Modelling and Control of Multi-terminal MVDC

Distribution Network
Patrobers Simiyu†, IEEE Student Member; Ai Xin IEEE, Senior Member; Lawrence Bibaya; IEEE
Student; Vedaste Ndiyishimiye; George Adwek, Girmaw T. Bitew,

 system topology[7],[8],. The grid can be radial or mesh similar

Abstract—The revolution in the power industry has motivated to the MVAC distribution. In the radial topology, each
a paradigm shift in the multi-terminal voltage source converter converter is connected to a central node. Its availability is
(VSC) medium voltage DC (MVDC) distribution network lower in case the central node fails as the entire network
application from ship to power grid with numerous market-based
collapses. In contrast, the mesh structure has converters
applications. However, research in MVDC networks in the power
grid is in its embryonic stages, hence exploring the existing connected in a ring architecture enabling power to flow from
primary control schemes in the well-developed VSC-HVDC both ends of every node hence improving reliability. These
systems as the benchmark for adoption remains an interesting features significantly impacts on the system stability and
research subject. Therefore, this paper proposes a 10kV radial security hence the need for appropriate coordinated control
multi-terminal MVDC distribution network in PSCAD/EMTDC strategy[9]. In multi-terminal MVDC grids, the DC voltage is
with solar PV integrated and a preliminary study involving
directly related to the power balance and power flow similar to
modelling and comparison of the master-slave, voltage margin
and droop control schemes undertaken. The results show that the the AC frequency in AC system thus the DC voltage
droop control is the most responsive and flexible achieving the coordinated control is fundamental in the operation and
best network DC voltage control and active power balance in the stability of the multi-terminal DC (MTDC) grid [9],[10].
MVDC distribution network. The growing research interests in Research in MVDC distribution networks is in its infancy
MVDC control provides an impetus into future study on adaptive stage, thus the question of whether or not the existing control
droop control schemes..
schemes in VSC-HVDC are applicable to the MVDC is a
Index Terms—MVDC distribution network, master-slave,
subject for exploration[11]. The master-slave and DC voltage
voltage margin, droop control, primary level control. droop control are the most fundamental DC voltage control
strategies in the MTDC HVDC systems. In addition, the
I. INTRODUCTION voltage margin is typically a modification of the master-slave
control method[10].
T HE liberalization and climate change mitigation in the
power industry has motivated a paradigm shift in the
multi-terminal voltage source converter (VSC) medium
In the master-slave strategy, one converter terminal (master)
regulates the DC voltage in constant DC voltage mode while
voltage DC (MVDC) distribution network application from the others (slaves) are configured in constant power or AC
voltage modes. The master controls the DC bus voltage and
ship to the power grid[1],[2]. These has driven the need for
distributes power references to the slaves using a
MVDC distribution network for de-risking VSC-HVDC
transmission networks, AC distribution network reinforcement, communication system. It is the simplest in design as well as in
implementation and has high dynamic response/controllability
transmission/distribution level renewable energy (RE)
with little risk of undesirable interaction between various
integration, rail transport applications as well as urban and
voltage regulators. However, an outage of the master terminal
rural electrification[3],[4],[5]. Unlike the conventional
can result in over-voltage/under voltage and subsequent
medium voltage AC (MVAC) system, the MVDC distribution
system constitutes fewer power conversion stages making it collapse of the entire grid. The control scheme also needs
accurate and high speed communication system between the
more efficient presenting high prospects for future
converter terminals for coordinating the control scheme. In
development hence proposed as an alternative to the AC
distribution in commercial and industrial applications [1]-[6]. addition, the master terminal power rating should be
adequately large and the connected host AC system sufficiently
MVDC distribution system typically rated 1.5-30kV, may
consist of intermittent RE and variable loads in a diverse strong to accommodate possible power imbalances in the
MTDC grid[12],[13],[14]. The practicability of the
master-slave control declines with increase in the network size
P. Simiyu, Ai Xin, L. Bibaya, B. T Garmaw, V. Ndiyishimiye, E.
Mohammed and W. Kunyu are with the North China Electric Power University; hence the strategy lacks the N-1 security as it can easily lose
Beijing 102206, China ([email protected]). any one of its N components and stop operating [10].
G. Adwek, is with the Hebei University of Technology; Tianjin, 300130, The voltage margin method undertakes the DC voltage
China
control in which the constant power and constant DC voltage

POWERCON2018 Paper NO. 201804270000807 Page1/13 2733

978-1-5386-6461-2/18/$31.00 ©2018 IEEE
modes are combined in each converter terminal. The DC AC Load

voltage reference value of each terminal differs by a given

margin and the voltage control priority given to the terminal
with the lowest voltage reference. The primary DC voltage
control in the MTDC grid is transferred from one terminal with AC
the smallest reference to the others with the largest in a DC
cascading way[9],[15]. This method is highly reliable as the MVAC I
10kV MVDC Bus
MVAC II
AC DC
DC voltage working point is fixed and it is possible to precisely DC AC
control the power flow in the MTDC grid. However, as one
converter terminal controls the DC voltage at a time, it is
vulnerable to unsatisfactory transient performance especially DC DC
DC DC
in a large network. Additionally, when the number of terminals
increase, configuring their voltage margins to satisfy the
steady-state and dynamic grid requirements is challenging Battery
besides its slow dynamic response[15],[16]. PV
System
In the DC voltage droop control, the droop controller is
configured in each terminal to enable the voltage control task in
the MTDC grid to be distributed amongst the converter Fig.1. The Multi-terminal MVDC Distribution Network Model
terminals [15]. It is considered the most reliable as the grid
can’t collapse even when one voltage controlling converter II. MULTI-TERMINAL MVDC DISTRIBUTION NETWORK
DYNAMIC MODELS
fails after a severe contingency as the remaining terminals
share the voltage control responsibility by adjusting their The modeling concepts of each of the multi-terminal MVDC
power being absorbed in line with the voltage deviation. distribution network model in Fig. 1.0 are outlined prior to
Besides, it is easy to configure the droop control scheme assembly and validation in PSCAD/EMTDC.
without rigorously modifying the control system of the other A. Three-Level VSC NPC
converters. Moreover, many terminals can be incorporated
The three phase 3-level VSC neutral point clamp (NPC)
with the droop controller to form a large MTDC grid and
power converter topology is chosen for the MVDC distribution
enhance stability and reliability without the need for fast
network. This topology is attractive because its control system
communication system[9]-[13]. The main drawback of the
and modulation scheme are simple to implement. The 3-level
droop control method is derived from the inherent trade-off
NPC has lower total harmonic distortion (THD) than the
between the DC voltage deviations and the power sharing
conventional 2-level VSC which minimizes the network
accuracy. The DC voltage working point in the droop strategy
filtering needs [21]. The absence of dynamic voltage sharing
is not fixed hence the power flow can’t be controlled
problems is the main reasons that places the 3-level VSC
precisely[17],[18]. Consequently, many adaptive droop control
practically higher on the market for medium voltage and high
methods have been proposed in literature to simultaneously
power applications [22].
and effectively fulfil the constraints on DC voltage and power
The NPC VSC can be applied separately as a rectifier and
deviations[16],[18],[19],[20].
inverter when connected to an AC power source and load
As interests in MVDC distribution networks increase, the
respectively. On the other hand, it can serve as an integral
DC voltage control methods previously popular in the HVDC
bidirectional AC-DC power conversion equipment [23]. All
systems become the reference point for investigations.
the three configurations of the three phase 3-level NPC
Therefore, this paper proposes a 10kV radial multi-terminal
converters are identical hence the discussion focuses on the
MVDC network in PSCAD/EMTDC with solar PV integrated
DC-AC converter from which the operation of the others will
and a preliminary study involving modelling and comparison
be instinctive. A typical 3-level half-bridge (HB) VSC NPC
of the master-slave, voltage margin and droop control schemes
consists of a series connection of two 2-level HBs as switching
undertaken at primary level. The network dynamic models are
sets with some minor modifications per phase leg as shown in
shown in. Fig. 1.0.
Fig. 2. The two HBs are connected via two clamping diodes per
phase that form the neutral point that divides the DC link into
two equal halves each equipped with a DC capacitor. Each
phase comprise three switching states in which two control
signals are complimentary to avoid short-circuiting the DC link.
The gating signals are binary in nature with ‘1’ and ‘0’ when
the switch is ‘ON’ and ‘OFF’ respectively as shown in Fig. 2.

