Industrial Power Transformers -- Operation and maintenance [6]

Home | Articles | Forum | Glossary | Books


The subject of transformer protection falls naturally under two main headings.

These are:

• Protection of the transformer against the effects of faults occurring on any part of the system.

• Protection of the system against the effects of faults arising in the transformer.

Protection of the transformer against faults occurring in the system Considering firstly the means to be adopted for protecting the transformer itself against the effects of system faults, three distinct types of disturbances (apart from overloads) have to be provided for. These are:

(1) short circuits;

(2) high-voltage, high-frequency disturbances including lightning;

(3) pure ground faults.

To this list could be added ferroresonance, which can occur under certain conditions in any system containing capacitance and inductance elements such as those associated respectively with cables and transformers. The problem usually arises when some external system disturbance causes a transformer to go into saturation thus greatly changing its inductance. This may lead to excess voltages and currents arising on the system which can cause damage to transformers and other plant. Although certain protective equipment may operate under ferroresonance conditions, ferroresonance is not normally regarded as a 'fault' in the normal sense of the word, rather as a condition to be avoided by careful system design. The non-linearity of core steel is a property which exists and cannot be eliminated. Whilst the design of transformers to operate at low flux densities might reduce the likelihood of core saturation, this would lead to very uneconomic designs and it is generally considered that it would have very little effect on the conditions which can lead to ferroresonance.

Short circuits

System short circuits may occur across any two or even three lines, or, if the neutral point is solidly grounded, between any one line and the grounded neutral.

As pointed out in Section 6.5, the effect of a system short circuit is to produce overcurrents, the magnitude of which are dependent on the short-circuit level of the system feeding the fault, the voltage which has been short circuited, the impedance of the circuit up to the fault and the impedance of the fault itself.

The short-circuit currents produce very high mechanical stresses in the equipment through which they flow, these stresses being proportional to the square of the currents. The magnitude of these short-circuit currents can be limited by increasing the system impedance, usually incorporating this into the supply transformers. Unfortunately, increasing system impedance increases the regulation and this is not usually acceptable because of its effect on system performance and operation. On EHV and HV systems close control of sys tem voltage is required in order to control power and VAr flows. On HV and MV systems there are close statutory limits on voltage variation at consumers' supply terminals which are necessary to ensure that the consumers' equipment will function correctly, particularly the starting of motor drives. Although on the EHV and HV systems the transmission authorities are able to make use of on-load tapchangers on transformers and other devices such as VAr compensators to control system voltages, it is desirable from the transformer practical and economic viewpoint that the extent of tapping ranges is limited, for the reasons explained in Section 4, and on MV systems tappings are usually only selectable off-circuit, so that no means of continuous voltage control is available. Consequently the system designer is normally striving to achieve minimum regulation by keeping supply impedance as low as possible, limited only by the fault interruption capability of the available switchgear. Whereas some years ago the capability of the supply transformers to withstand the resulting short-circuit currents also provided an important constraint on selection of the system fault level, nowadays transformer manufacturers must be prepared to supply a transformer which is capable of withstanding whatever fault level the system designer decides is necessary, so that modern transformers designed to comply with the current issue of EN 60076 are capable of withstanding, without damage whilst in service, the electromagnetic forces arising under short-circuit conditions, as determined by the asymmetrical peak value of the current in the windings, which is normally taken as 2.55 times the short-circuit current.

In recent times the widespread adoption of solid-state 'soft-start' equipment for 415 V motor drives has generally reduced motor starting currents so that regulation of medium voltage systems may no longer be quite so critical to the system designer. This might enable the smaller distribution transformers pro viding 415 V supplies to have higher impedances and consequently lower short circuit withstand capability. In reality however, this is unlikely to have much impact on distribution transformer specifications and designs since once low impedance and a high level of short-circuit withstand strength has been shown to be possible this will tend to dictate accepted design practices and cost-savings resulting from a reduction in this will prove to be minimal.

High-voltage, high-frequency disturbances

High-voltage, high-frequency surges may arise in the system due to lightning, external flashover on overhead lines, switching operations and to the effects of atmospheric disturbances. These surges principally take the form of travel ling waves having high amplitudes and steep wavefronts, and often successive surges may follow rapidly upon one another. On account of their high amplitudes the surges, upon reaching the windings of a transformer pose a significant threat to the winding insulation. The effects of these surge voltages may be minimized by designing the windings to withstand the application of a specified surge test voltage and then ensuring that this test value is not exceeded in service by the provision of suitable surge protection installed adjacent to the transformer terminals.

