33 452 H , NHB COMMERCIAL DYNAMO DESIGN ORIGINAL PAPERS ON COMMERCIAL DYNAMO DESIGN BY WILLIAM L. WATERS, M.E., E.E. Engineer in Charge of High Speed Machinery, Westinghouse Electric Manufacturing Co. ; formerly Chief Engineer, National Elec- tric Co. , and Consulting Engineer, Westinghouse Air Brake and Canadian Westinghouse Cos. FIRST EDITION NEW YORK JOHN WILEY & SONS LONDON: CHAPMAN & HALL, LIMITED 1911 K^' COPYRIGHT 1911 BY WILLIAM L. WATERS "// is not knowledge, but ignorance, that begets confidence." CHARLES DARWIN ' ' / hold every man a debtor to his profession ; from the which, as men of course do seek to receive countenance and profit, so ought they of duty to endeavor themselves by way of amends to be a help and ornament thereunto. " FRANCIS BACON 343G52 INTRODUCTION Commercial Engineering. The nineteenth century was nota- ble for the achievements of the engineer, and there is little doubt that the men responsible for this pioneer work were engineers in the broadest sense of the word. They were engineers in their ability to manipulate the forces and materials of nature ; but they were also far-sighted and level-headed men of affairs in their ap- preciation of the economic value of their work, and in the man- agement of their projects. These men, by their genius and enterprise, and by their hard common sense and solid achieve- ment, forced the world to recognize the engineer as destined to play a leading part in the future. Public attention became focussed on the engineer, and numbers of men were attracted to the profession, cither through natural inclination or through hope of profit. But as was to be expected, a large percentage of these men were imperfectly educated, and possessed only a specialized talent in certain directions, without any of the broad- minded comprehensive spirit of the pioneers. This change soon reacted on public opinion, and the engineer lost his high stand- ing; so that the public and the world of commerce came to re- gard him as a specialized "crank," a species of high-grade artisan, who though useful enough in his place, was devoid of any ability to take a sane and comprehensive view of a situation, or to man- age an undertaking in which he might be interested. The com- mercial world, though distrusting the ability of an engineer to manage his enterprises, came to realize more and more their iii iv INTRODUCTION economic value and financial possibilities. The result of this was, that the enterprises which the engineering world proposed, gradually came to be exploited by business men, who had to a greater degree the confidence of the financial powers. These men, who attempted to manage such undertakings, were usually altogether ignorant of the work they took up, and it is not sur- prising that this development produced indifferent results. En- gineers, being effectively cut off from the commercial side of their work, became narrower; while the increasing specialization left the commercial men in still greater ignorance of the products they were handling, and their management became more and more wasteful. The inefficiency of this arrangement has gradually be- come evident to all, and it is now generally recognized that the executive and commercial head of any large engineering enter- prise must posess some engineering knowledge. It is also realized that, given an opportunity for a broader education, an engineer can be readily trained to become an efficient and high-grade com- mercial executive, while it is almost impossible to instil an en- gineering knowledge into one whose training has been restricted to the commercial world. As a result of this, engineers are being given opportunities to obtain a broader commercial education, and it is recognized in high financial quarters that in the future they will have to depend on the engineering profession for high- grade operating men in all engineering enterprises. Already we find engineers at the head of a number of the large concerns, and it is probable that the time is not far distant when it will be the exception to find the operation of an engineering under- taking in the hands of any other than an engineer. Mr. John I. Beggs. It was my good fortune to work under Mr. John I. Beggs, one of the first and one of the most prominent of the commercial engineering executives in this country, at a time when I was so interested in purely engineering problems that I was in danger of drifting away from broader interests. For twenty-five years in active touch with the electrical engineer- ing industry, as a manufacturer, or as an organizer and operator of public service corporations, obtaining his engineering and INTRODUCTION v commercial knowledge through hard personal experience, Mr. Beggs was pre-eminently fitted to guide and encourage the youthful engineer. Continually improving and rendering com- mercially practicable much of the apparatus used in connection with electric lighting or railway interests, and doing this with the purpose of obtaining results, rather than of being accorded credit or public recognition, he was one of the first to realize that the engineer was destined to become the dominant factor in large engineering corporations. .Giving his young engineers a free hand in all branches of commercial and engineering work, only guiding and checking them where neces- sary, he developed organizations and men in a way, which, while it made the men his grateful and enthusiastic admirers, estab- lished his reputation as one of the foremost organizers in the engineering industry, and left his mark on the methods and apparatus of to-day. In inscribing these few papers to Mr. John I. Beggs, I am only taking an opportunity of express- ing my indebtedness, and my appreciation of the education it was to be in touch with him. I have entitled the papers "Commercial " Design, because engineering questions of purely academic interest are made subservient to those broader issues, where design and operation must be considered in their relation to the commercial success and future of the undertak- ing. It is in this feature of my work that I have to ac- knowledge the influence of Mr. Beggs, and to state that I have in a great measure to thank him for any small success I may have had. History of Papers. These papers, which have been written at different periods during the past six years, are the result of fifteen years' experience as a designing and manufacturing engineer in some of the most important electrical manufacturing concerns in Europe and America. They cover various subjects relating to the design of electrical power machinery, in which I happened to have been interested at different times, and they are reproduced here in the hope that they may be more accessible, to any whom they may interest, than they would be in the pro- vi INTRODUCTION ceedings of the various institutions. The student will find there is no lack of treatises on dynamo design by capable men, and this book makes no pretense to be a comprehensive work. It is merely offered as a supplement to such treatises, as a series of articles, which written by one who is actively engaged in prac- tical manufacturing, may cover a few points of special interest to the student or designing engineer, that could not be satis- factorily covered in a more general work. December, 1910. CONTENTS PAGE INTRODUCTION .... iii COMMERCIAL ALTERNATOR DESIGN 1 DOUBLE CURRENT GENERATORS IN THEIR RELATION TO DOUBLE CURRENT SUPPLY 23 PREDETERMINATION OF SPARKING IN DIRECT CURRENT MACHINES . 29 ROTARY CONVERTERS AND MOTOR-GENERATORS . . . .45 SHUNT AND COMPOUND ROTARY CONVERTERS FOR RAILWAY WORK . 69 THE NON-SYNCHRONOUS OR INDUCTION GENERATOR IN POWER STATION WORK 75 MODERN DEVELOPMENTS IN SINGLE PHASE GENERATORS . . . 103 PERFORMANCE SPECIFICATIONS AND RATINGS 113 INPUT-OUTPUT EFFICIENCY TESTS 119 DIRECT CURRENT TURBO GENERATORS 123 [Presented before the American Institute of Electrical Engineers, June 29, 1903.] COMMERCIAL ALTERNATOR DESIGN THE design of alternators has been treated times without number, but usually the commercial element in the design, i.e., the relation of factory cost to selling price, has been ignored. An engineer has been defined as a man "who can do for one dollar what any fool can do for two," and as this definition applies in connection with machine design, the man who can design the cheaper machine to satisfy a given specification is the better designing engineer. Speaking generally, there is no type of alternator that will compare with the internal revolving field construction, in which each pole carries a separate field coil of edge-on copper strap. The revolving armature is cheaper for high frequencies and low voltages, the inductor type is good for small 60 cycle high- speed machines, while the disc alternator with no iron in the armature is an excellent machine for high frequencies. But these, though sufficiently satisfactory in their own limited field, do not compare with the revolving field type for general work. Fundamental Types of Construction. The revolving field alternator took its present form about 1892 when Mr. C. E. L. Brown designed some alternators which were practically modern machines; while in 1893 Mr. S. Z. de Ferranti* installed some * I understand from Mr. C. E. L. Brown that the Brown Boveri Maschin- enfabrik deserve to some extent the credit for the design of the Portsmouth alternators, though Mr. Ferranti was the first to use edge-on copper strap for field magnets. 1 1 COMMERCIAL ALTERNATOR DESIGN FIG. 1. COMMERCIAL ALTERNATOR DESIGN FIG. 2. 4 COMMERCIAL ALTERNATOR DESIGN 210 K.W. alternators at Portsmouth, England, which were of similar design to those of Mr. Brown. Before that date alter- nators of this type were of a clumsy amateur design, and their performance w r as, generally speaking, very poor. Strangely enough, these two engineers, after bringing out a high grade design, apparently abandoned further development, and their machines to-day are almost identical with their machines of ten years ago. It has taken the different manufacturing concerns a long time to recognize the superiority of the Brown and Ferranti type for standard work, and it is practically only during the last three or four years that this type has been generally adopted. The result is that, except for a few minor details, the construc- tion of these alternators has been improved very little since they were first introduced; while the excellence, from a commer- cial point of view, of the electrical design of Mr. Brown's early machines seems hardly recognized even yet by some engineers; and we have alternators on the market to-day which are for a given performance decidedly more expensive than those which he designed ten years ago. The armature frame in the Brown type of machine was only a skeleton cast iron frame for clamping the laminations to- gether, and was provided with large ventilating holes; while the ends of the armature coils stood out from the laminations, quite free and exposed to the full windage of the magnet- wheel. The numerous holes gave excellent cooling effect, but they reduced the strength and stiffness of the frame, so that the armature had to be stiffened by a series of tie-rods or struts. This con- struction, which saves material at the expense of labor, has become standard with German and Swiss firms, though on account of its unsightly appearance it has never found favor in this country. This type of alternator, shown in Fig. 1, has retained practically its original form up to the present time and developments that have taken place have been mostly in the Ferranti type. American and English engineers have followed the Ferranti COMMERCIAL ALTERNATOR DESIGN FIG. 3. 6 COMMERCIAL ALTERNATOR DESIGN type shown in Fig. 2, and made the armature frame stiff enough to stand without bracing. The trouble with this construction was originally the poor ventilation of the armature. Ventilating spaces were either not used, or if they were, there was no proper circulation of air through them. The end connections on the armature winding and the ends of the coils were packed tight together, or were closed in by cover-plates permitting no ventila- tion. Thus, the armature winding was usually the hottest part of the machine; so that even allowing temperature rises of 45 C., these machines could not be rated as they should, solely on account of the poor ventilation. When American and English engineers took up the revolving field type of alternator, the badly ventilated Ferranti type was adopted, and the great importance of ventilation was not recognized, so that the de- velopment of alternator design in America and England has been comparatively slow. The improvement in ventilation which has recently taken place in this type of alternator is really the greatest forward step that has been made, and it has given the designer immense help in increasing the output of machines. Fig. 3 shows an old, badly ventilated armature, while Figs. 4 and 8 show a more up-to-date well ventilated machine. In Figs. 4 and 8 it will be seen that where the ends of the armature coils cross one another they are separated by an air space, and that the end covers are provided with ventilating holes, to allow a circulation of air around the coils; thus the armature winding, instead of being the hottest part of the machine, becomes the coolest. The armature core is well provided with vent spaces both at the centre and at the ends, and the air passing through these vents is free to escape at the back of the core. This type of armature coil has the additional advantages that if lightning gets into the machine or a coil is burnt out, the damage is con- fined to one coil, so that we do not have half a dozen burnt out as usually happens. Also, all the coils on the armature are alike and made on the same former, so that the question of spare coils is simplified. The difference in the cooling effect between COMMERCIAL ALTERNATOR DESIGN FIG. 4. 8 COMMERCIAL ALTERNATOR DESIGN these different designs may not seem to be much on paper, but it results in the difference between a temperature rise of 45 and one of 25, on actual test. It means that we need only take into consideration efficiency and regulation in designing a machine, knowing well that if these are satisfactory we can guarantee a temperature rise of 25, even on low-speed machines. When designing any machine we have the choice of taking a large diameter and making the machine short, or of taking a FIG. 5. small diameter and making the machine long. The difference in the cooling effect between these two is obvious from Figs. 5 and 6. In Fig. 5 the machine is small in diameter and long, the poles are close together and the winding crowded, while all the heat from the field-coils has to be dissipated from the small exposed surface at the ends of the coils. In Fig. 6 the alternator is large in diameter and short, and practically the whole surface of the field- coil is available for cooling. In addition, the field coils being COMMERCIAL ALTERNATOR DESIGN 9 separated more from one another and the peripheral speed higher, the cooling effect on the armature is much greater. The machine in Fig. 6, being built on a large diameter, will require heavier castings and present greater difficulties in handling, but the fact that the designer is not restricted by the temperature rise, gives him so much more latitude, that he will easily offset this slight extra expense by a cheaper design generally, and will in addition have a much cooler machine. The difference in FIG. 6. cooling effect between a construction with armature and mag- nets as shown in Figs. 3 and 5 and one with armature and magnets as shown in Figs. 4 and 6 is so obvious, that it is quite surprising to find the poorly ventilated type still on the market. The only inference to be drawn is that the firms building them have not given the subject due thought. Specification. In the electrical part of the design, the first thing to be decided is the specification to which the machine 10 COMMERCIAL ALTERNATOR DESIGN is to be built. The firm with which I am connected has adopted as standard: A temperature rise of 35 C. on a continuous full load run; A temperature rise of 50 C. on 50 per cent over-load for two hours; A regulation of 5 to 7 per cent on non-inductive load, accord- ing to the size of the machine; And gives a guarantee that all machines will without damage give continuously 25 per cent current over-load at zero power factor. Detail Design. Given the specifications, we have next to decide what diameter we shall make the machine What magnetic densities to take in the iron? What current density in the conductors? What percentage of the pole pitch shall the pole face be? What air gap? Of course, these questions can only be answered off-hand as the results of experience. But generally speaking we can, after a few trials, decide on the best design. We have only to consider the efficiency, the regulation, and the cost; as with a good design the temperature need not be considered. The efficiency of a machine within ordinary limits practically depends on the magnetic densities in the iron and the current densities in the copper. The higher the densities, the cheaper and the less efficient the machine. The copper loss in the armature is usually between 1 and 2 per cent. Apart from the efficiency this is decided by the regulation, because if we allow only 5 to 7 per cent voltage drop on P F = 1, we cannot well have more than 2 per cent of this as C R drop. This means that in low-speed machines with a large number of poles, the current density in the armature is very low, while in high-speed machines with few poles the current density can be much higher. In practice it varies from 1200 to 3000 amperes per square inch. The iron densities do not vary much in standard machines, as the most economical densities are very fairly constant and independent of the speed; and any attempt to obtain higher efficiencies by de- creased iron densities will increase the cost rapidly. COMMERCIAL ALTERNATOR DESIGN 11 The best ratio of pole face to pole pitch is largely a matter of opinion. If it is large, say 70 per cent, then the E.M.F. coeffi- cient (the Kapp coefficient) is reduced, and the total flux of the machine increased and hence the magnets made heavier. On the other hand, with a wide pole face we have more teeth to carry the flux, and for a given tooth-density the machine is shorter. But the armature core-plates are