Industrial Power Transformers -- Operation and maintenance [8]

Home | Articles | Forum | Glossary | Books


OPERATION UNDER ABNORMAL CONDITIONS

By definition, according to EN 60076, 'normal' service conditions for a power transformer are at an altitude of not greater than 1000 m above sea level, within an ambient temperature range of -25ºC to +40ºC, subjected to a wave shape which is approximately sinusoidal, a three-phase supply which is approximately symmetrical and within an environment which does not require special provision on account of pollution and is not exposed to seismic disturbance.

The inference, then, is that 'abnormal' means any conditions which fall out side these boundaries. Some 'abnormalities' are, however, more likely to be encountered in practice than others and, in this section, abnormal will be taken to mean certain operating conditions which differ from those identified on the transformer nameplate, namely:

• at other than rated power,

• at ambient temperatures which may not conform to the averages set out in EN 60076,

• at other than rated frequency,

• at other than rated voltage,

• at unbalanced loading.

It is also increasingly common for transformers to be operated with wave shapes which are not sinusoidal because of the large amount of equipment now installed which utilizes thyristors or other semiconductor devices which generate high levels of harmonics. Although such high levels of harmonics constitute abnormal operating conditions in accordance with the above definition, the problem is one which is particularly associated with rectifier transformers and is therefore considered in Section 7.12 which covers these in detail.

Seismic withstand requirements are now also occasionally included in specifications for transformers supplying strategically essential systems, for example, emergency reactor cooling supplies for nuclear power stations. Transformers are normally built with a high degree of ruggedness in order to withstand forces occurring on short circuit, as explained in Section 4.7, so compliance with seismic requirements mainly involves firmly anchoring the unit down and bracing the core to withstand the lateral seismic forces. No generally accepted rules have, as yet, emerged for the provision of measures to cater for these forces and it is not therefore proposed to discuss this subject in greater detail.

The first two of the conditions listed above are the ones which are most frequently encountered in practice and they are, of course, interrelated.

Transformers are rarely required to operate continuously at near constant load and in the short to medium term ambients may differ significantly from the annual averages on which EN 60076 ratings are based. Generally users recognize that it is uneconomic to rate a transformer on the basis of the peak loading which only occurs for limited periods each day and, in addition, in temper ate climates where lighting and heating loads cause winter loading peaks to be very much higher than those arising during the summer months, it is usually considered desirable to expect to obtain a degree of overloading capability during periods of low ambient temperature. Hence it is necessary to find a means of assessing the extent to which recurring loads over and above the EN rating might be permitted and of converting a cyclic loading pattern into an equivalent continuous rating, or the extent of overload which a temporary reduction in ambient might allow.

Operation at other than rated load or other than EN 60076 ambients

Operation at other than rated load will result in hot spot temperature rises differing from those corresponding to rated conditions and, as explained in Section 4.5, rated temperature rise is based on a hot spot temperature of 98ºC with a 20ºC ambient. This hot spot temperature is considered to result in a rate of ageing which will provide a satisfactory life expectancy. It has already been stressed in the earlier section, and it is worth stressing again, that there is no 'correct' value of hot spot temperature. The value of 98ºC has been selected as a result of testing in laboratory conditions and any attempt to draw too significant a conclusion as to true life expectancy from such laboratory testing must be avoided because of the many other factors which also ultimately affect service life. Consequently other values of hot spot temperature must be equally tenable and other ratings besides the EN rating must be equally permissible, particularly if it is anticipated that these ratings will not be required to be delivered continuously and if it is recognized that 20ºC may not always be representative of many ambients in which EN rated transformers are required to operate.

The question then is to decide what variation from 98ºC should be permitted.

To do this it is necessary to revisit the conclusions concerning insulation ageing which were discussed in Section 4.5. These were that for those periods for which the hot spot temperature is above that corresponding to normal ageing, insulation life is being used up at faster than the rate corresponding to normal life expectancy. In order to obtain normal life expectancy, therefore, there must be balancing periods during which insulation life is being used up less rapidly. Expressed in quantitative terms the time required for insulation to reach its end of life condition is given by the Arrhenius law of chemical reaction rate:


where L is the time for the reaction to reach a given stage, but which might in this case be defined as end of life

T is the absolute temperature

and a and ß are constants.

Within a limited range of temperatures this can be approximated to the simpler Montsinger relationship

L = e^-p theta

where p is a constant

theta is the temperature in degrees Celsius.

