Industrial Power Transformers-- Transformer construction (part 2)

Home | Articles | Forum | Glossary | Books



In describing the basic principles of a two-winding transformer, it has been assumed that the windings comprise a discrete primary and secondary, each being a cylinder concentric with the wound limb of the core which provides the low reluctance path for the interlinking flux. Whether of single-phase or three-phase construction, the core provides a return flux path and must, there fore, enclose the windings, as shown in FIG. 12. As well as dictating the overall size of the transformer, the size of the two concentric windings thus dictate the size of the window that the core must provide, and hence fix the dimensions of the core which, for a given grade of core steel and flux density, will determine the iron losses. The designer must aim for as compact a winding arrangement as possible. Militating against this are the needs to provide space for cooling ducts and insulation, and also to obtain as large a copper cross-section as possible in order to minimize load losses.

FIG. 10 (a) Winding leakage flux paths - no shunts. (b) Winding leakage flux paths modified by the installation of flux shunts

FIG. 11 Flux shields for main leads

FIG. 12 Arrangement of windings within core window

The following section describes how the best compromise between these conflicting objectives is achieved in practice. First it is necessary to look more closely at the subject of load losses. By definition the load loss of a transformer is that proportion of the losses generated by the flow of load current and which varies as the square of the load. There are three categories of load loss which occur in transformers:

(1) Resistive losses, often referred to as I^2 R losses.

(2) Eddy current losses in the windings due to the alternating leakage fluxes cutting the windings.

(3) So-called stray losses in leads, core framework and tank due to the action of load-dependent stray alternating fluxes.

More will be said about the third of these later. At the moment it is appropriate to examine the losses which occur in windings. These are by far the most significant proportion.

Resistive losses, as the term implies, are due to the fact that the windings cannot be manufactured without electrical resistance and therefore cannot be eliminated by the transformer designer. There are, however, ways normally open to the designer whereby they can be reduced. These are as follows:

• Use of the lowest resistivity material. This, of course, normally means high conductivity copper.

• Use of the lowest practicable number of winding turns.

• Increasing the cross-sectional area of the turn conductor.

Minimizing the number of winding turns means that a core providing the highest practicable total flux must be used. This implies highest acceptable flux density and the largest practicable core cross-section. The penalty of this option is the increase in core size (frame size) which, in turn, increases iron weight and hence iron loss. Load loss can thus be traded against iron loss and vice versa. Increased frame size of course increases the denominator in the expression for percent reactance so that l, the axial length of the winding, must be reduced in order to compensate and maintain the same impedance, although there will be a reduction in F, the winding ampere-turns by way of partial compensation (since a reduction in the number of turns was the object of the exercise). Reduction in the winding axial length means that the core leg length is reduced, which also offsets the increase in core weight resulting from the increased frame size to some extent. There is thus a band of one or two frame sizes for which the loss variation is not too great, so that the optimum frame size can be chosen to satisfy other factors, such as ratio of fixed to load losses or transport height (since this must be closely related to the height of the core).

The penalty for increasing the cross-section of the turn conductor is an increase in winding eddy current loss. (In addition to the increase in the size of the core window and hence overall size of the core.) Eddy current loss arises because of the leakage flux cutting the winding conductors. This induces volt ages which cause currents to flow at right angles to the load current and the flux. The larger the cross-section of the turn the lower will be the resistance to the eddy current flow and hence the larger the eddy currents. The only way of increasing resistance to the eddy currents without reducing the turn cross section is to subdivide the turn conductor into a number of smaller strands or sub conductors individually insulated from each other ( FIG. 13) and transposing these along the length of the winding. The practical aspects of transposition will be described below in the section dealing with winding construction. In reality, although the winder will prefer to use a reasonably small strand size in order that he can bend these more easily around the mandrel in producing his winding, in general the greater the number of strands in parallel the more costly it becomes to make the winding, so a manufacturer will wish to limit the number of these to the minimum commensurate with an acceptable level of eddy current loss --- more on this later. In addition the extra inter-strand insulation resulting on the increased number of strands will result in a poorer winding space factor providing yet another incentive to minimize the number of strands.

FIG. 13 Section of LV and HV windings showing radial and axial cooling ducts.

As explained above, eddy currents in winding conductors are the result of leakage flux, so a reduction in leakage flux results in smaller eddy currents.