correspondingly deeper, so that the only saving is a slight decrease in the length of mean turn of the armature winding. The larger the per- centage of pole face to pole pitch the greater the magnetic leak- age, and to a certain extent the less the synchronizing power of the alternator. So that there is little to be gained by much variation of this ratio, and it seems advantageous to keep it !ow, say between 60 and 65 per cent. The air gap is decided by the regulation of the machine. An immense amount has been written on various theoretical methods of calculating the regulation of alternators, but broadly speaking the regulation depends on the ratio of the ampere-turns on the armature to the ampere-turns for the air-gap. In an alternator the armature conductors are cut by magnetic lines due to the armature current, i.e., the armature self-induction flux; and by magnetic lines due to the magnet current. But the self-induction of the armature varies with the magnitude of the current in the armature and magnets, and with their relative position; while the useful flux due to the magnets varies with the current in the armature and field, on account of the varying perme- ability of the iron and the magnetic leakage. So it is obvious the conditions to be taken into account are so complicated that it becomes quite impossible to treat them theoretically, without making so many assumptions that the results, even when obtained, cannot be directly applied. What a designer has to do is to work theoretically through a few simple special cases himself and the results will give him an idea on what lines to work. Then by means of experimenting on a number of ma- chines he can develop an empirical method for calculating the regulation. Afterwards as he gets more and more experience 12 COMMERCIAL ALTERNATOR DESIGN with alternators, he introduces further refinements, and taking the regulation curves obtained by empirical methods he corrects them a little by eye. Speaking generally from a designer's point of view, an alter- nator should be calculated for a certain regulation on a low powder-factor load, say P F = 0. For, if the machine is satis- factory on low power-factors, it will be satisfactory for non- inductive loads, while the converse is not true. Other things being equal, the larger the air-gap the better the regulation on the low power-factors. But the leakage coefficient of the machine is an important factor, and this increases with the air- gap. This coefficient is, of course, taken into account in drawing the no load saturation curve; but we have to remember that on full load of low power-factor the leakage coefficient is much increased (the leakage is often doubled), on account of the additional ampere-turns required on the magnets to overcome the demagnetizing ampere turns on the armature. So that if the leakage coefficient is already high, and if the density in the magnet iron is also high, we run a considerable risk of saturating the magnet circuit so that we cannot obtain the rated voltage on loads of a low power-factor. It was this trouble that caused inductor machines to become obsolete for low power-factor loads, .as they are particularly sensitive to leakage and are always worked at high densities. Speaking generally, if the no load leakage coefficient of a machine is higher than 1.25, and if the density in the magnets is greater than 100,000 lines per sq. in., the designer has to be very careful or he will be in difficulties. The regulation on non-inductive loads is not affected by the length of the air-gap to the same extent as the regulation on low power-factors. So machines which are only intended for lighting or rotary converter work usually can be economically designed with a smaller air-gap and higher densities than machines for motor work. In Europe practically every alternator sold has to operate motors, so that the regulation either for P F = 0.8 or for P F = has to be guaranteed. In America, on account of the COMMERCIAL ALTERNATOR DESIGN i:J patent situation, induction motors are used only to a limited extent, and as a result of this it has become standard practice to sell machines on a regulation guarantee for non-inductive, rather than for inductive loads. This is very unsatisfactory. Almost every load that an alternator has to carry is to a certain extent inductive, e.g., arc lamps, transformers on light loads, rectifiers, induction motors, and synchronous motors, unless the excitation is carefully adjusted. And as the regulation of an alternator on P F = 0.9") is usually about twice as bad as on P F --= 1, it is obvious that a more satisfactory guarantee would be to give regulation on inductive loads. Practically the only exception to this is the case of an alternator for use exclusively for running rotary converters. And even with a compound- wound rotary converter and an inductive line, the power-factor is usually low and the current lagging for light loads; while if the rotaries have to be started from the alternating current side, a generator with poor regulation on low power-factors is very noticeable and may give trouble. It is extremely difficult to measure the regulation for P F --= 1 on any machine with good regulation, while on a large machine it is practically impossible. The result can only be figured as the difference between two large quantities, and there are so many disturbing features that the result when obtained is not very accurate. On the other hand, it is quite easy to measure the regulation on a very low power-factor by loading on to a second machine running as a synchronous motor, the first one operating as a generator, and varying the excitation of the motor and generator till full load current is flowing at full-load voltage. The power-factor in such a test will be very low and can, with sufficient accuracy, be taken zero. Alternators can be designed so as to satisfy a close regulation specification for non-inductive loads and yet be almost worthless for carrying loads of low power-factor. So that as such machines can be made cheaper than if they were required to give a reasonable regulation on inductive loads, there is a temptation for manufacturers to take advantage of the fact that the regula- 14 COMMERCIAL ALTERNATOR DESIGN tion is only guaranteed on P F = 1, to install one of these cheaper machines. It is probably this fact which is responsible for the number of alternators having poor regulation on low power-factors, that have been installed in this country. It would certainly be an advantage from the customer's point of view, and probably in the end from that of the manufacturer also, if the regulation were guaranteed for a load of low power-factor. This would make it necessary from the com- mercial standpoint to alter somewhat the lines on which modern alternators are designed, but the cost of the machines would not necessarily be much increased. A modern alternator gives, say, 7 per cent regulation on P F = 1, and 22 per cent on P F = 0.8. When operating with a normal power-factor of about 0.85, and a regulation of about 17 per cent, it does not help the station engineer to know that if he had a non-induct- ive load he would have good regulation. Such a machine could be re-designed on somewhat different lines, so as to have 6 per cent regulation on P F = 1, and 12 per cent on P F --= 0.8, and about one per cent lower efficiency without increasing the cost appreciably. Such a machine would be much more satis- factory for general work and could probably be sold for consid- erably more than the machine designed only for work on non-inductive loads. Other things being equal, the regulation of an alternator is better the more saturated the magnet circuit, and this applies to low power-factor loads as well as to high. It can be considered simply as an experimental fact, or the explanation can be accepted that the voltage drop in an alternator is partly due to the reaction of the armature ampere turns, and that the effect of a definite percentage change in the ampere-turns is less when the magnets are saturated than when they are not. Obviously the part of the magnetic circuit to saturate is the magnet core, as the less its cross section, the less its perimeter and the less the weight of the magnet copper. In the type of magnet shown in Fig. 7, it is impossible to saturate the pole core, and the large amount of iron and copper necessary always makes this design COMMERCIAL ALTERNATOR DESIGN 15 needlessly extravagant. It, however, possesses the advantage that the voltage can be raised 25 or 30 per cent, if desired, to compensate for an extraordinary line drop; though usually it is sufficient if an alternator is capable of having its voltage increased 15 per cent when carrying full load. When designing an alternator for a given output we can adopt either a strong armature and a weak field, or a weak armature and a strong field; and generally speaking, the stronger the armature, the cheaper the machine but the worse the regulation. So to design cheap machines with good regulation, it is necessary to take advantage of everything that will better our regulation, i.e., we must work with a long air gap and we must saturate our poles. But a long air gap results in large leakage, and as I pointed out before, a machine with large leakage and saturated poles is the most difficult machine to design. To make a uniform success of such machines, the designer must have had considerable experience with the type of alternator in question, and must be a very careful worker. In fact, when I first began designing alternators I was told to put plenty of iron and plenty of copper into the magnets, and that if I did this I would be safe. I think for a beginner the advice is good and that he could not do better than commence with a conservative and simple design like that shown in Fig. 7. But for a designer who has had considerable experience, it is well to figure more closely, as there is quite 20 per cent in the cost to be saved by so doing. Suppose we have decided to adopt a large air-gap and yet wish to keep down the leakage. There are several things which will help us in this, but making the pole pitch large and decreas- ing the length of the magnet pole are the most important. Adopting a large pole pitch results in making the machine larger in diameter, and shorter. Beyond certain limits this in- creases the cost, and it is a question to be decided by the designer as to when the advantages obtained from the larger pole pitch are offset by the increased diameter and weight of the castings. Decreasing the length of the magnet pole core reduces the leakage. Decreasing this length also results in decreases 16 COMMERCIAL ALTERNATOR DESIGN the radiating surface, increasing the depth of the magnet wind- ing and hence increasing the length of mean turn of the magnet coil somewhat at the same time. It also slightly decreases the ampere-turns for the magnetic circuit. If we use a large pole pitch, giving plenty of space between the poles, together with a short armature and high peripheral speed, we can easily avoid the increase in temperature due to decrease in radiating surface. So that the limiting factor in reducing the length of FIG. the pole core becomes the additional weight of copper due to the increased length of mean turn on the magnet winding, caused by the extra depth of the winding. With good design we can usually reduce the length of pole core to about one inch for every 1500 ampere turns required on full load; so that our leakage coefficient will generally not exceed 1.25, which is not excessive. To show the effect of these various points on the design of a machine, let us take a definite example. COMMERCIAL ALTERNATOR DESIGN 17 Output 750 K.W. 60 cycles, 100 R.P.M., 72 poles, 2200 volt. Specification to be: Electrical efficiency at full load, 95% Regulation 7% for P F = 1 16% for P F = 0.8 257 P F = Temperature rise on full load PF = 1, Armature 30 C., Magnets, 20 C. This low magnet temperature being adopted so that the temperature will not become excessive with the increased losses which will result when operating on loads of low power-factor. A is a machine which has a pole pitch and diameter large enough to use round poles, and has saturated pole cores. It has a strong armature and strong magnets. B has a smaller pole pitch, so that it is a longer machine, and has unsaturated fields. It has a weak armature and field, and the magnet winding is crowded. A. B. C. Internal diameter of armature Length of armature core 207* 6f 161" 13i" 207" 61" Pole pitch 9* 7" 9* Air gap 5/16" 1/4" 5/16" Peripheral speed, feet per min Slots per pole 5400 6 4200 6 5400 6 Turns per coil . 5 3 4 Magnet core section . round rectangular round Induction in magnet core, per sq . in. Regulation P F I 110000 7% 9.5000 6 8% 110000 5 6% P F = .8 15 5% 16% 10% PF = 24% 25% 16% Losses magnet C 2 R watts Armature C 2 R. " 12500 10700 8500 7250 17000 12000 Iron loss " 15200 24000 19300 Efficiency 95 1% 95 0% 94 0% Temperature rise of armature .... magnets . . . 22 C. 15 C. 30 C. 16 C 26 C. 26 C Weight magnet copper . . Ibs. 1800 :i.s( H 2300 " poles ' " wheel. . 1420 14000 4500 15000 1780 16000 " armature copper . ' 1425 1200 810 " " laminations . ' frame ' Cost of above material 5500 22000 $1,645 7500 18000 $2,150 6600 22000 $1,710 18 COMMERCIAL ALTERNATOR DESIGN So on the principal items that enter into the cost of material, the saving is about 25 per cent, due to the use of strong armature and magnets, large pole pitch, and short saturated magnet cores. Probably the saving of the cost of the complete machine would be about 20 or 25 per cent. The designs of these two machines are a little exaggerated, but they show very well the saving in cost that can be made. C is the same machine as A, but designed with a weaker armature, so as to have better regulation, especially on low power-factors, at the expense of a lower efficiency. The cost is about the same. The chief points for a cheap design for an alternator with good voltage regulation on inductive loads are strong, saturated magnets, a reasonably large pole pitch, and as large an air-gap as can be used without excessive leakage. In machines of small output with a large number of poles, it is impossible to get a really economical design. The diameter is decided by the number of poles and little is gained by making the machine less than 5 or 6 inches long, so the cost is not reduced very much with the output. Generally speaking, if the output of the machine is less than 10 K.W. per pole, the design is unnecessarily expensive, while machines in which the length of the armature is about equal to the pole pitch are usually the most economical. It is for this reason that 60 cycle alternators for small outputs, and 120 or 133 cycle alternators of all outputs are usually made belt-driven; the saving in cost by doing this often being 50 per cent. In continental Europe, where 50 cycles is the' usual practice, belt-driven alternators have never met with much favor; the universal custom being to direct-connect the alterna- tor to a low-speed engine. The result of this has been that the fly-wheel type of alternator has practically become standard for this work, and the poles of the alternator are simply bolted to the rim of the fly-wheel. This type allows considerable saving in cost in small 50 or 60 cycle machines and possesses so many other advantages that it is being introduced into this country. COMMERCIAL ALTERNATOR DESIGN 19 Steam Turbine Driven Alternators. Alternators for direct connection to steam turbines have lately come into prominence; the chief consideration in these machines being, of course, the high speed at which they operate. In order to prevent the length becoming excessive, the diameter of a turbo alternator is made as large as possible, and peripheral speeds of from 12,000 to 20,000 feet per minute are adopted. The mechanical stresses in the metallic parts of the magnets are high, but a con- servative factor of safety can bo maintained if high-grade materials are employed. The greatest difficulty consists in arranging the mechanical stresses on the insulation of the rotor in such a way that the insulation is not damaged. If this is not done, and if the winding and constituent parts of the rotor are not firmly fixed so that relative motion cannot take place under the influence of the centrifugal forces, continual trouble will result due to changing of balance. The electrical design of a turbo alternator is much the same as that of a belt-driven machine, except that the speed being very high the efficiency is good ; so that the magnetic and current densities in the armature are limited only by magnetic saturation and the difficulty of dissipating the heat in the extremely long armatures used on these machines. The polo pitch and the ampere turns on the armature being large, the magnets are of necessity very strong and the air gap large, so that the question of magnetic leakage becomes important. In turbo alternators just as in low-speed machines, saturated magnets with armature and magnets as strong as can conveniently be adopted, result in economical designs; but as the mechanical conditions are so much more severe, good mechanical design has a more important influence on the cost than is the case in a slow-speed machine. General Comparison. When commencing to work out a machine an experienced designer can usually estimate the rnost economical diameter to adopt; and he knows from experience approximately the number of ampere-turns he can take per inch periphery on the armature for an alternator of given pole pitch and type. This decides the number of turns on the COMMERCIAL ALTERNATOR DESIGN FIG. 8. Armature Winding of 275 K.W., 600 R.P.M., 3 Phase, 60 Cycle Generator. COMMERCIAL ALTERNATOR DESIGN 21 armature and the ampere-turns on the magnets. He then completes the first rough design, and working out the perform- ance curves, he can usually see very quickly in what way to modify it so as to obtain the best design possible under the circumstances. The speed and frequency are the chief factors in deciding the design of a machine, but the voltage, the con- ditions of operation, the equipment of the factory in which it is to be built, the facilities for obtaining castings and for shipping the completed machine, all are points which affect the design and have to be considered by the practical designer, since the prime object in a commercial design is rather to make profits for the manufacturing company than to produce the most perfect machine. The points which have to be taken into consideration arc so numerous and varied that it is impossible to give general rules for practical design. All that can be done is to give general directions and then it is a question of ability and experience until the engineer can produce the best results. Neglecting for a moment the designs of Mr. C. E. L. Brown, the greatest change in the design of alternators in the last ten years is the improved ventilation and the increased magnet strength. In 1893 we were working with air-gap densities of 25,000 to 30,000 and magnets with 3,500 to 4,000 ampere-turns per pole on full load, while to-day we have air-gap densities of 60,000 to 70,000 and magnets giving anything up to 20,000 ampere-turns per pole on full load, for ordinary belt-driven or engine-type alternators, and up to double this on steam-turbine- driven machines. The change has been made so gradually that it has been hardly noticeable, but the effect can be seen if I give the dimensions on two machines designed and tested, one in 1894 and the other in 1903, this latter machine being shown in Fig. 8. Both the designs are typical of the condition of the alternator design at those dates. COMMERCIAL ALTERNATOR DESIGN 1894 1903 Output ( 70 K.W. 3 Phase } 275 K W 3 Phase Speed ( 50 K.W. Single Ph. 600 R.P.M. 600 R P M Cycles 60 60 Type of magnets Lauffen Type. Standard Revolving Internal diameter armature .... Length of armature laminations Armature ampere-turns 37" ior 1380 Field Type. 38" 10 1050 Ampere-turns on magnets Full load efficiency, per cent Regulation P F = 1 4700 90 11% 7500 94 5 5% " P F = .8 Would not give volts 14% Full load temperature rise of ar- mature . . . 