See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/283518093

A PMU-based fast real-time sub-synchronous oscillation detection

application

Article · July 2015

DOI: 10.1109/EEEIC.2015.7165461

CITATIONS READS

8 214

6 authors, including:

Luigi Vanfretti José Luis Dominguez-Garcia

Rensselaer Polytechnic Institute IREC Catalonia Institute for Energy Research
236 PUBLICATIONS   2,302 CITATIONS    69 PUBLICATIONS   1,258 CITATIONS   

SEE PROFILE SEE PROFILE

Muhammad Shoaib Almas

KTH Royal Institute of Technology
38 PUBLICATIONS   409 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

INCITE: Innovative controls for renewable source integration into smart energy systems View project

Power System Test Cases for EMT-type Simulation Studies View project

All content following this page was uploaded by Luigi Vanfretti on 08 September 2018.

The user has requested enhancement of the downloaded file.

 A PMU-Based Fast Real-Time Sub-synchronous Oscillation Detection
Application
Luigi Vanfretti, Maxime Baudette, José Luís Domínguez-García,
Austin White, Muhammad Shoaib Almas, Jan Ove Gjerde,

Abstract—The share of wind power has strongly increased fast dynamics, due to the low data rate of conventional meter-
in electricity production, raising several issues concerning its ing and the lack of time synchronization among all metering
integration to power grids. Unexpected dynamic phenomena, devices. Figure 1 presents a comparison between SCADA
such as oscillatory events around 13 Hz have been recorded in
the US by Oklahoma Gas & Electric (OG&E). Such interactions and PMU measurements for this phenomenon. The utilization
differ from traditional and well studied inter-area oscillations, of PMUs enables observability for phenomena occurring at
and the ability to detect them is beyond the measurement frequencies up to 15-25 Hz across wide geographical regions,
capabilities of most of the existing measurement equipment and opening new wide perspectives for monitoring and control
monitoring tools in Energy Management Systems (EMS) systems. applications based on synchrophasors. These new tools will
This paper presents the development and implementation
of algorithms for PMU-based real-time fast sub-synchronous help in acquiring a better knowledge on the challenges brought
oscillation detection, focusing on the aforementioned case. by the increase of intermittent energy sources.
The paper focuses on the tool itself and its algorithms, briefly
presents an approach carried out for testing and validating it.
Experience from the use of the tool at OG&E is also described.
Index Terms—Power system oscillations; monitoring applica-
tion; PMU; wind farm; sub-synchronous oscillations.

