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

net/publication/3275306

A CT Saturation Detection Algorithm Using Symmetrical Components for

Current Differential Protection

Article  in  IEEE Transactions on Power Delivery · February 2006

DOI: 10.1109/TPWRD.2005.848654 · Source: IEEE Xplore

CITATIONS READS
61 934

2 authors:

Nicholas Villamagna P.A. Crossley

4 PUBLICATIONS   90 CITATIONS   
The University of Manchester
268 PUBLICATIONS   4,423 CITATIONS   
SEE PROFILE
SEE PROFILE

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

Power Quality issue View project

Multi-zone Current Differential Protection for Transmission Networks View project

All content following this page was uploaded by P.A. Crossley on 24 November 2014.

The user has requested enhancement of the downloaded file.

38 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

A CT Saturation Detection Algorithm

Using Symmetrical Components for
Current Differential Protection
Nicholas Villamagna and Peter A. Crossley, Member, IEEE

Abstract—A method of symmetrical component analysis for the the expense of protection sensitivity. Raising the bias threshold
detection of current-transformer (CT) saturation in a numerical has a detrimental effect on the relay sensitivity as it prevents the
current differential feeder protection relay is presented in this detection of in-zone resistive faults.
paper. The performance of the differential relay is investigated
for various faults on a typical Electro-Magnetic Transients Pro- When considering a protection CT, the flux density required
gram/Alternative Transients Program (EMTP/ATP) simulated to drive the normal load current and low values of fault current
transmission feeder. The simulator includes the effects of CT through the relays connected to the CT are normally well within
saturation. A comparison between simulation and tests conducted the linear region of the B-H curve [1]. Consequently, the excita-
on an analog model testbench are also evaluated. The results tion current is small and the secondary current is the ratio replica
show a high degree of similarity and illustrate the effect that CT
saturation imposes on the sensitivity and stability of the protection of the primary current. However, if the fault current is high and
scheme. An algorithm is presented that shows significant improve- contains a dc offset, then unless a large over-dimensional CT
ment in sensitivity on internal faults while still maintaining a high has been selected, the flux in the core will enter the saturated re-
level of stability on external faults and nonfault events. gion of the B-H curve. During the period when the flux is within
Index Terms—Current-transformer (CT) saturation, gradient, this region, the excitation current will be high and the secondary
hysteresis data (HYSDAT), rate of change, sliding data window, current will be severely distorted.
symmetrical component, testbench. Saturation of a CT resulting from a high current in-zone fault
is unlikely to affect the operation of the differential protection.
I. INTRODUCTION The magnitude of the differential current will be reduced but it
will still exceed the bias threshold by a considerable margin. The

