Chinese Journal of Electrical Engineering, Vol.7, No.

4, December 2021

Distributed Control Framework and Scalable

Small-signal Stability Analysis for Dynamic
Microgrids
Yuxi Men, Yuhua Du and Xiaonan Lu*
(College of Engineering, Temple University, Philadelphia 19122, USA)

Abstract: As a growing number of microgrids (MGs) has been integrated into the modern power grids, the interconnection and
applicable cooperation among multiple MGs motivate the development of networked MGs. Dynamic MGs, as an advanced networked
MGs structure, can not only integrate multiple MGs into the distribution system but also fulfill the requested system network
reconfiguration with improved flexibility. A general distributed control approach for networked MGs is reviewed. A distributed
control framework for dynamic MGs operation is developed, along with an extensible architecture with considerations of large-scale
distributed energy resources (DERs) integration. A scalable small-signal stability analysis is conducted per the proposed distributed
control strategies and the conditions under which the system is exponentially stable are derived. At last, the effectiveness of the
proposed control framework and stability analysis are verified using a 6-bus test feeder.
Keywords: Dynamic microgrids, distributed control, networked microgrids, small-signal stability analysis

[3-4]
MGs . For instance, in networked MGs, a MG can
1 Introduction1
support its neighboring MG to restore loads if there is
The increasing integration of inverter-interfaced generation deficiency caused by extreme events (e.g.,
distributed generators (DGs) has motivated the emerging power outages induced by natural disasters). However,
technology of microgrids (MGs). A MG is comprised of there exist many technical challenges in networked MGs
a cluster of DGs and loads and has explicitly identified management. For example, the system’s operating
electrical boundaries. It is capable of operating in both characteristics would become increasingly complex due
islanded and grid-connected modes. In other words, each to the heterogeneous dynamics of each MG and the
MG along the distribution feeder can not only operate as interactions among the MGs through both physic and
a dispatchable source at the point of common coupling [5]
cyber networks . As the framework of networked MGs
(PCC) but also operate autonomously without the becomes complicated, additional control efforts are
[1-2]
energization from the main power grid . required to ensure proper system operation
[6-7]
.
Additional system operating challenges are An advanced structure of networked MGs, known
presented when multiple adjacent MGs are [8]
as dynamic MGs, has recently been favored .
interconnected, which motivates the development of Compared with the conventional networked MGs that
networked MGs. The operating characteristics of each operate under static topologies and interact through
individual MG within the network could be various in static PCCs, dynamic MGs have flexible electric
load profile, power generation, and dynamic response, boundaries and interact with their adjacent MGs at
while the system operational resiliency could be multiple points of interconnection (POIs). Furthermore,
enhanced through proper coordination among the dynamic MGs could enable optimal network
reconfiguration plans for the system operators with
Manuscript received September 20, 2021; revised November 25, 2021; increased flexibility. Specifically, if there is a
accepted December 5, 2021. Date of publication December 31, 2021; date of
current version December 18, 2021. contingency in the power system and a network
* Corresponding Author, E-mail: xiaonan.lu@temple.edu
Digital Object Identifier: 10.23919/CJEE.2021.000037 reconfiguration is requested for system restoration, the
 50 Chinese Journal of Electrical Engineering, Vol.7, No.4, December 2021

topology of the distribution feeder could be seamlessly structure of conventional MGs, networked MGs, and
varied by changing the on-off status of target switches dynamic MGs is shown in Fig. 1, and the comparison
[9]
in the context of dynamic MGs . A schematic among them is summarized in Tab. 1.

Fig. 1 Comparison among the structure

Tab. 1 Comparison among the conventional, networked, and dynamic MGs

Type Conventional MGs Networked MGs Dynamic MGs

A cluster of inverter-interfaced DGs and loads Multiple interconnected conventional MGs An advanced structure of networked MGs
Characteristics Identified electric boundaries Fixed electric boundaries Flexible electric boundaries
Operating in both islanded and grid-connected modes Static PCCs with external system Dynamic POIs with adjacent MGs

