HV SHUNT REACTOR SECRETS

FOR PROTECTION ENGINEERS

By
Zoran Gajić
ABB Sweden
Västerås, Sweden

Birger Hillström
ABB Sweden
Västerås, Sweden

Fahrudin Mekić
ABB Inc.
Allentown, PA 18106

Presented to:
th
30 Western Protective Relaying Conference
Spokane, Washington
October 21-23, 2003
 HV SHUNT REACTOR SECRETS
FOR PROTECTION ENGINEERS

Z. Gajić, B. Hillström F. Mekić

ABB Sweden ABB Inc.
Västerås, Sweden Allentown, PA 18106

Abstract: Viewed in the substation yard, an HV, oil immersed, shunt reactor does not differ
much from a power transformer, but in reality it is not that simple. There are distinct differences
between construction and operating characteristics of these two devices.

In order to explain the properties of shunt reactors numerous current and voltage waveforms
either captured as disturbance recordings in the field or simulated by ATP [7] will be presented.
On all these figures the nomenclature for current and voltage signals, as shown in Figure 1
below, will always be used.

C
IA
IB 3Io
IC

UA UB UC

IaN IbN IcN

IN

Figure 1: Shunt Reactor Current & Voltage Signals

All presented current and voltage signals will be expressed in per unit system with shunt reactor
rated data as a base.
1
I. BASIS ABOUT HV SHUNT REACTORS
1.1 INTRODUCTION
Shunt reactors are designed for connection to the ends of high voltage transmission lines or to
high-voltage cables for the purpose of controlling the line voltage by absorbing reactive power.
Let us look at the equivalent circuit of the transmission line and see shunt reactor effect on the
line parameters.
Is Z Ir

Vs Vr
Y/2 Y/2

Figure 2: Transmission line equivalent circuit (uncompensated line)

In Figure 2, Vs and Is are the sending-end voltage and current, and Vr and Ir are the receiving-
end voltage and current.
Writing Kirchoff Voltage Law equation for the circuit in Figure2,

Vr * Y
Vs = Vr + Z ( Ir + ) (1)
2
Y *Z
Vs = (1 + ) * Vr + Z * Ir (2)
2
Also writing a Kirchoff Current Law equation at the sending end,

Vr * Y Vs * Y
Is = Ir + + (3)
2 2
Y *Z Y *Z
Is = Y (1 + ) * Vr + (1 + ) * Ir (4)
4 2

Vs   AB  Vr 
 Is  = CD   Ir  (5)
    

where:

2
 Y *Z
A = D = 1+ ( perunit ) (6)
2
B = Z (Ω ) (7)
Y *Z
C = Y (1 + )( S ) (8)
4

Example 1:

A three-phase line, completely transposed 345kV, 124 miles has the following positive sequence
constants:

z = 0.0515 + j 0.563(Ω / mile)

y = j 6.76 *10 −6 ( S / mile)

From equations (6), (7) and (8)

A = D = 0.9706∠0.159 0
B = 70.29∠84.78 0
C = 8.277 *10 − 4 ∠90.08 0

From (5), the no-load receiving-end voltage is

Vs 345.8
VrNL = = = 356.3kVLL
A 0.9706

Figure 3 summarizes these results, showing a high receiving-end voltage at no-load and a low
receiving-end voltage at full load. This voltage regulation problem becomes more severe as the
line length increase.

3
 V(x)
V RNL
No-load

SIL
Vs V RSIL VS

Full-load

Short-circuit
V RFL

V RSC 0
Sending end Receiving end

Figure 3: Voltage profiles of an uncompensated line

Assume that identical shunt reactors are connected from each phase to neutral at both ends of the
same line during light load conditions, providing 75% compensation (the reactors are removed
during heavy load conditions).

In this case the line constants are:

Z = z ∗ l = 70.29∠84.78(Ω)
Y = 8.4 * 10 −4 (1 − 0.75) = 2.1 *10 − 4 ∠90 0 ( S )

From equations (6), (7) and (8)

A = D = 0.993∠0.04 0

From (5), the no-load receiving-end voltage is

Vs 345.8
VrNL = = = 348.2kVLL
A 0.993
It may be concluded from the previous example that reactors reduce overvoltages during light
load conditions. However, shunt reactors can reduce line loadability if they are not removed
under full-load conditions.
There are two general types of shunt reactors. One is dry-type reactor of an air core or core-less
design. These reactors are limited to voltages up to 34.5kV and are often installed on the tertiary
of a transformer.
4
HV, oil immersed, shunt reactors are the most compact and cost-efficient means to compensate
reactive power generation of long-distance, high-voltage power transmission lines, or extended
cable systems during light load conditions.

Two main application of the reactor can be identified, referring to Figure 4.:
• Shunt reactors that are continuously in service, generally used for EHV and long HV
lines/cables
• Switched shunt reactors are applied in the underlying system and near load centers

It is common for shunt reactors to be installed at both ends of EHV lines, and sized to prevent the
line voltage from exceeding design value when energized from one end. Since there is usually
some uncertainty as to which end of a line may be energized (or de-energized) first, shunt
reactors are usually installed at both ends of line.

