9. KONFERENCA SLOVENSKIH ELEKTROENERGETIKOV – Kranjska Gora 2009
CIGRÉ ŠK B5-8
Optimal earth fault protection for compensated MV
distribution networks
Martin Čelko
ABB Oy, Distribution Automation – Finland
E-mail: martin.celko@fi.abb.com
Janne Altonen
ABB Oy, Distribution Automation – Finland
E-mail: janne.altonen@fi.abb.com
Ari Wahlroos
ABB Oy, Distribution Automation – Finland
E-mail: ari.wahlroos@fi.abb.com
Olli Rintamäki
ABB Oy, Distribution Automation – Finland
E-mail: olli.rintamaki@fi.abb.com
Abstract – This paper deals with earth fault protection for compensated MV distribution networks. First, we review
main characteristics and purpose of the compensated MV distribution network. Next, we describe the known
methods for detecting earth faults in compensated distribution MV network. We conclude and describe what could
be the optimal earth fault protection for compensated MV distribution networks.
I. INTRODUCTION
This paper focuses on earth fault (EF) protection in
medium voltage distribution networks with a
compensated neutral point.
The way the neutral point is earthed is a basic
problem which greatly affects the number of technical
and also economic solutions of power systems [1].
The earthing of the distribution network via the coil is
one of the most popular methods over the last years in
electrical network design by which better power
supply quality and reliability can be achieved. The
main advantage of this earthing arrangement is the
self-extinguishing of most of the arcing faults and the
possibility of continuing the network operation during
a sustained earth fault. Consequently the number of
autoreclosings, interruptions and outages of the power
supply for the customer is significantly reduced, and
thus running expenses for the utility can be brought
down.
These networks are also known as resonant earthed,
or according to the inventor, as Petersen coil earthed
networks. The arc suppression coil was invented by
W. Petersen in 1916 as the result of his pioneering
work in investigating earth fault phenomena [2].
II. BASICS OF COMPENSATED NETWORKS
The idea of earth fault current compensation is to
cancel the network earth capacitance by an equal
inductance connected to the neutral with a
corresponding decrease in the EF current as shown in
Fig. 1.
The main benefit of earth fault current compensation
is that most of the single phase-to-earth faults are
cleared by themselves. This, in addition to the small
fault currents, is due to the fact that when the arc is
extinguished, the compensation coil continues to
produce a voltage equal to the phase to neutral
voltage, so that the voltage in the fault spot (across the
arc) rises slowly and restriking is unlikely.
Fig. 1. EF in a compensated network.
9. KONFERENCA SLOVENSKIH ELEKTROENERGETIKOV – Kranjska Gora 2009
CIGRÉ ŠK B5-8
In compensated networks the phase-to-earth
voltages of the two sound phases are, in case of zero
fault resistance, equal to phase-to-phase voltage as
represented in Fig. 2. This must be taken into account
when selecting the insulation of the equipment in
order to minimize the risk of double phase-to-earth
faults.
I ef
U0 =
(
1 2
1 2
) + (3wC0 −
)
wL
R0
In the case of complete compensation, the neutral
voltage can be simplified as follows:
U0
R0
=
E
R0 + R f
Fig. 2. Phasor diagrams of the currents and voltages during a
single phase-to-earth fault in a compensated network.
(3)
(4)
In the above equations it was assumed that no
additional neutral resistor RL is used. If needed, the
effect of RL can be taken into account by replacing R0
in the above equations by the parallel connection of
RL and R0.
III. METHODS FOR DETECTING EARTH FAULTS IN
COMPENSATED NETWORKS
Fig. 3. Equivalent circuit for the EF in a compensated network.
The equivalent circuit for the EF in a compensated
network is shown in Fig. 3. The circuit is a parallel
resonance circuit, and if the compensation coil is
tuned exactly to the system capacitance, the fault
current has only a resistive component. This resistive
current is due to the resistances of the coil and
distribution lines together with the system leakage
resistances R0. Often the earthing equipment is
complemented with a parallel resistor RL, the task of
which is to increase the earth fault current in order to
make selective relay protection possible.
Using the equivalent circuit of Fig. 3, we can write
the following for the earth fault current:
I ef =
E 1 + R02 (wC0 −
1 2
)
wL
( R f + R0 ) 2 + R 2f R02 (wC0 −
(1)
In compensated networks directional EF protection
is commonly applied together with residual
overvoltage protection as a back-up.
The directional EF protection measures the residual
current I0 and the neutral voltage U0. The selectivity
of the protection is based either on measurement of
the angle ϕ between U0 and I0 or on the active
component I0cosϕ. Often the magnitude of this
component is very small and needs to be increased by
means of a resistor connected in parallel to the
compensation coil. Typical characteristic of the
“phase angle measurement” and the “I0cosϕ
measurement” is shown in Fig. 4.