I. I NTRODUCTION
Recent concerns about the environmental impact of tra-
ditional electricity generation in the western world has led
to a strong increase in renewable sources of energy. Since
the 1990s, wind power has been the fastest growing power
generation technology [1], a trend expected to continue as
several countries have politically engaged themselves into
large investments in wind power [2], [3].
Wind power is one kind among different intermittent gener-
ation sources, that brings several challenges to power system
Figure 1. Comparison example between PMU and monitoring through
dynamics. For example, its integration in existing power SCADA during a fast dynamic event
systems can involve transient stability issues [4]. Unexpected
dynamic behavior is now appearing in the form of sub- This paper presents a PMU based monitoring tool developed
synchronous oscillations. Oklahoma Gas & Electric (OG&E) for detecting the phenomenon depicted on Fig. 1, as well as a
has recently measured with PMUs [5] sub-synchronous oscil- testing method used during the development. The algorithms
latory events resulting from interactions between wind farms used in the tool are presented in Section II, the developed ap-
at frequencies around 3-15 Hz. These oscillations were also plication is introduced in Section III. Testing of the application
observed at the consumer level in the form of flicker [6]. via replay of archived data is described in Section IV, where
Traditional monitoring tools build on Supervisory Control comparisons are made with another tool developed at OG&E.
And Data Acquisition (SCADA) systems cannot detect these
The research was supported partly by EIT KIC InnoEnergy “‘Smart Power II. A LGORITHMS FOR FAST O SCILLATION D ETECTION
Project” under Action 2.6 “PMU-based power system tools”, Nordic Energy
Research through the STRONg2 rid project, Statnett SF, the STandUP for The effects of high frequency oscillations presented in the
Energy Collaboration Initiative. Introduction are undesirable [5], monitoring algorithms should
L. Vanfretti, M. Baudette, and M.S. Almas are with the SmarTS Lab, therefore enable fast oscillation detection from PMU measure-
Electric Power Systems Department, KTH Royal Institute of Technology,
Teknikringen 33, SE-1004, Stockholm, Sweden. e-mail: [email protected] ments. Traditional monitoring tools for inter-area oscillations
J.L. Domínguez-García, is with the Catalonia Institute for Energy Research, estimate the frequency and damping for each oscillatory mode
Sant Adrià del Besòs, 08930, Spain. with two separate algorithms. A similar strategy is adopted in
A. White is with Oklahoma Gas & Electric, TN, USA.
L. Vanfretti and J.O. Gjerde are with the Research and Development this case, with one algorithm dedicated to the estimation of the
Division of Statnett SF, The Norwegian Power System Operator, Nydalen amount of energy in the oscillations and the other dedicated
Allé 33, 0484 Oslo, Norway. to frequency estimation. The three algorithms used by the
978-1-4799-2442-4/13/$31.00 c 2015 IEEE proposed PMU application are described next.
A. Fast Oscillation Detection leaving to the user the choice to activate one of them or
The proposed oscillation detection algorithm in this paper both simultaneously. A non-parametric method can be used
builds from work in [7]. As highlighted in [8], it is desirable when the frequency of the oscillations within a certain range
for a fast oscillation detection tool to provide information is unknown, whereas a parametric method is appropriate when
about oscillatory behavior at different bands of the spectrum. the knowledge about the number of possible oscillations within
The spectrum of frequency of interest with potential oscillatory a range is known.
activity starts from 0.1 Hz and up to the maximum frequency
according to the Nyquist-Shannon criterion considering the PARAMETRIC
sampling frequency of the PMUs (up to 50 or 60 Hz reporting INPUT
AR SPECTRAL
ESTIMATION (YW)

rate). To cover such a broad span of frequencies, multiple PHASOR

SIGNAL
PRE
PROCESS
BAND-PASS
FILTER
PSD

instances of the algorithm can be executed in parallel. Each NON-PARAMETRIC

WELCH’S METHOD

instance can be configured independently and thus monitor

different frequency ranges. The following ranges can be used
for configuration: Figure 3. Diagram of the algorithm for Spectral Estimation
• 0.10 Hz - 1 Hz: Inter-area modes, e.g. system-wide
electromechanical swings. 1) Non-Parametric Welch’s Method: Welch’s power spec-
• 1 Hz - 3 Hz: Local-area modes.
trum estimation is a method based on the standard Peri-
• 3 Hz - 15 Hz: High frequency oscillations, e.g. wind
odogram. The method attempts to increase the readability of
farms controller interactions. the Power Spectrum Density (PSD) by reducing the noise,
• 15 Hz - 25 Hz: Other sub-synchronous oscillations, e.g.
however it decreases the PSD’s frequency resolution.
sub-synchronous resonance. The method has the following procedure: (1) The input
signal is split into overlapping segments of length M (the
As shown in Fig. 2, after pre-processing the real-time
overlapping rate is set by the user). (2) Each segment is
measurements, each frequency range previously mentioned,
windowed (the window is chosen by the user). (3) The Fast
can be separated by four different band-pass filters set to the
Fourier Transform (FFT) is computed for each windowed
boundary frequencies of each range.
segment. (4) The PSD is obtained by averaging all the resulting
spectra, thereby reducing the final variance. For a complete
BANDPASS
SUMMING
DELAY + LP FILTER
description of the method, refer to Peter Welch’s original
article [9].
FILTER MOVING
0.1 – 1 Hz
LP FILTER
- AVERAGE
LEVEL
COMPA
TRIGGER
FLAG
RISON
Because this method is based on FFT computations, it will
TRIGGER LEVEL
HP FILTER

