IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO.

12, DECEMBER 2014 6421

Modeling and Analysis of Harmonic Stability in an

AC Power-Electronics-Based Power System
Xiongfei Wang, Member, IEEE, Frede Blaabjerg, Fellow, IEEE, and Weimin Wu

Abstract—This paper addresses the harmonic stability caused Continuous research efforts have been made to investigate
by the interactions among the wideband control of power convert- the instability in the ac power-electronics-based power systems.
ers and passive components in an ac power-electronics-based power However, many of the studies focus on the low-frequency os-
system. The impedance-based analytical approach is employed and
expanded to a meshed and balanced three-phase network which is cillations caused by the constant power control for convert-
dominated by multiple current- and voltage-controlled inverters ers as constant power loads [7]–[9] or constant power genera-
with LCL- and LC-filters. A method of deriving the impedance ra- tors [10], [11], the phase-locked loop (PLL) for grid-connected
tios for the different inverters is proposed by means of the nodal converters [12]–[14], and the droop-based power control for
admittance matrix. Thus, the contribution of each inverter to the converters in islanded microgrids [15]–[17]. Apart from such
harmonic stability of the power system can be readily predicted
through Nyquist diagrams. Time-domain simulations and experi- oscillations associated with the outer power control and grid
mental tests on a three-inverter-based power system are presented. synchronization loops, the interactions of the fast inner current
The results validate the effectiveness of the theoretical approach. or voltage control loops may also result in harmonic instabil-
Index Terms—Current-controlled inverter, harmonic stability, ity phenomena, namely harmonic-frequency oscillations (typ-
impedance-based analysis, power-electronics-based power system, ically from hundreds of hertz to several kilohertz) due to the
voltage-controlled inverter. inductive or capacitive behavior of converters in this frequency
range [18]–[22]. Furthermore, this harmonic instability may also
be generated or magnified by the control of converters in an in-
I. INTRODUCTION teraction with harmonic resonance conditions introduced by the
high-order power filters for converters and parasitic capacitors of
HE PROPORTION of power electronics apparatus in elec-
T tric power systems has grown in recent years, driven by
the rapid development of renewable power sources and variable-
power cables [23]–[26]. Such phenomena have been frequently
reported in renewable energy systems and high-speed railway
[2], [3], [5], [26]–[30], and are challenging the system stability
speed drives [1]. As a consequence, power-electronics-based and power quality. It is, therefore, important to develop the effec-
power systems are becoming important components of electri- tive modeling and analysis approach for the harmonic stability
cal grids, such as renewable power plants [2], [3], microgrids [4], problem in the ac power-electronics-based power systems.
and electric railway systems [5]. These systems possess supe- A general analytical approach for the harmonic stability prob-
rior features to build the modern power grids, including the lem is to build the state-space model of the power system, and
full controllability, the sustainability, and the improved effi- identify the oscillatory modes based on the eigenvalues and
ciency, but also bring new challenges. High-order harmonics eigenvectors of the state matrix [31]. However, unlike conven-
tend to be aggravated by the high-frequency switching oper- tional power systems where the dynamics are mainly determined
ation of power converters, which may trigger the parallel and by the rotating machines, the small time constants of power
series resonance in the power system [6]. The interactions of the converters require the detailed models of loads and network
wideband control systems for power converters with each other dynamics in the power-electronics-based power systems [7].
and with passive components may manifest instability phenom- Thus, the formulation of the system matrices may become com-
ena of a power-electronics-based power system in the different plicated, and the virtual resistors are usually needed to avoid
frequency ranges [2]–[5]. the ill-conditioned problems [32]. To overcome these limits, the
component connection method (CCM) is introduced for the sta-
bility analysis of ac power systems including the high-voltage
Manuscript received September 8, 2013; revised December 17, 2013; ac- direct current transmission lines [33]. The CCM is basically
cepted February 8, 2014. Date of publication February 14, 2014; date of current a particular form of state-space models where the power sys-
version August 13, 2014. This work was supported by European Research Coun- tem components and network dynamics are separately modeled
cil (ERC) under the European Union’s Seventh Framework Program (FP/2007–
2013)/ERC Grant Agreement 321149-Harmony. Recommended for publication by a set of two vector–matrix equations. This results in the
by Associate Editor J. Clare. sparsity of the state equations and reduces the computation bur-
X. Wang and F. Blaabjerg are with the Department of Energy Technol- den of formulating the system transfer matrices. Furthermore,
ogy, Aalborg University, 9220 Aalborg East, Denmark (e-mail: [email protected];
[email protected]). the component interactions and the critical system parameters
W. Wu is with the Department of Electrical Engineering, Shanghai Maritime for the different oscillatory modes can be more easily deter-
University, Shanghai, China (e-mail: [email protected]). mined [34], [35].
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. Apart from the state-space analysis, the impedance-based
Digital Object Identifier 10.1109/TPEL.2014.2306432 approach, which is originally introduced for the design of input

0885-8993 © 2014 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution
requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
6422 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 12, DECEMBER 2014