The tripping in case of “phase angle measurement”
characteristic is permitted if the residual current I0
exceeds the setting and the neutral voltage U0 is above
the setting and the phase angle between I0 and U0 is in
the range ϕ0+/-∆ϕ, where ϕ0=0° and ∆ϕ=+/-80° (or
+/-88°).
In case the “I0cosϕ measurement” characteristic is
used the tripping is initiated if both the active
component I0cosϕ and the neutral voltage U0 exceed
the setting value.
1 2
)
wL
In the case of complete compensation, the above can
be simplified as follows:
I ef =
E
R f + R0
The neutral voltage
correspondingly as follows:
(2)
can
be
calculated
Fig. 4. Relay characteristics of the directional EF protection in
compensated networks where a) is phase angle measurement and b)
is I0cosϕ measurement.
In completely compensated networks the residual
current is usually mostly resistive (Fig. 1, ΣI02) and
9. KONFERENCA SLOVENSKIH ELEKTROENERGETIKOV – Kranjska Gora 2009
CIGRÉ ŠK B5-8
Fig. 5. Extended relay characteristic of directional EF protection
used in compensated networks.
IV. INTERMITTENT EF PROTECTION
In compensated networks, especially in those with
underground cables, a special type of fault is
encountered – the intermittent EF. This kind of fault
tends to be difficult for conventional directional EF
(DEF) relays to detect due to highly irregular wave
shape of residual current [3]. Whereas residual
overvoltage (ROV) relays used typically as a
substation back-up protection have better chances for
fault detection because of more steady behavior of
residual voltage. Due to this fact intermittent EF can
often cause non-selective tripping of the substation.
Intermittent EF can be characterized as a series of
cable insulation breakdowns because of reduced
voltage withstand of it. As shown in Fig. 6 the fault is
initiated as the phase-to-earth voltage exceeds the
reduced insulation level of the fault point and
extinguishes mostly itself as soon as the fault current
crosses zero for the first time.
0.4
residual current (kA)
0.3
Current (kA)
Voltage x 102 (kV)
the sensitivity of the relay located in the faulty feeder
is about the same for both types of characteristics.
This is not the case for the relay in the sound feeder
where the residual current is almost all capacitive
(Fig. 1, ΣI01) and its phase angle is very close to the
operation area of the relay characteristic. If there is an
excessive phase displacement error in the instrument
transformer used for measuring the residual current
then a nuisance tripping may happen. In the “I0cosϕ
measurement” principle the security margin can be set
higher and the risk of malfunction is somewhat
smaller than in the “phase angle measurement”
principle.
In compensated networks the protection relay must
also be able to operate selectively in case the
compensation coil is temporarily out of operation and
the network is operated as isolated. Traditionally, the
relay characteristic is changed by an external control
command received from the auxiliary switch of
compensation coil disconnector. In modern relays this
command can be received via the communication. In
Fig. 5 another solution is shown, where the relay
characteristic is wide enough to cover both types of
network. The application of this so called extended
operating sector makes the changing of the
characteristic unnecessary.
0.2
residual voltage x 102 (kV)
0.1
0
-0.1
5
10
cycles
-0.2
-0.3
-0.4
Fig. 6. Typical waveforms of residual current and voltage
measured from a substation during an intermittent EF.
The electrical characteristics of the intermittent fault
can be described by forming a simplified equivalent
circuit of the network involved and by modeling the
fault point as a nonlinear resistor and ‘a spark gap’
with certain type of characteristic current-voltage
curve as shown in Fig. 7.
When an EF occurs, for example, in phase A, the
phase-to-earth voltage becomes shorted by the fault.
The energy stored in the phase-to-earth capacitance of
the faulty phase discharges and this discharge current
transient can be measured in the faulty phase of any
feeder of the substation (e.g. IvA_DISCHARGE, Fig. 7).
It can be further concluded that because the healthy
phase-to-earth voltages increase during the fault, the
phase-to-earth capacitances of the healthy phases are
initially charged by a transient called the charge
current transient. The total charge current transient of
the whole network flows in the faulty phase of the
faulty feeder (IB,C_CHARGE, Fig. 7). The healthy phases
of the healthy feeders experience only a part of the
total charge current being proportional to the
respective portion of the total phase-to-earth
capacitance (IvB_CHARGE and IvC_CHARGE, Fig. 7).
9. KONFERENCA SLOVENSKIH ELEKTROENERGETIKOV – Kranjska Gora 2009
CIGRÉ ŠK B5-8
I0v
I0 j
U0
counter_pos
max_count_j
spike detection_j
0
counter_neg
drop-off time_j
0
operate time delay_j
0
start signal_j
0
spike detection_v
max_count_v
Fig. 8. Filtered and sampled residual current and voltage signals
(top) in relation to the principal operation of the spike detection
method (bottom). Indices _j and _v stand for faulty and healthy
feeders respectively.
