1910.] WALKER: DESIGN OF TURBO FIELD . 319

THE DESIGN OF TURBO FIELD MAGNETS FOR ALTERNATE-CURRENT GENERATORS WITH SPECIAL REFERENCE TO LARGE UNITS AT HIGH SPEEDS. By MILES WALKER, Member. (Paper received November 10, 1909, and read in London on March 10, 1910.)

SUMMARY. This paper re-opens the controversy, between the salient pole type of field and the cylindrical type. Reasons are given why the latter type is better suited for obtaining the greatest possible output from a given diameter. The methods adopted by various makers of supporting the windings are described, and the possible limits of output considered.

THE NECESSITY OF PROVIDING LARGE GENERATING UNITS OF HIGH SPEED. In the discussion which followed the paper by Dr. Kloss, " Practical Considerations in the Selection of Turbo ," before the Manchester Section of the Institution of Electrical Engineers in November, 1908, an argument arose as to whether the field magnets of turbo-generators should be made with salient poles or should be of the cylindrical type with a distributed winding. Some difference of opinion exists among designers on this point, and each method has been adopted in quite a large number of successful turbo-generators. In the opinion of the author the matter is one of great importance, because, in the future, makers of turbo-generators will be called upon to build machines of larger capacity than ever before contemplated, and the makers of steam turbines to drive them will call for extremely high speeds, so that the turbo-generator of the future will have to be capable of giving an extremely large output from a diameter which is not excessive, and will have to be constructed in a manner which permits of a high factor of safety. If we look at the growth in the kilowatt capacity of machines during the last thirty years, we are driven to the conclusion that the kilowatt capacity of large units in the immediate future will be as great as 15,000 or 20,000 k.w. In the year 1880 a 10-k.w. machine was considered large; in 1885 a 100 k.w. ; in 1890 a 300 k.w.; in 1895 a 500 k.w.; in 1900 a 1,000 k.w.; in 1905 a 5,000 k.w.; in 1910 we have 320 WALKER : DESIGN OF TURBO FIELD MAGNETS [March 10th, io,ooo-k.w. steam turbine-driven generators and 17,000-k.v.a. water turbine-driven generators. The cheapening of the cost of generation of electricity and the reduction in the capital required per kilowatt installed are already being felt" in the widening of the field in which electricity is being used, so that in the near future we may expect the output of electricity stations in all our large towns and manufacturing districts to be enormously increased, and the great requisite will be large generating stations at low capital cost. For capital cost is one of the main items which go to make up the price at which electricity can be supplied. These large units will in many cases consist of impulse turbines of very high speed and very large capacity.

THE SALIENT POLE TYPE versus THE CYLINDRICAL TYPE OF FIELD MAGNET. The author has therefore thought it would be well to open again the discussion upon the salient pole type and the cylindrical type of

FIG. 1. field magnet, and to consider the advantages and disadvantages of each of these types, with special reference to machines of very large output and very high speed. Let us consider first the essential features of a field magnet and the duties which it has to perform. We will take for granted that the field magnet is the rotating element, because it is difficult to support satisfactorily a rotating armature winding wound for a high voltage, and it is not desirable to collect current generated at 6,000, or 11,000 volts from slip-rings. Fig. 1 shows a section through the iron parts of a stationary arma- 1910.] FOR ALTERNATE-CURRENT GENERATORS. 321 ture with the winding turbo-. The duty of the field magnet, which we will take as having four poles, is to produce a magnetic flux which crosses the gap in a radial direction, and returns through the iron of the armature back to the core of the magnet. To produce this flux, we shall require a winding placed so that it produces a - motive force in the direction indicated by the arrows in Fig. i. The Two Factors in the Output.—The output of a field magnet is proportional to the product of two factors:— 1. The cross-section of the iron of the path indicated by the arrows (Fig. i). 2. The ampere-turns carried by the copper winding.

FIG. 2.

