Switching in-rush currents
It is often noticed when switching in a transformer on no-load that the ammeter registers an initial current rush (which, however, rapidly dies down) greatly in excess of the normal no-load current and sometimes even greater than the normal full-load current of the transformer. In the latter case, it may seem at
first glance that there is a fault in the transformer. Upon considering the problem fully, however, and bearing in mind the characteristics of iron-cored apparatus, the true explanation of the transient current rush will become clear. The initial value of the current taken on no-load by the transformer at the instant of switching in is principally determined by the point of the voltage wave at which switching in occurs, but it is also partly dependent on the magnitude and polarity of the residual magnetism which may be left in the core after previously switching out. There are six limiting conditions to consider, namely:
(a) Switching in at zero voltage - no residual magnetism.
(b) Switching in at zero voltage - with maximum residual magnetism having a polarity opposite to that to which the flux would normally attain under equivalent normal voltage conditions.
(c) Switching in at zero voltage - with maximum residual magnetism having the same polarity as that to which the flux would normally attain under equivalent normal voltage conditions.
(d) Switching in at maximum voltage - no residual magnetism.
(e) Switching in at maximum voltage - with maximum residual magnetism having a polarity opposite to that to which the flux would normally attain under equivalent normal voltage conditions.
(f ) Switching in at maximum voltage - with maximum residual magnetism having the same polarity as that to which the flux would normally attain under equivalent normal voltage conditions.
(a) Switching in at zero voltage: no residual magnetism.
Under normal conditions the magnetic flux in the core, being 90º out of phase with the voltage, reaches its peak value when the voltage passes through zero.
Due to this phase displacement it is necessary for the flux to vary from a maximum in one direction to a maximum in the opposite direction in order to pro duce one-half cycle of the required back e.m.f. in the primary winding, so that a total flux is embraced during the half cycle corresponding to twice the maximum flux density.
At the instant of switching in, there being no residual magnetism in the core, the flux must start from zero, and to maintain the first half cycle of the voltage wave it must reach a value corresponding approximately to twice the normal maximum flux density.
FIG. 63 Transient flux density conditions when switching in a transformer
at the instant V _ 0. No residual magnetism. Dotted lines represent the normal
steady flux density Bn and the transient component B.
This condition, together with the succeeding voltage and flux density waves, is shown in FIG. 63 and it will be seen that the rate of change of flux (upon which the magnitudes of the induced voltages depend) is nearly the same, throughout each cycle, as the normal flux density which is symmetrically placed with regard to the zero axis and which corresponds to the steady working conditions. The maximum values of the flux density, upon which the magnitude of the no-load current depends, vary gradually from a figure initially approaching twice the normal peak value in one direction only, down to the normal peak value disposed symmetrically on each side of the zero axis.
As the magnitude of the no-load current is dependent upon the flux density, it follows that the current waves also will initially be unsymmetrical, and that they will gradually settle down to the steady conditions. While, however, in the case of the flux density the transient value cannot exceed twice the normal, the transient current reaches a value very many times the normal no-load cur rent and can exceed the full-load current.
The reason for this current in-rush is to be found in the characteristic shape of the B/H curve of transformer core steel, which is shown in FIG. 64, and from this it will be seen that the no-load current at twice the normal flux density is increased out of all proportion as compared with the current under steady conditions.
FIG. 64 Typical B/H curve showing relationship between maximum flux density
and no-load current.
FIG. 65 illustrates this current in-rush phenomenon, and the total current may be considered to consist of the normal no-load current and a drooping characteristic transient current superimposed upon it. Due to the initial high saturation in the core, the current waves may be extremely peaked and contain prominent third harmonics.
In practice, the transient flux does not actually reach a value corresponding to twice the normal flux density, as the voltage drop, due to the heavy current in-rush flowing through the resistance of the entire primary circuit during the flux variation from zero to twice the maximum, is greater than the drop occur ring with the normal flux distribution. Consequently a somewhat smaller back e.m.f. is to be generated by the varying flux, so that the latter does not reach a value corresponding to 2Bmax, but remains below this figure, the more so the higher the resistance of the primary circuit.
(b) Switching in at zero voltage: with maximum residual magnetism having a polarity opposite to that to which the flux would normally attain under equivalent normal voltage conditions.
