Professional Documents
Culture Documents
AbstractData packets based high-speed digital communications have opened the door for new types of applications in power
system protection and control. The main obstacle in adopting
Ethernet-type networks for time-sensitive communications was
the data transfer characteristic of these networks. With baud
rates in the gigabit-per-second range, the industry has started
adopting the Ethernet networks as a viable alternative, not only
because of their lower cost. This paper presents an open-system
approach in which protective hardware equipment is designed
to work in a clustered environment, sharing resources, data, and
diagnostic functions with similar units. The Ethernet network,
being capable to accommodate transparently a wide range of
protocols, is used to connect the protective hardware. A model
was developed for the Ethernet network to be incorporated into
the power system simulation. The characteristics required from a
digital relaying algorithm to be able to work with signal samples
sent across data packets based networks are described in this
paper. The proposed solution is implemented using a modified
adaptive least-square-error algorithm and tested on a protection
system that was integrated into the power system model.
Index TermsData communication, open systems, power
system protection, power system relaying, protective relaying,
sampled data systems.
I. INTRODUCTION
Manuscript received January 24, 2005; revised March 16, 2005. Paper no.
TPWRD-00049-2005.
E. Demeter is with SaskPower, Regina, SK, S4P 0S1, Canada (e-mail:
edemeter@ieee.org).
T. S. Sidhu is with the University of Western Ontario, London, ON, N6A 5B9,
Canada.
S. O. Faried is with the University of Saskatchewan, Saskatoon, SK, S7N
5A9, Canada.
Digital Object Identifier 10.1109/TPWRD.2005.855628
to have a reliable protection scheme. Various types of synchronized phasor measurements have been proposed to solve the
communication delay problem. The requirements and performance of synchronized phasor measurement schemes have been
outlined in [2].
Microprocessor-based digital protection and control relays
offer a multitude of advantages over their analog counterparts.
Besides being able to perform a full range of protective and control functions, they offer numerous monitoring, analyzing, and
recording capabilities. The fact that the information is processed
and stored digitally makes their interconnection a natural next
step. Market forces and the lack of coordination between manufacturers in the past have resulted in numerous proprietary applications aimed at interconnecting digital protection, control,
and monitoring apparatus.
Until recently, communication with and between digital apparatus in substations was limited mainly to remote configuration and supervisory control and data acquisition (SCADA)
applications. The solutions that have emerged required dedicated lines of communication. For limited functionality, slow
RS-232 and RS-485 are used to exchange SCADA data between substation monitoring and control equipment. Intelligent
electronic devices (IED) have started adopting peer-to-peer and
process-bus communications over Ethernet networks with limited real-time data-sharing capabilities [3], [4].
Small changes to the substation configuration usually result
in extensive changes to the communication infrastructure and
in configuration software. Present practices have resulted in a
pyramidal structure with a multitude of data concentrators, remote terminal units (RTUs), protocol converters, data archiving
computers controlling, and monitoring field devices connected
through a web of cables. Such a structure makes high-speed
monitoring and control practically impossible.
The reasons why Ethernet applications in power system
protection and control have experienced a slower acceptance
curve include the data delivery characteristics, lack of industrial
components, physical media, security concerns, and lack of an
application layer standard. Recent developments have started
addressing these issues, making Ethernet the solution of choice
in many monitoring and control applications [5].
The networks best-effort data transmission was a significant drawback to adopting the Ethernet for real-time applications. A solution for obtaining synchronized phasors when signal
samples arrive across these networks is proposed. The solution
can be applied for remote data acquisition applications in the
protection, monitoring, and control of power systems. The proposed digital relaying algorithm takes into account the delays and
quality of service issues experienced during the transmission.
DEMETER et al.: OPEN-SYSTEM APPROACH TO POWER SYSTEM PROTECTION AND CONTROL INTEGRATION
31
32
the same SONET optical backbone, but utilizing separate, dedicated cards [7].
