Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

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

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
1

A Hybrid Power Sharing Control to Enhance the


Small Signal Stability in DC Microgrids
Abd-Alelah Derbas, Morteza Kheradmandi, Member, IEEE, Mohsen Hamzeh, Member, IEEE, and
Nikos D. Hatziargyriou, Life Fellow, IEEE

Abstract—Low-frequency power oscillations pose serious prob- [6]. The method is widely employed in DC microgrid due to its
lems to the operation of Distributed Generations (DGs) equipped simplicity and economic considerations. However, the voltage
with droop control. This paper presents a new hybrid power- regulation and the limited power-sharing accuracy, which
sharing control for damping low-frequency power and current
oscillations in DC microgrids. The proposed control consists appear as conflicting objectives, preclude the widespread adop-
of a decentralized droop control together with a centralized tion of the method. Much research has been carried out to
average current sharing control. A small-signal equivalent model overcome this limitation of the control method [7]–[9]. An
is analytically derived for the proposed hybrid control and is Average Current Sharing (ACS) method is proposed in [10]
incorporated into the DC microgrid model. The parameters as a decentralized secondary control to improve the voltage
of the PI controllers involved in the average current sharing
control units are tuned individually to achieve a satisfactory regulation in low-voltage DC microgrids. The stability consid-
performance in a stand-alone operation of DGs with a local erations are not involved in the reported study. A distributed
load. The contributions of control signals contained in the hybrid cascade secondary control is proposed in [11], in which the
control signal to the damping characteristic are analyzed. The inner loop is devoted to regulate the bus voltage and the
dynamic response of the DC microgrid under a variety of dis- outer loop to enhance the power-sharing. A distributed voltage
turbance scenarios is investigated. The power-sharing accuracy
of the proposed control is also examined under the investigated restoration control with a two-layer multi-agent framework is
scenarios. The method performance is also investigated for the presented in [12]. The proposed controller is supported with
oscillations caused by constant-power loads. Sensitivity analyses a communication link with a limited bandwidth. A similar
are conducted to evaluate the impact of the loading level and line voltage feedback is used in [13] with voltage feedback from
parameters on the damping ratio associated with the dominant all converters. A communication-less control algorithm is
low-frequency modes. The effectiveness of the control method is
examined in case of losing a number of communication links. The employed to achieve accurate load sharing between droop
impact of the delay in the communication links on the method controlled converters. As well as the power-sharing accuracy
performance is also examined. The simulations are conducted on and voltage deviation; the dynamic performance under dis-
a multi-source DC microgrid, and the results are verified in the turbances [14], performance in disconnection/reconnection of
OPAL-RT real-time simulator. communication channels [15], changes in a grid configuration
Index Terms—Average current sharing control, DC microgrid, and plug-and-play feature of DGs [16] constitute the most
droop control, low-frequency oscillation, small-signal analysis, important considerations in designing power-sharing control
voltage source converter. strategies, which have received little attention in previous
studies. It is to be noted that using multi-parallel secondary
I. I NTRODUCTION controls to regulate voltage and to compensate power-sharing
control worsens the dynamic performance. [11].

D C microgrids have been drawing more and more attention


due to the increasing trend toward using renewable
energy resources that rely on power electronics interfaces such
The stability considerations are among the serious chal-
lenges of DC microgrid operation. A common case of the
problem arises out of the negative incremental impedance
as photovoltaic and fuel cell systems. The growing use of DC characteristics associated with the constant power loads con-
loads has increased the tendency toward DC microgrids [1]– trolled by closed-loop controls. A model predictive control is
[3]. DC microgrids are controlled hierarchically over various presented in [17] to eliminate the adverse effect of such loads
time horizons: primary control, secondary control, and tertiary by using the injected current of the energy storage system.
control. These controls may deal with different issues such as The problem is dealt with in [18] by adjusting the virtual
voltage regulation, power-sharing, and dynamic performances. impedance, which is introduced into the output impedance
[4], [5]. In addition, these controls can be exercised through a of DG. A solution referred to as the DC electric springs are
centralized or a decentralized control [4], [5]. Droop control proposed in [19] for stabilizing DC power distribution systems.
is a typical example of a decentralized power-sharing method The low-frequency power/current oscillations with low
damping ratios is an issue caused in voltage source converter
Abd-Alelah Derbas and Morteza Kheradmandi are with the Faculty of
Electrical Engineering, Shahid Beheshti University, Tehran, Iran; Mohsen DGs controlled by droop method [20]. Adjusting the virtual
Hamzeh is with the School of Electrical and Computer Engineering, College of impedance can bring about the mitigation of low-frequency
Engineering, University of Tehran, Tehran, Iran; and Nikos D. Hatziargyriou oscillations. Increasing line impedance can lessen the current
is with the School of Electrical and Computer Engineering, the National
Technical University of Athens, Athens, Greece. (emails: a derbas@sbu.ac.ir; oscillations. This, however, gives rise to higher voltage de-
kheradmandi@sbu.ac.ir; mohsenhamzeh@ut.ac.ir, nh@power.ece.ntua.gr.) viations, a lowered power-sharing accuracy, and an increased