It will therefore be evident that in a transformer having a low leakage reactance, winding eddy currents are less of a problem than one with high reactance. Physically this can be interpreted by examination of Eq. (2.1), Section 2, which shows that a low leakage reactance is associated with a long or large winding axial length, l, that is a tall slim design will have less leakage flux than a short squat design and will therefore tend to have less winding eddy currents ( FIG. 14). It will also be apparent that with a tall slim arrangement the leakage flux is largely axial and it can be shown that when this is the case, it is only necessary to subdivide the conductor in the direction perpendicular to the leakage flux, that is, in the radial dimension. With the short squat winding arrangement the flux will also have a significant radial component, particularly near to the ends of the windings, so that the conductor must be subdivided additionally in the axial dimension. (In theory this would only be necessary near the ends of the windings, but it is not generally feasible to change the number of conductors mid-winding.) Another method of controlling winding eddy currents, mentioned in the previous section, is the use of flux shunts to modify the leakage flux patterns with the aim of ensuring that these do not pass through windings and where they do so their path will be predominantly axial (FIG. 10). Such measures will only tend to be economic in the larger high-impedance transformers where winding eddy currents prove particularly problematical. In practice, manufacturers find it is economic to limit eddy cur rent loss to about 25 percent of that of the resistive loss, although the degree of sophistication necessary to achieve this will vary greatly according to the circumstances and in low-impedance designs the level might easily be considerably less than this without resort to any special features.

FIG. 14 Leakage flux paths in tall and squat windings.

Winding construction

Section 3 briefly considered the requirements for copper as used in transformer windings and explained why this material is used almost exclusively.

Before discussing the details of transformer windings further it is necessary to look a little more closely at winding conductors.

Mention has already been made in the previous section that winding conductors for all transformers larger than a few kVA are rectangular in section ( FIG. 13). Individual strands must be insulated from each other within a winding conductor and, of course, each conductor must be insulated from its neighbor.

This is achieved by wrapping the strands helically with paper strip, and at least two layers are used, so that the outer layer overlaps the butt joints in the layer below. The edges of the copper strip are radiused in order to assist in paper covering. This also ensures that, where strands are required to cross each other at an angle, there will be less 'scissor action' tending to cut into the insulation. Where conditions demand it, many layers of conductor insulation can be applied and the limit to this is determined by the need to maintain a covered cross-section which can be built into a stable winding. This demands that, particularly when they have to have a thick covering of insulation, winding conductors should have a fairly flat section, so that each can be stably wound on top of the conductor below. In practice this usually means that the axial dimension of the strand should be at least twice, and preferable two and a half times, the radial dimension. Conditions may occasionally require that a conductor be wound on edge.

This can be necessary in a tapping winding. Such an arrangement can be accept able if made with care, provided that the winding has only a single layer.

Low-voltage windings

Although the precise details of the winding arrangements will vary according to the rating of the transformer, the general principles remain the same throughout most of the range of power transformers. When describing these windings it is therefore convenient to consider specific cases and it is, hope fully, also of help to the reader to visualize some practical situations.

Generally the low-voltage (LV) winding of a transformer is designed to approximately match the current rating of the available LV switchgear so that, regardless of the voltage class of the transformer, it is likely to have an LV current rating of up to about 2400 A. Occasionally this might extend to 3000 A and, as an instance of this, the majority of the UK power stations having 500 and 660 MW generating units installed have station transformers with a nominal rating of 60 MVA and rated LV windings of 11 kV, 3000 A. This current rating matched the maximum 11 kV air-break circuit-breakers which were available at the time of the construction of these stations. For the LV winding of most transformer, therefore, this is the order of the current involved. (There are transformers outside this range, of course; for an 800 MVA generator transformer, the LV current is of the order of 19 000 A.)

