Page 1 of 11
Journal for Studies in Management and Planning
Available at http://edupediapublications.org/journals/index.php/JSMaP/
e-ISSN: 2395-0463
Volume 02 Issue 9
September 2016
Available online: http://edupediapublications.org/journals/index.php/JSMaP/ P a g e | 67
Grid Voltage Synchronization for Distributed
Generation Systems under Grid FaultConditions
Mr.V.BALU & RACHAKONDA CHANDRASHEKAR
Abstract—The actual grid code requirements for the grid
connection of distributed generation systems, mainly wind and
photovoltaic (PV) systems, are becoming very demanding. The
transmission system operators (TSOs) are especially concerned
about the low-voltage-ride-through requirements. Solutions based
on the installation of STATCOMs and dynamic voltage regulators
(DVRs), as well as on advanced control functionalities for the
existing power converters of distributed generation plants, have
contributed to enhance their response under faulty and distorted
scenarios and, hence, to fulfill these requirements. In order to
achieve satisfactory results with such systems, it is necessary to
count on accurate and fast grid voltage synchronization algo- rithms, which are able to work under unbalanced and distorted
conditions. This paper analyzes the synchronization capability of
three advanced synchronization systems: the decoupled double
synchronous reference frame phase-locked loop (PLL), the dual
second order generalized integrator PLL, and the three-phase
enhanced PLL, designed to work under such conditions. Although
other systems based on frequency-locked loops have also been
developed, PLLs have been chosen due to their link with dq0
controllers. In the following, the different algorithms will be pre- sented and discretized, and their performance will be tested in an
experimental setup controlled in order to evaluate their accuracy
and implementation features.
Index Terms—Electric variable measurements, electrical engi- neering, frequency estimation, frequency-locked loops, harmonic
analysis, monitoring, synchronization.
I. INTRODUCTION
THE power share of renewable energy-based generation
sys- tems is supposed to reach 20% by 2030, where wind and
photovoltaic (PV) systems are assumed to be the most outstand- ing examples of integration of such systems in the electrical
network.
The increased penetration of these technologies in the
electrical network has reinforced the already existing concern
among the transmission system operators (TSOs) about their
influence in the grid stability; as a consequence, the grid
connection standards are becoming more and more restrictive
for distribution generation systems in all countries
In the actual grid code requirements (GCRs), special con- straints for the operation of such plants under grid voltage fault
conditions have gained a great importance. These requirements
determine the fault boundaries among those through which a
grid-connected generation system shall remain connected to the
network, giving rise to specific voltage profiles that specify the
depth and clearance time of the voltage sags that they must
withstand. Such requirements are known as low voltage ride
through (LVRT) and are described by a voltage versus time
characteristic [7].
Although the LVRT requirements in the different standards
are very different, as shown in [8], the first issue that generation
systems must afford when a voltage sag occurs is the limitation
oftheirtransient response, in order to avoid its protective discon- nection from the network. This is the case, for instance, of fixed
speed wind turbines based on squirrel cage induction generators,
where the voltage drop in the stator windings can conduct the
generator to an overspeed tripping, as shown in [9]. Likewise,
variable speed wind power systems may lose controllability in
the injection of active/reactive power due to the disconnection
of the rotor side converter under such conditions [10], [11].
Likewise, PV systems would also be affected by the same lack
of current controllability.
Solutions based on the development of auxiliary systems,
such as STATCOMs and dynamic voltage regulators (DVRs),
have played a decisive role in enhancing the fault ride through
(FRT) capability of distributed generation systems, as demon- strated in [12]–[16]. Likewise, advanced control functionalities
for the power converters have also been proposed [17], [18].In
any case, a fast detection of the fault contributes to improving
the effects of these solutions; therefore, the synchronization
algorithms are crucial.
In certain countries, the TSOs also provide the active/reactive
power pattern to be injected into the network during a voltage
sag; this is the case for the German E-on [2] and the Spanish
Red Eléctrica Española (REE) [3]. Thistrend has been followed
by the rest of the TSOs; moreover, it is believed that this
operation requirement will be extended, and specific demands
for balanced and unbalanced sags will arise in the following
versions of the grid codes worldwide [19].
Regarding the operation of the distributed generation systems
under balanced and unbalanced fault conditions, relevant con- tributions, such as [20]–[29], can be found in the literature.
