|Home | Articles | Forum | Glossary | Books|
Previous sections have examined those features which are common to the majority of power transformers. This section takes a closer look at specific classes of transformer to identify the special aspects which are required for each particular application.
It is appropriate to commence this examination of specific transformer types by considering generator transformers and in this context these are taken to mean those step-up transformers directly connected to a generator output terminals in large generating stations. The break-up of the state monopoly and privatization of electricity supply arrangements in the UK since 1990 has lead to an increase in the number of individual combined heat and power, wind power and other installations of embedded generation having ratings ranging from a few hundred kW to a few MW and connected to the local distribution network at 3.3 or 11 kV. It is generally uneconomic to make any special provision in the design of generator transformers for these small installations and usually normal distribution transformer design practices are acceptable, although it may be necessary to incorporate on-load tapchangers to ensure the generator output can be passed into the local network.
Most large modern generators are designed for operation at voltages between about 11 and 30 kV. The generator designer aims to use as high a voltage as practicable so as to limit the stator current necessary to achieve the required output.
Increase in the machine voltage significantly beyond the minimum necessary requires more insulation on the generator stator windings, thus increasing its size and cost. Hence machines of around 150 MW generally operate at 11 kV and have line currents of around 9260 A at 0.85 power factor, whilst those at 660 MW, in the UK, usually operate at 23.5 kV with line currents of about 19 000 A at 0.85 power factor. Since the generation is usually located away from the load centers for economic transmission it is necessary to greatly reduce these output currents, so at most power stations the generator output voltage is immediately stepped up by means of a generator step-up (GSU) transformer, and nowadays power stations are designed on the unit principle, so that each generator will have its own dedicated step-up transformer. In the case of combined cycle gas turbine (CCGT) plants this can mean having four step-up transformers, three associated with the gas turbines (each rated about 150 MW) and one with the steam turbine (rated about 250 MW) on a single unit. All four of these transformers may be connected together via isolators at 400 kV and switched by a single 400 kV circuit breaker to the 400 kV transmission system.
Generator transformers thus frequently have a wide voltage ratio, maybe 11/420 kV in the case of the abovementioned 150 MW unit. The rating must be sufficient to allow the generator to export its full megawatt output at 0.85 power factor lagging or 0.95 power-factor leading or, alternatively, half of full megawatt output at 0.7 power-factor lead, so that for a generator output of 150 MW, the transformer rating will need to be 150/0.85, which is approximately 176 MVA, and for 660 MW it must be 660/0.85 which is approximately 776 MVA.
In the UK some of the first CEGB 660 MW generators were designed to deliver full output at 0.8 power factor which, after subtracting the power requirements absorbed by the unit auxiliaries, led to a maximum output power of 800 MVA, so that for the sake of standardization the CEGB generator transformer rating was generally fixed at that level.
The important criteria which influence the generator transformer design are as follows:
• The HV volts are high - often 400 kV nominal or higher.
• The LV current is high - about 19 000 A for an 800 MVA, 23.5 kV transformer.
• The impedance must be lower than that resulting from the simplest design for this rating -- a figure of about 16 percent is generally specified over a wide range of ratings and variation with tap position must be kept to a minimum to simplify the system design and operation.
• An on-load tapchanger is generally used to allow for variation of the HV system volts and generator power factor. LV volts will generally remain within +/-5 percent under the control of the generator automatic voltage regulator (AVR). It should be noted that there is an alternative view prevailing amongst some utilities who see on-load tapchangers as a source of unreliability in an area where high availability and load factor is paramount.
These utilities prefer to have a very limited range of off-circuit taps on the generator transformer, say +/-2.5 and +/-5 percent, and control unit voltage and power factor entirely by means of the generator AVR. This approach, however, requires a very much larger and more costly AVR, and the holders of this viewpoint tend to be in the minority.
• The transport weight must be within the limits laid down by the appropriate transport authorities and the available transport vehicles.
• Transport height must meet the maximum limit permitted by the need to pass under any road bridges on the route to site or to the port of loading onto a vessel, if the transformer is to be transported by sea.