31 C 23 C. Total weight of copper, Ibs. 730 680 Total weight of machine " .... Total cost of material 7400 $455 11000 $490 (Two bearing machine and 15c. copper) The output has been increased about four times and we have a much better machine as regards performance, while the cost is about the same. The older machine having such a low out- put needs a large amount of unnecessary material, to reduce the losses so as to give a reasonable efficiency. While in the larger output machine we can afford a considerably higher loss without reducing the efficiency, so that the weight of mate- rial is little more; and the better mechanical design has made the cost of material in the two machines about the same. Speak- ing generally, the whole result has been accomplished by using stronger magnets and higher densities throughout, which are possible on account of the improved ventilation. Alternating current design has rather stagnated of late on account of the limited competition, and most of the recent developments have come from the other side of the water. But I think it has been shown that, from the point of view of dollars and cents, it is certainly worth while to spend time and ability somewhat lavishly in designing alternators. [Paper presented before the Northwestern Electrical Association, June 10, 1904.] DOUBLE-CURRENT GENERATORS IN THEIR CONNEC- TION WITH DOUBLE-CURRENT SUPPLY THE relative advantages of direct and alternating current supply are now tolerably well recognized. The great advantage of alternating current is the ease with which high voltages can be handled, and the facility with which the voltages can be trans- formed by means of stationary transformers; while the disad- vantage is that in the present state of the art it is unsatisfactory for street railway work, and also to a certain extent for elevator or variable speed motors. With direct current, exactly the reverse is the case; it is unsuitable for high voltage work, but gives good results in all classes of motor work. The obvious result of this has been the adoption of double-current supply in situations where both these advantages and disadvantages are important. So that usually in small towns alternating current is supplied for lighting and direct current for traction work while in larger towns direct current is supplied for the down- town districts, where the motor load is important, and alternating current for up-town districts, where the load is almost entirely a lighting one. Thus we very often have both alternating and direct current supply from the same power station. The question of double-current supply from one station is usually settled by the installation of both alternating and direct current sets, each set generally having its own engine. This solution is hardly regarded as satisfactory because the motor 23 24 DOUBLE-CURRENT GENERATORS load, having its maximum during the day, and the lighting load its maximum in the evening, the result is that the alternating current sets are idle during the day and the direct current sets are idle in the evening, so that we have only about half the plant in use at one time. It is obvious that a saving in first cost and in operating expense will be made if the two systems are tied together in some way, so that they can help out one another at times of heavy load. This can be done by having both an alternator and a direct-current generator coupled to one engine, or by having double-current generators, or by tying the supply circuits together by rotary converters and motor-generator sets. A modification of the latter method which has recently be- come popular consists in installing alternating current gen- erators only, and providing rotary converters to transform a portion of the power to direct current. From the point of view of the station engineer the double- current generator should be the best solution. The efficiency is higher than when rotary converters or motor generator sets are used, and it ought to be considerably cheaper than either of the other methods. The objection to it is that the voltage on the alternating current side of the generator bears a definite ratio to that on the direct-current side, so that one cannot be varied without the other; while the variation of the load on one side affects to a slight extent the voltage on the other. The relative importance of these objections has, of course, to be decided in each individual case. From the manufacturer's point of view the objection to double-current generators is that they arc special machines, and usually require new designs and special patterns or tools. Of course, we can take any direct-current generator, provide it with collector rings, and use it to supply alternating current; but the difficulty is usually that the frequency is unsuitable. An alternator can be built for any commercial frequency, speed, or voltage, without any serious difficulty; but for a direct-current generator, given the speed, voltage, and output, the question of commutation and cost decide within narrow limits the number DOUBLE-CURRENT GENERATORS of poles, and hence indirectly the frequency. If it is necessary to change this number of poles considerably in order to obtain a special frequency, there is often difficulty with the design, which results in increased cost. So if some latitude can be allowed in selecting the frequency and speed for double-current generators, it is advisable to choose them so that, if possible, the generator does not differ very much from some standard direct-current machine. The following table gives the frequency of the alternating current that could be obtained from standard direct-current generators. The number of poles and the speeds of the machines are taken from those of the National Electric Company, but there is little difference in these respects between the machines of the various manufacturers, so that they can be regarded as applying approximately to all standard makes of generators. ENGINE-DRIVEN. BELT-DRIVEN. STEAM Ti -RHINE DRIVEN. 250 Volt. 500 Volt. 250 Volt. 500 Volt. 2.50 Volt. 500 Volt. 25 15 15 40 40 50 14 14 33 33 90 90 100 14 13 35 35 70 70 250 14 12 27 23 60 60 500 12 10 31 25 . 75 50 750 14 10 70 1000 15 13 1500 20 16 2500 22 18 Considering as standard frequencies for double-current generators 25, 40, and 60, it is evident that some of these stand- ard direct-current machines could be very conveniently used as double-current generators with only a slight change in speed. A standard 2,500 K.W., 250 volt engine type machine would make a good 25 cycle double-current generator, the only changes nec- essary being to provide collector-rings, and to increase the air gap and put more copper on the field magnets to make the regulation satisfactory when operating as an alternating-current 26 DOUBLE-CURRENT GENERATORS generator. These changes would not increase the cost of the generator more than 20 per cent. On the other hand, if we took a 500 K.W., 500 volt engine type machine, very radical changes would be required to make this into a 25 cycle double-current generator, as it would be necessary to increase the number of poles from 10 to 24 or 30, which would practically double the cost of the machine. But if we make this 500 K.W. machine belt-driven instead of -direct-connected to a slow-speed engine, it is evident from the table of frequencies that a standard machine would be satisfactory for 25 cycles. Twenty-five or forty cycle machines are not in any way difficult to build at the most it is a question of special designs and patterns. But with 60 cycle generators we begin to have difficulties with the commutator on account of the high periph- eral speed. Sixty cycle, 600 volt, double-current generators and rotary converters can undoubtedly be built, but at the present date they are not such reliable machines as those for lower frequencies, and there is no brush gear now on the market which is quite satisfactory for the peripheral speeds necessary in a 60 cycle, 600 volt machine. The higher the speed of a standard direct-current machine for a given output, the higher the frequency; thus we would expect that the higher the frequency of a double-current genera- tor the higher the speed at which it should operate; and it appears from the table that the most satisfactory 60 cycle double-current generators will be those driven by steam turbines. Direct current generators for direct connection to steam turbines and suitable for operation under American con- ditions, are not at present on the market, but in all probability they will be shortly; and it appears probable that this type of machine will solve the problem of 60 cycle double-current generators. Generally speaking then, 25 cycle double-current generators, if of large size, can be direct-connected to a steam engine, while for smaller units than 500 K.W., they are better belt-driven. Forty cycle machines should always be belt-driven if the cost DOUBLE-CURRENT GENERATORS 27 is to be reasonable, while for 60 cycle double-current generators apparently the only reasonable solution is to have steam turbine driven sets. Of course, double-current generators can be made for any frequency or voltage up to 60 cycle, 600 volts, and at any speed desired, it is only a question of cost; but to obtain a reasonable price or delivery, and to have a unit which will at some time in the future have a second-hand value better than scrap, it would be advisable to consider the above table of frequencies and outputs when laying out a station for double- current supply with double-current generators. NOTE [Dec., 1910] In the past three years great improvements have been made in the construction of commutators and brush gear for operation at high speeds; this work having been done in connection with development of 60 cycle rotaries and direct-current turbo generators, which should be as reliable in operation as the cor- responding 25 cycle and slow-speed units. The result of this work has been to revolutionize the methods of constructing high speed commutators and brush gears, so that at the present time commutators are being built to operate perfectly, with reason- able attention, at the speeds required by 60 cycle rotaries and direct-current turbo generators. This being the case, the state- ment above, in regard to the reliability of 60 cycle 600 volt double-current generators and rotary converters should be modified accordingly. [Presented before the American Institute of Electrical Engineers, May 17, 1904.] PREDETERMINATION OF SPARKING IN DIRECT CURRENT MACHINES SPEAKING generally, dynamo design did not become an art until after the old two pole smooth-core Siemens and Edison machines came into extensive use for electric lighting. The original design of these machines was more or less guesswork ; but after a few machines had been made to operate satisfactorily, the engineer was able to lay out a complete line of machines, design- ing them partly by his engineering intuition, and partly by some empirical rules, which he decided on as he built successive machines. The armatures were designed more from a mechani- cal than from an electrical standpoint, their length being limited by the stiffness of the shaft rather than by questions of com- mutation; while they were unvcntilated, and in consequence the output was limited by heating. The armatures being of the smooth-core type, the self-induction of the armature coils was usually so small, even with the length of armatures in general use, that it was unnecessary to consider it in connection with the tendency of the machine to spark. It was, however, generally recognized that if the magnets were too weak the machine was liable to spark, so the length of the air-gap was usually deter- mined by some empirical rule obtained by experiment. When slotted armatures were first adopted extensively they were de- signed along the same lines as smooth-core armatures. They were so badly ventilated that the output was limited by heating to about one-half that of a modern armature; but in spite of 29 30 SPARKING IN DIRECT CURRENT MACHINES this it was found necessary to use carbon brushes to obtain good commutation. To economize in tools several different lengths of armatures were frequently built on the same diameter, while to economize space, the armatures were often built smaller in diameter and longer than they otherwise would be. Experience with these different forms of armature made it very evident that a long armature had a greater tendency to spark than a short one; and this became especially noticeable as the ventilating was improved and the output correspondingly increased on account of the cooler operation. Previous to this a great deal had been written on the theory of commutation in dynamos, but had been ignored by the prac- tical designers, who had more faith in experimental results. But this bad behavior of long armatures as regards sparking, called attention to the theoretical work, and designers began to consider whether or not the self-induction of the armature coils did not, after all, decide the amount of current the machine would commutate without sparking. In the first attempts to take into account the self-induction of the commutated coil, the self-induction of a one-turn coil was considered as being simply proportional to the length of the armature core; that is, the shape or size of the slot, the number of coils per slot, and the self-induction of the end connections, were all neglected. 'This gave a very simple formula for the self-induction : I L = ln\ Where I = length of armature n = number of turns per coil, And the self-induction E.M.F. of commutation (the reactance voltage as it was called), which is an estimate of the difficulty of commutating the current, was given by E = In 2 if i being the current per coil and / the frequency of commuta- tion. This formula gave satisfactory results when applied to ma- chines designed along the same general lines. The allowable value of the reactance voltage could be obtained from experiment SPARKING IN DIRECT CURRENT . MACHINES 31 on one machine, and used in the design of other similar machines. But, if applied to machines which were designed differently, the formula showed wide discrepancies; so it soon became recognized that the formula was at best only a rough approximation. Early slotted machines were designed with one coil per slot; two coils per slot obviously saved insulation space and were soon tried, but it was found that generally if this were done every other bar on the commutator became badly marked. As it was imperative to save space in car motors, three coils per slot were adopted, and in extreme cases four, or even five coils per slot were used. It was generally found, however, that whenever more than one coil per slot was used some of the commutator- bars were marked, and that it was possible to count the number of coils per slot by the recurrence of the marking on the com- mutator. This marking was attributed to the inequality caused by using a small number of slots, and so the general rules were adopted to use as many slots as possible and to make small machines with one coil per slot and large machines with only two coils per slot. It was also noticed that the dead coils necessary in certain multipolar wave-windings often caused some of the commutator bars to be marked. This was naturally attributed to the dis- symmetry produced in the winding, and it became generally recognized that anything tending to produce inequality in the commutation conditions, such as few slots, or many coils per slot, or dead coils, tended to make perfect commutation more difficult. With increased competition came the necessity of cheapening the cost of building these machines; designers then returned to the construction of several coils per slot. In reducing the amount of copper on the armatures to save in the cost of ma- terial, it naturally happened that shallow slots were used. And it was found that with these wide and shallow slots it was possible to obtain good commutation with several coils per slot, under conditions where it would be quite impossible with the old deep and narrow slots. Obviously this was due to the lesser self- 32 SPARKING IN DIRECT CURRENT MACHINES induction of a wide slot compared to a narrow one, and it was soon acknowledged that the shape of the slot should be con- sidered in calculating the self-induction of the commutated coil. When designing an armature for small self-induction it would be natural to make it large in diameter and short in length ; that is, with a large pole-pitch. But in carrying this to an extreme it was found that it did not give the good results expected. It was suggested that this result