SUMMING
BANDPASS
FILTER
DELAY + LP FILTER
MOVING LEVEL
highlight all the content of the power frequency spectrum,
1– 5 Hz
LP FILTER
- AVERAGE COMPA
TRIGGER
FLAG which is an intrinsic property of non-parametric estimation.
TRIGGER LEVEL RISON
INPUT HP FILTER
VOLTAGE
PHASOR
PRE
PROCESS SUMMING This property is especially useful for signals, which actual
DELAY +
frequency content is unknown.
BANDPASS LP FILTER
FILTER MOVING
5 – 15 Hz - AVERAGE
LEVEL
COMPA
TRIGGER

2) Parametric Auto-Regressive Methods: These methods

LP FILTER FLAG
TRIGGER LEVEL RISON
HP FILTER

BANDPASS
SUMMING
DELAY + LP FILTER
use an auto-regressive mathematical model of the input signal
FILTER
15 – 25 Hz
LP FILTER
-
MOVING
AVERAGE
LEVEL
COMPA
TRIGGER
FLAG
for estimating power spectral density. In this way, they inte-
TRIGGER LEVEL RISON
HP FILTER grate available knowledge about the input signal to improve
spectral estimation.
Figure 2. Diagram of the adaptation of the algorithm for fast real-time Auto-regressive models have been applied for estimating
oscillation detection power system frequency content, as described in [10], [11].
The output of the high-pass filter provides a measure of For this specific reason, this model is chosen in the Monitoring
the "oscillatory activity" at the selected frequency range. The Tool. The specific method utilized is, however, left to the
Root Mean Square (RMS) value of this output is used for choice of the user. All of the implemented methods work
energy computation, implying that the following computations on the principle of model fitting. They differ on the method
are performed sequentially: squaring, averaging and finally used to optimize the fitting process. For further details on
computing the square-root of the signal. A low-pass moving parametric methods refer to [12] and for further details on
average filter is used to extract the main trend of the squared their application to power systems, refer to [11]. It is worth
signal. This is necessary so that a persistent and stable signal is mentioning that during experimentations, the Burg Lattice
provided to the forthcoming trigger level comparison. Finally, method has been particularly efficient.
a trigger level comparison indicates if the computed energy The order of the model is a very important parameter.
exceeds a pre-set level. Its value cannot be fixed in advance because it is mainly
dependent on the number of oscillations to be identified, and
B. Frequency Estimation thus, the frequency content itself. In the case of the Monitoring
The frequency estimation algorithm, depicted in Fig. 3, is Tool where the frequency spectrum of interest is divided into
comprised by two different algorithms running in parallel, four frequency ranges, the order can be different in each of the
frequency bands. It is worth mentioning that too small orders average filter, where the average is computed on the 10
will lead to a smooth spectrum and some modes might be left preceding samples and 9 following samples. Compute
unidentified. On the other hand, too large orders might lead to the standard deviation of the resulting signal.
the identification of artificial modes and their appearance on 5) Create the upper (U) and lower (L) limits of the con-
the estimated spectrum. fidence interval by adding/subtracting k to (M), k the
sensitivity factor, is set by the user.
C. Data Pre-Processing:
Outlier Removal and Down Sampling Each element of (A) is then compared to (U) and (L). If
Ui > Ai > Li the element is preserved, else it is dropped and
The algorithms for estimating the energy and frequency of
will be replaced by an interpolated value. The interpolation
oscillations described previously involve filtering, averaging
is performed by taking as inputs the elements of the signal,
and spectral estimation processes. These processes require
the indexes of theses elements and all the indexes at which an
reliable data. In the case of PMUs, measurements are reported
element will be returned. Each returned element is either an
at a high rate (30, 50 or 60 samples per second) and transmitted
input element if its index is on both lists or an interpolated
over IP networks, and might contain erroneous or missing
value if its index is only on the second list. The chosen method
measurements. The pre-processing of input data thus appears
for interpolation is linear interpolation, as it gave the best
necessary to satisfy the requirements imposed by the methods
results during testing. Testing results showed that it has no
used in the detection tool.
divergence risks compared to methods involving higher order
The pre-processing implementation was inspired from the
polynomials.
prior work on PMU data pre-processing available in [13],
An example of the outlier removal algorithm applied to real-
which were executed off-line on archived data. The Monitoring
time PMU data is shown in Fig. 5. The data points removed
Tool processes the measurements in real-time, and thus, the
are single points far from the neighbouring values, likely
following algorithm was developed.
corresponding to measurement errors. The outlier removal
In general, the variation of frequency in a power system
effectively performs the tasks described above.
is the result of the interaction between varying loads and the
generation that follows the same variations with an inertial
delay. It should thus be a rather smooth process that implies
that any value is expected to be within a confidence interval, as
shown in Fig. 4, which can be determined from neighboring
values and intrinsic system properties. The performance of
the outlier removal algorithm is determined by the level of
confidence of the interval used.
50.1 (a) Outlier removal inactive (b) Outlier removal active
50.08 Figure 5. Screen shots of the frequency signal with and without outlier
50.06
removal
50.04
After removing outliers, the signal is processed further to
Frequency (Hz)