Investigators have not always agreed on the criteria for which L is representative of end of life, but for the purposes of this evaluation this is not relevant and of more significance is the rate of ageing. This is the inverse of the lifetime, that is

v = Me e^p theta

where M is a constant which is dependent on many factors but principally moisture content of the insulation and availability of oxygen. Additionally the presence of certain additives such as those used for the production of thermally upgraded paper (see Section 3.4) can have a significant effect on its value.

Most important, however, is the fact that the coefficient of temperature variation, p, can be generally regarded as a constant over the temperature range 80-140ºC and it is widely agreed that its value is such that the rate of ageing doubles for every 6 K increase in temperature for most of the materials currently used in transformer insulation.

Relative aging rate

If 98ºC is then taken as the temperature at which normal ageing rate occurs, then the relative ageing rate at any other temperature ?h is given by the expression

V = _ _ aging rate at /ageing rate at 98 C

(eqn. 61)

This expression may be rewritten in terms of a power of 10 to give

(eqn. 62)

This is represented in Fig. 130 and by Table 18.

Example. 10 hours at 104ºC and 14 hours at 86ºC would use (10 _ 2) _ (14 _ 0.25) _ 23.5 hours life used in 24 hours operation.


FIG. 130 Life line Table 18

Equivalent life loss in a 24-hour period

It may be required to find the time t hours per day for which the transformer may be operated with a given hot spot temperature ?h, with the complement to 24 hours corresponding to a sufficiently low temperature for negligible life loss; then the hours of life loss are given by tV and for tV to equal 24:

(eqn. 63)

Equation (eqn. 63) gives the number of hours per day of operation at any given value of ?h that will use one days life per day. Table 19 gives values of t for various values of ?h.

Table 19

It can happen that it is required, for limited periods of time, to operate at higher temperatures than those associated with normal daily cyclic loading and accept the more rapid use of life for those periods, for instance the loss of a unit in a group. If a daily loss of life 2, 5, 10 … times the normal value is assumed, the corresponding 'hot spot' temperatures will be 6ºC, 14ºC and 20ºC higher than given in Table 19, but ?h must not exceed 140ºC.

As a general rule, the transformer will be loaded in such a way that daily operation with use of life higher than normal will not extend over periods of time which are an appreciable proportion of normal expected life duration. In these conditions it will not be necessary to keep a record of the successive loads on the unit.

Determination of hot spot temperature for other than rated load In all of the foregoing discussion load capability has been related to hot spot temperature. The effect on hot spot temperature at rated load of variation in ambient is simple to deduce; one degree increase or reduction in ambient will result, respectively, in one degree increase or reduction in hot spot temperature. The question which is less simple to answer is, how does hot spot temperature vary with variation in load at constant ambient? To consider the answer to this it is necessary to examine the thermal characteristics of a transformer, which were discussed in Section 4.5.

Hot spot temperature is made up of the following components:

• Ambient temperature

• Top oil temperature rise

• Average gradient

• Difference between average and maximum gradient of the windings In EN 60076 the last two terms are on occasions taken together to represent maximum gradient. Maximum gradient is then greater than average gradient by the 'hot spot factor.' This factor is considered to vary between 1.1 for distribution transformers to 1.3 for medium sized power transformers. The last term thus varies between 0.1 and 0.3 times the average gradient.

Effect of load on oil temperature rise Mean oil rise is determined by the dissipation capability of the cooling surface and the heat to be dissipated. The heat to be dissipated depends on the losses.

At an overload k times rated load the losses will be increased to:

Fe + k^2 Cu

where Fe and Cu are the rated no-load and load losses respectively.

As the excess temperature of the cooling surface above its surroundings increases, cooling efficiency will tend to be increased, that is the oil tempera ture will increase less than pro-rata with the increased losses to be dissipated.

This relationship may be expressed in the form increased losses rated losses where ?o is the oil temperature rise, with suffixes 1 and 2, respectively, indicating the rated and the overload conditions.

EN 60076, Part 2, which deals with temperature rise, gives values for the index x which are considered to be valid within a band of +/-20 percent of the rated power, these are:

0.8 for distribution transformers having natural cooling with a maximum rating of 2500 kVA

0.9 for larger transformers with ON.. cooling 1.0 for transformers with OF.. or OD.. cooling

The inference to be drawn from the above values is that with OF.. and OD.. cooling, the coolers are already working at a high level of efficiency so that increasing their temperature with respect to the surroundings cannot improve the cooler efficiency further.