was due to the fact that the self- induction of the end connections had been neglected, and that in armatures with large pole-pitch and short length of core, the self-induction of the end connections was comparable with that of the conductor embedded in the slots. In the light of these experiences it is evident that the design of a direct-current machine in regard to sparking is a compromise between a number of conflicting conditions. It is not possible to obtain a formula which will give a strict measure of the corn- mutating qualities of all machines; but by taking into considera- tion the more important conditions which affect the sparking, it is possible to obtain one which will give fairly accurate results when applied to machines similarly designed, and which will give some idea of the tendency to spark when applied to machines of widely different design. Such a formula, when it has been applied to numerous machines of different types, so that the allowable values for the sparking constant have been determined, can be taken as a fair working formula, and can be placed in the same category as empirical formulae for determining the regu- lation of alternators. Such formula) are not intended to reduce designing to mere slide-rule work, but are intended simply to give an idea as to the experimental results to be expected from an individual design. As outlined above, the most important conditions to be taken into consideration are the self-induction electromotive force of the commutated coil, and the inequalities introduced by the conditions of commutation. Electromotive Force Due to Self-induction. This is given by the formula : E = Self-induction of one coil X number of coils SPARKING IN DIRECT CURRENT MACHINES 33 commutated in series X current in coil X frequency of commu- tation. The self-induction of one coil = (self-induction of one con- ductor embedded in the slot + self-induction of one end connec- tion) X (number of turns per coil) 2 . The self-induction of one conductor embedded in the slot = Ik. Where I is the length of the core and A; is a constant depending on the dimensions of the slot. By determining the self-induction of a large number of slots we find that this constant k can, with sufficient accuracy, be taken as a function of the ratio r, where Width of slot. Depth of slot. A curve can be plotted connecting r and k, determined ex- perimentally from tests on a number of armatures; such a curve is shown in Fig. 9. The self-induction of the end connections can be taken as = length of end connections X constant c / . And as the length of end connection is approximately proportional to the pole pitch, the self-induction of two end connections can, with sufficient accuracy, be written = p c. Hence the self-induction of one coil = n 2 (I k -f p c). The number of coils commutated in series, N, is, of course, one in a parallel or lap- wound armature, and equal to the number of pairs of poles in a series or wave-wound armature. The current per coil i in a two-circuit series or wave-wound armature is equal to one-half the total current in the machine, while in a parallel or lap-wound armature it equals the total current divided by the number of poles. The self-induction pressure of the commutated coil is then given by V = n 2 (Ik + p c) N if Where / is frequency of commutation, and is equal to the number of commutator bars X speed in rev. per min. 3 34 SPARKING IN DIRECT CURRENT MACHINES The width of the brush is neglected in calculating the fre- quency of commutation, since it is found by experiment that within the ordinary limits of practice the thickness of the brush has little effect on the operation of a machine, unless the current density is excessive. The probable explanation of this is that a thicker brush gives more time for commutation to take place; but it also means that more coils are commutated at the same FIG. 9. time, thus introducing the effect of mutual induction. These two effects apparently counterbalance each other to a great extent. Inequalities Due to Conditions of Commutation. These are due to the use of few slots; more than one coil per slot; and to dead coils. If there is only one coil per slot the use of few slots does not SPARKING IN DIRECT CURRENT MACHINES 35 in itself affect commutation, unless the number of slots is ex- tremely small; for though the slot may move through an ap- preciable arc while the coil is being commutated, the conditions are exactly the same for every coil when it is commutated. So there is no tendency to inequality in the conditions, and if it is possible to set the brushes so that one coil can be commutat- ed satisfactorily, then commutation will be satisfactory for all the coils. But if the number of slots is extremely small, say less than six per hole, then the coil will move in such a widely varying magnetic field, and will come so close to the strong field under the pole-tip while it is being commutated, that the local currents under the brush are liable to produce marking of the commuta- tor-bars even if the brushes apparently do not spark. Of course this is only important in very low voltage machines and it is unnecessary to take it into account in any constant which is to be a criterion of the tendency to spark. It is sufficient to say that the number of commutator-segments in the polar-gap, that is, the arc between the two pole-shoes, must never be less than two and should generally be three or more. With more than one coil per slot inequalities are introduced: due to the difference in the value of the self-induction of the various coils; and due to their commutation under different conditions. The self-induction of all the armature coils will be the same when there are only two coils per slot, as it is obvious that the configuration of the conductors and neighboring iron is the same for both coils. But when there are three or more coils per slot the self-induction of the various coils will vary, as they occupy different relative positions in regard to the iron; the self-induc- tion of the center coil being less than the self-induction of the outer coil. Investigating conditions at the point of commuta- tion in a modern generator by means of a pilot-brush, it is found that commutation usually takes place at a point where there is practically no resultant magnetic field; that is, at a point where the magnetic field of the armature just counterbalances the average field due to the magnets. In other words, there is 36 SPARKING IN DIRECT CURRENT MACHINES resistance commutation; the armature current is commutated by the varying resistance of the brush, rather than by a reversing E.M.F., due to passing through a magnetic field. This being the case it is only necessary to consider the self-induction of FIG. 10. Showing Position of Armature at Beginning and End of Commutation Period. those coils which have the greatest self-induction. If these are commutated satisfactorily by means of the varying resistance of the brushes, then the coils which have a smaller self-in- duction will also be satisfactorily commutated. Hence the variation in the self-induction of the coils need not be con- SPARKING IN DIRECT CURRENT MACHINES 37 sidered, and in the formula all that it is necessary to consider is the self-induction of those coils which have the greatest self- induction. The chief inequality introduced into the commutation by the adoption of more than one coil per s'ot is due to the various FIG. 11. Armature with Two Coils per Slot Showing Position of Armature when the Two Coils are being Commutated. coils in the slot being commutated when they are in different magnetic fields. This is evident from Fig. 11, which shows the position of the armature when the first and the last coil in the slot are being commutated. If the brushes are set so that the magnetic field is right for the first coil it will be incorrect for the 38 SPARKING IN DIRECT CURRENT MACHINES last one, and vice versa. So whenever the machine is loaded to its limit the commutating conditions may be so bad for some of the coils, that in time some of the commutator-bars will become pitted and the well-known regularly recurring marking of the commutator-bars will develop. The question is how to take this inequality into account in the sparking formula. To do this, we make the assumptions that the magnetic field varies uniformly from the neutral point to the pole-tip, and that in order to obtain perfect commutation FIG. 12. it is necessary to move the brushes from a position on the neutral point at no load, to a position half-way between the neutral point and the pole-tip at full load. Calling the distance be- tween the neutral point and the pole-tip 2 d, and assuming the brushes fixed on the line P half-way between the pole-tip and the neutral point, then if any coil is commutated when it is at Q distant "a" from P, the machine will only commutate perfectly a current corresponding to 1 - of full load. a Just what this assumption means can be seen from Fig. 13. SPARKING IN DIRECT CURRENT MACHINES 39 Abscissae represent positions along the polar-gap corresponding to Fig. 12, and ordinates represent E.M.F's. The line N A gives the E.M.F. induced at various points by the conductor moving in the field due to the magnets. C P, the ordinate of the line B B, gives the E.M.F. necessary to reverse the full-load current 7 in the coil. If the coil is commutated at the position Q instead of at P then the commutation conditions will be perfect only DQ for a current Hence we assume that if we have sev- eral coils per slot, and that if in consequence of this we have to Q P Q' Circumference of Armature FIG. 13. commutate some of our coils in a position E Q and E' Q', then the current which the machine will carry without sparking is reduced in the ratio - that is CP ' NP' It is very easy to figure out what this inequality amounts to in any particular case. Take 20 slots per pole, 3 coils per slot, and pole-face = 75 per cent of pole-pitch. There are 2.5 slots be- tween the neutral point and the pole-tip. Assuming that the con- ditions are perfect for the centre coil, the outer coils are 0.33 slot pitch distant from this most favorable position. And 1.25 40 SPARKING IN DIRECT CURRENT MACHINES slots corresponding to variation from no load to full load, hence 33 an equality of 0.33 slot- pitch gives an inequality factor -'- 1 ._ > = 0.26; so that the sparking constant should be multiplied by the inequality factor 1.26. Curves can. very easily be plotted for different numbers of slots per pole and coils per slot, in order to facilitate the calcu- lation of this inequality factor. Such curves are shown in Fig. 15. The assumptions on which this calculation is based are to a great extent rational, and though we cannot pretend that the FIG. 14. Armature with Twenty Slots per Pole and Three Coils per Slot. calculation has a rigid basis, yet it is probably as correct as the other sparking calculations, and used with discretion it gives fairly, reliable results. The inequality introduced by the use of a dead coil on the armature is similar to that due to several coils per slot. The dead coil produces a break in the uniformity of the winding; and if the position of the brush is correct for commutation of the coil immediately on one side of the dead coil then it will be one segment out of the correct position, for the coil immediately on the other side of the dead coil. The inequality introduced can be calculated, and allowed for, in exactly the same way as SPARKING IN DIRECT CURRENT MACHINES 41 we estimate the inequality due to several coils per slot. Assum- ing that the brush is in a mean position, then it will be half a segment out of position for the two coils which are next in position to the dead coil. So making the same assumptions as before; if there are n commutator-segments per pole, and if =4 Coils i>er Slot 8=3 Coils per Slot C = 2 Coils per Slot 1.0 10 20 30 Slots per Pole FlG. 15. the pole face = 75 per cent of pole-pitch, then the inequality Q is equivalent to . Thus if there are 20 segments per pole, a dead coil produces an equality equal to 40 per cent of the load, and the inequality should be introduced into the sparking con- stant by the factor 1.4. A curve can readily be plotted between 42 SPARKING IN DIRECT CURRENT MACHINES the inequality factor and the number of commutator segments per pole. Such a curve is shown in Fig. 17. General Formula. Combining all the various factors which affect sparking we get as our complete formula for a sparking constant C = rf (Ik + pc)NifPQ. P being the inequality factor resulting from a number of FIG. 16. Armature with Dead Coils Showing Positions of Armature when the Two Coils Next to Dead Coil are being Commutated. coils per slot, and Q the corresponding factor due to the presence of a dead coil. This formula is not put forward as being scientifically exact, but as an empirical formula which has gradually been built up SPARKING IN DIRECT CURRENT MACHINES 43 as the result of experience , different terms having been added to the formula from time to time when it was found necessary to take different conditions into account. As the formula stands it gives good results, when we know the value of C which can be allowed for the particular design of machine considered. The relative values of C that have been found allowable in different cases arc somewhat as follows: 2-pole 20 4-polc, scries two-circuit winding 35 f>-pole, " .")() 4-pole, multiple wound 30 6-pole, " 35 gradually increasing to 24-pole, multiple wound 50 These relative values, of course, only apply when the machines in each class are designed with similar constants. That is, they should have approximately the same densities in the teeth, and approximately the same ratio of ampere turns per pole on the armature (armature 1 reaction) to the ampere-turns required for the teeth and air gap. If these vary much it is difficult to get consistent results. The brush-gear and the current density in the brushes also play an important part in the sparking. If the brush-gear is weak mechanically, or if the commutator is in bad condition, sparking is sure to take place; while with the average carbon brush, burning will usually take place when the current density reaches 50 amperes per sq. in. The shape of the pole-tips has some effect on the operation of the machine. But as long as they are not too close together, and as long as they are shaped so that the commutation field varies gradually, the exact shape need give us no concern. The density in the armature core (behind the teeth) has also some effect on the allowable sparking constant; and if the core is highly saturated a higher constant can be used than if it is unsaturated. 44 SPARKING IN DIRECT CURRENT MACHINES Assuming that all these conditions are uniform and satisfac- tory, the variation in the allowable value of C, found in actual practice, shows that the formula does not take into account all the conditions that affect the sparking, so the formula must be 1.9 1.8 1.7 1.8 1..') 1.4 1.3 1.S 1.1 1 \ V \ \ \ \ \ ^ \ 10 20 30 FIG. 17. Commutator Segments per Pole-pitch. used with considerable discretion. It cannot be claimed that it is in any way accurate, but it can be considered an empirical working formula, capable of giving good results when carefully used ; and as such it is put forward. [Presented before the American Institute of Electrical Engineers, June 19, 1905.] ROTARY CONVERTERS AND MOTOR-GENERATORS AT the present time the alternating current motor in a motor-generator set of 100 K.W. capacity or larger, is usually a synchronous motor; an induction motor is seldom used for this purpose. The reasons for this are, that the lagging current taken by an induction motor renders it undesirable at the end of a long line, that from an operating standpoint the mechanical construction of an induction motor makes it less reliable than a synchronous motor, and that the cost of an induction motor has been heretofore appreciably higher than that of the corresponding synchronous motor. The usual objections to the synchronous motor that it has a low starting torque and that it requires external excitation do not apply to the case of a synchronous motor used in a motor-generator set, as a high starting torque is unnecessary and as there is always some way of exciting the motor whether it is coupled to a direct- current or to an alternating current generator. It has thus become practically standard practice to use synchronous motor- generator sets in all sizes except where the output is too small for a standard synchronous motor. This being the case it is only necessary to consider synchronous motor-generator sets in comparison with rotary converters. Motor-generators and rotary converters can be discussed from two points of view; that of the operating engineer, or that of the designer and manufacturer. As the operating point of view is probably most familiar to engineers, that will be considered first. 45 46 ROTARIES AND MOTOR-GENERATORS Cost and Floor Space. The main points that concern the engineer when installing transforming machinery are the first cost of the machinery, its efficiency, and its reliability and flexibility of operation. Incidentally, the floor space occupied, and sundry other things have to be taken into consideration. The cost of a motor-generator or rotary converter, or rather the price at which it is sold by the manufacturer, depends upon the output and the speed, and incidentally upon the competition among the firms that are trying to secure the business. The choice of the speed for either machine being usually left to the manufacturer, is as high as is consistent with good mechanical and electrical design. The following table gives speeds in R.P.M. which may be regarded as more or less standard for such machines of different output, frequencies, and voltages. 25 CYCLES MOTOR-GENERATORS. ROTARY CONVERTERS. 250 Volts. 600 Volts. 250 Volts. 600 Volts. 250 750 750 500 750 500 500 500 300 500 1000 250 250 187 250 1500 214 214 150 214 2000 187 187 125 167 60 CYCLES MOTOR-GENERATORS. ROTARY CONVERTERS. 250 Volts. 600 Volts. 250 Volts. 600 Volts. 