All types of surge protection aim at attaining the same results, namely that of shunting surges from lines to ground or line to line to prevent their reaching the transformer. Protection may take the form of a rod gap, known as a co-ordinating gap, connected across the transformer bushings and designed to flash over at a given voltage level, or alternatively surge arresters may be used. Until quite recently surge arresters employed several spark gaps in series with a non-linear resistor material, normally silicon carbide, and, although this type is still used in significant quantities on rural distribution networks at 33 kV and below, elsewhere these have now been almost entirely superseded by the gapless metal oxide variety. The arresters are connected from each line to ground, or they may be occasionally connected from line to line. When a high-voltage surge reaches the arrester the metal oxide becomes conducting or the spark gaps break down and the disturbance is discharged through the device by reason of the fact that at the high-voltage involved the arrester resistance is low. As the surge voltage falls the arrester resistance automatically increases and prevents the flow of power current to ground or between lines. An arrester of this type is therefore entirely automatic in action and self extinguishing.

The internal surge impedance of a transformer winding is not a constant single valued quantity but has a range of values corresponding to the frequencies of the incident surge waveform. Changes in the surge impedance due to oscillation and decay of the surge voltages within the windings do not appreciably affect the terminal conditions. Moreover, the transformer terminal impedance is so high when compared with the line surge impedance, that its assigned value, so long as it is of the right order, has little influence upon the shape of the resulting surge waveform given. The wave diagrams in this section show the variation of voltage and current with time at the transformer terminals and not the phenomena occurring throughout the windings subsequent to the application of the surge waves to the terminals. Studies of the latter are given in Section 6.5.

Consider, first, what happens when rectangular finite voltage and current waves reach a transformer from an overhead line, there being no protective apparatus installed to intercept the disturbance. The amplitudes of the waves in the overhead line and at the transformer terminals depend upon the respective values of their surge impedance, which is given by the formula:

Z _ _(L/C) where Z is the surge impedance in ohms

L is the inductance in henrys

C is the capacitance in farads of the circuit concerned.

L and C may be taken for any convenient length of circuit.

When any travelling waves of voltage and current pass from a circuit of a certain surge impedance to a circuit of a different surge impedance, such waves in their passage to the second circuit undergo changes in amplitudes.

The oncoming incident waves when reaching the transition point between the two circuits are, if the surge impedances of the two circuits are different, split up into two portions, one being transmitted into the second circuit, and the other reflected into the first. The transmitted waves always have the same sign as the incident waves, but the reflected waves may have the same or opposite sign to the incident waves depending upon the ratio of the two surge impedances. This applies both to the voltage and the current waves.

If the incident waves are of finite length, the reflected waves travel back into the first circuit alone, and they are only transient waves in that circuit. If, on the other hand, the incident waves are of infinite length, the reflected waves in their passage backwards along the first circuit combine with the tails of the incident waves, so that the resultant waves in the first circuit are a combination of the two respective incident and reflected waves.

Given the amplitudes of the incident voltage and current waves and the surge impedances of the two circuits, the transmitted and reflected waves may be calculated by means of the formulae given in Table 5. The table gives formulae for determining the conditions when the incident waves are finite in length, and they are based on the assumption that no distortion of the shape occurs, due to losses in the circuit.

Table 5 Reflection of rectangular traveling waves at a transition point. No protective device

FIG. 79 Diagram showing incident, transmitted, and reflected voltage waves assuming an amplitude of 20 kV for the incident wave

FIG. 80 Diagram showing incident, transmitted and reflected current waves, assuming an amplitude of 57.1 A for the incident wave

Figures 79 and 80 show the incident, transmitted, and reflected voltage and current waves respectively, assuming the incident voltage wave to have an amplitude of 20 000 V, and the incident current wave an amplitude equal to 20 000/Zl

_ 57.1 A, as Zl

is assumed to equal 350 Ohm. The transmitted and reflected waves are constructed from the formulae given on the diagrams (distortion being ignored), and it will be seen that a voltage wave arrives at the transformer terminals having an amplitude considerably higher than that of the original incident wave. It is due to this sudden increase of voltage at the transformer terminals that so many failures of the insulation on the end turns of windings have occurred in the past, as the increased voltage may be concentrated, at the first instant, across the first few turns of the winding only, though ultimately voltage is distributed evenly throughout the whole winding.

The transmitted current wave is correspondingly smaller in amplitude than the incident current wave and, as such, is usually of no particular danger.

These diagrams show clearly that where the surge impedance of the second circuit is higher than that of the first, in comparison with the incident waves, the transmitted voltage wave is increased and the transmitted current wave is decreased, while the reflected waves have such signs and amplitudes as to sat isfy the equations Vt

_ Vi

_ Vr It

_ Ii

_ Ir

That is, both transmitted voltage and current waves are equal to the sum of the respective incident and reflected waves.