Z
G 52 52 G
52 52

Equivalent Pi of the
Long Line
Y/2 Y/2

Figure 4: One-line diagram of line-connected switched shunt reactors

The shunt capacitance depends on type of transmission line, length of line and line voltage. A
long distance 345kV transmission line will have a shunt capacitance around 3.14 µF /mile
(1.12Mvar/mile). The corresponding shunt capacitance for the 345kV cable is almost 20 times as
large as or about 22.4Mvar/mile. The shunt capacitance will be increased by increasing the
transmission voltage (proportional to the square of the transmission voltage).

5
1.2 SHUNT REACTOR GENERAL DESIGN CONCEPTS

Two different ways are used in building reactors, commonly referred to as “gapped core” and
“coreless” [1] & [2]. The gapped core reactor has a subdivided limb of core steel with air gaps
inside the winding – and no limb at all for the coreless concept.
It is easy to verify that the gapped core concept becomes more advantageous as the loss
evaluation rate increases and particularly at higher system voltages. This is due to the higher
energy density that can be achieved in a gapped core reactor compared to a coreless reactor.

SHUNT REACTOR OPERATING CHARACTERISTICS

Linearity
For normal operating voltages there is a linear relationship between applied voltage and reactor
current (i.e. a small increase in voltage will result in a proportional increase in current). Magnetic
fluxes and flux densities are also proportional to the time integral of the applied voltage. With a
voltage of sinusoidal shape the fluxes and flux densities are also proportional to the voltage. The
deviation from a true sinusoidal shape in line voltage is in general negligible for normal
operating voltages.
As the magnetic flux to a great extent has its path in magnetic core steel the core steel will get
saturated for flux densities above a certain level, the saturation point. Below and up to the
saturation point only a small current is needed to magnetize the core steel and the extra current
needed to reach a marginal increase flux density is small. Once above the saturation point the
extra current needed to further increase the flux density will be large.

Harmonic content
Steady state harmonics in reactor current arise from partial saturation in the magnetic circuit.
These effects are in fact very small, and without practical importance for relaying and
communication interference. Of all harmonics the third harmonic will be dominant. In the reactor
neutral the third harmonics in the three phases add together and act like a zero sequence current.

Asymmetry between phases

The tolerances on asymmetry between phases of a three-phase reactor or between single-phase
units forming a three-phase bank can be judged by the amount of residual harmonics. The result
is a zero sequence current in the neutral connection. Standards are realistic, but better tolerances
are possible to achieve. A usual figure is 0.5 %.

6
 II. HV SHUNT REACTOR SWITCHING
2.1 SWITCHING IN OF REACTORS, INRUSH CURRENT

The switching in of a reactor gives rise to inrush current – a transient phenomenon related to
saturation in the shunt reactor magnetic circuit. In principle, it is the same story as inrush current
of a transformer, but there are differences. A reactor core keeps no remanence, because of the air
gaps, which makes the whole thing easier. However, the damping of the asymmetric condition –
“the dc component” – is slow, due to the inherent low losses in a shunt reactor. It is therefore
necessary to keep this phenomenon in mind when designing the relay protection system for HV
shunt reactors.
The instantaneous current values during shunt reactor switching in can be visualized from the
Figure 5. Input data for all figures in this chapter are obtained from actual disturbance recordings
in the field. Depending on the switching instant the currents might have a dc component. The
worst condition is when the reactor phase is closed in at zero voltage. The flux will increase with
the voltage-time-area during the first half-cycle to a value twice the maximum flux in normal
operation. The current is proportional to the flux density, until reactor core saturation occurs.
Above the point of saturation the current will increase faster than the flux.

99,2MVA, 440kV, 60Hz Reactor

2
Current [pu]

17 18 19 20 21 22 23 24 25 26 27 28 29 30
Cycles
IA
IB
IC

Figure 5: Shunt Reactor Phase Currents during Asynchronized Switching

Without saturation, the first peak of the current with full dc offset would be 2 ⋅ 2 = 2.82 times
rated current. The actual current peak might rise to a value in between 3 and 5.5 times depending
on the particular shunt reactor design details. One of the time intervals when reactor core goes
into saturation is clearly marked if Figure 5.
For a three-phase reactor the different phases will experience different degrees of dc offset. The
combination of the individual phase current offsets will give a neutral current rich in harmonics
and also with possibly dc offset from the zero line as shown in Figure 6 or Figure 14.

