Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Duan 2009

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Study of the Inrush Current Identification Using

the Improved Half-Cycle Fourier Analysis


Duan Jiandong Wu Chang Yu Jianming
Department of Electrical Engineering Department of Electrical Engineering Department of Electrical Engineering
Xian University of Technology Xian University of Technology Xian University of Technology
Xian, China Xian, China Xian, China
e-mailduanjd@xaut.edu.cn e-mailRabbi5585@sohu.com e-mailjianmingyu@xaut.edu.cn
AbstractAmong the traditional schemes of current differential the CT transient saturation, the inrush current will emerge
protection for power transformer, the principles of second symmetrical characteristic relatively, which will result in the
harmonic restraint and dead angle restraint to distinguish inrush decrease of the dead angle in inrush current waveform [3].
current from fault current are influenced such many factors as These factors have great effect on the reliability of the inrush
the decrease of transformer core magnetic saturation and CT current identifying, so far as to misidentification. So, it is very
transient saturation. In fact, the inrush current is characterized necessary to improve or make a further research on the inrush
of its peaky wave and its relatively big dead angle compared with current identification method.
the internal fault current. So the paper proposes a novel method Generally, there are great differences between inrush
to identify the inrush current using the improved half-cycle
current waveform and fault current waveform, since the
Fourier analysis (HCFA). The improved HCFA can accurately
extract the fundamental component and reflect the characteristic
extraction and calculation of waveform characteristic is more
difference of the inrush current waveform. The proposed method simple and can be easily applied, the identification method of
is tested by a large amount of EMTDC simulation on various inrush current based on waveform characteristics has been
conditions, the dynamic analogous test data from different extensively studied [6,7,10]. Based on the difference of
transformers. These test results indicate that the novel method waveform characteristics between inrush current and
to identify inrush current is basically correct and feasible. non-inrush current, a novel method to identify the inrush
current based on the improved half-cycle Fourier analysis is
Keywords - Transformer protection, Inrush current, Internal proposed in this paper.
fault current, Fourier transform
II. GENERAL WAVEFORM CHARACTERISTIC OF INRUSH
I. INTRODUCTION CURRENT AND FAULT CURRENT
Power transformers, are important equipments for power Transformer core saturation is the basic reason why the
transmission and transformation in power system, they have a transformer inrush current is formed [4]. In normal condition,
decisive effect on reliable transmissionflexible distribution the transformer core is unsaturated; core work at the linear
and safe service of electric power. So the safety operation of area, at this time, the exciting current in the windings is very
transformers is directly associated with the safety and small, approximate to zero. Once the transformer is operated
stabilization of power system. However, the correct operation or abnormal conditions occur, with the change of core flux, the
rate of transformer protection is relatively low for a long time. core may become saturable, and exciting current in the
According to the statistic, correct operation rate of transformer windings is rather high (can reach 10 times of rated current in
protection for 220kv and above power system in 2004 was serious situation), which called inrush current, and inrush
only 79.05% [1]. Therefore, the performance of transformer current is characterized of its peaky wave and its big dead
protection is extremely urgent to be improved. angle generally [5], as shown in Fig.1.
Now, Current longitudinal differential protection is used as The transformer core magnetic flux change periodically
the primary transformer protection for its simple principle and between linear condition and saturation condition, inrush
high sensitivity. However, differential principle is suitable for current is shown in Fig.1, a lot of harmonics included, its
only circuit that satisfies Kirchhoff's current law, while there fundamental component and harmonic component will change
is some relationship between transformers internal magnetic widely and periodically with the change of core flux,
circuits, and exciting current will exist in excitation circuit, especially, the fundamental component of inrush current is
then the circuit cannot meet Kirchhoffs current law largely different in different time zone.
essentially. Especially, inrush current turns to be the main Typical fault current waveform when transformer internal
source of unbalanced current and may induce mis-operation of fault occur which is shown in Fig.2, basically, its sinusoidal
differential protection. The inrush current is mainly identified wave which is composed of decaying DC component and a
by the second harmonic restraint principle and dead angle small amount of harmonic component, and the fundamental
restraint principle [2]. But, with the decrease of transformer component and harmonic component in fault current have no
core materials magnetic saturation point, the second harmonic change basically (very small) as time progress.
content in inrush current will reduce; on the other hand, due to
This work was supported in part by the National Natural Science
Foundation of China (50707026), the Science Research Project Fund of
Shaanxi Provincial Education Department07JK329, and the National
Natural Science Foundation of Shaanxi Province (SJ08E220).

