3:T eq

4.2.1. Gain parameters of DC-link voltage control loop for different grid
impedance values
Two systems having different grid impedance values, addressed
in Tables 1 and 2, are considered in this section to identify the gain
parameters of the outer control loop (i.e. DC-link voltage control
loop).
System 1: The considered system in the current paper.
System 2: The system implemented in [19], which has the same
base frequency xb and a ¼ 3 as given in Table 1.
From (34), since T c ¼ x 1C pu and C pu ¼ x :C:Z1 . Hence, the gain
b b dcbase

parameter in terms of DC-link grid impedance is given as follows:

C:Z dc;base
K pv ;pu ¼ ð36Þ
3T eq
Hence, Table 2 shows the computed gain parameters. From
Fig. 6. Time response of outer loop transfer function at multiple values of a: these Tables, it can be observed that the system implemented in
the current paper, has a high value of C:Z dc;base . Consequently, the
computed gain parameters become high than that of another sys-
tem in [19]. Therefore, the computed values of DC-link voltage con-
troller’s gain parameters varies according to the DC-link
capacitance and base impedance values.

4.3. Manual tuning for DC voltag

A manual tuning method has been used for tuning of DC-link

voltage controller to thoroughly study the effect of the variation
of gain parameters (K p ; K i ) on the output voltage response. Two
case studies have been considered:

4.3.1. Case study #1

To investigate the influence of the K i on the behavior of voltage
response, the K p is maintained at zero. On the other hand, the K i is
started with value of 5 and then increased progressively, while
observing the DC voltage response to identify the optimal value
of K i at which a better output response is achieved. After that, K p
is started to increase, and the dc voltage response is observed,
recorded and then described in the following figures.
Extensive simulation studies have been carried out for the DC
voltage response due to the change in the K i at a fixed value of
K p ffi 0. A sample of such results is illustrated in Fig. 8. From this
figure, it can be concluded that by increasing the value of K i , the
overshoot is reduced and the delay in the output response is
reduced (i.e. faster output response) consequently. On the other
Fig. 7. Bode plot of outer loop transfer function at multiple values of a. hand, such increase in the K i value, enabled a continuous decrease
in the rise time as shown in Fig. 9, in which a significant decrease is
exhibited when K i is changes from 20 to 30. It is noteworthy that
the critical value of K i is occurred at K i ¼ 150, which has a smallest
value of rise time. afterwards, an very small increase in the rise
time is observed. Hence, the value of K i ¼ 150 can be considered
Table 1
as the better integral gain value in this system.
System parameters. Similar to that performed previously with K i , K p is gradually
increased until an acceptable DC voltage response is attained.
Parameter Values
System #1 System #2
C:zdc;base 333:5 0:003014 Table 2
Gain parameters values.
C dc;pu 0:00000795 0:88
Lp:u 0:00334 0:15 Parameter Values
Rp:u 0:000476 0:01
System #1 System #2
F switching 20 kHz 10 kHz
xb 377 rad=s 377 rad=s K pv 2223333 10:0474
Ta 25 ls 50 ls T Iv 0:00045 0:0009
T eq 50 ls 100 ls K Iv 4940740000 11163:78

2769
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

noteworthy that by increasing K p gradually from 0, the DC voltage

response begins to damp gradually. Moreover, it is worth to note
that the output system response has an underdamping oscillation
at small values of K p . Once the value of the increased the system
response reaches its critical damping then it switched to over-
damping oscillation with high frequency. The relation between
the K p and maximum overshoot, shown in Fig. 11, illustrates that
the maximum overshoot is decreased by increasing K p .
As can be seen from Fig. 10 and Fig. 11, a better output
response/oscillation is identified at the critical value of K p ¼ 20,
while the lowest maximum overshoot is achieved at K p ¼ 40.
Hence, to confirm which value of K p that results in a better DC volt-
age response, the bode plot diagram has been plotted for the con-
sidered values of K p as illustrated in Fig. 12. From that figure, it a
larger phase margin at low frequency has been detected at
K p ¼ 20 compared to that obtained at K p ¼ 20. Hence, it can say
that the values of gain parameters which give a better DC voltage
response are (K i ¼ 150) and ðK p ¼ 20).
Fig. 8. DC voltage behavior with fixed value of K p under different value of K i .

