Industrial Power Transformers -- Special features of transformers for particular purposes [part 15]

Home | Articles | Forum | Glossary | Books


A work on transformers would not be complete without a description of reactors. The two are, of course, very different, but they also have a lot of similarities, and reactors have been described as transformers without secondary windings. That is indeed, a good starting point from which to consider them and to understand the way in which they operate.

In a power system reactors have two main functions:

(1) They can be used to provide reactance, which restricts current flow and thereby reduce system short-circuit levels.

(2) They can be used to absorb lagging VArs, or MVArs, to assist in the control of system load flow and voltage.

In the former role they are known as series reactors since they are installed in series with the supply and in order to perform their current limiting function they must have a relatively low reactance, or impedance, of between, say, 5 and 10 percent on rating.

In the latter role they are termed shunt reactors and must have a reactance, on rating, of 100 percent because in order to absorb VArs they must be connected as a load, that is between phases, in delta, or between phase and neutral, in star.

Series and shunt reactors therefore have many similar design and constructional features but it is the differing requirements with regard to reactance that determines the way in which these features are incorporated and the way in which they differ from transformers. Provision of the required parameters introduces problems not encountered in the design and construction of transformers, so that not every manufacturer skilled in the production of transformers will wish to manufacture reactors, and the field tends to be limited to a small number of specialists.

Like transformers, reactors can be oil immersed, or they can be dry type, air insulated air cooled. The larger types are generally oil immersed and, as has been the case with transformers, it is primarily oil-immersed reactors that will be examined in this volume.

Shunt reactors

It has been said that a reactor is a transformer without a secondary winding. If a transformer has its primary winding connected to an HV supply with the secondary open circuited, no current, except for a very small magnetizing current, will flow. Only by allowing load current to flow in the secondary to produce a demagnetizing m.m.f. acting on the core will load current be drawn through the primary winding.

The situation would be the same with a shunt reactor connected to an HV sup ply. If by drawing a magnetizing current from that supply a flux of comparable magnitude to that in the core of a transformer were allowed to become established, no current, other than that small magnetizing current, would flow, and the reactor would not perform its duty of absorbing MVArs. To enable the reactor to function, therefore, the reluctance of the magnetic circuit must be greatly increased, so that the flux density is very much reduced.

This is the basic difference between reactors and transformers.

The reluctance of the magnetic circuit can be increased by a variety of measures from the introduction of air gaps into the iron circuit to dispensing with the iron circuit altogether.

FIG. 50 Gapped iron-core reactor

There are two basic forms of shunt reactor construction. These are:

(1) Oil immersed gapped iron cored.

(2) Oil immersed magnetically shielded coreless.

Reactors with gapped iron cores are most like transformers in their appearance and construction. In a three-phase reactor, a core of superficially similar format to a normal transformer core carries one winding on each limb, similar to a transformer winding. The core however differs from a transformer core in that 'gaps' are inserted into the axial length of the wound limbs by the insertion of distance pieces made from non-magnetic material -- this may be press board or glass-reinforced resin to cope with the combination of temperature and mechanical loading. These normally make up no more than about 1 or 2 percent of the iron path length. Such a device is shown diagrammatically in FIG. 50. The flux density at rated supply voltage is determined by the number of turns in the winding in a similar way to that of a transformer. The external laminated yokes provide a return flux path and limit the level of leak age flux which enters the tank and adjacent steelwork reducing any extraneous heating of these components. Increases in the supply voltage above nominal will lead to increases in flux density, so that as saturation is approached the relationship between applied voltage and reactance becomes non-linear. The ratio between the length of the magnetic steel circuit and the total length of the gaps determines the degree of this non-linearity and the voltage level at which onset occurs. A more significant effect of saturation is the problem of 'flux fringing' at the air gaps. This is the distortion of the flux where it crosses the gaps, as shown in FIG. 51. This effectively means that there is a portion of the flux that is entering the laminations at an angle to their preferred axis of magnetization, which locally increases eddy current loss and vibration - which means noise. In addition, some of this fringing flux will enter the winding and increase the eddy current loss in the windings. Reduction in the fringing flux can be achieved by reduction in gap length, but this necessitates more gaps and reduces the mechanical rigidity of the core. Optimizing the design involves achieving the best compromise between these conflicting factors. The extent to which the flux enters the laminations at an angle to their main axis can be reduced by stacking laminations radially, for example as shown in FIG. 52.