filters in dc–dc converters [36], provides another attractive way

to analyze the harmonic stability problem. Similar to the CCM,
the impedance-based analysis is also built upon the models of
components or subsystems. However, instead of systematically
analyzing the eigenproperties of the state matrix as in the
CCM and other state-space models, the impedance-based
approach locally predicts the system stability at each point of
connection (PoC) of the component, based on the ratio of the
output impedance of the component and the equivalent system
impedance [37]. Thus, the formulation of the system matrices
can be avoided, and the contribution of each component to the
system stability can be readily assessed in the frequency domain.
Also, the output impedance of the component or subsystem
can be accordingly reshaped to stabilize the overall power
system [38]. Therefore, the impedance-based method provides Fig. 1. Simplified one-line diagram of a three-phase power-electronics-based
a more straightforward and design-oriented stability analysis ac power system.
compared to the CCM and other state-space models. Several
applications of the impedance-based approach for the harmonic
stability analysis can be found in ac power-electronics-based
power systems, e.g., the cascaded sour-load inverter system [39],
the parallel grid-connected converters with LCL-filters [22],
and the parallel uninterruptible power supply inverters with
LC-filters [25]. However, in all these cases, the network dy-
namics are often overlooked, which affect the derivation of the
equivalent system impedance, and few of them have considered
the systems including multiple voltage- and current-controlled
converters. Fig. 2. Block diagram of the CCM applied for the built power system.
This paper attempts to fill in this gap by expanding the
impedance-based analysis to a three-phase meshed and bal-
The current-controlled inverters operate with a unity power fac-
anced power system, in which a voltage-controlled and two
tor. In such a system, the presence of shunt capacitors in the
current-controlled inverters with LC- and LCL-filters are inter-
LC- and LCL-filters of inverters and power cables brings in res-
connected. The harmonic instability that results from the in-
onant frequencies, which may interact with the inner voltage and
teractions among the inner voltage and current control loops
current control loops of voltage- and current-controlled inverters
of these inverters and passive components is closely studied.
resulting in the harmonic-frequency oscillations and unexpected
By the help of the nodal admittance matrix, a simple method
harmonic distortion. On the other hand, the dynamic interactions
of deriving the impedance ratio at the PoC of inverters is pro-
between the inner control loops of inverters may also trigger the
posed to assess how each inverter contributes to the harmonic
existing resonant frequencies in the power system. This conse-
instability of the power system. The theoretical analysis in the
quently necessitates the use of CCM or impedance-based anal-
frequency domain is performed based on the linearized models
ysis to reveal how inverters interact with each other and with
of inverters with inner control loops. The results are validated
the harmonic resonance conditions in the system.
by the nonlinear time-domain simulations and experiments on
Since this research is concerned with the harmonic instabil-
a three-inverter-based power system.
ity owing to the dynamics of inner control loops, the dc-link
voltages of inverters are assumed to be constant. Also, the grid
II. SYSTEM MODELING AND ANALYSIS TECHNIQUES synchronization loop for the current-controlled inverters is de-
This section first describes the structure of the built power- signed with the bandwidth lower than the system fundamental
electronics-based power system in this study, and then reviews frequency. Thus, only the subsynchronous oscillation may be
the CCM and the impedance-based approach for modeling and induced by the grid synchronization [12]–[14]. Under these as-
analysis of harmonic stability in the power system. sumptions, only the inner voltage and current control loops are
modeled in this study. The previous studies have shown that the
A. System Description inner control loops themselves can be simply modeled by the
single-input and single-output transfer functions in the station-
Fig. 1 shows a simplified one-line diagram for the balanced
ary αβ-frame [20], [25]–[27], [39]–[43].
three-phase power-electronics-based power system which is
considered in this study, where a voltage-controlled and two
B. Component Connection Method
current-controlled inverters are interconnected as a meshed
power network through power cables. The voltage-controlled Fig. 2 shows the block diagram of the CCM applied for the
inverter regulates the system frequency and voltage amplitude. built power system, where the CCM decomposes the overall
WANG et al.: MODELING AND ANALYSIS OF HARMONIC STABILITY IN AN AC POWER-ELECTRONICS-BASED POWER SYSTEM 6423

system into three subsystems by inverters and the connection

network. Two current-controlled inverters are modeled by the
Norton equivalent circuits [22], while the voltage-controlled
inverter is represented by the Thevenin equivalent circuit [23].
Consequently, the composite inverter model can be derived in
the following:
y(s) = Gcl (s)u(s) − Gcd (s)d (s), (1)
where y(s) and d(s) are the output vector and disturbance vector
of inverters, respectively
y(s) = [V1 (s), ig 2 (s), ig 3 (s)]T (2)
T
d(s) = [ig 1 (s), V2 (s), V3 (s)] . (3)
Gcl (s) and Gcd (s) denote the closed-loop reference-to-
output and the closed-loop disturbance-to-output transfer ma-
trices, respectively, which depict the unterminated dynamic be-
havior of inverters seen from the PoC and can be given by
Gcl (s) = diag[Gclv ,1 (s), Gcli,2 (s), Gcli,3 (s)] (4)
Gcd (s) = diag[Zov ,1 (s), Yoi,2 (s), Yoi,3 (s)] (5)
Fig. 3. Impedance-based equivalent models for (a) the voltage- and (b) current-
where Gclv ,1 (s), Gcli,2 (s), and Gcli,3 (s) are the voltage and controlled inverters.
current reference-to-output transfer functions of voltage- and
current-controlled inverters, respectively. Zov ,1 (s), Yoi,2 (s),
disturbance-to-output transfer functions of inverters as transfer
and Yoi,3 (s) are the closed-loop output impedance and
matrices in the impedance-based approach [38]. Instead, the
admittances of the voltage- and current-controlled inverters,
impact of a given inverter on the overall system stability is de-
respectively.
termined by a minor feedback loop composed by the ratio of
The dynamic of the connection work can be represented by
the inverter output impedance or admittance, Zov ,1 or Yoi,m ,
the transfer matrix Gnw (s) as
and the equivalent impedance or admittance for the rest of the
d(s) = Gnw (s)y(s). (6) system, Zlv ,1 or Yli,m [40]. Furthermore, due to the scalar-type
models for inner control loops of inverters, the impedance ratios
Thus, the closed-loop response of the overall power system can be depicted by single-input and single-output transfer func-
can be derived as tions [39], which significantly simplify the harmonic stability
y(s) = [I + Gcd (s)Gnw (s)]−1 Gcl (s)u(s) (7) analysis compared to the CCM and other state-space models.
Hence, the impedance-based approach is preferred in this work.
where the transfer matrix [I +Gcd (s)Gnw (s)]−1 predicts the From Fig. 3, the closed-loop transfer functions of inverters
harmonic instability of the power system, provided that the can be derived as follows:
unterminated behavior of inverters Gcl (s) are stable. In addition
to the eigenvalues technique, the transfer matrix can also be V1 (s) 1
∗ = Gclv ,1 (s) Z (s)
(8)
evaluated in the multivariable frequency domain by means of V1 (s) 1 + ov ,1Z l v , 1 (s)
the generalized Nyquist stability criterion [44], [45].
It is obvious that the main superior feature of the CCM com- ig m (s) 1
∗
= Gcli,m (s) Y (s)
. (9)
pared to the other state-space models is to decompose the power ig m (s) 1 + oi,m Y l i , m (s)
system into multiple decentralized feedback loops by inverters,
and thus the effect of inverters controllers and the associated It is clear that if the unterminated dynamic behavior of invert-
physical components on the system oscillatory modes is more ers, Gclv ,1 and Gcli,m , are stable, the stability of voltage and
intuitively revealed. Furthermore, the decentralized stabilizing current at the PoC of inverters will be merely dependent on the
control loops may be developed by means of the CCM [46]. minor feedback loops composed by the following impedance
ratios:
C. Impedance-Based Approach
Zov ,1 (s) Yoi,m (s)
Fig. 3 depicts the equivalent circuits of the voltage- and Tm v (s) = , Tm c (s) = (10)
Zlv ,1 (s) Yli,m (s)
current-controlled inverters applied for the impedance-based
analysis. It is interesting to note that this approach also sepa- which are also termed as the minor feedback loop gains [40].
rately models the internal dynamics of inverters by means of the Based on these minor feedback loop gains, the specifications of
output impedance and admittance. However, differently from the inverters output impedances to preserve the system stability can
CCM, there is no need of stacking the reference-to-output and be derived.
6424 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 12, DECEMBER 2014