POWERCON2018 Paper NO. 201804270000807 Page2/13 2734

To coordinate the two 2-level HBs and control the average where idq are state variables, vdq are control inputs and edq the
value of the terminal voltage, the pulse-width modulation disturbance inputs.
(PWM) scheme is applied to control the duty ratios of the HBs d
[16],[23]. L idq edq  vdq  ridq r jZ L
S1 S2 S1' S2' Voltage dt (2)
S1
Vdc 1 1 1 0 0 +Vdc Taking the Laplace Transform, the output current is obtained
2 2 0 1 1 0 0
in (3).
S2 3 0 0 1 1 -Vdc
Vdc a sLid ed  vd  rid  jZ Liq
S1' sLiq eq  vq  riq  jZ Lid
(3)
-Vdc The d and q axes components are respectively cross coupled
2 S2'
due to the cross terms ωLiq and ωLid. Therefore, a feed-forward
b
compensator is provided for independent control of the dq axes.
Two identical d and q axes closed loop inner current controller
Fig. 2. 3-Level VSC NPC Phase Leg with a proportional integral (PI) controller for excellent
In the MVDC network in Fig. 1, the VSC NPC is connected performance is shown in Fig. 4.
to the MVAC I through a reactor and an AC transformer. The
inner and outer controllers for the VSC primary control system
are shown in Fig. 3. The space-phasor of the AC side of the idq* + idq
PI
VSC-based system is expressed in terms of the dq-axis
components created by the phase locked loop (PLL). The PLL -
synchronizes the input and output signals’ phase and frequency
by fixing its output signal to be in step with the AC grid
voltage. Fig. 4 Simplified Closed-Loop Inner Current Controller
Reactor The main objective of the inner controller is to regulate idq to
PCC P e abc e abc X
VSC the corresponding references i*dq. The converter delay due to
r Pdc
Vabc the PWM is incorporated where Tω is the switching delay
C Vdc related to the switching frequency (fs) given in (4). Kp and Ki
t1 t2 Is pulses
A are proportional and integral gains respectively.
PWM
C 1
Measurements TZ (4)
2 fs
Vref
P, Q,Vabc i dq The loop gain l(s) is given in (5) whose pole is very close to
Inner the origin. This signifies a trend in which l(s) begins to drop
Outer i*d Current
Controllers q from a relatively low frequency. Thus, the pole is cancelled by
P* , Q* , Vdq* Controller
Fig. 3. VSC NPC Control
a compensator zero given as s=(-ki/kp) then the system’s closed
loop transfer first order function is given as (6).
Inner Current Controller
s  ki k p
The dynamic equations in three phase stationary coordinate
system describing the VSC system behavior based on the
l ( s)   s
Kp
Ls r
L
(5)
Kirchoff’s Voltage Law (KVL) [24],[25],[26] in Fig. 3 is given I d (s) 1
in (1), where eabc and vabc are the AC voltages at the point of Gi ( s ) (6)
I dref ( s ) Wi s 1
common coupling (PCC) and VSC terminal respectively.
where kp=L/τi and ki=r/τi while τi is the closed-loop system’s
d
eabc  vabc riabc  L iabc (1) time constant which is a design choice. As a rule of thumb, the
dt time constant should be small i.e 0.5-5ms for fast response of
Equation (1) is rewritten in terms of the AC-side currents in the inner current controller [23].
space-vector then transformed to the dq frame in (2) assuming To ensure a fast dynamic response of the inner current
steady-state operation conditions. The PLL ensures that its controllers, the modulus optimum technique was used for
output signal is in phase with the AC grid voltage input. In this tuning its PI parameters in the study because of its simplicity
case, the d-axis of the dq frame is locked with the AC voltage, e and fast response. Based on [27],[28], the modulus optimum
at the PCC to realize a decoupled control of active and reactive technique involves pole cancellation and optimization of the
power. This represents a two input-output dynamic system absolute value to unity. This technique is effectively used when

POWERCON2018 Paper NO. 201804270000807 Page3/13 2735

tuning PI controllers in the control system with one dominant where P, Pdc, and Pc are respectively the AC input power at the
time constant and the other minor time constant by cancelling PCC, power entering the DC-grid and energy stored in the
the largest time constant while the closed loop gain maintains DC-side capacitance. The current through the DC capacitance,
larger than unity at high frequencies. (ic) is given in (10).
Outer Controllers dVdc °­ 3ed id Pdc °½
In addition to the inner current controllers in the VSC, there ic C ®  ¾ (10)
are active and reactive power; DC voltage and AC voltage dt °¯ 2Vdc Vdc °¿
outer controllers. From Fig. 3, the apparent power S at the PCC The relation expressing DC voltage is derived in (11).
is given as (7) in dq frame [17]. dVdc 1 § 3ed id Pdc · 3ed § 2CPdc ·
S 3
e i*
2 dq dq
3
2  

ed  jeq id  jiq dt
¨ 
C © 2Vdc Vdc ¹
¸ ¨ id 
2Vdc C © 3ed ¹
¸ (11)