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
2

system loss [4], [21], [22]. A common practice is to include


physical damping elements referred to as the passive damping AC Grid
[20], [22]–[24].
An observer of dc voltage droop and current feed-forward S
control is presented in [23]. The observer measures the
feedback current of dc voltage droop and the feed-forward DG3
AC
control with no additional instrument. An active solution DC Microgrid DC

is presented in [20] to damp the low-frequency oscillations


through introducing a negative feedback of DG output current
to offset the output of voltage controller. A method based Load4
DG1
on a dynamic coefficient, referred to as the virtual inductive zl1 zl4
CB1
gain is proposed in [22] to actively damp the low-frequency
zl3 zl5 CB2
oscillations of droop control. The opposite effect on the power-
sharing accuracy and low-frequency oscillations has received Load1
little attention in the reported studies. DG2
zl2
This paper presents a new hybrid control system for enhanc- Load5 zl6
Load3
ing power management and attenuating low-frequency oscil- CB4 CB3
lations. The proposed control relies on including an Average
Current Sharing (ACS) control signal provided by a centralized
control into the decentralized droop control signal. A small- Load2
Load6
signal equivalent model is developed for the hybrid control
and is incorporated into a multi-source DC microgrid model. Fig. 1: The schematic of multi-source DC microgrid.
The mesh matrix method is used to evaluate the eigenvalues
of the dynamic system. The parameters of the PI controller
included in the ACS controller are tuned for a stand-alone il L io zth
operation of DGs with a local load to reach a desired damping + vL − iL +
ratio in the worst cases outlined. Eigenvalue analyses are +
carried out to evaluate the contribution of control units to the Vin +
− C vo RL vth
output impedance of DG. The dynamic performance of the −
proposed controller is evaluated for a variety of disturbances. −
The accuracy of the control method for load demand sharing +
is also investigated for the studied disturbance scenarios. kiv
The dynamic performance and also power-sharing accuracy PWM kpv +
s +
is investigated in the case of interruption in communication −
io 1
channels. Sensitivity analyses are also conducted to illustrate io 2
the impact of system loading and line parameters on the LPF
eigenvalues and damping ratio of the system. The dynamic −
LP

vh
F

performance is also investigated for the delay in communica- P + −


io 3 LPF K + PI + Rd LPF
tion channels. ics ∆vcs ∆vcd

II. P ROPOSED CONTROL SYSTEM LPF


Average Current Current Droop
A graphic illustration of a DC microgrid is displayed in Vref
Sharing Control Control
ioN
Fig. 1. The microgrid consists of a number of DGs which are
connected to each other by distribution lines and to the Point Fig. 2: The block-diagram of the proposed hybrid control
of Common Coupling (PCC) and the main AC grid through an method.
interlink converter. The DGs are assumed to be integrated into
the microgrid through DC/DC buck converters. The converters
are assumed to operate in the voltage control mode, in which of the PI block in the converter voltage control loop is assumed
case are referred to as the Voltage Source Converter DGs not to be saturated by entering into the over-modulation region.
(VSC-DGs). The DC microgrid is also assumed to include The grid can then be modeled by a Thevenin voltage source
local and remote loads, which are represented by a constant- vth together with a Thevenin impedance zth . A local load RL
impedance model. is also connected to the DG terminal. Using the superposition
rule, the output voltage of the converter can be expressed in
A. Equivalent Small-Signal Model of Single Bus DC Micro- terms of the hybrid control signal vh (s) and the output current
grid io (s) as follows:
The structure of the proposed control scheme is shown in
Fig.2 for a VSC-DG connected to a DC microgrid. The output vo (s) = H(s)vh (s) − zoc (s)io (s) (1)

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
3

where zoc (s) denotes the closed-loop output impedance of vL (s)


the converter without the hybrid controller and only with the zcs (s) zcd (s) zoc (s) io zth (s)


+
voltage control, and H(s) is a transfer function; which are + +
iL
calculated as follows:
vo (s) vo (s) + H(s)Vref vo RL vth

H(s) = |io (s)=0 zoc (s) = − |v (s)=0 . (2)
vh (s) io (s) h
− −
The hybrid control signal vh (s) is produced by combining
average current sharing and current droop control signals as
follows: Fig. 3: Equivalent small signal model of DG

vh (s) = Vref (s) − ∆vcd (s) + ∆vcs (s) (3)


output impedance of the closed-loop control system, respec-
where Vref (s) is the nominal voltage of the DC microgrid;
tively. The term vL (s) appearing in (8) is calculated as follows:
and ∆vcd (s) and ∆vcs (s) represent the output signals of the   X n
current droop control and average current sharing control units, ki ωcs
vL (s) = H(s)K kp + iL (s). (11)
respectively. The V-I characteristic of the droop control unit s s + ωcs m=1 m
is obtained from the figure as follows:
The equivalent model of the proposed closed-loop control
ωcd
∆vcd (s) = Rd io (s) (4) method obtained from (8) is illustrated in Fig. 3. As the
s + ωcd
figure suggests, the DG is modeled by an independent voltage
where ωcd represents the cut-off frequency of the low-pass source, a dependent voltage source, and a closed-loop output
filter contained in the current droop unit, and Rd represents impedance. The output current can be calculated from the
the static droop coefficient of the current droop controller. The figure as follows:
coefficient is determined based on the maximum allowable  
vo (s)
voltage drop at the rated current. The control signal of the vo (s) − vth = zth (s) io (s) − . (12)
average current sharing unit, ∆vcs , is written as follows: RL

ki
 Substituting vo (s) from (8) yields:
∆vcs = kp + (ics (s) − io (s)) (5) (RL + zth (s)) (H(s)Vref + vL (s)) − (RL Vth (s))
s io (s) = .
zvc (s) (RL + zth (s)) + RL zth (s)
where kp and ki denote the coefficients of the PI controller,
(13)
and ics represents the reference signal of the ACS loop, which
is produced as follows: The characteristic equation can then be obtained as follows:
  