This of course increases manufacturing cost.

FIG. 51 Simplified diagram illustrating flux fringing in the air gaps of a gapped-core shunt reactor.

FIG. 52 Cross-section of core limb showing one way in which laminations can be stacked radially.

In a magnetically shielded coreless reactor, the magnetic shield is arranged to surround the coils in much the same way as the yokes of a conventional gapped iron-core reactor. As in the gapped-core reactor the shield provides a return path for the coil flux thus preventing this from entering the tank, which would result in large losses and tank heating. The larger the cross-section of the shield the greater is the quantity of iron required, the larger is the tank and oil quantity, and the more costly the reactor. If the shield cross-section is reduced, the flux density under normal rated conditions increases and the tendency to saturate under conditions of high system voltage is increased.

Reactor windings are very similar to those of transformers. The requisite number of turns of the appropriate cross-section are wound as either a single helical coil or a stack of disc coils. A high-voltage shunt reactor may have non-uniform insulation so that the winding may be arranged to have its line lead at the center of the leg with two half-windings wound with the opposite polarity in each half of the limb. This arrangement also helps to increase the ratio of series to shunt capacitance and thus assists with impulse voltage distribution.

Of course other measures, such as those used in transformers are also available to improve the impulse voltage distribution if necessary. Shunt reactors are not subject to system fault currents as are transformers so at least one potential problem for the designer is avoided.

Series reactors

Series reactors are sometimes referred to as current limiting reactors and, as already identified, they are used for the purpose of limiting fault currents or restricting the short-circuit levels of transmission and distribution networks and works auxiliary systems, which includes those of power stations. The usual reason for wishing to limit short-circuit levels is to ensure that the sys tem will remain within the short-circuit capability of the system switchgear and, provided the requirements with regard to system regulation can be met, the use of current limiting reactors can often enable more economic fault ratings for switchgear to be employed. For the auxiliary systems of power stations, switchgear of high fault rating has been developed, primarily to make possible the direct-on-line starting of large drives, so the use of series reactors for these is the exception rather than the rule although in some instances these are installed between station and unit switchboards to limit fault levels when the unit and station transformers are paralleled for load transfer purposes.

Series reactors may have the same form of construction as described above for shunt reactors, but two other types may also be encountered, namely:

(1) Cast-in-concrete air cored.

(2) Oil-immersed electromagnetically shielded coreless.

Ideally, current limiting reactors should have no iron circuit because all iron circuits exhibit a non-linear saturating-type characteristic, so that, under the very overcurrent conditions which the reactor is required to protect against, there is a tendency for the reactance to be reduced. Hence, the prevalence of coreless reactors in this list.

The cast-in-concrete variety is therefore aimed at eliminating iron entirely and consists of a series of non-reinforced concrete posts supporting a helical copper conductor arrangement. The problems with these reactors result from the fact that they present extremely specialized manufacturing requirements, albeit that they are technically fairly crude. They tend to be sold in such small quantities that it is rarely worthwhile for a manufacturer to maintain the expertise required in their construction. The major problem is to cast the concrete posts with a sufficiently consistent quality that they can be guaranteed crack free, particularly since they are arranged in a circle of six, eight or more, all of which must be made without defects to achieve an acceptable reactor.

As a result of the above problems it is likely that enquiries for cast-in-concrete reactors to most electrical plant manufacturers will be met with a totally blank reaction.

