The two general types of construction used for shunt reactors are dry-type and oil-immersed. The construction features of each type, along with variations in design, are discussed under the headings which follow.

Dry Type
Dry-type shunt reactors generally are limited to voltages through 34.5 kV and are usually applied on the tertiary of a transformer which is connected to the transmission line being compensated. The reactors are of the air-core (coreless) type, open to the atmosphere, suitable for indoor or outdoor application. Natural convection of ambient air is generally used for cooling the unit by arranging the windings so as to permit free circulation of air between layers and turns.

The layers and turns are supported mechanically by bracing members or supports made from materials such as ceramics, glass polyester, and concrete. The reactors are constructed as single-phase units and are mounted on base insulators or insulating pedestals which provide the insulation to ground and the support for the reactor.

Because the dry-type shunt reactor has no housing or shielding, a high-intensity external magnetic field is produced when the reactor is energized. Care is thus required in specifying the clearances and arrangement of the reactor units, mounting pad, station structure, and any metal enclosure around the reactor or in the proximity of the reactor.

A closed metallic loop in the vicinity of the reactor produces losses, heating, and arcing at poor joints; therefore, it is important to avoid these loops and to maintain sufficient separation distances. Shielding may be required when it is not possible to arrange dry-type units in an equilateral-triangle configuration isolated from external magnetic influences. This shielding is required to limit the impedance deviation between phases. Deviation from impedance values for reactors will result in a deviation from the actual MVAR rating.

For the same range of applications, the primary advantages of dry-type air-core reactors, compared to oil-immersed types, are lower initial and operating costs, lower weight, lower losses, and the absence of insulating oil and its maintenance. The main disadvantages of dry-type reactors are limitations on voltage and kVA ratings and the high intensity external magnetic field mentioned above. Because these reactors do not have an iron core, there is no magnetizing inrush current when the reactor is energized.

The two design configurations of oil-immersed shunt reactors are coreless type and gapped iron-core type. Both designs are subject to low-frequency longtime constant currents during de-energizing, determined by the parallel combination of the reactor's inductance and line capacitance. However, the gapped iron-core design is subject to more severe energizing inrush than the coreless type.

Most coreless shunt reactor designs have a magnetic circuit (magnetic shield) which surrounds the coil to contain the flux within the reactor tank. The steel core-leg that normally provides a magnetic flux path through the coil of a power transformer is replaced (when constructing coreless reactors) by insulating support structures. This type of construction results in an inductor that is linear with respect to voltage.

The magnetic circuit of a gapped iron-core reactor is constructed in a manner very similar to that used for power transformers with the exception that small gaps are introduced in the iron core to improve the linearity of inductance of the reactor and to reduce residual or remanent flux when compared to a reactor without a gapped core.

Oil-immersed shunt reactors can be constructed as single-phase or three-phase units and are very similar in external appearance to that of conventional power transformers. They are designed for either self cooling or forced cooling.


Manual closing or autoreclosing without synchronization supervision at line terminals that are in close electrical proximity to turbine-generators can subject them to excessive shaft torques and winding stresses with resultant loss of life of the turbine-generator system.

These effects should be studied and evaluated before autoreclosing is initiated by tripping. It is preferable to re-energize a line at a terminal remote from the generator bus, check synchronism between the generator bus and line, and then close the breaker at the generator end.

In past years, considerable research and analysis focused on the stresses in the shafts and components of turbine-generators due to switching operations. There is little documentation of actual damage to, or failure of, turbine-generators resulting from autoreclosing or switching.

The effects of these stresses induced are cumulative and can be caused by normal switching operations or system faults. Therefore, autoreclosing can be a contributing factor to machine failure, but not necessarily the sole contributor.

An unsuccessful autoreclose attempt (particularly three-phase faults) close in to a generating plant can contribute to accelerated torsional fatigue on the turbine-generator shafts (ANSI C30.13-1977 [B1], IEEE Committee Report [B8], and Jackson et al. [B12]). This can be dealt with by not autoreclosing near generating plants or by blocking the autoreclose for close-in faults or three-phase faults.

Consideration should be given as well to the natural oscillatory frequency of the transmission line as autoreclosing can result in a resonance condition, which could contribute to other system problems. The operation of closing a breaker in the power system can result in the creation of power transients and current oscillations, which can stress or damage generating units located electrically close to that breaker.

These transients effect various components of the turbine-generator. The concern is the average initial power, dP, which occurs when the breaker is closed, and its effect in producing torsional stresses, primarily in the rotational members of the turbine-generator.