4.3.2. Case study #2

In order to investigate the impact of changing the K p on the out-
put system response, K i is set to 0 while the value of K p is gradually
changing. Based on this change, the value of K p that gives improved
system response can be identified. After that, a gradual increment
in the K i has been employed while observing system behavior.
Fig. 13 illustrates that at zero K i and small value of K p , the output
response of the system has a low oscillation and high steady state
error. While by increasing K p , the steady state error (SSE) is
decreased significantly and the output oscillation is almost con-
stant until reach to a critical value of K p ¼ 20. Behind such value,
an oscillated output response has been observed. Fig. 14 shows
the change of SSE with a gradually increasing of K p . Fig. 15
describes the change of the waveform displacement with increas-
ing the K p , in which the K p ¼ 20 results in the lowest displacement.
Therefore, for identifying the influence of K i on the output system
response, the K p is fixed at K p ¼ 20 that gives an improved system
oscillation and K i is increased gradually as shown in Fig. 16:
Fig. 16 indicates that by gradually incrementing the value of K i
Fig. 9. Rise time as a function of K i with fixed value of K p .
from 0 to 50, the steady state error is continuously decreasing until
it is vanished. Once the value reaches K i ¼ 50, any increase in K i
resulting an almost the same output system response with unrec-
ognized increase in the speed response. Hence, the value of
K i ¼ 150 results in an optimal value of SSE, overshoot and system
speed response at the instant of 1 s. Therefore, in the two case
studies, the values of the gain parameters that results in an

Fig. 10. Output DC voltage response at multiple values of K p while K i is fixed at

K i ¼ 150.

The K i is kept constant at the value that gives the lowest rise time
(K i ¼ 150) as mentioned, while K p is gradually increased. The DC
Fig. 11. Maximum overshoot response at multiple values of K p while K i is fixed at
voltage responses are described in Fig. 10. From this figure, it is
K i ¼ 150.

2770
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Fig. 14. Steady state error of the system output response at multiple values of K p
and K i ffi 0.

Fig. 12. Bode plot diagram for the considered two values of K p and K i ¼ 150.

Fig. 15. Waveform displacement of output at multiple values of K p and K i ffi 0.

Fig. 13. Output system response at multiple values of K p and K i ffi 0.

improved system response are K p ¼ 20 and K i ¼ 150, which reveal

that the value of K i represents 7  8 times the value of K p .
Based on the aforementioned case studies’ results, it can be con-
cluded that increasing the K p influences the DC voltage response by
achieving better damping effect, reducing the steady state error
and overshoot. After certain limit, high frequency oscillation starts
to increase with any further increase in the K p . On the other hand,
increasing the K i influences the DC voltage response by increasing
the speed system response, and eliminating the steady state error.

4.3.3. Studying of Poles/Zeros movement with changing the gain

parameters values Fig. 16. The output system response at multiple values of K i and fixed K p ¼ 20.
In this section the movement of the poles/zeros of the system
closed loop transfer function with changing the K p , and, K i has
been followed and discussed according to the two previous case According to (11), the system parameter values are C ¼ 0:0667,
studies as shown in the following tables. T eq ¼ 0:000005, where at the steady state operating conditions

2771
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

V d ¼ V dc . Hence, the system closed loop transfer function can be fore, PI controller parameters of DC voltage control loop can be
defined as: tuned by specifying the damping factor f as follows. According to
Figs. 17, and 20, when the system closed loop transfer function
1:5K p 1:5K
0:0667
 s þ 0:0667Tp iv has a one real pole and two conjugate pair poles, these real and
Gc:l ¼ ð37Þ
1:5K
0:00005  S þ S2 þ 0:0667p  s þ 0:0667Tp
3 1:5K conjugate pair poles can be defined. as
iv

preal ¼ a  r ð38Þ
Table 3 and 4 show the poles and zeros locations at (K p ffi 0, and
variable K i ) and at (K i ffi 150; and variable K p ), respectively. These
pconj ¼ r  jx ð39Þ
values have been plotted in Figs. 17 and 18, which represent Case
1. Table 5 and 6 show the poles and zeros locations at (K i ffi 0, and where a > 1 meaning that the real pole is always faster than the
variable K p ) and the poles and zeros locations at ðK p ffi 20, and vari- conjugate pair. According to [19,35], the characteristic equation of
able K i ), respectively. These values have also been plotted in the system can be stated as follows:
Figs. 19 and 20 that represent Case 2. Since the real pole is posi-