Fig. 7. Simplified equivalent circuit model of the network during
an intermittent EF. Indices _j and _v stand for faulty and healthy
feeders respectively.
The residual current transient that can be measured
in the beginning of the healthy feeders (I0v, Fig. 7) is a
sum of charge and discharge current transients
generated in each feeder (IvA_DISCHARGE, IvB_CHARGE and
IvC_CHARGE, Fig. 7). Whereas the residual current
transient that can be measured in the beginning of the
faulty feeder (I0j, Fig. 7) is a sum of charge and
discharge current transients of the healthy feeders
and the coil transient (IvA_DISCHARGE, IvB_CHARGE,
IvC_CHARGE and IL, Fig. 7). Moreover, it should be
noted that the polarities of I0j and I0v are opposite to
each other.
The frequency of the charge transient is typically
between 200 Hz and 1000 Hz. The frequency of the
discharge transient is generally much higher, practical
values being 4-20 times the frequency of the charge
transient. Because of the high frequency the
impedance of the compensation coil is also high,
which means that the transient of the EF current is
unaffected by the reactance magnitude of the
compensation coil (i.e. the degree of compensation).
Conventional DEF and ROV relays have been
designed to operate with more or less steady–state
fundamental frequency current and voltage sinewaves. It is therefore evident that problems arise
when the current and voltage waveforms are highly
irregular and a dedicated functionality against
intermittent EF is needed. Bearing this in mind
algorithms have been developed and implemented in
modern feeder terminals. The basic detection methods
are typically based on spike detection with certain
polarity and phase angle criterion methods.
Fig. 8 shows typical residual current waveforms
measured from both faulty (blue) and healthy (red)
feeder where the operation of the protection is based
on spike detection method.
Fig. 9 shows the operation of the protection based
on phase angle criterion method where the phase
angle difference between the residual current and
voltage phasors measured from both faulty (blue) and
healthy (red) feeder is shown. The protection starts if
the calculated phase angle difference is inside the
defined operating sector and in the same time the
residual current and voltage amplitudes are above the
set values.
120
60
0
-60
-120
jv
jj
extended
operating sector
drop-off time_j
0
operate time delay_j
0
start signal_j (internal)
0
start signal_j
0
Fig. 9. Phase angle difference between filtered and sampled
residual current and voltage signals (top) in relation to the principal
operation of the phase angle criterion method (bottom). Indices _j
and _v stand for faulty and healthy feeders respectively.
Variety of field tests has been performed in practical
utility networks. The general result from the tests was
that in the faulty feeder both detection methods gave
out the trip signals correctly. A few times in a healthy
feeder the DEF function and a couple of times the
intermittent earth fault function applying phase angle
criterion gave out false starting signal, but did not trip,
especially when the network was heavily under or
overcompensated and the feeder in question had a
high capacitive current contribution. It should be
noted that these algorithms are subject to constant
development work to further enhance the
performance.
9. KONFERENCA SLOVENSKIH ELEKTROENERGETIKOV – Kranjska Gora 2009
CIGRÉ ŠK B5-8
V. CONCLUSIONS
The basic ideas, features and advantages of
compensated MV distribution networks have been
described. The main focus has been on the directional
earth fault protection methods. Special attention has
been put to the challenge of protection relaying in
compensated networks with underground cables – to
the intermittent earth fault.
The directional earth fault protection is usually
based either on the “I0cosϕ measurement” principle,
or on the “phase angle measurement” principle. It is
typical to apply multiple stages, i.e. low-set and highset stages, which make it possible that the operating
speed and sensitivity requirements set on earth fault
protection by the authority can be optimally fulfilled.
For example, the low-set stage can be applied only for
alarming and set to detect as high resistive faults as
possible, whereas one or two high-set stages are used
for tripping with different operate time delays. This
also enables that the autoreclosure shots and
sequences can be different depending on the fault
current magnitude.
In compensated networks the directional earth fault
protection relay must be able to work selectively in
case the compensation coil is temporarily out of
operation. The relay characteristic shall be able to be
changed according to the actual network earthing
type. In modern relays this information or command
can be received via the communication.
Finally it could be concluded that the optimal earth
fault protection for compensated MV distribution
networks, especially for those with underground
cables, shall have multiple stage directional earth fault
protection completed with the dedicated protection
against intermittent earth faults.
References
[1]
[2]
[3]
M. Lehtonen, T. Hakola “Neutral Earthing and Power
System Protection“ ABB Transmit Oy, Relays and Network
Control, 1996
E. Bjerkan, T. Venseth, “Locating Earth-Faults in
Compensated Distribution Networks by means of Fault
Indicators”, International Conference on Power Systems
Transients, Canada, 2005
J. Altonen, O. Mäkinen, K. Kauhaniemi, K. Persson
“Intermittent earth faults – need to improve the existing
feeder earth fault protection schemes?“ CIRED 2003
Barcelona