S 3
2 ^  eqiq   j eqid  ed iq `
ed id
(7) It is observed that DC voltage can be controlled the same
way as active power by regulating id to its set point i*d as
The PLL effects the transformation of the AC voltage eabc at shown in Fig. 6. In this case, the compensator process the DC
the PCC to edq. The voltage vector oriented control (VOC) is voltage error in order to determine the set point i*d of the inner
used to decouple the active and reactive power in which the it current controller via a feed-forward compensation. The inner
adjusts the rotational speed of the dq-axis frame such that eq = 0 current controller ultimately regulates the DC voltage of the
by aligning the ed vector with the d-axis at steady-state. Thus, plant. The DC voltage controller alongside the reactive power
the active and reactive power injected to or absorbed from the controller in Fig. 5(b) forms the Vdc-Q control mode of a VSC
AC system at the PCC is given in (8). terminal.
3 2 P*
P ed id o id* Equivalent Inner Current
2 3ed DC Voltage Controller Controller
idmax G(S)
(8)
3 2Q* + + id* + ic
Q  ed iq o iq*  PI id 2ed 1 Vdc
2 3ed V*dc 3Vdc Cs
- + -idmax -
Therefore, based on (8), the active and reactive power at the iL
PCC can be controlled by id and iq respectively which Vdc
2 Pdc
constitutes the PQ control mode of a VSC terminal. Fig. 5 Pdc 3ed
shows the active and reactive power controllers. The
controllers’ outputs have limiter functions for restricting the
Fig. 6. VSC Controllers and the Plant
reference currents within permissible values. The active and From Fig. 5 and 6, it can be observed that id is involved in
reactive power flow can be effectively and independently both active power and DC voltage controllers. When the two
controlled by fast reference tracking of their corresponding are combined, a DC voltage droop controller is derived as
references commands; idq≡ idqref. shown in Fig. 7. The controller regulates the DC bus voltage
iqmax linearly by changing the active power reference in proportion
idmax
to the DC voltage, regulating power to its reference value
+ + PI
PI id* Q* iq* simultaneously with DC voltage. The slope of the droop
P*
control characteristics is the droop constant, Rdroop and its
- -idmax - -iqmax
i inverse the droop gain, Kdroop. This parameters determine the
controller sensitivity to changes in DC voltage with variations
P Q (b) in power. With the use of the PI controller, the steady-state
(a)
error of the DC voltage controller equals zero as shown in
Fig.5 Active and Reactive Power Controllers (12)[17].
Active power can be regulated as an intermediate variable by
id for ultimately controlling DC voltage to its corresponding P*  P  Kdroop (V *  V ) 0 (12)
reference in the DC grid. The power balance equation between
the AC input and the DC-side in Fig. 3 is given in (9). The
power losses of the VSC are ignored as the reactor reactance
has very small resistance.
P  Pdc  Pc 0
(9)
3
e i  Pdc  Vdc ic 0
2 d d

POWERCON2018 Paper NO. 201804270000807 Page4/13 2736

 with two poles at the origin like the outer controllers find this a
+ + suitable tuning approach which influences the system’s
Vdc_ref Kdroop PI Id_ref frequency response to low frequencies maximizing the phase
- + margin. Consequently, the system can tolerate more delays
Vdc which optimizes the control system’s behavior to disturbance
+
input [27],[28],[29].
P_ref
- B. MVAC Grid System
The AC grid system consists of the 3-phase source, the AC
P
bus, an ideal transformer and a phase reactor interfaced to the
Fig. 7. DC Voltage Droop Controller
MVDC grid via a bidirectional 3-phase 3-level VSC NPC
The reactive power output to or from a VSC can be regulated
modelled in II(A). To enhance effective filtering of the high
directly by iq as shown in (8) or indirectly for controlling AC
switching harmonics by the VSC system, an RC filter is
voltage at the PCC. From Fig. 3, (13) can be obtained between
incorporated in parallel with the reactor. The AC source
the PCC and the VSC AC terminal.
supplements the distributed RES in the MVDC distribution
e  v (r  jZ L) is (13) grid. The AC system is modeled as a balanced 3-phase voltage
where e is the AC voltage at the PCC, v the voltage at the VSC source at 50 Hz with low internal R-L impedance. The voltage
AC terminal, r and L are respectively the reactor resistance and source is star-connected with neutral grounding and controls its
inductance while is is the AC input current at the PCC. The parameters such as amplitude and frequency[30].
AC input current vector is expressed in (14). The primary and secondary transformers as well as converter
* * voltages in the AC grid system are respectively given as et1, et2
§S· § P  jQ · § P  jQ ·
is ¨e¸ ¨ e ¸ ¨ e ¸ (14) and V in Fig. 3. The voltage relations per phase across the
© ¹ © ¹ © ¹ reactor and the transformer are given in (17) and (18) where X
where S is the apparent power at the PCC earlier considered in and XL are their reactance respectively [30].
(7), while P and Q are the active and reactive power ­ X d (iabc ) ½
respectively. The AC voltage vector at the PCC is obtained as ª¬et 2abc  Vabc º¼ ®  riabc ¾
¯ Z dt ¿ (17)
(15).
X L d (iabc )
§ P  jQ · ª¬et1abc  et 2abc º¼
e v  (r  jZ L) ¨ ¸ Z dt
© e ¹ (18)
(15)
§ Pr  Z LQ · § PZ L  Qr · Taking the dq transformation of (17) gives (19) and (20).
e v ¨ ¸ j¨ ¸
© ¹ © ¹ ­ X d (id ) ½
e e
®  Xiq ¾ >et 2d  rid  Vd @
Considering only the real part equation (16) is realized. ¯ Z dt ¿ (19)
§ Pr  Z LQ ·
e v ¨ ¸ ­ X d (iq ) ½
© e ¹ (16) ®  Xid ¾ ª¬et 2 q  riq  Vq º¼
ZL
'e 'Q ¯ Z dt ¿ (20)
e
Thus, the AC voltage at the PCC can be indirectly controlled Similarly the dq transformation of (18) gives (21) and
by regulating the reactive power flow controlled by iq as its (22).
relation with active power regulated by id is very weak. Thus X L d (id )
ª¬et 2d  et1d º¼ X L iq  (21)
changes in the reactive power (ΔQ) can be used to control the Z dt
AC voltage at the PCC. Fig. 8 shows the AC voltage controller X d (iq )
of a VSC terminal. ªet1q  et 2q º X L id  L (22)
¬ ¼ Z dt
Qmax The current dynamics in the transformer in (21) and (22) are
+ PI neglected. Using the PLL to align the q-axis of the dq frame
e*
with the AC system voltage at the PCC et1, then taking et1d=0
- -Qmax
(23) and (24) apply.
e et 2d X L iq (23)
Fig. 8. AC Voltage Controller
ªet1q  et 2q º X L id (24)
The symmetrical optimum design criterion concepts were ¬ ¼
applied when tuning the outer controllers to obtain good Adding (19) and (20) to (23) and (24) respectively forms the
stability against disturbances. The open loop transfer function MVAC side model equations given in (25) and (26).