During the time corresponding to the lengths of the incident waves, the total voltage and current in the first circuit is equal to the sum of their respective incident and reflected waves, but after that period the reflected waves alone exist in the circuit.

From the preceding formulae and diagrams, when surges pass from one circuit to another the phenomena can be summarized as follows.

When a voltage wave passes from one circuit to another of higher surge impedance, both reflected and transmitted waves have the same sign as the incident wave, while the transmitted voltage wave is equal in amplitude to the sum of the incident and reflected waves. For the same circuit conditions the trans mitted current wave possesses the same sign as the incident current wave, but the reflected current wave has the opposite sign. The transmitted current wave is equal in amplitude to the sum of the incident and reflected waves.

At the transition point itself the total amplitudes of the voltage and current waves are always equal to the sum of the respective incident and reflected waves, bearing in mind, of course, the relative signs, and if the incident waves are infinite in length the same amplitudes extend throughout the whole of the circuit in which the incident waves arise. If, on the other hand, the incident waves are of finite lengths, the amplitudes of the resultant waves in the major part of the first circuit will be equal to the amplitudes of the reflected waves only.

The amplitudes of the transmitted and reflected waves are solely dependent on the ratio of the surge impedances of the two circuits.

The preceding notes have indicated that transmitted waves passing to a second circuit do so with rectangular fronts. That is, apart from any deformation of the waves which may occur due to losses in the circuits, the fronts of the transmitted waves are not modified in any way. If the second circuit is composed of inductive windings, such a rectangular fronted voltage wave represents a distinct danger to the insulation of the transformer. It becomes desirable, therefore, to modify the shape of the waveform from the steep rectangular form to a more gradual sloping one, and this can be achieved by the use of suitable surge diverters.

Surge protection of transformers [3]

[3. Reproduced by kind permission of the CEGB]

Modern practice of surge protection of transformers is aimed at preventing excessive voltage surges from reaching the transformer as a unit, that is, not only the HV and LV windings but also the bushings, where flashover and insulation breakdown will result in serious damage and system disconnections. In the UK surge protection is implemented by the addition of rod gaps or surge arresters adjacent to the transformer to shunt the surges to ground. These attenuate the surge magnitudes seen by the windings and their resulting insulation stresses to levels which can be withstood by suitably proportioned insulation distribution without causing resonant instability and dangerous oscillations within the windings.

Bushing flashover would generally protect the windings but this is not tolerable in practice for several reasons, notably the likelihood of damaging the bushing. The breakdown characteristic is most unfavorable and after initial breakdown to ground, via the bushing surface, tank and tank ground connection, the low impedance path to ground will allow a power-frequency current to flow if the system neutral is grounded or if two bushings flash over simultaneously.

This current will cause protective schemes to operate, leading to system disconnections even when the bushings are undamaged.

The desired characteristic is one where the path to ground presents a high impedance to normal supply frequency voltages but which falls to a low impedance under high-voltage transient conditions, followed by a rapid recovery to the original impedance levels as the voltage falls again.

The two methods commonly adopted to obtain surge protection are:

(a) co-ordinating rod gaps, and (b) surge arresters.

Both methods have advantages and disadvantages but are applicable on systems operating at voltages down to 3.3 kV which are reasonably insulated and where the cost of surge protection of these types can be justified for system reliability.

Screened co-ordinating rod gaps

Co-ordinating rod gaps have been fitted in the UK on bushings for many years in order to protect the windings of high-voltage transformers from external overvoltages. A disadvantage of co-ordinating rod gaps is the difference between the positive and negative polarity voltage protection levels, which in the case of 400 kV winding protection may be of the order of 300 kV for 100 µs wavefronts.

A further disadvantage is that for switching surges with wavefronts exceeding 100 µs there is a rapid increase in the operating voltage level with positive surges. On the assumption that a particular protective level at 100 µs wave front is adequate to protect the transformer winding, the 40 percent voltage withstand increase for wavefronts between 100 and 1000 µs for the positive wave could result in the withstand strength of the transformer winding being exceeded in the case of switching surges with long fronts. Such fronts may be encountered when switching transformer feeders where surges with wave fronts of several hundred microseconds have been measured during switching operations.

Screened co-ordinating gaps for fitment to 132, 275 and 400 kV transformer bushings have been developed by CERL whereby the polarity differential is considerably reduced such that the 99.7 percent probability of sparkover to wavefronts between (a) 1-10 µs, (b) 10-200 µs can be maintained at not greater than the specified (a) lightning, (b) switching surge withstand levels of the winding, and with a 95 percent probability of sparkover for wavefronts up to 500 µs.

