Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
I. INTRODUCTION
Manuscript received June 12, 2014; revised September 29, 2014 and
November 27, 2014; accepted January 05, 2015. Date of publication March
02, 2015; date of current version May 20, 2015. This work was prepared by
working group H-11 of the Relay Communications Subcommittee of the IEEE
Power System Relaying Committee of the Power Engineering Society. Paper
no. TPWRD-00690-2014.
K. E. Martin, Chair, is with the Phasor Measurement Systems, KenM Consulting, Portland, OR 97213 USA and also with the Electric Power Group,
Pasadena, CA 91101 USA.
Color versions of one or more of the gures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identier 10.1109/TPWRD.2015.2403591
0885-8977 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
TABLE I
REQUIRED PMU REPORTING RATES (FPS)
1515
The concept of different classes of performance was introduced with IEEE Standard C37.118-2005. Classes were differentiated as performance levels 0 and 1. Compared to the requirements for level 1, level 0 had relaxed requirements for
harmonics and out-of-band signal rejection, and narrowed frequency and magnitude ranges for measurement performance.
Level 1 was intended for use by controls and applications that
are sensitive to harmonics and to small signals that could be
aliased. Level 0 was for applications insensitive to small signals but needed minimum latency.
Since these requirements were described as levels of performance and level 1 had tighter requirements than level 0, many
users interpreted the higher the level, the better the performance
of their applications. Some users were specifying level 1 when
level 0 was better suited to their requirements.
In the 2011 standard, the label levels was changed to
classes. Rather than using a number (that seemed to imply
quality) for the class, a letter was chosen to indicate the type
of targeted application. P was chosen for the class of measurement that would be made with minimal delay but less
immunity to out-of-band interference. P class measurement
is intended to support applications (e.g., high-speed controls)
that generally require minimal delay in responding to dynamic
changes. M was chosen for the class of measurements that
requires out-of-band ltering to avoid signal aliasing where
longer latency is acceptable. M class measurement is intended
to support applications (e.g., certain control functions) that are
sensitive to signal aliasing but can tolerate longer delays.
It should be noted that the letters chosen for classes are only
an indicator of the differences in performance requirements.
They might be used as a general guideline in choosing a performance class, but certainly should not be used without full
consideration of the specic application requirements. A demanding application may have performance requirements beyond what is specied for either class. The system designer
should always choose the class with the specic system and application in mind.
IV. STEADY-STATE REQUIREMENTS
The 2005 standard dened performance requirements under
steady-state signal conditions. It introduced the total vector error
(TVE) method of evaluating the measured synchrophasor value.
Total vector error (TVE) combines the evaluation of the angle
and magnitude errors of the synchrophasor estimate as a single
error value. This method of evaluation was retained to maintain compatibility with the previous standard and its simplicity
in specifying error limits. Separate phase angle and magnitude
limits could be added to future revisions.
Phase angle is determined by the reference time in conjunction with the estimation algorithm. Its accuracy depends
on time synchronization, analog input components, digital
conversion, and signal processing. To avoid problems with
specifying internal operations or requirements that may not
be testable (for example, some PMUs use an internal global
positioning system (GPS) receiver and do not provide access to
the timing signal), the Working Group H11 (WGH11) decided
to only specify the result. This leaves it to the manufacturer to
optimize their design and allocation of the error budget to meet
1516
1517
1518
1519
Frequency is dened
(1)
And ROCOF is dened
(2)
These derivative calculations can be done using phasor
values computed at the point-on-wave rate data to improve
performance. This reduces the delay time between the phasor
values and frequency and ROCOF as well as reducing measurement noise.
Estimation of derivatives by differencing small sample steps
has been shown to create high noise due to quantized steps.
However in this case, the data used to compute the steps is
1520
highly correlated, so the close spacing of the measurement samples actually reduces the noise. Suppose the waveform sampling is every
s and the three phase-angle estimates at times
,
, and
are used to estimate frequency and
ROCOF. Due to the window of calculation they are based on almost the same data. For instance, in the 50/s estimator described
in the reference model, each PMU estimate is based on 143
samples. Assume that the three phase-angle estimates are determined using three sets of 143 samples, offset by
1 sample
(i.e.,
one sample interval, which is 1/800 s here). This requires a total of 145 samples. 141 of the samples will be used
by all three of the estimates, with slightly different weighting;
and two samples at the ends of this group will be used by only
one or two of the estimates, with very small weighting. Noise
present in any of the samples as well as the desired signal will
propagate to the phase-angle estimates to almost the same degree for all three estimates. The smaller
, the closer to identical the three estimates become, and the more the signal (including noise embodied in it) will cancel in the frequency and
ROCOF calculations. In the limiting case, if
is set to zero,
all three estimates will be identical (even with truncation in the
math, which otherwise does not cancel).