Gci (s) = Kpi , GPW M (s) = e−1.5T s s (14)

where Gci (s) is the P current controller and GPW M (s) de-
picts the effect of the digital computation delay (Ts ) and the
pulse width modulation (PWM) delay (0.5 Ts ) [42]. Then, by
including the outer voltage control loop, the voltage reference-
to-output transfer function and closed-loop output impedance
are derived in the following:
V1 (s) = Gclv ,1 (s)V1∗ (s) − Zov ,1 (s)ig 1 (s) (15)

Tv (s) Zoi (s)

Gclv ,1 (s) = , Zov ,1 (s) = (16)
1 + Tv (s) 1 + Tv (s)
Fig. 4. Voltage-controlled inverter with the multiloop control scheme.

Gv i (s)Gci (s)GPW M (s)ZC f ,1 (s)

Tv (s) = (17)
ZL f ,1 (s) + ZC f ,1 (s) + Gci (s)GPW M (s)

ZC f ,1 (s) [ZL f ,1 (s) + Gci (s)GPW M (s)]

Zoi (s) = (18)
ZL f ,1 (s) + ZC f ,1 (s) + Gci (s)GPW M (s)
Fig. 5. Block diagram of the multiloop voltage control system.

Kr v s
III. MODELING OF INVERTERS Gv i (s) = Kpv + (19)
s2 + ω12
A. Voltage-Controlled Inverters where Gv i (s) is the PR voltage controller and ω1 denotes the
Fig. 4 depicts the simplified one-line diagram of the voltage- system fundamental frequency. Tv (s) is the open-loop gain of
controlled inverter and the multiloop voltage control scheme. the control system. Zoi (s) is the open-loop output impedance
The control system is implemented in the stationary frame, in- obtained with the outer voltage control loop open and the inner
cluding the inner proportional (P) current controller and outer current loop closed.
proportional resonant (PR) voltage controller. It is worth men-
tioning that the three-phase inverters without neutral wire can be B. Current-Controlled Inverters
transformed into two independent single-phase systems in the Fig. 6 shows the simplified one-line diagram of the current-
stationary αβ-frame [42]. Furthermore, due to the assumptions controlled inverters and the associated control system. The
of the constant dc-link voltage and balanced three-phase opera- grid-side inductor current of the LCL-filter is controlled for
tion, the voltage-controlled inverter can be linearized based on the inherent resonance damping effect of the computation and
the LC-filter and modeled as a real scalar system by single-input modulation delays [47]. The PR controller in the stationary αβ-
and single-output transfer functions [23]–[25], [42], [43]. frame is adopted for the grid current control. The synchronous
Fig. 5 shows the block diagram of the multiloop voltage con- reference frame PLL is used for the grid synchronization [48].
trol system, where the following two transfer functions are used It is important to mention that the PLL has an important ef-
to describe the effect of the inverter output voltage VPW M ,1 and fect on the output admittance of the inverter in addition to the
grid current ig 1 on the filter inductor current iL 1 , respectively: current control loop. The inclusion of PLL effect will intro-
1 duce an unbalanced three-phase inverter model which needs to
YL i (s) = be described by transfer matrices [12]. However, it has been
ZL f ,1 (s) + ZC f ,1 (s)
found in recent studies that the PLL only affects the output
ZC f ,1 (s) admittance within its control bandwidth where the negative re-
Gii (s) = (11)
ZL f ,1 (s) + ZC f ,1 (s) sistance behavior may be introduced [14], and this problem can
where ZL f ,1 (s) and ZC f ,1 (s) are the impedances of the filter be avoided by reducing the bandwidth of PLL [13]. Hence, in
inductor and capacitor, respectively. Thus, the dynamic behavior this work, the bandwidth of the PLL is designed to be lower
of the inner current control loop can be given by than the system fundamental frequency in order not to bring in
any harmonic-frequency oscillations. Thus, the current control
Tc,1 (s) ∗ Gii (s) loop itself is linearized based on the LCL-filter and is modeled
iL 1 (s) = iL 1 (s) + ig 1 (s) (12)
1 + Tc,1 (s) 1 + Tc,1 (s) as a real scalar system.
Fig. 7 depicts the block diagram of the grid current control
where Tc,1 (s) is the open-loop gain of the inner current control
loop. The LCL-filter in itself is a two-input and single-output
loop, which is expressed as
system, where the PoC voltage Vm and the inverter output volt-
Tc,1 (s) = Gci (s)GPW M (s)YL i (s) (13) age VPW M ,m are the two inputs and grid current ig m is the
WANG et al.: MODELING AND ANALYSIS OF HARMONIC STABILITY IN AN AC POWER-ELECTRONICS-BASED POWER SYSTEM 6425

Fig. 8. Impedance-based equivalent circuit for the built power system.