The screened co-ordinating gap comprises, as in the standard co-ordinating gap, a horizontal gap between two rods or two loops, but the HV rod or loop is screened by a vertical toroid at right angles to the rod or loop and positioned so that the radiused rod or loop termination is in the same plane as the surface of the toroid facing the ground electrode. FIG. 81 illustrates the arrangement of a screened co-ordinating gap electrode for use on a 275 kV system.

FIG. 81 Screened co-ordinating gap electrode for use on a 275 kV system

Protection with surge arresters

When a travelling wave on a transmission line causes an insulator flashover any ground fault which may be established on the system after the surge has discharged may cause the relay protective gear to operate, disconnecting the line from the supply. To avoid this, the voltage flashover level of the line can be increased by means of larger insulators but this can only be done within limits because the higher the line insulator flashover voltage then the higher the value of the impulse wave transmitted to the transformer. Therefore some compromise has to be made between the risk of a line flashover and possible damage to equipment and thus it is necessary to know the impulse voltage strength of the transformers to be protected. This has led to the system of insulation co-ordination as set out in EN 60071 Guide for insulation co-ordination.

This is a two-part document; Part 1 covers terms, definitions, principles and rules; Part 2 is the application guide.

Quoting from the 1996 issue of Part 1: 'Insulation co-ordination (comprises) the selection of the dielectric strength of equipment in relation to the volt ages which can appear on the system for which the equipment is intended and taking into account the service environment and the characteristics of avail able protective devices.' Determination of electric strength is carried out by means of dielectric tests for which the test levels are related to the Highest voltage for equipment, Um, defined as 'The highest r.m.s. phase-to-phase volt age for which the equipment is designed in respect of its insulation as well as other characteristics which relate to this voltage in the relevant equipment standards.' For power transformers the relevant standard is, of course, EN 60076.

The response of protective devices to system overvoltages is very much dependent on the characteristics of those overvoltages, so EN 60071, Part 1, divides these into four classes as set out below:

(a) Continuous (power frequency) voltage: considered as having a constant r.m.s. value, continuously applied to any pair of terminals of the equipment.

(b) Temporary overvoltage: a power frequency overvoltage of relatively long duration which may be undamped or weakly damped and its frequency may be several times lower or higher than the system frequency.

(c) Transient overvoltage: which are further subdivided into slow-front, fast front, very fast-front and combined overvoltages. Slow is defined as having a time to peak of 20 µs, fast 0.1 µs, and very fast _0.1 µs, all with proportionately corresponding tails.

(d) Combined, consisting of two or more components applied simultaneously.

The standard then subdivides values of highest voltage for equipment into two ranges as follows:

Range I: Above 1 kV up to and including 245 kV and covering transmission and distribution systems.

Range II: Above 245 kV which covers mainly transmission systems.

Recommended test levels are then related to the value of highest voltage for equipment and the overvoltage class within which the equipment is required to operate. The protective device, surge arrester or spark gap, must not operate for temporary overvoltages such as, for example those due to system ground faults which can lead to voltages to ground on the non-faulted terminals of up to _3 times normal until the ground fault is cleared.

Selection of the appropriate power-frequency withstand level is related to the ground fault factor for the system and the plant location within that system.

Ground fault factor is defined as 'the ratio of the highest r.m.s. phase-to-ground power-frequency voltage on a sound phase during a fault to ground (affecting one or more phases at any point) to the r.m.s. phase-to-ground power frequency voltage which would be obtained at the location without the fault.' The ground fault factor is related to the grounding conditions of the system as viewed from the selected location. It is equal to the product of _3 times the 'factor of grounding' or 'coefficient of grounding' which has been used in the past.

The ground fault factor can be calculated from the symmetrical component parameters of the system R0, X0, and X1, where R0 is the zero phase sequence resistance; X0 is the zero phase sequence reactance; X1 is the positive phase sequence reactance subtransient reactance values being used for any rotating machines


For a solidly grounded system for which R X

0 1 1

and X X

0 1 3 the ground fault factor will not exceed 1.4.


For a non-effectively grounded system for which R X

0 1 1

and X X

0 1 3

the ground fault factor will be _3, that is 1.7.

(iii) For a system grounded via an arc suppression coil the ground fault factor will be 1.9.

The higher the ground fault factor, the higher should be the rated power-frequency 1 minute withstand voltage. Lightning impulse withstand voltage is tied to power-frequency withstand voltage so that once the latter has been determined the former is automatically fixed.