The three-level hierarchical control framework and the reactive power sharing. In Ref. [17], a
including primary control, secondary control, and wireless-based robust communication algorithm is
tertiary control has been adopted in networked and developed to improve the reliability of distributed
[10-11]
dynamic MGs . Droop control has been widely secondary control. The proposed algorithm combines
implemented at the primary level due to its communications and control functionalities to absorb
simplicity and effectiveness in automatic system the transmission errors by performing an averaging
[12-13]
stabilization . However, since there are no operation in every controller. As discussed, the
intercommunications between local primary distributed secondary control strategy could deliver the
controllers and other DG units, the secondary control same control efforts with improved system operation
[18]
is used to realize global regulations in MGs. Recent reliability and resilience .
research interests have mainly been focused on Due to the increasing number of DGs, it is
decentralizing the control strategies in the secondary impractical to implement a fully connected
control level when coordinating multiple DGs. communication network among all DGs to make the
Compared to the conventional centralized secondary system states globally accessible for regulations.
control strategy where a central controller is used to However, consensus-based algorithms can calculate
regulate all the DGs, the distributed secondary control the average of locally measured parameters in a
strategy implements a local secondary controller to distributed way using a sparse communication
[19]
each DG. Each DG controller exchanges its local network . The conventional consensus-based
measurements (e.g., frequency and voltage amplitude) algorithm, called average consensus, is only applicable
with its neighboring DGs using peer-to-peer to a group of static inputs. However, considering the
[14]
communications links . In Ref. [15], a distributed requirements of appropriate coordination among DGs
secondary control scheme is presented for the system in distributed control, it make senses to develop the
network reconfiguration in an unbalanced system. In tracking of dynamical linear consensus on time-
[20]
Ref. [16], a set of distributed averaging proportional varying inputs .
integral (DPAI) controllers are proposed for islanding The small-signal stability analysis is critical to
operation, which achieve frequency regulation and the operation of the dynamic MGs due to its
[21]
proportional active power sharing and allow for a complicated structure and operational dynamics .
tunable compromise between the voltage regulation Dynamic MGs have complicated control structures
 Yuxi Men et al.: Distributed Control Framework and Scalable Small-signal Stability Analysis for
Dynamic Microgrids 51

and operating dynamics. In the context of dynamic follows. Section 2 presents a review of control
MGs, it is challenging in deriving and expanding strategies for networked MGs. Section 3 proposes a
small-signal stability analysis accordingly to the distributed secondary control strategy for dynamic
general dynamic MGs system. In Ref. [9], the MGs operation, and introduce a scalable small-
sufficient conditions for dynamic MGs exponential signal stability analysis. The proposed work is
stability are derived; In Ref. [16], A small-signal validated using the case studies in Section 4. The
stability analysis has been presented for the proposed conclusion is drawn in Section 5.
voltage controller along with a performance study in
networked MGs. 2 Review of hierarchical control framework
and consensus-based distributed control
In this work, a distributed control framework
strategies
for dynamic MGs operation is presented, along with
a scalable architecture with considerations of
2.1 Primary control and secondary control
large-scale distributed energy resources (DERs)
integration. Furthermore, a scalable small-signal The conventional primary and secondary control
stability analysis is derived based on the proposed framework is depicted in Fig. 2. It is worth mentioning
distributed control strategies and the conditions that since the tertiary control relies on the present
under which the system is exponentially stable are electricity market and price, it is not the key point of
[10]
derived. The remainder of this paper is structured as this study .

Fig. 2 Conventional primary and secondary control framework

2.1.1 Primary control for the inductive system are represented as follows
The primary control coordinates multiple ωi = ω * − mi Pi (1a)
inverters to automatically stabilize the system’s
Ei = E ∗ − ni Qi (1b)
operating frequency and voltage. When the system
operates under the islanded mode, each dispatchable where ωi and Ei are the operating frequency and output
* *
inverter acts as a grid-forming voltage source inverter voltage amplitude; ω and E are the nominal
[10]
(VSI) . Hence, to guarantee an ideal output frequency and voltage amplitude; mi and ni are droop
impedance at the fundamental frequency, the voltage, coefficients; Pi and Qi are the active and reactive
active power, and reactive power are regulated using power flows, respectively.
the primary control framework consisting of inner 2.1.2 Secondary control
voltage and current control loops, droop control, However, DGs that are only implemented with
voltage reference generators, and optional virtual droop control will be regulated with steady-state
impedance loops. Among them, droop control, as an deviations. Referring to the droop characteristics in
adaptive control, is used to adjust frequency and Eqs. (1a) and (1b), Pi and Qi may not be guaranteed to
voltage magnitude based on the changing loading remain at zero, thus the operating frequency ωi and
conditions. When integrating multiple VSIs in parallel, output voltage amplitude Ei would deviate from their
* * [16]
the widely adopted P-f and Q-V droop characteristics nominal values (ω and E ), respectively .
 52 Chinese Journal of Electrical Engineering, Vol.7, No.4, December 2021

The secondary level control is implemented to among agents, and A is the adjacency matrix with
eliminate the steady-state error generated by the droop entries aij=aji, where aij=1 if there is communication
control. For the centralized secondary control, the between nodes i and j (i.e., (i, j)∈ε), or else aij=0.
system operating frequency (ωi) and voltage Fig. 3 presents a sample topology of the
magnitude (Ei) at PCC are measured and compared to communication network among the DGs in the
* *
their reference values (ω and E ), respectively. And networked MGs and its corresponding adjacency
[7]
then the errors processed by the compensators (ωseci matrix . Based on the communication network
and Eseci) are sent to the primary control level of the topology, the commonly used continuous-time average
inverter to restore the frequency and output voltage consensus algorithm can be written as
[10] n
magnitude . In other words, Eqs. (1a) and (1b)
*
xi (t ) = − k ′∑ aij ( xi (t ) − x j (t )) (3a)
become ωi=ω −miPi+ωseci while Ei=E−niQi+Eseci. A set j =1

of PI controllers could be used to realize the frequency 1

and voltage amplitude restorations
[14]
xi (t ) = ∑ xi (0)
n i∈ε
t →∞ (3b)

