tms *&*&?& m GIFT OF ASSOCIATED ELECTRICAL AND MECHANICAL ENGINEERS MECHANICS DEPARTMENT Engineering Library UNIV CALIF A E & M E ROOM UN IV CALIF A E & M E l - GENERAL LECTURES ON ELECTRICAL ENGINEERING BY CHARLES PROTEUS STEINMETZ, A. M., Ph. D. Consulting Engineer of the General Electric Company, Professor of Electrical Engineering in Union University, Past President, A. I. E. E. Author of "Alternating Current Phenonema," "Elements of Electrical Engineering," "Theory and Calculation of Transient Electric Phenonema and Oscillations." Edited by JOSEPH Le ROY HAYDEN Robson & Adee, Publishers Schenectady, N. Y. Engineering Library Copyright 1908 by Contents First Lecture General Review 7 Second Lecture General Distribution 21 Third Lecture Light and Power Distribution 35 Fourth Lecture Load Factor and Cost of Power. ... 49 Fifth Lecture Long Distance Transmission 61 Sixth Lecture Higher Harmonics of the Generator Wave 77 Seventh Lecture High Frequency Oscillations and Surges 89 Eighth Lecture Generation 99 Ninth Lecture Hunting of Synchronous Machines. . 113 Tenth Lecture Regulation and Control 125 Eleventh Lecture Lightning Protection 135 Twelfth Lecture Electric Railway 147 Thirteenth Lecture Electric Railway Motor Char- acteristics 163 Fourteenth Lecture Alternating Current Railway Motors 175 Fifteenth Lecture Electrochemistry 197 Sixteenth Lecture The Incandescent Lamp 207 Seventeenth Lecture Arc Lighting 215 Appendix I. Light and Illumination 229 Appendix II. Lightning and Lightning Protection. . 259 789562 Preface HE following lectures on Electrical Engineering are general in their nature, dealing with the problems of generation, control, transmission, distribution and utilization of electric energy; that is, with the operation of electric systems and apparatus under normal and abnormal conditions, and with the design of such systems ; but the design of apparatus is discussed only so far as it is necessary to under- stand their operation, and so judge of their proper field of application. Due to the nature of the subject, and the limitations of time and space, the treatment had to be essentially descriptive, and not mathematical. That is, it comprises a discussion of the different methods of application of electric energy, the means and apparatus available, the different methods of carry- ing out the purpose, and the relative advantages and disadvant- ages of the different methods and apparatus, which determine their choice. It must be realized, however, that such a discussion can be general only, and that there are, and always will be, cases in which, in meeting special conditions,, conclusions regarding systems and apparatus may be reached, differing from those which good judgment would dictate under general and average conditions. Thus, for instance, while certain transformer con- nections are unsafe and should in general be avoided, in special cases it may be found that the danger incidental to their use is so remote as to be overbalanced by some advantages which they may offer in the special case, and their use would thus be PREFACE justified in this case. That is, in the application of general con- clusions to special cases, judgment must be exerted to deter- mine, whether, and how far, they may have to be modified. Some such considerations are indicated in the lectures, others must be left to fthe judgment of the engineer. The lectures have been collected and carefully edited by my assistant, Mr. J. L. R. Hayden, and great thanks are due to the publishers, Messrs. Robson & Adee, for the very credit- able and satisfactory form in which they have produced the book. CHARLES P. STEINMETZ. Schenectady, N. Y., Sept. 5, 1908. FIRST LECTURE GENERAL REVIEW N ITS economical application, electric power passes through the successive steps : generation, transmission, conversion, distribution and utilization. The require- ments regarding the character of the electric power imposed by the successive steps, are generally different, frequently contradictory, and the design of an electric system is therefore a compromise. For instance, electric power can for most pur- poses be used only at low voltage, no to 600 volts, while economical transmission requires the use of as high voltage as possible. For many purposes, as electrolytic work, direct current is necessary; for others, as railroading, preferable; while for transmission, alternating current is preferable, due to the great difficulty of generating and converting high voltage direct current. In the design of any of the steps through which electric power passes, the requirements of all the other steps so must be taken into consideration. Of the greatest importance in this respect is the use to which electric power is put, since it is the ultimate purpose for which it is generated and transmitted ; next in importance is the transmis- sion, as the long distance transmission line usually is the most expensive part of the system, and in the transmission the limitation is more severe than in any other step through which the electric power passes. The main uses of electric power are : General Distribution for Lighting and Poiver. The relative proportion between power use and lighting may vary from the distribution system of many small cities, in which *!<*: GENERAL LECTURES practically all the current is used for lighting, to a power distribution for mills and factories, with only a moderate lighting load in the evening. The electric railway. Electrochemistry. For convenience, the subject will be discussed under the subdivisions: 1. General distribution for lighting and power. 2. Long distance transmission. 3. Generation. 4. Control and protection. 5. Electric railway. 6. Electrochemistry. 