If there is residual magnetism in the core at the instant of switching in and the residual magnetism possesses an opposite polarity to that which the varying flux would normally have, the phenomena described under (a) will be accentuated. That is, instead of the flux wave starting at zero it will start at a value corresponding to the polarity and magnitude of the residual magnetism in the core, and in the first cycle the flux will reach a maximum higher than out lined in (a) by the amount of residual magnetism. The theoretical limit is a flux which corresponds to a value approaching 3 times the normal maximum flux density, and at this value the initial current in-rush will be still greater.
FIG. 66 illustrates the resulting transient flux/time distribution, while the current in-rush will be similar to that shown in FIG. 65, except that the maximum values will be much higher and the in-rush current will take a longer time to reach steady conditions. In this case also the drop in voltage, due to the resistance of the primary circuit, operates to reduce the maximum flux density and consequently the current in-rush, but to a greater extent than in the case of (a).
(c) Switching in at zero voltage: with maximum residual magnetism having the same polarity as that to which the flux would normally attain under equivalent normal voltage conditions.
FIG. 65 Typical transient current in-rush when switching in a transformer
at the instant V= 0. in _ normal no-load current, it switching current in-rush.
FIG. 66 Transient flux density conditions when switching in a transformer
at the instant V = 0 and with residual magnetism in the core and opposite to
the normal flux density. Dotted lines represent the normal steady flux Bn and
the transient component B.
FIG. 67 Flux density conditions when switching in a transformer at the
instant V = 0 and with residual magnetism in the core equal to and of the same
polarity as the normal flux density. No transient condition.
The converse of (b), where the residual magnetism possesses the same polarity as that which the changing flux would normally attain, results in a diminution of the initial maximum values of the flux, and consequently of the current in-rush.
If the value of the residual magnetism corresponds to maximum flux density the flux will follow its normal course and the no-load current in-rush will be avoided. FIG. 67 illustrates the flux/time distribution. If, however, the residual magnetism corresponds to a flux density lower than the maximum, the initial flux waves are unsymmetrically disposed about the zero axis, the more so the lower the value of the residual magnetism. FIG. 68 illustrates this, and a current in-rush occurs according to the maximum value of the flux.
(d) Switching in at maximum voltage no residual magnetism
FIG. 68 Flux density conditions when switching in a transformer at the
instant V = 0 and with residual magnetism in the core equal to half the normal
flux density and of the same polarity as the normal flux density. Dotted lines
represent the normal steady flux Bn and the transient component B
FIG. 69 Transient flux density conditions when switching in a transformer
at V _ Vmax and with residual magnetism in the core equal to but in the opposite
direction to the normally increasing flux density. Dotted lines represent the
normal steady flux Bn and the transient component Bt
FIG. 70 Transient flux distribution conditions when switching in a transformer
at V _ Vmax and with residual magnetism in the core equal to and in the same
direction as the normally increasing flux density. Dotted lines represent the
normal steady flux Bn and the transient component Bt
In this case at the instant of switching in, the flux should be zero, due to its 90º phase displacement from the voltage, and as we have assumed there is no residual magnetism in the core, the desired conditions are obtained which pro duce the normal steady time distribution of the flux. That is, at the instant of switching in the flux starts from zero, rises to the normal maximum in one direction, falls to zero, rises to the normal maximum in the opposite direction and again reaches zero, the wave being symmetrically disposed about the zero axis. The no-load current, therefore, pursues its normal course and does not exceed the magnitude of the normal no-load current.
(e) Switching in at maximum voltage: with maximum residual magnetism having a polarity opposite to that to which the flux would normally attain under equivalent normal voltage conditions In this case the residual magnetism introduces the transient components, so that the initial flux waves are unsymmetrically disposed about the zero axis, high initial maximum flux values are attained, and in the case where the residual magnetism has the same value as corresponds to the normal maximum flux density the current in-rush will have a value corresponding approximately to twice the normal maximum flux density. This is shown in FIG. 69.
(f ) Switching in at maximum voltage: with maximum residual magnetism having the same polarity as that to which the flux would normally attain under equivalent normal voltage conditions This is the converse of the foregoing case, and the initial flux waves will again be unsymmetrically disposed about the zero axis. For the same value of residual magnetism the total maximum flux would be the same as in case (e), but both flux and current waves would initially be disposed on the opposite side of the zero axis. FIG. 70 illustrates this case.