If cast-in-concrete reactors are employed care is needed in the design of the floor and the building to house the reactor to ensure that any reinforcing of concrete in these is not influenced or affected by the large magnetic field which the reactor produces in service. Personnel should also be aware of these fields and avoid having anything affected by magnetic fields in their possession, particularly heart pacemakers, when carrying out routine inspections even at some distance from the reactor.

Series reactors with gapped iron cores are similar in construction to the shunt variety except that the total gap length will be greater than for shunt reactors in order to reduce their flux density and also their reactance. Unlike shunt reactors they can be subjected to currents many times greater than their continuous rated current, and in fact, withstanding these currents is an important part of their duty.

Like transformers, series reactors are subjected to large electromagnetic forces under fault conditions. Since each limb has only one winding, there can be no significant axial unbalance such as can be experienced in a transformer, so there will be no major end forces on winding supports. There remains an axial compressive force and an outward bursting force on the coils. The latter is resisted by the tensile strength of the copper which is usually well able to meet this but the winding must be adequately braced to prevent any tendency for it to unwind. Since series reactor windings normally have fewer turns than transformer outer (HV) windings this aspect often requires more careful consideration than for a transformer.

The axial compressive force can, after repeated over-current applications, result in a permanent compression of the winding insulation with the result that windings can become loose. This must be prevented by the application of sufficient axial pressure during works processing to ensure that all possible shrinkage is taken up at that time.

In a magnetically shielded coreless series reactor, the magnetic shield is arranged to surround the coils in the same way as described above for shunt reactors and, as for shunt reactors, the shield provides the return path for the coil flux thus preventing this from entering the tank and leading to increased tank losses and heating. The larger the cross-section of the shield the greater is the quantity of iron required, the larger is the tank and oil quantity, and the more costly the reactor. However, in the case of a series reactor, if the shield cross-section is reduced, the flux density under normal rated conditions increases and the tendency to saturate under short-circuit currents is greater, thus bringing about a greater impedance reduction. A wise precaution when purchasing such a reactor is to specify that the impedance under short-circuit conditions shall not be less than, say, 90 percent of the impedance at normal rated current. FIG. 53 shows the internal arrangement of a three-phase, 30 MVA, 11 kV, 16 percent reactance magnetically shielded reactor.

In many respects the electromagnetically shielded reactor appears the most attractive in that it offers the advantage of constant reactance. In practice, this benefit is usually reflected in the cost. The arrangement of the shield for a single phase reactor is shown in FIG. 54. The shield, which may be of copper or aluminum, provides a path for currents which effectively eliminate the return flux at all points outside the shield. The flow of shield current does, of course, absorb power which appears as heating in the shield. In addition to the balancing effect of the shield currents on flux outside the shield, there is some reduction of the flux within the coil, hence there is a reduction in its reactance. It can be shown, however, that this is independent of the current within the coil and is determined only by the inductance of the coil and the mutual inductance between coil and shield. As in the case of the magnetically shielded reactor, therefore, there is a need to strike an economic balance between physical size, as determined by the size of the shield, and the unwanted reduction of reactance produced by placing the shield too close to the reactor coil. In practice the effective reactance of the coil and shield combination is made about 90 percent of the coil reactance alone.

FIG. 53 Magnetically shielded three-phase reactor 30 MVA, 11 kV, 16% _ 50 Hz, shown out of its tank (ABB Power T&D Ltd)

FIG. 54 Electromagnetically shielded reactor

Testing of series reactors

Testing of all reactors can present problems to the manufacturer, which are not encountered in the testing of transformers. To a certain extent this results from the fact that they are made in very much smaller quantities than transformers and so manufacturers do not equip themselves with the specialized test equipment necessary to deal with them.

Series reactors create two difficulties: one is concerned with proving the performance under short-circuit, the other with proving the adequacy of the interturn insulation.

Proving performance under short-circuit not only involves demonstrating that the reactor will withstand the fault currents which are very likely to be a similar magnitude to those in transformers but, for a magnetically shielded or gapped-cored reactor, also establishing the reactance reduction which occurs under short-circuit conditions.