The voltage ratio is such that the current in the high-voltage (HV) winding is an order of magnitude lower than this, say, up to about 300 A. In most oil-filled transformers utilizing copper conductors, the current density is between 2 and 4 A/mm^2 , so the conductor section on the LV winding is of the order of, say, 50 x 20 mm and that on the HV winding, say, 12 x 8 mm. As explained in Section 1, the volts per turn in the transformer is dependent on the cross sectional area of the core or core frame size. The frame size used depends on the rating of the transformer but, since, as the rating increases the voltage class also tends to increase, the volts per turn usually gives an LV winding with a hundred or so turns and an HV winding with a thousand or more. In practice, the actual conductor sizes and the number of turns used depend on a good many factors and may therefore differ widely from the above values. They are quoted as an indication of the differing problems in designing LV and HV windings. In the former, a small number of turns of a large-section conductor are required; in the latter, a more manageable cross-section is involved, but a very much larger number of turns. It is these factors which determine the types of windings used.

The LV winding is usually positioned nearest to the core, unless the transformer has a tertiary winding (which would normally be of similar or lower voltage) in which case the tertiary will occupy this position:

• The LV winding (usually) has the lower test voltage and hence is more easily insulated from the earthed core.

• Any tappings on the transformer are most likely to be on the HV winding, so that the LV windings will only have leads at the start and finish and these can be easily accommodated at the top and bottom of the leg.

The LV winding is normally wound on a robust tube of insulation material and this is almost invariably of s.r.b.p. This material has high mechanical strength and is capable of withstanding the high loading that it experiences during the winding of the large copper-section coils used for the LV windings.

Electrically it will probably have sufficient dielectric strength to withstand the relatively modest test voltage applied to the LV winding without any additional insulation (see Section 3.4 regarding dielectric strength of s.r.b.p. tubes when used in oil-filled transformers).

The hundred or so turns of the LV winding are wound in a simple helix, using the s.r.b.p. tube as a former, so that the total number of turns occupy the total winding axial length, although occasionally, for example, where the winding is to be connected in interstar, the turns might be arranged in two helical layers so that the two sets of winding ends are accessible at the top and bottom of the leg. As explained in Section 2, winding length is dictated by the impedance required, so that the need to accommodate the total turns within this length will then dictate the dimensions of the individual turn.

FIG. 15 Transverse section of core and windings, showing axial cooling ducts above and below windings and dovetailed spacers which form radial ducts

FIG. 16 Developed section of an eight-strand conductor showing transposition of strands

Between the winding base tube and the winding conductor, axial insulation board (pressboard) strips are placed so as to form axial ducts for the flow of cooling oil. These strips are usually of a dovetail cross-section ( FIG. 15) so that spacers between winding turns can be threaded onto them during the course of the winding. Axial strips are usually a minimum of 8 mm thick and the radial spacers 4 mm. The radial cooling ducts formed by the spacers are arranged to occur between each turn or every two turns, or even, on occasions, subdividing each turn into half-turns.


It has already been explained that the winding conductor of an LV winding having a large copper cross-section, is subdivided into a number of sub conductors, or strands, to reduce eddy current loss and transposing these throughout the length of the winding. Transposition is necessary because of the difference in the magnitude of the leakage flux throughout the radial depth of the winding. If the strands were not transposed, those experiencing the higher leakage flux would be subjected to higher induced voltages and these voltages would cause circulating currents to flow via the ends of the winding where strands are of necessity commoned to make the external connections.

Transposition ensures that as nearly as possible each strand experiences the same overall leakage flux. There are various methods of forming conductor transpositions, but typically these might be arranged as shown in FIG. 16.

If the winding conductor is subdivided into, say, eight sub-conductors in the radial dimension, then eight transpositions equally spaced axially are needed over the winding length. Each of these is carried out by moving the inner conductor sideways from below the other seven, which then each move radially inwards by an amount equal to their thickness, and finally the displaced inner conductor would be bent outwards to the outer radial level and then moved to the outside of the stack.

Continuously transposed strip

Even with an arrangement of transpositions of the type described above and using many sub-conductors, eddy currents in very high-current windings (perhaps of 2000 A or greater) cannot be easily limited in magnitude to, say, 25 percent of the resistance losses as suggested above. In addition, transpositions of the type described above take up a significant amount of space within the winding. As a result, in the early 1950s, manufacturers introduced a type of continuously transposed conductor. This enables a far greater number of trans positions to be carried out. In fact, as the name suggests, these occur almost continuously in the conductor itself before it is formed into the winding. Although the 'continuous' transpositions result in some loss of space within the conductor group, this amounts to less space within the winding than that required for conventional transpositions, so that there is a net improvement in space factor as well as improved uniformity of ampere-conductor distribution. FIG. 17 shows how the continuously transposed conductor (c.t.c.) is made up. It has an odd number of strands in flat formation insulated from each other by enamel only and these are in two stacks side by side axially on the finished winding.