ωsec i = kω _ P (ω * − ωi ) + kω _ I ∫ (ω * − ωi )dt (2a) where xi(t) refers to the static state of agent i at the
time t (i=1, …, n); xi(0) is the initial value at t=0; k ′
Esec i = kE _ P ( E * − Ei ) + kE _ I ∫ ( E * − Ei )dt (2b)
is a measurement of the interaction strength that
where kω_P, kω_I, kE _ P , and kE _ I are the control gains [22]
modifies the system states among agents . Eq. (3b)
of the secondary control. indicates all the states in Eq. (3a) asymptotically
[22]
2.2 Consensus-based distributed secondary control achieve the average consensus .

In Eqs. (2a) and (2b), the conventional

centralized secondary controller is used to regulate the
frequency and voltage amplitude of all the DGs.
However, a significant drawback of the centralized
secondary controller is that the system could cease the
services if any failures occur in the controller. In
contrast, the distributed secondary control strategy
uses a local secondary controller in each DG to
communicate with its neighboring DGs, which
enhances the reliability of the system.
Furthermore, as the system configuration
becomes complex and the flexible network
reconfiguration is requested in the system (e.g.,
Fig. 3 Communication network and adjacency matrix
dynamic MGs), it is not applicable to build a fully
connected communication network among all DGs for 2.2.2 Dynamic consensus algorithm
control implementation. However, the limitation can Compared to the average consensus that tracks
be released using consensus algorithms over a sparse the average of static states, the average of dynamic
communication network. Two types of input signals could be tracked using the dynamic
consensus-based algorithms that can be applied in the consensus. The continuous-time dynamic consensus
distributed secondary control of dynamic MGs are also algorithm is expressed as
introduced in Section 2.2. n
xi (t ) = zi (t ) − k ′ ∑ aij ( xi (t ) − x j (t )) (4a)
2.2.1 Average consensus algorithm i , j∈ε

The communication network of the multi-agent 1

system can be illustrated as a graph G=(V, ε, A),
xi (t ) = ∑ zi (t )
n i∈V
t →∞ (4b)

where V={V1, V2, ..., Vn} represents the mark of each where zi(t) refers to the dynamic state of agent i at the
agent, ε⊆V×V refers to effective communications time t (i=1, …, n); zi(0) represents the initial value at
 Yuxi Men et al.: Distributed Control Framework and Scalable Small-signal Stability Analysis for
Dynamic Microgrids 53

t=0 and xi(0)=zi(0); Eq. (4b) indicates all the system controllers are written as
states in Eq. (4a) reach the dynamic consensus without ⎧ ωsec = ωM0 + ωM1 + ωM2 + ωM3
[20]
steady-state errors . ⎪
⎪ω M0 = − kp1 ∑ aij ( Pi ′ − Pj′)
⎪
3 Distributed control strategies for dynamic ⎨ ω M1 = − kf (ωi − ω * ) − (1 − λi )kp1 ∑ μSSW ( Pi ′ − PN′ ) (5)
⎪
MGs and scalable small-signal stability ⎪ωM2 = −λi kp2 ( PSS′ W − 0)
analysis ⎪ ω = −ηi kθ (θSSW − 0)
⎩ M3
⎧ Esec = EM0 + EM1 + EM2 + EM3
3.1 Dynamic MGs distributed control strategies ⎪
⎪ EM0 = kq1 ∑ aij (Qi′ − Q′j )
Compared to the networked MGs, the dynamic MGs ⎪
⎨ EM1 = −(1 − ηi )kv (φi − E * ) − (1 − λi )kq1 ∑ μSSW (Qi′ − QN′ )
with flexible electric boundaries demonstrate higher ⎪
⎪ EM2 ′ − 0)
= −λi kq2 (QSSW
scalability and resilience for stabilizing the system
⎪
operation, but it requires a more complex system ⎩ EM3 = −ηi kv ( ESSW − E * )
architecture and control framework. The requested (6)
network reconfiguration is realized in the context of where ηi and λi are a pair of binary variables to
th
dynamic MGs by changing the status of smart switches indicate the operation mode of the i DG; μssw
(SSWs), which are used to achieve the boundary variation represents the status of SSW (μssw=1 as closed and
[23]
of dynamic MGs . Moreover, a concept of minimum μssw=0 as open); kf, kp1, 2, kv, kq1, 2 and kθ are the
MG (min-MG) that is the smallest set of DGs and SSWs designed control gains; Pi′ and Qi′ are the active and
to support local loads is presented as the basic building reactive power output in per unit, respectively; ⎯P′N,
block to construct dynamic MGs. As shown in Fig. 1c, the and⎯Q′ N are the average active and reactive power
dynamic MGs are assumed as the min-MGs, and the SSW generation of DGs in per unit, respectively; P′SSW and
located at the position of dynamic POI can identify the Q′SSW are the per-unit active/reactive power flow at the
[9]
electric boundary of min-MGs . target SSW; θSSW and ESSW are the voltage phasor
The regulations developed for the system seamless mismatch and amplitude at the target SSW;
network reconfiguration are fulfilled at the secondary Eoi=Ei−∫aij(Eoi−Eoj)dt is the observed average DG
control level. Based on the different requirements of the operating voltage that can be regulated as rated (i.e.,
*
system reconfiguration, the various operation modes Eoi=Eoj=E ) using the dynamic consensus-based
[24]
(i.e., Type I, II, and III modes) included in the average observer ; as shown in Fig. 4, ωM0 and EM0
distributed secondary control level are developed are the designed control variables for active and
below. Note that each DG can only activate one reactive power sharing within the min-MG; ωM1, ωM2,
operation mode at a time to avoid introducing ωM3 and EM1, EM2, EM3 are the designed control
th
contradictory regulation efforts. For the i DG, variables for the operation of Type I, II and III modes,
the distributed secondary frequency and voltage respectively. They are detailed as follows.