978-1-4244-2487-0/09/$25.00 2009 IEEE


IV. ALGORITHM OF INRUSH CURRENT IDENTIFICATION
The practical criterion based on the improved half-cycle
Fourier analysis is given in the following. Assume that the
transformer differential protective system is started, for the
sampling time t1, t2,,tN in a period of time (for example a
period of power frequency) in the following, utilize the
 improved half-cycle Fourier analysis equation (1) and (2) to
Fig.1 Typical inrush current Fig.2 Typical fault current work out the fundamental current peak values I1, I2,,IN in a
half period of power frequency after the sampling time. For
III. IMPROVED HALF-CYCLE FOURIER ANALYSIS convenience, transform the fundamental current peak value
Analysis above indicates that, the change degree of the into relative value:
fundamental component in inrush current and fault current is Ik ' = Ik / I (k = 1,2, , N ) (3)
of significant difference as time progress. By detecting the
difference of fundamental current component in different time Where I is the mean value of I1, I2,, IN, then, work out the
zone, we can find a way to identify inrush current. If there is value of discriminant parameter D:
great change of fundamental current content in different time D=
1
[( I1' 1) 2 + ( I 2' 1) 2 + + ( I N ' 1) 2 ] (4)
zone, the current is judged as inrush current. This is the new N
principle thought to identify inrush current in this article. Parameter D reflect the variance of fundamental currents peak
The most important part in this algorithm is the extraction values actually (relative value). According to D, inrush current
of fundamental current component. At present, the full wave is identified:
Fourier algorithm is adopted to extract fundamental z If D>, differential current is judged to be inrush current;
component, the full-wave Fourier analysis reflect the synthesis z If D<, it is judged to be fault current.
characteristics of waveform in one cycle. Besides, half-cycle In the criterion, is a threshold value that is very small.
Fourier analysis is also adopted to extract fundamental Detailed theoretical derivation, simulation study and dynamic
component, this algorithm can not only increase response analogous test, all of the above indicate that make =0.05 is
speed but also can reflect the characteristics of the peak values relatively appropriate.
of fundamental wave in half cycle. However, traditional Assume that is the initial angle of the improved
half-cycle Fourier analysis is affected greatly by non-periodic half-cycle Fourier analysis. Then algorithm for real and
DC component [8,9]. An improved method based on the imaginary part of the fundamental peak value is:
improved half-cycle Fourier analysis is adopted to extract T
fundamental component of current waveform in different time +
4 100 2
zone, the improved algorithm is the composition of the real a1 ( ) = i ( t ) cos( 1t ) dt (5)
part of traditional half-cycle Fourier analysis and T
100
Mann-Morrison algorithm [10,11,12]. First of all, use
a ( + ) a1 ( )
traditional half-cycle Fourier analysis to get the real part IRe(k) b1 ( ) = 1 (6)
of fundamental current component, k is sampling point; then 2 sin( )
according to the Mann-Morrison algorithm, the imaginary part Mean value of the fundamental peak values:
of fundamental current is obtained ( is sampling interval): 1 2 2 2
I (k + 1) I Re (k + 1) A= a1 ( ) + b1 ( ) d (7)
I Im (k ) = Re (1) 2 0
2 sin( ) Discriminant parameter D is:
Then the peak value of fundamental is : 2 2
1 2 a1 ( ) + b1 ( )
2 2
I1 (k ) = I Re (k ) + I Im (k ) (2) D= ( 1) 2 d (8)
2 0 A
Use the improved half-cycle Fourier analysis to extract the From reference [4] we know that single-phase transformer
fundamental component from the current shown in Fig.1 and inrush current can be described as follows:
Fig.2 respectively, the fundamental component change as the 0 < S
time progress as shown in Fig.3. Fig.3(a) is the fundamental
i (t ) = 1 BS Br (9)
change condition of inrush current, Fig.3(b) is that for fault L U m [cos cos(1t + ) B ] S
1 m
current, obviously, the wave show that, as time go on, the
fundamental peak values of inrush current change greatly, but Where Um represent the peak value of rating voltage, SBS
its stable on the whole for fault current. represent saturation flux and saturation flux density
respectively, Br represent remnant flux density, Bm represent
rating flux density corresponding to Um, L represent the
inductance in transformer excitation loop, 1 represent
angular frequency of fundamental, represent the initial
angle when transformer switch on.
For large-scale power transformers, the possible minimum
value of BS is 1.15 times the value of Bm, the possible
R
(a) inrush current (b) fault current maximum value of Br is 0.9 times the value of B, when is 0 ,
Fig.3 Fundamental current curve by Improved half-cycle Fourier analysis
and other conditions are constant, the dead angle of inrush A. Simulation of identification algorithm on typical conditions
current is minimum, secondary harmonic content is also On some typical transformer operation or fault conditions,
minimum, that is the practically serious condition for the different current waveformd of transformer are shown in
analysis of inrush current [4], the minimum value of Fig.5(a)~(d), As shown in Table I, D values of discriminant
discriminant parameter D worked out on this condition is parameter and identification results are presented for the
0.0897. identification algorithm. If D is greater than threshold value
From reference [14] we can get the mathematic model of =0.05, the current is judged to be inrush current, or else judged
fault current: to be fault current, from Table I, we can see, these