 X N ωcd ki
ωcs zoc (s) + H(s) Rd + H(s) kp + ×
ics (s) = K io (s) (6) s + ωcd s
s + ωcs j=1 j  
RL + (rth + slth ) + RL (rth + slth ) = 0. (14)
where ωcs represents the cut-off frequency of the low-pass
filters used in the average current sharing control unit, and
B. Small Signal model of DC Microgrid With Communication
K is the contribution of the target DG to the total load. The
Delay
summation appearing in the figure gives the sum of the output
currents of DGs, which equals the sum of load currents. The The ACS units of DGs receive signals from the other DGs
average current sharing control signal is then written as for generating signal ∆vcs . However, the signal communi-
! ! cation may be subject to a considerable delay. The output
  n
ki ωcs X signal ∆vcs at instant t will equal ∆vcs at t − δ, where δ
∆vcs (s)= kp + K iL (s) −io (s) (7)
s s + ωcs m=1 m represents the delay time. The delayed ∆vcs signal may even
give rise to instability in case of long delays. The delays of
Substituting (4) and (7) in (3) and then in (1) yields transmitting signals from DGs to the control center and from
vo (s) = H(s)Vref + vL (s) − zvc (s)io (8) the control center to DGs are lumped together as a single delay.
To linearize the communication delay function, the time delay
where zvc (s) represents the closed-loop output impedance of is divided into m time-intervals with length dtn , as depicted in
the proposed control, which is calculated as follows: Fig. 4. To model the delay communication of DC microgrid,
it is assumed that DG j is the sender and DG i is the receiver.
zvc (s) = zoc (s) + zcd (s) + zcs (s) (9)
Current io,j of DG j is then sampled and sent to the DG i at
where instant tjn , n = 1, 2, ..., m, to build the sequence of io,j . The
ωcd

ki
 signal samples of io,j are received by ACS controller of DG
zcd (s) = H(s) Rd zcs (s) = H(s) kp + . (10) i with time delay δ at instant tin = tjn + δ. It is assumed that
s + ωcd s
the signal magnitude remains unchanged. The ACS control
The impedances zcd (s) and zcs (s) denote the contributions responds based on the supplied signal value io,j of t = tin ,
of the current droop and average current sharing units to the which is equal to that at t = tjn , whereas the actual current

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
4

vcs (t ) vcs (t1 ) vcs (t ) vcs (t1 )


kp=0
kp3 =0.056 Increasing kp
DG3
kp2 =0.044
DG2
kp1 =0.037
DG1

dt n
vcs (t −  ) vcs (t ) 

Fig. 4: communication delay in the output signal ∆vcs of the kp =0.1

ACS controller.

DGj
DG1 io , j (t 0j ) io , j (t1j ) io , j (t 2j ) i 'o , j
... Fig. 6: Dominant low-frequency modes for varying propor-
tional gain kp .
DG2 DG3 ...
DGi the rise and settling times and the maximum overshoot.
...
Frequency-response analyses are conducted to obtain the PI
io , j (t 0j +  )
i i i
i (t ) i (t ) i (t )
o,j 0 o,j 1 o,j 2
controller parameters with the open-loop converter for the
cases with/without the voltage controller for a bandwidth of
Fig. 5: The process of communication delay in the sent signal. 2 kHz and a phase margin of 18◦ . The coefficients are obtained
equal to 10 and 200 for kpv and kiv , respectively. To set the
parameters of the PI controller in the ACS controller, it is
i0o,j at t = tin may be different, as sketched in Fig. 5. The
assumed that the DGs are operated stand-alone connected to
delay function can be expressed as e−sδ . The received signal
a local load of 2 kW through a distribution line. The line
io,j (t = tin ) can be calculated as:
resistance rl is assumed to be zero to provide the worst-case
io,j (s) = e−sδ × i0o,j (s), (15) conditions. The line inductance ll is assumed equal to 4.4 µH.
To identify an appropriate proportional gain kp , the integral
which can be written as follows: gain ki is set to zero, and the proportional gain is changed
δ
io,j (s) e−s 2 from 0 to 0.1. An eigenvalue analysis is conducted, as in
= δ . (16)
i0o,j (s) es 2 (14), to calculate the dominant low-frequency modes, which
are shown for the individual DGs in Fig. 6. As can be seen,
Using the Taylor series, the numerator and the denominator the DGs lose the stability for k values smaller than 0.006,
p
can be expanded as follows: 0.007, and 0.008, respectively, and the stability is improved
2 3
io,j (s) 1 − δ/2s
+ δ2s/4 − δ3s/8 for increasing proportional gain. To achieve a damping ratio
= . (17) of 0.7 and an overshot of 5%, the proportional gains should be
i0o,j (s) 1 + δ/2s 2
+ δ2s/4 + δ3s/8
3