Fig. 4 Proposed distributed secondary control framework

 54 Chinese Journal of Electrical Engineering, Vol.7, No.4, December 2021

Type I mode (ηi=0 and λi=0) is designed for DG and PCC, respectively.
dynamic MG operating under static topology without Since the frequency is a global signal that
*
reconfiguration requests. Under Type I mode, ωi=ω , converges fast enough in the system, the delay in
*
Pi' = P' N , and ⎯Eoi=E , Qi' =⎯Q' N as the controller adjusting the output frequency can be neglected, while
converges, meaning that the operating frequency and any delay in adjusting the output voltage magnitude is
[16]
voltage are regulated as rated, and proportional active modeled as a first-order low-pass filter
and reactive power sharing is achieved among the DGs ωc
Ei = ( E ∗ − ni Qi + Esec i ) (8)
between the neighboring min-MGs. s + ωc
Type II mode (ηi=0 and λi=1) is designed for where ωc is the cut-off frequency.
networked MG network reconfiguration when a closed Based on Eqs. (1a), (1b), (7a), (7b) and (8), the
SSW is requested to open. To guarantee seamless system system can be written as
topology transition, power flows through the target SSW Ei E *
δi = −mi δ i + ωseci (9a)
are eliminated before it re-opens. Under Type II mode, Xi
* *
besides ωi=ω and ⎯Eoi=E , it is also observed that P'SSW=0 E*
E i = − Ei − ni cos(δ i ) Ei + e + Esec i (9b)
and Q'SSW=0 as the controller converges, meaning that the Xi
*
active and reactive power flow at the target SSW is where it is assumed that sin δi≈δi and EPCC≈E are
* *
minimized for seamless re-opening. constants and so that e=(niE /Xi+1)E is constant and it
Type III mode (ηi=1 and λi=0) is designed for is not considered in the following stability analysis.
networked MG operating when requesting an open Then the linearized approximation method is used
SSW to close. To achieve seamless system topology to convert the nonlinear system into a linear system. It
transit, voltage phasors on both sides of the targeted is assumed that Ei≈E' and cos δi≈cos δ' are constants in
SSW are synchronized before the target SSW Eqs. (9a) and (9b), respectively. The linearized system
*
re-closes. Under Type III mode, besides ωi=ω , Pi' = modeling with the proposed frequency and voltage
P'N, and Qi' = Q'N, it is also observed that θ SSW=0 and regulations is thereby modeled as follows
*
ESSW=E as the controller converges, meaning that the
δ = − m M δ + ω
i i i i (10a)
sec i
phase and voltage mismatches at the target SSW are
ωc−1 Ei = − Ei − ni Ni Ei + Esec i (10b)
minimized for seamless re-closure. * *
where Mi=E′E /Xi and Ni=(E /Xi)cos δ′ are constants,
3.2 Scalable small-signal stability analysis respectively.
For presentation purposes, it is denoted that
In this paper, the time domain small-signal method is
M=diag(Mi), N=diag(Ni), m=diag(mi) and n=diag(ni).
used to approximate the nonlinear function at the
The state-space representations of Eqs. (10a) and (10b)
equilibrium point by its first-order partial derivative. Since
are expressed as
the obtained linear model parameters are closely related to
 =WX
X (11a)
the selection of the equilibrium point, the large-signal δ δ δ