FIG. 71 Flux space and time distribution at 30-degree intervals in a three-phase
core-type transformer with star/star-connected windings
FIG. 72 Transient flux density conditions when switching in a three-phase
core-type transformer with star/star-connected windings at the instant Va =0;
Vb = 0.866 Vmax and Vc = 0.866 Vmax and with residual magnetism in phase A equal
to twice the normal flux density and in phases B and C equal to half the normal
flux density. Dotted lines represent the transient components BAt , BBt and BCt.
The foregoing remarks are strictly applicable to single-phase transformers operating as such, but the principles can also be applied to polyphase transformers or banks so long as one considers the normal magnetic relationship between the different phases. That is, curves have been given which relate to one phase only, but the principles apply equally well to polyphase transformers, providing each phase is treated in conjunction with the remaining ones.
A single instance will suffice to show what is meant, and for this purpose consider a three-phase core-type star/star-connected transformer switched in under the same conditions as (b) which is illustrated in FIG. 66. FIG. 71 shows the normal main flux space and time distribution at intervals of 30º, and the number of lines in the cores indicate the relative flux density in each. Due to the usual three-phase relationship, the transformer can only be switched in when any one phase, say A, is at zero voltage, so that the remaining two phases B and C will each give a voltage at the instant of switching in equal to 86.6 percent of the maximum of each phase, one being positive and the other negative. Similarly, if the transformer has previously been switched out so that the residual magnetism in phase A of the core has a value corresponding to the maximum flux density and a polarity opposite to that which the flux would normally attain to under equivalent normal voltage conditions, the residual magnetism in each phases B and C will have a value corresponding to half the normal flux density in each phase, and a polarity opposite to the residual magnetism in phase A. The current in-rushes in the three-phases will therefore not be equal, but they will be modified by the flux conditions, which are shown in FIG. 72. It is only necessary to refer to the B/H curve and hysteresis loop to obtain an approximation of the current value corresponding to each value of flux density.
The flux waves have been drawn sinusoidal in order to present the illustrations in the clearest manner, but the actual shape of the flux and current waves will be determined by the connections of the transformer windings and the type of magnetic circuit. In a three-phase core-type transformer with star/star connected windings the normal flux wave may contain small third harmonics, and may therefore be flat topped, while the no-load current will be a sine wave. With a delta-connected winding on either primary or secondary side the normal flux wave will be sine shaped, while the no-load current may contain third harmonics and be peaked.
FIG. 73 Construction for determining the non-sinusoidal waveform of flux
density with a sine wave of no-load current.
Note: induced voltage wave determined by differentiation of the flux wave.
FIG. 74 Construction for determining the non-sinusoidal waveform of no-load
current with sine waves of flux and induced voltage
Figures 73 and 74 show the method of obtaining the non-sinusoidal wave shapes of flux and no-load current from the hysteresis loop of the core material when the no-load current and flux, respectively are sine waves. In the initial transient stages the saturation of the cores will accentuate the higher harmonics so that in-rush current will have much higher peak values than can be deduced from ammeter readings. The B/H curve and hysteresis loops at various maximum flux densities would have to be available if more accurate theoretical determinations of the current values were to be made, but even then further difficulties would arise from the unequal form factors of the current waves in the three-phases. These would be particularly marked in star/star connected transformers and, due to the tendency to balance out the currents, the neutral point would be deflected and the phase voltages unbalanced.
It is also worthy of note that transformers with butt-type yokes do not retain residual magnetism so much as if the yokes and cores are interleaved.
In-rush current may therefore be less with butt yoke transformers, though the disadvantages of this form of construction for ordinary power transformers far outweigh any advantage which might be gained in respect of a minimized cur rent in-rush.
In passing it might be mentioned that these heavy current in-rushes were not experienced in the early days of transformer design on account of the relatively low flux densities which were then employed. The loss characteristic of transformer steel has improved considerably so that much higher flux densities are now utilized and the prevalence of heavy current in-rushes with modern transformers is due to this. These in-rushes are higher the lower the frequency for which the transformer is designed, as the lower the frequency the higher can the flux density be, which will still keep the iron loss to a reasonable figure.