set to 0.037 at 30 Hz for DG1, 0.044 at 36 Hz for DG2, and


The higher-order terms than the third order are neglected. 0.056 at 45.5 Hz for DG3. To identify an appropriate integral
By using mathematical reduction operations, (17) is written as gain ki , the proportional gains are set to the marginal values
follows: obtained in the preceding study. The line parameters are also
16 4 chosen the same as in the preceding study. The dominant low-
io,j (s) −1 δ3 + δ2 frequency modes are depicted in Fig. 7 for DG1 for varying
= + . (18)
i0o,j (s) 2
s3 + 2sδ + δ4s2 + δ83
2
s3 + 2sδ + δ4s2 + δ83 integral gains from 0 to 1. As indicated, the figure is plotted
for the marginal proportional gain of 0.006. As can be seen,
III. STABILITY ANALYSIS the stability is lost if the integral gain exceeds 0.31.
The eigenvalues sensitivity analysis method is used for tun-
ing controller parameters and analyzing the dominant modes B. Contribution of Control Units to Output Impedance
of low-frequency.
To investigate the contribution of the control units to the
damping performance of the proposed controller, it is assumed
A. Tuning the Controller Parameters initially that only the current droop control is used to control
The DGs are assumed to operate in the voltage control the DG. To study the effect of the current droop, the frequency
mode and are equipped with PI voltage controllers. The PI response of the output impedance of the closed-loop system,
controllers are used to eliminate the steady-state error and to calculated in (9), is evaluated over a range of frequencies
give a satisfactory dynamic characteristics. This controller is with and without the current droop control. The results are
designed regardless of the outer loops and whose parameters displayed in Fig. 8 for DG1. As the figure suggests, in lower
are tuned to produce desirable performance indices, including frequencies, the output impedance of the closed-loop system is

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
5

No-Power-sharing Control
Increasing ki
DR =0.78 at
Only droop Control f =31.2Hz
DR =0.41 at
f =10.8Hz

ki =0.1
ki =0.31 Hybrid Control
DR =0.01
DR =0 DR =0.78 at
f =31.2Hz

Fig. 7: Dominant low-frequency modes for varying integral Fig. 9: Locus of output impedance poles and zeroes with
gain ki . various control methods for DG1.

PCC

zl1 zl3 zl4

I1 I2 I3
zl2 zl5

R L4
+ vL1

+ vL2
+ vL3
− −
zvc1 RL1 RL3

zvc2 zvc3 R L5
RL2
Fig. 8: The variation of output impedance with various control +
− VS1
methods for DG1. + +
− VS2 − VS3
DG1
DG2 DG3
influenced strongly by the current droop controller. This illus-
trates that the droop control offers the capability of damping Fig. 10: The proposed equivalent model
the low-frequency oscillations for frequencies below 15 Hz. In
frequencies between 15 to 100 Hz, however, using the droop
controller produces a reduction in the output impedance, which in Fig.1. The converters have been fully modeled, including
results in lower damping for the oscillations in this range of the inner loop and the droop loop with no order reduction. The
frequencies. This variation in the output impedance gives rise control model is first described by the differential equations,
to the creation of a pair of low-frequency complex conjugate and the equations are written in the s-domain. The equivalent
poles. In higher frequencies, neither the current droop nor the model of the microgrid is shown in Fig. 10.
current sharing unit makes a notable impact on the output As can be seen, the DGs are represented by using the devel-
impedance due to the low-pass filters used. The variation of oped model by dependent and independent voltage sources in
output impedance is also depicted for the case where ACS series with closed-loop output impedances, calculated for the
control unit is also included to control the DG. As can be seen, individual DGs. The characteristic equation can be obtained
the magnitude of the output impedance with hybrid control by using the mesh impedance matrix to calculate the dominant
exceeds that with droop control for all frequencies, which low-frequency modes. The microgrid is also assumed to have
illustrates the effectiveness of using the hybrid control signal. a radial configuration, in which there is a single path from
The locus of poles and zeroes of output impedance is depicted DGs and loads to the PCC. It is initially assumed that no load
in Fig. 9. As indicated in the figure, incorporating the ACS unit is connected to bus 6, and the connections toward this bus
into the control signal yields a significantly higher damping are all open. The small-signal model of DGs, obtained in Fig.
ratio in comparison to the case where only the current droop 3, can be incorporated to yield the small-signal model of the
unit is used. island DC microgrid, as shown in Fig. 10. The figure provides
a simplified small-signal model for the DC microgrid.
C. Generalization of Small Signal Model to DC microgrids Small-signal analysis is conducted to evaluate the per-
The DC microgrid is assumed to operate in island mode. formance of the proposed method on system dynamics. As
To include the obtained DG model in a microgrid, a multi-bus mentioned earlier, the conventional mesh impedance matrix
DC microgrid comprising three DGs is assumed, as displayed method can be used to obtain the characteristic equation of

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
6

TABLE II: Parameters of distribution lines

DR =0.068 DR =0.81 rl (mΩ) ll (µH) rl (mΩ) ll (µH)


f =35.7Hz f =47.6Hz
DR =0.8
zl1 3.46 2.2 zl4 3.46 2.2
DR =0.026 f =26.4Hz zl2 3.46 2.2 zl5 6.91 4.4
f =25.9Hz
zl3 6.91 4.4 zl6 69.1 44

TABLE III: Parameters of controllers

Rd kp ki
DG1 0.057 0.037 0.1
DG2 0.114 0.044 0.1
DG3 0.228 0.056 0.1
a b

Fig. 11: Low frequency modes of DC microgrid with (a) droop


control (b) proposed hybrid control. io1

TABLE I: Nominal power of DGs and load powers

io #123 (A)
Generation (kW) kW kW
DG1 5 Load1 0.5 Load4 1.5
DG2 2.5 Load2 0.5 Load5 1 io2
DG3 1.25 Load3 0.7 Load6 2