ωc−1X
= WE X E (11b)
model is performed to find out the equilibrium point E

where Xδ=[δ1, …, δn, ωsec1, …, ωsecn] ; XE=[E1, …, En,

[25] T
before the small-signal stability analysis .
Esec1, …, Esecn] ; The corresponding system matrices
T
The system under study is assumed to be
inductive, and the active and reactive power flows Pi Wδ and WE are given by
and Qi at bus i are given by
[26]
⎡ −mM I ⎤
Wδ = ⎢ ⎥ (12a)
Pi =
Ei EPCC
sin δ i ⎣ kf mM − kp1H h M − kf I ⎦
(7a)
Xi ⎡ −I − nN I⎤
WE = ⎢ ⎥ (12b)
⎣ − k v T − kq1K h N 0 ⎦
2
Ei EPCC EPCC
Qi = cos δ i − (7b)
Xi Xi where Hh and Kh (h=1, 2, 3) are the coefficient
where Ei is the output voltage magnitude; EPCC is the matrices that correspond to different operation modes,
voltage magnitude at PCC; Xi and i are the equivalent respectively; It is assumed that the variables nL and nR
th
reactance and phase angle mismatch between bus i are the numbers of DGs on the left and right sides of
 Yuxi Men et al.: Distributed Control Framework and Scalable Small-signal Stability Analysis for
Dynamic Microgrids 55

the target SSW so that T=[[1]nL/nL, [1]nL,nR/nR; the min-MGs at each side of the target SSW.
[1]nR,nL/nR, [1]nR/nL] describes the average voltage ϕi By using the same method in Eq. (14), when kv=0,
th
observed at the i DG. the characteristic polynomial of WE is expressed as
The following lemmas are used for the det( sI − WE ) = det(α 2 ) ⋅ det(β 2 ) = 0 (16)
subsequent stability analysis. where α2=sI+I+nN and det(α2)=0 satisfied Re(s)<0 for
[27] 2
Lemma 1 : if M is positive definite and N is a all its roots; similarly, β2=s I+s(I+nN)+(kq1K1N) and
positive-definite scalar, then MN=NM and M+N are det(β2)=0 satisfy Re(s)<0 for all its roots according to
positive-definite. Lemma 2. Since eigenvalues are a continuous function
[16]
Lemma 2 : if both A1 and A2 are positive of matrix parameters, the system is exponentially
2
definite, the characteristics polynomial det(s I+sA1+ stable if kv>0 is sufficiently small.
A2)=0 has all its roots satisfy Re(s)<0, i.e., all In Type II mode, based on the power balance
characteristic roots are in the left-half complex plane. between generation (Pi, Pj) and consumption (PL)
[9]
Lemma 3 : if M=diag(Mi) is a positive-definite when the system operates autonomously, the active
[28]
scalar, L is a Laplacian matrix of a connected graph , power flow in per-unit P'SSW at the target SSW can be
and r is a positive constant, L+r[1]nM is positive expressed as
definite.
In Type I mode, the system matrix of the
Pssw
∑P −P
i∈nL
i L ∑P −P
i∈nR
j L
(17)
frequency regulation (h=1) is ′ =
Pssw =2 =2
⎡ −mM I ⎤
∑ *
Pssw ∑ *
Pssw ∑ *
Pssw
Wδ1 = ⎢ ⎥ (13) Therefore, the system matrix of the frequency
k
⎣ f mM − k H
p1 1M − k f I⎦
regulation (h=2) is
[1]nL [1]nR
where H1 = diag(L nL + I nL + , L nR + I nR + ); ⎡ −mM I ⎤
nR nL Wδ2 = ⎢ ⎥ (18)
mL , nR ⎣ kf mM − kp1H 2 M − kf I ⎦
LnL, nR=diag( ∑a
j =1
ij )–AnL, nR are the Laplacian matrices
where H 2 = diag(L nL , L nR ) + P.
at the each side of the target SSW, respectively. The matrix P=2kp2diag([1]nL, nL, [1]nR, nR) introduces
Referring to Schur complement, the characteristic the through active power at the target SSW with
polynomial of Wδ is proper direction. Referring to Lemma 3, H2 is
det( sI − Wδ ) = det(α1 ) ⋅ det(β1 ) = 0 (14) positive-definite so that H2M is positive-definite based
where α1=sI+kfI and det(α1)=0 satisfies Re(s)<0 for all on Lemma 1. Finally, Similar to the derivations in
2
its roots; β1=s I+s(mM)+kp1H1M and det(β1)=0 Type I mode, the system is exponentially stable if
satisfies Re(s)<0 for all the roots by referring to Mi≈Mj.
Lemma 2. Therefore, in this scenario, the system is The reactive power flow in per-unit Q'SSW at the
exponentially stable if Mi≈Mj. target SSW can be derived similarly in Eq. (17) so that
The system matrix of the voltage regulation (h=1) the system matrix of the voltage regulation (h=2) for
for Type I mode is Type II mode is
⎡ −I − nN I⎤
WE1 = ⎢ ⎥ (15) ⎡ −I − nN I⎤
⎣ − k v T − kq1K1N 0 ⎦ WE2 = ⎢ ⎥ (19)
⎣ −k v T − kq1K 2 N 0 ⎦
[1]nL
where K1 = diag(L nL + I nL + N −1 , L nR + I nR +
nR where K 2 = diag(L nL , L nR ) + Q.
[1]nR The matrix Q=2kq2diag([1]nL, nL, [1]nR, nR) introduces
N −1 ) .
nL the through reactive power at the target SSW with proper
In Eqs. (13) and (15), both H1 and K1 describe direction. Referring to Lemma 3, K2 is positive definite so
the communication among DGs within the individual that K2N is positive definite as per Lemma 1. Similar to
min-MG and the communication among DGs between Type I mode, the system is exponentially stable if kv>0 is
 56 Chinese Journal of Electrical Engineering, Vol.7, No.4, December 2021