t
M
identification results on typical conditions are correct.
i (t ) = I 0e + I k cos(k1t + k ) (10) TABLE 1 INRUSH CURRENT IDENTIFICATION RESULT ON TYPICAL CONDITIONS
k =1
where I0 represent the peak value of decaying DC component, Transformer conditions Value of D
Identification
represent the decaying time coefficient of decaying DC result
Unload switch on and unsymmetrical
component, Ik represent the peak value of fundamental and kth 1
inrush current is generated
0.3720 Inrush current
harmonic, k represent the initial angle of fundamental and Unload switch on and symmetrical
2 0.1143 Inrush current
kth harmonic, 1 represents the fundamental angular inrush current is generated
frequency. 3 Internal single phase to ground fault 0.0066 Fault current
From reference [15] we can know that, when using the Switch on with no-load and internal
4 0.0012 Fault current
single phase to ground fault
half-cycle Fourier analysis to extract fundamental peak value,
the third harmonic and above have little effect on that
algorithm, and the half-cycle Fourier analysis itself can filter
odd harmonics, so the mathematic model of fault current can
be simplified as:
t
2
i (t ) = I 0e + I k cos(k1t + k ) (11)
k =1 
(a) unsymmetrical Inrush current (b) symmetrical Inrush current
where parameter selection range are as follows: the peak value
of decaying DC component: I0=0~2I1; the decaying time
coefficient of decaying DC component ~PV; the peak
value of second harmonic: I2=0~1I1; the initial angle of second
R
harmonic:   ~; The Initial angle of fundamental
R R
  ~ ; The maximum value of discriminant parameter D