POWERCON2018 Paper NO. 201804270000807 Page5/13 2737

 ­ X d (id ) ½ VSC

®   X  X L  iq ¾ ª¬rid  Vd º¼ (25) Reactor

¯ Z dt ¿
­
°X d (i ) ½
°
  X  X L  id ¾ ª¬et1q  riq  Vq º¼
q Inner Vdq_ref
® (26) Vdq_ref(pu) PI
Current
abc
PWM
°
¯ Z dt °
¿ Controller
dq

Vabc
Vdq PLL
C. AC Load and Load-Side VSC
A constant impedance-constant current-constant power (ZIP) abc Vabc
Idq dq Iabc
static AC load model is connected to the AC system’s bus
Fig. 10 Load-Side VSC Controller
which in turn is linked to the MVDC grid via a 3-phase VSC
NPC. It is a 3-phase block widely used for implementing a D. Battery and Buck-Boost Converter
3-phase balanced load as a series combination of RLC elements The lead-acid battery model is adopted for the study as it is
as shown in Fig. 9. The equations describing the AC load model the cheapest on cost per kW basis, mature and widely used
are described in [16] using a polynomial in (27) and (28), where battery energy source storage (BESS) technology. When
Pi, Qi and Vi are the real power, active power and voltage at the analyzing the dynamic behavior of the lead acid battery, the
ith bus respectively. The variables with a 0-subscript denote the electrical equivalent models are preferable to the mathematical
nominal values while those with 1,2 and 3 represents the and electro-chemical models. The ideal, linear and Thevenin are
fraction of the constant; Z, I and P of the AC load respectively, the common equivalent circuit models[31]. The linear model
hence, Pi1+Pi2+Pi3=1 and Qi1+Qi2+Qi3=1. shown in Fig. 11 is the modest choice representing the battery as
ª
    º
2 a voltage source with a series internal resistance [32],[33].
V V
Pi Pi 0 « pi1 V i  pi 2 V i  pi3 » (27)
¬ i0 i0
¼ Ibat
+
ª
    º
2
V V
Qi Qi 0 « qi1 V i  qi 2 V i  qi 3 » (28) EMF +
Rint
Vbat
¬ i0 i0
¼ -

It can be observed that the active and reactive power -

absorbed by the ZIP load at the ith bus are proportional to the Fig.11 Linear Equivalent Circuit Model
square of the applied voltage. It should also be noted that at a The EMF and the internal resistance are functions of the
given frequency, the AC load exhibits a constant impedance battery state of charge (SOC) as shown in (29) and (30) while
[16]. For this study, the connection configuration, the nominal the SOC is calculated using (31). Vmin and Rmin are battery
voltage and frequency as well as active and reactive power are terminal voltage and internal resistance at the initial 0% SOC
specified. respectively while Kv and KR are the battery constants
determined by the battery type [32].
Pi EMF Vmin  Kv SOC (29)
Qi Rint Rmin  K R SOC (30)
Va STATIC
t
Vdc LOAD 1
³I
Vb
Vc SOC  bat (W )dW (31)
Cmax
0
The battery interfaces the MVDC distribution grid via the
bi-directional DC-DC converter. Its charging and discharging
Fig. 9 Load-Side VSC NPC connected to the Fixed AC load cycles are regulated using the battery controller. The controller
The mathematical model of the load-side VSC is similar to continually monitors the stored and the reference power as well
that presented in II(A). The reactive power controller in Fig. 5(b) as the corresponding voltage differences and regulates the
and the AC voltage controller in Fig. 8 are used to derive the charging and discharging regimes as shown in the flow chart in
load-side AC VSC controller in Fig. 10. The load-side VSC Fig. 12 [32],[34]. The constant current charging mode is
ensures that the amplitude and frequency of the AC supply recommendable for initial charging of the battery from a low
voltage to the passive AC load are constant and the PI is used to SOC. As the battery terminal voltage rises to about 90% of the
reduce the steady-state error. reference voltage, the constant voltage charging mode switches
on to prevent damages arising from over charging. The
acceptable voltage differences is taken as 2% [33].

POWERCON2018 Paper NO. 201804270000807 Page6/13 2738

 Monitoring Pdc diL
and Vdc L U o - DU d (34)
dt
1 di
YES
Pdc>0
NO D  (U o - L L ) (35)
Ud dt
NO YES NO YES In the boosting mode, the voltage and current relations; the
-2%<(Vdc-Vdc*)/(Vdc*)<2% -2%<(Vdc-Vdc*)/(Vdc*)<2% average voltage and duty cycle of the converter are given in (36),
(37) and (38) respectively.
diL
Constant Voltage Constant Current Constant Voltage Constant Current
Ud L ....t on
Charging Charging Discharging Discharging dt (36)
diL
Ud Uo L ....t off
dt
Buck-Boost DC-DC Converter diL
L U d - Uo (1 - D) (37)
Fig. 12 Battery Control Strategy dt
The bidirectional DC-DC converter in Fig. 13 operates in
1 di
buck-boost mode where the battery is charged from the MVDC D 1 (U d - L L ) (38)
network to its nominal SOC in the bucking mode then Uo dt
discharged back in the boosting mode. The actuating signals of
the IGBT switches T1 and T2 are complementary in the PWM where, D is the duty cycle, Ud and iL are the battery voltage and
control mode to accordingly charge and discharge the battery current respectively while Uo is the MVDC bus voltage.
[34],[35]. The control scheme of the buck and boost modes can be
realized from (35) and (38) as shown in Fig. 14. In each case,
T2
L the charging and discharging currents of the battery as well as
IL the output voltages of the converter can be regulated by
+
adjusting the converter duty cycle, D.
VSC

Ud T1 Uo

- +
+
PI IL_ref PI Uo_ref Dref
Ud_ref 1/(Ud) PWM
Fig. 13 Buck-Boost DC-DC Converter
- -
Assuming a steady-state operation, the voltage across the +

inductor in one switching cycle is zero as given in the inductor Ud IL Uo Signal

Generator
voltage balance equation in (32) [35].
Ts ton Ts (a)
³ u dt ³ u dt  ³ u dt
DC-DC
L L L 0
(32)
0 0 ton
U d ton  (U o )toff 0
1
where the switching period is given by Ts. +
+
During the bucking mode, the voltage and current relations Uo_ref
+ PI IL_ref PI Ud_ref Dref
PWM
1/(Uo) -
based on Lenz’s Law including the average voltage and the - - +
duty cycle of the converter are given by (33), (34) and (35)
Uo IL Ud Signal
respectively. Generator
diL
Uo L ....t on (b)
dt (33) Fig. 14 Control Scheme of the buck-boost DC-DC Converter; a). Buck Mode
di
Uo Ud L L ....t off b). Boost Mode
dt E. Solar PV and Boost DC-DC Converter
The solar PV is one of the fastest growing RE integrated into
the distribution network. When modelling a PV system, the