For this condition, the permissible limit for dP or dI at the generator terminals are 0.5 per unit based on the rated load and power factor. Regardless of the cause of initial disturbance, autoreclosing times in excess of 10 s appear long enough to allow the oscillations from the initial disturbance to die out.

Turbine-generators when subjected to high-speed autoreclosing can resonate at the natural frequencies of the turbine and shaft. These transient torques will cause cyclic stress variations in the generator shaft resulting in cumulative fatigue damage when they exceed material fatigue limits.

This results in reduced component life of shafts, retaining rings, and rotors. In extreme cases, these torsional vibrations have led to growing oscillations resulting in shaft damage. Some of the more recent papers on the subject of shaft fatigue as a result of high-speed autoreclosing (ANSI C30.13-1977 [B1]) suggest that simple measures such as dP or dI cannot be correlated directly.

Transient torque studies that quantify the impact of high-speed autoreclosing can be performed to calculate the impact on the turbine-generators. This study would require a detailed turbine-mass representation and is generally performed using the Electro Magnetic Transient Program (EMTP). This study can then provide a basis for evaluating the need for torsional monitors/relays on the turbine-generators.

The torsional monitoring devices monitor the turbine-generator shaft for torsional oscillations by providing torsional mechanical response evaluation, shaft torsional stress, and fatigue evaluation, and can be used by the operator to assess torsional impact of an event on a unit. The torsional protective devices continuously monitor the turbine-generator shaft and provide trip output contacts when shaft fatigue reaches predetermined levels.
As a result of the apparent risk to turbine-generator life, most utilities have modified their autoreclosing practices to some form of the following:

a) Autoreclose by synchronism check only
b) Allow a minimum of a 10-s delay prior to any autoreclose attempt
c) Use single pole tripping and allow autoreclose on single phase faults only
d) Autoreclose lines with tapped generation only under dead-line conditions
e) Use no autoreclosing near generation


The ultimate surge voltage protection is obtained through arrester voltage ratings as low as system grounding conditions will permit during normal and abnormal system conditions. Initially, however, when the surge arrester was adopted as the basic protection device, the equipment design (coordination of major insulating structures) assumed that an “ungrounded neutral” or “100% rated” arrester would be used, unless otherwise specified.

In time, after successful service experience with 100% rated arresters (100% of maximum line-line voltage), it was reasoned that lower rated arresters would be suitable on grounded neutral systems. On these systems, the TOV on the unfaulted phases during a line-to-ground fault would bear the same relationship to arrester rating as “maximum line-line voltage” in an ungrounded system.

An “effectively grounded” system was then defined in terms of the symmetrical-component sequence resistances and reactances, for which the TOV on an unfaulted phase does not exceed 80% of the maximum line-to-line voltage. Under this condition, an arrester rated at 80% of maximum line-to-line voltage was deemed applicable, and it was classified as a “grounded neutral” arrester.

The use of a “grounded neutral” arrester with lower protective levels enabled designs in some electrical equipment, such as transformers, to have reduced insulation levels with adequate protection. Reduced insulation allowed reduction in size, weight, and cost. Subsequently, still lower rated arresters were commonly applied whenever the grounding was significantly better than “effective,” particularly at system voltages where these reductions were significant (above 230 kV).

Usually the TOV produced by a system ground fault is greater than that produced by other causes (generator overspeed, ferroresonance, harmonics, etc.). An exception to this might occur on systems where the coefficient of grounding is less than 80%. The rating of gapped silicon-carbide surge arresters generally exceeded the TOV due to a phase-to-ground fault on the system where it was applied.

This criterion was based on the assumptions that the maximum TOV is produced by a ground fault and that the arrester might operate due to a surge while there was a ground fault on another phase. The arrester had then to seal off against the TOV, which was sustained until the fault was interrupted.

There were some arresters that sealed off against voltages higher than their rating. Overvoltage characteristics for these arresters were published in the late 1960s or early 1970s. This feature has sometimes been utilized to provide lower protective levels.

An important consideration for selecting a metal-oxide arrester is the maximum continuous operating voltage (MCOV); however, the arrester will also be subjected to TOVs. A conservative criterion is that the TOV should not exceed the duty cycle voltage rating of the arrester. However, metal-oxide arresters can have thermal capability for TOVs in excess of the duty-cycle rating for specified times, and data and curves of TOV versus allowable time of the overvoltage are available.


IEEE Std C37.48-1997 covers in detail the application guidelines for high-voltage external capacitor fuses.