The cross-section of the path would be largest if the consisted of a solid cylinder of iron filling the whole space, in which case there would, of course, be no room for the copper. The ampere-turns would be greatest if the whole space were filled with a copper winding, in which case there would be no iron for the flux. The best theoretical arrangement would be a disposition of copper and iron something like that shown in Fig. i, in which the iron and copper are put in those places where they are most useful, and least interfere with the presence of one another. Such an ideal disposition of iron and copper would, of course, be difficult to carry out in commercial manufacture, but the type of rotor which most nearly approximates to this, will be the one which will give the greatest output in the smallest space. One of the main considerations which determines the arrangement 322 WALKER: DESIGN OF TURBO FIELD MAGNETS [March 10th,

of iron and copper on a turbo-rotor is the necessity of supporting the parts against the great centrifugal forces. The Salient Pole Construction.—The salient pole at first sight appears to offer very great facilities for supporting the copper winding, because it can be made with a projecting lip all round the pole. Fig. 2 shows the field magnet of one of the 4-pole turbo-generators lately described before this Institution.* This is mechanically a very good construction. In cases where it is decided to go to extremely high peripheral speeds, and additional support is required for the independent coils, the type shown in Fig. 3 may be employed. Here the coil on each

FIG. 3. pole consists of four parts wound in slots cut out of a solid steel forging. Or if it is desired to construct the pole of laminations, the type shown in Fig. 4 may be employed, and great mechanical strength at the same time obtained. Very successful turbo-generators have been built according to the 'methods illustrated in Fig. 2, 3, and 4. It will be observed, however, from these figures that there is between all the poles a space which is not utilised at all, and that is the very space where the copper winding should be, according to the ideal arrangement in Fig. 1. This space can be utilised to a certain extent for the purpose of ventilation It is, however, frequently occupied by bronze coil supports which are found necessary to prevent side bulging of the coils. With the coils placed as shown in Figs. 2, 3, and 4, the cross-section * G. Stoney and A. A. Law : " High-speed Electrical Machinery," Journal of-the Institution of Electrical Engineers, vol 41, p. 300, 1908. 1910.] FOR ALTERNATE-CURRENT GENERATORS. 323 of the pole is necessarily reduced, and therefore one of the factors of output (No. i above) is reduced. It.will be seen that the reduction of the total flux of a generator reduces its output in a way that can never be quite compensated for by increasing the ampere-turns, because the capacity of the armature is in any case limited to a definite number of ampere-wires per inch of periphery on account of heating considerations, whereas the flux-carrying capacity of the armature iron can be extended by increasing the outside diameter of the punchings. A type of field magnet which very effectually gets away from the necessity of side supports for .the coils is that shown in Fig. 5, where each pole consists of a number of round, poles placed side by side on a central boss. This construction, however, has the drawback of still

FIG. 4.

further reducing the cross-section of the iron path, but it leaves a large amount of space available for ventilation. With the salient pole type the ends of the coils are easily supported by a projection from the ends of the poles. The coils, moreover, are comparatively simple in form and cheap to construct, and the length of mean turn is not great for a given cross-section of iron enclosed. The coils can be easily replaced. The Cylindrical Type Construction.—The advantage of the cylindrical type with distributed winding, as illustrated in Figs. 6 to 15, is that by placing the winding in slots between iron teeth we can utilise a great part of the periphery both for carrying flux and for carrying ampere-turns. In order to obtain a generator of good regulation, it is necessary to have the ampere-turns at no load upon the field magnet greater than the ampere-turns on the armature. Where the iron body of the pole and the iron core of the armature offer very little magnetic 324 WALKER : DESIGN OF TURBO FIELD MAGNETS [March 10th, reluctance, it is necessary to introduce into the magnetic circuit some part, such as a large air-gap, which will offer a large reluctance, and enable the ampere-turns on the field magnet at no load to be of the required amount. It is therefore seen that the interference with the magnetic circuit caused by the slots in Fig. 6 is rather beneficial than otherwise, for the flux path being constricted in these teeth, there is a drop in magnetic potential just as there is in an air-gap. The teeth and slots may therefore be taken to replace in some measure a large air-gap ; in fact, saturated teeth employed in this manner give an even better characteristic than an air-gap, as is shown in connection with Fig. 20 below. Again, the space taken up by the teeth does not seriously interfere with the space required for the copper, because the iron presents a large surface to which heat can be conducted from the copper, and in that way enables the copper to be worked at a higher current density than otherwise would be the case.

FIG. 5.

Thus we see that with the arrangement shown in Figs. 7 to 12 the iron and copper, though each, to a certain extent, encroaches upon the space required for the other, mutually assist one another. It will be seen, too, in Fig. 6, that the iron below the slots offers such a large cross-section for the flux path that we can easily spare space for a number of large holes which traverse the field magnet from one end to the other, and supply air to a number of radial air ducts. Thus with the cylindrical field magnet we obtain a very large cooling surface on the insulation of the copper winding.by which the heat can be carried from the copper to the iron, and also very large surface within the ventilating ducts, by which the heat can be carried from the iron to the air. It is found in practice that copper placed in such slots can carry twice as many ampere-turns per square inch as copper in a wound coil i\ in. in thickness. There is a distinct advantage in supporting comparatively small sections of copper winding in independent slots, because by so doing one can preserve a high factor of safety in the FIG. 6.