sufficiently small and Ni≈Nj. dynamic MGs framework and the controllers
In Type III mode, since SSW is requested to open, developed for each operation mode, a 6-bus test
the system stability analysis is applied for each system with five DGs is established in Fig. 5, where
individual min-MG. according to Millman’s the system consists of two min-MGs, one SSW, and
[29]
theorem , the measured voltage phasor at the target five DGs with local loads, respectively. The requested
SSW can be written as system topology variation (Topologies #1-#4) is
n implemented in the context of dynamic MGs. Initially,
VSSW (1 + jθSSW ) = ∑c E
i =1
i i (1 + jδ i ) (20) both the main breaker and the SSW are closed, so that
system operates under grid-connected mode (Topology
n
#1). It is then assumed that the distribution feeder is
where ci=XP/Xi and (XP) − =
1
∑(X )
i =1
i
−1
represents the
disconnected from the main power grid due to a grid
system equivalent parallel reactance. contingency. The main breaker is open, and the system
Thus, the system matrix of the frequency operates autonomously (Topology #2). The networks
regulation (h=3) is are possessively reconfigured as per the system operator
requests (Topology #3 and Topology #4). The detailed
⎡ −mM I ⎤
Wδ3 = ⎢ ⎥ (21) system parameters are shown in Tab. 2.
⎣ kf mM − kp1H 3M − kf I ⎦

kθ E ′ci
where H3 = Ln + [1]n,n diag( )M−1 =
kp1 [1]1,n diag(ci )Ei
L n + θ.
The matrix θ represents the phase mismatch at
the target SSW compared to the common reference.
When Mi≈Mj, M−1 is positive definite scalar so that H3
is positive-definite referring to Lemma 3. Similar to
Type I and II modes, the system is exponentially stable
if Mi≈Mj.
The system matrix of the voltage regulation (h=3)
for Type III mode is

⎡ −I − nN I⎤
WE3 = ⎢ ⎥ (22)
⎣ −k v T − kq1K1N 0 ⎦

kv
where K 3 = Ln +
kq1
[1]n,n diag(ci )N −1 = L n + V.

The matrix V represents the voltage magnitude

mismatch at the target SSW. When Ni≈Nj, N−1 is a
diagonal positive definite scalar so that K3 is positive
definite based on Lemma 3. Similar to Type I and II
modes, the system is exponentially stable if Ni≈Nj. The
value of kv does not affect the system stability in this
mode since the average DG voltage regulation T is
disabled.

4 Case studies
To validate the performance of the proposed Fig. 5 Dynamic MGs reconfiguration schemes
 Yuxi Men et al.: Distributed Control Framework and Scalable Small-signal Stability Analysis for
Dynamic Microgrids 57

Tab. 2 System setting and control parameters Topology #4 in Fig. 5 shows the proposed controller
Parameter Value performance for system reconfiguration with SSW
Nominal operating frequency
377
closure. The requested reconfiguration is executed
ω*/Hz
with the Type III mode being activated. Specifically,
Nominal output voltage amplitude
E*/V
392 Fig. 6 shows that the SSW receives request at t=t6=22
Rated active power flow at target
s, changes the load profile at t=t7=24 s, and turns off
100
SSW P*ssw/kW the SSW at t=t8=30 s. After t>t8, the proposed
Rated reactive power flow at target
100 controller switches back to Type I mode and
SSW Q*ssw /kVar
proportional power sharing among DGs 1-5 is
Line resistance RL/Ω 0.016 8
continuously achieved.
Line inductance LL/mH 0.054