obtained from the algorithm is about 0.0001 with all these (c) phase to ground fault (d) unload and phase to ground fault
conditions considered. Fig.5 Differential current waveform of transformer
V. EMTDC SIMULATION TEST STUDY
B.Simulation test for the algorithm to identify inrush current
For the Method to identify inrush current based on the
For the conditions transformer switch on with unload,
improved half-cycle Fourier analysis, lots of simulation is
light-load, full-load, internal fault, operate with internal fault,
proceeded using the electromagnetic transient simulation
etc, the performance of the algorithm is further studied on
software EMTDC. System simulation model is shown in
concrete conditions such as different switch angle, different
Fig.4, the transformer adopt Y/y0 rejoining method, 250MVA
residual magnetism, different fault condition and so on,
for capability, the voltage grade is 400/132kV, three DC
simulation result is shown in Table II-V. Limit to the length of
current supply is used to simulate residual magnetism.
the article, only discriminant parameter D values in phase A
Simulation conditions contain transformer switch on with
are given.
unload, light-load, full load, operate with internal fault, switch
on with external fault, internal fault, etc. Differential current of TABLE 2 D VALUES FOR TRANSFORMER UNLOAD SWITCH ON
transformers high voltage side and low voltage side id is what Switch Residual magnetism
the Inrush current identification algorithm analyzes. angle r = 0 S r = 0.7 S r = 0.8 S
0o 0.1608 0.1521 0.1502
30o 0.2640 0.1433 0.1901
60o 0.3646 0.1764 0.0931
90o 0.0039* 0.1477 0.2983