T HE BASIC operating principle of current differential

protection is to calculate the difference between the
current entering and leaving the protected zone. The simplicity
concerning problem caused by an in-zone fault is the effect on
the future stability of the protection, due to the remanent flux left
in the CT after the in-zone fault has been cleared. Since the load
of comparing the current flowing into a feeder with the current current is unlikely to reduce the remanent flux to a low value,
flowing out is very attractive, particularly when one considers the flux in the core will continue to cycle around the remanent
the complexities of setting a distance protection scheme. It is value, creating a minor B-H loop. This will continue until the
inherently selective, allowing for sensitive detection of internal next high-current fault occurs. Then, depending on the point on
faults, while ignoring other events such as external faults, load wave (POW) at which a fault occurs, the remanent flux in the
current, and power swings. Another important benefit is that core may either increase, driving the CT into deeper saturation,
simultaneous tripping at both transmission-line terminals can or decrease, reducing or eliminating the level of saturation. The
be affected, irrespective of the relative current contribution former case may cause a major problem for a differential pro-
from the sources at both ends. tection relay if a second fault occurs outside the protected zone
The protection operates when the differential current exceeds (external fault) and one of the CTs has a high value of rema-
a set bias threshold. For external faults, the differential current nent flux. In this case, the differential current should be zero,
should be zero, but errors caused by CT saturation can result in a but it will no longer be a zero value and may exceed the bias
nonzero value. To ensure stability (i.e., to prevent maloperation, threshold, causing the differential protection to maloperate. In
the operating threshold is raised by increasing the bias setting). practice, stability is maintained even when an external fault oc-
This prevents maloperation by raising the setting of the relays curs by raising the bias threshold or by selecting a large over-di-
in proportion to the magnitude of the throughfault current. En- mensional CT that avoids saturation. The former decreases pro-
hanced stability is ensured by increasing the bias threshold at tection sensitivity, while the latter is not cost effective. Other
techniques [2], [3] describe how a saturation detection algorithm
Manuscript received February 26, 2004; revised June 17, 2004. This work can be applied to transmission feeders protected by a differen-
was supported in part by the U.K. Engineering and Physical Science Research tial protection scheme. These techniques along with the pro-
Council, in part by the National Grid Company, in part by Scottish Power, in part
by Hathaway Corporation, and VATECH Reyrolle. Paper no. TPWRD-00081- posed algorithm provide alternative means in coping with CT
2004. saturation (i.e., maintaining system stability without hindering
The authors are with The School of Electrical and Electronic Engineering, the sensitivity of the differential protection scheme).
Queen’s University of Belfast, Belfast BT9 5AH, U.K. (e-mail: n.villamagna@
ee.qub.ac.uk; p.crossley@ee.qub.ac.uk). The general concept of symmetrical components was first de-
Digital Object Identifier 10.1109/TPWRD.2005.848654 veloped by Fortescue in 1918 [4]. He introduced them for the
0885-8977/$20.00 © 2006 IEEE
VILLAMAGNA AND CROSSLEY: CT SATURATION DETECTION ALGORITHM 39

Fig. 2. The 400-kV EMTP system model.

Fig. 1. The 415-V transmission-line testbench.

decomposition of complex steady-state phasors. Symmetrical

components allow any unsymmetrical set of three-phase cur-
rents (or voltages) to be expressed as the phasor sum of three
symmetrical ac current (or voltage) components. The symmet-
rical component transformation matrix is given by (1)
Fig. 3. Symmetrical component evaluation within a current differential relay.

(1)
fault and SW2 is closed for an external fault. The type of fault
is determined by a user-selectable switch (i.e., LLL, LL, SLG,
where . and LLG). The line and source reactance and resistance can also
Symmetrical components is a powerful tool for the analytical be set by the user.
treatment of asymmetrical conditions in a three-phase system. From Fig. 1, the output current signals, from the real CTs, are
The proposed algorithm computes the positive-sequence ( ve captured by a six-channel digital oscilloscope. The discrete data
sequence), negative-sequence ( ve sequence), and zero-se- are then transported to MATLAB for processing.
quence components of the differential current and also monitors Fig. 2 is the Electro-Magnetic Transients Program (EMTP)
the rate of change of the sequence component currents. power system network model configured to represent a typical
Under normal operating conditions, the current phasors on a U.K. 400-kV interconnected power system in which power can
transmission feeder are approximately symmetrically balanced flow in either direction.
and become significantly unbalanced when an asymmetrical The system model is a 100-km transmission feeder with
fault occurs. Under normal operating conditions or during a six CT models [5]–three at the sending end and three at the
three-phase fault (LLL), the behavior of the feeder is deter- receiving end. The generated data are then transported to
mined using only the ve sequence current. For interphase MATLAB for processing. The EMTP CT model is described
faults clear of ground (LL), ve and ve sequence currents in Section III. The physical implementation of the current
occur and for single-line and double-line-to-ground faults (SLG differential protection is described in Fig. 3.
and LLG), all of the three sequence currents exist. Fig. 4 describes the functional elements in the current dif-
Transformation of the differential current from the phase do- ferential relay model implemented using MATLAB. The relay
main to the sequence component domain allows the differen- model is described in Section IV.
tial protection scheme to more sensitively detect the system
changing from a symmetrical condition to an asymmetrical fault III. EMTP CT MODEL
condition. Applying this concept to detect CT saturation gives
The transient response of CTs and the correct models in an
an early indication of a CT being driven into saturation.
EMTP simulation are very important in the evaluation of high-
speed relaying systems [6]. The model is especially important
II. SYSTEM MODEL for studying CT saturation, harmonics, and their affect on the
Fig. 1 describes the structure of an analog relay testbench performance of the protective relay. The model allows the user
protected by a global positioning system (GPS) synchronized to represent the effects of residual flux left in the CT after inter-
current differential relaying scheme. The analog 415-V three- ruption of a previous fault.
phase testbench is configured to represent a single-end-fed two- The hysteresis effects in the CT were modeled in EMTP
terminal transmission line. The testbench allows for in-zone and using the Type-96 nonlinear element and the auxiliary program
external faults to be applied. SW1 is closed to apply an in-zone HYSDAT [5]. HYSDAT automatically generates a hysteresis
40 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

