Embed Size (px)
Citation preview
CC II RR EE DD 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
CIRED2005 Session No 3
DIRECTIONAL RELAYS WITHOUT VOLTAGE SENSORS FOR DISTRIBUTION NETWORKS: USE OF SYMMETRICAL COMPONENTS AND EFFECT OF THE DISTRIBUTED GENERATION
Marc PETIT,
Supélec – France [email protected]
Patrick BASTARD, Supélec – France
Xavier LE PIVERT, Armines – France
Christophe POULAIN, Schneider-Electric – France [email protected]
INTRODUCTION Fortescue’s theory introduced in 1918 [1] allows to study a 3-phase coupled electrical network as 3 independent uncoupled circuits. It supposes the electrical network is symmetric and linear: self and mutual impedances of each phase are equal. The decomposition gives 3 independent sequences: zero (o), positive (1) and negative (2) sequences (symmetrical base). Real (A, B, C) and symmetrical (0, 1, 2) signals are linked by:
32
2
1
0
2
2
2
2
2
1
0
11
111
11
111
31
πj
C
B
A
C
B
A
eawithSSS
aaaa
SSS
andSSS
aaaa
SSS
=⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡=
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡
These three sequences are very easy to calculate with the numerical techniques currently used in protection devices. Currents and/or voltage signals are acquired through sensors, and then the phasors (module and argument) are calculated with a Fourier transform, and finally the Fortescue’s transform gives the symmetrical sequences. The development of numerical techniques allows the conception of new protection algorithms using the symmetrical components. The calculations of symmetrical components in various fault types and charge disequilibrium are widely described in numerous books [2]. In a previous paper [3] the use of symmetrical components to remove voltage sensors in directional relays was presented. The established criterion uses the argument of the ratio I2 / I1sound (I1 sound is the positive sequence current before the fault and I2 the negative sequence current with fault). Nevertheless the distributed generation was not taking into account. Now the electricity market liberalisation in many countries will increase the number of independent generators (winding generators). These generators must be taken into account in the conception of new algorithms and in the behaviour analysis of existing algorithms. Indeed, the generators’ behaviour is different in the positive and negative sequences and the power flows in the feeders will be modified. MODELISATION OF A FAULTY NETWORK To study a faulty network with the symmetrical components, the network is subdivided in two parts: the sound part and the faulty part. The sound part is symmetric and the Thevenin’s equivalent schemes of the zero, positive and negative sequences are calculated viewed from the fault point using the
equivalent impedances of each element. The fault is considered as an unsymmetrical 3-phase charge and its equations in the symmetrical base are calculated from its real characteristics and Fortescue’s transformation. Then three electric schemes are deduced in the symmetrical base, the fault being represented as three current sources. In the case of a phase-to-earth fault (main faults encountered) on the phase “A”, the current sources are: where Zfault is the fault impedance. EQUIVALENT SCHEMES OF A SYNCHRONOUS MACHINE In the following example the distributed generation (DG) is described by a synchronous machine with its three equivalent schemes in the (0,1,2) domain. Such a machine is sometimes used for windmill with its frequency converter for a variable speed use. Nevertheless DFIG asynchronous generator are more and more used for large power as it enables a control of both active and reactive powers. But regarding to the present study there is little difference because both generators have similar X0, X1, X2 reactance values. The value of EDG (emf of the source in the positive sequence) depends on the active and reactive powers transferred by the DG to the network. The value chosen for X1 depends on the phenomena we want to characterize:
• behaviour before the fault (steady-state) X1 • behaviour just after the fault X1’’ (sub-
transient regime) • transient regime X1’
X1’ is the value commonly used for the positive sequence impedance to study the faulty network behaviour because a detector must make its decision in the 0,1 – 1 s range. It is rarely acceptable to expect the complete steady-state (t > 1 s). DESCRIPTION OF THE DISTRIBUTION NETWORK The studied network consists of a busbar energized by a HV network through two HV / MV transformers (fig.1). Four detectors are placed to detect faults downstream the busbar (A & B) or faults on the feeders (C & D). All the simulations were made with EMTP.