• Reliability and availability must be as high as possible, since without the generator transformer, unit output cannot be made available to the transmission network.
There are also a number of other criteria which although less important will also have a bearing on the design. These are:
• Because of the high-load factor, both load and no-load losses must be as low as possible.
• In view of the direct connection to the extra-high-voltage (EHV) transmission system, a high impulse strength is required.
• Noise level must be kept below a specified level.
• Very little overload capability is necessary since generator output cannot normally be increased, although most gas-turbine generators achieve maximum output with an inlet-air temperature of 0ºC so that generator transformers associated with these often have a designated rating at 0ºC which is a few percent higher than the EN 60076 rating. Many fossil fuelled steam turbines are capable of about 4 percent increase in output at 0ºC with some loss of efficiency by bypassing part of the feed-heater train so that generator transformers for these units are required to match this output.
Generator transformers may be subjected to sudden load rejection due to operation of the electrical protection on the generator. This can lead to the application of a sudden overvoltage to the terminals connected to the generator. Very little documentary evidence is in existence concerning the likely magnitude of the overvoltage since monitored full-load rejections on generators are not the type of tests which are carried out every day. CEGB carried out some testing in the 1970s and concluded that the likely magnitude was of the order of 135 percent of normal voltage and this might persist for up to 1 minute. As a result of this testing the CEGB specification for generator transformers contained a clause requiring that this level and time of overvoltage should be withstood without damage. Manufacturers of generator transformers maintained that before the overvoltage had reached this level the core would become saturated and would thus draw a very large magnetizing current at a very low power factor which would have the effect of pulling the generator voltage down. The feeling was that the end result would be a magnitude of about 125 percent volts persisting for something less than 1 minute. EN 60076, Part 1, specifies that generator transformers subject to load rejection should withstand 140 percent volts for 5 seconds. This is probably pessimistically high but for an optimistically short duration.
General design features
The extensive list of required characteristics given above places considerable constraints on the design of the generator transformer. For even quite modestly rated generator transformers in the UK, where transport is invariably by road, the most limiting factor is that of transport weight coupled with travelling height. For a transformer of 400 kV, the high voltage requires large internal clearances which means increasing size and, as can be seen from the expression for leakage reactance increased HV to LV clearance has the effect of increasing the reactance, and hence the impedance. This tendency to increase reactance would normally be offset by an increase in the axial length of the winding but, for a large generator transformer, even when measures are adopted such as reduced yoke depth for the core, necessitating outer return limbs resulting in a five-limb core, the stage is soon reached where further increases cannot be obtained because of the limit on transport height.
A significant reduction in leakage reactance for given physical dimensions can be obtained by adopting an arrangement of windings known as 'split concentric.' This is shown in FIG. 1(a). The HV winding has been split into two sections, with one placed on either side of the LV winding. This is not too inconvenient for a transformer with graded HV insulation, since the inner HV winding is at lower voltage and can therefore be insulated from the grounded core without undue difficulty. The reason why this arrangement reduces leak age reactance can be seen from FIG. 1(b) and (c), which give plots of leakage
flux both for simple-concentric and split-concentric arrangements having the same total m.m.f. It can be shown that the leakage reactance is proportional to the area below the leakage flux curve, which is significantly less in the split concentric design. The price to be paid for this method of reducing the leakage reactance which, in reality, means significantly reducing the physical size for a given rating, is the complexity involved in the increased number of windings, increased number of leads and increased sets of interwinding insulation. For simplicity, the tapping winding has not been shown in FIG. 1(c). With this split-concentric arrangement, it is usual to locate the taps in a separate winding below the inner HV winding. As taps are added or removed, the ratio of the HV split is effectively varied and this has the effect of producing relatively large changes in leakage reactance. This undesirable feature is a further disadvantage of this form of construction.