Transpositions are effected by the top strip of one stack moving over to the adjacent stack as the bottom strip moves over in the opposite direction. The conductor is moved sideways approximately every 50 mm along its length. In addition to the enamel covering on the individual strands, there is a single vertical paper separator placed between the stacks and the completed conductor is wrapped overall with at least two helical layers of paper in the same manner as a rectangular section conductor. Manufacture of the continuously transposed conductor involves considerable mechanical manipulation of the strands in order to form the transpositions and was made possible by the development of enamels which are sufficiently tough and resilient to withstand this. The introduction of continuously transposed strip has been particularly beneficial to the design of large transformers, which must be capable of carrying large currents, but its use is not without some disadvantages of which the following are most significant:

• A single continuously transposed conductor stack which might be up to, say, twelve strands high, and two stacks wide wrapped overall with paper, tends to behave something like a cart spring in that it becomes very difficult to wind round the cylindrical former. This problem can be limited by the use of such strip only for large-diameter windings. It is usual to restrict its use to those windings which have a minimum radius of about 30 times the overall radial depth of the covered conductor.

• When the covered conductor, which has significant depth in the radial dimension, is bent into a circle, the paper covering tends to wrinkle and bulge. This feature has been termed 'bagging'. The bagging, or bulging, paper covering can restrict oil flow in the cooling ducts. The problem can be controlled by restricting the bending radius, as described above, and also by the use of an outer layer of paper covering which has a degree of 'stretch' which will contain the bagging such as the highly extensible paper described in Section 3. Alternatively some allowance can be made by slightly increasing the size of the ducts.

• Joints in continuously transposed strip become very cumbersome because of the large number of strands involved. Most responsible manufacturers (and their customers) will insist that a winding is made from one length of conductor without any joints. This does not, however, eliminate the requirement for joints to the external connections. It is often found that these can best be made using crimped connectors but these have limitations and very careful control is necessary in making the individual crimps.

• A high degree of quality control of the manufacture is necessary to ensure that defects in the enamel insulation of the individual strands or metallic particle inclusions do not cause strand-to-strand faults.

In the late 1990s a variant of continuously transposed conductor was introduced for which the paper outer wrapping is replaced by a netting tape.

FIG. 17 Continuously transposed conductor

Because continuously transposed cables generally find their use in LV windings for which the voltage between adjacent turns is modest, the outer paper wrapping is largely providing stability for the conductor bundle rather than insulation between turns. If, for this outer wrapping, the paper is replaced by netting, the thermal drop which occurs across the paper can be eliminated without sacrificing conductor stability. This type of c.t.c. has found quite wide spread use where voltage consideration allows it. An additional feature that can be incorporated is to utilize an epoxy resin for the strand coating that does not totally cure until the transformer windings are processed, so that conductor bundles acquire their maximum strength when the coils have been wound, thus greatly improving their short-circuit strength.

High-voltage windings

Mention has already been made of the fact that the high-voltage (HV) winding might have 10 times as many turns as the low-voltage ( LV) winding, although the conductor cross-sectional area is considerably less. It is desirable that both windings should be approximately the same axial length subject to the differing end insulation requirements, see below, and, assuming the LV winding occupied a single layer wound in a simple helix, the HV winding would require 10 such layers. A multilayer helical winding of this type would be somewhat lacking in mechanical strength however, as well as tending to have a high voltage between winding layers. (In a 10-layer winding, this would be one-tenth of the phase voltage.) HV windings are therefore usually wound as 'disc windings'. In a disc winding, the turns are wound radially outwards one on top of the other starting at the surface of the former. If a pair of adjacent discs are wound in this way the cross-over between discs is made at the inside of the discs and both 'finishes' appear at the outer surfaces of the respective discs. The required number of disc pairs can be wound in this way and then connected together at their ends to form a complete winding. Such an arrangement requires a large number of joints between the pairs of discs (usually individual discs are called sections) and so has been largely superseded by the continuous disc winding.