faultZZZZEIII
3210
1210 +++===
CC II RR EE DD 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
CIRED2005 Session No 3
Fig.1 : MV network fed by two transformers and with the DG on a radial feeder. The detectors A, B and C are studied. The fault is located at one of the four indicated places. The characteristics of the network are: HV network characteristics : 63 kV, 560 MVA, 10 km HV line (0,3 Ω/km) Transformers’ characteristics: 63 / 20 kV, 30 MVA, uCC = 15% DG characteristics (type 59 EMTP model): round rotor, Ra = 0.005, Xf = 0.0665, Xd = Xq = 2.5, X’
d = X’q = 0.25, X’’
d = X’’
q = 0.15,T’d0 = 5 s, T’
q0 = 1 s, T’’d0 = 50 ms, T’’
q0 = 80 ms, Grounding system: resistive or compensated. Charges: power factor = fp = 0,9. In our example the MV busbar is fed through two HV/MV transformers. So directional relays are indispensable in the following two situations: (i) fault upstream the busbar and (ii) fault on capacitive lines (cabled lines). When a fault occurs upstream the detector A, the fault current will be driven in the low impedance loop represented by the two transformers and the busbar. Then the detector B will also see a fault current and will open the breaker. To prevent such a maloperation, a directional relay with a voltage sensor is classically used to locate the fault downstream or upstream the detector. But the voltage sensors generate an additional cost. To suppress these voltage sensors, an algorithm using the symmetrical components was proposed in a previous paper [3]. This algorithm uses the argument of the ratio I2 / I1sound (I1 sound is the positive current before the fault) to detect a fault upstream or downstream the detector. Unfortunately the argument of this ratio becomes very uncertain when a distributed generator (DG) is connected to a MV feeder, especially for a compensative grounding [4]. In the paper [4] the argument of the ratio I2 / I1sound was precisely studied using analytics formula, and then the limits of this ratio were identified and confirmed by the EMTP simulations. The main drawback comes from the positive current that depends on the power flows on the feeders. These power flows depend both on charges power and power injected by the DG. Then we proposed to use communication techniques between each detector (A, B or C) and the detector (D) on a feeder without DG. This solution is no more a local one. Then in this paper we propose a local solution, always using the symmetrical components of the currents: • Use of the zero and negative sequence components that
don’t depend on power flows. The restriction is that the zero sequence only exists for a fault to ground.
USE OF THE ZERO AND NEGATIVE SEQUENCE CURRENTS Description If a detection algorithm with only local criteria is required (due to investment costs) the positive sequence current can not be used when a detector sees a generator downstream (detectors A, B and C). Then information must be extracted from zero and negative sequences but it induces that only fault to ground can be detected. We will discuss from the analytics formula of the I2 / I0 ratio that does not depend on the active and reactive power flows before the fault. The approximated analytics formulae of the I2 / I0 ratio for the detectors A, B & C are given in the table I. These formulae show that arguments depend on both neutral grounding (ZN) and capacitive current. Figures According to these approximated formulae, the sign of the I2 / I0 ratio arguments’ allows to distinguish an upstream fault from a downstream one. Nevertheless, the distinction between the two areas can be quite thin. So to get a better precision, we work in the complex plan and we try to define “upstream” and “downstream” areas with (if necessary) an uncertainty area. These areas will be defined from the simulations made with EMTP under the following conditions: • DG power : 1 – 2 or 5 MVA • Rfault = 1 – 10 – 50 – 100 – 200 – 500 – 1000 – 1500 Ω • Neutral grounding (40 – 400 – 1200 – 2400 Ω) for
aerial lines (length = 25 – 40 – 60 – 80 – 120 km) • Compensating grounding (RN = 40 – 400 – 1200 –
2400 Ω and tuning = 0,8 – 1,2 – 1,4) for cabled lines (length = 8 – 10 – 12 – 15 – 20 km)
TABLE I - Analytics formulae of the I2 / I0 ratio for the detectors A, B & C
Fault upstream Fault downstream
Detectors A & B
Detector C
ZN : grounding impedance X2 : negative sequence impedance of the DG C0 : zero sequence capacitance of the ligne (per km) ltot : total length of the feeders lC : length of the feeder with the DG
Zn
Zn
~
HV network~
Zn
Zn
~~
HV network~~ DG
(A)
(B) (D)
(C)ZchC
Zn
Zn
~
HV network~
Zn
Zn
~~
HV network~~ DG
(A)
(B) (D)
(C)ZchC
NtotO
NtotO
CC
CC
ZljCZljC
jXZjXZ
ωω
3132
2 2
2
++
×++ ( )NtotO
CC
ZljCjXZ
jXω32
2 2
2 +×+
NCO
NtotO
CC
CC
ZljCZljC
jXZZ
ωω
332
2 2
+×
+NCO
NtotO
CC ZljCZljC
jXZjX
ωω
3232
22
2
2
++
×+
CC II RR EE DD 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
CIRED2005 Session No 3
In the figure 2 (compensative grounding) there is no overlap between the “UP” and “DOWN” areas of the detectors A & B, but the areas are very close especially for RN = 40 Ω . Then phasors measurement errors will generate maloperation. For the detector C, the two areas are clearly separated. In the figure 3 (resistive grounding) we can see that an uncertainty exists for detectors A & B because the areas “UP” and “DOWN” overlap. This uncertainty mainly corresponds to low grounding (40 and 400 Ω) whatever the other parameters. As a resistive grounding is simulated with aerial lines the feeder with the detector C is weakly capacitive so a directional relay is useless that is why no results is presented. Fig. 2: I2 / I0 ratio in the complex plan. Compensative grounding. “UP”: the fault is upstream the detector. “DOWN”: the fault is downstream the detector.