Effect of load on winding gradients The heat transfer between windings and oil is considered to improve in the case of ON.. and OF.. transformers for increased losses, that is the increased heat to be dissipated probably increases the oil flow rate, so that the winding gradient also increases less than pro-rata with heat to be dissipated, which is, of course, proportional to overload factor squared. EN 60076-2, gives the following values

wo wo1 2

_ k y

where ??wo is the winding/oil differential temperature, or gradient, with additional suffixes 1 and 2, respectively, to indicate the rated and overload conditions. The index y is then

1.6 for ON.. and OF.. cooled transformers

2.0 for OD.. cooled transformers

EN 60076-2, places limits on the accuracy of the above as within a band of _10 percent of the current at which the gradient is measured, however it does state that this limitation, and that placed on the formula for extrapolation for oil temperature indicated above, should be applied where the procedure is used for the evaluation of test results subject to guarantee. In other circumstances the method may give useful results over wider ranges.

Example 1. The above method may be used to estimate the hot spot temperature of a 30/60 MVA 132/33 kV ONAN/ODAF transformer when operated at, say, 70 MVA. The transformer has losses of 28 kW at no load and load losses of 374 kW on minimum tapping at 60 MVA. On temperature rise test the top oil rise was 57.8ºC and the rise by resistance was; LV, 69.2ºC; HV, 68.7ºC on minimum tapping. The effect of changes in ambient can also be included. Let us assume that the ambient temperature is 10ºC.

The transformer temperature rise test certificate should indicate the value of the mean oil rise and the winding average gradients. If this information is not available, for example if no temperature rise test was carried out, these values will have to be estimated. Top oil rise at 60 MVA can be measured by a thermometer placed in the top tank pocket. Oil temperature rise on return from the cooler can be similarly measured at the tank oil inlet. Mean oil temperature rise is the average of these two figures. Let us assume that either from the test certificate or by measurement, mean oil rise is found to be 49.8ºC. Then, LV winding gradient = 69.2-49.8 = 19.4ºC, and

HV winding gradient = 68.7-49.8 = 18.9ºC.

At 70 MVA, the overload factor is 70/60 = 1.167

New top oil rise 57.8

The critical gradient is the LV winding _ 19.4ºC, at 1.167 times rated load this will become:

19.4 x 1.167^1.6 = 24.8

hence, hot spot temperature = 10 + 77.3 + 1.3 x 24.8 = 119.5ºC.

By reference to Table 19 it can be seen that this overload may be carried for up to 2 hours/day with the remainder of the time at a load which is low enough to cause minimal loss of life. Alternatively, provided this daily over load is only imposed for a matter of a few weeks, normal load may be carried for the remainder of the day with only negligible loss of life.

Normally a transformer such as the one in the above example would have pumps and fans controlled from a winding temperature indicator which would mean that these would not be switched in until a fairly high winding tempera ture was reached, however if the overload is anticipated, pumps and fans can, with advantage, be switched in immediately. This will delay the time taken to reach maximum hot spot temperature and, although cooler losses will be incurred, these will to some extent be offset by the lower transformer load loss resulting from the reduction in winding copper temperature.

During any period of overloading there will be a time delay before the maximum hot spot temperature is reached. This will have two components:

• The time for the windings to reach equilibrium with the oil at the new level of gradient.

• The time taken for the complete transformer to reach equilibrium with its surroundings.

The first of these, the winding time constant, is likely to be of the order of minutes, say, between 5 and 20 minutes and it is normally neglected. The second, the transformer oil time constant, or simply transformer time constant, will be a great deal longer, probably between 1 and 3 hours. The hot spot temperature variation for a daily loading duty of the form indicated in Fig. 131 will be as shown, with an exponential increase at the commencement of the overloading and a similar decay at the end of the overload period. In terms of use of life the areas under these exponential curves are equal, so the times spent in the heating and cooling phases will partly cancel out and may therefore be ignored. This will not be entirely true however because rate of ageing is proportional to 2 (or 10) raised to a power of temperature (see Eqs (eqn. 58) and (eqn. 59) above). Ignoring the time delays results in the introduction of a small factor of pessimism which is no bad thing. Recognition of the time delay can be particularly beneficial, however, where the overload would take the final hot spot temperature above 140ºC. By definition, for an overloading period equal to the time constant for the oil, the rise in top oil temperature at the end of this period will be approximately 63 percent of its ultimate value. If the time constant for the oil is 2 hours and its ultimate rise, say, 45ºC, 63 percent of this is only 28.4ºC, some 16.6ºC lower, and this will not be reached until after 2 hours.