This has the same configuration when completed as a sectional disc winding but is wound in such a way as to avoid the need for it to be wound as separate disc pairs. When the 'finish' of a disc appears at the outside radius, it is taken down to the mandrel surface using a tapered curved former. From the surface of the mandrel, a second disc is then built up by winding outwards exactly as the first. When this second complete disc has been formed, the tension is taken off the winding conductor, the taper former removed and the turns laid loosely over the surface of the mandrel. These turns are then reassembled in the reverse order so that the 'start' is the crossover from the adjacent disc and the 'finish' is in the centre at the mandrel surface. The next disc can then be built upwards in the normal way. A section of continuous disc winding is shown in FIG. 18.

FIG. 18 Arrangement of continuous disc winding

The operation as described above has been the method of producing continuous disc windings since they were first introduced in about the 1950s. Whilst it may sound a somewhat complex procedure to describe, a skilled winder makes the process appear simple and has no difficulty in producing good quality windings in this way. There are, however, disadvantages of this method of winding. The most significant of these is associated with the tightening of those discs which must be reversed. After reassembling the individual turns of these discs to return the winding conductor to the surface of the mandrel, a procedure which requires that the turns are slack enough to fit inside each other, the winder must then re-tighten the disc to ensure that the winding is sufficiently stable to withstand any shocks due to faults or short-circuits in service. This tightening procedure involves anchoring the drum from which the conductor is being taken and driving the winding lathe forward. This can result in up to a meter or so of conductor being drawn from the inside of the disc and as this slack is taken up the conductor is dragged across the dovetail strips over which the disc winding is being wound. To ensure that the conductor will slide easily the surface of the strips is usually waxed, but it is not unknown for this to 'snag' on a strip damaging the conductor covering. And, of course this damage is in a location, on the inside face of the disc, which makes it very difficult to see.

The other disadvantage is minor by comparison and concerns only the labor cost of making a continuous disc winding. The process of laying out the disc turns along the surface of the mandrel and reassembling them in reverse order requires skill in manipulation and it is the case that a second pair of hands can be beneficial. In fact when labor costs were very much lower than at present it was standard practice for a winder engaged in producing a continuous disc winding to have the services of a laborer throughout the task.

Nowadays such practices are considered to be too costly but nevertheless in many organizations the winder will seek the assistance of a colleague for the more difficult part of the process, which also has cost implications.

FIG. 19 Winding in progress - horizontal lathe (Peebles Transformers)

FIG. 20 Winding in progress - vertical lathe (Peebles Transformers)

Both of the above problems associated with the manufacture of continuous disc windings have been overcome by the introduction of the vertical winding machine which has been used by some manufacturers for many years but whose use became more widespread in the 1980s. From the earliest days of transformer manufacture it has been the practice to wind conductors around horizontal mandrels of the type shown in FIG. 19. FIG. 20 shows a modern vertical axis machine which has replaced some of the horizontal axis types in the winding shops of some of the more advanced manufacturers of large high-voltage transformers. On these machines production of continuous disc windings is a much more straightforward and reliable procedure. Using such a machine, the first disc is wound near to the lower end of the mandrel building up the disc from the mandrel surface, outwards, in the normal manner. Then the next disc is wound above this, starting from the outer diameter, proceeding inwards in a conical fashion, over a series of stepped packing pieces of the type shown in FIG. 21. When this 'cone' has been completed, taking the conductor down to the mandrel surface, the packing pieces are removed allowing the cone to 'collapse' downwards to become a disc. This procedure requires only a very small amount of slackness to provide sufficient clearance to allow collapse of the cone, so the tightening process is far less hazardous than on a horizontal machine and furthermore the process can easily be carried out single handed. Vertical machines allow the production of windings of considerably superior quality to those produced using the horizontal type but their installation requires considerably greater capital outlay compared with the cost of procuring and installing a horizontal axis machine.

The HV winding requires space for cooling oil flow in the same way as described for the LV winding and these are again provided by using dovetail strips over the base cylinder against the inner face of the discs and radial spacers interlocking with these in the same way as described for the LV. Radial cooling ducts may be formed either between disc-pairs or between individual discs.

Before concluding the description of the various types of HV winding it is necessary to describe the special type of layer winding sometimes used for very high-voltage transformers and known as a shielded-layer winding.