Throughout the 1960s, at the time of building most of the CEGB 500 MW units, the split-concentric arrangement was the most common form adopted for 570 and 600 MVA three-phase generator transformers. This enabled these transformers to be transported as three-phase units within limits of about 240 tonnes transport weight and 4.87 m travelling height, albeit most of them had very high flux densities, a nominal 1.8 Tesla, and relatively high losses in order to keep the material content to the minimum. It was still necessary to resort to reduced yoke depths and five-limb cores to meet the maximum height limit. The use of aluminum tanks was another measure employed, and, in the case of one manufacturer, a very sophisticated lightweight steel tank having a girder frame and stressed-skin construction was used. In fact, a transformer of 735 MVA, three phase, was transported within these height and weight limits to Hartlepool power station, although its HV winding was only 275 kV. However, at the time of the adoption of the 660 MW unit size for the Drax power station at the end of the 1960s, it was decided to make the transition to single-phase units. These have many advantages and will be described in greater detail in the following section.
Single-phase generator transformers
With the adoption of single-phase construction, transport weight for 800 MVA and probably even larger transformers ceases to impose such a major constraint on the transformer designer. Travelling height continues to impose some restriction, but the designer is usually able to deal with this without undue difficulty, and as will become clear, single-phase construction provides even more scope than does three-phase for reduced yoke depths.
It is now normal practice to design for a nominal flux density at nominal voltage of 1.7 Tesla since this offers a better margin of safety below saturation than the figure of 1.8 Tesla previously allowed and it is good practice, in addition, to specify that this should not exceed 1.9 Tesla at any point in the core under any operating condition. When originally framed, this specification clause was intended to allow for some local increase in flux density in the region of core bolt holes. Now that boltless cores are normally used this reason is eliminated, but since it can never be assumed that the distribution of flux will always be uniform at all points in the core, some margin is nevertheless desirable. As indicated above, the generator AVR will normally ensure that the voltage applied to the LV terminals will remain within the range of +/-5 percent of nominal so it may appear that variation in flux density due to possible variation of applied voltage will likewise not be very great. However it is necessary to consider the operating situation.
In order that the machine can export maximum rated VArs at 0.85 power-factor lag under all normal conditions of the 400 kV transmission network, that is at a voltage of up to 5 percent high and a frequency down to 49.5 Hz in the UK, it will be necessary for its excitation to be increased to such a level as to over come the regulation within the transformer at this condition and still have a high enough voltage at the HV terminals to match that of the 400 kV network at its highest normal voltage condition of 420 kV. Regulation within the transformer at a load of a times full-load current may be calculated by reference to Eq. (1.7):
percentage regulation cos sin __
_ aV V a RX () ff 2 200 2 () VV XR cos sin ff _
For a large generator transformer VR will be so small as to be negligible, so this expression may be rewritten:
percentage regulation sin __ aV a V X X f f 2 2 20 (cos) 0
Taking VX equal to 16 percent, the regulation at full load (a =1) may thus be calculated at 9.35 percent. Overall, therefore, under these conditions the generator output voltage must be increased by (9.35 + 5) = 14.35 percent above nominal in order to export the required VArs. This voltage applied to the generator transformer would represent too great an increase above normal and to avoid the need for this it is normal practice to specify an open circuit HV voltage for the generator transformer of 432 kV, that is 8 percent above nominal. The generator would thus only be required to increase its voltage by (14.35 - 8) = 6.35 percent in order to produce the required voltage at the transformer HV terminals. Thus the transformer flux density would be increased by this amount plus any increase resulting from possible reduced frequency of the network, that is it could be as low as 49.5 Hz, or 1 percent low. Total overall increase in flux density is thus (6.35 + 1) = 7.35 percent. This would result in a nominal flux density under this condition of 1.0735 x 1.7 = 1.825 Tesla.
It will be noted that in the above discussion no allowance has been made for tappings on the transformer HV winding. In reality, however, flexibility of operation will be assisted by the use of these. Typically, in the UK, a 400 kV generator transformer might have tappings on the HV winding of +6.67 to -13.33 percent in 18 steps of 1.11 percent. (Some of the first UK 800 MVA generator transformers were provided with HV tappings at +2 to -18 percent in 18 steps but these were found to be somewhat limiting in VAr exporting capability under some conditions at some locations on the system.) Normally these would be used for control of VAr import and export so that the 6.35 percent increase in output voltage called for in the above example could be achieved by tapping up by the appropriate amount on the transformer tapchanger and would normally be done in this way. As mentioned above, by so doing, the necessary continuous rating of the generator excitation system is greatly reduced.