FIG. 7.

1910.] FOR ALTERNATE-CURRENT GENERATORS. 325

VOL. 45 326 WALKER: DESIGN OF TURBO FIELD MAGNETS [March 10th, pressures thrown on the insulating materials, and by avoiding a great number of successive layers of insulation one is able to avoid troubles due to shrinkage of the insulating material. A drawback to be set off against this is that the length of mean turn for a given area of iron enclosed is greater than in the salient pole construction. Methods of Supporting the End Connections.—The main difficulty with the early cylindrical field magnets having windings distributed in slots was the supporting and cooling of the connections from one slot to another at each end of the rotor. Various methods have been adopted for making these end connections. One type is illustrated in Fig. 7 in which the connections form a barrel winding, similar to that used on direct-current armatures. This type has the following advantages :

FIG. 9.

All the conductors are made of the same shape, and the connections between them are very easily made by means of thimbles placed over the ends. The insulation of the winding can be carried out very effectually, and when well constructed and secured by wire or solid metal rings placed over the end connections, it preserves its balance extremely well. Another type is shown in Fig. 8. This consists of concentric coils and is perhaps the best type of winding to employ where the exciting current is small (say 50 to 100 amperes supplied at a high voltage), and where, therefore, a great number of comparatively small conductors must be placed in each slot. This winding may be supported by a metal ring placed over the ends, as shown in Fig. 8, or by means of a sheath strengthened by numerous rivets or bolts. Another method of supporting the end connections is shown diagrammatically in Fig. 9. Here the end connections consist of copper straps bent into the required shape, insulated from one another, and assembled so as to form a solid fan-shaped block. Four of these FIG. IO.

FIG. II.

FIG. 12.

1910.] FOR ALTERNATE-CURRENT GENERATORS. 327 blocks are mounted on a' saddle and V-shaped grooves are cut as shown in Fig. 10. The whole is then assembled between steel cheeks having V-shaped protections, mica V-rings being placed in the grooves, just as in the construction of a direct-current commutator. Fig. II shows the assembled end connectors being pressed on to a field magnet on which the field bars have already been placed in the slots. Fig. 12 shows a field magnet completed. A great number of field magnets have been constructed in this manner on machines intended for very high peripheral speeds, and they have proved very satisfactory both mechanically and electrically.

("J

FIG. 13.

The important points to be aimed at in the designing of the wind- ing of turbo-generator field magnets are as follows :— 1. Although the field magnet is excited at a comparatively low voltage—125 or 240 volts—the insulation must be carried out with the very greatest care. It is found that the dust which settles on the revolving parts of turbo-generators can, under certain circumstances, be so highly compressed by centrifugal forces that it becomes a much better conductor than the dust found on slow-speed machines. Cases have been experienced where pressures as low as 50 volts have caused the current to leak over considerable distances of .dirty insulation. 2. The insulation should be of such a nature that it does not permit the conductors to move in a radial direction after it has become dry. 328 WALKER: DESIGN OF TURBO FIELD MAGNETS [March 10th; 3. Allowance must be made for expansion and contraction of the copper. 4. Provision must be made for the cooling of the end connectors. 5. The design should be such that a repair of one part of the winding can be carried out with as little disturbance as possible of the remainder of the winding. Field Form.—The flux distribution on no load on the face of the armature of a salient pole machine is shown in Fig. 13A. At full load the field becomes distorted as shown in Fig. 13B, and if the armature is constructed with open slots, there is a tendency for the peak in the wave of the field form to produce ripples in the E.M.F. wave-form. These ripples, though of little consequence in the operation of most alternating-current machinery, may give rise to dangerous voltages