However, there can be situations where expected surge voltages might be higher in proportion to power-frequency overvoltages than might normally be expected for example equipment connected to wood-pole overhead lines with ungrounded cross-arms. Conversely, there will also be situations where the magnitude of lightning surges will be likely to be lower. These include substations connected to overhead lines which are protected by protective ground wires or to overhead lines connected via at least 1 km of cable having a grounded metallic sheath.

For voltages above 245 kV the ability to withstand switching surges is not adequately tested by the power-frequency overvoltage test, so it is necessary to carry out a separate switching surge test.

By reference to the tables of dielectric test levels set out in Section 5.2, it will be noted that there are a number of options for each voltage class. These arise because of the application of insulation co-ordination principles to the various situations in which the equipment can be installed. Particularly for equipment in the higher-voltage range the economic savings to be gained from a careful matching of the insulation level to the degree of exposure to surges is most apparent. It will be evident, for example, that a 420 kV transformer designed for lightning impulse testing at 1300 kV, a figure generally specified by European purchasers where protection by surge arresters is to be employed, will be less costly than a similar unit but designed for a 1425 kV lightning impulse level as specified in the UK where protection is to be by co-ordinating rod gaps.

Whilst the use of screened co-ordinating gaps on the transformer bushing can provide an acceptable level of protection for transformers in countries which have a relatively low incidence of lightning, their use in locations where lightning is very frequent would not afford the degree of protection attainable by the use of surge arresters. For many system designers, therefore, the use of surge arresters is regarded as the standard practice.

Part 2 of the application guide, EN 60071, expands considerably on this theme. It is a very comprehensive and readable document covering a complex subject for which attempting further detailed summary is beyond the scope of this guide. For those involved in system design and plant selection, or simply wishing to have further information on the subject, reference to the document is recommended.

Selection of surge arrester protective characteristics

Some guidance on selection of surge arresters is given in EN 60071, Part 2, however there are a number of documents which deal specifically with surge arrester characteristics and selection. For clarity these are listed below, all are parts of the EN 60099 (IEC 99) series:

Part 1: Non-linear resistor type gapped surge arresters for AC systems

Part 4: Metal oxide surge arresters without gaps for AC systems

Part 5: Selection and application recommendations

Part 3: Artificial pollution testing of surge arresters is not likely to be of interest to transformer engineers.

These documents are lengthy and very comprehensive and in a work such as this it is only possible to give an outline of the general principles involved.

Prospective users are therefore advised to consult the appropriate standard specification before specifying surge arresters.

Selection of gapped silicon carbide arresters

As indicated above, the superior performance of metal oxide arresters has meant that the use of silicon carbide non-linear resistor type surge arresters with series spark gaps nowadays tends to be restricted to equipment to the lower end of voltage in Range I, say, up to about 72.5 kV. The main risk for transformers in such systems arises from induced and direct lightning strokes to the connected overhead line. In the case of cable connected transformers only overvoltages due to faults and switching surges can occur. The basic characteristics of this type of arrester are their rated voltage, their sparkover voltages for lightning and switching surges, their nominal discharge currents and their residual voltages at these currents. Also relevant is the continuous operating voltage, long duration discharge class, pressure relief class and pollution withstand capability.

Rated voltage must be adequate to withstand temporary overvoltages resulting from ground faults on one phase causing a voltage rise on a healthy phase at a time when the arrester may be called upon to operate on this healthy phase.

EN 60099, Part 5, recommends an iterative procedure for the selection of surge arresters as shown in the diagram FIG. 82. This is as follows:

• Determine the necessary continuous operating voltage of the arrester with respect to the highest system operating voltage.

FIG. 82 Flow diagram for the selection of surge arresters

• Determine the necessary value for the rated voltage of the arrester bearing in mind the temporary overvoltages as identified above.

• Estimate the magnitude and probability of the expected lightning discharge currents through the arrester (see below) and the transmission line discharge requirements and select nominal discharge current, the high current impulse value and the line discharge class of the arrester.

• Select the pressure relief class of the arrester with respect to the expected fault current.

• Select a surge arrester which fulfils the above requirements.

• Determine the necessary rated lightning and switching impulse withstand voltages of the transformer taking into account the switching overvoltages (not necessary to consider switching overvoltages below 72.5 kV).

• Determine the necessary rated lightning impulse voltage considering

- the representative impinging lightning overvoltage surge as it is determined by the lightning performance of the overhead line,

- the substation layout,

- the distance between the surge arrester and transformer,

- the rated insulation level of the transformer in accordance with IEC 71-1.