io3

the system. A 3 by 3 mesh impedance matrix is obtained


to calculate the general characteristic equation. The system
low-frequency dominant modes are depicted in Fig. 11 with
various control methods. As can be seen, the hybrid control
method provides higher damping ratios in comparison to the Fig. 12: Dynamic response for the load variation at bus 4.
droop control. While the damping ratios associated with the
two complex conjugate poles with the droop control method
are 0.068 and 0.026 at frequencies 35.7 Hz and 25.9 Hz, The static droop coefficients are chosen in such a way that
respectively, the proposed method produces damping ratios of the output voltages of converters remain within the allowed
0.81 and 0.8 at frequencies 47.6 Hz and 26.4 Hz, respectively. bounds of %5 when the maximum source powers are drawn
from the converters. Various scenarios are now examined to
IV. S IMULATION R ESULTS confirm the observations.
The control scheme is examined in an island multi-bus 1) Load change: It is assumed that the load value at bus 4
DC microgrid comprising three DGs, as shown in Fig. 1. rises from 1.5 kW to 2.5 kW at t = 0.5 s. Figure 12 shows the
The simulation model is developed in MATLAB/SIMULINK, variation in the output currents of DGs with the droop control
and the results are verified by the OPAL-RT simulator. The and with the proposed control. It can be observed that by using
microgrid base power and the nominal power of the interlink only the droop control, the output currents of DGs 1,2 and
converter are assumed equal to 10 kW. The microgrid nominal 3 exhibit oscillations at 25.3 Hz, whereas with the proposed
voltage is 48 V, and the input voltage is assumed to vary hybrid control, the DG currents vary with no oscillations.
between 55-65 V. The nominal power of DGs and also the load 2) DG connection/disconnection: To investigate the dy-
powers are given in Table I. The switching frequency is set namic response of connection/disconnection of a DG, it is
to 20 kHz.The cut-off frequency of low-pass filters is chosen assumed that DG3 is disconnected from the microgrid at
equal to 5 Hz. It is true that a higher cut-off frequency can re- t = 0.5s by opening CB1 . After settling down at a new
sult in improved low-frequency oscillations. However, the cut- operation point, the DG is reconnected to the microgrid at
off frequency should remain far away from the high-frequency t = 2.5s. The output currents are shown in Fig. 13. As can
oscillations to provide an appropriate damping coefficient for be seen, the output currents oscillate at 26.5 Hz when DG3 is
these frequencies. The inductance, resistance, and capacitance disconnected and at 25.3 Hz when reconnected if the droop
of converters are identical and equal to 2.5 mH, 0.01 Ω, and control method is used; whereas by using the proposed hybrid
50 µF, respectively. The parameters of the distribution lines control, the low-frequency oscillations are eliminated.
are given in Table II. 3) Change in Configuration: To investigate the perfor-
The parameters chosen for the PI controller of the ACS mance of the control method for a change in microgrid
unit and for the current droop unit are shown in Table III. As configuration, a radial configuration change is assumed to
can be seen, identical values are chosen for the integral gains. occur by closing breakers CB2 and CB3 to radially integrate

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
7

io1
io #123 (A)

io #123 (A)
io2
io3

Fig. 13: Dynamic response for dis/reconnecting DG3. Fig. 15: Dynamic response for a change from radial to meshed
and meshed to radial configuration.

A. Power-Sharing Accuracy
An appropriate power-sharing should be ensured among
DGs when the DC microgrid operates in island mode. The
power-sharing error is evaluated for the four simulation sce-
io #123 (A)

io1
narios dealt with in the preceding subsection. The results
io2 are shown in Figs. 16-(a), (b), (c), and (d) with the droop
io3 control and with the proposed hybrid control. As the figures
illustrate, with the droop control, the power-sharing exhibits
errors within the range [-1 2]% for the load increase and
dis/reconnecting DG scenarios. The error amounts to 11.2%
and 6.1%, respectively, for the change in microgrid con-
figuration in radial and meshed operation. The error with
the proposed hybrid control is substantially improved and
Fig. 14: Dynamic response for a radial change in configuration. is confined below 0.6% in the worst case. To compare the
performance of the proposed method, a comparison is made
with [21] to evaluate the two main objectives of the dynamic
load 6 into the grid. After settling down at a new operating performance and the power-sharing accuracy. To compare, it
point, the load and the distribution line are disconnected from is assumed that the microgrid develops a meshed configuration
the network by opening the mentioned circuit breakers at as in scenario 4. The output currents and the power-sharing
t = 2.5 sec. Figure 14 compares the performance of the accuracy are shown in Fig.s 17 and 18, respectively, for the
proposed control with the droop control. As it is seen, by proposed method and for the method of [21]. As it can be
using the droop control, the output currents of DGs 1 and 2 observed, the two methods exhibit suitable transient responses
oscillate at 26.6 Hz and DG3 at 35.8 Hz in introducing load for the mentioned disturbance. However, the method of [21]
and line, and at 26.5 Hz for DGs 1 and 2 and 34.6 Hz for exhibits power errors within the range [-2.2 6]%, while the
DG3 in disconnecting from the network. On the other hand, error with the proposed method is confined below 0.6%.
with the proposed hybrid control, the output currents move to
the new operating point with no oscillation.
4) Change to Mesh Topology: It is assumed that the mi- B. Sensitivity Analysis
crogrid develops a meshed configuration by closing breakers 1) System Loading: To analyze the effect of loading on
CB2 , CB3 , and CB4 at t = 0.5s. After stabilizing at the the low-frequency modes of DC microgrid, the load powers
new operating point, the microgrid is assumed to return to are increased individually from the base power by 10%. The
the previous radial configuration at t = 2.5s by opening simulation results indicate that the loading makes little impact
the mentioned breakers. The output currents are shown in on the eigenvalues and the damping ratio. The results are
Fig. 15. As can be seen, with the droop control, DGs 1 presented for the variation of load power on bus 1 from 500
and 2 oscillate at 26.5 Hz in closing the loop and also in to 550 W in Fig. 19.
disconnecting the breakers, whereas DG3 oscillates at 38.6 Hz 2) Line Parameters: The value of the damping ratio is
in loop closure and at 32.5 Hz in loop opening. In either cases, influenced by system parameters, including the parameters
with the proposed hybrid control, the system moves to the new of the distribution lines. In order to analyze the effect of
operating conditions with no oscillation. line resistance, it is assumed that the resistance of all lines

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
8

DG3
DG3 DG2 DG2
DG2
DG1
DG1
DG1

DG1
DG1
DG1
DG3 DG2
DG2 DG3
DG2

(a) (b)

DG2 DG2
DG2
DG3 DG3
DG3
DG1 DG1
DG1

DG1
DG1
DG3 DG1
DG3
DG3
DG2
DG2
DG2

(c) (d)
Fig. 16: Power-sharing error for a (a) load power increase, (b) dis/reconnecting DG3, (c) radial change in topology, (d) and
change to/from meshed topology.