tioned far away from the origin as indicated in the previous tables, s3 þ ðp þ 2rÞ:s2 þ 2pr þ r2 þ x2 :s þ p r2 þ x2 ð40Þ
transient effects corresponding to such remote pole are small, last Since the damping factor defined as
for short time and can be neglected. Hence, only the two dominant
r
poles have been indicated in the following figures. From Fig. 17, it f ¼ cos b ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð41Þ
can be seen that when K p ffi 0, and K i is increased gradually, the r2 þ x2
system can be considered as unstable system since the two domi- By rearranging the damping factor equation in terms of x, the
nant poles are not located at the Lift Hand Side (L.H.S) but are damping factor is related to the roots as follows:
located and moved along the imaginary axes. !
Fig. 18 shows the locations of the two dominant poles when 1
f2
x¼  r2 ð42Þ
K i ¼ 150 and K p is increased gradually, which indicates that at a f2
small value of K p , the system has two conjugate pair dominant
By substituting the relations, the characteristic equation
poles located on the L.H.S. Therefore, the system is stable and
becomes:
has an under damping response. While with increasing the value
!
of K p these two conjugate pair poles approaching the real axis of 1
f2
the L.H.S gradually until become two equal real roots hence the s þ ða þ 2Þ:r:s þ 2a þ 1 þ
3 2
:r2 :s
f2
system in this state has a critical damping response. After that !
any increasing in the K p converts the poles into two distinct real 1
f2
poles and the system response converts to an over damping þ 1þ :a:r3 ð43Þ
f2
response and then overdamping with high frequency. From Figs. 19,
and 20, it can be seen that when K i ffi 0; and K p is changed gradu- Comparing this characteristic equation with the characteristic
ally, one of the two dominant poles located in the origin at any equation of original closed loop transfer function expressed in
value of K p and the other dominant one moves along the L.H.S of (11), for the steady state operating conditions, the controller
the real axis with increasing K p parameters can be expressed in terms of a and f in the following
When the K i starts to increase at constant K p ¼ 20, the pole that manner.
is located in the origin starts to move along the L.H.S of real axis
1 þ 2a:f2 2C
gradually hence the system become more stable with increasing K pv ¼  ð44Þ
the value of K i . f2 :ða þ 2Þ2 3:T eq


T eq : 1 þ 2a:f2 :ða þ 2Þ
4.4. Pole placement method T iv ¼ ð45Þ
a
Pole-placement approach is used for positioning the closed- Hence, pole placement method makes the gain parameters val-
loop poles at the suitable location, depending on the transient- ues depend on the damping factor f, and a. Hence, the closed loop
response and/or frequency–response requirements (i.e. speed and poles of the system can be located in the desired location according
damping ratio) to achieve the appropriate output response. There- to the specified value of a and f. Fig. 21 which shows the change of

Table 3
Poles and Zeros at K p ffi 0, and Variable K i .

Gain Parameters Poles Zeros

Kp Ki Two dominant poles Third pole

0:00001 5 P1 ¼ þ11i
20000
4:9973  105

P2 ¼
11i
0:00001 20 P1 ¼ þ21i
20000
1:9990  106
P2 ¼
21i
0:00001 50 P1 ¼ þ34i
20000
4:9973  106
P2 ¼
34i
0:00001 100 P1 ¼ þ47i
20000
9:9950  106
P2 ¼
47i
0:00001 150 P1 ¼ þ58i
20000
1:4918  107
P2 ¼
58i
0:00001 200 P1 ¼ þ67i
20000
1:999  107
P2 ¼
67i
0:00001 300 P1 ¼ þ83i
20000
3:0288  107
P2 ¼
83i

2772
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Table 4
Poles and Zeros at K i ¼ 150, and Variable K p .

Gain Parameters Poles

Kp Ki Two dominant poles Third pole Zeros

1 150 P1 ¼
11 þ 57i
19978
15:0124

P1 ¼
11
57i
2:5 150 P1 ¼
28 þ 51i
19944
59:9900
P2 ¼
28
51i
3:5 150 P1 ¼
39 þ 43i
19921
42:8635
P2 ¼
39
43i
4:5 150 P1 ¼
51 þ 29i
19898
33:3335
P2 ¼
51
29i
5 150 P1 ¼
56 þ 14i 1:9887
30:0310
P2 ¼
56 þ 14i
5:157 150 P1 ¼
58 þ 1i
19884
29:0866
P2 ¼
58 þ 1i
5:158 150 P1 ¼
59
19883
29:0816
P2 ¼
58
5:16 150 P1 ¼
64
1:9882
29:0701
P2 ¼
54
10 150 P1 ¼
211
19773
14:9924
P2 ¼
16
20 150 P1 ¼
453
19540
7:5022
P2 ¼
8
40 150 P1 ¼
940
1:9056
3:7495
P2 ¼
4