Fig. 6. Current-controlled inverter (m = 2, 3) with a grid current control
scheme.

where Gcg i (s) is the PR current controller.

IV. ANALYSIS OF HARMONIC STABILITY

To perform the impedance-based analysis of harmonic stabil-
ity in the built power system, a method of deriving the impedance
Fig. 7. Block diagram of the grid current control loop.
ratio at the PoC of each inverter is proposed in this section. It
is noted from Fig. 3 that the equivalent system impedance seen
output. Thus, the following two transfer functions are used to from the PoC of the inverter is indispensable in the impedance
model the LCL-filter plant: ratio. Hence, an equivalent system impedance derivation proce-
 dure is developed by the help of the nodal admittance matrix
ig m (s) 
Yg i , m (s) = = and used in the following analysis.
VP W M , m (s) V m (s )= 0

ZC f , m (s)
ZC f , m (s)ZL f , m (s) + ZL g , m (s)ZL f , m (s) + ZC f , m (s)ZL g , m (s) A. System Equivalent Circuit
(20) Fig. 8 depicts the impedance-based equivalent circuit for the
 power system shown in Fig. 1. The power cables are represented
ig m (s) 
Yo , m (s) = = as the Π-section models to include the effect of parasitic capac-
Vm (s)  V P W M , m (s )= 0 itors. Also, to facilitate the formulation of nodal admittance
ZL f , m (s) + ZC f , m (s) matrix, the Thevenin model of the voltage-controlled inverter is
ZC f , m (s)ZL f , m (s) + ZL g , m (s)ZL f , m (s) + ZC f , m (s)ZL g , m (s) converted to the Norton circuit, where Yov ,1 = 1/Zov ,1 .
To obtain the equivalent system impedance at each PoC of
(21)
the inverter, the system nodal admittance matrix (Ync ) including
where ZL f ,m (s), ZL g ,m (s), and ZC f ,m (s) denote the the closed-loop output admittances of inverters, as highlighted
impedances of the LCL-filter inductors and capacitor, respec- by the dot–dashed line in Fig. 8, is derived as
tively. Yo,m (s) is the open-loop output admittance. From Fig. 7,
⎡ ⎤ ⎡ ⎤
the closed-loop response of the current control loop can be de- Yov ,1 Gclv ,1 V1∗ V1
rived as follows: ⎢ ⎥ ⎢ ⎥
⎢ Gcli,2 i∗g 2 ⎥ = Ync ⎢ V2 ⎥ (26)
⎣ ⎦ ⎣ ⎦
ig m (s) = Gcli,m (s)i∗g m (s) − Yoi,m (s)Vm (s) (22) Gcli,3 i∗g 3 V3
Tc,m (s) Yo,m (s)
Gcli,m (s) = , Yoi,m (s) = (23)
1 + Tc,m (s) 1 + Tc,m (s)
⎡ ⎤
where Gcli,m (s) and Yoi,m (s) are the current reference-to- Yo v , 1 + 2Yp −Yp −Yp
output transfer function and closed-loop output admittance, re- ⎢ ⎥
⎢ −Yp Yo i , 2 + 2Yp + YL , 1 −Yp ⎥
Yn c = ⎢ ⎥
spectively. Tc,m (s) is the open-loop gain of current control loop, ⎣ ⎦
which is given by −Yp −Yp Yo i , 3 + 2Yp + YL , 2
(27)
Tc,m (s) = Gcg i (s)GPW M (s)Yg i,m (s) (24) where Yp is the cable admittance, and YL ,1 and YL ,2 are the
Kr g i s admittances for loads 1 and 2 connected to Buses 2 and 3, re-
Gcg i (s) = Kpg i + (25)
s2 + ω12 spectively. Then, by inverting Ync , the nodal impedance matrix
6426 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 12, DECEMBER 2014

(Znc ) is given by TABLE I

MAIN ELECTRICAL PARAMETERS OF THE POWER SYSTEM
⎡ ⎤
Z11 Z12 Z13
⎢ ⎥
−1
Znc = Ync =⎢ Z
⎣ 21
Z22 Z23 ⎥
⎦ (28)
Z31 Z32 Z33

where the elements are expanded in (29) as shown at the bottom

of the page.
It is worth noting that the diagonal elements of the nodal
impedance matrix are the equivalent system impedances seen
from the equivalent current sources of inverters, which include
the closed-loop output impedances of inverters and the equiva-
lent system impedances at the PoC of inverters. Consequently,
the equivalent system impedances at the PoC of inverters can be
derived by the following relationship:

Zov ,1 Zlv ,1 1
Z11 = , Z22 =
Zov ,1 + Zlv ,1 Yoi,2 + Yli,2
1
Z33 = . (30)
Yoi,3 + Yli,3

Furthermore, comparing the term Z11 in (30) with (8), it Together with (26), (28), and (29), the closed-loop response
is seen that the closed-loop response of the voltage-controlled of the equivalent current sources can be derived as
inverter can also be expressed by ⎡ ⎤ ⎡ ⎤
ig ,1 Yov ,1 Gclv ,1 V1∗
⎢ ⎥
⎢ ig 2 ⎥ = Yno Znc ⎢⎣ Gcli,2 i∗g 2 ⎦ .
⎥
(33)
V1 (s)
= Gclv ,1 (s)Z11 (s)Yov ,1 (s) (31) ⎣ ⎦
V1∗ (s) ig 3 ∗
Gcli,3 ig 3

where Z11 Yov ,1 is the closed-loop gain of the minor feed- Comparing (33) with (9), it can be found that the second
back loop. Similarly, the closed-loop responses of the current- and third diagonal elements of the closed-loop transfer matrix
controlled inverters can also be found by means of the nodal Yno Znc represent the closed-loop gains of the minor feedback
admittance matrix (Yno ) derived at the PoC of inverters, which loops for the current-controlled inverters.
is highlighted by the dashed line in Fig. 8
B. Impedance-Based Analysis
⎡ ⎤⎡ ⎤
⎡ ⎤ 2Yp −Yp −Yp V1
ig 1 Table I lists the main electrical parameters of the studied
⎢ ⎥⎢ ⎥
⎣ ig 2 ⎦ = ⎢ −Yp 2Yp + YL ,1 −Yp ⎥ ⎣ V2 ⎦ . (32) power system. Table II gives the parameters of the voltage
⎣ ⎦
ig 3 −Yp −Yp 2Yp + YL ,2 and current controllers for inverters. The cables use the same
V3 Π-section models, and two current-controlled inverters are also

 
Yn o designed with the same parameters for the sake of simplicity.