Fig. 4. MATLAB relay model. Fig. 5. Dual bias current differential characteristic.

of 6400 samples/s (128 samples/cycle). The Fourier routine em-

loop when the saturation point (flux current: ) is speci-
ploys a sliding 20-ms data window that calculates a new 50-Hz
fied. This routine contains predefined trajectories in the
phasor every 2 5 ms.
plane to decide which path to follow when the flux increases
or decreases. The shape of the loop is a function of the core
material and the geometry [7]. EMTP supports one type of V. CT SATURATION ALGORITHM
material; the ARMCO M4-oriented steel. The CT knee-point The proposed algorithm monitors only the zero-sequence
voltage is given by (2) component of the differential current for detecting CT satura-
tion. For every 2 5-ms shift in the DFT data window, the rate
(2) of change of the zero-sequence differential current component
is computed by (5)
where is the root mean square (rms) knee-point voltage,
is the system frequency, and is the instantaneous flux. Since (5)
the model parameters are derived from CT test data, a degree of
validity is implied [6].
where is the zero-sequence differential current gradient
with respect to the bias current. is the initial (1st) zero-
IV. RELAY MODEL sequence differential current value computed with re-
A current differential relay was developed within MATLAB lating to the first DFT sliding data window and is the
[8]. The algorithm applied in MATLAB consists of a finite-im- initial phase domain bias current evaluated. The instant of a sig-
pulse-response (FIR) lowpass filter, a discrete Fourier trans- nificant rate of change [3] of differential current clearly marks
form (DFT) routine, and a model that implements a segregated- the start of CT saturation. The polarity of the gradient
phase dual slope percentage bias current differential protection distinguishes an in-zone fault from an external fault (see results,
scheme [9]. The operating characteristic of the dual slope, bi- Section VI and VII). For an external fault involving CT satu-
ased current differential element for a two-terminal feeder is ration, a “positive” significant rate-of-change value
shown in Fig. 5. The dual slope restraint characteristic is a form is classified as an external fault. For a “negative”
of adaptive restraint in which the magnitude of the restraint gradient value, an in-zone fault is classified.
quantity is increased for high-current conditions where CT sat- When a CT is driven into saturation, the third harmonic is
uration becomes more probable. largely responsible for the sharp peak in the waveform of the
The operating criteria is described by (3) and (4) CT current [11] (Fig. 9). By Fourier series expansion, [12] de-
scribes how all triplen harmonics are entirely zero sequence.
(3) That is, the similarity between the third harmonic component
(4) and the fundamental zero-sequence component is that they are
both cophasal. Therefore, since CT saturation always produces
where is the relay operating current. Equation (3) is applied a third harmonic component, a zero-sequence component will
when and (4) when . The sensitivity always be present in the differential current signal from a set of
setting of and in (3) is an important factor for the differ- saturated CTs. This is irrespective of whether the fault involves
ential relay because it is more susceptible to fixed errors, such a ground path. Therefore, for a LL fault clear of ground, the al-
as charging current and CT errors when [10]. gorithm is able to detect the zero-sequence current component
The FIR filter and the DFT routine process the input data and if a CT is driven into saturation. Where in [3], the phase domain
extract the power frequency current phasors at a sampling rate rate of change of differential current is monitored, the proposed
VILLAMAGNA AND CROSSLEY: CT SATURATION DETECTION ALGORITHM 41

Fig. 7. Testbench SLG (A 0 G) external fault.