POWERCON2018 Paper NO. 201804270000807 Page7/13 2739

characteristics of the PV cell must be taken into consideration PPVmax-array are the maximum array voltage, current and power
as the fundamental generator unit [36]. Typically, a PV cell respectively while the I mp and Vmp are respectively the
delivers about 2W at 0.5V but the cells are connected in maximum power current and voltage.
series-parallel on a PV module to generate a high amount of VPV max  array
power. A PV array is formed with a string of modules in Ns
Vmp
parallel with each string comprising series modules. The PV
PPVMax  array
system output characteristics depend on the solar irradiance, I PV max  array (41)
the cell temperature and the output voltage [35],[37]. N s x Vmp
The performance of a PV system is usually investigated I PV max  array
under the standard test conditions (STC) of 1000 WM -2 Np
I mp
irradiance, AM 1.5 average solar spectrum and 250C cell
temperature. However, since the commercial PV modules do The commercial PV manufacturers provide product
not behave the same as in the STC in practice, appropriate PV datasheet for standard PV module characteristics; namely:
models are employed to represent the PV systems [35]. The maximum power (Pmax); Vmp; Imp; open-circuit voltage (Voc);
general equivalent PV model is moderate and most preferable short-circuit current (Isc); temperature coefficient of
[38],[37].Fig. 15 shows the general equivalent PV cell and open-circuit voltage (KV) and temperature coefficient of
module/array model. short-circuit current (KI). This data is generally used to
Np generate the electrical characteristics such as the
Rs I
+
Ns Rs I
+ current-voltage (I-V) and power voltage (P-V) curves of the
Np
PV cell/module/array [36],[39].
Ns
Rsh Vm
D0
Rsh V Ns
Np The maximum power point (MPP), where the operation of
Isc Im
Ish
the entire PV system occurs at its highest efficiency typically
ID Ish
-
- varies under changing environmental conditions of
(a) (b)
Fig. 15 General Equivalent Circuit Model: a). PV Cell b). PV Module/Array temperature and irradiance. An MPPT system is incorporated
The equation for the equivalent PV cell in Fig. 15 can be in the boost DC-DC converter in Fig. 16. The converter
derived using the basic Kirchoff’s current law. The equation of operates under a control strategy similar to that in Fig. 14(b)
the PV cell model is given in (39). The equation for a enabling the PV system interfacing the MVDC operate at
module/array configuration with Ns series and Np parallel maximum efficiency.
modules is adapted from [38] where (39) is modified and the The perturb and observe (P&O) and the incremental
subscript ‘m’ used to denote the module current and voltage in conductance are the most commercially employed MPPT
(40). The Isc, Io, A, Ns_base, Rs and Rsh represents the short algorithms for stand-alone and grid-interactive PV systems.
circuit current, dark-saturation current, the ideality factor, However, the incremental conductance is the most preferable
original series cells, series and shunt resistances respectively. because of its more accurate MPPT, higher efficiency and fast
Others include the Boltzman’s constant K, the elementary convergence. The method can also track accurately even under
charge q and temperature, T. The series and parallel resistance rapidly varying environmental conditions [40],[41] hence
denotes the internal cell and leakage resistance respectively. It chosen for the study.
is assumed that the photo-generated current, Iph>>Io hence L Diode
IL
Iph=Isc. The combination of series and parallel modules is also +
accounted for in the module/array equation, where Np=Ns=1
for a PV cell; NP =1 and NS for series number of cells in a PV
module and NS and NP for series-parallel number of modules in Ud IGBT Uo
a PV array.
I I sc  I D  I sh
ª K ˜Tq˜(AV˜NsI ˜_Rbase
s)
˜ º ª V  I ˜ Rs º (39) -
I I sc  I o «e  1»  « » Fig. 16 DC-DC Boost Converter
¬ ¼ ¬ Rsh ¼
The incremental conductance method is based on the slope
ª q (Vm Im˜Rs NNps ) º ª V  I ˜ R ˜ Ns º of the P-V curve; where the slope is positive on the left of the
I m N s ˜ I sc  N p ˜ I o ««e  1»»  « » (40)
K ˜T ˜ A˜N s _ base ˜N s
m m s Np
« N » MPP, zero at the MPP and negative on the right of the MPP as
«¬ »¼ ¬« Rsh ˜ N s
p ¼» given in (42).
The PV array output is scaled by utilizing Ns and Np
determined using (41); where VPVmax-array, IPVmax-array and

POWERCON2018 Paper NO. 201804270000807 Page8/13 2740

 dP Input: V(t); I(t)
!0
dV
dP ΔI= I(t)-I(t-Δt)
0 (42) ΔV= V(t)-V(t-Δt)
dV
dP
0 NO
ΔV=0
YES
dV
The incremental conductance aims to determine the voltage
operating point at which the PV module instantaneous YES YES
ΔI/ΔV=(-I/V) ΔI=0)
conductance equals the incremental conductance. Thus from
the basic equations in (43); NO NO

P VI YES
ΔI/ΔV>(-I/V) ΔI>0
YES

dP d ( IV ) dI dI (43)
I | I V NO
N NO
dV dV dV dV
Increment Vref Decrement Vref Decrement Vreff Increment Vref
The conditions for the true MPPT are given in (44) [42].
dP
0
dV
dI
I V
I(t-Δt)=I(t)
0 (44) V(t-Δt)=V(t)
dV
dI I
 Return
dV V Fig. 17 The Incremental Conductance Method Flow Chart [41]
Equation (44) indicates that the MPP can be obtained by
comparing the instantaneous conductance (I/V) to the III. RESULTS AND DISCUSSION
incremental conductance (dI/dV). Equations (42) and (44) can
A. MVDC Distribution Network Model in PSCAD/EMTDC
be used to derive the incremental conductance algorithm. The
relations before the MPP, at the MPP and after the MPP are The 10kV radial MVDC distribution network in Fig. 1 is
respectively given in (45) [41],[42]. modelled and simulated in PSCAD/EMTDC. The converters at
the MVAC I and II as well as that at the AC load are
dP dI I
! 0 or  !0 galvanically separated from the AC side by an equivalent
dV dV V transformer reactance. In the PV system, the incremental
dP dI I
0 or  0 (45) conductance algorithm is used for MPPT. The key parameters
dV dV V
dP dI I of the network are given in Table 1.
 0 or  0 The performance of the MVDC network is assessed for
dV dV V
The expressions in (45) define the three operation zones for master-slave, voltage margin and droop control. In the
the incremental inductance method and the flow chart is given investigations, different operation situations such as start-up
in Fig.17. In this method, the MPP is determined by comparing behavior including step changes in PV output at 2s, normal
the incremental conductance (IΔ/VΔ) to the PV array operation and operation under some contingency are
conductance (I/V). When (I/V = IΔ/VΔ), the output voltage at considered. At t=3s and t=6s, an LLL-G fault is applied at the
Vref is the VMPP and the controller maintains this voltage until AC load and MVAC I respectively for 0.1s whereas at t=9s, the
MVAC I is disconnected.
the solar irradiation changes hence change in I and the
In the master-slave control, the MVAC I is the master while
process is repeated. The MPPT regulates the PWM control
MVAC II the slave. The PV is operated at MPPT while the AC
signal of the boost converter until: (∂I/∂V) + (I/V) = 0 is
load controller regulates its terminal output voltage. At the
satisfied by adjusting its duty cycle. In this method, the peak
battery, the charge controller continually monitors the charging
power of the PV array lies above 98% of its incremental
and discharging regimes though the emphasis for the study is on
conductance[40].
the charging operation. A two-stage voltage margin controller
reviewed in [10] is applied in which the DC voltage control
priority is given to MVAC I where the lowest and highest DC
voltage values for the margin are 9.8kV and 10.2kV
respectively. The minimum, reference and maximum active
power are given as -2.8MW, -3.0MW and 3.2MW respectively
for MVAC II while the reference power for MVAC I is 6MW.
The PV, battery and AC load controller remain the same as in