Despite the disadvantage of multilayer HV windings identified above namely that of high voltages between layers and particularly at the ends of layers; electrically this winding arrangement has a significant advantage when used as a star-connected HV winding having a solidly earthed star point and employing non-uniform insulation. This can be seen by reference to FIG. 22. If the turns of the winding are arranged between a pair of inner and outer 'shields,' one connected to the line terminal and the other to earth, the distribution of electromagnetic voltage within the winding will be the same as the distribution of capacitance voltage if the outer and inner shields are regarded as poles of a capacitor, so that the insulation required to insulate for the electromagnetic voltage appearing on any turn will be the same as that required to insulate for a capacitative voltage distribution. This provides the winding with a high capability for withstanding steep-fronted waves such as those resulting from a lightning strike on the line close to the transformer. (The next section of this section deals in detail with this subject.) FIG. 22(a) shows the ideal arrangement for a shielded-layer winding. FIG. 22(b) shows how such a winding might typically be manufactured in practice. As will be apparent from FIG. 22(b), this type of winding has very poor mechanical strength, particularly in the axial direction, making it difficult to withstand axial clamping forces. There is also a problem associated with the design of the shields. These must be made of very thin conducting sheet, otherwise they attract a high level of stray loss and additionally the line-end shield, being heavily insulated, is difficult to cool, so there can be a problem of local overheating. The shields must have an electrical connection to the respective ends of the winding and making these to flimsy metallic sheets in such a manner that they will withstand a lifetime of high 100 Hz vibration, is not easy. These difficulties and in particular the complex insulation structure required, make this type of winding very costly to manufacture. Consequently designers have concentrated on improving the response of disc windings to steep-fronted waves. This work has been largely successful in recent years, so that shielded layer type windings are now rarely used.


FIG. 22 Shielded-layer windings

Insulation between layers tapers - maximum at 'open' end, minimum at 'crossover' end

Insulation at layer ends increases with distance in winding from neutral end


Outer electrostatic shield

Inner electrostatic shield


(a) Theoretically ideal arrangement

Insulation wraps between layers

Line Outer electrostatic shield Inner electrostatic shield Neutral

(b) Typical arrangement used in practice


Tapping windings

Thus far it has been assumed that power transformers have simply a primary and secondary winding. However, practically all of them have some form of tapping arrangement to allow both for variations of the applied voltage and for their own internal regulation. In the case of distribution and small auxiliary transformers these tappings will probably allow for _5 percent variation, adjustable only off-circuit. On larger transformers tappings of _10 percent, or more, might be provided selectable by means of on-load tapchangers. More will be said later about the subject of tappings and tapchangers. However, it is convenient at this stage to describe the tap windings themselves.

Most power transformers have the tappings in the HV winding for two reasons. Firstly, it is convenient to assume that the purpose of the tappings is to compensate for variations in the applied voltage which, for most transformers, except generator transformers, will be to the HV winding. (Generator transformers are a special case and will be discussed more fully in Section 7.) As the applied voltage increases, more tapping turns are added to the HV winding by the tapchanger so that the volts per turn remain constant, as does the LV winding output voltage. If the applied voltage is reduced, tapping turns are removed from the HV winding again keeping the volts per turn constant and so retaining constant LV voltage. From the transformer design point of view, the important aspect of this is that, since the volts per turn remains constant, so does the flux density. Hence the design flux density can be set at a reasonably high economic level without the danger of the transformer being driven into saturation due to supply voltage excursions.

The second reason for locating tappings on the HV side is that this winding carries the lower current so that the physical size of tapping leads is less and the tapchanger itself carries less current.

Since the tappings are part of the HV winding, frequently these can be arranged simply by bringing out the tapping leads at the appropriate point of the winding. This must, of course, coincide with the outer turn of a disc, but this can usually be arranged without undue difficulty.

In larger transformers, the tappings must be accommodated in a separate tapping winding since the leaving of gaps in one of the main windings would upset the electromagnetic balance of the transformer to an unacceptable degree so that out of balance forces in the event of an external fault close to the transformer could not be withstood. The separate tapping winding is usually made the outermost winding so that leads can be easily taken away to the tapchanger.

The form of the winding varies greatly and each of the arrangements has their respective advantages and disadvantages. Before describing separate tapping windings further it should be noted that it is always significantly more costly to place the taps in a separate layer because of the additional interlayer insulation that is required. It is always preferable therefore to accommodate the taps in the body of the HV if this is at all possible.