Fig. 6. Current differential relay flowchart.

algorithm enhances sensitivity by monitoring the zero-sequence

differential current component to detect CT saturation.
The flowchart of the proposed algorithm describing CT sat-
uration detection is depicted as Fig. 6. The sequence domain
transformation as described by (1) computes the ve sequence,
ve sequence and zero-sequence differential current. Detection
of CT saturation is only initiated when the bias current is greater
than or equal to , (i.e., ) (Fig. 5) (i.e., where the
threshold setting is determined by the slope). The slope
is the higher percentage bias setting designed to improve relay
stability against CT saturation under heavy throughfault current Fig. 8. Phase-A differential current loci: external fault.
(external fault) conditions [13]. With p.u., this en-
sures that detection of CT saturation is only applied under heavy
with a balanced prefault load current. For external faults, an X/R
throughfault current conditions (external faults), where the relay
ratio of 40 was chosen, emulating the power system time con-
operates when as described in (4).
stant seen by the CTs for a fault at the source. The ratio for
Once the bias current is within the region of the slope
both sets of CTs was selected as 1:1. A CT burden was selected
threshold (i.e., ) and a positive rate-of-change value
to cause deep saturation (dependent on the point-on-wave of
in the zero-sequence differential current is detected (Figs. 8,
the fault). A number of tests were conducted in order to cap-
12, and 14) and as given by (5), an inhibit signal prevents the
ture a range of CT saturation levels (arbitrary point-on-wave
relay from initiating a “trip” command. The CT saturation
switching). The result chosen is one of the more severe cases
detection is then reset (i.e., released) once the magnitude of
of CT saturation.
the zero-sequence differential current is approximately equal
to zero . This ensures that the CT is no longer Fig. 7 is the CT current for a SLG external fault. At
operating in a saturated condition. the maximum dc offset, the phase-A CT current at the remote
end of the protected zone is driven into saturation for approxi-
mately 100 ms.
VI. TESTBENCH RESULTS
Fig. 8 describes the phase-A differential current verses bias
For the analog model power system (testbench), the CTs mea- current characteristic. The parameter settings for the dual bias
sure the primary current signals caused by a short-circuit fault. slope are p.u., p.u., ,
All tests conducted on the testbench are SLG faults .
42 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

Fig. 10. In-zone SLG (A 0 G) resistive fault.