Fig. 18. Location of the two dominant Poles at K i ¼ 150, and variable K p :
Fig. 17. Location of the two dominant Poles at K p ffi 0, and variable K i :

According to (44)-(45), the system parameters values and let

poles position with changing the value of damping factor from f ¼ 0:707; anda ¼ 2,50,100, the gain parameter values can be cal-
0:1to1:2 at a fixed value of a ¼ 2 [19] indicates that when the value culated. Therefore, in order to specify which value of a that pre-
f < 1 the system has two conjugate pair poles and operates in ferred to be used, simulation of the output response has been
underdamping. With increasing the value of f from 0.1 to 0.9 these done for these three values in the next section (see Fig. 23).
two poles move toward the real axis until become two equal real
poles at f ¼ 1 so the system operates in critical damping. After that
by increasing the value of f > 1 the two poles become two distinct 5. Simulation verification
real poles and the system operates in overdamping. Since the value
of a constant in this case so all poles are located at the same loca- MATLAB/Simulink program has been utilized for validating the
tion of the real axis. performance of the modulus optimum, symmetrical optimum and
Fig. 22, shows the effect of changing f on the transient response. pole placement tuning method by analyzing the output system
In which increasing f, resulting a decreasing on the overshoot and response of VSC-HVDC system. Since the gain parameters of the
the speed of response. Hence, the desired performance of the out- pole placement method are tuned according to the desired tran-
put waveform can be obtained by specifying the damping factor f. sient response so it has been used to prove the effectiveness of
To determine the effect of a on the output response, the gain the symmetrical optimum and manual tuning method. The system
parameter K pv ; andK iv have been computed for different value of output response at gain parameters; K i ¼ 150, and K p ¼ 20, which
aat fixed f ¼ 0:707. as shown in Table 7. are tuned by the manual tuning method, is also simulated. More-
2773
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Table 5
Poles and Zeros at K i ffi 0, and Variable K p .

Gain Parameters Poles Zeros

Kp Ki Two dominant poles Third pole

1 0:00001 P1 ¼
23
19977
1:0004e
05

P2 ¼0
2:5 0:00001 P1 ¼
56
19944
4:0021e
06
P2 ¼0
10 0:00001 P1 ¼
227
19773
1:0005e
06
P2 ¼0
20 0:00001 P1 ¼
460
19540
5:0024e
07
P2 ¼0
50 0:00001 P1 ¼
1196
18804
2:0010e
07
P2 ¼0

Table 6
Poles and Zeros at K p ¼ 20, and Variable K i .

Gain Poles Zeros

Parameters
Kp Ki Two dominant poles Third pole

20 1 P1 ¼
460
19540
0:0500

P1 ¼0
20 5 P1 ¼
460
19540
0:2500
P2 ¼0
20 20 P1 ¼
459
19540
1
P2 ¼
1
20 100 P1 ¼
458
19540
2:5000
P2 ¼
3
20 150 P1 ¼
455
19540
5:0000
P2 ¼
5
20 300 P1 ¼
444
19540
14:9924
P2 ¼
16
20 400 P1 ¼
439
19540
19:9998
P2 ¼
21
20 500 P1 ¼
433
19540
24:9997
P2 ¼
27
20 1000 P1 ¼
402
19541
49:9995
P2 ¼
57

Fig. 19. Location of the two dominant Poles at K i ffi 0, and variable K p :

Fig. 20. Location of one of the two dominant poles at K p ¼ 20, and variable K i :

over, the output DC voltage response of the pole placement

method at three different values of gain parameters are simulated. according to the desired transient response Moreover, the symmet-
The following figures demonstrate the output system response rical optimum method gives a maximum phase margin equal 53.1°
of the tuning methods. Fig. 24 and Fig. 25 indicate that the sym- and better rise and settling time as indicated in Figs. 26 and 27.
metrical optimum method results in almost the same transient Hence, it can be concluded that the symmetrical optimum method
response of the DC-link voltage and current as resulted from the gives a satisfied output DC-link voltage transient response. While,
pole placement method which it’s gain parameters are tuned Fig. 28 and Fig. 29 reveal that the steady state response of the out-

2774
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Fig. 21. Conjugate poles position.