The sample data in this analysis includes errors in the signal,
analog front end, timing, and analog-to-digital sampling. These
are correlated in the phasor estimation and minimized. However
errors caused by truncation or round off in the phasor estimation algorithm itself (dependent on the CPU/DSP word length
and other implementation considerations) do not cancel since
they are created independently for each estimate after the waveform samples are taken and are not correlated. Also noise that is
inband or leaks through the phasor lters, such as those introduced by the OOB test, can cause frequency and ROCOF errors.
C. Assessment of Frequency Performance
The tests described in the standard use input signals are lownoise sinusoids with limited, well-dened changes in magnitude
and phase. The standard notes that the real power system signals consist of noisy, rapidly changing values. Artifacts are introduced by nonlinear loads, line switching, reactor switching,
and others. This situation is further exacerbated by the proliferation of nonsynchronous power sources, such as wind, solar,
and HVdc.
As rst- and second-order derivatives of phase angle,
frequency and ROCOF may be adversely affected by these
real-world conditions. Though they are theoretically invaluable
quantities for power system protection and control, the standard
notes that they should be used with caution.
PMU frequency and ROCOF estimates are not intended to replace IEEE Standard C37.2 device 81 (under or over frequency)
or 81R (rate-of-change of frequency) elements as found in conventional protection and control schemes for power networks.
In order to comply with the requirement of having a single
time-tag for a set of measurements from a PMU (a PMU reporting message), their performance and accuracy are dictated
by the phasor ltering system, the primary purpose of which has
been to achieve adequate synchrophasor measurements. For use
in any application, the PMU frequency and ROCOF measurements performance and accuracy should be checked against
This limit is required for both classes of PMU for all reporting
rates. It is the time difference between the actual time of a step in
phase or magnitude of the input signal and the time that the PMU
output reaches 50% of the step size. This requirement limits the
maximum inaccuracy of the time stamp relative to a change in
the input being reported in the synchrophasor output.
The second limit on latency appeared in Table 11 of Clause
5.5.9, PMU reporting latency compliance. The limit is Maximum PMU Reporting Latency (in seconds) and is limited to
no more than 2/(Reporting Rate) for P class and 7/(Reporting
Rate) for M Class. PMU reporting latency is the interval
from the time reported by the time stamp to the time that the
synchrophasor information becomes available at the output of
the PMU. This requirement assures the data transmission from
the PMU has bounded delay, as is required for real-time applications. IEEE Std. C37.118.1-2011 described Measurement
reporting latency in ambiguous terms. Amendment IEEE Std.
C37.118.1a-2014 replaced Measurement reporting latency
1521
1522
REFERENCES
[1] IEEE Standard for Synchrophasors for Power Systems, IEEE Standard
C37.118-2005 (Re. IEEE Standard 1344-1995).
[2] IEEE Standard for Synchrophasor Measurements for Power Systems,
IEEE Standard C37.118.1-2011.
[3] IEEE Standard for Synchrophasor Data Transfer for Power Systems,
IEEE Standard C37.118.2-2011.
[4] IEEE Standard for Synchrophasor Measurements for Power SystemsAmendment 1: Modification of Selected Performance Requirements, IEEE Standard C37.118.1a-2014.
[5] A. G. Phadke, J. S. Thorp, and M. G. Adamiak, A new measurement
technique for tracking voltage phasors, local system frequency, and rate
of change of frequency, IEEE Trans. Power App. Syst., vol. PAS-102,
no. 5, pp. 10251038, May 1983.
[6] IEEE Standard for Synchrophasors for Power Systems, IEEE Standard
1344-1995 (R2001).
[7] K. Martin, J. F. Hauer, and T. Faris, PMU testing and installation
considerations at the Bonneville Power Administration, presented at
the Power Eng. Soc. Gen. Meeting, Tampa, FL, USA, 2007.
[8] R. M. Moraes, Y. Hu, G. N. Stenbakken, K. E. Martin, J. E. Alves, A. G.
Phadke, H. A. R. Volskis, and V. A. Centeno, PMU interoperability,
steady-state and dynamic performance tests, IEEE Trans. Smart Grid,
vol. 3, no. 4, pp. 16601669, Dec. 2012.