250 720 720 720 900 500 514 514 450 600 1000 240 240 225 300 1500 189 189 189 240 2000 150 150 150 189 In comparing the cost of motor-generators and rotaries it may be assumed that it will always be necessary to use trans- formers with the latter in order to get the comparatively low ROTARIES AND MOTOR-GENERATORS 47 alternating current voltage required. With motor-generators, on the other hand, the motor can be wound to take the high- tension current without the interposition of transformers, unless the line voltage exceeds 15,000 volts. In estimating the costs of motor-generator sets it is assumed that no transformers are necessary. In general, any table of relative costs of motor- generators and rotary converters should be accepted with a certain amount of reserve; as each individual installation must be considered by itself and the costs of the various items com- pared. The cost of the switchboard and cables should also be considered, and in this respect the motor-generator is usually cheaper than the converter. The following table gives the cost, 500 K.W. 600 VOLTS, 25 CYCLES Rotary Converter. Motor-Generator. Cost $4500 + 2700 = $7200 $9000 Efficiency 1.25 load " i 91.5 91 5 88. 87 5 " 0.75 " " 005 " 91.0 88 5 85.5 81 Floor soace . . 60 + 50 = 110sa. ft. 85 so. ft. 500 K.W. 600 VOLTS, 60 CYCLES Rotary Generators. Motor-Generator. Cost . $4700 + 2300 - $7000 $8700 Efficiency 1.25 load . 90 5 88 " 1 90 5 87 5 " 0.75 " 89 5 85 5 " 0.50 " ... 86 5 81 Floor space 70 + 50120 sq ft 90 sq f t efficiency, and the floor space, required for rotary converters and motor-generators of different outputs. The rotaries are assumed to operate in connection with three single phase 6,600 volt transformers, and the motor-generators to operate on 6,600 48 ROTARIES AND MOTOR-GENERATORS volts without transformers. The efficiencies are the combined efficiencies of the sets; rotaries and transformers in the one case, and motors and generators in the other. In the case of rotaries, under the head of cost and floor-space, the first figure refers to the rotary and the second to the transformers. 1500 K.W., 275 VOLTS, 25 CYCLES Rotary Converter. Motor-Generator. Cost $18000 + 6300 $24300 $21000 Efficiency 1.25 load 93 5 90 5 " 1 93 5 90 " 075 " 92 5 88 " 050 " 90 5 85 Floor space 240 -f 125 = 345 sq ft 320 sq ft The above are sale prices f.o.b. factory and do not include freight or erection charges. All necessary rheostats and shunts are included, but no induction regulators for the rotaries. In all three cases it will be noticed that the rotary converter and transformers are the more efficient, the difference in efficiency being about 3 per cent at full load and about 6 per cent at half load. The value of this difference in efficiency has to be decided in each case by the cost of producing the extra kilowatt-hours. In a water-power plant the efficiency is an unimportant feature; in a steam plant, where the cost of fuel is high, it is quite im- portant. The floor space taken up by a rotary converter and its transformers is about 25 per cent greater than that taken up by a two-bearing motor-generator set. The floor space is only of importance in the case of a sub-station in a city where real estate is valuable, and in such cases the transformers could be placed in a gallery over the rotaries if desired. As the rotaries them- selves only take up about two-thirds the floor-space of a motor- generator set, the advantage would, with this arrangement, be with them. Operating Characteristics. The relative desirability of rotary converters and motor-generators from the operating point of ROTARIES AND MOTOR-GENERATORS 49 view depends upon the question of their reliability and flexibility of operation. As regards flexibility, the motor-generator is of course by far the better. With motor-generators the power- factor of the motor may be adjusted to unity, or if desired a leading current may be introduced into the line without affecting the operation of the motor, while the voltage on the direct current side may be adjusted within wide limits either by use of the shunt rheostat, or by compounding. Neither of these adjustments can be applied conveniently to a rotary. In a rotary converter the ratio of the voltage on the direct current side to that on the alternating current side is practically constant; that is, any drop or rise of voltage on the line affects proportionally the direct current voltage of the rotary. In attempting to regulate the power factor of the rotary or to in- troduce leading or lagging currents into the line by varying the field strength, we are liable to alter the alternating current voltage at the end of the line, and hence to affect the direct current voltage of the rotary. Variation of the shunt current within wide limits is also objectionable, because it often has a tendency to produce hunting. Theoretically a rotary converter can be compounded or over-compounded, or rather the line can be over-compounded, producing the same effect on the direct current terminal voltage as if the rotary itself was over-compounded. This may be done by introducing an artificial self-induction into the line, and producing by means of a series winding on the converter fields a leading current approximately proportional to the load on the rotary. This leading current will then raise the voltage of the line because of the self-induction present. This compounding is, however, at best a rough method and can be used only on sys- tems in which exactness of voltage is unimportant, as it is some- what difficult to adjust the self-induction and the series winding to give the required effect. With this method the power factor of the rotary and that of the system are also varied within wide limits as the load varies, and a change of voltage affects all other machinery on the line. So that the cases are limited in which 4 50 ROTARIES AND MOTOR-GENERATORS this method can be used to regulate the direct-current voltage on a rotary. A method of voltage control with rotary converters is used on some of the Edison systems. An induction regulator is inserted in the alternating current circuit between the trans- former and the rotary, and is usually controlled from a switch- board by means of a pilot motor. Such a regulator increases the cost of the apparatus about 20 per cent, decreases the total efficiency about 1 per cent, adds about one-third to the floor space required by the transformers, and introduces additional complications into the system. Usually, therefore, the necessity for employing an induction regulator is a strong argument against the use of rotary converters in the particular installation considered. A rotary converter is more liable to hunt and to flash over on short circuits, and is a somewhat more complicated piece of apparatus than a motor-generator set. On the other hand, a syn- chronous motor wound for a pressure of over 6,600 volts is not so reliable as a transformer wound for the same voltage. Generally speaking, from the point of view of reliability of operation, there is little choice between 25 cycle rotaries and 25 cycle motor-generators, although in a 60 cycle installation the advan- tage is decidedly in favor of the motor-generator. There is no doubt that satisfactory 60 cycle rotaries can be made up to 600 volts, but their design is more difficult than that of 25 cycle rotaries, so that it is natural they should require more attention than motor-generator sets. Motor-generators have another advantage over rotary con- verters in that they are not so liable to hunt. Of course, hunt- ing can be prevented, but not usually without introducing some corresponding disadvantage. Dampers may be placed on the pole faces, but with the disadvantage of causing some loss in ef- ficiency; or extreme uniformity of engine speed may be obtained at the expense of a heavy fly-wheel; while the small line drop that is usually found necessary for the reliable parallel operation of rotary converters requires considerable expense for conduc- ROT ARIES AND MOTOR-GENERATORS 51 tors. Synchronous motors are not so liable to hunt as rotary converters, and the conditions that are good enough to insure the satisfactory parallel operation of alternators are usually all that are required to prevent hunting in synchronous motors. From the operating engineer's standpoint, a motor-generator is preferable to a rotary converter in almost every respect, except as to efficiency and cost ; and even as to cost a motor- generator is the cheaper for low voltages and large outputs. Consequently, when comparatively cheap medium size units are wanted, and close voltage regulation is unimportant, rotary converters are used. But when large units are desired, and the voltage regulation is important, as in incandescent lighting, motor-generators are employed. This applies to both 60 and 25 cycles. So far, rotary converters have only been considered for transforming from alternating current to direct current. In regard to inverted rotary converters; that is, rotaries for trans- forming from direct current to alternating current, almost the same remarks apply. In addition, however, inverted rotaries are subject to another disadvantage: that the power factor of the load on the alternating current side affects the magnetic flux, and in consequence the speed and frequency of the ro- tary. A heavy inductive load on the alternating current side tends to make the rotary run away. This, of course, can be prevented by an automatic speed limit device; or, to a cer- tain extent, by separately exciting the rotary from an under saturated exciter which it drives mechanically, or by making the armature of the rotary very weak in comparison with its field magnet. But these devices are makeshifts, and none of them can keep the speed absolutely constant under these conditions. And when the speed varies it also causes variation in the speed of all induction and synchronous motors, driven from the rotary, which is highly objectionable. This, combined with its other faults, makes an inverted rotary usually less desirable than a motor-generator. 52 ROTARIES AND MOTOR-GENERATORS Design of Rotary Converters and Motor-Generators. Generally speaking, and within reasonable limits, the higher the speed of a machine the less is its cost, so that it is to the interest of the manufacturer to run at as high a speed as possible. (See NOTE p. 64.) The permissible speed for any alternator is limited only by mechanical considerations, while the maximum speed at which a direct- current generator or a rotary converter of a given output can be run is limited by the operating charac- teristics in regard to sparking. Given the approximate speed at which a direct-current machine will run, the number of poles which it should have, is fixed within narrow limits by questions of sparking and economy of design. As the number of poles and speed determine the frequency, it is easily seen how the choice of speed for which a rotary of given output, frequency, and voltage may be built is limited. Suppose a 1,000 K.W., 25 cycle, 600 volt rotary is to be designed: to insure the cheapest machine the number of poles must be as few as possible so that the speed can be high. There are 1,670 amperes to commutate; this, to a great extent, deter- mines the number of poles. On laying out the design, it is found that 8 or 10 poles will suffice, but that 12 poles will be more conservative. Let 250 rev. per min. be decided upon. Assuming a pole-pitch of 21 in., we obtain an armature diameter of 80 in., 24 slots per pole, two coils per slot, length of armature core 13.5 in. The slots are comparatively narrow, only 0.4 in. wide, so that solid pole-faces or copper dampers can be used. This makes a very good machine. Fig. 18 shows such a machine. If a 1,000 K.W., 600 volt motor-generator is to be designed, the most suitable number of poles on the direct current generator can be decided on, and the speed may be made as high as is consistent with good commutation. The speed may be as high as 300 rev. per min., but, as in the case of the rotary, a more conservative machine would result if the speed were kept dow r n to 250 rev. per min. Twelve poles is a suitable number, and as a somewhat better sparking constant is required than in a ROTARIES AND MOTOR-GENERATORS 53 rotary, the armature is built with a slightly larger diameter. An armature of 80 in. diameter, with 10 slots per pole and three coils per slot, is satisfactory. The width of slot is not limited, as laminated pole-faces are to be used, so that a wide slot can be adopted with its consequent reduction in the self-induction of the commutated coil. This gives an armature length of 10.5 in., airl also makes a very good machine. FIG. 18. 1,000 K.W. 600 Volt, 25 Cycle Rotary Converter. This is one instance in which we find that the rotary con- verter and motor-generator will operate at the same speed. In this case, the cost of the rotary and transformers will be less than the corresponding motor-generator set. Suppose, on the other hand, a 1,500 K.W., 275 volt, 25 cycle machine is to be designed. It is found that the minimum number of poles for a rotary of this output and voltage is about 54 ROTARIES AND MOTOR- GENERATORS 20. This gives a 20 pole rotary at 150 rev. per min., with an armature 130 in. diameter, 13 in. long, 24 slots per pole, and one coil per slot. The corresponding generator for the motor-generator set may be run at 250 rev. per min. by making it with 18 poles. This gives: armature 110 in. diameter, 9.5 in. long, 14 slots per pole, and two coils per slot. In this case the cost of the motor-generator will be less than that of the corresponding rotary and transformers, on account of higher speed at which the motor-generator can be run. Such a motor-generator is shown in Fig. 19. Speaking generally from the designer's standpoint, there is little to choose between the difficulty of designing a rotary converter and that of designing a direct current generator of the same output, speed, and voltage. The design of a rotary is subject to more limitations than that of a direct current generator. The number of commutator segments per pair of poles must be divisible by the number of phases, and the number of slots per pair of poles should preferably be also divisible by the same number. Also the relative dimensions of the slot and air gap are limited by the fact that eddy currents must be avoided in the solid pole-faces, or copper dampers, which are usually employed to prevent hunting. On the other hand, the absence of armature reaction in a rotary converter is a consider- able point in its favor as regards tendency to sparking. Investigating the conditions of commutation in a direct- current generator by means of a pilot brush, when the machine is operating at full rated output with brushes set in the normal position, it is generally found that resistance commutation is taking place; that is, the brushes are advanced just far enough for the armature cross magnetization field to neutralize the direct field due to the magnets at the point of commutation. As the load on the machine is increased, the increased armature- reaction causes the resultant field at the point of commutation to become of the opposite sign to that which would be required for perfect commutation, thus tending to make the brushes ROTARIES AM) MOTOR-GENERATORS 5o spark. At the same time the increased current, which has to be commutated, also has a tendency to make the brushes spark unless the resistance of the brush contact is sufficiently high. Thus assuming the generator is delivering the maximum current that the brushes could commutate by the varying resistance of their contacts if they were in a zero resultant magnetic field, any increase of load on the direct current generator will cause the brushes to spark, for two reasons: first, because of increased 899 B FIG. 19. 1,500 K.W. 6,600 Volt, 3 Phase, 25 Cycle and 300 Volt D.C. 250 R.P.M. Motor-Generator. armature-reaction; second, because the current becomes too great to be taken care of by resistance commutation, even assum- ing no armature-reaction. In a rotary converter, on the other hand, the armature- reaction effect is not present, and the brushes may bo assumed at all times to be either in a neutral field or in one that is help- ing the commutation. The result of this is that a direct current machine operated as a rotary converter will carry, as regards sparking, heavier overloads with fixed brushes than will the 56 ROTARIES AND MOTOR-GENERATORS same machine as a generator. The sparking does not appear to increase so rapidly in a rotary converter when the load is raised. The result of this is that the sparking constant in a rotary is usually permitted to be about 25 per cent higher than in a generator, and in consequence of this a rotary converter may be designed for a lower peripheral speed, and with a longer armature core than the corresponding generator. There being no resultant armature-reaction in a rotary converter, some manufacturers design such machines with a high armature-reaction and low volts per commutator bar; that is, with a strong armature and weak field. This design requires less material in the construction and tends to low r er the cost of the machine. But this saving is not so pronounced as one might think at first sight, as the labor cost is increased quite materially when the number of coils on the armature and the number of commutator bars are increased. In addition, a strong armature is not conducive to good operation in a rotary converter; it tends to make the brushes flash badly or even to flash over when starting from the alternating current side, or when hunting, and it also reduces the synchronizing power of the machine. Considering the question from all points of view, it is usually found most satisfactory to design rotary converters with about the same armature reaction and volts per bar as the corresponding direct current machines. Rotary Converter Armature Winding. The copper loss in the armature of a polyphase rotary converter is usually consider- ably less than in the corresponding direct current generator, so that such rotaries are often designed with a much smaller cross-section of copper than would be used in a generator. This is bad practice. The copper loss in a rotary converter armature is not equally distributed, the loss in the bars nearest the collector leads being usually much greater than in those midway between the leads. And although the difference in temperature at the end of a temperature run cannot generally be detected by a thermometer, the difference is very appreciable in the case of sudden overload at a low power factor. And cases- ROTARIES AND MOTOR-GENERATORS 57 are on record where the armature bars connected to the leads