FIG. 131 Simplified load diagram for cyclic daily duty

IEC loading guide for oil-filled transformers

The principles outlined above have been used as a basis for compiling loading guides for oil-immersed transformers, for example IEC 60354 (in the UK BS 7735) Loading guide for oil-immersed transformers. Whilst the use of such guides can greatly simplify the process of assessing loading capability, it is always beneficial to have a good understanding of the theory involved. As well as aiding an appreciation of the precise effect on the transformer of operating at other than rated load, it is clearly preferable to be able to perform a calculation for a particular transformer using loss values and gradients specific to that transformer than to rely on guides which must of necessity make many assumptions. It will be seen from the example worked above, that if a transformer has high maximum gradients, say approaching 30ºC, which is not untypical of many OD. type transformers (IEC 60354, 1991, assumes 29ºC), then its ability to carry overloads will be considerably less than that of a transformer having maximum gradients of, say 25ºC or less, since the effect of overloading for OD.. transformers is to increase gradients in accordance with a square law. For an overload of 25 percent, 1.252 _ 30 _ 46.8ºC, whereas 1.252 _ 25 _ 39.06, so an OD.. transformer having the lower maximum gradient will have a rate of using life of less than half that of the transformer with the higher gradient at an overload of 25 percent.

It should also be noted that a transformer with a low ratio of load loss to no-load loss will also be capable of slightly greater overloading than one for which this ratio is higher, since it is only the load loss which will increase under overload, and this in proportion to the overload squared. Loading guides must assume a typical value for the ratio of the load to no-load loss. IEC 60354 assumes a ratio of 5 for ONAN distribution transformers and 6 for all other types. Just how widely actual transformers can vary in practice will be apparent from the example of the 30/60 MVA transformer used in the overload calculation above. The figures are for an actual transformer and it can be seen that the ratio is 374/28 _ 13.4 to 1. Variation of this ratio has less an effect on top oil temperature, and hence hot spot temperature, than does variation in gradient. If the transformer in the above example is assumed to have the same total losses, that is 402 kW, but split so that the ratio is the IEC assumed value, that is 57.4 and 344.6 kW, respectively, and the top oil rise recalculated for a load of 70 MVA, it will be found that this equates to 75.7ºC, only 1.6ºC lower.

Continuous loading at alternative ambients in accordance with IEC 60354 Table 20, reproduced from IEC 60354, gives factors for continuous loadings which will result in a hot spot temperature of 98ºC for varying ambient temperatures and for each type of cooling, thus enabling the continuous loading capability for any ambient temperature to be calculated.

Table 20 Acceptable load factor for continuous duty K24 at different ambient temperatures (ON, OF and OD cooling)

Cyclic loading in accordance with IEC 60354 IEC 354 may also be used to give indication of permissible daily loading cycles. Loading patterns are deemed to consist of a simplified daily cycle having the form shown in Fig. 131, above. Symbols used in the guide have the following meanings:

K1 is the initial load power as a fraction of rated power K2 is the permissible load power as a fraction of rated power (usually greater than unity)

t is the duration of K2, in hours

?A is the temperature of cooling medium, air or water

__ and where S1 is the initial load power

S2 is the permissible load power and

Sr is the rated power

The values of K1, K2 and t must be selected to obtain as close a match as possible between the actual load cycle and the overload basic cycle of Fig. 131. This can be done on an area for area basis as shown in Fig. 132, reproduced from IEC 60354. For the not uncommon case where there are two peaks of nearly equal amplitude but different duration, the value of t is deter mined for the peak of longer duration and the value of K1 is selected to correspond to the average of the remaining load as shown in the example of Fig. 133. Where the peaks are in close succession, the value of t is made long enough to enclose both peaks and K1 is selected to correspond to the average of the remaining load, as shown in Fig. 134.


FIG. 132 Load cycle with one peak


FIG. 133 Load cycle with two peaks of equal amplitude and different duration


FIG. 134 Load cycle with peaks in close succession


FIG. 135 Permissible cyclic loading duties for ONAN distribution transformers for normal loss of life at 20 degr. C ambient

A series of loading curves for varying ambients, of which Fig. 135 is a

typical example, are then provided to enable permissible cyclic loading to be deduced. The guide lists the thermal characteristics which have been assumed in drawing up the curves and recommends that for normal cyclic loading the load current should not exceed 1.5 times rated current and the hot spot temperature should not exceed 140ºC. For large power transformers (over 100 MVA) it recommends that these should not exceed 1.3 times rated current and 120ºC, respectively. For all transformers it recommends that top oil temperature should not exceed 105ºC. The following examples shows how the tables may be used.

Example 2. A 2 MVA ONAN distribution transformer has an initial load of 1 MVA. To find the permissible load for 2 hours at an ambient temperature of 20ºC, assuming constant voltage:

?A _ 20ºC K1 _ 0.5 t _ 2 hour FIG. 135 gives K2 _ 1.56, but the guide limit is 1.5. Therefore the permissible load for 2 hours is 3 MVA (then returning to 1 MVA).