50.02 feed the frequency estimation algorithms by mean removal and

50 high-pass filtering. This aids to attenuate frequency compo-
49.98 nents related to the action of generator governors and loads,
49.96 which are close to 0 Hz and lower than electro-mechanical
49.94 modes.
Lower limit
49.92 Upper limit Furthermore, according to Shannon’s theorem, all the fre-
49.9
Frequency
quency content below frequency f1 can be restored if the
1 1.5 2 2.5 3 3.5 4 4.5
Time (s)
5 5.5 6 6.5 7 7.5 8
sampling frequency is at least 2 ⇥ f1 . The typical reporting
frequencies of PMUs are 30, 50 or 60 Hz, which correspond
Figure 4. Example of a confidence interval for the frequency to a highest observable frequencies 15, 25 or 30 Hz. The
observed frequency in the tool can be much lower, thus, a step
The confidence interval suggested here is built according to of down-sampling is added to remove redundant data1 . This
the following steps: step is preceded by a low-pass filter acting as an anti-aliasing
1) The input signal: (f ). filter.
2) Create one copy (A) of the input signal delayed by 10 The implementation was carried out with a band-pass filter
samples: Ai = fi+10 , where i is the time index. for each frequency range configured. The down-sampling
3) Filter one copy (M) of the input signal withP a moving factor is calculated from the cut-off frequency of the band-pass
fi t
average filter of order equal to 20: Mi = 0<t<20
20 . 1 Down-sampling is applied for frequency estimation and not performed in
4) Subtract (M) from (A). The combination of a delayed the oscillation detection algorithm in order to avoid lowering the reaction
copy (A) and (M) is equivalent to a having a moving times of the algorithm.
filter, considering Shannon’s criterion. An upper limit equal to
10 was added to the down-sampling factor, motivated by the
limited size of the rolling window of buffered data.