FIG. 2 shows various arrangements of core and windings that can be adopted for single-phase transformers. In FIG. 2(a), the core has one wound limb and two return yokes. Alternatively, both limbs could be wound, as shown in FIG. 2(b), but this increases the cost of the windings and also the overall height, since the yoke must be full depth. It would be possible to reduce the yoke depth by providing two return yokes as in FIG. 2(c) but this adds further complexity and is therefore rarely advantageous. Some manufacturers reduce the yoke depth still further by using four return yokes (FIG. 2(d)). FIG. 3 shows the core and windings of a single-phase 267 MVA, 23.5/432 kV generator transformer having one-limb wound and with two return yokes. This has a transport weight of 185 tonnes and a travelling height of 4.89 m.
A further benefit of single-phase construction is that should a failure occur, it is very likely to affect one phase only, so only that phase need be replaced and, being more easily transported, spare single-phase units can be kept at strategic central locations which can then serve a number of power stations.
This led to the concept of interchangeable single-phase generator transformers which were developed for the majority of the CEGB 660 MW units. For this the electrical characteristics of impedance and voltage ratio must be closely matched on all tap positions and, of course, the physical sizes and arrangements of connections for HV and LV windings must be compatible. Each single-phase unit must have its own on-load tapchanger, driven from a single drive mechanism mounted at the end of the bank. Tapchangers must thus be compatible in that all must drive in the same sense and all must have the same number of turns for a tapchange. The tapchangers must be located so that the drive shafts will align. The location of inlet and outlet cooling oil pipes must correspond on all units. FIG. 4 shows the arrangement of an 800 MVA bank of single-phase units and details all the items which must align to provide complete interchangeability.
Both ends of each winding of a single-phase unit are brought out of the tank so that the HV neutral has to be connected externally, as well as the LV delta.
The former is arranged by bringing the groundy end of each HV winding to a bushing terminal mounted on the top of the tapchangers. These can then be solidly connected together by means of a length of copper bar, of adequate cross-sectional area for the HV current at full rating and which is taken via any neutral current transformer and connected to the station ground.
On the early CEGB single-phase banks, the LV delta was connected by means of an oil-filled delta box which spanned the three tanks. This can be identified in FIG. 4. It was split internally into three sections by means of barrier boards so that the oil circuits of the three tanks were kept separate. It was recognized that phase-to-phase faults were possible within the delta box and that greater security could be obtained by the use of an external air-insulated phase-isolated delta which was, in fact, an extension of the generator main connections. This was subsequently adopted as the standard arrangement, so that the LV connections to each single-phase unit are made via a pair of bushings mounted on a pocket on the side of the transformer tank. The use of air-insulated phase-isolated delta connections has the added advantage that it enables the oil circuits of the three phases to be kept entirely separate, so that, in the event of a fault on one phase, there will be no contamination of the oil in the other phases.
The HV connections may be via air bushings or SF6-insulated metal-clad trunking. The interface is therefore the mounting flange on the tank cover, as can be seen from FIG. 4.
FIG. 5 shows an 800 MVA generator transformer bank installed at the Drax power station before erection of the acoustic enclosure.
The procedure for commissioning generator transformers is similar to that adopted for other transformers and generally as described in Section 5.4, but one aspect deserving of some extra attention arises from the very high cur rents which are normally associated with generator transformer LV windings and connections. These can lead to additional stray fluxes in the vicinity of LV terminals and busbars which can give rise to severe local overheating. It is most important to ensure that there are no closed conducting paths encircling individual phase conductors and that any cladding surrounding LV busbars and connections does not form closed paths permitting circulating currents. Such structures must, of course, be solidly bonded to ground but care needs to be taken to ensure that the bonding is at one single point only for each individual section of the structure.