Fig.3: I2 / I0 ratio in the complex plan. Resistive grounding. For the detectors A & B an uncertainty zone has been defined. Regarding to the detectors A and B upstream the busbar (fig.3), it is more difficult especially for a low value of RN and a low capacitive current. The more precise is the phase measurement, the easier it will be to make the good decision. Finally it will depend on the current transformers (CT) precision that is respectively ± 0.5°, ± 1°, ± 3° for TPX, TPY and TPZ protection CT. Influence of phasors measurements errors In the table II we indicate the influence of phasors measurement errors on the I2 / I0 algorithm. As an example, we only analyse the behaviour of the detectors A and B for a compensating grounding. The parameters (DG power, Rfault, grounding and length) are those given previously. After calculating a ratio I2 / I0, the error (module: r and argument: θ) is introduced as I2 / I0×r.ejθ. It represents the cumulative error over all the measurement sequence (CT, AC-DC converter and phasor calculator). We simulated 960 cases. The table II shows that the algorithm is really reliable for module errors less than 2% and argument errors in the range [-2° , +1°]. A detailed analysis of the errors shows that they only occur for Rn = 40 Ω. Nevertheless such a low resistance value is seldom used for compensative grounding. Typical values are rather 400 Ω. In this case the algorithm becomes completely reliable for a compensative grounding whatever the errors in the ranges of the table II.
-1 -0.5 0 0.5 1 1.5 2 2.5 3-1
-0.5
0
0.5
-1 -0.5 0 0.5 1 1.5 2 2.5 3-1
-0.5
0
0.5
-1 -0.5 0 0.5 1 1.5 2 2.5 3-1
-0.5
0
0.5
frontier
DOWN
UP
Detectors A & B
-5 -4 -3 -2 -1 0 1-20
-15
-10
-5
0
5
-5 -4 -3 -2 -1 0 1-20
-15
-10
-5
0
5
-5 -4 -3 -2 -1 0 1-20
-15
-10
-5
0
5
frontierDOWN
UP
Detector C
0.6 0.7 0.8 0.9 1 1.1 1.2-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
UP
DOWNuncertainty
frontier
Detectors A & B
0.6 0.7 0.8 0.9 1 1.1 1.2-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.6 0.7 0.8 0.9 1 1.1 1.2-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
UP
DOWNuncertainty
frontier
Detectors A & B
CC II RR EE DD 18th International Conference on Electricity Distribution Turin, 6-9 June 2005
CIRED2005 Session No 3
TABLE II - Influence of the phasors measurement errors on the I2 / I0 algorithm. The error, r.ejθ, represents the cumulative error over all the measurement sequence. Behaviour of the detectors A and B for a compensating grounding. 960 simulated cases. Errors only occur for Rn = 40Ω.
Module error r
-5% -2% -1% 0% 1% 2% 5% Number of errors
Up Down Up Down Up Down Up Down Up Down Up Down Up Down-5° 0 0 0 31 0 55 0 67 0 85 0 93 0 109
-2° 0 0 0 0 0 0 0 0 0 0 0 2 0 46
-1° 0 0 0 0 0 0 0 0 0 0 0 0 0 5
0° 31 0 0 0 0 0 0 0 0 0 0 0 0 0
1° 63 0 7 0 1 0 0 0 0 0 0 0 0 0
2° 90 0 51 0 34 0 12 0 1 0 0 0 0 0 Arg
umen
t er
ror
θ
5° 178 0 127 0 116 0 102 0 97 0 86 0 54 0 TABLE III - Summary of the reliability of the I2 / I0 algorithm for the detectors A, B and C. The grounding is resistive or compensated. The reliability was tested with measurement errors in the range ±5% for the module and ±5° for the argument and with the parameters given in the text.
Resistive grounding Compensative grounding Detectors A & B Detectors A & B Detector C
Algorithm I2 / I0
Failures due to areas overlap for low neutral resistance and low capacitive current
No failure Except for Rn = 40Ω No failure
CONCLUSION In this paper we have presented a local algorithm to suppress voltage sensors in the directional relays for medium voltage networks with distributed generation. This algorithm only uses the symmetrical components of the currents: it calculates the ratio of the negative and zero sequences of the currents at the fundamental frequency: I2 / I0.
In the complex plan, areas can be defined to recognise a fault upstream or downstream the detector. As the algorithm uses the zero sequence, only phase-to-earth faults (which represent the main part of the faults) can be detected. The table III shows that the presented algorithm is better adapted to a compensative grounding network: the I2 / I0 ratio used for the A & B detectors allow a complete reliability. If Rn > 100 Ω, the I2 / I0 ratio is a good solution for all the detectors. Finally the better the measurement precision, the better the reliability. To increase the algorithm’ reliability when the transformers neutral points are resistively grounded, a supplementary algorithm must be found because the I2 / I0 ratio is not reliable enough.
REFERENCES [1] C.L. Fortescue, 1918, Proc. of the American IEE, 37, pp
629-716 [2] P. Anderson, Analysis of faulted Power Systems, Wiley-
IEEE Press, June 1995. [3] X. Le Pivert, P. Bastard and I. Gal, ” How symmetrical
components may help to suppress voltage sensors in directional relays for distribution networks” in Proceedings 17th CIRED, 2003.
[4] M. Petit, X. Le Pivert, P. Bastard and I. Gal,
“Symmetrical components to suppress voltage sensors in directional relays for distribution networks: effect of distributed generation” in Proceeding 8th Int. Conf. on Developments in Power Systems Protection, 2004, pp 575-578
