Вы находитесь на странице: 1из 30

# HV SHUNT REACTOR SECRETS FOR PROTECTION ENGINEERS

By Zoran Gaji ABB Sweden Vsters, Sweden Birger Hillstrm ABB Sweden Vsters, Sweden Fahrudin Meki ABB Inc. Allentown, PA 18106

Presented to: 30 Western Protective Relaying Conference Spokane, Washington October 21-23, 2003
th

## HV SHUNT REACTOR SECRETS FOR PROTECTION ENGINEERS

Z. Gaji, B. Hillstrm ABB Sweden Vsters, Sweden F. Meki ABB Inc. 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.
A B C
IA IB IC UA UB UC 3Io

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 Y/2 Y/2

Vr

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 receivingend voltage and current. Writing Kirchoff Voltage Law equation for the circuit in Figure2,

Vs = Vr + Z ( Ir + Vs = (1 +

Vr * Y ) 2

(1) (2)

## Y *Z ) * Vr + Z * Ir 2 Also writing a Kirchoff Current Law equation at the sending end, Vr * Y Vs * Y + 2 2 Y *Z Y *Z Is = Y (1 + ) * Vr + (1 + ) * Ir 4 2 Is = Ir +

Vs AB Vr Is = CD Ir

(3) (4)

(5)

where:

A = D = 1+
B = Z ( )

Y *Z ( perunit ) 2

(6) (7)

C = Y (1 +

Y *Z )( S ) 4

(8)

Example 1: A three-phase line, completely transposed 345kV, 124 miles has the following positive sequence constants:
z = 0.0515 + j 0.563( / mile) y = j 6.76 *10 6 ( S / mile)

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

A = D = 0.97060.159 0 B = 70.2984.78 0 C = 8.277 *10 4 90.08 0 From (5), the no-load receiving-end voltage is VrNL = Vs 345.8 = = 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.

V(x)

V RNL

SIL Vs

V RSIL

VS

Short-circuit

V RFL

V RSC
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 = 70.2984.78() Y = 8.4 * 10 4 (1 0.75) = 2.1 *10 4 90 0 ( S ) From equations (6), (7) and (8)

A = D = 0.9930.04 0 From (5), the no-load receiving-end voltage is Vs 345.8 = = 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. VrNL =
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

52 52 52

52

## 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 F /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).

## 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 %.

## 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
4

2 Current [pu]

17

18

19

20

21

22

23 Cycles

24

25

26

27

28

29

30

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.

1.75

17

18

19

20

21

22

23 Cycles

24

25

26

27

28

29

30

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.
2

Current [pu]

4.5

5.5

6.5

7 Cycles

7.5

8.5

9.5

10

IA IB IC

1.25

4.5

5.5

6.5

7 Cycles

7.5

8.5

9.5

10

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

## 1 Current & Voltage [pu]

5.22

5.44

5.67

5.89 Cycles

6.11

6.33

6.56

6.78

IA IB IC UA UB UC

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

## 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 4 3 Current [pu] 2 1 0 1 2

20

30

40

IC

50 Cycles

60

70

80

## 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).

2.25

IC RMS value

1.69

[pu]

1.13

0.56

20

30

40

TRMS DFT

50 Cycles

60

70

80

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.
4

## 150MVAr, 220kV, 50Hz Reactor

CT Saturation Instant

Current [pu]

10

15 Cycles

20

25

30

35

IC

13

1.5

10

15

20

DFT IC

25 Cycles

30

35

40

45

50

## 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 doesnt 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. Lets have a look into the disturbance-recording file captured by numerical relay, which is shown in Figure 14.