Yoi,2 Yoi,3 + Yoi,2 YL ,2 + Yoi,3 YL ,1 + 2Yoi,2 Yp + 2Yoi,3 Yp + YL ,1 YL ,2 + 2YL ,1 Yp + 2YL ,2 Yp + 3Yp2

Z11 =
Ynom
Yov ,1 Yoi,3 + Yov ,1 YL ,2 + 2Yov ,1 Yp + 2Yoi,3 Yp + 2YL ,2 Yp + 3Yp2
Z22 =
Ynom
Yov ,1 Yoi,2 + Yov ,1 YL ,1 + 2Yov ,1 Yp + 2Yoi,2 Yp + 2YL ,1 Yp + 3Yp2
Z33 =
Ynom
Yoi,3 Yp + YL ,2 Yp + 3Yp2 Yoi,2 Yp + YL ,1 Yp + 3Yp2 Yov ,1 Yp + 3Yp2
Z12 = Z21 = , Z13 = Z31 = , Z23 = Z32 =
Ynom Ynom Ynom
Ynom = 3Yp2 (Yov ,1 + Yoi,2 + Yoi,3 + YL ,1 + YL ,2 ) + 2Yp (Yov ,1 Yoi,2 + Yov ,1 Yoi,3 + Yov ,1 YL ,1 + Yov ,1 YL ,2 )
+ (2Yp + Yov ,1 )(Yoi,2 Yoi,3 + Yoi,2 YL ,2 + Yoi,3 YL ,1 + YL ,1 YL ,2 ) (29)
WANG et al.: MODELING AND ANALYSIS OF HARMONIC STABILITY IN AN AC POWER-ELECTRONICS-BASED POWER SYSTEM 6427

TABLE II
CONTROLLER PARAMETERS OF INVERTERS

Fig. 9. Frequency responses for the open-loop gains of the voltage-controlled

inverter, T v (s), and the current-controlled inverters, T c (s).

Fig. 10. Nyquist plots of the minor feedback loop gains of inverters in the
unstable case. (a) Full view. (b) Zoom in (−1, j0).
Fig. 9 shows the frequency responses of the inner control
loop gains, Tv (s) and Tc,m (s), for the voltage- and current-
controlled inverters, respectively. The stable unterminated reduced (Kpi = 5, Kpv = 0.05). It is clear that all minor feed-
dynamic behaviors of inverters at the PoC are observed with back loops become stable. This fact again indicates that the
the controller parameters listed in Table II. This provides a the- voltage-controlled inverter with the controller parameters given
oretical basis for using the minor feedback loops to assess the in Table II causes the harmonic instability in the power system
harmonic stability of the power system. even if it has a stable unterminated dynamic behavior.
Fig. 10 depicts the Nyquist diagrams of the minor feed-
back loop gains of the voltage- and current-controlled inverters. V. SIMULATION AND EXPERIMENTAL RESULTS
Since the impedance ratios are described by the single-input and
To validate the impedance-based stability analysis in the fre-
single-output transfer functions, the Nyquist stability criterion
quency domain, the power system in Fig. 1 is built in the nonlin-
can be directly used to evaluate the interactions among the inner
ear time-domain simulations by using the MATLAB and PLECS
control loops of inverters and the network dynamics. It is seen
Blockset, and the experimental tests.
that the minor feedback loop for the voltage-controlled inverter
is unstable, whereas the minor feedback loops for the current-
controlled inverters are stable. This implies that the impedance A. Simulation Results
interaction at the PoC of the voltage-controlled inverter (Bus 1) Fig. 12 shows the simulated grid currents of inverters with the
causes the harmonic instability in the power system, and the ad- electrical constants and controller parameters given in Tables I
mittance interactions at the PoC of current-controlled inverters and II. The simulated bus voltages are shown in Fig. 13. It
have no contribution to the harmonic instability. is seen that the harmonic-frequency oscillation occurs in the
Fig. 11 shows the Nyquist diagrams of the minor feedback power system, which confirms the frequency-domain analysis
loop gains after adjusting the controller parameters for the in Fig. 10.
voltage-controlled inverter. In this case, the proportional gains In contrast, Fig. 14 shows the simulated grid currents of in-
of the P current controller and the PR voltage controller are verters after reducing the proportional gains of controllers for
6428 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 12, DECEMBER 2014

Fig. 13. Simulated bus voltages in the unstable case.

Fig. 11. Nyquist plots of the minor feedback loop gains of inverters in the
stable case. (a) Full view. (b) Zoom in (−1, j0).

Fig. 14. Simulated grid currents of inverters in the stable case.

the voltage-controlled inverter (Kpi = 5, Kpv = 0.05). The

simulated voltage at each bus of the system is shown in Fig. 15.
It is obvious that the harmonic instability phenomenon shown
in Figs. 12 and 13 becomes stabilized in this case. This agrees
well with the theoretical analysis in Fig. 11 and also verifies that
the harmonic instability in the power system is caused by the
Fig. 12. Simulated grid currents of inverters in the unstable case. voltage-controlled inverters rather than the current-controlled
inverters.
WANG et al.: MODELING AND ANALYSIS OF HARMONIC STABILITY IN AN AC POWER-ELECTRONICS-BASED POWER SYSTEM 6429

Fig. 15. Simulated bus voltages in the stable case.