FIG. 23 Interleaved helical tapping winding having four taps in parallel of five turns per tap.

One common arrangement for a separate tapping winding is the multistart or interleaved helical winding. This is shown diagrammatically in FIG. 23.

These windings usually occupy two layers but may occasionally have four layers. The arrangement is best described by using a practical example.

Consider a transformer with a 275 kV star-connected HV winding having a tapping range of _10 to _20 percent in 18 steps of 1.67 percent per step. It has already been suggested that a typical HV winding might have about 1000 turns total. In general, transformers for higher voltages, particularly at the lower end of the MVA rating range, tend to be on smaller frames in relation to their class so that the number of turns tends to be higher than the average. In this example, and being very specific, assume that the HV winding has 1230 turns on principal tap, so that each tap would require:

167/100 x 1230 = 20 54 turns

This means that the tapping winding must provide approximately twenty and a half turns per tap. Of course, half turns are not possible so this would be accommodated by alternating 20 and 21 turn tapping steps. (In practice the designer would need to be satisfied that his design complied with the requirements of BS EN 60076, Part 1, as regards tolerance on voltage ratio for all tap positions. This might necessitate the adjustment of the number of turns in a particular tap by the odd one either way compared with an arrangement which simply alternated twenty and twenty-one turn tapping steps.) One layer of the tapping winding would thus be wound with nine (i.e. half the total number of taps) sets of conductors in parallel in a large pitch helix so that, say, twenty turns took up the full axial length of the layer. There would then be an appropriate quantity of interlayer insulation, say duct-wrap-duct, the ducts being formed by the inclusion of pressboard strips, followed by a further layer having nine sets of twenty-one turns in parallel. The winding of the layers would be in opposite senses, so that, if the inner layer had the starts at the top of the leg and finishes at the bottom, the outer layer would have starts at the bottom and finishes at the top, thus enabling series connections, as well as tapping leads, to be taken from the top and bottom of the leg. (As stated earlier, the voltage induced in all turns of the transformer will be in the same direction regardless of whether these turns are part of the LV, HV or tapping windings. In order that these induced voltages can be added together, all turns are wound in the same direction. This difference in sense of the windings therefore depends upon whether the start is at the top of the leg or at the bottom, or, since most windings are actually wound on horizontal mandrels, whether the start is at the left or the right. In the case of buck/boost tapping arrangements - the winding output voltage is in some cases reduced by putting in-circuit tappings in a subtractive sense, that is 'buck,' and in other cases increased by putting in-circuit tappings in an additive sense, that is 'boost.' The windings them selves are, however, still wound in the same direction).

The helical interleaved tap winding arrangement has two advantages:

(1) By distributing each tapping along the total length of the leg a high level of magnetic balance is obtained whether the taps are in or out.

(2) Helical windings with a small number of turns are cheap and simple to manufacture.

It unfortunately also has disadvantages, the first of which is concerned with electrical stress distribution and is best illustrated by reference to FIG. 24.

Manufacturers design transformers in order to meet a specified test condition so it is the electrical stress during the induced overvoltage test which must be considered. A transformer having an HV voltage of 275 kV may be subjected to an induced overvoltage test at 460 kV and it is permitted in BS EN 60076 to induce this test voltage on the maximum plus tap, that is _10 percent in this example, so that 460 kV must be induced in 110 percent of the winding turns.


FIG. 24 Arrangement of two-layer helical interleaved tapping winding

Part axial section of tapping windings

(b) Physical arrangement


FIG. 24(b) shows a part section of the tapping layers. It will be apparent from this that it is not advisable to allocate the tapping sections in numerical order, otherwise in the outer layer at the end of the first turn, tapping 1 will be immediately adjacent tapping 17 and in the inner layer, tapping 2 will be adjacent tapping 18. The diagram shows one possible way of distributing the taps so as to reduce the voltage differences between turns which are physically close together. In this arrangement the start of tapping 17 is separated from the start of tapping 1 by the width of three turns. The test voltage appearing between the start of tapping 1 and the start of tapping 17 is that voltage which is induced in 16 tapping steps, which is,

16 167 100 460 110 100 111 74 ____ .