The energy stored in the healthy capacitors of one series group of parallel-connected capacitors will discharge into the failed capacitor unit of that group and its fuse. The fuse shall be able to interrupt the energy supplied by the parallel group of capacitor units when they are charged to their peak voltage.

If the capacitor bank design has an available discharge energy higher than the capacitor units or expulsion fuses can withstand, current-limiting fuses with adequate energy rating should be considered. When ungrounded wye capacitor banks are supplied in an enclosure, current-limiting fuses shall be used to eliminate the arc products that occur with the use of an expulsion fuse.

These arc products in the confined enclosure could cause further evolution of the fault. Current-limiting fuses may also be required on enclosed single-group ungrounded wye banks that are designed with two bushing units. In this design, the first bushing is used for the phase connection, the second bushing used for the neutral connection, and the case connected to ground.

This arrangement requires the capacitor fuses to interrupt system fault current in the event of a failure of the unit insulation near the phase bushing. NEMA CP1-1988 [B11] suggests a parallel energy limit of 15 kJ (4650 kvar) for all film dielectric capacitors.

Expulsion fuses are frequently applied with higher parallel energy (to 30 kJ) (Mendis et al. [B9]). This higher energy application is acceptable if the total available discharge energy of the bank does not exceed the discharge energy rating of the fuse or the capability of the faulted capacitor unit.

To determine proper fuse selection, the capacitor unit case rupture curve shall be available from the manufacturer.

Case rupture curves are different for different capacitor unit constructions and designs. The total clearing curve of the fuse or fuse link is then compared to the case rupture curve; adequate protection is assured if the total clearing curve of the fuse is to the left of and below the rupture curve of the capacitor unit.

Other important considerations for external fuse selection and operation include the following:

— Fuses should be designed and rated for the externally fused capacitor bank application.
— Fuses should provide for the fast isolation of a faulted capacitor unit.
— Voltage interruption capability of the fuse shall be coordinated with the voltage withstand capability of the capacitor unit.
— Fuses shall handle the transient inrush and outrush current.
— Fuses shall be designed for the current loadability requirements, including harmonics and adequate allowance factors.
— Fuses shall be designed for the inductive and capacitive current interruption capability.
— Fuse characteristics shall coordinate between the different shunt bank protection schemes and the characteristics of the fuses (that is, expulsion, current-limiting, or a combination of both).


The application of large shunt capacitor banks with switched parallel banks in high-voltage transmission systems involves a number of considerations, one of which is grounding. It is generally recommended that the neutral of capacitor banks be grounded only to systems that are effectively grounded.

In the event of a phase-to-ground fault, a grounded capacitor bank neutral in an otherwise ungrounded system may lead to high transient overvoltages in the system and capacitor bank as a result of restriking of the arcing fault to ground.

One of the main advantages associated with neutral grounding concerns the severity of the recovery voltage across the first pole of the switch to clear, interrupting the charging current of the capacitor bank.

The recovery voltage across the first pole to open consists of trapped charges on the capacitors and the variation in the 60 Hz voltage of the system. Due to system parameters and capacitor bank size, the recovery voltage can be approximately two times normal peak voltage when the bank is grounded.

On an ungrounded bank, the magnitude of the first peak of the recovery voltage can be as high as three times the peak system line-to-ground voltage when the bank is switched. Because recovery voltage is a critical factor in determining the capability of a switching device to switch capacitive reactive power, it may be desirable (in terms of switch performance) to ground the neutral of shunt capacitor banks.

IEEE Std C37.04-1979 and ANSI C37.06-1997 recommend that both the shunt capacitor bank and the system be grounded at voltage levels of 121 kV and above. Many capacitor banks of higher voltage are installed ungrounded, but the circuit breaker manufacturer should be consulted for the application of a breaker if these conditions are not met.

While many shunt capacitor banks are directly connected to a high-voltage substation bus, switched capacitor banks may be applied to tertiaries of power transformers that are connected to the line or possibly to the bus. Grounding the neutral of the wye-connected capacitor bank should be done only on an effectively grounded system.

For instance, the delta tertiary of the auto transformer represents an isolated source; grounding the capacitor bank neutral makes this side of the transformer capacitively grounded. Overvoltages may be experienced during line-to-ground faults for certain ratios of X0/X1, depending on system, transformer, and capacitor bank parameters.

If the neutral is to be grounded on a system that is not effectively grounded, the application should be thoroughly analyzed for proper application of surge arresters, bank configuration, bank switching devices, etc.