Fig. 17. Measured grid currents of inverters in the unstable case.

and maximum 1000-V root mean square is used for voltage

measurement.
First, the unstable case that is based on the controller param-
eters given in Table II is tested. Figs. 17 and 18 depict the mea-
sured grid currents of inverters and the measured bus voltages,
Fig. 16. Hardware picture of the built laboratory test setup.
respectively. It is observed that the same harmonic-frequency
oscillation arises in the experimental test as in the simulation
results shown in Figs. 12 and 13. Furthermore, in both the sim-
ulation and experimental test, this harmonic-frequency oscilla-
B. Experimental Results
tion propagates into the whole power system and leads to the
Fig. 16 shows a hardware picture of the built power- unexpected harmonic disturbance to the current control loops of
electronics-based power system in laboratory. Three Danfoss the current-controlled inverters. Consequently, even though no
frequency converters are employed to operate as a voltage- unstable condition is brought by the current-controlled invert-
controlled inverter and two current-controlled inverters. The ers, as shown in Fig. 10, the harmonic-frequency oscillations
control algorithms of inverters are implemented in DS1006 are still present in the grid currents of current-controlled in-
dSPACE system, in which the DS5101 digital waveform output verters. Hence, it is difficult to identify the source of harmonic
board is adopted to generate switching pulses in synchronous instability in the power system by merely observing the simu-
with the sampling circuit [42]. The current transducer LA 55-P lations and experimental results. The impedance-based analysis
and voltage transducer LV 25-P are used to acquire current and in the frequency domain thus becomes important to reveal how
voltage signals, respectively, for the digital control circuit. each inverter contributes to the harmonic stability of the power
In the experimental tests, the bus voltages of the power system.
system and grid currents of inverters are recorded by using Then, the stable case with the reduced proportional gains of
the digital oscilloscope with the 250-kS/s sampling rate and the controllers for the voltage-controlled inverters (Kpi = 5,
the 100-kHz effective bandwidth. The current probe with the Kpv = 0.05) is tested. Figs. 19 and 20 show the measured grid
100-kHz bandwidth is adopted for the current measurement. The currents of inverters and bus voltages, respectively, where the
differential mode voltage probe with the 25-MHz bandwidth stable operation of the power system is clearly observed. This
6430 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 12, DECEMBER 2014

Fig. 18. Measured bus voltages in the unstable case. Fig. 19. Measured grid currents of inverters in the stable case.

matches well with the simulations results shown in Figs. 14 and

15, and further validates the impedance-based stability analysis
in Fig. 11. Together with Figs. 17 and 18, the experimental tests
point out that the interaction between the voltage control loop
of the voltage-controlled inverter and the rest of network results
in harmonic-frequency oscillations propagating into the power
system.

VI. CONCLUSION
This paper discusses a modeling and analysis procedure for
the harmonic stability problem in the ac power-electronics-
based power systems. Two attractive stability analysis methods,
i.e., the CCM and impedance-based approach, have been briefly
reviewed. It has been found that the impedance-based approach
provides a more computationally efficient and design-oriented
analysis tool than the CCM. The impedance-based approach
was expanded to a three-phase meshed and balanced network,
where the harmonic instability resulting from the interactions of
the inner control loops for the voltage- and current-controlled
inverters was studied. A method for deriving impedance ratios
was developed based on the system nodal admittance matrix.
Time-domain simulations and experimental results have shown
that the proposed approach could be a promising way to ad-
dress the harmonic instability in the ac power-electronics-based
power systems. Fig. 20. Measured bus voltages in the stable case.
WANG et al.: MODELING AND ANALYSIS OF HARMONIC STABILITY IN AN AC POWER-ELECTRONICS-BASED POWER SYSTEM 6431