7
 99,2MVA, 440kV, 60Hz Reactor
1.75

0.92
Current [pu]

0.0833

0.75
17 18 19 20 21 22 23 24 25 26 27 28 29 30
Cycles
IN
3Io

Figure 6: Shunt Reactor Neutral Currents during Asynchronized Switching

The time to more or less fully balanced operation around zero flux in the core may be fairly long
often in order of seconds, but such condition is of no harm for the shunt reactor itself. In recent
years, so-called point on wave closing relays are available from switchgear manufacturers. By
using these relays switching of different power system devices, including shunt reactors, can be
performed without a disturbance to the rest of the power system. Typical current waveforms
during the synchronized shunt reactor switching is shown in Figures 7, 8 & 9.
150MVAr, 220kV, 50Hz Reactor
2

1
Current [pu]

2
4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10
Cycles
IA
IB
IC

Figure 7: Shunt Reactor Phase Currents during Synchronized Switching

8
 150MVAr, 220kV, 50Hz Reactor
1.25

0.75
Current [pu]

0.25

0.25
4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10
Cycles
IN
3Io

Figure 8: Shunt Reactor Neutral Currents during Synchronized Switching

In order to obtain such disturbance free shunt reactor switching, circuit breaker poles must be
precisely closed in three consecutive phase voltage peaks as shown in the following figure.

150MVAr, 220kV, 50Hz Reactor

2

1
Current & Voltage [pu]

2
5 5.22 5.44 5.67 5.89 6.11 6.33 6.56 6.78 7
Cycles
IA
IB
IC
UA
UB
UC

Figure 9: Shunt Reactor Phase Currents & Voltages during Synchronized Switching

9
2.2 SHUNT REACTOR DISCONNECTION

Disconnection of small reactive current was at one time regarded as a dangerous operation
because of the risk of current chopping and resulting switching overvoltage. Modern surge
arresters are fully capable of handling this condition, and besides, the tendency of the circuit
breaker to chop reactor current is not so pronounced for typical HV shunt reactor rated current
values [1], [3].

However, the primary current chopping causes another, and maybe less known, transient
phenomenon, which appears in the CT secondary circuit. This phenomenon is manifested as an
exponentially decaying dc current component in the CT secondary circuit (see Figure 16, for
typical example). This secondary dc current has no corresponding primary current in the power
system. The phenomenon can be simply explained as a discharge of the magnetic energy stored
in the magnetic core of the current transformer. However these discharge secondary currents are
typically very small for shunt reactors and pose no effect on the reactor protection schemes with
numerical relays.

10
 III. NUMERICAL PROTECTION RELAY RESPONSE
DURING SHUNT REACTOR SWITCHING IN

All numerical relays utilize so-called sampling technique of the input current and voltage signals.
Typically 12 to 32 samples per fundamental power system cycle are used depending on the
particular relay design. From these samples numerical relays calculates root-mean-square values
of the input quantities by using different type of digital filters. These RMS values are then
typically processed by different protective functions (i.e. phase and ground overcurrent)
In order to apply correct relay settings for shunt reactor protection application, it is of outmost
importance to understand the relay digital filter response to typical input current waveforms,
which can be encountered.

Figure 10 represents typical current waveform during reactor switching in.

99,2MVA, 440kV, 60Hz Reactor

5

3
Current [pu]

2
20 30 40 50 60 70 80
Cycles
IC

Figure 10: Typical Shunt Reactor current waveform during switching in

Response of two different types of digital filters will be investigated.

1. TRMS (i.e. True RMS filter), which extracts equivalent RMS value from the input signal.
Therefore this filter includes the dc component and higher harmonic components from
the input quantity into its output result
2. DFT (i.e. digital Fourier filter), which extracts only RMS value of the fundamental
component from the input signal. This filter effectively suppresses the dc component and
higher harmonic components in the input quantity.

11
From Figure 11 it is obvious that the overcurrent relays which use DFT filtering technique can be
set more sensitive than the relays which use TRMS filter for its operation. Similar results can be
obtained if similar analysis is performed for the neutral point current as well. All setting
recommendation in this document will be given for relays, which utilize DFT filtering technique
(i.e. relays which effectively suppress the dc component and higher harmonic components in the
input quantity).

IC RMS value
2.25

1.69
[pu]

1.13

0.56

0
20 30 40 50 60 70 80
Cycles
TRMS
DFT

Figure 11: Digital Filter Output for input signal shown in Figure 10

12
 IV. POSSIBLE PROTECTION PROBLEMS
DURING SHUNT REACTOR SWITCHING
It is well known fact that one of the principal difficulties with shunt reactor protection scheme is
false operation during reactor energizing and de-energizing [4]. As explained previously, during
this period relatively high and long lasting dc current component typically causes most problems
for protective relays. If the protection relays maloperate this typically happen some hundreds of
millisecond or even 1 to 2 seconds after circuit breaker closing. What is most difficult to
understand is why this problem often happens randomly and not with every reactor switching
attempt. Most problems are typically encountered with restricted ground fault protection,
differential protection and ground fault protection during switching. Therefore performance of
these three relays during switching in of the shunt reactor will be explained here in more details.