Fig. 24. Transient response of the output DC-link voltage in symmetrical optimum,
manual and pole placement tuning methods.

Fig. 22. Poles location at different damping factor from 0.1 to 1.2 by step 0.1.

Table 7
The gain parameters at different value of a.

Parameter Values
a¼2 a ¼ 50 a ¼ 100 Fig. 25. Transient response of the reference component of the current in symmet-
K pv 333:38 33.53 17:258 rical optimum, manual and pole placement tuning methods.
T pv 0:0002999 0.00265 0:0052
K Iv 1111637 12652.83 3351:61

Fig. 26. The current controller time-domain characteristics due to a step change in
Fig. 23. The effect of changing f on the transient response due to step response at case of symmetrical optimum approach.
fixed value of a = 2, and multiple values of f = 0.3, 0.5, 0.7 and 0.9.

2775
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Fig. 30. The output system response due to manual tuning approach.

put DC-link voltage and currents in case of symmetrical optimum

Fig. 27. Bode plot diagram for the open loop controller transfer function of DC-link
method have a high value of oscillation as compared with the pole
voltage tuned by symmetrical optimum approach.
placement and manual tuning methods. Figs. 30 and 31 indicate
that although the manual tuning method is seen to be an imprac-
tical method, since it cannot be applied if the system is growing up
or there is a change in the problem type, it gives a better output
DC-link voltage and currents responses in both transient and
steady state response. Fig. 32 represents the output DC voltage of
pole placement method at three different values of a. This figure
indicates that the oscillation of the output DC voltage depends
on the value of a, in which its oscillation decreases by increasing
the value of a since the real pole of the current system is located
far away from the conjugate pair poles as indicated in Section 4.3.
Therefore, in case of a system which has a real pole located far
away from the conjugate pair poles, pole placement method has
a better output response at high value of a as compared with the
system given in [19], which gives a better DC voltage output
response at lower value of a. Furthermore, from these figures, it
can be seen that pole placement method has an underdamping
response because its closed-loop transfer function has conjugate
pair poles.
Fig. 28. The output system response of symmetrical optimum tuning approach. Fig. 31 and Fig. 33 indicate that the modulus optimum tuning
method gives a better tuning behavior of the inner control loop

Fig. 29. Measured compared to reference active current component due to

symmetrical optimum tuning approach.
Fig. 31. Measured compared to reference active current component due to manual
tuning approach.

2776
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

Fig. 32. System response at different values of a ¼ 2; 50; 10, while fixing f ¼ 0:707
in case of pole placement approach.
Fig. 34. The current controller time-domain characteristics due to a step change in
case of modulus optimum approach.

Fig. 33. Measured compared to reference active current component of the pole
placement tuning approach at a ¼ 100and fixed value of f ¼ 0:707:

Fig. 35. The current controller frequency-domain characteristics due to a step

since the main target of the inner control loop is to tracking the ref- change in case of modulus optimum approach.

erence and measured current signals. Moreover, Fig. 34 and Fig. 35

shows that the modulus optimum tuning method provide a better
reduced from 1 to 0.8 MW at the instant of 1 s. Moreover, it can
transient response and maximum phase margin equals 65.5°.
be confirmed that the independent active/reactive power control,
Therefore, the oscillation of the current in Fig. 29 results as a feed-
in which the active power doesn’t follow the change of the reactive
back from the symmetrical tuning method. Fig. 36 verifies that an
power from 0 to
0.2 at time 1.2 s.
independent active/reactive power control is properly achieved.
A comparison between the three DC-link voltage tuning meth-
Since, the reactive power doesn’t affect when the active power is
ods has been summarized as the following.