Example 3. With ?A _ 20ºC, an ONAN distribution transformer is required to carry 1750 kVA for 8 hours and 1000 kVA for the remaining 16 hours each day. To find the optimum rating required to meet this duty. Assuming constant voltage, we have K K 2 1 1750 1000 175 __ .


FIG. 136 Illustration of Example 3

On the curve of Fig. 135, first plot the line K2/K1 _ 1.75 (Fig. 136), then at the point where this intersects the curve for t _ 8, the values of K1 and K2 are K2 _ 1.15 and K1 _ 0.66 so that the rated power is

Sr

___ 1750 115 1000

066 1520

kVA Emergency cyclic loading

Example 3, above, enables the best rating of transformer to be selected to meet a known cyclic duty. On occasions it may be necessary to overload a transformer on a cyclic basis when it was not originally intended to be so loaded and even though some shortening of life might be entailed. IEC 354 terms this 'Emergency cyclic loading' and provides a series of tables covering all cooling types for a range of loading duties. Table 21, which is Table 27 of the guide, provides information relating to emergency loading of OD medium and large power transformers for 2 hours.

Example 4. What is the daily loss of life and the hot spot temperature when the 30/60 MVA transformer of Example 1, above, is loaded at 70 MVA for 2 hours in an ambient of 10ºC?

K1 _ 1.0, K2 _ 1.167, ?A _ 10ºC, t _ 2 hours

Table 21 shows that V _ 2.4, ??h _ 103.7 for an ambient temperature of 20ºC. (By linear interpolation between K2 _ 1.1 and K2 _ 1.2, which is reasonable for hot spot temperature, somewhat optimistic for V). Taking account of the actual ambient temperature of 10ºC we have:

Loss of life _ 2.4 _ 0.32 _ 0.77 'normal' days ?h _ 103.7 _ 10 _ 113.7ºC.

The above hot spot temperature is a little lower than the figure of 119.5 calculated in Example 1 which corresponds to exactly one days loss of life per day.

It will be noted that there is a reference in Table 21 to a Table 1 which gives a value of maximum permissible hot spot temperature. This is, of course, Table 1 of IEC 60354. For completeness this is reproduced as Table 22, how ever the reader should refer to IEC 60354 for a full explanation of its position in relation to maximum hot spot temperature.

The above examples give some indication of the information which is available in IEC 60354 and the way in which it can be used to determine the loading capability of an oil-filled transformer. For a fuller explanation of over loading principles for all sizes of transformers and types of cooling reference should be made to the document itself.

Table 21 OD medium and large power transformers: t _ 2 hours. Permissible duties and corresponding daily loss of life (In 'normal' days)

To determine whether a daily load diagram characterized by particular values of K1 and K2 is permissible and to evaluate the daily loss of life entailed, proceed as follows:

Hot-spot temperature:

Add the hot-spot temperature rise given in the table to the ambient temperature. If the resulting hot-spot temperature exceeds the limit stated in Table 1, the duty is not permissible.

Limitations on overloading

Although in the previous paragraphs emphasis has been placed on the arbitrary nature of 98ºC as a hot spot temperature for 'normal' rating in normal ambients and the flexibility built into the rating of transformers designed on this basis, before concluding it is appropriate to add a few words of caution.


Table 22 Current and temperature limits applicable to loading beyond nameplate rating

Care should be taken, when increasing the load on any transformer, that any associated cables and switchgear are adequately rated for such increases and that any transformer ancillary equipment, for example tapchangers, bushings, etc., do not impose any limitation. The voltage regulation will also increase when the load on a transformer is increased.

The mineral oil in the transformer should comply with BS 148 and should be maintained at least in accordance with BS 5730. Consideration should be given to closer monitoring of the oil in accordance with the procedures out lined in Section 6.7.

For normal cyclic duty, the current should not exceed 1.5 times rated value.

Hot spot temperature should never exceed 140ºC. For emergency duty, currents in excess of 1.5 times rated value are permissible provided that the 140ºC hot spot temperature is not exceeded, that the fittings and associated equipment are capable of carrying the overload and that the oil temperature does not exceed 115ºC. The limit of 115ºC for the oil temperature has been set bearing in mind that the oil may overflow at oil temperatures above normal. Depending on the provision made for oil expansion on a particular transformer, the oil may overflow at temperatures lower than 115ºC.