It is rarely possible to measure the reactance at the full short-circuit level, so that the usual approach is to measure impedance at 50, 75 and 100 percent of rated current. For three-phase reactors, this is normally obtained from volt age and current measurements taken with the windings temporarily connected in star for the purposes of the test. A curve plotted from these values can then be extrapolated to the short-circuit level. Since this will involve consider able extrapolation (although the iron part of the circuit should operate below the knee point of the magnetizing curve -- even at the short-circuit current) it is usual, as a type test, to make a reactance measurement on one coil fully removed from the shield. This establishes an absolute minimum reactance which may be used as an asymptote for the extrapolated reactance curve.

Alternatively, depending on the rating, it is possible that one unit might be taken to a specialized short-circuit testing station.

The normal method of proving the interturn insulation of a transformer is to carry out an induced overvoltage test during which a voltage of twice the normal interturn voltage is developed. Such a test would not be very effective for a series reactor since the 'normal' voltage between turns will be very small and even increasing this to twice its normal value is unlikely to give rise to any particularly searching stress.

The usual solution is to apply an impulse test to each line terminal in turn which will generate a more significant voltage between turns. The test level is usually the same as would be applied to the same voltage class of transformer.

The usual practice is to apply two full-wave shots preceded and followed by a reduced (between 50 and 70 percent) full-wave application.

Shunt reactors have a much higher volts per turn in normal service than do series reactors so in theory an overvoltage test is possible and could be a worthwhile test of the interturn insulation. However, whereas a transformer will have an LV winding through which to supply the unit for an overvoltage test this cannot be done for a reactor. For a 400 kV reactor, even if the manufacturer had a testing transformer that could supply 630 kV for an overvoltage test, it would be extremely unlikely that this testing transformer could do so and also supply the MVAr that would be required. So, like series reactors, shunt reactors are generally tested using a lightning impulse test to prove inter turn insulation. The major insulation of both types can be tested by means of a separate source test, and in fact, since the series reactor must be connected at line voltage, whereas a shunt reactor can have its neutral end solidly grounded and therefore have non-uniform insulation, it is possible that it will be the series reactor that will have the higher separate source voltage test.

The problems of noise and vibration have already been mentioned above, and these are more relevant in the case of a shunt reactor than for a series reactor, since the latter will only operate at high flux density and therefore produce high noise and vibration levels whilst carrying fault current. It is therefore necessary to specify noise level tests at rated volts at full MVAr on all shunt reactors.

For both types, winding losses will occur at very low power factors so that it is essential that loss measurements should be made using precision wattmeters of the appropriate low power factor rating. For a shunt reactor full-load loss can only be supplied at rated voltage, so the test plant will be required to sup ply the full MVAr at this voltage.

Other tests are more straightforward and similar to the tests which would be carried out on a transformer, so that a full test series might consist of:

• winding resistance;

• oil samples;

• loss measurement;

• impedance measurement;

• zero phase-sequence impedance;

• noise level;

• temperature rise test;

• separate source voltage test, including measurement of partial discharge;

• impulse test;

• oil samples (repeat);

• insulation resistance;

• magnetic circuit and associated insulation applied-voltage test.


1. Copper Development Association (1996) Copper for Busbars. Copper Development Association, 5 Grovelands Business Center, Boundary Way, Hemel Hempstead, HP2 7TE, UK.

2. British Electricity International (1992) Vol D, 'Generator Main connections,' Section 4, pp. 287-324, 'Protection,' Section 11, pp. 868-947. Modern Power Station Practice, Third edition. Pergamon Press, Oxford.

3. Energy Networks Association (2001) Engineering Recommendation G5/4, Planning Levels for Harmonic Voltage Distortion and the Connection of Non-linear Equipment to Transmission Systems and Distribution Networks in the UK. Energy Networks Association, 18 Stanhope Place, Marble Arch, London W2 2HH.

4. Corbyn, D.B. (1972) 'This business of harmonics.' Electronics and Power, June, 219-223.

Top of Page

PREV.   NEXT   Guide Index HOME