2777
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

(a) Before controlling (b) After controlling

Fig. 36. Output active/reactive power after and before controlling.

influence of gain parameters on the output DC voltage response

Method Transient Steady state Applicability and poles/zeros movement are further investigated and simulated.
response response From the theoretical analysis and simulation results, it can be
Symmetrical Symmetrical Symmetrical Simple observed that increasing the proportional gain can achieve a better
optimum optimum optimum implementation damping on the output system oscillation and decrease the over-
method method method method shoot and steady state error, while increasing the integral gain
gives a gives a high can eliminate the steady state error and increase the speed
satisfied oscillation in response of the system. On the other hand, modulus optimum tun-
transient steady state ing method gives a better tuning behavior. Furthermore, symmet-
response. response. rical optimum tuning approach provides an oscillatory output DC-
Pole Pole pole pole placement voltage in the steady state response, while its transient response is
placement placement placement method can be acceptable. Pole-placement method gives satisfied transient and
tuning method gain method has a applied in the steady state responses.
method parameters better steady underdamping
are tuned state response only.
Declaration of Competing Interest
according to response at
the desired high value of
The authors declare that they have no known competing finan-
transient a.
cial interests or personal relationships that could have appeared
response.
to influence the work reported in this paper.
Manual Manual Manual manual tuning
tuning tuning tuning method is
method method method impractical References
gives a gives a better method.
satisfied steady state [1] OETTINGER G. A strategy for competitive, sustainable and secure energy.
transient response. Report, European Union Report, 2011.
[2] Scarlatti F. Controllo di un inverter con connessione ad una linea HVDC per
response impianto eolico, Msc thesis, 2014.
[3] Peake O. The history of high voltage direct current transmission. Australian J
Multi-Discip Eng 2010;8(1):47–55.
[4] Asplund G, Eriksson K, Svensson K. DC transmission based on voltage source
converters. In: CIGRE SC14 Colloquium, South Africa. p. 1–7.
[5] Lu D, Wang X, Blaabjerg F. Impedance-based analysis of DC-link voltage
6. Conclusion dynamics in voltage-source converters. IEEE Trans Power Electron 2018;34
(4):3973–85.
Voltage-source converter-based HVDC (VSC-HVDC) transmis- [6] Lorenzen SL, Nielsen AB, Bede L. Control of a grid connected converter during
weak grid conditions. In: 2016 IEEE 7th International Symposium on Power
sion system are gaining higher attention due to their merits com- Electronics for Distributed Generation Systems (PEDG). p. 1–6.
pared to LCC-HVDC counterparts. In such transmission systems, [7] Chen X, Chen J, Gong C, Wang H. Impedance-based analysis of grid-connected
controlling the DC-voltage is mandatory while ensuring an ade- inverter in high impedance grids. In: 2013 IEEE Conference on Industrial
Electronics and Applications (ICIEA). p. 1284–9.
quate output response and system stability which are studied, ana-
[8] Alenius H. Modeling and electrical emulation of grid impedance for stability
lyzed and simulated in this paper. Several control approaches, studies of grid-connected converters, Msc, 2018.
employed in VSC, have been considered in this paper, in which [9] Moharana A, Dash P. Input-output linearization and robust sliding-mode
the current controller is tuned by modulus optimum tuning controller for the VSC-HVDC transmission link. IEEE Trans Power Delivery
2010;25(3):1952–61.
method, and DC-link voltage controller is tuned by symmetrical [10] Glad T, Ljung L. Multivariable and Nonlinear Methods. New York, NY, United
optimum, manual and pole placement methods. Moreover, the States: Taylor and Francis; 2000.