III. M ONITORING A PPLICATION

The Monitoring Tool has been implemented as a real-
time graphical analysis tool, which could be used by system
operators or wind farm owners. LabVIEW was chosen as
the software development environment due to the availability
of two different real-time mediators, namely Statnett’s Syn-
Figure 7. Screen shot of the configuration tab of one module of the
chrophasor SDK [14] and BableFish [15]. These data medi-
Monitoring Tool
ators are compliant with the IEEE C37.118.2 standard [16]
for PMU data exchange and deliver the PMU measurements
in the LabVIEW development environment. In addition, Lab- The graphical display on the left has several components,
VIEW also allows designing a graphical user interface in a the most important are the three LEDs Low Activity, High
straightforward manner. Activity and Danger!!!, corresponding to the three thresholds
The tool is provided with real-time PMU measurements, defined in the general configuration tab. The comparison is
allowing monitoring of oscillatory events at a frequency up made on the latest sample, whose value is displayed by the
to the Nyquist limit. This broad frequency spectrum contains field Energy. The graph provides a history to easily corroborate
different categories of phenomena, mentioned in Section II. the energy computed with the input signal displayed. Given
The tool has been implemented for monitoring four frequency that the input signal is filtered by three FIR filters, the output
ranges simultaneously. The elements in four instances are signal is shorter than the input signal. Moreover the order of
named Modules. The resulting Graphical User Interface (GUI) the filters have a strong influence on the length of the output
is depicted in Fig. 6 (for concision purposes, only the interface signal. This can be modified by the user to adapt to their needs,
of the Replay Tool introduced in Section IV is shown in this measurement features and particular power network.
paper). The Monitoring Tool presented in this paper was compared
The interface includes a data display that is a simple against an in-house tool developed by OG&E. OG&E’s tool
representation of the buffered input signals received. The runs on a production-grade server and is shown in Fig. 8. It
signals are displayed one at a time, letting the user choose provides detection of the oscillations and email notifications.
from the list box on top of the graph in Fig. 6 (the frequency is The application was developed in VB.net and utilizes the
displayed in this case). This graph is also useful for calibrating digital signal processing exocortex library for FFT oscillation
of the outlier removal algorithm. detection2 . It is configured via a set of parameters, which can
The top left part of the GUI is used to configure the tool be tested against archived data for assessing their relevance.
and divided in five tabs. The tab “Options” gathers the general
configuration of the tool related to the data acquisition as well
IV. R EPLAY E XPERIMENT
as state LEDs. The remaining tabs “Module 1 to 4” are four
identical tabs used for the configuration of the estimation and In the previous Section the Monitoring Tool was introduced
detection modules, see Fig. 7. The configuration should be and all the options for its configuration have been presented.
done after launching and initializing the tool. While the options related to connectivity aspects do not require
Figure 7 presents an interface with two separate blocks. knowledge about the power system analyzed, the configuration
The block on the left is used to configure the band-pass filter, of the processing algorithms, especially for frequency estima-
the parameters are common for both the Oscillation Detection tion, have to be calibrated according to the properties of the
and the Frequency Estimation algorithms. The block on the power system. A good solution is to use archived data to run
right is dedicated to the Frequency Estimation algorithm and off-line experiments.
contains the parameters for each of the methods mentioned The original idea was to replay archived data from a PDC,
in Section II. It also includes the parameters for spectral broadcast it as an output stream and use the Monitoring Tool.
averaging. Finally the user is able to choose in the list box The diversity of archive techniques and file structures raised
if both methods are to be used simultaneously or just one of the issue that a PDC, such as OpenPDC [17], was not flexible
them. enough and not suitable to replay the archived data format
available. This led to the development of the Replay Tool.
Once the configuration is set, the outputs of the algorithms
are displayed in a Module, highlighted on Fig. 6. The graph A. Replay Tool
on the right presents the power spectrum density, which The Replay Tool has been developed using the same code
is the output of the Frequency Estimation algorithm. It is as the real-time tool. However, the software related to input
scaled according to the parameters of the band-pass filter,
highlighting the content of interest. 2 http://www.exocortex.org/dsp
 Figure 6. Screenshot of the tool reading the file of the oscillation case at OG&E, shown in Fig. 8

identical to the Monitoring Tool. The processing algorithms

are identical. This tool has specific additional features. For
example it allows to scroll along the replayed data, which
can be useful to get a quick overview of the content of the
selected file. The Replay Tool works, otherwise, exactly like
the Monitoring Tool.