For ordinary power transformers it has been suggested that residual magnetism may be greatly minimized if the load on the transformer is switched off before the primary circuit is opened. In this case when the transformer is finally switched out of circuit the only current flowing will be the normal no-load current which will be lagging behind the applied voltage by an angle usually between 70º and 90º. As it is generally found that a circuit breaker opens a circuit at or near zero current, this will correspond to a point at or near the maximum point on the voltage wave so that the flux in the core will be nearly zero. If the transformer is switched out of circuit on load, zero current will, in the case of a non-inductive load, correspond nearly to zero voltage, so that the residual magnetism left in the core would be a maximum; but the more inductive the load, the less likelihood is there of switching out at zero voltage.
Theoretically the residual magnetism may almost be eliminated by gradually reducing the applied voltage before switching the transformer out of circuit, while a further possible method would be to provide some kind of contact mechanism which would ensure the switch opening the circuit at maximum voltage. It should be remembered, however, that in any case with polyphase transformers it is not possible to switch out all phases at the maximum volt age, and consequently this last method would, at the best, only result in zero magnetism in one phase and something between zero and the maximum in the remaining phases. Both of the last two methods are objectionable, however, in so far as they involve additional equipment.
It has also been suggested at various times that switching in current in-rushes may be minimized by slowly closing the switch in the exciting circuit.
The idea underlying this suggestion is that as the switch contacts approach one another, a point will be reached just prior to actual closing at which a spark will bridge the contacts of one phase at the maximum peak voltage. At this instant the normally varying flux will have zero value and consequently current in-rushes will be avoided. This effect could, of course, only occur with one phase, and consequently the method could theoretically only be applied with perfect success to single-phase transformers. In the case of polyphase transformers, the remaining phases will each have a voltage at some value between zero and the maximum, and therefore abnormal flux distribution occurs in these phases, which will produce current in-rushes similar to those previously described. When considering this method, however, the fact should not be lost sight of that arcing at the switch contacts is liable to produce high-frequency voltage oscillations, the more so the more slowly the switch is closed. As this type of disturbance is generally very liable to produce a breakdown in the transformer windings, the method of slowly switching in for the purpose of avoiding current in-rushes is not one which can easily be recommended.
At one time circuit breakers controlling transformers were sometimes fitted with buffer resistors and auxiliary contacts so that the resistors were connected in series with the transformer when switching in, being subsequently short circuited upon completion of the switching operation. These buffer resistors are, however, no longer employed but as a matter of academic interest FIG. 75 illustrates the effect of a buffer resistor on the transient switching in cur rent in-rush of a 20 kVA transformer, the resistor having such a value that it takes 5 percent of the normal supply voltage at no-load. An objection to these current in-rushes is the mechanical forces which are exerted between coils at the instant of switching in. While these certainly die down more or less rap idly, the conductors are strained to some extent and the insulation between individual conductors may become compressed in places, while in other places the normal mechanical pressure due to the winding process may be released, so that the mechanical rigidity of the coils as a whole becomes entirely altered.
FIG. 75 Current in-rush when switching in a 20 kVA transformer.
That is, in some parts adjacent conductors may be slack, while in others they may be compressed tightly, and with repeated switching in operations there may be a risk of failure of the insulation between turns of the windings.
Cases have been known in which a transformer switched in under particularly adverse conditions has moved in its tank, and this introduces the possibility of damage to connections between coils and connections from coils to terminals, resulting in open circuits in the windings concerned.
Among the minor disadvantages are the tripping of main switches, blowing of fuses, and the operation of relays, but while these are often annoying, they are not serious.
Short-circuit currents
It has already been shown that abnormal currents may occur in the primary windings under certain adverse conditions when switching in a transformer on no-load, but much heavier currents may flow in both primary and secondary windings when a transformer momentarily supplies its heaviest load: that is, when a short circuit occurs across the secondary terminals. We thus have four distinct current conditions to which a transformer may be subjected, these being:
(a) transient switching no-load current in-rush,
(b) steady no-load current,
(c) steady normal load current,
(d) transient short-circuit current.
The currents which represent a danger to the windings are the transient currents only, viz. (a) and (d), and of these two the latter is the one against which special precaution must be taken, as the resulting currents set up severe mechanical stresses in the windings.