These solutions are based on advanced control systems that
need to have accurate information of the grid voltage variables
in order to work properly, something that has prompted the im- portance of grid synchronization algorithms. In power systems,
the synchronous reference frame PLL (SRF PLL) is the most
extended technique for synchronizing with three-phase systems
[30]. Nevertheless, despite the fact that the performance of SRF
PLL is satisfactory under balanced conditions, its response can
be inadequate under unbalanced, faulty, or distorted conditions
[31]–[33].
In this paper, three improved and advanced grid synchroniza- tion systems are studied and evaluated: the decoupled double
synchronous reference frame PLL (DDSRF PLL) [34], the dual
Page 2 of 11
Journal for Studies in Management and Planning
Available at http://edupediapublications.org/journals/index.php/JSMaP/
e-ISSN: 2395-0463
Volume 02 Issue 9
September 2016
Available online: http://edupediapublications.org/journals/index.php/JSMaP/ P a g e | 68
second order generalized integrator PLL (DSOGI PLL), [35]
and the three-phase enhanced PLL (3phEPLL) [36]. Their per- formance, computational cost, and reliability of the amplitude
and phase detection of the positive sequence of the voltage,
under unbalanced and distorted situations, have been evaluated
according to experimental grid fault patterns extracted from
[37] and [38], which have been reproduced in a real scaled
electrical network.
In the following sections, the discrete representation of each
PLL will be detailed (see Section IV), due to its great impor- tance in the final implementation of the control, after a brief
description of the different structures (see Section III). Finally,
their behavior will be tested in an experimental setup (see
Section VI), and their performance will be discussed, partic- ularly taking into account the accuracy of the positive-sequence
detection and their computational cost (see Section VII) when
considering different faulty scenarios, covering the response in
front of sags, frequency changes, and harmonic immunity.
II. GRID SYNCHRONIZATION SPECIFICATIONS
BASED ON GCR
Even though several works are published within the field of
grid synchronization, almost all of them are centered on ana- lyzing the individual dynamic performance of each proposal,
without first determining a time response window within the
dynamic behavior of the system under test, which would be
considered to be satisfactory.
In this paper, in order to evaluate the response of the grid
synchronization topologies under test, a common performance
requirement for all the structures has been established in this
section, considering the needs that can be derived from the
LVRT requirements.
Despite the fact that the detection of the fault can be carried
out with simpler algorithms, as shown in [39] and [40], the
importance of advanced grid synchronization systemsliesin the
necessity of having accurate information about the magnitude
and phase of the grid voltage during the fault, in order to inject
the reactive power required by the TSO.
In the German standard [2], it is stated that voltage control
must take place within 20 ms after the fault recognition, by
providing a reactive current on the low voltage side of the
Fig. 1. E-on voltage support requirement in the event of grid fault.
Fig. 2. REE voltage support requirement in the event of grid fault.
generator transformer to at least 2% of the rated current for each
percent of the voltage dip, as shown in Fig. 1. 100% reactive
power delivery must be possible, if necessary.
A similar condition is given in the Spanish grid code, where
the wind power plants are required to stop drawing inductive
reactive power within 100 ms of a voltage drop and be able to
inject full reactive power after 150 ms, as shown in Fig. 2.
Considering these demands, this paper will consider that the
estimation of the voltage conditions will be carried out within
20–25 ms, as this target permits it to fulfill the most restrictive
requirements, in terms of dynamical response, available in
the grid codes. This condition will be extended to frequency
estimation; although this parameter is more related to secondary
control algorithms than LVRT, the same time window between
20 and 25 ms will be considered in this work for the detection
of the disturbance.
III. DESCRIPTION OF THE THREE
SYNCHRONIZATION SYSTEMS
Many of the positive-sequence detection algorithms are
based on SRF PLLs [32]. Despite having a good response under
balanced conditions, their performance becomes insufficient in
unbalanced faulty grids (95% of cases), and their good opera- tion is highly conditioned to the frequency stability, which is
incompatible with the idea of a robust synchronization system.
Many authors have discussed different advanced models, which
Page 3 of 11
Journal for Studies in Management and Planning
Available at http://edupediapublications.org/journals/index.php/JSMaP/
e-ISSN: 2395-0463
Volume 02 Issue 9
September 2016
Available online: http://edupediapublications.org/journals/index.php/JSMaP/ P a g e | 69
abc
Fig. 3. DDSRF-PLL block diagram.
are able to overcome the problems of the classical PLL, using
frequency and amplitude adaptive structures which are able to
deal with unbalanced, faulty, and harmonic-polluted grids. In
the framework of these topologies, three PLL structures will be
discussed and evaluated in this paper.