4.1 CURRENT TRANSFORMER PERFORMANCE DURING SWITCHING IN OF SHUNT REACTOR

It should be noted that HV shunt reactors are typically switched in and out at least once per day
or even more often depending on the power system loading patterns. As shown in Chapter II
during switching in of shunt reactor relatively high and long lasting dc current component might
appear in one or more phases. This current waveform moves the operating point of CT magnetic
core on the hysteresis curve in one direction and when the dc component diminish it leaves the
main CT with certain level of residual (i.e. remanent) flux. During normal operation reactor
current is always around 1pu and therefore of a relatively low magnitude, which is never big
enough to move the operating point towards the origin. Therefore when next switching attempt
comes, depending on the moment of switching, residual flux in the CT core can increase or
decrease. Thus this mechanism will sooner or later cause CT saturation during reactor switch in
operation. This CT saturation then causes problems for protective relays, which lose the correct
information about the primary current and therefore can maloperate. Such CT saturation event is
captured by numerical relay disturbance recorder and it is shown in Figure 12.
150MVAr, 220kV, 50Hz Reactor
4

CT Saturation
3
Instant

2
Current [pu]

1
0 5 10 15 20 25 30 35
Cycles
IC

Figure 12: Phase CT saturation during Shunt Reactor Switching in

13
 150MVAr, 220kV, 50Hz Reactor
1.5

1
Current [pu]

0.5

0
0 5 10 15 20 25 30 35 40 45 50
Cycles
DFT IC

Figure 13: Influence of Phase CT saturation on current DFT value calculation

This type of CT saturation is reflected in the CT secondary side as:

• loss of information about primary dc component
• reduced current magnitude

Figure 13 represents the DFT filter output value for the input current waveform as shown in
Figure 12.

4.2 RESTRICTED GROUND FAULT RELAY PERFORMANCE DURING SWITCHING IN OF REACTOR

Modern numerical relays typically offer restricted ground fault protection of a low impedance
type. This gives the following benefits to the end user:
• this relay can be applied with different type of CTs at the reactor bushing and at reactor
neutral point (i.e. CTs doesn’t need to be identical)
• main CTs can be shared with other relays
• no galvanic connection is necessary between CTs at the reactor bushing and at reactor
neutral point
• in case of an internal fault no high voltages will appear in the CT secondary wiring

Typically these restricted ground fault relays of a low impedance type calculate the differential
current as a difference between zero-sequence currents at the reactor bushings and the reactor
neutral point. As additional operating criteria they often use directional principle (i.e. product
type relays). However for shunt reactor protection these sometimes might not be enough to
prevent maloperations. Let’s have a look into the disturbance-recording file captured by
numerical relay, which is shown in Figure 14.

14
 150MVAr, 220kV, 50Hz Reactor
2

1
Current [pu]

1
0 5 10 15 20 25 30 35
Cycles
IN
3Io

Figure 14: Zero-sequence currents during Shunt Reactor Switching in

The problem is that when one or more phase CTs saturate false 3Io current appears at the reactor
bushings. Unfortunately this very often manifests as the current of opposite polarity in
comparison with the neutral point current, which then causes the directional restricted ground
fault relay (i.e. product type relay) to maloperate during reactor switching in.
Calculated phase angle difference between neutral point current and zero-sequence current at the
reactor bushing for the above event is shown in Figure 15:

Phase Angle Between IN & 3Io

100

80

60
Angle [deg]

40

20

20
5 10 15 20 25 30 35
Cycles
3Io-IN

Figure 15: Calculated phase angle difference between IN and 3Io currents

Obviously it is necessary to have some additional means to restrain low impedance, restricted
ground fault relay from maloperations during shunt reactor switching in. One very effective
method is to check the amount of second harmonic component in the shunt reactor neutral point
current and adaptively prevent relay operation if the preset limit is exceeded.
15
4.3 DIFFERENTIAL RELAY PERFORMANCE DURING SWITCHING IN OF REACTOR

Modern numerical relays typically offer differential protection of a low impedance type. This
gives the following benefits to the end user:
• this relay can be applied with different type of CTs at the reactor bushing and at reactor
star point (i.e. CTs doesn’t need to be identical)
• main CTs can be shared with other relays
• no galvanic connection is necessary between CTs at the reactor bushing and at reactor
star point
• in case of an internal fault no high voltages will appear in the CT secondary wiring

Here the situation is little bit easier because the relay measures essentially the same current on
both ends of the protected winding. However again the long lasting dc component can cause
uneven saturation of the two CTs and cause the relay maloperations. Let’s have a look into the
disturbance recording file captured by numerical relay, which is shown in Figure 16.

99.2MVA, 440kV, 60Hz Reactor

4

2
[pu]

2
0 2.08 4.17 6.25 8.33 10.42 12.5 14.58 16.67 18.75 20.83 22.92 25
Cycles
IC
IcN

Figure 16: Phase C winding currents during shunt reactor switching in

As it can be seen in Figure 16 due to uneven CT saturation on the two winding ends differential
protection had unwanted operation and it has disconnected the shunt reactor from the power
system. Thus if sensitive setting is required for the differential protection (i.e. 10-15% of the
reactor rated current) it might be necessary to have some additional means to restrain low
impedance differential protection relay from maloperations during shunt reactor switching in.
One effective method is to enable second harmonic blocking feature commonly readily available
in numerical transformer differential relay. Second possibility is to delay the restraint differential
protection operation only during reactor switching. In the same time in order to have secure
operation for heavier internal fault, the unrestrained differential level can be typically set down
to 200% and without any time delay.