IEC 60354, 1991, states that for certain emergency conditions the hot spot temperature may be allowed to reach 160ºC. The question then is what constitutes such an emergency. It should be noted that when the hot spot temperature reaches 140-160ºC, gas bubbles may develop which could hazard the electrical strength of the transformer. It is clearly most undesirable to add to an existing emergency, possibly caused by the failure of a transformer, by creating conditions which might lead to the failure of a second unit.

Operation at other than rated voltage and frequency Considering initially variation from rated frequency, it can be stated that it is not usually possible to operate a transformer at any frequency appreciably lower than that for which it was designed unless the voltage and consequently the output are correspondingly reduced. The reason for this is evident if the expression connecting voltage, frequency and magnetic flux given in Section 1.1, Eq. (1.4), is recalled. This is

E/N _ 4.44 Bm Af _ 10_6

where E/N is the volts per turn, which is the same in both windings

Bm is the maximum value of flux density in the core, Tesla

A is the net cross-sectional area of the core, mm^2

f is the frequency of supply, Hz.

Since, for a particular transformer A and N are fixed, the only variables are E, Bm and f, and of these Bm is likely to have been set at the highest practicable value by the transformer manufacturer. We are therefore left with E and f as the only permissible variables when considering using a transformer on a frequency lower than that for which it was designed. The balance of the equation must be maintained under all conditions, and therefore any reduction in frequency f will necessitate precisely the same proportionate reduction in the voltage E if the flux density Bm is not to be exceeded and the transformer core not to become overheated. The lower the frequency the higher the flux density in the core, but as this increase is relatively small over the range of the most common commercial frequencies its influence on the output is very slight, and therefore the reduction in voltage and output can be taken as being the same as the reduction in frequency.

Operation at higher than rated frequency but at design voltage is less likely to be problematical. Firstly, the danger of saturation of the core is no longer a threat since increased frequency means a reduction in flux density. There will be some increase in winding eddy current loss which will probably increase as the square of the frequency. The impact of this will depend on the magnitude of the eddy current losses at rated frequency but for transformers smaller than 1 or 2 MVA and frequencies within about 20 percent of rated frequency, this will probably be acceptable. Changes in hysteresis and eddy current components of core loss, both of which increase with frequency, will probably be balanced by the reduction in flux density as can be seen by reference to the expressions for these quantities which were given in Section 3.2. These were Eqs (3.1) and (3.2) respectively:

Hysteresis loss, Wh _ k1fBn max W/kg and Eddy current loss, We _ k2 f 2 t 2 B2 eff /? W/kg

where k1 and k2 are constants for the material

f is the frequency, Hz

t is thickness of the material, mm

? is the resistivity of the material

Bmax is maximum flux density, T

Beff is the flux density corresponding to the r.m.s. value of the applied voltage.

n is the 'Steinmetz exponent' which is a function of the material.

Considering next the question of using a transformer on voltages different from the normal rated voltage, it can be stated very definitely that on no account should transformers be operated on voltages appreciably higher than rated voltage. This is inadmissible not only from the point of view of electrical clearances but also from that of flux density, as will be clear from Eq. (1.4) which was recalled earlier. It should be noted, whilst considering this aspect of operation at higher than rated voltage, that many specifications state that the system voltage may be capable of increasing by 10 percent above its rated value. It is important that in this circumstance the designer must limit the design flux density to such a value as will ensure that saturation is not approached at the overvoltage condition. This usually means that the nominal

flux density must not exceed 1.7 Tesla at any point in the core. If the transformer has an on-load tapchanger under automatic control and there is any possibility that this might be driven to minimum tap position whilst system volts are high, then the design flux density must be selected so as to ensure that saturation is not approached under this fault condition, which might require that this be kept as low as 1.55 Tesla.


FIG. 137 Current distribution due to a single-phase load on polyphase transformers or transformer groups. Note: in all cases the dotted lines indicate the phase angle of the single-phase load currents


FIG. 138 Current distribution due to single-phase load on polyphase transformer groups. Note: in all cases the dotted lines indicate the phase angle of the single-phase load currents

Operation with unbalanced loading

In considering the question of unbalanced loading it is easiest to treat the subject from the extreme standpoint of the supply to one single-phase load only, as any unbalanced three-phase load can be split up into a balanced three-phase load and one or two single-phase loads. As the conditions arising from the balanced three-phase load are those which would normally occur, it is only a question of superposing those arising from the single-phase load upon the normal conditions to obtain the sum total effects. For the purpose of this study it is only necessary to consider the more usual connections adopted for supplying three phase loads. The value of current distribution is based upon the assumption that the single-phase currents are not sufficient to distort the voltage phasor diagrams for the transformers or transformer banks. This assumption would approximate very closely to the truth in all cases where the primary and secondary currents in each phase are balanced. In those cases, however, where the primary current on the loaded phase or phases has to return through phases unloaded on the secondary side, the distortion may be considerable, even with relatively small loads; this feature is very pronounced where three-phase shell-type transformers and banks of single-phase transformers are employed. Figures 6.137-6.139 show the current distribution on the primary and secondary sides of three- to three-phase transformers or banks with different arrangements of single-phase loading and different transformer connections. These diagrams may briefly be explained as follows.