DG2

DG3
DG1 DG2
DG3
DG1

DG2
DG3

DG1
method of [21] proposed method
method of [21] proposed method

Fig. 17: Dynamic response comparison with [21]. Fig. 18: Power-sharing error comparison with [21].

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
9

Increasing rlines
rnominal
10*rnominal

Fig. 19: Variation of the dominant modes for a 10% increase


(a)
in loading of bus 1.

are uniformly increased while remaining the inductances un- znominal


Increasing zlines
changed. Figure 20-(a) shows the effect of line resistance on
10*znominal
the eigenvalues with the droop control. As can be seen, the
damping ratio is improved as the resistance increases. The
damping ratio varies from 0.068 to 0.989 as the resistances
are augmented by 10 times the nominal values. In Fig.20-(b),
the variation of eigenvalues are shown for varying resistances
and inductances while keeping the l/r ratio constant. The
damping ratio changes from 0.068 to 0.997 as the resistances
and inductances change by 10 times the nominal values.
As indicated formerly, the microgrid exhibits non-oscillatory
behavior with the proposed hybrid control. Figures 21-(a),(b)
illustrate the variation of damping ratio with the proposed
hybrid and droop method with varying line resistances. As the
(b)
figures imply, the damping ratio is less impacted by the line
resistances with the proposed method than the droop method. Fig. 20: Variation of modes with droop control for increasing
(a) line resistance, (b) and line impedance.

C. Control Method Performance for Constant-Power Loads


shown in Fig. 23 for varying delays extending from a short
To ensure a satisfactory performance of the proposed con- 2ms delay to long 10, 50, 80, 100, and 250ms delays. As can
troller for the oscillations caused by the nonlinear loads, it is be seen, the microgrid, though severely impacted, maintains
assumed that a constant-power load (CPL) with rated power the stability even for the long communication delays. The
0.5 kW is connected radially to bus 4 through a line with the dynamic response tends to that obtained by the droop control
impedance equal to zl3 . The load is modeled by a DC-DC given in Fig. 4 as the delay increases.
buck converter operated in voltage-control mode supplying a
resistive load. This scenario is also implemented in OPAL-RT- E. Communication Link Interruption
OP5600, as shown in Fig. 22.
The control method is associated with the potential risk of
It is assumed that the load power is increased by 50% at
failures in communication links. To investigate the impact of
t = 1s. The figure shows the variation in the output currents
this failure on the performance of the control method, four
of DGs with the droop control and with the proposed control.
failure scenarios are assumed: interruption in the communica-
As can be seen, the method shows satisfactory power-sharing
tion channels of the individual three DGs and also interruption
as well as stability properties.
in the whole communication system. It is assumed that the
propagation channels are interrupted at t = 1s and restored 3s
D. Delay in Communication Links after at t = 4s. Figure 24 shows the output current of DGs
The delays of transmitting signal from DGs to the control for the four scenarios of failed communication. As it is seen,
center and from the control center to DGs are lumped together the microgrid maintains stability for all studied scenarios. The
as a single delay. The study conducted for scenario 1, in which system performance declines to that of the droop control for
load at bus 4 is increased. The transient output currents are the case of interruption in the whole communication system. In

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
10

0.8
DF (%)

0.6

io #123 (A)
0.4

0.2

0
1 2 3 4 5 6 7
R lines (times)
(a)

0.8

Fig. 23: Dynamic response in several cases of delay time in


DF (%)

0.6
communication link.
0.4

0.2

0
0.5 1 1.5 2 2.5 3 3.5
R lines (times)

(b)

Fig. 21: Variation of damping ratio for increasing line resis-


tance with (a) droop control, (b) and proposed hybrid control.
io #123 (A)
io #123 (A)

channel#1
channel#2
channel#3
cut_link

Fig. 24: Dynamic response under failed communication link


Fig. 22: Dynamic response for an increase in constant power condition.
load.

of interruption in communication links. A small-signal model


case of interruption in one of the three channels, the deficit or
is developed for the proposed hybrid control and is incorpo-
excessive power is distributed among the remaining connected
rated into a multi-source DC microgrid to evaluate the system
DGs based on new contribution coefficients. The results of
eigenvalues. Evaluating the output impedance demonstrates
the power-sharing error for DGs 1, 2, and 3 are shown in
that the included ACS signal makes a substantial contribution
Figs. 25-(a), (b) and (c), respectively. As it can be observed,
to the output impedance, thus leading to a significant reduction
the interrupted DG experiences the highest error settling at a
in the current oscillations. The results show that the hybrid
minimal error upon restoring the communication channel.
control produces a high performance in damping out the
low-frequency oscillations for various disturbance scenarios
V. C ONCLUSION extending from load change to source interruption and grid
This paper presents a new hybrid control system to enhance configuration changes. The method also provides a satisfactory
the small-signal stability and the power-sharing control of DC performance for the oscillations caused by constant-power
microgrids. Using a hybrid signal provides the control system loads. The results also indicate that the control method pro-
with the capability to function at a certain level even in the case duces a considerable reduction in power-sharing error for the

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
11

examined disturbances. The control method is demonstrated


to give a superior performance in the case interruption in
communication channels. The method is also demonstrated to
give a satisfactory performance for the delay in propagation
channels. The sensitivity analyses show that system loading
makes little impact on the eigenvalues and the damping ratio.
The variation in line resistances, on the other hand, makes a
significant impact on the damping ratio. Yet, with the proposed
hybrid method, the damping ratio is less impacted by the line
parameters than the droop method.