TABLE 3 D VALUES FOR TRANSFORMER SWITCH ON WITH LOAD


Switch on with
Switch on with light-load
Switch angle full-load (cos=0.8)
r = 0 S r = 0.7 S r = 0.8 S r = 0 S
0o 0.2826 0.2890 0.0944 0.2108
30o 0.2951 0.2956 0.3085 0.2446
60o 0.3154 0.3106 0.3089 0.3351
90o 0.0023* 0.1850 0.2350 0.0006*
Fig.4 Transformers EMTDC simulation model
TABLE 4 D VALUES FOR TRANSFORMER SWITCH ON WITH INTERNAL FAULT accurately, and no mistaken judgment for fault current when
Phase A internal earth fault Phase A transformer with external fault and internal fault, its
Switch
angle r = 0 S r = 0.7 S r = 0.8 S
turn-to-turn obviously the dynamic analogous data have verified the
fault(4.6%) feasibility of the proposed method.
0o 0.0068 0.0082 0.0085 0.0032
30o 0.0045 0.0058 0.0063 0.0023 VII. CONCLUSIONS
60o 0.0013 0.0016 0.0021 0.0002
90o 0.0000 0.0000 0.0000 0.0000 Inrush current waveform in transformer differential current
is characterized of its peaky wave and its big dead angle
TABLE 5 D VALUES FOR TRANSFORMER WITH INTERNAL FAULT relatively. Based on this, new method to identify inrush
Phase A
current is proposed, which adopts a fast improved Fourier
Switch Phase A internal Analysis integrating the real part of half-cycle Fourier analysis
turn-to-turn
angle earth fault and Mann-Morrison analysis Theory analysis, EMTDC
fault(4.6%)
0o 0.0066 0.0032 simulation test indicate that this method can identify inrush
30o 0.0044 0.0023 current and internal fault current effectively, and also have a
60o 0.0013 0.0002 good identification effect on symmetrical inrush current.
90o 0.0000 0.0000
REFERENCE
In Table II-V, rS are core residual magnetism and
saturation flux respectively. [1] Yu-lan ZHOU, Yu-ling WANG, Man-yong ZHAO, "Statistics of opera-
tion situation of protective relayings and automation devices of power
The simulation results shown in Table II-V indicate that all systems in China in 2004," Power System Technology,2005,29
the discriminant parameter values are greater than the (16):42-48.
threshold value =0.05 whether transformer with load on or [2] Zhi-guo HAO, Bao-hui ZHANG, Yun-long CHU, "Trend and situation
off in different conditions, and far greater than the of distinguish technology of transformer magnetizing inrush current,"
discriminant parameter values when transformer with internal Transformer,2005,42(7):23-27.
fault (generally, values of D are less than 0.05 when [3] Ben-teng HE, Xi-dong XU, "Protection based on wave-comparison,"
transformer with internal fault). Furthermore, the last line in Proceedings of the CSEE, 1998,18(6):395-398.
Table II and Table III, the values which with * are less than [4] Wei-jian WANG, Bing-yun HOU, "Theoretical fundamentals of large-
scale generator-transformer unit protection (The second edition),"
0.05, that is because core residual magnetism is 0, meanwhile Beijing: China Electric Power Press,1989.
the voltage change rate is very small, then result in the core [5] Wei-jian WANG, "Generator and transformer relaying protection
transient flux is very small, core can not be saturated, so it can application (The second edition)," Beijing:China Electric Power Press,
not generate inrush current. Lots of EMTDC simulation test 2005.
indicate that, the Method proposed to Identify the Inrush [6] Zheng-qing HAN, Shi-bin GAO, Qun-zhan LI, "New method to identify
Current Based on the improved half-cycle Fourier analysis is inrush current based on half-wave Fourier analysis," Automation of
correct, and with high-sensitivity. Electric Power Systems,2005,29(14): 60-63.
[7] Jun CHEN, Wei YAN, Jia-sheng CHEN, Zi-gang GUO, Quan- rong
VI. VERIFICATION WITH DYNAMIC ANALOGOUS DATA SHEN, "Novel identification method for magnetizing inrush current
based on the comparability of different Fourier algorithm," Automation
TABLE VI D VALUES FOR TRANSFORMER DYNAMIC ANALOGOUS of Electric Power Systems, 2006, 30(6): 21-24.
Test conditions Phase A Phase B Phase C
[8] Fu-de JIN, Le HUANG, "The compensating characteristic analysis of
0.0030 0.0041 0.0042 Fourier algorithm," Relay, 2005,33(21):45-50.
External metallicity fault of transformer
0.0001 0.0003 0.0002 [9] Jing GAO, Jian-yong ZHENG, Zhen-don PAN, "Study of improved
0.1463 0.1768 0.2169 Fourier algorithm for digital protection in power system," Relay, 2002,
0.1614 0.2302 0.2686 30(10):16-20.
Transformer unload switch on and inrush 0.2007 0.1829 0.1841 [10] Yong-li LI, Chao-ying CHEN, Jia-li HE, "A fast algorithm based on
current is generated 0.2740 0.3153 0.2020 half-cycle Fourier algorithm for protective relaying," Power System
0.2212 0.1392 0.1867 Technology, 1996,20(1):52-55.
0.1601 0.2233 0.2592 [11] A G Phadke, J S Thorp, "Computer relaying for power system,"
Transformer turn-to-turn fault (3%) 0.0004 0.0001 0.0001 Research Studies Press Ltd. Taunton, Somerset, England, 1988.
Switch on with internal phase A-B fault [12] B J Mann, I F Morrison, "Digital calculation of impedance for
0.0266 0.0310 0.0860 transmission line protection," IEEE Trans on PAS,1971;90(1): 270-279.
of on medium voltage side
Switch on with turn-to-turn fault (3%) 0.1153* 0.1369* 0.2112* [13] Jia-le SUONAN, Zai-bin JIAO, Yi-ning ZHANG, Wen-tao LIU, Dong
LIU, "A fast algorithm to identify inrush current based on waveform
Using the dynamic analogous data of relay protection in factor," Power System Technology, 2006,30(11):71-76.
750kV system of China Electric Power Research Institute, [14] Shi-wen SHI, "Large-scale generator-transformer unit protection,"
verification for the method to identify inrush current proposed Beijing: China Electric Power Press,1987.
in this paper is presented [13]. Discriminant parameter values [15] Bin LI, Yong-li LI, Jia-li HE, "Accurate and fast filtering algorithm for
fundamental component," Automation of Electric Power Systems,2006,
of inrush current for transformer with external fault, switch on 30(10):39-43.
with no-load, internal fault, are shown in Table.VI, except for
the transformer internal fault with small windings is not
identified, as shown in the last line, discriminant parameter
values of inrush current worked out by the method are greater
than 0.05 obviously, then the inrush current is identified

You might also like