REFERENCES [23] X. Wang, F. Blaabjerg, Z. Chen, and W. Wu, “Resonance analysis in par-
allel voltage-controlled distributed generation inverters,” in Proc. Annu.
[1] F. Blaabjerg, Z. Chen, and S. B. Kjaer, “Power electronics as efficient IEEE Appl. Power Electron. Conf. and Expo., 2013, pp. 2977–2983.
interface in dispersed power generation systems,” IEEE Trans. Power [24] X. Wang, F. Blaabjerg, and Z. Chen, “Autonomous control of inverter-
Electron., vol. 19, no. 5, pp. 1184–1194, Sep. 2004. interfaced distributed generation units for harmonic current filtering and
[2] P. Brogan, “The stability of multiple, high power, active front end voltage resonance damping in an islanded microgrid,” IEEE Trans. Ind. Appl.,
sourced converters when connected to wind farm collector system,” in vol. 50, no. 1, pp. 452–461, Jan./Feb. 2014.
Proc. Euro. Conf. Power Electron. and Appl., 2010, pp. 1–6. [25] M. Corradini, P. Mattavelli, M. Corradin, and F. Polo, “Analysis of parallel
[3] J. H. Enslin and P. J. Heskes, “Harmonic interaction between a large num- operation of uninterruptible power supplies loaded through long wiring
ber of distributed power inverters and the distribution network,” IEEE cables,” IEEE Trans. Power Electron., vol. 25, no. 4, pp. 1046–1054, Apr.
Trans. Power Electron., vol. 19, no. 6, pp. 1586–1593, Nov. 2004. 2010.
[4] J. Rocabert, A. Luna, F. Blaabjerg, and P. Rodriguez, “Control of power [26] S. Zhang, S. Jiang, X. Lu, B. Ge, and F. Z. Peng, “Resonance issues and
converters in AC microgrids,” IEEE Trans. Power Electron., vol. 27, damping techniques for grid-connected inverters with long transmission
no. 11, pp. 4734–4749, Nov. 2012. cable,” IEEE Trans. Power Electron., vol. 29, no. 1, pp. 110–120, Jan.
[5] E. Mollerstedt and B. Bernhardsson, “Out of control because of 2014.
harmonics—An analysis of the harmonic response of an inverter loco- [27] J. Agorreta, M. Borrega, J. Lopez, and L. Marroyo, “Modeling and control
motive,” IEEE Control Syst. Mag., vol. 20, no. 4, pp. 70–81, Aug. 2000. of N-paralleled grid-connected inverters with LCL filter coupled due to
[6] Z. Shuai, D. Liu, J. Shen, C. Tu, Y. Cheng, and A. Luo, “Series and parallel grid impedance in PV plants,” IEEE Trans. Power Electron., vol. 26,
resonance problem of wideband frequency harmonic and its elimination no. 3, pp. 770–785, Mar. 2011.
strategy,” IEEE Trans. Power Electron., vol. 29, no. 4, pp. 1941–1952, [28] L. H. Kocewiak, J. Hjerrild, and C. L. Bak, “Wind turbine converter con-
Apr. 2014. trol interaction with complex wind farm systems,” IET Renewable Power
[7] N. Bottrell, M. Prodanovic, and T. C. Green, “Dynamic stability of a Generation, vol. 7, no. 4, pp. 380–389, Jul. 2013.
microgrid with an active load,” IEEE Trans. Power Electron., vol. 28, [29] Z. He, H. Hu, Y. Zhang and S. Gao,“Harmonic resonance assessment to
no. 11, pp. 5107–5119, Nov. 2013. traction power-supply system considering train model in China high-speed
[8] N. Jelani, M. Molinas, and S. Bologani, “Reactive power ancillary service railway,” IEEE Trans. Power Del., to be published, 2013
by constant power loads in distributed AC systems,” IEEE Trans. Power [30] H. Lee, C. Lee, and G. Jang, “Harmonic analysis of the Korean high-
Del., vol. 28, no. 2, pp. 920–927, Apr. 2013. speed railway using the eight-port representation model,” IEEE Trans.
[9] B. Wen, D. Boroyevich, P. Mattavelli, Z. Shen, and R. Burgos, “Ex- Power Del., vol. 26, no. 2, pp. 979–986, Apr. 2006.
perimental verification of the generalized Nyquist stability criterion for [31] P. Kundur, Power System Stability and Control. New York: McGraw-
balanced three-phase AC systems in the presence of constant power Hill, 1994.
loads,” in Proc. IEEE Energy Convers. Congr. and Expo., 2012, pp. 3926– [32] N. Pogaku, M. Prodanovic, and T. C. Green, “Modeling, analysis and
3933. testing of an inverter-based microgrid,” IEEE Trans. Power Electron.,
[10] T. Messo, J. Jokipii, J. Puukko, and T. Suntio, “Determining the value vol. 22, no. 2, pp. 613–625, Mar. 2007.
of DC-link capacitance to ensure stable operation of a three-phase photo- [33] G. Gaba, S. Lefebver, and D. Mukhedkar, “Comparative analysis and
voltaic inverter,” IEEE Trans. Power Electron., vol. 29, no. 2, pp. 665–673, study of the dynamic stability of AC/DC systems,” IEEE Trans. Power
Feb. 2014. Syst., vol. 3, no. 3, pp. 978–985, Aug. 1988.
[11] J. M. Espi and J. Castello, “Wind turbine generation system with optimized [34] S. Lefebver and R. Dube, “Control system analysis and design for an
DC-link design and control,” IEEE Trans. Ind. Electron., vol. 60, no. 3, aerogenerator with eigenvalue methods,” IEEE Trans. Power Syst., vol. 3,
pp. 919–929, Mar. 2013. no. 4, pp. 1600–1608, Nov. 1988.
[12] L. Harnefors, M. Bongiorno, and S. Lundberg, “Input-admittance calcu- [35] S. Lefebver, “Tuning of stabilizers in multi-machine power systems,”
lation and shaping for controlled voltage-source converters,” IEEE Trans. IEEE Trans. Power App. Syst., vol. PAS-102, no. 2, pp. 290–299, Feb.
Ind. Electron., vol. 54, no. 6, pp. 3323–3334, Dec. 2007. 1983.
[13] B. Wen, D. Boroyevich, P. Mattavelli, Z. Shen, and R. Burgos, “Influence [36] R. Middlebrook, “Input filter considerations in design and application of
of phase-locked loop on input admittance of three-phase voltage-source switching regulators,” in Proc. IEEE Ind. Appl. Soc. Annu. Meet., pp. 366–
converters,” in Proc. 28th Annu. IEEE Appl. Power Electron. Conf. and 382, 1976.
Expo., 2013, pp. 897–904. [37] M. Belkhayat,“Stability criteria for ac power systems with regulated
[14] T. Messo, J. Jokipii, A. Makinen, and T. Suntio, “Modeling the grid loads,” Ph.D. dissertation, School Electr. Comput. Eng., Purdue Univ.,
synchronization induced negative-resistor-like behavior in the output Dec. 1997.
impedance of a three-phase photovoltaic inverter,” in Proc. IEEE Fourth [38] J. Sun, “Small-signal methods for AC distributed power systems–A re-
Intl. Symp. Power Electron. for Distributed Generations System, 2013, view,” IEEE Trans. Power Electron., vol. 24, no. 11, pp. 2545–2554, Nov.
pp. 1–8. 2009.
[15] E. A. Coelho, P. Cortizo, and P. F. Gracia, “Small signal stability for [39] R. Turner, S. Walton, and R. Duke, “A case study on the application of the
parallel-connected inverters in stand-alone ac supply systems,” IEEE Nyquist stability criterion as applied to interconnected loads and sources
Trans. Ind. Appl., vol. 38, no. 2, pp. 533–542, Mar./Apr. 2002. on grids,” IEEE Trans. Ind. Electron., vol. 60, no. 7, pp. 2740–2749, Jul.
[16] M. Marwali, J. W. Jung, and A. Keyhani, “Stability analysis of load sharing 2013.
control for distributed generation systems,” IEEE Trans. Energy Convers., [40] J. Sun, “Impedance-based stability criterion for grid-connected invert-
vol. 22, no. 3, pp. 737–745, Sep. 2007. ers,” IEEE Trans. Power Electron., vol. 26, no. 11, pp. 3075–3078, Nov.
[17] E. Barklund, N. Pogaku, M. Prodanvoic, C. H. Aramburo, and 2011.
T. C. Green, “Energy management in autonomous microgrid using [41] L. Harnefors, “Modeling of three-phase dynamic systems using com-
stability-constrained droop control of inverters,” IEEE Trans. Power Elec- plex transfer functions and transfer matrices,” IEEE Trans. Ind. Electron.,
tron., vol. 23, no. 5, pp. 2346–2352, Sep. 2008. vol. 54, no. 4, pp. 2239–2248, Aug. 2007.
[18] J. D. Ainsworth, “Harmonic instability between controlled static convert- [42] S. Buso and P. Mattavelli, Digital Control in Power Electronics. CA,
ers and a.c. networks,” Proc. Inst. Elect. Eng., vol. 114, pp. 949–957, Jul. USA: Morgen & Claypool Publishers, 2006.
1967. [43] S. Hiti, D. Boroyevich, and C. Cuadros, “Small-signal modeling and con-
[19] F. Wang, J. Duarte, M. Hendrix, and P. Ribeiro, “Modelling and analysis trol of three-phase PWM converters,” in Proc. Conf. Rec. IEEE Ind. Appl.
of grid harmonic distortion impact of aggregated DG inverters,” IEEE Soc. Annu. Meet., 1994, pp. 1143–1150.
Trans. Power Electron., vol. 26, no. 3, pp. 786–797, Mar. 2011. [44] Z. Yao, P. G. Therond, and B. Davat, “Stability analysis of power systems
[20] R. Turner, S. Walton, and R. Duke, “Stability and bandwidth implications by the generalized Nyquist criterion,” in Proc. Int. Conf. Control, Mar.
of digitally controlled grid-connected parallel inverters,” IEEE Trans. Ind. 1994, vol. 1, pp. 739–744.
Electron., vol. 57, no. 11, pp. 3685–3694, Nov. 2010. [45] A. MacFarlane and I. Postlethwaite, “The generalized Nyquist stability
[21] L. Harnefors, L. Zhang, and M. Bongiorno, “Frequency-domain passivity- criterion and multivariable root loci,” Int. J. Control, vol. 25, pp. 81–127,
based current controller design,” IET Power Electron., vol. 1, no. 4, Jan. 1977.
pp. 455–465, Dec. 2008. [46] S. Lefebvre, D. P. Carroll, and R. A. DeCarlo, “Decentralized power mod-
[22] X. Wang, F. Blaabjerg, M. Liserre, Z. Chen, J. He, and Y. Li, “An active ulation of multiterminal HVDC systems,” IEEE Trans. Power App. Syst.,
damper for stabilizing power-electronics-based AC systems,” IEEE Trans. vol. PAS-100, no. 7, pp. 3331–3339, Jul. 1981.
Power Electron., vol. 29, no. 7, pp. 3318–3329, July 2014.
6432 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 29, NO. 12, DECEMBER 2014