R EFERENCES
channel#1
channel#2
channel#3 [1] N. Hatziargyriou, H. Asano, R. Iravani, and M. Chris, “Microgrids,”
cut_link IEEE Power and Energy Magazine, vol. 5, no. 4, pp. 78–94, 2007.
[2] F. Katiraei, R. Iravani, N. Hatziargyriou, and A. Dimeas, “Microgrids
management,” IEEE Power and Energy Magazine, vol. 6, no. 3, pp.
54–68, 2008.
[3] D. Dong, I. Cvetkovic, D. Boroyevich, W. Zhang, R. Wang, and
P. Mattavelli, “Grid-interface bidirectional converter for residential DC
(a) distribution systems—Part one: High-density two-stage topology,” IEEE
Transactions on Power Electronics, vol. 28, no. 4, pp. 1655–1666, 2013.
[4] T. Dragičević, X. Lu, J. C. Vasquez, and J. M. Guerrero, “DC
microgrids—Part I: A review of control strategies and stabilization
techniques,” IEEE Transactions on Power Electronics, vol. 31, no. 7,
pp. 4876–4891, 2016.
[5] A. A. Derbas and M. Hamzeh, “A new power-sharing method for
improving power management in DC microgrid with power electronic
interfaced distributed generations,” in 2019 27th Iranian Conference on
Electrical Engineering (ICEE). IEEE, 2019.
[6] Y. Han, X. Ning, P. Yang, and L. Xu, “Review of power sharing, voltage
restoration and stabilization techniques in hierarchical controlled DC
microgrids,” IEEE Access, vol. 7, no. -, pp. 149 202–149 223, 2019.
[7] P. Wang, X. Lu, X. Yang, W. Wang, and D. Xu, “An improved distributed
secondary control method for DC microgrids with enhanced dynamic
current sharing performance,” IEEE Transactions on Power Electronics,
channel#1 vol. 31, no. 9, pp. 6658–6673, 2015.
channel#2
channel#3
[8] S. Peyghami, H. Mokhtari, P. C. Loh, P. Davari, and F. Blaabjerg,
cut_link “Distributed primary and secondary power sharing in a droop-controlled
LVDC microgrid with merged AC and DC characteristics,” IEEE Trans-
actions on Smart Grid, vol. 9, no. 3, pp. 2284–2294, 2018.
[9] P. Prabhakaran, Y. Goyal, and V. Agarwal, “Novel nonlinear droop
control techniques to overcome the load sharing and voltage regulation
issues in DC microgrid,” IEEE Transactions on Power Electronics,
(b) vol. 33, no. 5, pp. 4477–4487, 2018.
[10] S. Anand, B. G. Fernandes, and J. Guerrero, “Distributed control to
ensure proportional load sharing and improve voltage regulation in low-
voltage dc microgrids,” IEEE transactions on power electronics, vol. 28,
no. 4, pp. 1900–1913, 2012.
[11] T. V. Vu, S. Paran, F. Diaz-Franco, T. El-Mezyani, and C. S. Edrington,
“An alternative distributed control architecture for improvement in the
transient response of dc microgrids,” IEEE Transactions on Industrial
Electronics, vol. 64, no. 1, pp. 574–584, 2017.
[12] F. Guo, L. Wang, C. Wen, D. Zhang, and Q. Xu, “Distributed voltage
restoration and current sharing control in islanded dc microgrid systems
without continuous communication,” IEEE Transactions on Industrial
Electronics, vol. 67, no. 4, pp. 3043–3053, 2020.
[13] A. Kirakosyan, E. F. El-Saadany, M. S. El Moursi, A. Hassan Yazdavar,
and A. Al-Durra, “Communication-free current sharing control strategy
for dc microgrids and its application for ac/dc hybrid microgrids,” IEEE
channel#1 Transactions on Power Systems, vol. 35, no. 1, pp. 140–151, 2020.
channel#2
channel#3 [14] J. Ahmad, M. Tahir, and S. K. Mazumder, “Improved dynamic perfor-
cut_link mance and hierarchical energy management of microgrids with energy
routing,” IEEE Transactions on Industrial Informatics, vol. 15, no. 6,
pp. 3218–3229, 2019.
[15] R. Majumder, G. Ledwich, A. Ghosh, S. Chakrabarti, and F. Zare,
“Droop control of converter-interfaced microsources in rural distributed
generation,” IEEE Transactions on Power Delivery, vol. 25, no. 4, pp.
(c) 2768–2778, 2010.
[16] S. Dasgupta, S. N. Mohan, S. K. Sahoo, and S. K. Panda, “A plug and
Fig. 25: Power-sharing error under failed communication link play operational approach for implementation of an autonomous-micro-
condition for (a) DG1, (b) DG2 and (c) DG3. grid system,” IEEE Transactions on Industrial Informatics, vol. 8, no. 3,
pp. 615–629, 2012.