Load resistance R1, R2, R3, R4,

1.92, 3.84. 1.92, 1.28, 3.84
R5/Ω

Load inductance L1, L2, L3, L4,

1, 2, 1, 0.67, 2
L5/mH

Droop coefficients m, n 0.000 01, 0.000 02

Control gains kf, kp1, kp2, kv, kq1, kq2

8, 1, 5, 40, 80, 1 000, 2
and kθ

4.1 Case I: System dynamic response with load

change

Topology #1 in Fig. 5 indicates that the system

operates in the grid-connected mode with the main
breaker and the target SSW closed. The corresponding
time-domain waveforms are shown in Fig. 6, when
t<t1=4 s, the system frequency, and voltage are regulated
by the upstream network. Further, Topology #2 in Fig.
5 illustrates the proposed controller performance for
system operation without the request of
reconfiguration. When t≥t1=4 s in Fig. 6, the system
starts operating in islanded mode with Type I mode
being activated. When t=t2=8 s, the load profile is
changed on each bus (e.g., parallel a large impedance
load). As seen, the system frequency and voltage are
regulated as reference values, and accurate
proportional active and reactive power sharing is
achieved within each MG .
Topologies #3 in Fig. 5 explains the proposed
controller performance for system reconfiguration
with SSW opening. The requested reconfiguration is
Fig. 6 System dynamic response
executed with the proposed controller under Type II
mode being activated. First of all, SSW receives the 4.2 Case II: System stability with control parameter
change
reconfiguration request at t=t3=10 s as shown in Fig. 6,
where the power flowing through SSW is mitigated In this case, the effectiveness of the presented
before reconfiguration. Second, the load profile is small-signal stability analysis is verified using various
varied on each bus by paralleling a large load at control parameters. As mentioned, for the
t=t4=15 s. Finally, the SSW is opened at t=t5=20 s. voltage/reactive controller of both Type I and Type II
 58 Chinese Journal of Electrical Engineering, Vol.7, No.4, December 2021

modes, the system is exponentially stable if the control linearized system is illustrated leveraging specific
gain kv>0 is sufficiently small. As a consequence, the sufficient conditions, respectively. Finally, through the
voltage/reactive controller of Type II mode is taken as system reconfiguration testing and the real-time
an example to explain the consistency between the simulation results, the developed works are validated.
theoretical analysis and simulation results. When kv is
References
positive but not sufficiently small (e.g., kv=300), the
system is unstable, as shown in Fig. 7a. On the other [1] R H Lasseter. Microgrids. Proc. IEEE Power Eng. Soc.
hand, the corresponding eigenvalue distributions Winter Meet., 2002, 1: 305-308.
derived from the theoretical stability analysis is [2] F Nejabatkhah, Y W Li. Overview of power management
depicted in Fig. 7b, where the eigenvalue λ1 is located strategies of hybrid AC/DC microgrid. IEEE Trans. Power
at the right-half plane and this result agrees with the Electron., 2015, 30(12): 7072-7089.

simulation results in Fig. 7a. [3] N Hatziargyriou. Microgrids: Architectures and control.
Hoboken: Wiley, 2014.
[4] Z Wang, B Chen, J Wang, et al. Networked microgrids for
self-healing power systems. IEEE Trans. Smart Grid,
2016, 7(1): 310-319.
[5] Y Du, X Lu, J Wang, et al. Dynamic microgrids in
resilient distribution systems with reconfigurable
cyber-physical networks. IEEE J. Emerging and Selected
Topics in Power Electron., 2021, 9(5): 5192-5205.
[6] Y Du, Y Men, X Lu, et al. Coastal community resiliency
enhancement using marine hydrokinetic (MHK) resources
and networked microgrids. Technical Report at Argonne
National Laboratory, United States, 2021.
[7] Q Zhou, M Shahidehpour, A Paaso, et al. Distributed
control and communication strategies in networked
microgrids. IEEE Commun. Surv. Tutor., 2020, 22(4):
2586-2633.
[8] M E Nassar, M M A Salama. Adaptive self-adequate
microgrids using dynamic boundaries. IEEE Trans. Smart
Grid, 2016, 7(1): 105-113.
[9] Y Du, X Lu, J Wang, et al. Distributed secondary control
strategy for microgrid operation with dynamic boundaries.
Fig. 7 System operation status with control parameter change IEEE Trans. Smart Grid, 2019, 10(5): 5269-5282.
(kv=300) [10] J M Guerrero, J C Vasquez, J Matas, et al. Hierarchical
control of droop-controlled AC and DC microgrids: A
5 Conclusions general approach toward standardization. IEEE Trans. Ind.
Electron., 2011, 58(1): 158-172.
In this paper, based on the hierarchical control
[11] X Lu, J M Guerrero, K Sun, et al. Hierarchical control of
framework and the consensus-based averaging parallel AC-DC converter interfaces for hybrid microgrids.
approach, a distributed control strategy is proposed for IEEE Trans. Smart Grid, 2014, 5(2): 683-692.
dynamic MGs operation. The developed operation [12] J M Guerrero, L G de Vicuna, J Matas, et al. A wireless
modes regulate the system operating frequency and controller to enhance dynamic performance of parallel
voltage and guarantee proportional power sharing inverters in distributed generation systems. IEEE Trans.
among the DGs. Moreover, the corresponding Power Electron., 2004, 19(5): 1205-1213.
small-signal stability analysis is developed for the [13] Y W Li, C Kao. An accurate power control strategy for
presented controllers along with various operation power electronics-interfaced distributed generation units
modes, where the exponential stability of the operating in a low voltage multibus microgrid. IEEE
 Yuxi Men et al.: Distributed Control Framework and Scalable Small-signal Stability Analysis for
Dynamic Microgrids 59