current trajectories of both and . That is, a significant

rate of change in both and is no-
Fig. 9. Local and remote-end phase–a DFT data window. ticed and where by applying (5), value.
For in-zone resistive fault tests, an X/R ratio of 16 was chosen.
This is the typical value for a 400-kV transmission line in the
Each consecutive point (i.e., “o” and “x”) marks a 2 5-ms
U.K. The in-zone resistive faults were applied at the midpoint
shift in the DFT sliding data window. The “o” points mark the
of the transmission line with all CTs operating in ideal condi-
locus of the phase-A differential current and “x” is the
tions (i.e., CTs operating in the linear region of the B-H curve).
zero-sequence differential current . Note: is the
Fig. 10 shows how the differential current loci change
total phase current component, equal to the summation of the
as the through load current is increased. From Fig. 10, it can be
, , and zero sequence differential
seen that an increase in through load current results in a reduc-
currents. Fig. 8 shows that immediately after fault inception,
tion in the sensitivity of the relay.
there is an increase in bias current with no appreciative increase
It should be noted that Fig. 8 is a transient trajectory and
in differential current (i.e., the horizontal trajectory moving
Fig. 10 is a steady-state trajectory. The “dashed line” ( ve
from p.u.). The point of the “significant
derivative slope) in Fig. 10 signifies the transient trajectory as
rate of change” and marks
the fault resistance changes. Comparing Fig. 8 with Fig. 10,
the start of the current distortion caused by the saturation of the
it can be seen that the significant rate-of-change polarity of
CT core. From this point, a rapid increase in differential current
and is positive in Fig. 8 and that
is noticed. Both and have the same locus trajectory,
transient trajectory (denoted by the dashed line), ,
with being approximately one-third of the magnitude
depicted in Fig. 10 is negative as the in-zone fault resistance
of . This is as expected since the fault is and
changes. A significant rate of change with a positive gradient
. Also note that the significant rate of change com-
occurs when CT saturation is caused by a heavy throughfault
mences when 5 p.u. This agrees with
current (external fault) and a negative gradient occurs for an
p.u.), where CT saturation becomes more probable
in-zone fault.
due to a heavy throughfault current (external fault). The CT
saturation detection is initiated once the bias current exceeds
VII. EMTP SIMULATION RESULTS
the setting and then begins to monitor the rate of change of
the zero-sequence differential current (Fig. 6). For simulations involving CT saturation, both external and
Fig. 9 shows the DFT sliding data windows of the phase-A in-zone faults were considered. A transmission-line X/R ratio
current signal. Each data window marks a 2 5-ms shift. of 40 was chosen for both cases. To emulate the worse case sce-
Fig. 9(a) shows that at the point of fault inception, the local nario, the CTs at the receiving end were prevented from being
and remote-end current rapidly increases and there is no distor- driven into saturation.
tion in the current signals. This is seen in Fig. 8, where the trajec- The in-zone resistive fault simulations all involve SLG
tory shows an increase in bias current (i.e., throughfault current) faults, with all of the CTs operating in an unsaturated con-
with no developed differential current. Fig. 9(b) corresponds to dition. An X/R ratio of 16 was chosen and all of the faults were
the next shift of DFT data window. The window captures the applied at the midpoint of the transmission line.
start of the distortion in the remote-end current. At this point, For all of the tests, a CT ratio of 2000:5 with a knee-point
a significant rate of change in differential current is developed. voltage of 300 V given by (2) and a resistive burden of 1 5
This is seen in Fig. 8 as the first initial rise in the differential was chosen.
VILLAMAGNA AND CROSSLEY: CT SATURATION DETECTION ALGORITHM 43

Fig. 11. EMTP: SLG (A 0 G) external fault.

Fig. 13. Sending and receiving-end phase-A DFT data windows.