POWERCON2018 Paper NO. 201804270000807 Page9/13 2741

the master slave control. In the droop control, the MVAC I & II terminals but with lower oscillation amplitudes resulting in a
are each equipped with a droop controller. A droop coefficient dip in DC voltage by 15% in the master-slave control and about
of 3 is used for effective sharing of the control responsibility in20% in both voltage margin and droop control schemes. It can
the network. The other converter controllers remain the same be observed that the oscillations in active power and voltage
as in the master slave scheme. after the fault are more significant on the PV output for the
TABLE 1 KEY PARAMETERS OF THE MVDC NETWORK voltage margin. The master-slave and droop control have
Parameters Nominal
similar characteristics with less severe transients on the PV
Value
MVDC grid Voltage 10kV output. The DC voltage oscillations settle fastest after 0.5s in
MVDC Link Capacitance 4000μF droop control and longest in 1.3s with the voltage margin.
MVAC I/II Utility Voltage & Power 5kV; 6/3MW Similarly, the power oscillations with droop method decay
Reactor Inductance (equiv. Transform.) 0.005H
AC Frequency 50Hz fastest after 0.6s and longest in 1.2s with voltage margin.
PV Array Power 4MW When a three phase fault is applied at the MVAC I (t=6s), a
PV Array Equiv. Output Voltage 7kV more severe transient occur at all the network terminals under
Boost Converter Inductance 1H
PV Capacitance 1000μF each control scheme. Dangerous power swings 20times the
PV-Boost side Capacitance 2500μF nominal value occurs at the battery with each of the control
Battery Voltage & Power 6kV; 0.67MW strategies. However, extensive power oscillations are observed
Buck Converter Inductance 1H
Battery Side Capacitance 11000μF
on the PV system when each under master-slave and voltage
AC load Voltage & Power 280V; 3MW margin control. Consequently the DC voltage falls to zero for
master-slave and to a significantly low voltage value in voltage
B. Results
margin than in droop control method. The oscillations in active
Fig. 18 shows the start-up and steady-state performance power especially in the PV output takes shortest to decay in
features of the master-slave, voltage margin and droop control 0.5-0.9s with the droop control and longest in about 2s with
strategies. During start up, the MVDC grid voltage has a voltage margin. When the MVAC I (t=9s) is disconnected, a
modest delay time of 0.12s, rise time of 0.19s, overshoot of 10% fall in active power in the network arise causing some
and settling time of 0.78s in the droop control method. The oscillations that results in the least voltage dip of 11% when
master-slave strategy closely follows with a delay time of 0.12s, with the droop control and the largest of 15% under the voltage
a rise time of 0.22s, an overshoot of 17% and a settling time of margin. The supply fall in MVAC I prompts more PV output
2s. The voltage margins registers a delay time of 0.17s, a rise with little MVAC II supplement for grid support.
time of 0.22s, an overshoot of 10% and a settling time of 1.3s. 14.0
12.0
In this way, the droop control method, takes the shortest time 10.0
Vd (kV)

8.0
6.0
possible for the DC voltage to be fast-tracked to the set value 4.0
2.0
than the other two. 0.0

There are also startup transients mainly within the first 1s in 20.0
AC Grid I(MW)

the active power at all the terminals in the grid as expected

under each control strategy with the greatest at the PV system
0.0
and the loads but decay as the steady state sets in. The droop
8.0
AC Grid II (MW)

control shows the most modest start-up features closely 6.0

4.0
2.0
followed by the master-slave due to faster responses than the 0.0
-2.0
voltage margin. The step increase in the active power output -4.0
-6.0
-8.0
from the PV system at 2s, results in some active power
4.0
decrease in the MVAC I with the master-slave and voltage 3.0
PV(MW)

2.0
margin resulting in a 3% DC voltage increase in the MVDC 1.0
0.0
-1.0
grid. It can be observed that the step changes have less than 1% -2.0
-3.0
increase in DC voltage when droop control is applied. Besides, 0.0
AC Load (MW)

the use of droop shows the most desirable steady-state

operation where the DC voltage is highly controlled, hence the
flow of active power in and out of the MVDC network. -16.0

Fig. 19 shows the contingency features for the master-slave, 1.00

Pev
Battery (MW)

voltage margin and droop control strategies. When a three

phase fault is applied at the AC load (t=3s), a surge in active
power 5times it’s nominal value occur with all the control -4.00
0.0 2.0 4.0 6.0 8.0 10.0
methods. A similar effect occurs at the other converter Time(s)
(a)

POWERCON2018 Paper NO. 201804270000807 Page10/13 2742

 14.0 14.0
12.0 12.0
Vdc kV)

Vdc (kV)
10.0 10.0
8.0 8.0
6.0 6.0
4.0 4.0
2.0 2.0
0.0 0.0
AC Grid I (MW)

20.0 20.0

AC Grid I (MW)
0.0
0.0
10.0
AC Grid II (MW)

9.0

AC grid I (MW)
-10.0
-8.0
4.0
3.0 5.0
PV(MW)

2.0

PV (MW)
1.0
0.0
-1.0
-2.0
-3.0
-4.0
0.0
AC Load (MW)

1.0
AC Load (MW)

-16.0

1.00 -16.0
Battery (MW)

1.00
Battery (MW)

-4.00
0.0 2.0 4.0 6.0 8.0 10.0
Time (s) -4.00
(b) 0.0 2.0 4.0 6.0 8.0 10.0
14.0 Time (s)
12.0 (a)
Vdc (kV)

10.0
8.0 14.0
6.0 12.0
Vdc kV)

10.0
4.0
2.0 8.0
0.0 6.0
4.0
AC Grid I (MW)

2.0
20.0 0.0
AC Grid I (MW)

20.0

-5.0

8.0 0.0
6.0
AC Grid II (MW)

4.0
10.0
AC Grid I (MW)

2.0
0.0
-2.0
-4.0
-6.0

4.0 -10.0
3.0
PV (MW)

2.0 5.0
1.0
PV (MW)