If a conductor carries current a magnetic field is set up round the conductor in the form of concentric circles, the density of the field at any point being directly proportional to the current in the conductor and inversely proportional to the distance between the conductor and the point considered. If two conductors both carrying current are in close proximity to each other they will each be subjected to the influence of the magnetic field surrounding the other, and in the case of adjacent conductors carrying currents in the same direction the magnetic fields will produce a force of attraction between the two conductors, while with currents flowing in opposite directions the magnetic fields mutually repel each other and a repulsion force is set up between the conductors. For a given current and spacing between the two conductors the value of the forces is the same, irrespective of whether they are attractive or repulsive.
If, now, the above principles are applied to transformer coils it will be seen that any one coil, either primary or secondary, carries current so that the cur rents in opposite sides flow in opposite directions, and repulsion forces are thus set up between opposite sides so that the coil tends to expand radially outwards in just the same way as does a revolving ring or other structure due to centrifugal force. The coil thus tends to assume a circular shape under the influence of short-circuit stresses, and therefore it is obvious that a coil which is originally circular is fundamentally the best shape, and is one which is least liable to distortion under fault conditions. From this point of view the advantages of the circular core type of construction are obvious.
In a coil composed of a number of wires arranged in a number of layers, each having a number of turns per layer, such as may be the case with HV windings, the wires situated in the same sides of the coil carry current in the same direction, and therefore attract one another and tend to maintain the homogeneity of the coil. In addition to the radial forces set up in the individual windings tending to force the coils into a circular shape, other repulsion forces exist between primary and secondary windings, as these windings carry currents flowing in opposite directions. The directions of these repulsion forces are shown in FIG. 76 for circular and rectangular coils under the conditions of:
(i) coincident electrical centers, (ii) non-coincident centers.
When the electrical centers coincide it will be seen that if the coils are of the same dimensions, repulsion forces normal to the coil surfaces only exist, but if the electrical centers do not coincide, a component at right angles to the force normal to the coil surfaces is introduced which tends to make the coils slide past one another. A similar component at right angles to the normal component is introduced, even if the electrical centers are non-coincident and the dimensions of the coils are different. In this case, the system consisting of the primary and secondary coils is balanced as a whole, but adjacent sides of primary and secondary coils are liable to distortion on account of the sliding components introduced. In actual practice, both with core-type and shell-type transformers, the sliding component of the mechanical forces between primary and secondary coils is the one which has been responsible for many failures under external short-circuit conditions, particularly of some of the older transformers having low impedances. In passing it should be remembered that often it is not possible to preserve the coincidence of electrical centers at all ratios when transformers are fitted with voltage adjusting tappings.
FIG. 76 Mechanical forces on transformer coils.
The value of the current flowing under external short-circuit conditions is inversely proportional to the impedance of the entire circuit up to the actual fault, and, so far as the transformer itself is concerned, the most onerous condition which it has to withstand is a short circuit across the secondary terminals. The question of how the transformer impedance is affected by matters of design is dealt with elsewhere in this guide, and it is only necessary to point out here that as the impedance voltage is that voltage required to circulate full load current in the transformer windings, the short-circuit current bears the same relation to the normal full-load current as does the normal full line volt age to the impedance voltage, the latter being expressed in terms of the full line voltage. Expressed in equation form, the connection between short-circuit current and impedance voltage is as follows:
(eqn. 60)
where ISC is the primary or secondary short-circuit current
IFL is the primary or secondary full-load current
Vz is the percentage impedance voltage
It must be remembered that the short-circuit current derived from the above equation presumes maintenance of the full line voltage under fault conditions, but in fact the line voltage is generally maintained for the first few cycles only after the first instant of short circuit, and then only on the larger systems having sufficient MVA of generating plant behind the fault. Therefore, at the first instant of short circuit the current reaches a value given by Eq. (eqn. 60) but as the line voltage drops the value of the short-circuit current similarly falls until the transformer is automatically switched out of circuit.