FIG. 139 Phasor diagrams showing unbalanced loadings on a delta/star, three-phase, step-down transformer.

(a) Star/star; single-phase load across two lines

With this method of single-phase loading the primary load current has a free path through the two primary windings corresponding to the loaded secondary phases, and through the two line wires to the source of supply. There is, there fore, no choking effect, and the voltage drops in the transformer windings are those due only to the normal impedance of the transformer. The transformer neutral points are relatively stable, and the voltage of the open phase is practically the same as at no-load. The secondary neutral point can be grounded with out affecting the conditions.

The above remarks apply equally to all types of transformers.

(b) Star/star; single-phase load from one line to neutral With this method of single-phase loading the primary load current corresponding to the current in the loaded secondary must find a return path through the other two primary phases, and as load currents are not flowing in the secondary windings of these two phases, the load currents in the primaries act as magnetizing currents to the two phases, so that their voltages considerably increase while the voltage of the loaded phase decreases. The neutral point, therefore, is considerably displaced. The current distribution shown on the primary side is approximate only, as this will vary with each individual design.

The above remarks apply strictly to three-phase shell-type transformers and to three-phase banks of single-phase transformers, but three-phase core-type transformers can, on account of their interlinked magnetic circuits, supply consider able unbalanced loads without very severe displacement of the neutral point.

(c) Star/star with generator and transformer primary neutrals joined; single phase load from one line to neutral

In this case the connection between the generator and transformer neutral points provides the return path for the primary load current, and so far as this is concerned, the other two phases are short circuited. There is therefore no choking effect, and the voltage drops in the transformer windings are those on the one phase only, due to the normal impedance of the transformer. The transformer neutral points are relatively stable, and the voltages of the above phases are practically the same as at no-load. The secondary neutral point can be grounded without affecting the conditions.

The above remarks apply equally to all types of transformers.

(d) Delta/delta; single-phase load across two lines

With this connection the loaded phase carries two-thirds of the total current, while the remainder flows through the other two phases, which are in series with each other and in parallel with the loaded phase. On the primary side all three windings carry load currents in the same proportion as the secondary windings, and two of the line wires only convey current to and from the generator. There is no abnormal choking effect, and the voltage drops are due to the normal impedance of the transformer only. The type of transformer does not affect the general deductions.

(e) Star/delta; single-phase load across two lines On the delta side the distribution of current in the transformer windings is exactly the same as in the previous case, that is, two-thirds in the loaded phase and one-third in each of the other two. On the primary side the corresponding load currents are split up in the same proportions as on the secondary, and in value they are equal to the secondary currents of the different phases multiplied by _3 and multiplied or divided by the ratio of transformation, according to whether the transformer is a step-up or step-down. The primary neutral point is stable.

The above remarks apply equally to all types of transformers.

(f ) Delta/star; single-phase load across two lines Single-phase loading across lines of this connection gives a current distribution somewhat similar to that of (a), except that the currents in the two primary windings are 1/_3 times those occurring with the star primary, while all the three lines to the generating source carry currents in the proportions shown instead of two lines only carrying currents as in the case of the star primary.

There is no choking effect, and the voltage drops in the windings are due only to the normal impedance of the transformer. The transformer secondary neutral point is relatively stable and may be grounded. The voltage of the open phase is practically the same as at no-load.

The above remarks apply equally to all types of transformers.

(g) Delta/star; single-phase load from line to neutral With this connection and single-phase loading the neutral, primary and secondary windings on one phase only carry load current, and on the primary side this is conveyed to and from the generating source over two of the lines only. There is no choking effect, and the voltage drops in the transformer windings are those corresponding only to the normal impedance of the transformer. The secondary neutral point is stable and may be grounded without affecting the conditions. The voltages of the open phase are practically the same as at no-load. The type of transformer construction does not affect the general deductions.