0.0
-1.0
-2.0
-3.0

0.0 -3.0
AC Load (MW)

1.0
AC Load (MW)

-16.0

1.00 -16.0
Battery (MW)

1.00
Battery (MW)

-4.00
0.0 2.0 4.0 6.0 8.0 10.0
Time (s) -4.00

(c) 0.0 2.0 4.0

Time (s)
6.0 8.0 10.0

Fig. 18 Start-up and Steady-State Characteristics for: a). Master-Slave, b). (b)
Voltage Margin and c). Droop Control Strategies

POWERCON2018 Paper NO. 201804270000807 Page11/13 2743

 14.0
12.0 a longer settling time. Thus the droop control increases the
10.0
MVDC distribution network dynamic response and flexibility
Vd (kV)

8.0
6.0
4.0
2.0
guaranteeing better DC voltage control and active power
0.0
balance.
20.0
However, the DC voltage droop strategy has its main
AC Grid I (MW)

limitation given as the inherent trade-off between DC voltage

deviations and power sharing which makes precise control of
-5.0
active power flow challenging [17],[18]. Consequently,
8.0
research in more adaptive droop control methods has emerged
AC Grid II (MW)

6.0
4.0
2.0 to simultaneously and effectively fulfil these constraints in
0.0
-2.0
-4.0
MTDC grids. For instance, Chaudhuri et al., proposed an
-6.0 adaptive/variable droop scheme for effective power sharing in
5.0
MTDC grids [16], [19]. A modified droop control for DC
PV (MW)

voltage restoration in MTDC system was presented by

Korompili and Monti [18]. The latest proposal is the use of the
-3.0 fuzzy-logic based adaptive droop control in MTDC HVDC by
1.0
Chen et al. [20]. From these adaptive droop proposal schemes,
AC Load (MW)

a better compromise between the DC voltage deviation and

power sharing is made than the conventional droop control.
-16.0 Therefore, the adaptive droop control strategies presents great
1.00
prospects for operation control of the future MTDC grid
Battery (MW)

applications likely to stir greater motivation in MVDC

development.
-4.00
0.0 2.0 4.0 6.0 8.0 10.0
Time (s) IV. CONCLUSION
(c)
Fig. 19 Contingency Characteristics for: a). Master-Slave, b). Voltage Margin The MVDC grid modelling and the comparative study of the
and c). Droop Control Strategies master-slave, voltage margin and droop control of a 10kV
radial multi-terminal VSC MVDC distribution network was
C. Discussion
undertaken in PSCAD/EMTDC. The results show that the
The master-slave control and voltage margin are typically MVDC network under droop control increases greatest
centralized primary control strategies. However, the dynamic response and adaptability that achieves the best DC
master-slave’s faster response speed gives it a more voltage control and active power balance than the master-slave
competitive advantage in start-up phase with lower influences and voltage margin methods. The growing research interests in
of start-up oscillations in active power and DC voltage. During MVDC control motivates future study into a more adaptive
steady-state operation, the two are largely similar but during droop control scheme.
transient conditions such as a three phase fault on the slack bus
MVAC I, the reliability and power sharing which is a key ACKNOWLEDGMENT
consideration increases with the voltage margin while in the
master slave decreases as the priority for DC voltage control This work was supported by 2016 National Key R & D
can be shifted from the faulted converter to the other converter Program of China to support Low-Carbon Winter Olympics of
terminal in the network. In this MTDC grid operation, the Integrated Smart-Grid Demonstration Project
master-slave strategy contrasting voltage margin, is too rigid to (2016YFB0900500) and the Beijing Natural Science
adapt to the new network changes arising from contingency Foundation (No.3182037)
scenarios.
The voltage margin and droop control depict closely related REFERENCES
steady-state performance characteristics in which active power [1] G. F. Reed, B. M. Grainger, A. R. Sparacino, E. J. Taylor, M. J.
Korytowski, and Z. Mao, “Medium Voltage DC Technology
and DC voltage is well regulated. However, the slow response Developments, Applications and Trends,” in Proc. 2012 Cigre US
in the voltage margin causes the MVDC network to be National Committee, pp. 1–7.
vulnerable to the undesirable oscillatory start-up as well as [2] G. F. Reed, B. M. Grainger, A. R. Sparacino, and Z. H. Mao, “Ship to
Grid: Medium-Voltage DC Concepts in Theory and Practice,” IEEE
persistent contingency related transients. This oscillatory Power Energy Mag., vol. 10, no. 6, pp. 70–79, 2012.
control problem arises as the bandwidth of its constant DC [3] S. Hay, C. Cleary, G. Mcfadzean, J. Mcgray, and N. Kelly, “MVDC
voltage control is lower than that of the droop controller hence Technology Study-Market Opportunities and Economic Impact,”
2015.