FIG. 14. when the generator is feeding a large, system of cables, whose capacity happens to give to the system a natural period of vibration correspond- ing, or nearly corresponding, with the frequency of the ripples. For this reason armatures provided with open slots are not as good as armatures with semi-enclosed slots unless the number of slots per pole is very great, the air-gap very long, or the field-form entirely free from peaks, such as shown in Fig. 13A. Ripples which on a no-load wave- form are scarcely perceptible may, when the capacity of the circuit assumes a certain critical value, be exaggerated, so that the E.M.F. wave-form assumes a shape like that illustrated in Fig. 14. In this case the crest-factor * may be very greatly augmented and a serious pressure thrown on • the cables. An E.M.F. wave-form closely resembling a sine wave-form and free from ripples both at no load and on full load is of considerable commercial importance. * " Crest-factor " is a term proposed by Dr. Gisbert Kapp for the ratio between the maximum value and the square root of mean square value. In the original paper the author used the term " form-factor." 1910.] . FOR ALTERNATE-CURRENT GENERATORS. . 329 Let us take an example : When a designer of a distribution system is deciding upon the voltage which he shall adopt for his high-tension transmission, he considers all the risks of breakdowns to his cables and • apparatus, and chooses that particular voltage which in his mind is the most suitable for his requirements. If he choose a higher voltage he will save copper ; if he choose a lower voltage, he will make his insulation safer. Now suppose that he fixes upon 6,600 volts, and then buys a generator which has a crest-factor under certain conditions of load of 1*9. With a crest-factor, of 1*9 on a virtual voltage of 6,600, the maximum voltage on his cables will be 12,500. This is the voltage which the insulation has to stand. If he had bought a generator with a true sine wave-form, whose crest-factor is only 1*41, he could with equal safety—so far as insulation is concerned—have employed a virtual voltage of 8,900. In so doing he would have saved 50 per cent, of the copper in the mains. The field-form of the cylindrical type of field magnet with a distri- buted winding is shown in Fig. 13c. The dotted line shows the distribution of magneto-motive force along the air-gap. The full line shows the form taken by the field. It will be seen that owing to the saturation of the teeth near the centre of the pole the corner of the field-form is nicely rounded off. Fig 13D shows the full-load field- form. Here again we have a nicely rounded curve free from elevated peaks. In order to arrive at the wave-form of E.M.F. yielded by a generator of this type of field, it is necessary to take a number of these field-forms, corresponding to the number of slots in 2 phase-bands of a 3-phase armature, displaced from one another by the amount of difference in phase between the various slots and to make a summation of them all. This yields an E.M.F. wave-form so nearly approximating to a sine wave as to be hardly distinguishable from it at no load, and at full load it only suffers a slight distortion. ! Voltage Regulation.—In order to make an ordinary alternate- current generator regulate well, it is necessary to have a large ratio between the ampere-turns on the field and the ampere-turns on the armature. With turbo-generators it is not usual, to have this ratio greater than 2 ; in many turbo-machines it is considerably less. In these cases it is usual to saturate the field magnet to a fairly high point, so that when the load is thrown off the extra ampere-turns in the exciting coils which were necessary to keep up the voltage on load shall not be able to create too great a rise in voltage. There is con- siderable danger in employing a very highly saturated field magnet, because it is always difficult to be sure of the quality of iron, and if the saturation is carried too far, it may be found impossible to obtain the voltage of the machine on heavy loads at a low power factor. It is important in this connection to consider at what part of the pole the saturation occurs. If the generator is of the salient pole construction and the saturation occurs at the root of the pole, there is much more danger of being unable to obtain full voltage than if the saturation occurs in teeth distributed near the periphery field of the 330 WALKER: DESIGN OF TURBO FIELD MAGNETS [March 10th, magnet. The reason of this is that the leakage is greater with the salient pole than with the cylindrical type. The leakage flux from the pole combines with the working flux in producing the saturation, and as the leakage flux is almost proportional to the ampere-turns on the pole, we increase the leakage and consequently the saturation at the same time that we increase the field current. Fig. 15 shows the shape of the saturation curve of a machine with salient poles both at no load and at full load, where the saturation has been designed to be rather excessive. It will be seen that at heavy load on a low power factor the saturation curve is almost asymptotic to a horizontal line, so that though the field current may be raised to very high values, the voltage cannot be increased beyond a certain " 10,000 9,000 — — 8,000 / 1/ 7,000 V 6,000 f 5,000 1 4?