The DFT window in Fig. 13 captures a section of the first

two cycles of CT saturation. Fig. 13(a) captures the prefault
and fault inception. Fig. 13(b) is shifted 2 5 ms from the first
window, capturing the start of the “significant rate-of-change”
point. Fig. 13(c) is shifted 4 5 ms from the first window, cap-
turing the entire first half a cycle showing maximum current dis-
tortion. Fig. 13(d) is shifted 37 5 ms (second cycle) from the
first window, capturing the “deep saturation” (max. and
, at min. ) cycle. Taking the “numeric area of integra-
tion” (i.e., the area under the curve), we see that in Fig. 13(d),
the area under the curve is less than in Fig. 13(c). This gives rise
to a reduced bias current against a maximum rise in differential
current, as seen in Figs. 12 and 14 (denoted “deep saturation”).
Fig. 12. Phase-A differential current loci: external fault.
For the LL external fault, the phase-B differential
current loci are shown. A residual flux of 80% was chosen
Two types of external faults that cause CT saturation were for the phase-B sending-end CT. Both the phase-B and phase-C
considered: i) SLG ; and ii) LL faults. CTs at the sending end were driven into saturation, with all other
In both cases, the point on wave of fault inception was such as CTs remaining in the unsaturated state. The differential current
to cause maximum asymmetry in phase-A (for SLG fault) and loci of phase-B are described in Fig. 14.
phase-B (for LL fault). Observing Fig. 14(a) and (b), the positive-sequence
For the SLG external fault, the phase-A CT at the sending and the negative-sequence differential current trajecto-
end is driven into saturation half a cycle after fault inception, as ries show close similarities. Fig. 14(c) shows that the zero-se-
shown in Fig. 11. A residual flux of 80% was selected for the quence differential current trajectory is less than one-
sending-end phase-A CT. All other CTs remained in an unsatu- third of the magnitude of . This is as expected since for the
rated condition. LL fault, the zero-sequence differential current is due only to
Within the second cycle, the CT is driven further into satu- the predominately large third harmonic component which exists
ration, denoted as “deep saturation.” This is depicted in Fig. 12 only when the CT is saturated (i.e., it has the same characteristic
where the bias current decreases when the differential current as the zero-sequence component). In this case, for an LL fault,
increases. the zero-sequence differential current is less than the zero-se-
Fig. 12 shows the developed phase-A differential current quence differential current developed for an SLG fault, since
characteristic due to saturation of the sending-end CT. Note: there is no ground path. Noting that for an LL fault in which
both and have the same locus trajectory, with the CT does not saturate, only the positive-sequence and nega-
being approximately one-third of the magnitude of . tive-sequence current component exist in which .
44 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 1, JANUARY 2006

Fig. 15. Phase-A differential current loci: in-zone fault.

Fig. 16. In-zone SLG (A 0 G) resistive fault.

driven into saturation. A residual flux of 80% was chosen for
the phase-A sending-end CT. All of the other CTs operated in
the unsaturated condition.
Referring to Fig. 15, the initial significant rate of change of
and both lie between a perpendicular gradient from
the point of prefault, to a negative gradient. For the perpendic-
ular gradient, the derivative is infinite (i.e., for a tangent at 90 ,
an infinite value is evaluated). Also noting that at the point of
fault inception, there is an immediate rise in differential current
and no immediate increase in bias current. This trajectory
(Fig. 15) notably differs from the trajectory seen in Fig. 8,
Fig. 12, and Fig. 14, which are characteristic to external faults.
Since the bias current at the point of fault inception is less
than the setting (Fig. 15), a “trip” command from the relay
can be initiated, (providing it exceeds the slope threshold)
Fig. 14. Phase-B differential current loci: external fault.
(i.e., the CT saturation detection is bypassed).
Fig. 16 describes the steady-state differential current
Fig. 15 describes the differential current trajectory for an SLG loci for various resistive in-zone faults against an increase in
in-zone fault when the phase-A sending-end CT is through load current. The “dashed line” indicates the transient
VILLAMAGNA AND CROSSLEY: CT SATURATION DETECTION ALGORITHM 45