B. Results
The experiments carried out with the Replay Tool used
archived data obtained from OG&E, containing measurements
from different locations during oscillatory events at a fre-
quency around 13 Hz, as mentioned in the Introduction.
These experiments served, during the development and im-
plementation of the algorithms, to verify that the algorithms
worked properly and calibrate their configurations for other
experiments.
Figure 6 presents the Replay Tool during the replay of such
oscillatory event. The beginning of the oscillatory phenomenon
Figure 8. Screenshot of OG&E’s FFT tool
can be clearly identified in the frequency (display on the top
left), as well as in both active and reactive power (displays
on the top right). It can also be noticed that the Module on
measurements has been developed to use archived PMU the bottom right, corresponding to the frequency range [12 -
data instead of real-time PMU data. The interface is almost 15 Hz], that the Danger!!! indicator is activated only a few
seconds after the beginning of the phenomenon. Additionally [6] A. White (OGE), S. Chisholm (OGE), H. Khalilinia (WSU), Z. Tash-
the oscillatory frequency is identified to 13.4 Hz on the PSD. man (WSU), and M. Venkatasubramanian (WSU), “Analysis of Subsyn-
chronous Oscillations at OG&E,” NASPI-NREL Synchrophasor Tech-
Figure 8 presents OG&E’s tool analysing the same archived nology and Renewables Integration Workshop - Denver, CO., June 7
data. As it can be seen on the screenshot a strong oscillatory 2012.
activity, above the defined threshold, is occurring at 12 Hz. [7] J. Hauer and F. Vakili, “An oscillation detector used in the BPA power
system disturbance monitor,” IEEE Transactions on Power Systems,
Additional oscillatory components can be noticed around vol. 5, no. 1, pp. 74 –79, Feb. 1990.
1.4 Hz, 5.5 Hz, 8 Hz, 9.4 Hz and 10.7 Hz. The main 12 Hz [8] D. Trudnowski, “Fast real-time oscillation detection,” Proc. North Amer.
component is also detected with the Monitoring Tool and it SynchroPhasor Initiative (NASPI) Work Group Meeting, Orlando, FL.,
February 29 - March 1 2012.
activates the Danger!!! threshold as shown on Fig. 6. All the [9] P. Welch, “The use of fast fourier transform for the estimation of power
other components can also be identified on Fig. 6, validating spectra: A method based on time averaging over short, modified peri-
the Monitoring Tool’s algorithms. odograms,” IEEE Transactions on Audio and Electroacoustics, vol. 15,
no. 2, pp. 70 – 73, jun 1967.
Additional testing and validation experiments have been [10] L. Vanfretti, S. Bengtsson, V. Aarstrand, and J. O. Gjerde, “Applications
carried out by using hardware-in-the-loop setups with real of spectral analysis techniques for estimating the Nordic grid’s low fre-
quency electromechanical oscillations,” in 16th IFAC Symposium on Sys-
PMU devices in two different laboratory environments. One tem Identification, ser. IFAC Proceedings Volumes (IFAC-PapersOnline),
setup involved the use of a real-time digital simulator with no. 16. IFAC, 2012, pp. 1001–1006, invited Paper, Special Session:
a model reproducing the sub-synchronous oscillatory phe- “Development of System ID Methods for Power System Dynamics”.
[11] J. Sanchez-Gasca, “Identification of electromechanical modes in power
nomenon [18], [19]. In the second setup, the simulator was systems,” Special Publication, TP462, IEEE Power & Energy Society,
replaced by power emulators, producing real power oscilla- 2012.
tions [20]. [12] M. H. Hayes, Satistical digital signal processing and modeling. John
Wiley & Son, 1996.
Finally, paper [21] evaluates the impact of a laboratory set- [13] L. Vanfretti, S. Bengtsson, and J. O. Gjerde, “Preprocessing Synchro-
up when performing validation experiments on this tool. nized Phasor Measurement Data for Spectral Analysis of Electrome-
chanical Oscillations in the Nordic Grid,” European Transactions on