Determination of lightning discharge currents

Generally arrester currents resulting from lightning strokes are lower than the current in the stroke itself. If a line is struck directly, travelling waves are set up in both directions from the point of the strike. Line insulators will probably flashover thus providing parallel paths to ground. More than one conductor may be struck so that two or more surge arresters operate to share the cur rent. Only in the case of a direct strike close to the terminal arrester where no flashover occurs before operation of the arrester will the arrester be required to carry the full current of the lightning discharge. The probability of this occur ring can be greatly reduced by the use of shielding for the line and/or the sub station of the type described above. When a lightning strike occurs to a shield wire the current flows to ground through the line support structures, poles or towers, which carry the shield wire. This will raise the potential of the top of the support structure which may result in a flashover to the line. This is known as a back-flashover. The incidence of back-flashovers can be reduced by correct selection of the insulation level for the line and ensuring that the resistance of the support structure ground connections is kept to a minimum.

Bearing in mind the above factors, the importance of the transformer and degree of protection required, EN 60099, Part 5, recommends that 5 kA arresters are normally adequate for distribution systems of voltages up to 72.5 kV. 10 kA arresters should be used for the higher voltages in Range I and may also be used below 72.5 kV for important installations or in areas of high lightning incidence.

Selection of metal oxide arresters

Metal oxide arresters are now the preferred means of protection for systems having voltages in Range II but may, of course, also be used for voltages in Range I. Their main advantage is the lack of series gaps which means that they have a very much faster response time. For example, a lightning strike within one span of a terminal tower can result in a rate of rise exceeding 1200 kV/µs.

With a gapped arrester a delay due to the gaps of, say, 0.5 µs will result in a peak voltage of 600 kV being 'let through' to the transformer. The response time of the gapless arrester is of the order of 10-15 ns and therefore no such problem exists. Basic characteristics of this type of arrester are the continuous operating voltage, the rated voltage, the nominal discharge current and the residual volt ages at nominal discharge current, at switching impulse current and steep front current. For given continuous operating and rated voltages different types of arresters and, therefore, different protection levels exist. Careful attention must be given to ascertaining the correct continuous operating voltage for this type of arrester in view of the high degree of current/voltage non-linearity of the material and the fact that this has a negative temperature coefficient of resistance up to voltages in the order of 10 percent above rated voltage which could lead to thermal runaway if the continuous operating voltage and/or the rated voltage were underestimated. In selection of arresters of this type it is also necessary to consider requirements with regard to line discharge class, pressure relief class and pollution withstand capability. The selection procedure is thus similar to that described for gapped arresters. As regards current rating; for voltages up to 72.5 kV, 5 or 10 kA rated arresters may be used according to the same criteria as discussed above for gapped arresters; for higher voltages in Range I, 10 kA arresters are normally adequate, and for voltage Range II 10 kA arresters are considered adequate up to 420 kV, thereafter 20 kA arresters should be used.

Connection of surge arresters

Mention has already been made of the desirability of connecting the surge arrester as close as possible to the equipment it is intended to protect. It is also important that the neutral end should have a low resistance connection to ground of adequate cross-sectional area. At currents of 10 kA or more, any resistance in these connections will mean a significant additional voltage is seen at the transformer terminals over and above the nominal protection level of the arrester.

FIG. 83(a) shows a typical installation adjacent the HV terminal of a substation transformer. The lengths d, d1 and d2 must be kept as short as possible. In the case of a small distribution transformer, possibly pole mounted, where there is no ground mat, the installation should take the form shown in FIG. 83(b).

FIG. 83 Schematic diagram for the surge arrester connection to the protected object.

Surge arrester operation and construction

(a) Gapped silicon carbide surge arresters A typical gapped surge arrester consists of a series of non-linear resistor discs separated by spark gaps. The resistor unit is made from a material having a silicon carbide base in a clay bond. The size of grit used, method of mixing, moisture content and ? ring temperature play important parts in determining the characteristics of the final product and all are very carefully controlled. The finished material is a semi conductor, and its voltage/current characteristic is a curve with a pronounced knee-point beyond which the curve is fairly linear typically as shown in the curve, FIG. 84. The material has a negative temperature coefficient of resistance. At constant voltage the current increases by about 0.6 percent/ºC. To maintain equilibrium the dissipation of heat must be adequate. Its specific heat is 0.84 Ws/g/ºC between 20 and 300ºC.

FIG. 84 Typical surge diverter voltage/current characteristic for gapped silicon carbide surge arrester

FIG. 85 shows a cross-sectional drawing of a typical gapped silicon car bide type station-class surge arrester. The ROTARC gaps provide the means of preventing power follow-through current following diversion of the surge.