16
4.4 GROUND OVERCURRENT RELAY PERFORMANCE DURING SWITCHING IN OF REACTOR

Numerical ground overcurrent relay might maloperate during reactor switching if it is set too
sensitive. Typically in such cases either pickup current value or time delay are increased.
However, another very effective method for such type of problem is to enable second harmonic
blocking feature for ground overcurrent relay, which is readily available in certain numerical
protections. Then the relay will check the second harmonic component level in the measured
input current and prevent relay operation if the preset limit is exceeded.

V. SHUNT REACTOR BEHAVIOUR DURING

EXTERNAL AND INTERNAL FAULTS
Shunt reactors are connected in parallel with the rest of the power network. As shown in
Appendix II shunt reactor can be treated as a device with the fixed impedance value. Therefore
the individual phase current is directly proportional to the applied phase voltage (i.e. I=U/Z).
Thus during external fault condition, when the faulty phase voltage is lower than the rated
voltage , the current in the faulty phase will actually reduce its value from the rated value.
Depending on the point on the voltage wave when external fault happens the reduce current
might have superimposed dc component. Such behavior is verified by an ATP simulation and it
is shown in Figure 17.
150MVA, 220kV, 50Hz Reactor
1.6

1.08

0.57
Current [pu]

0.47

0.98

1.5
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Cycles
IA
IB
IC

Figure 17: External Phase A to Ground Fault, Reactor Phase Currents

As a result, shunt reactor unbalance current will appear in the neutral point as shown in Figure
18. However, this neutral point current will typically be less than 1 pu irrespective of the location
and fault resistance of the external fault.

17
 150MVA, 220kV, 50Hz Reactor
2.5

1.17
Current [pu]

0.17

1.5
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Cycles
IN
3Io

Figure 18: External Phase A to Ground Fault, Reactor Zero-sequence Currents

Similarly during an internal fault the value of the individual phase currents and neutral point
current will depend very much on the position of the internal fault. Assuming that due to the
construction details, internal shunt reactor phase-to-phase faults are not very likely, only two
extreme cases of internal phase to ground fault scenarios will be presented here.

In the first case the Phase A winding to ground fault, 1% from the neutral point has been
simulated in ATP. As a result the phase currents on the HV side (i.e. in reactor bushings) will be
practically the same as before the fault as shown in Figure 19.

150MVA, 220kV, 50Hz Reactor

1.6

1.08

0.57
Current [pu]

0.47

0.98

1.5
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Cycles
IA
IB
IC

Figure 19: Internal Phase A Winding to Ground Fault, Phase Currents

18
However phase A current at the shunt reactor star point and common neutral point current will
have very big value due to so-called transformer effect. These currents can be so high to even
cause CT saturation as shown in Figure 20 for the common neutral point current.

150MVA, 220kV, 50Hz Reactor

15

8.33
Current [pu]

1.67

5
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Cycles
IN
3Io

Figure 20: Internal Phase A Winding to Ground Fault, Zero-sequence Currents

This type of the internal fault shall be easily detected and cleared by the differential, restricted
ground fault or neutral point ground overcurrent protection, but not by reactor HV side
overcurrent or HV residual ground fault protections.

In the second case the Phase A to ground fault, just between the HV CTs and shunt reactor
winding (i.e. shunt reactor bushing failure) has been investigated. In this case the currents have
opposite properties. The phase A current on the HV side is very big (limited only by the power
system source impedance and fault resistance), while the phase A current in reactor star point
will have very small value due to a fact that phase A winding is practically short-circuited.
As a result, shunt reactor unbalance current will appear in the neutral point. However, this
neutral point current will typically have a value around 1 pu (i.e. similar value as during external
ground fault).

That type of the internal fault (i.e. shunt reactor bushing failure) shall be easily detected and
cleared by the differential, restricted ground fault or HV side overcurrent or residual ground fault
protections. Neutral point ground overcurrent protection can operate with the time delay.

For internal ground fault in some other location in-between these two positions the shunt reactor
currents will have values somewhere in the range limited by this two extreme cases.

19
 VI. SHUNT REACTOR TURN-TO-TURN PROTECTION SCHEMES

Turn-to-turn faults in shunt reactor present a formidable challenge to the protection engineer.
The current and the voltage changes encountered during such fault are very small and therefore
sensitive and reliable protection against turn-to-turn faults is difficult to achieve. At the same
time the longitudinal differential protection offers no protection at all for such faults. Hence
special protection schemes shall be employed.