in a large rotary have been fused before the other bars got dangerously hot. For heavy railway work the section of the copper in a rotary armature ought to be at least equal to that in the corresponding generator. As regards heating, six phase rotary converters have an advantage over two or three phase, but as they result in extra complications in the cables and switchboard they are seldom employed except in large units. In any case a six phase rotary may have to operate three phase at some time, so it should be designed simply as a three phase machine with three extra collector rings. This being the case, the remarks before made in regard to the section of the armature copper and heating of conductors also apply to six phase machines. One advantage sometimes claimed for the six phase rotary is that having a greater number of equipotential connections on the armature than either a two or three phase, there is a greater tendency towards equality in distribution of current between the various sets of brushes on the commutator. This would be true if the collector rings were the only equipotential connections on the armature. But in addition to the collector rings, the armature winding of the modern high speed rotary is usually provided with an equipotential connection for every second slot, while in 60 cycle or large 25 cycle rotaries where the commutating conditions are more severe, one connection to every slot is frequently employed. So it is obvious that, in this re- spect at least, the six phase rotary as usually constructed pre- sents no advantage over the two or three phase. Undoubtedly the larger the number of phases in the armature winding of a rotary converter, the more accurately and uniformly do the alternating and direct currents in the conductors neutralize one another, and in consequence, the less the pulsation of the armature reaction and the less the variation in the commutating conditions. But with the modern well proportioned rotary converter, this variation and pulsation is not a serious matter, so that it should not influence the choice of the number of phases 58 ROTARIES AND MOTOR-GENERATORS any more than the necessity for additional equipotential con- nections should. An important feature in the design of the armature of a rotary converter is often overlooked, and that is equality or balancing of the phases. If the windings of the different phases on a rotary armature are not all exactly equal, and placed on the armature in an exactly similar and symmetrical position with regard to one another, then the phases will be unbalanced, with the result that the load will not be balanced among the phases, and that there will be a greater tendency to hunting. If the three ammeters in the three phases of such a rotary are watched, the load can be seen changing from one phase to an- other. This is the reason why rotary converters with series wound armatures are usually unsatisfactory. It is often im- possible to balance the phases. A 6 pole, three phase rotary having a series wound armature with 224 coils, must have the rings connected to coils 1-26-51. An 8 pole three phase rotary having a multiple wound armature with 520 coils, must have its rings connected to coils 1-44-86. The phases of both these armatures are unbalanced. To have the phases perfectly balanced the number of commutator bars per pair of poles should not only be divisible by the number of phases, but the number of slots per pah- of poles should also be divisible by the same number. The reason for this is that to have the different phase windings all symmetrically and similarly placed on the armature, all the coils that are connected to the phase leads must be in the same relative positions in their slots. Hunting of Rotary Converters and Synchronous Motors. Rotary converters are more liable to hunt than synchronous motors. Generally speaking, conditions of operation and design that will enable two alternators to operate satisfactorily in parallel without hunting, will also enable one of them to operate satisfactorily as a synchronous motor when driven by the other machine under the same conditions. A single rotary driven from an alternating current system, will not operate under given conditions quite so well as regards hunting, as a motor- ROTARIES AND MOTOR-GENERATORS 59 generator. The armature reaction of a rotary converter is considerably higher in proportion than that of a synchronous motor. Therefore, its synchronizing power is less and the fact that the direct current side is so intimately connected to the alternating current side, makes it peculiarity sensitive to hunting. But the main difficulties with rotaries in regard to hunting are experienced when two or more rotaries are running in parallel on the same alternating current system, and feeding into the same direct current system. These difficulties are especially marked when the rotaries are running in parallel upon the same alternating current and the same direct current bus-bars. Under similar conditions motor-generators are no more sen- sitive to hunting than under the conditions of singly operated units. But the difficulties with rotaries are often so serious, that operating engineers have found it necessary to insist that manufacturers provide damping or anti-hunting devices for all machines intended to be connected in this way, and also to install artificial choke coils between the collector rings and the alternating current bus-bars, in order to limit the interchange of current. Artificial damping devices of various types and forms have been tried, but the only one in extensive use at the present time is a heavy grid of copper embedded in the pole face. Another construction now in use for accomplishing the same result, consists in solid pole faces so shaped, and with the armature slots and air gap so proportioned, that eddy currents due to the teeth are avoided. As far as can be seen from practical operation, these two methods of preventing hunting seem equally effective. They both enable rotaries to be run in parallel satisfactorily, as long as the variation in speed of the engines and the pressure drop in the feeders is not excessive. As regards the relative effects of a copper grid damper and a solid pole face upon the efficiency under working conditions, it is difficult to speak with any degree of accuracy as so much depends on the uniformity of speed of the engines. But it is probable there is a constant loss in the copper-grid damper, 60 ROTARIES AND MOTOR-GENERATORS when, as is often the case, it is used with large armature teeth and small air gaps. Solid pole faces require a larger air gap or narrower slot, which in turn demands more copper on the magnets, this being especially the case in 60 cycle converters. On the other hand, from a mechanical point of view it is a nuis- ance to have to attach auxiliary copper grids or dampers to any pole face, while the cost of the dampers themselves is not insignificant. Generally speaking, then, there is little choice between solid pole-faces and auxiliary dampers, so it would perhaps be best to advise the use of solid pole-faces in all cases, as they are simpler and more mechanical. Mechanical Design. Though rotary converters have been used for quite a number of years, yet their detail design seems to have received less care than has been given to generator design. A rotary in a sub-station carrying a street railway or interurban load is usually subjected to rough treatment, and consequently should be of robust design. All parts should be as accessible as possible in order to facilitate repairing. Two features in the design of rotaries that are often faulty are the alternating current collector gear and leads, and the starting resistance used for starting from the direct current side. The alternating current end of a rotary is often designed sim- ilarly to the old revolving-armature alternators; that is, the leads are strap-copper soldered to the armature conductor and attached to the armature end-plate with a few cleats. While the collector-rings are mounted solid upon the shaft, separated and insulated by fiber or wood discs and bushings, the leads being embedded in this insulation where they pass through or are connected to the rings. This construction is shown in Fig. 20, and it is hardly necessary to state that it is unsatisfactory for heavy railway work. Solid copper leads, unless very carefully and solidly cleated to the armature end plates, are liable to break due to vibration, while soldered joints are liable to melt under overloads. Collector-rings mounted solid on the shaft are liable to break down due to warping or cracking of the insulation, and to get hot on overloads due to poor cooling NOTARIES AND MOTOR -GENERATORS 61 facilities. Also when the rings are insulated by wooden discs projecting between the rings they cannot be easily turned off when they become cut or grooved by the metal brushes. The most substantial construction is probably to use cable for the leads and to connect them to the winding by special lugs riveted and silver-soldered to the armature conductors. The rings should be carried on arms projecting from a spider and should be freely FIG. 20. Rotary Converter Armature and Collector-Rings. open to the air for cooling, and be easily accessible for turning off whenever they become grooved or uneven from wear. In fact they should be designed exactly like the collector gear of an up- to-date revolving-field alternator. When this construction is adopted a temperature rise of over 15 degrees is rarely attained on normal load. Such collector-rings are shown in Fig. 21. ROTARIES AND MOTOR-GENERATORS Regarding a starting resistance for starting a rotary or a motor-generator from the direct current side, it is often for- gotten that such work is much more severe than starting a motor. A rotary or a motor-generator has to be run up to speed and then synchronized, and when synchronizing the speed has r , FIG. 21. A.C. Collector-Rings on Rotary Converter. to be exact. If the voltage on the direct current system is vary- ing, it often requires several minutes to synchronize, and if it is varying suddenly, the speed can only be adjusted by means of the starting resistance because shunt control is not quick enough. Starting resistances for rotaries or motor-generators should be designed so that the last steps can be kept in circuit ROTARIES AND MOTOR-GENERATORS 63 for at least five minutes without overheating. An oil-cooled starting resistance that has been designed by the National Electric Co. for this work is shown in Fig. 22. The resistance coils are of iron wire supported on porcelain insulators and brazed to heavy brass terminals. It is designed so that the temperature rise of the oil will be 150 Cent, in five minutes with all the resistance coils carrying their maximum rated cur- rent. This type of resistance gives excellent results, and has the additional advantages of being quite cheap and almost fireproof. The usual starting switch, with overload and no-load release, FIG. 22. Starting Resistance for Rotary Converter or Motor-Generator. is neither necessary nor desirable for rotary converter work; it is too complicated and expensive and not reliable enough for such duty. A standard multiple-contact switch is all that is required. Such a switch requires very little space and may be mounted on the rotary panel. This paper has not been written with the idea of advocating the use of rotary converters instead of motor-generators, or vice versa, but more with the idea of comparing them generally, their advantages and disadvantages, and of pointing out some of the characteristic features of each machine. The question as to which type of machine should be used in any given case 64 ROTARIES AND MOTOR- GENERATORS can only be decided after every feature of the situation has been duly considered. Broadly speaking, however, the tendency to-day is toward motor-generator sets in lighting systems and rotary converters on traction systems. This seems to be per- haps the most rational conclusion. Motor-generators and rotary converters, all things con- sidered, are more difficult to design and build than the ordinary standard engine-type generator. They are high-speed machines and usually operate in sub-stations where the conditions of operation and the supervision are not of the best. The question of their reliability in continuous operation should therefore receive the most careful consideration from the designer and manufacturer; and as suggestions and criticisms from operating engineers are of the utmost value they should always be welcomed and investigated, in order to determine whether or not they contain features valuable enough to warrant the modification of standard designs. NOTE [Dec., 1910] In the past two years the standard speed for rotary con- verters and motor-generators has appreciably increased, this being due partly to improvements in mechanical design; partly to the development of commutating poles for direct current generators and to a certain extent for rotary converters; and partly to our increased knowledge of power system phenomena and to the education of po\ver station engineers, which has per- mitted closer commercial designing. The present commutating pole units require careful adjustment, but, after adjustment are much less sensitive to operating conditions, and require less attention and maintenance. As stated above, increasing the speed of any unit tends to decrease the cost, though this is true only within certain limits. This is exemplified by the direct current generators which are now built for operation at the very high speed required for direct connection to steam turbines, and whose cost is, on account of the mechanical difficulties of construction due to the ROTARIES AND MOTOR-GENERATORS 65 high speed, appreciably greater than that of the corresponding medium speed unit. Given the approximate speed at which a direct current generator is to operate, and adopting a commutat- ing pole construction, the number of poles which the machine should have is fixed within narrow limits by questions of cost, efficiency, and mechanical design. If we carefully design a unit of given rating, whether A.C., or D.f. (with commutating poles), for a number of different speeds, and figure the efficiency STANDARD ROTARY CONVERTER AND MOTOR- GENERATOR SPEEDS (1010) 25 C'YCLKS MOTOR-GENERATORS. NOTARY CO.VVKRTKR*. K.W. 250 Volts. 600 Volts. i.'.-,o Volts. GDO Volts. 250 750 750 750 750 500 750 750 500 500 1000 500 500 300 375 1500 375 375 214 250 2000 3f /{aliny. The question of uniformity in the rating of generators and motors is in a decidedly unsatisfactory condition at the present time, a< almost every manufacturer and every purchaser rates his generators or motors, and specifies their performance in a different way. The result of this is that when an operator decides a new generator is required he specifies as nearly as possible a unit he considers suitable for the work, buys one that the manufacturer estimates will fill his specifications, and then proceeds to test, to find what output he can obtain from it under the conditions that exist in his station. The operating engineer and the manufacturer are both to blame for this state of affairs; the manufacturer because he often designs and builds his machine to take care of some arbitrary theoretical, rather than practical, conditions of operation; and the operating en- gineer because he does not more closely study the operating conditions of his machinery, and insist that the manufacturer supply generators or motors suitable to his requirements. So long as the mechanical construction of a machine is satisfactory and such that repairs can be easily carried out, no further in- quiries are usually made by the purchaser, provided the name- plate carries the nominal rating required. The tendency is to pay altogether too much attention to the figures stamped on the name-plate when buying a generator or motor. The result is that almost every engineer knows examples of units *with the same nominal rating, built either by the same or by different 8 113 114 SPECIFICATIONS AND RATINGS manufacturers, some of which under certain operating conditions are capable of carrying 50 per cent more load than others. The remedy for this, of course, is to have some uniform and rational method of rating adopted by all manufacturers, and then to specify and select machines suitable for the work they have to perform. Temperature and Power-Factor. The two most important points in which specification and performance guarantees are unsatisfactory seem to be: 1. The temperature rise on alternating and direct current generators or motors; and 2. The power-factor of the load for which guarantees are made on alternating current generators. The present system of temperature guarantees, which con- sists in stating the maximum temperature rise in any part of the machine, under numerous different conditions of load and for varying periods, is both unsatisfactory and irrational. The temperature limit of output in any generator or motor is not decided by the temperature rise, but by the absolute temperature of the insulation. At one of the specified loads the armature iron may be the hottest part, at another the field coils, while at a third it may be the armature winding or commutator. As each of these parts has a different limiting temperature to which it can be subjected without damage, it is obvious that such a sys- tem of uniform guarantees is misleading. The operating engin- eer is not practically interested in knowing that the temperature rise of his generator, as measured by a thermometer on the out- side of the winding, is 35 degrees on full load for 24 hours, 45 degrees on 25 per cent overload for 24 hours, and 55 degrees on 50 per cent overload for one hour. What he must know is the maximum load he can safely carry continuously, and in some cases the overload he can safely carry for two or three hours. This load depends, obviously, on the room temperature and on the limiting temperature to which the insulation can be sub- jected without damage. The most rational method of tempera^ ture rating is, then, to specify the maximum continuous rating SPECIFICATIONS AND RATINGS 115 at which the unit can be safely operated with a certain room temperature, e.g., 25 degrees centigrade; and where desirable the safe two or three hour overload, with the same room tempera- ture, can also be