[47] J. Yin, S. Duan, and B. Liu, “Stability analysis of grid-connected inverter Weimin Wu received the B.S. degree from
with LCL filter adopting a digital single-loop controller with inherent the Department of Electrical Engineering, Anhui
damping characteristic,” IEEE Trans. Ind. Inform., vol. 9, no. 2, pp. 1104– University of Science and Technology, Huainan,
1112, May 2013. China, in 1997, the M.S. degree from the Depart-
[48] S. K. Chung, “A phase tracking system for three phase utility interface ment of Electrical Engineering, Shanghai University,
inverters,” IEEE Trans. Power Electron., vol. 15, no. 3, pp. 431–438, May Shanghai, China, in 2001, and the Ph.D. degree from
2000. the College of Electrical Engineering, Zhejiang Uni-
versity, Hangzhou, China, in 2005.
He worked as a Research Engineer in Delta Power
Electronic Center, Shanghai, from July 2005 to June
2006. Since July 2006, he has been a Faculty Member
at Shanghai Maritime University, where he is currently an Associate Professor
in the Department of Electrical Engineering. He was a Visiting Professor in the
Xiongfei Wang (S’10–M’13) received the B.Sc. de- Center for Power Electronics Systems, Virginia Polytechnic Institute and State
gree from Yanshan University, Qinhuangdao, China, University, Blacksburg, from September 2008 to March 2009. From November
in 2006, and the M.Sc. degree from the Harbin In- 2011 to February 2014, he has been a Visiting Professor in the Department of
stitute of Technology, Harbin, China, in 2008, both Energy Technology, working at the Center of Reliable Power Electronics. He
in electrical engineering, and the Ph.D. degree in en- has coauthored about 60 papers in technical journals and conferences. He holds
ergy technology from Aalborg University, Aalborg, of five patents. His areas of interests include power converters for renewable
Denmark, in 2013. energy systems, power quality, smart grid, and energy storage technology.
He was a Visiting Student at Hanyang Univer-
sity, Seoul, Korea, from 2007 to 2008. Since 2009,
he has been with the Aalborg University, Aalborg,
Denmark, where he is currently an Assistant Profes-
sor in the Department of Energy Technology. His research areas include the
power electronics for renewable energy systems, distributed generations, mi-
crogrids, and power quality issues. He is a Member of CIGRE/CIRED JWG
C4.24.

Frede Blaabjerg (S’86–M’88–SM’97–F’03) re-

ceived the Ph.D. degree in 1992 from Aalborg
University, Aalborg, Denmark.
He was with ABB-Scandia, Randers, Denmark,
from 1987 to 1988. He became an Assistant Professor
in 1992, an Associate Professor in 1996, and a Full
Professor of power electronics and drives in 1998.
His current research interests include power electron-
ics and its applications such as in wind turbines, PV
systems, reliability, harmonics, and adjustable speed
drives.
Dr. Blaabjerg has received 15 IEEE Prize Paper Awards, the IEEE PELS Dis-
tinguished Service Award in 2009, the EPE-PEMC Council Award in 2010, and
the IEEE William E. Newell Power Electronics Award 2014. He was an Editor-
in-Chief of the IEEE TRANSACTIONS ON POWER ELECTRONICS from 2006 to
2012. He has been a Distinguished Lecturer for the IEEE Power Electronics
Society from 2005 to 2007 and for the IEEE Industry Applications Society
from 2010 to 2011.