It is also important to pay particular attention to the connections between the LV bushings of the transformer and its external busbar connections from the generator. Nowadays generator busbars are generally of phase-isolated construction with aluminum conductors and an external aluminum sheath. To allow for relative movement between the transformer and the busbars these are connected by means of bolted flexible links, or connectors, bridging between two sets of palms, one on the busbars and a similar set on the transformer bushings. In the UK these flexible links consist of woven copper braided connectors; in some countries foil laminates are preferred. Woven braid has the advantage that it allows relative movement in three perpendicular directions, whereas a laminate only allows movement in two directions. Users of laminates consider that braids can become abraded and ultimately disintegrate due to vibration and work hardening of the copper. CEGB experience was that braids are superior to laminates provided the connection is carefully designed to allow sufficient flexibility in the braids and to ensure that none of them are overloaded in terms of current density.
Current sharing is a very important aspect and, to provide this, particularly for currents in excess of 10 000 A it is desirable that the connection palms are arranged in an as close as possible to a circular configuration as explained in the earlier section describing high current bushings. The four sides of a square represents a satisfactory approximation to this at 10 000 A, but this needs to be octagonal at 14 000 A. Such arrangements ensure that all the current paths through the connectors have equal reactance, which is more important in determining equal current sharing than equal resistance. It is nevertheless important to obtain a low contact resistance for each joint. This is obtained by ensuring that these have adequate contact surface area -- generally this needs to be 5 or 6 times the conductor cross-sectional area, surfaces must be clean and
flat and adequate contact pressure must be maintained. The design of and practical aspects associated with the making of heavy current joints is outside the scope of this guide and those seeking further information should consult works such as Copper for Busbars  or Modern Power Station Practice .
On a completely new installation the turbo-generator will need to have an initial proving run and this usually involves running it for 72 hours at continuous full load. The opportunity should be taken during this run to monitor temperatures around the generator transformer, particularly in the vicinity of the LV busbars and connections. The usual procedure is to use temperature monitoring 'stickers' attached to terminal palms and braided flexible connectors which are inside enclosures and thus cannot be approached when the unit is in operation. Temperatures on the outer surfaces of enclosures and other such accessible locations may be monitored whilst the unit is in operation by means of a contact thermometer or using infrared thermal imaging equipment. It is usual to specify that the temperature of terminal palms should not exceed 90ºC for plain copper or 110ºC if both palm and connector is silver plated. The temperature of external cladding should not exceed 80ºC in those locations where an operator could make accidental contact with it from ground level.
Performance and reliability
The generator transformer is the one transformer on a power station for which no standby is provided. It must be available for the generator output to be connected to the transmission network. For a large high efficiency generating unit in the UK, high reliability is required. If its output is not available, this results in loss of earnings for the generating company and necessitates the running of less efficient plant which is likely to be more costly to the Electricity Pool.
It is difficult to set down design rules for high reliability. Design experience may identify features which might detract from reliability and studies are occasionally carried out such as that by CIGRE reported in Section 6.7, but it is difficult to apply lessons which may be learned from these and be sure that every potential source of trouble has been avoided. Large generator transformers are produced in relatively small numbers, so there are no long production runs which can be used to eliminate teething troubles. One factor which can aid reliability, therefore, is to repeat tried and proven design practices wherever possible, even over many years, thus reducing the occasions on which teething troubles might occur.
Another design rule for high reliability is to 'keep it simple.' This is not always easy for an item as complex as a large generator transformer but, nevertheless, as explained above, a degree of simplification was achieved by the change from three-phase to single-phase units. This also meant that there was no longer the same emphasis on keeping sizes and weights to an absolute minimum and so there was a consequent relaxation of another of the pressures which threatened reliability.
A further option which must be considered is that of adopting a more onerous testing regime than that included in EN 60076. This has been discussed in here. This was part of the strategy adopted by CEGB in the 1980s and is considered to have been instrumental in improving the reliability obtained from large generator transformers.
The final factor to be considered is the level of monitoring and maintenance to be applied. For large generator transformers oil samples for dissolved gas analysis should be taken at least at 3 monthly intervals. Dryness of the oil must be maintained at levels in the region of 0.1 percent and action taken as discussed in here on any step changes in dissolved gas levels.