given. Usually, in a modern, well ventilated generator or motor, the temperature reaches its maximum after a three or four hour run, so that the two or three hour overload is about the same as the maximum continuous rating. In this case the system of temperature guarantees reduces to a single guarantee of the maximum safe load which the unit can carry continuously, with the specified room temperature; and it should be noted that this maximum load is greater the lower the temper- ature of the air cooling the machine. This system of maximum rating has been in use to some extent for the past year, and it would seem that the sooner it is adopted universally, the sooner will purchasers have a rational idea in regard to the temperature limitations of the machines they arc buying. Few engineers seem to appreciate the effect of a low power- factor on the operation of an alternator, and few operating engi- neers consider the power-factor of the load an important point when installing new machinery. When new generators are to be bought, we regularly find 100 per cent power-factor machines specified for a station operating with a power-factor varying from 65 to 85 per cent; and the purchaser becomes suspicious when he is told that a standard 100 per cent power-factor generator would not carry much more than 50 per cent of its rated kilovolt- amperes if operating on his system, or that he ought to buy a more expensive generator which is designed and guaranteed for a 75 to 80 per cent power-factor load. A standard alternator designed for 100 per cent power-factor load (i.e., for rotary converter or synchronous motor work) is usually designed with a comparatively saturated magnetic circuit, and, unless extreme- ly liberally designed, such a machine will not hold up voltage with full rated K.V.A. at 80 or 90 per cent power-factor. If such a machine were required for 80 per cent power-factor, it would be re-designed with an unsaturated magnetic circuit, given a rating of about 75 per cent of its nominal K.V.A. rating at 100 per cent 116 SPECIFICATIONS AND RATINGS power-factor, and possibly a higher temperature rise specified for the field coils. Any method of giving alternators a different rating for every operating power-factor would probably be too complicated for practical work. It has, therefore, been proposed to continue to give all machines a nominal rating in kilo volt-amperes at 100 per cent power-factor, and in addition to give the maximum load which they will safely carry at various lower operating power- factors. This maximum load at low power-factors, is decided for some machines by the question of holding up voltage, and for others by the heating of the field coils, so that for cases in which temperature is the limit the maximum load should be referred to a definite room temperature, e.g., 25 degrees centigrade. This method of rating alternators gives for the purpose of comparison a nominal rating at 100 per cent power- factor, and in addition gives the purchaser exact information as. to the operative characteristics of the proposed unit under any particular condition of load. If the operating engineer knows the power-factor at which the machine will be required to operate, he should have no difficulty in deciding as to the suitability of the unit for his requirements. The question of power-factor is equally important in rotary converters and synchronous motors. Generally speaking, the power-factor should always be adjusted to 100 per cent, unless for some special reason, definitely specified and understood at the time the machine was purchased. A case recently came to the writer's knowledge in which a system was operating its rotaries at 90 per cent power-factor, and when the manufacturer pro- tested on behalf of his generators and rotaries, the operating engineer stated that he considered 90 per cent a " mighty good " power-factor. Possibly it would be for induction motors, but for rotary converters it is a " mighty bad " one. Synchronous motors are often used to correct the power-factor of the line, but when they are installed with that intention the maximum kiloyolt- amperes at the required power-factor should be specified, exactly as in the case of an alternating current generator. A synchron- SPECIFICATIONS AND RATINGS 117 ous motor designed to operate at 100 per cent power-factor, is just as unsuitable for operating at a low power-factor as an alternator would be in a similar case. Limiting Conditions. Neglecting the question of efficiency, the limit for operating conditions should, in all cases, be decided by the resultant injury to the machine. The limit of tempera- ture rise is decided by the damage to the insulation, or to the mechanical construction, of the part of the machine considered. Some insulation will not stand continuously a temperature higher than 90 degrees centigrade without deterioration; other insulation will stand 300 degrees safely. The temperature limit on a commutator or collector-rings is usually decided by the shrinkage of insulation or unequal expansion of the materials, causing loss of mechanical balance, or loss of accuracy on the wearing surface. The limit of the allowable sparking on a commutator or collector-rings is the resulting temperature rise, the damage to the surface of the commutator or collector-rings, and the disintegration of the brushes. All these effects must be considered in relation to the duration of the specified load ; and as in such cases it is difficult for the purchaser to decide, without an actual test, the amount of damage that will result from a certain condition of operation, he must, to a great extent, de- pend upon the guarantees of the manufacturer; which guaran- tees, however, will be subsequently checked by the actual opera- tion of the machine in service. Testing and Specifications. Testing to determine in what degree a unit meets the specified detailed performance guarantees is always a very difficult question. It is almost impossible to get an accurate direct test of the voltage regulation of any alternator, as the result is measured only as the difference of two high readings, and a variation in any one of the conditions of test affects the result. Unless made in a laboratory, away from masses of iron which would affect the accuracy of the in- struments, an input-output efficiency test of a motor-generator can not be made with a greater accuracy than 2 or 3 per cent on account of the impossibility of obtaining accurate instrument 118 SPECIFICATIONS AND RATINGS readings under practical conditions. In both of these cases, the direct method of test has to be abandoned in favor of an indirect method, which enables more accurate results to be obtained. Temperature tests are very difficult to carry out accurately, and unless careful precautions are taken by experienced observers, it is often impossible to be sure of results to 5 degrees. Generally speaking, the customer will more profitably spend his time in investigating the conditions of operation, and in specifying a machine suitable to operate under these conditions, than in making tests to determine regulation, temperature rise, and other similar characteristics which will be of only doubtful accuracy and value when made. If a generator or motor is specified suitable for the work, then the most satisfactory and convincing test for the machine is the manner in which it performs its work; and by means of suitable performance specification this should be made something definite, and something to which the customer can hold the manufacturer. If this were done, we should have fewer disputes on the question of whether or not a machine satisfies its contract guarantees, we should have fewer unsuitable units installed, and I think we should have better operating con- ditions on most power systems. [Presented before the National Electric Light Association, June 3, 1909.] INPUT-OUTPUT EFFICIENCY TESTS ON ROTARY CONVERTERS AND MOTOR-GENERATORS THE objection to the input-output method of testing the efficiency of rotary converters or motor-generator sets is in its inaccuracy. This is due to the fact that an error of one per cent in any reading results in a one per cent error in the efficiency; and when it is considered that it is very difficult to duplicate in- strument readings with a greater accuracy than from one or two per cent, the importance of this objection can be realized. In such a test we have two readings to take, one of the input and one of the output ; and an error of one per cent in each of these readings may result in an error of two per cent in the efficiency. The reason for the popularity of the input-output method of measuring efficiency is the extreme simplicity of the test : as it only requires simultaneous readings to be taken of the current, voltage, and watts on both the alternating and direct current sides of the machine. The fact that it is a simple test which is easily made, and that it gives direct readings without any calculation, has led engineers to often overlook the fact that its accuracy under the usual conditions of test is very doubtful. Measuring instruments, even when carefully calibrated, cannot be relied upon to give readings to a greater accuracy than one per cent under practical operating conditions, and the error is often nearer two or three per cent. The reason for this is that the presence of large masses of iron or stray magnetic fields affects their accuracy very considerably, and these local conditions are 119 120 INPUT-OUTPUT EFFICIENCY TESTS continually changing and rarely the same for any two tests. In addition to the inaccuracies of the instruments, errors in obser- vation also affect the results, and it requires extremely careful work on the part of accurate and experienced observers to repeat readings consistently to one per cent. As a result of this, it is very doubtful whether the readings taken under practical operat- ing conditions, in the factory or in the sub-station, can be ob- tained with a greater average accuracy than two per cent. And, when it is considered that this inaccuracy of two per cent may occur in both the input and output readings, resulting in a possible error of four per cent in the combined efficiency, we can see that such a test is valueless to check guarantees. An additional source of inaccuracy in a commercial input- output sub-station efficiency test is that due to the power-factor of any synchronous alternating current machine tested, being other than 100 per cent, and that due to the increased loss caused by inaccurate setting and poor condition of the brushes in a direct current machine. There are a number of such conditions which affect the efficiency as tested by this method, which are apt to be overlooked, or which are controlled with difficulty under practical conditions. When we contrast the above objections with those which can be raised against the separate loss method, we realize that the extra complication of this method is justified by the more accurate results obtained. In the separate loss method each individual loss is measured separately, and a possible error of two or three per cent in the measurement of the losses, will usually not make an error of more than 0.25 per cent in the combined effi- ciency. Each loss can be measured under favorable conditions, or under the particular conditions which are being considered, and when the test is completed there can be no suspicion that the accuracy of the results has been influenced by unknown factors. Separate measurement of the individual losses also shows at once how the losses are distributed, and allows the operating characteristics of the machine under various con- ditions of load to be defined more accurately. INPUT-OUTPUT EFFICIENCY TESTS 121 The input-output method of testing rotary converters and motor generators has long been considered convenient and sufficient by operating engineers, on account of the ease with which readings can be obtained. But, from what has been said above, it will be realized that the additional time and expense of measuring the efficiency by the separate loss method is well justified by the more accurate results, and by the additional information obtained in regard to the characteristics and opera- tion of the machine. The input-output method can be used as a rough check in cases where accuracy is not of importance, but in all cases where an accurate efficiency test is required there seems to be little doubt that the separate loss test is the only one which can be relied upon. [Presented before the National Electric Light Association, May 26, 1910.] DIRECT CURRENT TURBO GENERATORS Direct Current Supply. Large lighting and railway central stations supplying power to direct current systems are gradually abandoning the use of direct current generating apparatus, and are instead generating alternating current, which is transformed to direct current after distribution to sub-stations. But the number of small power stations and isolated plants generating direct current is increasing yearly, while the use of direct current for excitation of alternators, or for train lighting, also creates a large demand for small and medium size direct current gener- ators. It has long been recognized that this demand is most satisfactorily met by some form of direct current steam turbine driven set. The difficulty in building such a unit is to design a direct current generator which will operate satisfactorily at the high speed required for the economical operation of the steam turbine. The advantages of the steam turbine set are smaller floor space and lower maintenance, due to absence of reciprocating parts. And as these advantages are often relatively important, several methods have been employed for obtaining direct current from a steam turbine driven generator. The most important of these are: 1. A direct current generator direct connected to the steam turbine, and designed to operate satisfactorily at the high speed required. 2. A unipolar generator. 123 124 DIRECT CURRENT TURBO GENERATORS 3. An alternating current synchronous or induction generator direct connected to the steam turbine, and supplying current to a rotary converter which transforms the alternating to direct current. The rotary converter method has, up to the present time, proved to be the most conservative arrangement, and prob- ably will continue so for large units. The unipolar generator has been used in a few special cases, but both it and the combined rotary and alternating current generator, are at the present time being gradually displaced by the direct-driven direct current turbo generator for all small capacity units. Historical. Direct current turbo generators have been built in Europe for the past fifteen years, but then- design and opera- tion has not until recently been sufficiently satisfactory for them to be considered under American conditions. The early direct current turbo generators were built with smooth core armature and copper brushes, so that the operation was poor. Recently, however, several European manufacturers have built direct current turbo generators in sizes up to 1,250 K.W. and 4,000 amperes which have proved much more satisfactory. All the machines above referred to operated with metallic brushes, and naturally suffered from the handicap of excessive maintenance cost. But about two years ago, when the direct current turbo generator was extensively adopted, the question of mainten- ance became so serious, that operating engineers took matters into their own hands and insisted on replacing the metallic brushes by high-grade carbon or graphite brushes. In a number of instances these European direct current turbo generators, which were originally built and shipped from the factory to operate with metallic brushes, have, because of the difficulty of keeping this brush gear in running condition, been modified after installation, so as to operate with carbon brushes. The result of this has been that European manufacturers are now adapting their machines where possible to operate with carbon brushes. American manufacturers realized early that the direct current turbo generator would never be a satisfactory commercial DIRECT CURRENT TURBO GENERATORS 125 machine until it could be built with carbon brushes. But it is only within the last five years, that the skill of designers and manufacturers has been equal to constructing direct current generators to operate satisfactorily with carbon brushes at speeds materially higher than those of the standard belt-driven generator. By careful design with auxiliary commutating poles and by accurate shop-work it has been possible to build motor- generators for doub e the speed which was formerly considered FIG. 28. 50 K.W., 125 Volt, 3,000 R.P.M., D.C. Turbo Generator. possible, while the development of direct current generators for coupling to steam turbines has proceeded under the same con- ditions. It can now be considered that the direct current turbo generator has been developed, suitable for satisfactory operation with carbon brushes under American conditions. They are being built in sizes from 10 to 300 kilowatts at 125 volts, and from 50 to 500 kilowatts at 250 volts, while designers are working on still larger units. Although 600 volt generators do not seem to be in a great demand for this type of unit, a number 126 DIRECT CURRENT TURBO GENERATORS have been in satisfactory operation for some time, notably a 1,000 K.W. unit (consisting of two 500 K.W. generators coupled to one steam turbine), operating at 1,500 R.P.M., which was in- stalled by the North Shore Railway Company, California, in 1907. Designs. The following is a list of approximately standard speeds which have been found most suitable for these generators: K.W. R.P.M. Volts. 