The performance of Ethernet-based and TCP/IP-based networks under typical power systems applications traffic have
been investigated in the industry. The performance of ten IED
devices that were connected both in single shared collision domain and in switched full-duplex collision-free domains show
that in both 10 Mbps and 100 Mbps operations latency can vary
depending on numerous factors [8]. The results showed that
with a properly designed transmission system, on average, data
sampled at 1 kHz can be transmitted with minimum delay using
IP over Ethernet networks. Other studies and projects proved
that values sampled at a frequency of 4 kHz can be transmitted
over an Ethernet-based process-control bus [9].
III. COMMUNICATION-BASED PROTECTION AND CONTROL
Depending on their performance, the protection and control
applications response time can range from very fast, for local
protective relays, to relatively slow, for SCADA applications,
as depicted in Fig. 3.
A widely utilized solution to transmission delay problems is
to desensitize the protective device to a level that can accommodate the transmission delay without causing maloperations.
Such an approach can be tolerated in systems such as the differential protection of transmission lines, but is not applicable
to solutions where precise phase angle differences are to be accounted for between remote sites. A complete standard is presented in [10] that deals with the synchronized measurement and
transmission of phasor information.
A. Implied Sequential Sampling
When using sequential communications, the data samples
that arrive from the remote end are presumed to be in proper
sequence. The protection and control applications that are
using such communication channels do not time-stamp the
samples, but imply that the samples are in sequential order:
every sample is considered to have been obtained after the
previously received value and before the sample that will arrive
next. Any communication breakdown, even for the duration of
one sample, results in losing the functions of the relays.
B. Time-Stamped Sampling
A different set of applications are the ones that attach a time
stamp to every sample taken. These solutions are extremely sensitive to deviations of their internal clocks. Keeping the internal
clocks of the devices utilized in such applications synchronized
DEMETER et al.: OPEN-SYSTEM APPROACH TO POWER SYSTEM PROTECTION AND CONTROL INTEGRATION
33
Fig. 4. Open-system protection and control integration into the power system.
Fig. 6.
A. Data Packet
The data samples taken from voltages and currents are timestamped and tagged. The attached time-stamp represents the
moment at which the sample was taken, as obtained through
GPS. The GPS is used in the proposed application as the source
of a synchronized time reference.
The data packet that is transmitted on the high-speed network
can include information relating to the quality of the sample. In
fact, a wide range of information such as tags, limit violations,
and quality of data can and are included in the packet.
B. Adaptive Filtering
Fig. 5. Measured and received signals.
has been investigated for transmitting protection control functions [13]. Ethernet-based process-bus applications based on the
IEC standard 61850 have been proposed and tested.
Integrating several protective functions into a single relay is
becoming common practice. The proposed digital relaying algorithm allows a new level of integration of protective equipment
and functions into a comprehensive open-system application, as
represented in Fig. 4, which is an example of station-level integration. By making use of existing information technology solutions, such as clustered processing hardware, redundant networking, etc., the system components can share operational and
diagnostic functions, providing a new level of redundancy and
security.
for
Therefore
also be described by its system function
(4)
, with the coefficients
which represents a polynomial in
.
In case that a sample does not arrive when it is expected, the
first element of the filter (as seen in Fig. 7) will be undefined;
therefore the output will yield incorrect results. On the other
34
Fig. 7.
VI. IMPLEMENTATION
In order to test the procedure described in the previous section, the least-square-errors algorithm was used to develop the
required set of filters [14].
Decaying dc components and a reasonable number of harmonics can be calculated using this algorithm. The general form
of a measured signal is
(7)
Fig. 8.
where
is the magnitude of the harmonic .
The Taylor series expansion of a complex function around a
number is defined by
(8)
Starting from (8) and using as center
series expansion of the complex function
, the Maclaurin
can be written as
(9)
(5)
When the filter window shifts to get the new sample, during the
.
convolution, the missing value would be multiplied with
In case that the sample has not arrived (as seen in Fig. 8), a
is employed to
new set of filter coefficients for which
obtain the correct result. The system function of the new filter is
(6)
By applying the above steps, a new set of coefficients can
be obtained for each position of the missing sample inside the
filter window. Once the sample arrives and is recognized by the
system as being covered by the filter window, the sample is inserted into its proper position in the convolution sum, and the
original coefficients are reloaded.
