Beruflich Dokumente
Kultur Dokumente
Power Quality
Jian Sun
Department of Electrical, Computer and Systems Engineering
Rensselaer Polytechnic Institute, Troy, NY 12180-3590, USA
Abstract This paper presents an overview of ac power Once an averaged model is established for a converter, it can be
electronic system modeling and analysis methods. Characteristics linearized about a steady-state operation point to develop a small-
of such systems are first discussed, and their applications and signal model such that linear system theories can be applied to
significance in renewable energy and distributed generation (DG), study system stability and performance. The most widely used
transportation, and other mobile and autonomous power systems system analysis method is to partition the system into a source and
are highlighted. Typical stability and power quality problems
encountered in these systems are demonstrated by measurements a load subsystem, and to determine system stability by applying the
from several practical systems. Small-signal analysis methods, Nyquist stability criterion to the ratio between the source output
such as phasor-based analysis, transformation into dq reference impedance to the load input impedance [12]. Dynamic interactions
frame, reduced-order modeling, and harmonic linearization that between a converter and its input filter can be analyzed using the
are applicable to ac power electronic systems, are then reviewed same method [13]. The impedance-based system stability require-
and compared in terms of their applicability and limitations. ments can also be used to define impedance specifications for
Small-signal impedance modeling of rectifiers and inverters by individual converters such as system stability can be guaranteed by
harmonic linearization is reviewed in some detail. Limitation of the design [14]. However, such specifications are almost always too
existing impedance-based system criterion when applied to conservative, resulting in increased size and cost of system design;
current-source systems is identified and an extension of the theory
design for system stability remains a topic for future research.
is presented for application to grid-parallel inverters.
Ac power electronic system is a relatively new development.
I. INTRODUCTION The electric power grid is by far the largest power system that has
been in existence since George Westinghouse and Nikola Tesla
successfully demonstrated a polyphase electrical supply system at
As power converters (such as power supplies, motor drives and
the Worlds Columbian Exposition in Chicago in 1893. Despite the
UPS) that are traditionally built from discrete devices and compo-
large number of power supplies and other power electronic devices
nents become commodity products, research in power electronics
connected to the grid, however, the utility grid remained, until
is shifting towards miniaturization for low- and medium-power
recent years, largely an electromechanical system where power is
applications, as well as to integration and control of power
almost exclusively generated by large synchronous generators and
electronic systems encompassing multiple converters. For the
mostly consumed by linear and passive loads such as incandescent
purpose of this work, a power electronic system is defined as a
light bulbs and motors. The percentage of power processed by
power system where a significant portion of system power is
power electronic circuits and control has been relatively small and
processed by power electronic circuits and control such that their
their effects on overall system operation is negligible for most
operation has a dominant effect on system performance. The form
practical purposes. This, however, is changing rapidly in recent
of power distribution in such a system can be dc, ac, or a combi-
years due to the proliferation of renewable energy and distributed
nation of both, and the system will be referred to as dc, ac, or
generation technologies on the source side, and energy-efficient
hybrid power electronic system, respectively.
technologies such as solid-state lighting and variable-speed drives
Dc power electronic system has been a research subject in power on the load side. Similar changes are taking place in other, mobile
electronics for almost three decades now, started from satellite and and autonomous power systems such as those found on ground and
space station applications [1, 2] in the 1970s and 1980s, and air transportation systems. As a result, there is a great demand for
proliferated in the 1990s driven by telecom and computer applica- understanding of stability and power quality issues in ac power
tions where it is commonly referred to as distributed power electronic systems, as well as for methodologies and tools that can
systems (DPS) [3, 4]. Electric and hybrid-electric vehicle became be applied to analyze and integrate such complex systems.
another application of dc power electronic system technology in
more recent years. Many works have been published on the This paper presents an overview of ac power electronic system
modeling, analysis, and design of such systems [6-14], and a stability and power quality problems as we know today, and review
standard analysis method based on averaging and linearization has available system modeling and analysis methods. The next section
been developed. Since a switching converter represents a discon- presents several practical ac power electronic systems and discuss
tinuous system that is difficult to deal with analytically, averaging their characteristics. Possible system stability and power quality
is a necessary step to first remove the switching operation such that problems are demonstrated by measurement results in Section III.
an approximate, continuous (averaged) model can be developed. Different system analysis methods are reviewed and compared in
Various averaging techniques have been discussed in the literature, Section IV, followed by an overview of the recently developed
ranging from straightforward state-space averaging [6-9] to more harmonic linearization method (Section V) and an extension of the
sophisticated reduced-order modeling techniques for resonant and impedance-based method to current-source systems (Section VI).
soft-switching converters [10, 11]. Section VII summarizes the paper.
Zs +
Source Vs + Vl Zl Load
b)
Fig. 6. A fuel-cell inverter output voltage and current in a) standalone operation
with a resistive load, and b) grid-parallel operation. Fig. 7. Impedance representation of an interconnected source-load system.