(h) Interconnected star/star; single-phase load across two lines With this connection and method of loading, all the primary windings take a share of the load, and although in phase C there is no current in the secondary winding, the load currents in the two halves of the primary windings of that phase flow in opposite directions, so that their magnetic effects cancel. There is no choking effect, and the voltage drops in the transformer windings are those due to the normal impedance of the transformer only. With three-phase shell type transformers and three-phase banks of single-phase transformers the secondary neutral is not stable and should not be grounded unless the flux density is sufficiently low to permit this. With three-phase core-type transformers, how ever, the neutral is stable and could be grounded. The voltage of the open phase is practically that occurring at no-load.

(i) Interconnected star/star; single-phase load from one line to neutral With this connection and method of loading a partial choking effect occurs, due to the passage of load current in each half of the primary windings corresponding to the unloaded secondary windings. The voltage of the two phases in question, therefore, becomes increased on account of the high saturation in the cores and the voltage of the windings corresponding to the loaded phase drops. Both primary and secondary neutrals are therefore unstable and should not be grounded. The above remarks apply strictly to three-phase shell-type transformers and to three-phase banks of single-phase transformers. With three-phase core-type transformers the deflection of the neutral point is not so marked, and considerable out-of-balance loads can be supplied without any excessive deflections of the neutral points.

(j) Star/interconnected star; single-phase load across two lines With this connection and method of loading the secondary windings on all three limbs carry load currents, and therefore all the primary windings carry corresponding balancing load currents. The current distribution is clearly shown on the diagram, from which it will be seen there is no choking effect, and the transformer neutral points are stable if three-phase core-type transformers are used, and so may be grounded. On the secondary side the voltage of the open phase is practically the same as at no-load. The voltage drops in the transformer windings are those due only to the normal impedance of the transformer.

(k) Star/interconnected star; single-phase load from one line to neutral With this method of loading there is similarly no choking effect, as the primary windings corresponding to the loaded secondaries carry balancing load currents which flow simply through two of the line wires to the generating source.

The voltage drops in the transformer windings are those due only to the normal impedance of the transformer, and the voltages of the above phases are practically the same as at no-load. The secondary neutral is stable and can be grounded. The primary neutral can only be grounded, however, if the transformer unit is of the three-phase core type of construction.

(l) Delta/interconnected star; single-phase load across two lines With this connection and loading the general effect is similar to the star/inter connected star connection. That is, there is no choking effect, as the primary windings corresponding to the loaded secondaries take balancing load currents, although the primary current distribution is slightly different from that occurring with a star primary. The voltage drops in the transformer windings are those due only to the normal impedance of the transformer, while the volt age of the open phase is practically the same as at no-load. The secondary neutral is stable and can be grounded.

(m) Delta/interconnected star; single-phase load from one line to neutral With this connection and method of loading the results are similar to those obtained with the star/interconnected star, that is the primary windings corresponding to the loaded secondaries carry balancing load currents so that there is no choking effect. The voltage drops in the transformer windings are those due only to the normal impedance of the transformer, while the voltages of the open phases are practically the same as at no-load. The secondary neutral point is stable and can be grounded.

(n) Vee/vee With this connection and method of loading there is clearly no choking effect, as this is simply a question of supplying a single-phase transformer across any two lines of a three-phase generator. The voltage drops are comparable to those normally occurring, and the voltages of the open phases are practically the same as at no-load. The connection is, however, electrostatically unbalanced, and should be used only in emergency.

(o) Tee/tee; single-phase load across two lines, embracing the teaser and half the main windings With this connection and method of loading there is no choking effect, as the balancing load current in the corresponding primary windings has a perfectly free path through those windings and the two line wires to the generating source. The voltage drops in the windings are those due only to the normal impedance of the transformer, and the voltages of the open phases are practically the same as at no-load. The neutral points are stable and may be grounded.

It should always be remembered that it is impossible to preserve the current balance on the primary side of a polyphase transformer or bank and in the line wires and source of supply when supplying an unbalanced polyphase load or a pure single-phase load. In most cases the voltage balance is maintained to a reasonable degree, and the voltage drops are only greater than those occurring with a balanced load on account of the greater phase differences between the voltages and the unbalanced polyphase currents or the pure single-phase currents. The voltage drops become accentuated, of course, by the reactance of the circuit when the power factors are low.

FIG. 139 shows the phasor diagrams for typical unbalanced loading conditions on a delta/star three-phase step-down transformer where one, two and three separate single-phase loads are connected from lines to neutral. Voltage drops include transformer and cable or line drops. The triangles constructed on V' A, V' B and V' C show the resistive and reactive components of the total voltages across the respective loads. In diagram 1 the current IN in the neutral is the same as the load current IA; in diagram 2 the neutral current IN is the phasor sum of the load currents IA and IB, while in diagram 3, IN is the phasor sum of IA, IB and IC.

Top of Page PREV.   NEXT   Guide Index HOME