One such scheme, often used in certain countries, utilizes a fact that the HV shunt reactor
winding is often made of two half-windings connected in parallel (i.e. the HV lead is brought out
at the mid point of the winding, and the two neutral leads at the bottom and the top of the
winding). This gives the opportunity to install two CTs in the winding star point (i.e. one in each
winding part). Then so-called split phase differential protection can be utilized to detect turn-to-
turn faults. However this protection scheme have the following drawbacks:
• this special CT arrangement typically causes reactor manufacturing problems
• typically low CT ratio is required, which can cause longitudinal differential protection
problems during reactor switching in, if the same CTs are used for both differential
protections
• this scheme can be only used if the shunt reactor is specifically ordered with these CTs

Second turn-to-turn protection scheme for shunt reactors, successfully used in some other
counties, utilize the following facts:
• HV power system voltages are well balanced during normal load conditions
• Modern HV, oil immersed shunt reactors have very small manufacturing asymmetry
between individual phases
• Shunt reactor winding impedance is approximately proportional to the square of the
number of active turns
• Short circuit between some number of turns will cause the decrease of the winding
impedance only in the faulty phase and corresponding small raise of the shunt reactor
neutral point current
• Currents during turn-to-turn fault are of the small magnitude and they will not produce
any sufficient unbalance voltage
• Any external cause of neutral point current (i.e. external phase to ground fault) will cause
appearance of unbalance voltage which can be used to block the operation of turn-to-turn
protection scheme
• In case of a bigger winding turn-to-turn fault which might cause the sufficient voltage
unbalance, sensitive directional zero sequence relay connected on the shunt reactor HV
side and set to look into the reactor shall be capable to detect such fault

This protection scheme was developed even before multifunctional numerical relays were
available. To implement such shunt reactor turn-to-turn protection scheme within multifunctional
numerical relay utilizing its graphical configuration facilities, and readily available logical gates,
timers etc. shall not be a big problem for a protection engineer.

20
In order to verify above statements, shunt reactor behavior, for phase A winding 1% turn-to-turn
faults, is verified by an ATP simulation and it is shown in Figures 21 & 22. From these figures is
obvious that the above-described scheme can be successfully implemented if the power system
itself is well balanced.

150MVA, 220kV, 50Hz Reactor

1.6

1.08

0.57
Current [pu]

0.47

0.98

1.5
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Cycles
IA
IB
IC

Figure 21: Internal Phase A Winding turn-to-turn fault, Phase Currents

150MVA, 220kV, 50Hz Reactor

0.06

0.013
Current & Voltage [pu]

0.033

0.08
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Cycles
IN
3Io
3Uo

Figure 22: Internal Phase A Winding turn-to-turn fault, Zero-sequence Quantities

21
 VII. SHUNT REACTOR MECHANICAL FAULT DETECTION
Similarly to the power transformers, HV oil immersed shunt reactors typically have build-in
mechanical devices for internal fault or abnormal operating condition detection. Typically the
following built-in mechanical fault detection devices can be encountered within shunt reactor:
• gas detection relay (i.e. Buchholz relay) with alarm and trip stage
• sudden pressure relay
• winding temperature contact thermometer with alarm and trip stage
• oil temperature contact thermometer with alarm and trip stage
• low oil level relay

These mechanical relays are excellent compliment to the electrical measuring relays previously
explained. Typically it is recommended to arrange that these mechanical relays trip reactor
circuit breaker independently from electrical relays. However signals from mechanical devices
shall be connected to binary inputs of numerical relays in order to get time tagging information,
disturbance recording and event reporting in case of their operation.

VIII. TYPICAL SHUNT REACTOR CONTROL SCHEMES

The shunt reactors are generally designed for natural cooling with the radiators mounted directly
on the tank. However sometimes it is required to have some control action in the cooling circuit
depending on the status of the shunt reactor circuit breaker. The control action can be initiated by
the circuit breaker auxiliary contact or by operation of an overcurrent relay set to 50% of the
reactor rated current. By using overcurrent relay secure control action is obtained when reactor is
energized regardless the circuit breaker auxiliary contact status.

In order to improve power system performance, lately it is often required by the electrical
utilities to perform automatic shunt reactor in and out switching, by monitoring the busbar
voltage level. This functionality is quite easy to integrate into multifunctional, numerical relay.
However user must carefully check relay performance regarding the following points:
• over/under voltage relay with reset ratio or 1% or better is required for such application
• typically more than one over/under voltage level with independently settable time delays
are required within the relay
• over/under voltage relay shall be capable to operate only when all three voltages are
above/below set operate level or relay must be capable to measure and operate on the
value of the positive sequence voltage

22
 IX. TRADITIONAL SHUNT REACTOR PROTECTION
AND CONTROL SCHEMES

Usually multifunctional numerical protection relays are used for both power transformer and
shunt reactor protection. However, typically old protection schemes for shunt reactor protection,
with just a few protection functions are still specified and applied today.

Two such traditional protection arrangements are shown in the following two figures.

IA
IB
50/51
IC

87N

IN
50G/51G

Figure 23: Typical Shunt Reactor Protection Scheme No1

The first protection scheme utilizes restricted ground fault protection (i.e. 87N) as reactor unit
protection. This protection shall trip instantaneously for all internal phase to ground faults. For
internal phase-to-phase fault detection, overcurrent protection (i.e. 50/51) is utilized. Ground
overcurrent protection (i.e. 50G/51G) is used as backup protection for ground faults and as main
protection for circuit breaker pole disagreement condition.