14

## 150MVAr, 220kV, 50Hz Reactor

1 Current [pu] 0 1

10

15 Cycles

20

25

30

35

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:
100

## Phase Angle Between IN & 3Io

80

60 Angle [deg]

40

20

20

10

15 Cycles

20

25

30

35

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 doesnt 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. Lets have a look into the disturbance recording file captured by numerical relay, which is shown in Figure 16.
4

2 [pu] 0 2

2.08

4.17

6.25

8.33

10.42

12.5 Cycles

14.58

16.67

18.75

20.83

22.92

25

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.
1.6

1.08

0.47

0.98

1.5

8 Cycles

10

11

12

13

14

15

16

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

2.5

8 Cycles

10

11

12

13

14

15

16

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.
1.6

1.08

0.47

0.98

1.5

8 Cycles

10

11

12

13

14

15

16

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.
15

8 Cycles

10

11

12

13

14

15

16

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-toturn 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.

1.6

1.08

0.47

0.98

1.5

8 Cycles

10

11

12

13

14

15

16

IA IB IC

0.06

0.013

0.033

0.08

8 Cycles

10

11

12

13

14

15

16

IN 3Io 3Uo

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 IC

50/51

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 IC

50/51

50N/51N

87

IbN IaN

IcN

## 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-tophase 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 IC

50/51
#1

50N/51N

67N

87N
UA UB UC

59

27

59N 87

IbN IaN

IcN

50/51
#2

49

50/51
#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

Function
87=low impedance differential protection 87N=low impedance, restricted ground fault protection #1-50/51=HV overcurrent protection #2-50/51=HV overcurrent protection #3-50/51=HV overcurrent protection

Comment
Check suitability for shunt reactor application with relay manufacturer. Check suitability for shunt reactor application with relay manufacturer. Backup protection, sensitive for internal faults close to the reactor bushings. Backup protection, sensitive for internal fault close to the reactor star point. Used as circuit breaker failure protection and indication that reactor is energized for the cooling control logic. Shall be used with great care. Shunt reactor overload can only be caused by overvoltage in a power system. That is the exact time when reactors are required to be energized. Thus it might come in conflict with shunt reactor voltage/reactive power control functionality in the power system. Backup protection, sensitive for internal fault close to the reactor star point. Used for turn-to-turn fault detection logic. Backup protection, sensitive for internal faults close to the reactor bushings. Used for turn-to-turn fault detection logic. Used for turn-to-turn fault detection logic. Used for automatic shunt reactor control. Often more than one stage required.

Set restraint differential level to 10-15% with 2nd harmonic restrain set at 10%. Set unrestraint differential level 200%. Set differential level to 10%. Set operate angle for directional criteria to 65 deg. Relay shall include adaptive 2nd harmonic restrain feature. Set low set to 130% with time delay in between 0.6s and 1s. Set high set to 250% with time delay of 0.1s. * Set low set to 130% with time delay in between 0.6s and 1s. Set high set to 200% with time delay of 0.1s. * Set low set to 30% with appropriate time delay as CBF protection. Set high set to 50% in order to indicate that shunt reactor is energized. *

## Typical setting shown in percents of the shunt reactor rating

Specific manufacturing data are required in order to properly set this function. Possible to use winding/oil contact thermometer instead.

50G/51G=ground fault protection in reactor neutral point 50N/51N=ground fault overcurrent protection in reactor HV side 59N=unbalance overvoltage 67=directional ground fault protection 27&59=under/over voltage

Specific system data are required in order to properly set this function. Set low set to 20% with time delay in between 0.6s and 1s or even longer. Use 2nd harmonic blocking. Set high set to 175% with time delay of 0.1s. * Specific system data are required in order to properly set this function. Specific system data are required in order to properly set this function. Specific system data are required in order to properly set these functions.

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: Q = 3 *U * I = X * I 2 = U2 X (App_I.1) (App_I.2) (App_I.3) (App_I.4)

PTotal 0.002 * Q PCu 0.75 * PTotal 0.0015 * Q PFe 0.25 * PTotal 0.0005 * Q

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 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.

2 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-meansquare value, and the quotient between the peak value and the root-mean-square value exceeds value 2 .

28

BIBLIOGRAPHY
1. Carlson, ., 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, VstersSweden, 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 Gaji 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 Hillstrm 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 Meki 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