. kV, approximately

The width of three turns depends on the total length available for the tap ping layer. On a fairly small 275 kV transformer this could be as little as 2 m. In layer one 9 x 20 turns must be accommodated in this 2 m length, so three turns occupy,

3 2000 920 33 33 _ _ _ .mm

and the axial creepage stress is thus

111 74 33 33 335

. _ kV/mm

which is unacceptably high.

The situation could be greatly improved by opting for four layers of taps rather than two, arranged so that no more than half the tapping-range volts appeared in the same layer.

Whilst the numbers quoted above do not relate to an actual transformer they do illustrate the problem, also showing that design problems frequently arise in very high-voltage transformers at the lower end of the MVA rating band applicable to the voltage class in question.

Another way of resolving this problem would be to use either a coarse/fine or a buck/boost tapping arrangement. These require a more sophisticated tap changer but allow the tap winding to be simplified. They can be explained by reference to FIG. 25. With a coarse/fine arrangement (FIG. 25(a)) the tapping winding is arranged in two groups. One, the coarse group, contains sufficient turns to cover about half of the total tapping range and is switched in and out in a single operation, the other, the fine group, is arranged to have steps equivalent to the size of tapping step required and is added and subtracted sequentially either with or without the coarse group in circuit.

Physically, the coarse group would occupy a third helical layer and no more than half tapping-range volts would appear in one layer. With the buck/boost arrangement, the tapchanger is arranged to put taps in and out with such a polarity as to either add or subtract to or from the voltage developed in the main HV winding. Again, with this arrangement the taps could be contained in two helical layers so that these did not contain more than half tapping-range volts.

The second disadvantage of the helical tapping arrangement concerns its mechanical strength. Under short-circuit conditions an outer winding experiences an outward bursting force. Such a winding consisting of a small number of turns wound in a helix does not offer much resistance to this outward bursting force and requires that the ends be very securely restrained to ensure that the winding does not simply unwind itself under the influence of such a force. The twenty/twenty-one turns in the example quoted above can probably be adequately secured, however as the transformer gets larger (and the magnitude of the forces gets greater also) the frame size will be larger, the volts per turn increased and the turns per tap proportionally reduced, so the problem becomes more significant.


FIG. 25

Line; Main high-voltage winding 85% of total turns; Coarse tapping winding 15% of total turns; Fine tapping winding 9 × 1.67% of total turns

(a) Coarse/fine arrangement to provide a tapping range of ±15% in 18 steps of 1.67% per step

Line; Main high-voltage winding 100% of total turns; Buck/boost tapping winding 9 × 1.67% of total turns to be added or subtracted

(b) Buck/boost tapping arrangement to provide a tapping range of ±15% in 18 steps of 1.67% per step


The most common alternative to the use of interleaved helical tapping windings is to use disc windings. These at least have the advantage that they can be accommodated in a single layer. The number of turns in an individual tapping section must ideally be equal to an even number of discs, usually a single disc pair. Tapping leads are thus connected between disc pairs so the disc pairs may be joined at this point also, that is, it is just as convenient to make up the winding from sectional disc pairs as to use a continuous disc. This former method of manufacture is therefore often preferred. Another advantage of using a disc winding is that the discs can be arranged in the normal tapping sequence so that the full volts across the tapping range is separated by the full axial length of the tap winding.

A third possibility for the tapping winding is to utilize a configuration as for the disc wound taps described above but to nevertheless wind each tap section as a helix. This arrangement might be appropriate at the lower end of the size range for which a separate tapping winding is necessary so that the radial bursting forces under short-circuit are not too great. In the example quoted above, a figure of around 20 turns per tap would lend itself ideally to a disc arrangement having 10 turns per disc, that is, 20 turns per disc pair. The example quoted was however quite a high-voltage transformer. Often the number of turns per tap will be very much less, possibly as few as six or seven. Such a small number does not lend itself so well to a pair of discs and hence a helical arrangement must be considered, which raises the problem of accommodating the necessary number of turns in a single layer. It is here that it might be necessary to wind the conductor on edge. As previously stated, this can be done provided the winding is single layer and of a reasonably large diameter. In fact this might produce a stiffer winding, more able to withstand the radial bursting forces than one in which the conductor was laid flat.


Top of Page

PREV.   NEXT   Guide Index HOME