23
 IA
IB
50/51 50N/51N
IC

87

IcN
IbN
IaN

Figure 24: Typical Shunt Reactor Protection Scheme No2

The second protection scheme utilizes differential protection (i.e. 87) as reactor unit protection.
This protection shall trip instantaneously for all internal phase to phase and phase to ground
faults. Overcurrent protection (i.e. 50/51) is used as backup protection for internal phase-to-
phase faults. Residual overcurrent protection (i.e. 50N/51N) is used as backup protection for
ground faults and as main protection for circuit breaker pole disagreement condition.

Actually it shall be noted that the numerical multifunctional relays can offer much more
functionality than shown on the above two figures. Please refer to the following chapter to see
proposed shunt reactor protection scheme with multifunctional numerical protection relay.

24
 X. CONCLUSIONS
The paper has described a number of details regarding HV shunt reactors and their protection
and control schemes. In order to help the end user to properly select and apply multifunctional
numerical relays for HV shunt reactor protection and control, an example of possible application
of such relay, which utilizes DFT filtering technique, is presented in Figure 25.

IA
IB
50/51 50N/51N 67N
#1
IC

87N
UA
UB 59 27 59N
UC

87

IcN
IbN
IaN 50/51 49 50/51
#2 #3

IN
50G/51G

Figure 25: Example of complete HV shunt reactor protection and control scheme
with multifunctional, numerical relay

All proposed protection or control functions in Figure 25 are typically readily available in
multifunctional numerical transformer protection relays. However suitability of a particular relay
to be used for shunt reactor application shall be carefully evaluated.

Table 1 gives the summary about each function from Figure 25 as well as some typical setting
values [5]. The proposed settings shall be considered only as guidelines.

It is hoped that this paper will provide some guidance to those seeking assistance in HV shunt
reactor protection and control issues.

25
 Typical setting shown in percents
Function Comment
of the shunt reactor rating
87=low impedance Set restraint differential level to 10-15% with 2nd
Check suitability for shunt reactor
differential harmonic restrain set at 10%. Set unrestraint
application with relay manufacturer.
protection differential level 200%.
87N=low
Set differential level to 10%. Set operate angle for
impedance, Check suitability for shunt reactor
directional criteria to ±65 deg. Relay shall include
restricted ground application with relay manufacturer.
adaptive 2nd harmonic restrain feature.
fault protection
#1-50/51=HV Set low set to 130% with time delay in between
Backup protection, sensitive for internal 0.6s and 1s. Set high set to 250% with time delay
overcurrent
faults close to the reactor bushings.
protection of 0.1s. *
#2-50/51=HV Set low set to 130% with time delay in between
Backup protection, sensitive for internal 0.6s and 1s. Set high set to 200% with time delay
overcurrent
fault close to the reactor star point.
protection of 0.1s. *
#3-50/51=HV Used as circuit breaker failure Set low set to 30% with appropriate time delay as
overcurrent protection and indication that reactor is CBF protection. Set high set to 50% in order to
protection energized for the cooling control logic. indicate that shunt reactor is energized. *
Shall be used with great care. Shunt
reactor overload can only be caused by
overvoltage in a power system. That is
Specific manufacturing data are required in order
49=thermal the exact time when reactors are
to properly set this function. Possible to use
overload protection required to be energized. Thus it might
winding/oil contact thermometer instead.
come in conflict with shunt reactor
voltage/reactive power control
functionality in the power system.
50G/51G=ground Backup protection, sensitive for internal
Specific system data are required in order to
fault protection in fault close to the reactor star point. Used
properly set this function.
reactor neutral point for turn-to-turn fault detection logic.
50N/51N=ground Set low set to 20% with time delay in between
fault overcurrent Backup protection, sensitive for internal 0.6s and 1s or even longer. Use 2nd harmonic
protection in reactor faults close to the reactor bushings. blocking. Set high set to 175% with time delay of
HV side 0.1s. *
59N=unbalance Used for turn-to-turn fault detection Specific system data are required in order to
overvoltage logic. properly set this function.
67=directional
Used for turn-to-turn fault detection Specific system data are required in order to
ground fault
logic. properly set this function.
protection
Used for automatic shunt reactor
27&59=under/over Specific system data are required in order to
control. Often more than one stage
voltage properly set these functions.
required.

Table 1: List of functions for complete HV shunt reactor protection and control scheme

* These settings are proposed for HV shunt reactors with own circuit breaker. In case that the
HV shunt reactor is directly connected to the HV line without its own circuit breaker, these
settings have to be revised in order to prevent unwanted operation, when:
HV line is de-energized, due to low-frequency, long time constant transients which are
determined by the combination of line capacitance and shunt reactor inductance [4]&[6]
One HV line phase is open during dead time of single pole autoreclosing cycle

26
APPENDIX I
Basic relationships between shunt reactor most important quantities are shown here:

U2
Q = 3 *U * I = X * I 2 = (App_I.1)
X
PTotal ≈ 0.002 * Q (App_I.2)
PCu ≈ 0.75 * PTotal ≈ 0.0015 * Q (App_I.3)
PFe ≈ 0.25 * PTotal ≈ 0.0005 * Q (App_I.4)

where:
• Q is shunt reactor rated reactive power in VAr
• U is shunt reactor rated phase-to-phase voltage in V
• I is shunt reactor rated phase current in A
• X is shunt reactor rated reactance in Ω
• Ptotal is total shunt reactor losses in W (typically total shunt reactor losses are in order of
0,2% in accordance with references [1]&[2])
• PCu is total shunt reactor copper losses in W (typically shunt reactor copper losses are in
order of 75% of the total losses in accordance with reference [1])
• PFe is other shunt reactor losses (most of them are iron losses) in W

Above data are important in order to properly simulate HV shunt reactors.