4,000 / ff / 3,000 J

2,000 / 1,000 / 1 / 0 20 40 60 80 100 120 140 160 amperes. Exciting Current. FIG. 15.—Salient Pole Saturation Curves.

point. Fig. 16 shows the shape of the saturation curve of a generator with a cylindrical field with highly saturated teeth both at no load and on alow power factor load. Here the curve never becomes asymptotic to a horizontal line, because at the worst the space occupied by the copper and teeth will operate as a long air-gap, and there is no saturation in the body of the pole. An important advantage of the cylindrical type field magnet is that it is easy to arrange it so as to embody the compensating principle described on a previous occasion,* and make the armature reaction strengthen the field. This enables good regulation guarantees to be made without necessitating a very great ratio between the ampere- turns on the field and the ampere-turns on the armature. With this * Journal of the Institution of Electrical Engineers, vol. 34, p. 402, 1905. 1910.] FOR ALTERNATE-CURRENT GENERATORS. 331 type of machine it is possible to get good regulation (2 per cent, at unity power factor and 12 per cent, at o*8 power factor) even with the ampere-wires per inch of periphery as high as 1,000. It is also possible to employ a flux density in the air-gap as high as n Kapp lines per square inch. From these two quantities it is easy to calculate the possible output of a 4-pole generator running at 1,500 revs, per minute. Suppose that we decide not to go to a greater diameter than 44 in., or to a greater length than 70 in., though these figures could be increased by adopting special construction.

10,000 y y 9,000 / 8,000 / / / ,000

S3 6,000 1xf 5,000 1 J 4,000 J /f ,000

2,000 "f 1,000 l I

0 20 40 60 80 100 120 140 160 ampena. Exciting Current FIG. I6.—Cylindrical Field Saturation Curves. The possible output is easily calculated from the following con- siderations :— Let—

BK= maximum Kapp-lines per square inch in the air-gap. P == the periphery of the rotor in inches. / = the length of the rotor in inches. W = total conductors in all three phases. A = the amperes per phase in a 3-phase generator. Then- 6 Volts = 0-4 x revs, per minute xWxBK xPx/X io~ . Output in kilovolt-amperes = amperes X volts x 173 6 = 0*69 X revs, per minute xAxWxBKxPx/x 10- .

Taking A x W at 1,000 x P and BK at 11, we get— Output = 0-69 x 1,500 x 1,000 X 140 x 11 X 140 x 70 x io~6 = 15,500 k.v.a. 332 WALKER: PAPERS ON SHORT-CIRCUITING [March 10th, . By adopting special ventilating arrangements which will permit the bearings to be brought near to the ends of the rotor iron, the length of the rotor can be increased beyond 70 in. without giving the critical speed an unsuitable value. Then the output can be increased in pro- portion. Again, if we are content to have a poorer regulation we can increase the ampere-wires per inch of periphery considerably beyond 1,000, and in that way still further increase the output. In the above I have considered only the 4-pole machine. The outlet of a 2-pole generator is very much smaller, because the iron in the body of the pole becomes saturated long before the density in the air-gap reaches 11 Kapp-lines per square inch, and the ampere- turns cannot be made as great. With a 4-pole field it is possible to work as high as 11 Kapp-lines without incurring any trouble from unbalanced pull, because with the compensated type of generator the density in the gap is almost independent of the length of the air-gap. The rotor illustrated in Fig. 12 is for a 4,000-k.w. generator built for the Glasgow Corporation. It holds its voltage constant within 2 per cent, from no load to full load at 0*93 power factor. In this machine the density in the air-gap is 11 Kapp-lines, and by actual trial it was found that there was no serious unbalanced magnetic pull with a radial displacement of ^th in. An important feature of this type of rotor is that while it gives good inherent regulation at ordinary power factors, the armature current on short circuit can be made quite small;—that is, i'5 times full-load current for the permanent value and about 10 times full-load current for the instantaneous value. The low value of this current is of importance when we have to design the windings to withstand the forces to which they are subjected at the instant of short circuit.

DISCUSSION.

Mr Mr. W. A. CHAMEN : Personally I am more than usually interested Chamen. m these papers, because I have a 3,000-k.w. alternator designed by the author, and although we have had one or two short circuits, and have been rather alarmed at the noise and dust from the alternator when short circuits occurred on the distribution system, making one think that it is smoking, we have never had a breakdown. I think that speaks volumes, particularly when I tell you that it is very probably the first 2-pole alternator of 3,000-k.w. capacity running at 1,500 revs, per minute. There are plenty of 4-pole machines, but I have not been able to find any 2-pole machines of that size. As the author says, the strains in a 2-pole machine are very much greater, and the strains in our case certainly were very remarkable. We knew before the alternator was put in that we should have these strains to deal with, and I therefore took some trouble to look into the question of the clampings and fixings employed, and I am bound to say that I did not think anything could happen even if we did have a short circuit. As a matter of fact, the big inch studs holding the clamps have been bent in various directions and the coils themselves have been pushed out of