A. DDSRF PLL
The DDSRF PLL, published in [34] and [41], was developed
for improving the conventional SRF PLL. This synchronization
system exploits two synchronous reference frames rotating at
Fig. 4. DSOGI-PLL block diagram.
as explained in [35]. The diagram of the DSOGI PLL is shown
in Fig. 4. As it can be noticed, the ISC method is implemented
by the positive-sequence calculation block.
To apply the ISC method, it is necessary to have a set of
signals, vα–vβ, representing the input voltage vector on the αβ
stationary reference frame together with another set of signals,
qvα–qvβ, which are in quadrature and lagged with respect
to vα–vβ. In the DSOGI PLL, the signals to be supplied to
the ISC method are obtained by using a dual second order
generalized integrator (DSOGI), which is an adaptive band- pass filter based on the generalized integrator concept [42].
At its output, the DSOGI provides four signals, namely, v
r
and v
r , which are filtered versions of vα and vβ, respectively, and qvr r
the fundamental utility frequency, one counterclockwise and ofv
r
α and qvβ
, which are the in-quadrature versions r
another one clockwise, in order to achieve an accurate detection
of the positive- and negative-sequence components of the grid
voltage vector when it is affected by unbalanced grid faults. The
diagram of the DDSRF PLL is shown in Fig. 3.
When the three-phase grid voltage is unbalanced, the funda- mental positive-sequence voltage vector appears as a dc voltage
on the dq+1 axes of the positive-sequence SRF and as ac
voltages at twice the fundamental utility frequency on the dq−1
axes of the negative-sequence SRF. In contrast, the negative- sequence voltage vector will cause a dc component on the
negative-sequence SRF and an ac oscillation on the positive- sequence SRF. Since the amplitude of the oscillation on the
positive-sequence SRF matches the dc level on the negative- sequence SRF and vice versa, a decoupling network is applied
to signals on the dq positive/negative SRF axes in order to
cancel out such ac oscillations. Low-pass filters (LPFs) in
Fig. 3 are responsible for extracting the dc component from
the signal on the decoupled SRF axes. These dc components
collect information about the amplitude and phase angle of the
positive- and negative-sequence components of the grid voltage
vector.
Finally, the PI controller of the DDSRF PLL works on the
decoupled q-axis signal of the positive-sequence SRF (v
∗ )
and performs the same function as in an SRF PLL, aligning the
α and vβ
.
A conventional SRF PLL is applied on the estimated
positive-sequence voltage vector, v
+ , to make this synchro- nization system frequency adaptive. In particular, the v
+ volt- age vector is translated to the rotating SRF, and the signal on
the q-axis, v
+
, is applied at the input of the loop controller.
As a consequence, the fundamental grid frequency (ωr
) and
the phase angle of the positive-sequence voltage vector (θ
+
t )
are estimated by this loop. The estimated frequency for the
fundamental grid component is fed back to adapt the center
frequency ωr of the DSOGI.
C. 3phEPLL
The enhanced phase-locked loop (EPLL) is a synchroniza- tion system that has proven to provide good results in single- phase synchronization systems [43]. An EPLL is essentially
an adaptive bandpass filter, which is able to adjust the cutoff
frequency as a function of the input signal. Its structure was
later adapted for the three-phase case [44], in order to detect
the positive-sequence vector of three-phase signals, obtaining
the 3phEPLL that is represented in Fig. 5.
In this case, each phase voltage is processed independently
by an EPLL. This block filters the input signal and generates
two sinusoidal outputs of the same amplitude and frequency, positive-sequence voltage with the d-axis. This signal is free of v
r r ◦ r
n and jvn
, the second one being 90 with respect to vn
.
ac components due to the effect of the decoupling networks; the
bandwidth ofthe loop controller can be consequently increased.
The resulting signals constitute the input for the computational
unit. Owing to these in-quadrature signals, the instantaneous
positive-sequence voltage component, v
+
, can be estimated
B. DSOGI PLL
The operating principle of the DSOGI PLL for estimating
the positive- and negative-sequence components of the grid
voltage vectors is based on using the instantaneous symmetrical
component (ISC) method on the αβ stationary reference frame,
by means of using the ISC method.
IV. DISCRETE IMPLEMENTATION
The performance ofthe differentstructures under test isreally
dependent on their final digital implementation, particularly on