The initial value of the short-circuit current in-rush may be further modified by the normal conditions existing at the instant of short circuit, and in the worst case the initial value of the short-circuit current reaches twice the amount given by Eq. (eqn. 60) by what has been termed the 'doubling effect.' This doubling effect occurs when the actual short circuit is made at the instant when the voltage of the circuit is zero, and we will consider the two extreme cases when the short circuit occurs at the instant (a) when the voltage is passing through its maximum value, and (b) when the voltage is passing through zero. In FIG. 77 V represents the voltage wave, BM the flux wave, and IFL the wave of normal full-load current. If a short circuit takes place at the instant marked I in the diagram, the flux leading the voltage by 90º is zero, and as on short circuit the resulting current is in phase with the flux or nearly so, the short-circuit current should have a similar phase relationship. At the instant I the short-circuit current should therefore be zero, and if no current existed in the circuit at this instant the short-circuit current would pursue its normal course, reaching an initial maximum value corresponding to Eq. (eqn. 60), and it would be disposed symmetrically on either side of the zero axis, gradually and symmetrically dying down until the transformer was tripped out of circuit. This condition is shown by the current wave ISC in FIG. 77. On account of the presence of the normal full-load current which has a definite value at the instant I, the short-circuit current must initially start from that point and the resulting wave will be somewhat unsymmetrical, depending upon the ratio between the full-load and short-circuit currents and upon their relative power factors. This wave is shown at I_SC in FIG. 77.
FIG. 77 Short circuit at instant V = Vmax FIG. 78 Short circuit at
instant V = zero.
If, on the other hand, the short circuit occurs at the instant marked II in FIG. 78, the voltage is zero and the flux has a negative maximum value, so that the initial short-circuit current should also be at or near its negative maximum value. This cannot occur, as the short-circuit current cannot instantly attain the value corresponding to the position and value of the flux wave, but instead it must start from a value corresponding in sign and magnitude to the current already in the circuit at the particular instant, viz. the normal full-load current at the instant II. During the first cycle immediately following the short circuit the full voltage is generally active in producing abnormal short-circuit cur rents, and in order to maintain this voltage during the first half cycle, the short circuit current must vary from a maximum negative to a maximum positive value, that is the total change is twice that occurring when a short circuit takes place at maximum voltage and zero flux. This abnormal current wave, there fore, commences from the value of the full-load current in the circuit at the instant II, and from this point rises to a value approaching twice that obtained with a symmetrical short circuit, as shown at I_SC in FIG. 78. This explains the so-called 'doubling effect,' though, as a rule, the short-circuit current does not reach the full double value on account of resistance voltage drops. This highly abnormal wave is, of course, unsymmetrical, but dies down rapidly, giving ultimately the same symmetrical current distribution as when a short circuit takes place at the instant corresponding to maximum voltage.
The mechanical stresses set up in transformer windings vary as the square of the current flowing, and it will be seen, therefore, that the doubling effect may have serious consequences. For instance, in a transformer having an impedance of 5 percent the initial stresses under short-circuit conditions would be 400 times as great as those in the transformer under normal full-load conditions when making the short circuit at maximum voltage, but when making the short-circuit at zero voltage the resulting stresses in the windings would be approximately 1600 times as great as those under normal full-load conditions, on account of the doubling effect.
In practice these high mechanical stresses have been responsible for dam age to HV end coils of core-type transformers with concentrically disposed windings, though in such transformers the radial bursting tendency on the outer windings is not usually high enough to reach the elastic limit of the conductors. Similarly, the compressive stress on the inner windings is resisted by the mechanical rigidity inherent in such windings. With rectangular shell-type transformers employing flat rectangular coils arranged in sandwiched fashion, severe distortion and subsequent rupture of the ends of the coils projecting beyond the core has occurred from the same cause. With low-reactance transformers particularly, the HV end coils and also the coil clamping structure have been distorted, but as the forces which come into play have become more and more appreciated, transformer reactance has been increased and the coil clamping structure better designed and more adequately braced. Such coil clamps have been applied to compress the coils in an axial direction and to restrain them from moving under short-circuit conditions. Except in very special cases, radial coil supports are not necessary, though there is more justification for their use with rectangular type of coils. In the core-type transformer the coils are generally circular, and as this is inherently the best possible shape, the conductors are best able to stand the short-circuit stresses.
While on short circuit the phenomenon most to be feared is the mechanical stresses to which the windings and structure are subjected, and it must be remembered as somewhat of a paradox that this is so on account of the rapidity with which modern circuit breakers automatically disconnect such a fault from the supply. If for any reason the automatic means provided did not operate, the transformer would rapidly become overheated, and it would exhibit all the appearances associated with severe overloads. In a very short space of time short circuits between turns would take place and the windings would become destroyed.