10 6000 125 25 4500 125 50 3000 125 and 250 75 2800 100 2400 150 2200 200 2000 300 1800 500 1500 250 All of which machines can be satisfactorily built to operate with commutating poles when carefully designed. At one time en- gineers considered that a commutating pole of almost any design was a universal remedy for all commutating troubles, but ex- perience with direct current turbo generators and other high speed machines, has shown that this is very far from being true. The commutating pole must be proportioned as carefully as the other parts of a machine; and it was the neglect of this fact which caused the failure and abandonment of this device when first used many years ago. The two factors which limit the design of direct current turbo generators are the commutation, and the collection of large currents at high speeds. The commutating difficulties can be satisfactorily overcome in the generators given in the above list, if a properly designed interpole construction is used, though for generators of more special or more extreme ratings a complete system of distributed compensating winding in the pole-faces is usually necessary. The limiting speed for which it is possible to build large direct current turbo generators is decided by the maximum commutator peripheral speed, which can be conserva- DIRECT CURRENT TURBO GENERATORS 127 tively operated with the particular grade of brushes adopted. The following equation limits the design of the commutator: Commutator peripheral speed = Circumferential distance between brushes on the commutator X the number of poles X R.P.M. The minimum distance between brush-arms on the commutator, which can be conservatively allowed for any given voltage, is definitely fixed by the mechanical clearance necessary for accessi- bility and to prevent flashing, and by the space required for the necessary number of commutator segments per pole. The niun- FIG. 29. 100 K.W., 12.5 Volt, 2,400 R.P.M. , B.C. Turbo ( iem-rator. ber of poles is decided by the current the machine is to coin- mutate. The maximum commutator peripheral speed is de- cided by the standard of workmanship and by the type or quality of carbon brushes adopted; while the revolutions per minute should be fixed by the question of maximum economy for the steam turbine. Thus we have the above equation stating a relation between a number of factors, each one of which is subject to restriction. As an example of this we can consider a 500 K.W., 250 volt direct current turbo generator, to operate with a commutator 128 DIRECT CURRENT TURBO GENERATORS peripheral speed of 5,500 feet per minute. We have 2,000 amperes to commutate, which will require at least four, and preferably six poles. The minimum allowable distance between brushes on the commutator for a machine of this size and type is about 7 or 8 inches, while preferably it should be 10 or 12 inches. The more conservative figure would result in a 14-inch diameter commutator at 1,500 R.P.M. for a four-pole machine, or a 21-inch commutator at 1,000 R.P.M. for a six-pole machine, though adopting 7J inches distance between brushes we could operate the six-pole machine at 1,500 R.P.M. , using a 14-inch diameter commutator. But adopting the more conservative figure, and a four-pole machine at 1,500 R.P.M., we must com- mutate 1,000 amperes per pole, which will require approximately twenty-six 1 J" x I" brushes. Thus we will require a commutator 14 inches in diameter and approximately 56 inches long, or two commutators each 14 inches in diameter and 28 inches long. This example shows the difficulties in operating large direct current turbo generators at high speeds when a conservative design is followed ; and explains also why generators of extreme rating, in regard to voltage or current, become so difficult to build. Commutator and Brush-Gear. On account of the careful de- sign and accurate shop work required, the question of collecting large currents at high speed with carbon brushes is the most difficult problem in connection with the design and manufacture of direct current turbo generators. Flexible metallic brushes will operate whether the commutator runs true or not; while having low contact resistance, they are suitable for collection of large currents; and this explains why they were adopted univer- sally on the early European machines. The difficulty in operat- ing with this type of brush is due to the fact that it is almost impossible to entirely eliminate sparking, unless carbon trailing tips are used ; while it is necessary to keep the brushes carefully trimmed if the operation is to be at all reliable. If the brushes are not trimmed frequently, the trailing edge of a copper gauze DIRECT CURRENT TURBO GENERATORS Hi) or wire brush becomes ragged, and when the brush is in such condition a short-circuit or sudden violent change in load is liable to make the machine flash over. A new typo of copper- leaf-graphite brush has been used recently in Kurope with better results, but the operation cannot be considered satisfactory, and the cost is high. Carbon trailing-tips have been used with metallic brushes, but this results in a complicated and sensitive brush-gear, which is almost as difficult to manufacture and keep in operative condition as a brush-gear using entirely carbon brushes. The only reason for the adoption of a carbon trailing- tip and copper brush combination is that it makes possible the use of a smaller commutator than would be necessary with all carbon brush-gear. In addition to the excessive attention re- FIG. 30. Armature for 200 K.W., ll>r> Volt, 1,800 R.P.M., D.C. Turbo (Joiu'rator. quired, the life of metallic brush-gear of all types is so very short that the cost of maintenance usually becomes prohibitive; and this is the main reason why engineers consider that the only satisfactory solution of the commutator problem on a direct current turbo generator is the use of carbon or graphite brushes. The better the quality of these brushes, the more satisfactory the operation, but the greater the difficulty of obtaining span- brushes for renewals. It is an open question whether it is better commercially to design these machines to operate with ordinary good quality graphitic carbon brushes, or with some special high grade imported brush. There is no question, however, but that the direct current turbo generator should have only carbon or graphite brushes if it is to give satisfactory commercial service under American conditions. 9 130 DIRECT CURRENT TURBO GENERATORS The question of good operation with carbon brushes is a mechanical one, and requires a commutator which runs abso- lutely true under all conditions and at all times. Commutators to carry a large current at high speed are always relatively small in diameter and long. The diameter is fixed by the revolutions per minute, and the maximum peripheral speed which can be satisfactorily operated with the particular grade of carbon used, and with the degree of accuracy obtainable in the commutator manufacture. The peripheral speed usually adopted at the present time in America is from 4,500 to 6,000 feet per minute, although peripheral speeds 40 per cent higher than this have been used by European manufacturers with a special grade of brush. With the diameter fixed, the length of the commutator is decided by the questions of temperature rise and correct spac- ing of the requisite number of brushes. When the length of the commutator is greater than about 30 inches it becomes usually advisable to build two commutators of half the length instead of one of full length. These two commutators can be arranged either one at each end of the armature, or both in tandem at one end of the armature; in the latter the bars of the two commuta- tors being connected by suitable lugs. A difficulty wilich is ex- perienced with any long commutator, or with two commutators in tandem, is the lack of uniformity in the distribution of current between the different brushes on each brush-arm. With two commutators, one at either end of the armature, we have diffi- culty in distributing the current equally between the two com- mutators; this difficulty being especially marked if a single armature winding instead of two independent windings is used. The difficulty in obtaining uniform distribution of current is about the same with each of these three types of commutators, and it can be avoided only by selection of a suitable type of brush-holder, with good quality brushes of uniform quality, and a suitable arrangement of generator leads. The standard construction adopted for direct current turbo generators is a cylindrical commutator of the shrink-ring type. Radial commutators have been used to a certain extent in DIRECT CURRENT TURBO GENERATORS 131 Europe with good results, but the inaccessibility of the brushes has prevented their extensive adoption. The only advantage claimed for them is a reduction in over-all length, and less trouble due to vibration of the armature caused by lack of balance. This latter advantage is due to the fact that the operating surface of the commutator is at right angles to the shaft, and consequent- ly in the same plane as any vibration, instead of being at right angles to such a plane, as is the case with a cylindrical commu- tator. The standard cylindrical shrink-ring type of commutator is in small sizes built directly on the shaft, while in large sizes it is built on a bushing. The success or failure of a commutator depends upon extremely accurate shop work, and on the adop- tion of a design such that the deflections and stresses due to centrifugal action and temperature variation are moderate. Accurate and experienced shop work is the foundation of all good operation in direct current turbo generators, and it is this education and development of the shop as much as anything else which has rendered this type of generator possible. The manufacture of high-speed commutators differs from that of the corresponding low speed in that much greater accuracy is required. The micanite or mica used in the construction, in- stead of being a heterogeneous combination of mica and shellac, must be built up of carefully gauged and selected mica segments of uniform thickness regularly arranged and cemented together with the minimum amount of shellac. This micanite has to be suitably treated so that it takes its final dimensions before being placed in the commutator. Every element in the commutator, that is, the copper, micanite, bushing and shrink-rings, must be accurately gauged, and after the commutator is assembled it must also be carefully seasoned, so that there remains no possibil- ity of distortion or of change in the relative position of segments after the machine is placed in operation. Variation in tempera- ture and mechanical stresses are the primary causes of commu- tator mechanical trouble, and the more perfect the commutator the better will it stand these. The Y-ring type of commutator is unsuitable for long high-speed commutators of small diameter, 132 DIRECT CURRENT TURBO GENERATORS as with this construction it is difficult to keep the mechanical stresses within reasonable limits, and the advantage of the shrink-ring construction is that the stresses can be directly calculated and arrangements made to take care of them. The shrink-rings should be of high grade steel of sufficiently heavy section, so that the stresses due to centrifugal action become moderate. They must also be stiff enough to retain their circular form and to prevent any local distortion of the commutator. Practice varies in regard to undercutting the mica segments. When soft graphite brushes are used, undercutting the mica seg- ments is essential for good running, but with hard brushes it is not. Probably the best results are obtained on these high-speed commutators when graphite brushes and undercut mica are used. The undercut grooves should, however, be cleaned out occasion- ally to prevent the accumulation of carbon dust and dirt. Mechanical Construction. The mechanical construction of the armature is of great importance, since it is essential that the balance of the armature should not change after the machine is put in operation. This necessitates that the punchings do not become loose nor move on the shaft, and that the armature wind- ing does not move under the action of centrifugal forces. The punchings are usually either pressed on the shaft one at a time, or built up on a mandril bored out, and shrunk on the shaft, no intermediate spider being used on account of the small diameter. Opinion varies as to whether the armature coils are better held in position by wedges or by wire bands, but the most satisfactory arrangement seems to be the use of bands on the small, and wedges on large armatures. The end connections on the arma- ture are probably better held in position by steel wire bands. Bronze rings have been used for this purpose, but there is danger that they may become loose and change the balance of the armature, as it is very difficult to fix them securely. The ques- tion of insulation of the armature winding is extremely impor- tant, as the armature winding is exposed to carbon and copper dust from the commutator, and the collection of dirt on such a DIRECT CURRENT TURBO GENERATORS 133 high speed armature is very much greater than on the corre- sponding low speed. On account of this it is necessary to be extremely careful in insulating all bare metal on the armature, so that there will be no danger of flashing over dirty surfaces to ground ; while the insulation on the armature coils must be care- fully baked and pressed, so that there will be no shrinkage and consequent movement of the coils. The whole question of satis- factory armature and commutator construction lies in working out the numerous details in design and manufacture, so as to ob- tain an armature and commutator, satisfactory both mechani- cally and electrically at the time it is built, and so thoroughly seasoned before put in operation that it will not change appre- ciably with time. The question of vibration is one of the most serious difficulties to be considered in these machines. It is difficult to predeter- mine the critical speed of a direct current turbo generator armature; but it is very important that this critical speed of the generator, when coupled to the steam turbine, shall not be close to the normal running speed. This usually requires that the armature must be designed with the maximum possible diameter of shaft, and it is generally necessary to sacrifice the advantage of low commutator peripheral speed to enable a sufficiently stiff shaft to be used. The question of permanency of balance is equally important, and this requires that there be no relative movement of the component parts of the armature with time, and also that the shaft neither spring nor deflect under the in- fluence of the temperature variations obtained. Direct current turbo generators as built a few years ago would operate perfectly on test, when first built, but after running six months mechanical vibration and deterioration of commutator were frequently so great that they could no longer be considered commercial. One of the most satisfactory constructions for small units is a two bearing set, the turbine wheel being overhung and the two bearings self-aligning; as this construction obviates any trouble due to lack of alignment. With larger units, however, it is no longer suitable on account of the axial space required by the tur- 134 DIRECT CURRENT TURBO GENERATORS bine, and a three or four bearing set with a coupling, preferably a rigid one, beomes necessary. Such sets again require careful alignment and careful fitting of the coupling and bearings ; other- wise there will be trouble with vibration. Oil-ring lubrication is effective in the smaller sizes, but forced flow lubrication is usually required in capacities above 50 K.W., if the temperature of the bearing is to be kept within reasonable limits and operation to be reliable. Foreign practice is usually to completely enclose the arma- ture, except the commutator, and to supply cooling air from a special duct. This is hardly considered good American practice on account of the difficulty of access, and usually on small machines a semi-enclosed construction with natural cooling is adopted. On large machines, however, as the noise is appre- ciably more than in corresponding low-speed units, it may ultimately be found advisable to adopt a more enclosed construction. Present Situation. At the present time the direct current turbo generator can hardly be considered as commercially suit- able for the American market above 500 K. W. at 250 volt ; and the probability is that in larger sizes it will be necessory, for the present, to use an alternating current turbo generator and rotary converter as a substitute, though this substitute may be only temporary in the 750 K.W. and possibly the 1,000 K.W. sizes. Considerably larger sizes are at present in use in Europe, but it should be remembered that operating conditions there, are by no means as severe as they are here. A typical example of European direct current turbo generator installation which was recently inspected by the writer on a large steamship, exemplifies this latter point. It consisted of four units operating with me- tallic brushes; the normal load being sufficient only to fully load two machines, and the load being changed around from one unit to another. The generators after operating six days in this way were subject to three or four days' overhauling while the boat was in port, which overhauling consisted in replacing the brushes with a newly trimmed set, and in carefully sand-papering the DIRECT CURRENT TURBO GENERATORS 135 commutator and adjusting the brushes. With this attention the units gave very good satisfaction, but it is obvious that such results, and they are to be expected from the use of metallic brushes, make these machines unsuitable for the American market. It is the necessity of developing direct current turbo generators capable of operating with carbon brushes and a minimum amount of attention, that has caused American manu- facturers to delay in placing this type of machine on the market. At the present time such units can be considered commercial in the smaller sizes, while there is the possibility of larger units being developed in the future. V- *\M/ UNIVERSITY OF CALIFORNIA LIBRARY BERKELEY THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW Books not returned on time are subject to a fine of 50c per volume after the third day overdue, increasing to $1.00 per volume after the sixth day. Books not in RETURN TO the circulation desk ot any University of California Library or to the NORTHERN REGIONAL LIBRARY FACILITY Bldg. 400, Richmond Field Station University of California Richmond, CA 94804-4698 ALL BOOKS MAY BE RECALLED AFTER 7 DAYS 2-month loans may be renewed by calling (510)642-6753 1-year loans may be recharged by bringing books to NRLF Renewals and recharges may be made 4 days prior to due date. DUE AS STAMPED BELOW SEfsTTONILL JAN 1 5 1999 U. C. BERKELEY 12,000(11/95) ^343652 UNIVERSITY OF CALIFORNIA LIBRARY