Trans. Power Electron., 2009, 24(12): 2977-2988. [27] R A Horn, R A Horn, C R Johnson. Matrix analysis.
[14] Q Shafiee, J M Guerrero, J C Vasquez. Distributed Cambridge: Cambridge University Press, 1990.
secondary control for islanded microgrids: A novel [28] R Olfati-Saber, J A Fax, R M Murray. Consensus and
approach. IEEE Trans. Power Electron., 2014, 29(2): cooperation in networked multi-agent systems. Proc.
1018-1031. IEEE, 2007, 95(1): 215-233.
[15] Y Du, X Lu, H Tu, et al. Dynamic microgrids with [29] J Millman. A useful network theorem. Proc. IRE, 1940,
self-organized grid-forming inverters in unbalanced 28(9): 413-417.
distribution feeders. IEEE J. of Emerging and Selected
Topics in Power Electron., 2020, 8(2): 1097-1107. Yuxi Men (S’19) received the B.S. degree in
Faculty of Water Resources and
[16] J W Simpson-Porco, Q Shafiee, F Dörfler, et al. Hydroelectric Engineering from Xi’an
Secondary frequency and voltage control of islanded University of Technology, Xi’an, China in
2017 and the M.E. degree in Energy System
microgrids via distributed averaging. IEEE Trans. Ind.
Engineering Institute from Lehigh University,
Electron., 2015, 62(11): 7025-7038. Bethlehem, USA in 2018. He is currently
[17] Q Shafiee, Č Stefanović, T Dragičević, et al. Robust pursuing Ph.D. degree in Electrical and
Computer Engineering from Temple
networked control scheme for distributed secondary
University, Philadelphia, PA, USA. His
control of islanded microgrids. IEEE Trans. Ind. Electron., research interest includes modeling and
2014, 61(10): 5363-5374. control of inverter-dominated hybrid and networked AC and DC
microgrids.
[18] Z Cheng, J Duan, M Chow. To centralize or to distribute?
That is the question: A comparison of advanced microgrid Yuhua Du (S’17-M’21) received the B.S
management systems. IEEE Ind. Electron. Magazine, degree in Electrical Engineering from Xi’an
Jiaotong University, China in 2013 and Ph.D.
2018, 12(1): 6-24. degree in Electrical and Computer
[19] Y Du, H Tu, H Yu, et al. Accurate consensus-based Engineering from North Carolina State
University, USA in 2019. He was a Research
distributed averaging with variable time delay in support
Aide with Argonne National Laboratory in
of distributed secondary control algorithms. IEEE Trans. 2018. He is currently a Postdoctoral Fellow in
Smart Grid, 2020, 11(4): 2918-2928. Temple University. His research interests
include voltage source converter modeling and control, microgrid
[20] D P Spanos, R Olfati-Saber, R M Murray. Dynamic
distributed secondary control and development of microgrid
consensus on mobile networks. Proc. IFAC World Congr., hardware-in-the-loop testbed.
2005: 1-6.
Xiaonan Lu (S’12-M’13) received his B.E.
[21] Z Li, M Shahidehpour. Small-signal modeling and and Ph.D. degrees in Electrical Engineering
stability analysis of hybrid AC/DC microgrids. IEEE from Tsinghua University, Beijing, China, in
2008 and 2013, respectively. From
Trans. Smart Grid, 2019, 10(2): 2080-2095.
September 2010 to August 2011, he was a
[22] F M Atay. Consensus in networks under transmission guest Ph.D. student at the Department of
delays and the normalized Laplacian. IFAC Proc., 2010, Energy Technology, Aalborg University,
Denmark. From October 2013 to December
43(2): 277-282. 2014, he was a Postdoc Research Associate
[23] Y J Kim, J Wang, X Lu. A framework for load service at the Department of Electrical Engineering
and Computer Science, University of
restoration using dynamic change in boundaries of
Tennessee, Knoxville. From January 2015 to July 2018, he was with
advanced microgrids with synchronous-machine DGs. Argonne National Laboratory, first as a Postdoc Appointee and then as
IEEE Trans. Smart Grid, 2018, 9(4): 3676-3690. an Energy Systems Scientist. In July 2018, he joined the College of
Engineering at Temple University as an Assistant Professor. His
[24] Y Du, X Lu, B Chen, et al. Distributed average
research interests include modeling and control of power electronic
observation in inverter dominated dynamic microgrids. in inverters, hybrid AC and DC microgrids, and real-time
2020 IEEE Energy Conversion Congress and Exposition hardware-in-the-loop simulation.
Dr. Lu is the Associate Editor of IEEE Transactions on Industrial
(ECCE), 2020: 4627-4633. Electronics, the Associate Editor of IEEE Transactions on Industry
[25] X Wang, Y Song, M Irving. Modern power systems Applications, and the Editor of IEEE Transactions on Smart Grid. He
analysis. New York: Springer, 2010. serves as the Vice Chair of the Industrial Power Converters Committee
(IPCC) in the IEEE Industry Applications Society (IAS). He is also the
[26] J Machowski, J W Bialek, J R Bumby. Power system recipient of the 2020 Young Engineer of the Year Award in the IEEE
dynamics. 2nd ed. Hoboken: Wiley, 2008. Philadelphia Section.