POWERCON2018 Paper NO. 201804270000807 Page12/13 2744

[4] M. Stieneker and R. W. De Doncker, “Medium-Voltage DC [24] N. R. Chaudhuri, B. Chaudhuri, R. Majumder, and A. Yazdani,
Distribution Grids in Urban Areas,” in Proc. 2016 IEEE 7th Multi-terminal Direct-Current Grids; Modeling, Analysis, and
International Symposium on Power Electronics for Distributed Control. New Jersey: Wiley, 2014, pp.153-168.
Generation Systems (PEDG), pp. 1–7. [25] S. Cole and R. Belmans, “A proposal for standard VSC HVDC
[5] A. Darbandi, “Medium Voltage DC Distribution for Rural dynamic models in power system stability studies,” Electr. Power
Electrification.” Manitoba Research Center, 2018. Syst. Res., vol. 81, no. 4, pp. 967–973, 2011.
[6] D. Nilsson and A. Sannino, “Efficiency Analysis of Low- and [26] R. B. S. Cole J. Beerten, “Generalised Dynamic VSC MTDC Model
Medium-Voltage DC Distribution Systems,” in Proc. 2004 Power for Power System Stability Studies,” IEEE Trans. power Syst., vol.
Engineering Society General Meeting, pp. 2315–2321. 25, no. 3, pp. 1655–1662, 2010.
[7] G. L. Kusic, G. F. Reed, J. Svensson, and Z. Wang, “A Case for [27] C. Bajracharya, M. Molinas, J. A. Suul, and T. M. Undeland,
Medium Voltage DC for Distribution Circuit Applications,” in Proc. “Understanding of Tuning Techniques of Converter Controllers for
2011 IEEE/PES Power Systems Conference and Exposition, pp. 1– VSC-HVDC,” in Proc. 2008 Nordic Workshop on Power and
7. Industrial Electronics (NORPIE).
[8] J. J. Mesas, L. Monjo, L. Sainz, and J. Pedra, “Study of MVDC [28] J. W. Umland and M. Safiuddin, “Magnitude and Symmetric
System Benchmark Networks,” in Proc. 2015 International Optimum Criterion for the Design of Linear Control Systems: What
Symposium on Smart Electric Distribution Systems and is it and How Does it Compare with the Others?,” IEEE Trans. Ind.
Technologies (EDST), pp. 235–240. Appl., vol. 26, no. 3, 1990.
[9] J. Binkai and W. Zhixin, “The Key Technologies of VSC-MTDC [29] M. Machaba and M. Braae, “Explicit Damping Factor Specification
and its Application in China,” Renew. Sustain. Energy Rev., vol. 62, in Symmetrical Optimum Tuning of PI Controllers,” in Proc. 2003
pp. 297–304, 2016. First African Control Conference., pp. 399–404.
[10] P. Simiyu, A. Xin, G. T. Bitew, M. Shahzad, and L. K. Tuan, “A [30] M. Imhof and G. Andersson, “Dynamic Modeling of a VSC-HVDC
Review of the DC Voltage Coordinated Control Strategies for Converter,” in Proc. 2013 48th International Universities’ Power
Multi-terminal VSC MVDC Distribution Network,” in Proc. 2018 Engineering Conference (UPEC), pp. 1–6.
14th IET International Conference on AC and DC Power [31] Y. H. Kirn and H. D. Ha, “Design of interface circuits with electrical
Transmission, pp. 1–9. battery models,” IEEE Trans. Ind. Electron., vol. 44, no. 1, pp. 81–
[11] Y. Ji, Z. Yuan, J. Zhao, Y. Zhao, G. Li, and Y. Li, “Control Scheme 86, 1997.
for Multi-terminal VSC-based Medium- Voltage DC Distribution [32] A. R. Sparacino, B. M. Grainger, R. J. Kerestes, and G. F. Reed,
Networks,” in Proc. 2018 14th IET International Conference on AC “Design and Simulation of a DC Electric Vehicle Charging Station
and DC Power Transmission, pp. 1–8. Connected to a MVDC Infrastructure,” in Proc. 2012 IEEE Energy
[12] T. K. Vrana, J. Beerten, R. Belmans, and O. B. Fosso, “A Conversion Congress and Exposition (ECCE), pp. 1168–1175.
Classification of DC Node Voltage Control Methods for HVDC [33] Z. Wang, X. Li, G. Li, M. Zhou, and K. L. Lo, “Energy storage
Grids,” Electr. Power Syst. Res., vol. 103, pp. 137–144, 2013. control for the photovoltaic generation system in a micro-grid,” in
[13] CIGRE., “HVDC Grid Feasibility Study," (Working Group B4.52). Proc. 2010 5th International Conference on Critical Infrastructure,
2013. pp. 1–5.
[14] F. Mura and R. W. De Doncker, “Design Aspects of a [34] M. Han and S. Yao, Principles of Power Electronics. Beijing: North
Medium-Voltage Direct Current (MVDC) Grid for a University China Electric Power University Press, 2015, pp.161-199.
Campus,” in Proc. 2011 8th International Conference on Power [35] S. Arno, J. Klaus, I. Olindo, V. S. Rene, and Z. Miro, Solar Energy;
Electronics - ECCE Asia, pp. 2359–2366. The physics and Engineering of Photovoltaic Conversion,
[15] R. T. Pinto, S. F. Rodrigues, P. Bauer, and J. Pierik, “Comparison of Technologies and Systems. Cambridge: UIT Cambridge Ltd, 2016.
Direct Voltage Control Methods of Multi-terminal DC (MTDC) [36] SMA: “Planning of a PV Generator,” pp. 1–37, 2013.
Networks through Modular Dynamic Models,” in Proc. 2011 14th [37] H. Tsai, C. Tu, and Y. Su, “Development of Generalized
European Conference on Power Electronics and Applications, pp. Photovoltaic Model Using MATLAB / SIMULINK,” in Proc. 2008
1–10. World Congress on Engineering and Computer Science (WCECS).
[16] N. R. Chaudhuri, B. Chaudhuri, R. Mujumder, and A. Yazdani, [38] M. . Villalva, J. . Gazoli, and E. . Filho, “Comprehensive Approach
Multi-terminal Direct Current Grids; Modeling, Analysis and to Modeling and Simulation of Photovoltaic Arrays,” IEEE Trans.
Control. New Jersey: Wiley, 2014, pp.153-168. Power Electron., vol. 24, no. 5, pp. 1198–1208, 2009.
[17] T. M. Haileselassie and K. Uhlen, “Precise Control of Power Flow in [39] R. P. Sera D Teodorescu R, “PV panel model based on datasheets
Multiterminal VSC-HVDC using DC Voltage Droop Control,” in values,” IEEE Transactions on Power Electronics, 2007, vol. 4, pp.
Proc. 2012 Power and Energy Society General Meeting, IEEE, pp. 2392–2396.
1–9. [40] T. Esram and P. L. Chapman, “Comparison of photovoltaic array
[18] A. Korompili and A. Monti, “Adaptive Droop-based Voltage maximum power point tracking techniques,” IEEE Trans. Energy
Control in Multi-terminal DC Systems,” in Proc. 2017 IEEE Convers., vol. 22, no. 2, pp. 439–449, 2007.
Manchester PowerTech, pp. 1–7. [41] P. Joshi and S. Arora, “Maximum power point tracking
[19] N. R. Chaudhuri and B. Chaudhuri, “Adaptive Droop Control for methodologies for solar PV systems – A review,” Renew. Sustain.
Effective Power Sharing in Multi-terminal DC (MTDC) Grids,” Energy Rev., vol. 70, pp. 1154–1177, 2017.
IEEE Trans. Power Syst., vol. 28, no. 1, pp. 21–29, 2013. [42] D. Verma, S. Nema, A. M. Shandilya, and S. K. Dash, “Maximum
[20] X. Chen, L. Wang, H. Sun, and Y. Chen, “Fuzzy Logic Based power point tracking (MPPT) techniques: Recapitulation in solar
Adaptive Droop Control in Multiterminal HVDC for Wind Power photovoltaic systems,” Renew. Sustain. Energy Rev., vol. 54, pp.
Integration,” vol. 32, no. 3, pp. 1200–1208, 2017. 1018–1034, 2016.
[21] J. Rodriguez, Franquelo, L.G.G, Kouro, S. et al., “Multilevel
Converters: An Enabling Technology for High-Power Applications,”
in Proc. 2009 of the IEEE, vol. 97, no. 11, pp. 1786–1817.
[22] J. Rodríguez, J. I. Leon, S. Kouro, R. Portillo, and M. A. M. Prats,
“The Age of Multilevel Converters Arrives,” IEEE Industrial
Electronics Magazine, no. 2, pp. 28–39, 2008.
[23] A. Yazdani and I. Reza, Voltage Source Converters in Power
Systems; Modeling, Control and Applications. New Jersey: John
Wiley & Sons Inc, 2010, pp.127-160.

POWERCON2018 Paper NO. 201804270000807 Page13/13 2745