trajectory ( ve derivative) of the differential current (i.e., as the [6] D. A. Tziouvaras and P. McLean et al., “Mathematical models for cur-
fault resistance value changes during the fault period). rent, voltage and coupling capacitor voltage transformers,” IEEE Trans.
Power Del., vol. 15, no. 1, pp. 62–72, Jan. 2000.
Both Fig. 10 and 16 show the situation of how the relay is de- [7] M. Kezunovic and L. Kojovic et al., “Experimental evaluation of EMTP-
sensitized as the through load current increases. This compro- based current transformer models for protective relay transient study,”
mise between maintaining system stability and relay sensitivity IEEE Trans. Power Del., vol. 9, no. 1, pp. 405–413, Jan. 1994.
[8] N. Villamagna, P. A. Crossley, and H. Y. Li, “GPS synchronized cur-
is mainly related to the probability of the CT/CTs, being driven rent differential protection with adaptive bias operating characteristic,”
into saturation during a heavy throughfault current condition. in Proc. Int. Symp. Modern Electric Power Systems, Wroclaw, Poland,
2002.
[9] H. Ito and I. Shuto et al., “Development of an improved multifunctional
VIII. SUMMARY transmission line protection,” in Developments in Power System Protec-
tion, 2001, Conf. Pub. 479.
The major conclusion of this study is that the proposed al- [10] M. Yamaura, Y. Kurosawa, and H. Ayakawa, “Improvement of internal
gorithm allows for a differential relay to improve its’ ability to charging current compensation for transmission line differential protec-
be more sensitive while still maintaining the high level of pro- tion,” in Proc. Developments Power System Protection, 1997, Conf. Pub.
434.
tection stability. Due to the probability of CT saturation occur- [11] Members of the staff of Dept. Elect. Eng., Massachusetts Inst. Technol.,
ring during a heavy throughfault condition, the slope in (4) Magnetic Circuits and Transformers. New York: Wiley, 1946, pp.
is used to enhance protection stability when , but at 173–184.
[12] G. J. Wakileh, Power Systems Harmonics—Fundamentals, Analysis and
the expense of reduced protection sensitivity. By monitoring the Filter Design. Berlin, Germany: Springer, 2001, pp. 13–15.
zero-sequence differential current, the results presented show [13] GEC ALSTOM, Digital Current Differential Relay—Type: LFCB 102,
that the slope can be appreciably reduced or reshaped to pro- Techn. Manu.
vide increased sensitivity while still maintaining the high level
of stability. This is achieved by having bias threshold charac-
teristics in (4), sensitive to the zero-sequence differential cur-
Nicholas Villamagna received the Electrical Fitting (Electrical Technician)
rent for when bias . This also covers LL faults clear of diploma from the Fremantle Technical College, Perth, Western Australia, in
a ground path, where results have shown that when the CT sat- 1986. He received the B.Eng. (Hons.) degree in electrical engineering from
urates owing to LL faults, a zero-sequence differential current Brunel University, Brunel, U.K., in 1998 and the M.Sc. degree in electrical
power systems from the University of Manchester Institute of Science and
can be detected due to the third harmonic current component. Technology, Manchester, U.K., in 2000. He is currently pursuing the Ph.D.
The results obtained from the analog testbench with those degree in power system protection at Queen’s University of Belfast, Belfast,
from the simulation tests have shown a high degree of similarity. U.K.
He was in industry for 13 years, specializing in machines and control systems.
This confirms that for any further investigation, simulations in
EMTP will give realistic results.

REFERENCES Peter A. Crossley (M’96) received the B.Sc. degree from the University of
[1] P. A. Crossley, H. Y. Li, and A. D. Parker, “Design and evaluation of a Manchester Institute of Science and Technology (UMIST), Manchester, U.K.,
current differential relay test system,” IEEE Trans. Power Del., vol. 13, in 1977 and the Ph.D. degree from the University of Cambridge, Cambridge,
no. 2, pp. 427–433, Apr. 1998. U.K., in 1983.
[2] Y. C. Kang, S. H. Ok, and S. H. Kang, “A CT saturation detection algo- Currently, he is Professor of Electrical Engineering at Queen’s University of
rithm,” in Proc. PowerTech, Bologna, Italy, 2003. Belfast, Belfast, U.K. He had been involved in the design and application of
[3] SEG, Differential Protection Relay for Generators and Motors, Model digital protection relays and systems for 25 years, first with GEC, then with
IRD1-G, Tech. Manu. ALSTOM, UMIST, and later with Queen’s University of Belfast, Belfast, U.K.
[4] C. L. Fortescue, “Method of symmetrical coordinates applied to the so- He has published many technical papers on protection.
lution of polyphase networks,” Trans. AIEE, vol. 37, pp. 1027–1140, Dr. Crossley was the Chairman of the 2001 IEE Development in Power
1918. System Protection Conference in Amsterdam, The Netherlands, and is an active
[5] EPRI, Electromagnetic Transients Program (EMTP), Version 1, Revised member of various CIGRE, IEEE, and Institution of Electrical Engineering
Rule Book, Vancouver, CA, vol. 1, Apr. 1986. committees on protection.

View publication stats