The extra 'space' to accommodate the necessary quantity of spark gaps means that an efficient, fast acting pressure relief system must be used to avoid the undesired explosion of the porcelain housing, due to the passage of system short-circuit current, following arrester failure. The presence of moisture, at this time, due to housing seal failure means that even higher internal pressures can be developed during the passage of system short-circuit current. This is of course due to the considerable expansion of water when changing to steam.

In a gapped silicon carbide surge arrester the spark gap serves the purpose of interrupting the system follow-through current after a surge has discharged.

The multi-gap design is chosen because it is more sensitive to voltage variation and more effective in arc quenching. The electrodes are symmetrical, giving many of the advantages of a sphere gap, which is independent of polarity effect and quick in response to breakdown voltage. Uniform voltage grading across the gaps is attained either by direct resistance grading or by equalizing the capacitance values between the gaps. In a typical distribution class surge arrester, specially selected and graded mica discs are inserted between each pair of electrodes and, as they are very thin and have a high dielectric constant, the capacitance is relatively high, hence the stray capacitance to ground is relatively small. This further improves the uniformity of voltage grading across the gaps. The mica used has a small loss angle, almost independent of frequency, and the change in dielectric constant when subjected to impulse voltages is therefore negligible. The thin outer edge of each mica disc points towards the gap and under impulse conditions the concentration of charge around this mica is high, so that the resulting displacement current provides electrons which irradiate the gap, thus alleviating the darkness effect and improving the consistency of flashover.

FIG. 85 Cross-sectional drawing of a typical station-class surge arrester (Bowthorpe EMP Ltd.)

Under impulse conditions the arrester gaps are required to flash over at the lowest possible voltage and to interrupt the system follow-through current at the first current zero after the surge has discharged. These two requirements are somewhat conflicting since both the impulse voltage breakdown and the current interrupting characteristics of the gaps increase with the number of gaps in series. Also the more non-linear discs employed in series the smaller is the system follow-through current and thus fewer gaps are required to interrupt this current. In contrast to this principle there is the fact that the surge voltage across the arrester stack will be higher under surge conditions. Thus, some compromise must be made; if too many discs or gaps are incorporated in the arrester then adequate protection may not be provided, while too few may result in destruction during operation under surge discharge conditions.

(b) Metal oxide surge arresters A typical metal oxide surge arrester consists of a number of non-linear resistor discs arranged in series. The discs are sometimes spaced by metallic spacers in order to make the non-linear resistor assembly of comparable height to the required housing length. The housing may be porcelain or more recently a polymeric type material. The gapless metal oxide surge arrester has fewer components and is therefore of much simpler construction than the older gapped silicon carbide type.

It is extremely important to select the correct rated voltage, due to the highly non-linear characteristic (see FIG. 86). For example, at voltages in the order of rated voltage a 10 percent increase in voltage can cause a 300 percent increase in current.

FIG. 86 Current/voltage characteristic for gapless metal oxide surge arrester (Bowthorpe EMP Ltd).

Note: This curve is constructed from measurements taken with various forms of voltage/current stimulation, that is 2 ms current pulses, 8/20 ms current pulses and 4/10 ms current pulses

FIG. 87 Gapless metal oxide arrester in porcelain housing (Bowthorpe EMP Ltd.)

To avoid premature failure, the selected rated voltage must take into account all temporary overvoltages existing at the site where the surge arrester is installed:

(i) Porcelain housed gapless metal oxide surge arresters

FIG. 87 shows a sectional arrangement of a gapless metal oxide surge arrester in a porcelain housing. It is evident that this is very much simpler than that shown in FIG. 85. The peripheral air space, necessary for pressure-relief operation can be a source of internal ionization under polluted conditions which can cause premature failure of the surge arrester.

(ii) Polymeric housed gapless metal oxide surge arresters

FIG. 88 Cross-section of EG/HE Series gapless metal oxide arrester detailing the major features of arrester design (Bowthorpe EMP Ltd.)

FIG. 89 Installation of gapless metal oxide surge arresters with polymeric housings to protect 11 kV distribution transformer (Bowthorpe EMP Ltd.)

FIG. 88 shows a sectional arrangement of a gapless metal oxide arrester in a polymeric housing. The use of the gapless construction makes possible another major step forward in the design of surge arresters.

This is the use of polymeric housings in place of porcelain. This material not only permits a construction totally free of internal voids, which is thus impervious to moisture ingress, but also avoids the tendency to shatter associated with porcelain, hence eliminating the risk of disintegration under fault conditions and making unnecessary the incorporation of any pressure relief arrangement. The polymeric material is, of course, less expensive than porcelain and also very much lighter in weight. For example a 10 kA, 150 kV rated voltage, arrester weighs only around 85 kg. This greatly simplifies the installation arrangement as well as easing site handling at the time of installation. An installation of gapless arresters in polymeric housings protecting a 11 kV pole mounted distribution transformer is shown in FIG. 89 whilst FIG. 90 shows a set of 300 kV 20 kA polymeric housed arresters providing the protection for the transformers of a large EHV substation.