For power system protection applications the two-term approximation is considered sufficient; therefore (7) can be written
as
(10)
When the data stream delay is within the acceptable limits
and no samples are missing from the filter window, each of the
samples will satisfy (10).
for
. If the sampling rate
It can be considered that
is set, then the previous sample before the moment
was
, where
taken at
(11)
with
DEMETER et al.: OPEN-SYSTEM APPROACH TO POWER SYSTEM PROTECTION AND CONTROL INTEGRATION
35
..
.
(12)
The set of equations represented in (12) can be written in
matrix form as
(13)
where
..
.
The first row of matrix
..
.
..
.
Fig. 9. Received signal together with the phase angle and magnitude of the
sent signal obtained at the receiving end.
is
..
.
(14)
The matrix
can be calculated offline. The
values can also be obtained by using an online processing procedure. From numerical stability point of view, it is safer to use
the singular value decomposition of matrix , together with
the GramSchmidt orthogonalization, to obtain the optimum
solution of (13).
are obtained, the elements describing
Once the values of
(10) can be calculated. For example, the two elements of matrix
that will give information on the real and imaginary parts of
and . Therefore, it
the fundamental frequency signal are
can be written that
(15)
(16)
(17)
36
Fig. 13.
control.
Fig. 10.
Fig. 11.
The system performed well under various fault and communication media conditions, confirming the viability of the proposed solution. The first two plots of Fig. 12 represent the current signal magnitude and phase angle measured at Station A
during a line-to-ground fault at location F. The next two plots
are the magnitude and phase angle of the current signal measured at Station B, and calculated at Station A, using samples
received across the Ethernet network.
The output of two differential protective functions are shown
in the last two plots of Fig. 12. Both the alpha-plane differential(87-A) and the percentage differential (87-D) protective functions performed well with the proposed relaying algorithm [15].
The length of the designed filters can vary depending on the
performance of the transmission network. The test results shown
in Fig. 12 were obtained with a digital relaying algorithm that
was designed for 16 samples and less, over one half period of
the fundamental frequency.
With proper analog filtering and antialiasing in place at the
location where the samples are taken, the target length of the
filter can be reduced for faster response, or the number of filter
sets increased, to accommodate less performant communication
channels.
VIII. CONCLUSION
Fig. 12.
The digital relaying algorithm, together with the data transmission subsystem, was integrated into the test power system shown
in Fig. 10. The overall integrated model used to test the performance of the algorithm is presented in Fig. 11.
DEMETER et al.: OPEN-SYSTEM APPROACH TO POWER SYSTEM PROTECTION AND CONTROL INTEGRATION
37
Tarlochan S. Sidhu (SM94F04) received the B.E. (Hons.) degree from the
Punjabi University, Patiala, India, and the M.Sc. and Ph.D. degrees from the
University of Saskatchewan, Saskatoon, SK, Canada, in 1979, 1985, and 1989,
respectively.
He is currently Chair of the Department of Electrical and Computer Engineering and Professor and Hydro One Chair in Power Systems Engineering,
University of Western Ontario, London, ON, Canada. From July 1990 to June
2002, he was a Professor and Graduate Chair of the Department of Electrical
Engineering, University of Saskatchewan. He worked also with the Regional
Computer Center, Chandigarh, India; Punjab State Electricity Board, India; and
Bell-Northern Research Ltd., Ottawa, ON. His areas of research interest include
power system protection, monitoring, and control.
Dr. Sidhu is a Fellow of the Institution of Electrical Engineers of the U.K.,
Fellow of the Institution of Engineers, India, a Professional Engineer registered
in the Province of Ontario, and a Chartered Engineer in the U.K.
Sherif O. Faried (SM00) received the B.Sc. and M.Sc. degrees in electrical
engineering from Ain Shams University, Cairo, Egypt, and the M.Sc. and Ph.D.
degrees in electrical engineering from the University of Saskatchewan, Saskatoon, SK, Canada.
He is currently a Professor of Electrical Engineering, Graduate Chair, and
the Coordinator of the Power Systems Research Group at the University of
Saskatchewan. His research interest includes power system dynamics, reliability, and electrical machines.