The impedance-based method discussed above can be applied to In general, the response of the current to such perturbation contains
ac or hybrid ac-dc power electronic systems as well. The difficulty, harmonics at different frequencies because of nonlinearity. For the
however, is in obtaining the source and load impedance models. A purpose of small-signal analysis, one can ignore all harmonics and
converter operating with ac input or/and output exhibits similar consider only the response at the perturbation frequency:
nonlinear/discontinuous characteristics as in a dc-dc converter.
Although averaging can still be applied to remove model disconti- i(t) = I 1 sin ( 1 t + 1 ) + I sin ( p t + p + p ) + (7)
nuity, the resulting averaged model is nonlinear, and direct linear-
ization of such a model is impossible due to the fact that the The small-signal impedance at the perturbation frequency can be
converter operates along a periodic time-varying trajectory rather determined once I and p have been calculated:
than at a dc point as in dc power electronic systems. V
Several methods are available for dealing with the time-varying Z(j p) = ------- p (8)
I
trajectories in ac power electronic systems. These include the
phasor-based method widely used in conventional power system With phasor-based models, however, small-signal perturbation
analysis, the transformation into the dq reference frame, and the is introduced to the phasor, that is, to the amplitude (and phase) of
reduced-order modeling. These methods will be reviewed in the the voltage. As shown below, amplitude modulation is equivalent
following subsections to compare their utility and limitations. to superimposing two harmonic voltages at the same time:
A. Phasor Analysis v(t) = [ V 1 + V sin ( p t + p ) ] sin 1 t = V 1 sin 1 t +
Phasor analysis is an effective method for linear ac circuits and V V
------- cos [ ( p 1 )t + p ] ------- cos [ ( p + 1 )t + p ]
systems where the excitation is a single-frequency sine wave. The 2 2
phasor concept was originally developed for sinusoidal steady-state
analysis of linear circuits [30]. Conventional power system As the result, the amplitude of the current will be modulated by the
analysis has used it to model transmission lines and loads for same frequency
system stability analysis as well [31, 32]. The underlying
assumption for such analysis is that the phasor amplitude and phase i(t) = [ I 1 + I sin ( p t + p + p ) ] sin ( 1 t + 1 ) +
angle can vary over time and, hence, can both be treated as
dynamic variables. The concept of such time-varying phasors was and the ratio of the voltage perturbation to the resulting current
more formally developed in [33], where they were referred to as perturbation can be obtained and written in the same form as (8).
dynamic phasors to distinguish from conventional phasors used in However, this function is not compatible with the definition of
steady-state analysis. small-signal impedance discussed before. Therefore, although
phasor-based models can be linearized, the resulting small-signal
Phasor-based dynamic models can be directly linearized by voltage-to-current transfer functions cannot be used as impedances
using conventional small-signal analysis methods because both for use with the impedance-based system analysis method outlined
the amplitude and phase angle of a phasor are constant in at the beginning of this section. This was discussed in more detail
steady-state operation. One limitation of the conventional in [37] by using a single-phase PFC converter as example.
dynamic phasor theory is that it only applies to linear circuits
and systems. This is solved by the so-called generalized The impedance-based method enables a hierarchical approach to
averaging theory [34-36], in which each dynamic variable is system analysis by allowing a system to be divided into different
represented by a complex Fourier series with time-varying subsystems and the system behavior to be studied at their intercon-
amplitudes and phase angles. The Fourier series can be nection points. This is very attractive for large systems, particularly
truncated to include only the fundamental or with additional those with variable configurations and/or parameters, because
harmonic terms, and each such term can be considered a changes in one subsystem do not affect the impedance of the rest of
dynamic phasor [35]. A dynamic model can be developed for a the system, and adding or disconnecting a source or load can be
nonlinear circuit based on such Fourier representation by easily accommodated in the system model by adding or removing
applying the harmonic balance principle [34]. This method has the corresponding impedance element. Input and output impedance
been applied to model FACTS and other nonlinear devices for models can also be established from experimental measurements if
power flow and dynamic stability study [35, 36]. analytical models are not available. None of these is possible with
One limitation of the phasor-based method is that the models are phasor-based models because their incompatibility with small-
only valid below the fundamental frequency (or, in the case of signal impedance definition. As the result, the system will have to
generalized averaging, the frequency of the highest harmonic be treated as a whole and any change in any part of the system will
included in the model.) Therefore, power quality problems, such as require reformulation of the entire system model. This has not been
those discussed in III.B and III.C, cannot be studied using such a problem for traditional power system stability study, which
models. Another limitation, which is more subtle and has generated focuses at the transmission level where things do not change often.
some confusions in the past, is that small-signal impedance of Most renewable energies and other distributed resources, on the
nonlinear circuit cannot be obtained from linearizations of phasor- other hand, are integrated into the utility grid at low- to medium-
based models. To determine small-signal impedance of a nonlinear voltage distribution level where things change frequently, making
circuit, one first injects a voltage perturbation as defined below: the impedance-based approach much more attractive than the
phasor-based approach. The same is true for mobile and other
v(t) = V 1 sin 1 t + V sin ( p t + p ) autonomous power systems reviewed in Section II, where there are
(6) no transmission lines and there is usually only one voltage which