2778
D. Rabie, T. Senjyu, S. Alkhalaf et al. Ain Shams Engineering Journal 12 (2021) 2763–2779

[11] Henson MA, Seborg DE. Nonlinear process control. New Jersey: Prentice Hall [23] Durrant M, Werner H, Abbott K. Model of a VSC HVDC terminal attached to a
PTR Upper Saddle River; 1997. weak AC system. Proceedings of 2003 IEEE Conference on Control Applications,
[12] Yang B, Sang Y, Shi K, Yao W, Jiang L, Yu T. Design and real-time 2003. CCA 2003, vol. 1. IEEE; 2003. p. 178–82.
implementation of perturbation observer based sliding-mode control for [24] Li S, Haskew TA, Xu L. Control of HVDC light system using conventional and
VSC-HVDC systems. Control Eng Pract 2016;56:13–26. direct current vector control approaches. IEEE Trans Power Electron 2010;25
[13] Ramadan HS, Siguerdidjane H, Petit M, Kaczmarek R. Performance (12):3106–18.
enhancement and robustness assessment of VSC–HVDC transmission [25] Schauder C, Mehta H. Vector analysis and control of advanced static VAR
systems controllers under uncertainties. Int J Electr Power Energy Syst compensators. In: IEE Proceedings C (Generation, Transmission and
2012;35(1):34–46. Distribution), July 1993, vol. 140, no. 4, p. 299–306.
[14] Guan C, Pan S. Nonlinear adaptive robust control of single-rod electro- [26] Molinas M, Naess B, Gullvik W, Undeland T. Robust wind turbine system
hydraulic actuator with unknown nonlinear parameters. IEEE Trans Control against voltage sag with induction generators interfaced to the grid by power
Syst Technol 2008;16(3):434–45. electronic converters. IEEJ Trans Ind Appli 2006;126(7):865–71.
[15] Ruan S-Y, Li G-J, Jiao X-H, Sun Y-Z, Lie T. Adaptive control design for VSC-HVDC [27] Ruihua S, Chao Z, Ruomei L, Xiaoxin Z. VSCs based HVDC and its control
systems based on backstepping method. Electr Power Syst Res 2007;77(5– strategy. In: Proc. 2005 IEEE/PES Transmission & Distribution Conference &
6):559–65. Exposition: Asia and Pacific, Dalian, China. p. 1–6.
[16] Zhang L, Harnefors L, Nee H-P. Interconnection of two very weak AC systems [28] Yin M, Li G, Zhou M, Zhao C. Simplified Input-Output Linearization Control for
by VSC-HVDC links using power-synchronization control. IEEE Trans Power Wind Farm Interconnection Based on Multi-Terminal VSC-HVDC. In: Proc. 16th
Syst 2010;26(1):344–55. Power Systems Computation Conf. (PSCC 2008), Glasgow, Scotland, UK. p. 1–7.
[17] Meah K, Sadrul Ula A. Simple fuzzy self-tuning PI controller for multi- [29] Flourentzou N, Agelidis VG, Demetriades GD. VSC-based HVDC power
terminal HVDC transmission systems. Electr Power Compon Syst 2008;36 transmission systems: An overview. IEEE Trans Power Electron 2009;24
(3):224–38. (3):592–602.
[18] Yang B et al. Passive control design for multi-terminal VSC-HVDC systems via [30] Bajracharya C. Control of VSC-HVDC for wind power, Master of Science in
energy shaping. Int J Electr Power Energy Syst 2018;98:496–508. Energy and Environment, Norwegian University of Science and Technology,
[19] Bajracharya C, Molinas M, Suul JA, Undeland TM. Understanding of tuning June 2008.
techniques of converter controllers for VSC-HVDC. Proc. Nordic Workshop on [31] Rezek ÁJ, Coelho CA, Vicente JME, Cortez JA, Laurentino PR. The modulus
Power and Industrial Electronics (NORPIE/2008), Espoo, Finland, 2008. optimum (MO) method applied to voltage regulation systems: modeling,
[20] Ayari M, Belhaouane MM, Braiek NB, Guillaud X. Optimal control design for tuning and implementation. In: Proc. Int. Conf. on Power System Transients
VSC-HVDC systems. In: Proc. 2013 Int. Conf. on Electrical Engineering and (IPST), Rio de Janeiro, Brazil. p. 24–8.
Software Applications, Hammamet, Tunisia. p. 1–6. [32] Machaba M, Braae M. Explicit damping factor specification in symmetrical
[21] Haileselassie TM, Molinas M, Undeland T. Multi-terminal VSC-HVDC system optimum tuning of PI controllers. Proc. 1st African Control Conference, 2003.
for integration of offshore wind farms and green electrification of platforms [33] Aydin O, Akdag A, Stefanutti P, Hugo N.Optimum controller design for a multilevel
in the North Sea. Nordic Workshop on Power and Industrial Electronics AC-DC converter system. In: Proc. 20th Annual IEEE Applied Power Electronics
(NORPIE/2008), June 9-11, 2008, Espoo, Finland. Helsinki University of Conference and Exposition (APEC 2005), Austin, TX, USA. p. 1660–6.
Technology.; 2008. [34] Ogata K. Modern control engineering. 5th ed. NJ: Prentice Hall Upper Saddle
[22] Timbus A, Liserre M, Teodorescu R, Rodriguez P, Blaabjerg F. Evaluation of River; 2009.
current controllers for distributed power generation systems. IEEE Trans [35] Preitl S, Precup R-E. An extension of tuning relations after symmetrical
Power Electron 2009;24(3):654–64. optimum method for PI and PID controllers. Automatica 1999;35(10):1731–6.

2779