FIG. 90 300 kV polymeric housed surge arresters installed on the Statnet system in Norway (Bowthorpe EMP Ltd.)

Ground faults

Ground faults have different effects according to whether the neutral point of the system is grounded or isolated. In the first case, a ground fault represents a short circuit across one phase, and the same remarks regarding protection apply as outlined for the protection against short-circuit stresses. In the other case, where the neutral point is isolated, there are two conditions to consider:

first, when the ground is a sustained one, and second, when it takes the form of the so-called arcing ground. In the first of these two cases the voltage of the two sound lines is raised to full line voltage above ground, and after the initial surge the insulation stresses become steady, although increased by _3 above normal service conditions. The protection for the first condition is grounding of the neutral point, as explained elsewhere. In the second case, where the ground fault is unstable, such as at the breakdown of an overhead line insulator, high frequency waves are propagated along the line in both directions, and to protect the transformer against the effects of these waves some form of surge arrester gear may be installed in front of the transformer as outlined earlier in this section. Alternatively the neutral point may be grounded, thereby converting the ground fault into a short circuit across one phase.

Protecting the system against faults in the transformer

Consider next the means to be adopted for protecting the system against the effects of faults arising in the transformer; the principal faults which occur are breakdowns to ground either of the windings or terminals, faults between phases generally on the HV side, and short circuits between turns, usually of the HV windings.

The protection of transformers, in common with the protection of other electrical plant, is an area in which there has been a great deal of change in recent years, brought about by the development of digital solid-state relays.

The use of microelectronics makes possible the provision of high-reliability, rugged, compact and inexpensive relays having accurate tailor-made characteristics to suit almost any situation, so that the bulky and delicate electro magnetic devices on which protective gear has relied for so many years are becoming consigned to history.

However, the principles and objectives of transformer protection have not changed. It is simply the case that the protection engineer now has available relays which come much closer to meeting all his requirements and they will do so at a price which enables a degree of sophistication to be applied to the protection of a 500 kVA transformer which might hitherto have only been considered economically justified for one rated 30 MVA or more.

There are many thousands of electromagnetic relays in service and their life and reliability is such that they will continue to be so for a good many years.

The following description of transformer protection principles will therefore consider initially those 'traditional' schemes based on electromagnetic relays before considering briefly how these have been developed to make use of the latest technology.

Breakdowns to ground may occur due to failure of the major insulation of transformers or of bushing insulators, these failures being due to the absence of any external surge protective apparatus or upon the failure of such apparatus to operate. When such a breakdown occurs it is essential that the transformer is isolated from the supply with as little delay as possible.

For small transformers, single overload and ground leakage devices will pro vide the necessary degree of protection to ensure that the transformer is disconnected automatically from the circuit.

On larger transformers forming parts of important transmission or distribution networks, it is necessary to employ some form of automatic discriminative protective equipment. This will remove from the circuit only the faulty apparatus leaving the sound apparatus intact, while the disconnection is per formed in the shortest space of time and the resulting disturbance to the sys tem is reduced to a minimum. The automatic protective gear systems which are most commonly used are described in the following sections.

Comprehensive details of various forms of protective systems for generators, generator/transformer combinations, transformers, feeders and busbars are given in The J & P Switchgear Book (Butterworths).

FIG. 91 Differential rough balance protection scheme (GEC Measurements)

In considering the problems of protection across a power transformer, note should first be made of what is known as a differential rough balance scheme, as shown in FIG. 91. This scheme can be applied where existing overcurrent and restricted ground-fault protection has become inadequate but provision of a separate differential scheme is considered unjustified. By using overcurrent relays and current transformers, lower fault settings and faster operating times can be obtained for internal faults, with the necessary discrimination under external fault conditions.

The taps on interposing transformers are adjusted so that an inherent out-of balance exists between the secondary currents of the two sets of current transformers but is insufficient to operate the overcurrent relays under normal load conditions. With an overcurrent or external fault the out-of-balance current increases to operate the relay and choice of time and current settings of the IDMT relays permit grading with the rest of the system.

The scheme functions as a normal differential system for internal faults but is not as fast or sensitive as the more conventional schemes. It is, however, an inexpensive method of providing differential protection where IDMT relays already exist.


Top of Page

PREV.   NEXT   Guide Index HOME