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSG.2022.3156850, IEEE
Transactions on Smart Grid
12

[17] S. Yousefizadeh, J. D. Bendtsen, N. Vafamand, M. H. Khooban, Mohsen Hamzeh received the B.Sc. and M.Sc.
T. Dragičević, and F. Blaabjerg, “EKF-based predictive stabilization degrees from the University of Tehran, Tehran, Iran,
of shipboard DC microgrids with uncertain time-varying load,” IEEE in 2006 and 2008, respectively, and the Ph.D. degree
Journal of Emerging and Selected Topics in Power Electronics, vol. 7, from Sharif University of Technology, Tehran, Iran,
no. 2, pp. 901–909, 2019. in 2012, all in electrical engineering. From 2013 to
[18] M. Nawaz Hussain, R. Mishra, and V. Agarwal, “A frequency-dependent 2018, he was an Assistant Professor with Shahid
virtual impedance for voltage-regulating converters feeding constant Beheshti University. In 2018, he joined the School
power loads in a DC microgrid,” IEEE Transactions on Industry Ap- of Electrical and Computer Engineering, University
plications, vol. 54, no. 6, pp. 5630–5639, 2018. of Tehran. His research interests include photo-
[19] K.-T. Mok, M.-H. Wang, S.-C. Tan, and S. Y. Ron Hui, “DC electric voltaic systems, microgrid control, and applications
springs—A technology for stabilizing DC power distribution systems,” of power electronics in power distribution systems.
IEEE Transactions on Power Electronics, vol. 32, no. 2, pp. 1088–1105,
2017.
[20] N. Rashidirad, M. Hamzeh, and K. Sheshyekani, “An effective method
for low-frequency oscillations damping in multibus DC microgrids,”
IEEE Journal on Emerging and Selected Topics in Circuits and Systems,
vol. 7, no. 3, pp. 403–412, 2017.
[21] X. Lu, J. M. Guerrero, K. Sun, and J. C. Vasquez, “An improved
droop control method for DC microgrids based on low bandwidth
communication with DC bus voltage restoration and enhanced current
sharing accuracy,” IEEE Transactions on Power Electronics, vol. 29,
no. 4, pp. 1800–1812, 2014.
[22] M. Hamzeh, M. Ghafouri, H. Karimi, K. Sheshyekani, and J. M. Guer-
rero, “Power oscillations damping in DC microgrids,” IEEE Transactions
on Energy Conversion, vol. 31, no. 3, pp. 970–980, 2016.
[23] X. Li, L. Guo, S. Zhang, C. Wang, Y. W. Li, A. Chen, and Y. Feng,
“Observer-based DC voltage droop and current feed-forward control of
a DC microgrid,” IEEE Transactions on Smart Grid, vol. 9, no. 5, pp.
5207–5216, 2018.
[24] Y.-C. Jeung, D. C. Lee, T. Dragicevic, and F. Blaabjerg, “Design of
passivity-based damping controller for suppressing power oscillations in
DC microgrids,” IEEE Transactions on Power Electronics, 2020, Early
Access.

Nikos D. Hatziargyriou is professor Emeritus in


Power Systems at the National Technical University
of Athens. He has over 10 year industrial experience
as Chairman and CEO of the Hellenic Distribu-
tion Network Operator (HEDNO) and as executive
Vice-Chair and Deputy CEO of the Public Power
Abd Alelah Derbas received the B.Sc. degree from Corporation (PPC), responsible for the Transmission
Al-Baath University, Homs, Syria, in 2009, and and Distribution Divisions. He was chair and vice-
the M.Sc. degree from Shahid Beheshti University, chair of the EU Technology and Innovation Platform
Tehran, Iran, in 2018, both in Electrical Power on Smart Networks for Energy Transition (ETIP-
Engineering. He is currently working toward a Ph.D. SNET). He is honorary member of CIGRE and past
degree with Shahid Beheshti University. He received Chair of CIGRE SC C6 “Distribution Systems and Distributed Generation”.
an excellent Scholarship From the Syrian Ministry of He is Life Fellow Member of IEEE, past Chair of the Power System Dynamic
Higher Education and Scientific Research to Shahid Performance Committee (PSDPC) and Editor in Chief of the IEEE Trans on
Beheshti University, Iran (2015). He also received Power Systems. He is the 2017 recipient of the IEEE/PES Prabha S. Kundur
the Outstanding Best Foreign Student Award from Power System Dynamics and Control Award. He has participated in more
Shahid Beheshti University (2017). His research than 60 RD&D projects funded by the EU Commission, electric utilities and
interests include microgrids stability and power management, smart grids, industry for fundamental research and practical applications. He is author
power system dynamics, intelligent control, power electronics. of the book “Microgrids: Architectures and Control” and of more than 250
journal publications and 500 conference proceedings papers. He is included
in the 2016, 2017 and 2019 Thomson Reuters lists of the top 1% most cited
researchers and he is 2020 Globe Energy Prize laureate.

Morteza Kheradmandi (M’18) received the B.S.


and M.S. degrees in Electrical Engineering from
Sharif University of Technology, Tehran, Iran, in
1999 and 2001, respectively; and the Ph.D. degree
from Sharif University of Technology in collabo-
ration with the Institut National Polytechnique de
Grenoble, Grenoble, France. He is currently an
Associate Professor with the Faculty of Electrical
Engineering, Shahid Beheshti University, Tehran,
Iran. His research interests include power system
dynamics, microgrid control, and power system op-
eration and planning.

1949-3053 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: Murdoch University. Downloaded on March 09,2022 at 15:24:16 UTC from IEEE Xplore. Restrictions apply.

You might also like