27
APPENDIX II
Typical HV shunt reactor magnetizing characteristic is shown in Figure 26.

Shunt Reactor Characteristic

2.5

2
Voltage [pu]

1.5
1

0.5
0
0 1 2 3 4
Current [pu]

Figure 26: Typical magnetizing characteristic of a gapped core shunt reactor

The relation between voltage and current peak value in a HV shunt reactor can be described by
two lines, one below saturation and the other above saturation. The point where the two lines
intersect is called knee point. The relation between voltage and current peak value is shown in
Figure 26. The knee point usually corresponds from 125% to 135% of the rated voltage. The
slope of the saturated part is 20% to 40% of the slope in the unsaturated region.

Below the knee point the reactor current is sinusoidal (i.e. this means that the relation between
the current peak value and the root-mean-square value of the current is fix and equal to 2 ) and
its magnitude is directly proportional to the applied sinusoidal voltage. Thus for normal
operating voltages there is a linear relationship between applied voltage and reactor current (i.e. a
small increase in voltage will result in a proportional increase in current). Magnetic fluxes and
flux densities are also proportional to the time integral of the applied voltages. With a voltage of
sinusoidal shape the fluxes and flux densities are also proportional to the voltage. The deviation
from a true sinusoidal shape in phase voltage is generally negligible for normal operating
conditions of the power system.

For operation above the knee point the current peak value increases faster than the root-mean-
square value, and the quotient between the peak value and the root-mean-square value exceeds
value 2 .

28
BIBLIOGRAPHY
1. Carlson, Å., “Shunt Reactor Manual”, ABB Power Technology Products/Transformers,
Ludvika-Sweden, 2002-08-20
2. ABB Transformatori, “Pamphlet Shunt Reactors”, Roma-Italy
3. Switchgear Committee of the IEEE Power Engineering Society, “IEEE Application Guide
for Shunt Reactor Switching”, IEEE Std C37.015-1993
4. IEEE Power Systems Relaying Committee, “IEEE Guide for the Protection of Shunt
Reactors”, ANSI/IEEE C37.109-1988
5. Nylen, R., “Shunt Reactor and SVC Protection Application Guide”, ABB Relays, Västerås-
Sweden, 1988-11-17
6. Elmore, W. A., editor, Protective Relaying: Theory and Applications, Marcel Dekker, Inc.,
1994.
7. ATP is the royalty-free version of the Electromagnetic Transients Program (EMTP). For
more info please visit one of the following web sites: http://www.eeug.de/ or
http://www.ee.mtu.edu/atp/

BIOGRAPHICAL SKETCHES

Zoran Gajić was born in former Yugoslavia in 1965. He received his BSEE with honors from University
of Belgrade, Yugoslavia in 1990 and GDE in Computer Engineering from Witwatersrand University,
Johannesburg-RSA in 1995. Since 1993 he has been working in the area of power system protection and
control within ABB Group of companies, where he had various engineering positions. Currently he has a
position of Protection Application Specialist with ABB Automation Technologies, Substation
Automation, Sweden. He is a member of IEEE and PES. Zoran has published several technical papers in
the relay protection area. His main working areas are computer applications for protection and control of
electrical power systems, protection and control algorithms for microprocessor based relays and power
system simulation. Zoran is co-holder of two patents.

Birger Hillström was born in Sweden 1944. He received his M.Sc.E.E degree from Chalmers Technical
University in Goteborg Sweden, 1968. He was employed as development engineer at ASEA, Sweden,
1970 and is currently working as development project manager at ABB Automation Technologies,
Substation Automation, Sweden. During ten years he was relay laboratory manager and was then
responsible for development of a digitally controlled, analogue power system simulator and was after that
part of a specification group of a real time digital power system simulator. His special areas of interest are
transient network analysis, development and testing of line and transformer static and numerical relays.

Fahrudin Mekić was born in former Yugoslavia in 1967. He received his BSEE with honors from
Sarajevo University, Bosnia and Herzegovina in 1991 where he also worked as research assistant. He
received his MSEE degree from Istanbul Technical University, Turkey in 1996. Since 1996 he has been
working in the area of power system protection and control within ABB, where he had various
engineering positions. Currently he is Senior Application Engineer with the Substation Automation and
Protection Division, ABB Inc, in Allentown, PA. Fahrudin has published several technical papers in the
area of protection and reliability. He is currently responsible for the application and technical issues
associated with ABB relays. He is a member of IEEE.

29