The reader should refer to Section 4.8 for more detailed information regarding electromagnetic forces in transformer windings.
Geomagnetically induced currents
Geomagnetically induced currents (GICs) are not transients in the same sense as those others discussed in this section, although they are transient in so far as their more serious effects are present only for limited periods of time and at infrequent and irregular intervals. It is, therefore, appropriate to discuss them in this section. As the name suggests, these currents occur in certain transformers under certain conditions associated with activity involving the ground's magnetic field.
During periods of intense sunspot activity, it has long been know that streams of highly charged particles can be ejected and carried by solar wind to be captured by the ground's magnetic field. The interaction of these particles with the ground's field can result in sudden and severe magnetic storms which can lead to fluctuations in the normally stable field close to the ground's surface.
One of the best-known effects of this phenomenon is the so-called 'Northern Lights' or Aurora Borealis. The fluctuations in the magnetic field can induce electric fields in the ground's surface giving rise to potential differences of typically 1-5 V/km. This potential difference exists between transformer grounded star-points and can lead to circulating currents with the associated transmission lines completing the circuit. Currents of over 100 A, quasi DC in nature with a time period of about 15 minutes and persisting for several hours have been reported in the UK. In North America where transmission lines tend to be longer, such instances are reported to be more common with current magnitude and resultant damage to plant more severe.
Large transformers, particularly those with five limb cores or arranged as banks of single-phase units, are known to be particularly susceptible to dam age resulting from the passage of sustained GICs. When the level of GIC flowing is in the same order of magnitude as the normal magnetizing current of the transformer this will lead to a DC offset causing the magnetic circuit to be driven into half cycle saturation. The resultant increase in flux density causes significant increase in leakage flux which in turn, can cause severe overheating in components that would not normally be exposed to such a high level of flux, such as tank walls and core bolts. In addition the increase in harmonic content of the current waveform as a result of saturation will increase winding eddy current losses. Consequences can range from gassing in the oil in contact with the overheating component to component failure leading to loss of the unit.
During a severe magnetic storm in 1989 the level of GIC flowing caused almost simultaneous failure, as a result of profuse gassing, of two 400 kV autotransformers in the South of England. These were separated by over 200 miles, but at extreme ends of the east/west 400 kV network. At the same time in the USA, in New York, a generator step-up transformer failed catastrophically due to severe overheating as a result of high levels of leakage flux.
In the UK Scottish Power, who have been monitoring these effects since the 1990s, have recorded a DC current of 25 A in the neutral of a 400/275 kV 1000 MVA autotransformer at their Neilston substation and, using the same monitoring equipment, have been able to show correlation between currents measured in transformer neutrals on their transmission network and geomagnetic data recorded at the same time by the British Geological Survey Observatories.
It will be recognized from consideration of the above that only somewhat limited measures are available to mitigate the effects of GICs. However, with present day monitoring levels, periods of high geomagnetic activity can always be predicted with high accuracy, and of course, those transformers and lines that are susceptible are well known. One of the most worthwhile measures therefore, is to reduce V/f particularly for those parts of the network that are considered to be most at risk. Some reduction in flux density can be achieved by a modest reduction in voltage accompanied by a small increase in frequency, which whilst not necessarily eliminating the problem, will reduce the risk of damage to plant. System operators might quote license obligations and feed the event with additional MVArs. This only exacerbates the situation, but a GIC event should be treated as a severe weather emergency in the same way as hurricanes and snow. Then the license obligations can be waived.
Susceptible transformers can also be provided with on-line gas-in-oil monitors which can correlate d.g.a. levels to GIC activity and thus, in most cases, avoid costly unnecessary investigations into 'unexpected' increases in dissolved gases.
Where transformers are to be installed in locations known to be susceptible to high levels of GICs, this should be made known in the purchasers' specification. This will enable measures such as reduced strand sizes to be used for winding conductors, thus reducing winding eddy currents, and increased use of non-magnetic steel in proximity to the core, as well as additional external flux shunts, to reduce the magnitude of stray-flux heating. These measures will, of course, increase the cost of the transformer so should only be called for where a problem is known to exist.