Electrical Power, in press.
V. C ONCLUSION [14] L. Vanfretti, V. H. Aarstrand, M. Shoaib Almas, V. Peric, and J. O.
Gjerde, “A software development toolkit for real-time synchrophasor
This paper gives an example how synchonized phasor applications,” to appear in Proceedings of the IEEE PowerTech 2013.
data applications can be developed to help grid operators in [15] L. Vanfretti, I. Al Khatib, and M. Shoaib Almas, “Real-Time Data
Mediation for Synchrophasor Application Development Compliant with
monitoring and control of renewable energy sources when IEEE C37.118.2.” Washington DC: IEEE Innovative Smart Grid
unpredictable dynamic interactions arise. Conference (ISGT) North America, Feb 2015.
In addition, the development of this PMU application [16] “IEEE Standard for Synchrophasor Data Transfer for Power Systems,”
IEEE Std C37.118.2-2011 (Revision of IEEE Std C37.118-2005), pp. 1
showed that new applications can be conceived and imple- –53, 2011.
mented without relying on a monolithic software environment [17] “OpenPDC: The Open Source Phasor Data Concentrator,” online:
and within a relatively short time. http://openpdc.codeplex.com/.
[18] L. Vanfretti, M. Baudette, I. Al Khatib, M. Shoaib Almas, and J. Gjerde,
The paper did not cover the control actions that can be “Testing and validation of a fast Real-Time oscillation detection PMU-
taken to mitigate the unwanted dynamics discussed, as the Based application for Wind-Farm monitoring,” in First International
Black Sea Conference on Communications and Networking, Jul. 2013.
only available action at this time is to curtail the wind farm [19] M. Baudette, “Fast real-time detection of sub-synchronous oscillations
power output. One important aspect to further investigate is in power systems using synchrophasors,” Master’s thesis, KTH, Electric
how the use of synchrophasor data can help in providing Power Systems, 2013.
[20] M. Baudette, L. Vanfretti, G. Del-Rosario, A. Ruíz-Alvarez, J.-L.
rich measurements containing information on sub-synchronous Domíngez-García, I. Al-Khatib, M. S. Almas, I. Cairo, and J. O. Gjerde,
oscillations into the controls of Static Var Compensators, and “Validating a Real-Time PMU-Based Application for Monitoring of Sub-
other controllable devices. Recent applications to damping Synchronous Wind Farm Oscillations,” accepted, IEEE ISGT 2014.
[21] J. L. Domínguez-García, M. Baudette, G. Del-Rosario, A. Ruíz-Alvárez,
of low-frequency oscillations suggest that PMU-based sub- M. S. Almas, I. Cairó, and L. Vanfretti, “Validation Experiment Design
synchronous oscillation damping could be effectively ap- of a PMU–Based Application for Detection of Sub-Synchronous Oscil-
plied [22]. lations,” in 15th IEEE International Conference on Environment and
Electrical Engineering (EEEIC).
[22] K. Uhlen, L. Vanfretti, M. M. De Oliveira, A. Leirbukt, V. H. Aarstrand,
R EFERENCES and J. O. Gjerde, “Wide-area power oscillation damper implementation
and testing in the norwegian transmission network,” in IEEE Power and
[1] J. F. Manwell, J. G. McGowan, and A. L. Rogers, Wind Energy Energy Society General Meeting, 2012, 2012, pp. 1–7.
Explained: Theory, Design and Application. John Wiley & Sons, 2010.
[2] D. Milborrow, “Europe 2020 wind energy targets, eu wind power now
and the 2020 vision,” Windpower Monthly Special Report, March 2011.
[3] U.S. Energy Information Administration (EIA). (2012, Feb) Most
states have Renewable Portofolio Standards. [Online]. Available:
http://www.eia.gov/todayinenergy/detail.cfm?id=4850
[4] J. Wiik, J. O. Gjerde, and T. Gjengedal, “Impacts from large scale inte-
gration of wind energy farms into weak power systems,” in Proceedings
of the International Conference on Power System Technology, 2000, pp.
49 –54.
[5] A. White and S. Chisholm, “Relays become problem solvers,” Trans-
mission & Distribution world, November 2011.

View publication stats