7. Lighting. CHARACTER OF ELECTRIC POWER. Electric power is used as a. Alternating current and direct current. b. Constant potential and constant current. c. High voltage and low voltage. a. Alternating current is used for transmission, and for general distribution with the exception of the centers of large cities; direct current is usually applied for railroading. For power distribution, both forms of current are used ; in electrochemistry, direct current must be used for electrolytic work, while for electric furnace work alternating current is preferable. The two standard frequencies of alternating current are 60 cycles and 25 cycles. The former is used for general distri- bution for lighting and power, the latter for conversion to direct current, for alternating current railways, and for large powers. GENERAL REVIEW n In England and on the continent, 50 cycles is standard frequency. This frequency still survives in this country in Southern California, where it was introduced before 60 cycles was standard. The frequencies of 125 to 140 cycles, which were standard in the very early days, 20 years ago, have disappeared. The frequency of 40 cycles, which once was introduced as compromise between 60 and 25 cycles is rapidly disappear- ing, as it is somewhat low for general distribution, and higher than desirable for conversion to direct current. It was largely used also for power distribution in mills and factories as the lowest frequency at which arc and incandescent light- ing is still feasible; for the reason that 40 cycle generators driven by slow speed reciprocating engines are more easily operated in parallel, due to the lower number of poles. With the development of the steam turbine as high speed prime mover, the conditions in this respect have been reversed, and 60 cycles is more convenient, giving more poles at the same generator speed, and so less power per pole. Sundry odd frequencies, as 30 cycles, 33 cycles, 66 cycles, which were attempted at some points, especially in the early days, have not spread; and frequencies below 25 cycles, as 15 cycles and 8 cycles, as proposed for railroading, have not proved of sufficient advantage at least not yet so that in general, in the design of an electric system, only the two standard frequencies, 25 and 60 cycles, come into considera- tion. b. Constant current, either alternating or direct, that is, a current of constant amperage, varying in voltage with the load, is mostly used for street lighting by arc lamps; for all other purposes, constant potential is employed. Pi 1 2 GENERAL LECTURES c. For long distance transmission, the highest permis- sible voltage is used; for primary distribution by alternating current, 2200 volts, that is, voltages between 2000 and 2600; for alternating current secondary distribution, and direct current distribution, 220 to 260 volts, and for direct current railroading, 550 to 600 volts. i. GENERAL DISTRIBUTION FOR LIGHTING AND POWER. In general distribution for lighting and power, direct current and 60 cycles alternating current are available. 25 cycles alternating current is not well suited, since it does not permit arc lighting, and for incandescent lighting it is just at the limit , where under some conditions and with some genera- tor waves, flickering shows, while with others it does not show appreciably. Fi&. 1 The distribution voltage is determined by the limitation of the incandescent lamp, as from 104 to 130 volts, or about no volts, no volts is too low to distribute with good regu- lation, that is, with negligible voltage drop, any appreciable amount of power, and so practically always twice that voltage is employed in the distribution, by using a three-wire system, with no volts between outside and neutral, and 220 volts between the outside conductors, as shown diagrammatically in Fig. i. By approximately balancing the load between the two circuits, the current in the neutral conductor is very small, the GENERAL REVIEW 13 drop of voltage so negligible, and the distribution, regarding voltage drop and copper economy, so takes place at 220 volts, while the lamps operate at no volts. Even where a separate transformer feeds a single house, usually a three-wire distribu- tion is preferable, if the number of lamps is not very small. When speaking of a distribution voltage of no, some voltage anywhere in the range from 104 to 130 volts is employed. Exactly no volts is rarely used, but the voltages of distribution systems in this country are distributed over the whole range, so as to secure best economy of the incan- descent lamp. This condition was brought about by the close co-oper- ation, in this country, between the illuminating com- panies and the manufacturers of incandescent lamps. The constants of an incandescent lamp are the candle power for instance 16; the economy for instance 3.1 watts for hori- zontal candle power; and the voltage for instance no. By careful manufacture, a lamp can be made in which the filament reaches 3.1 watts per candle power economy at 16 c. p. within one-half candle-power; but the attempt to fulfill at the same time the condition, that this economy and candle power be reached at no volts, within one-half volt, would lead to a considerable percentage of lamps which would fall outside of the narrow range permitted in the deviation from the three con- stants; and so, if the same distribution voltage were used throughout the country, either a much larger margin of varia- tion would have to be allowed in the product, that is, the lamps would be far less uniform in quality as is the case abroad, or a large number of lamps would not fulfill the requirements, could not be used, and so would increase the cost of the rest. i 4 GENERAL LECTURES Therefore, all the efforts in manufacture are con- centrated on producing the specified candle power at the required economy, and the lamps are then sorted for voltage. This arrangement scatters the lamps over a considerable voltage range, and different voltages are then adopted by different distribution systems, so as to utilize the entire product of manufacture at its maximum economy. The result of this co-operation between lamp manufacturers and users is, that the incandescent lamps are very much closer to requirements, and more uniform, than would be possible otherwise. The effect however is, that the distribution is rarely actually no, and in alternating current systems, the primary distribution voltage not 2200, but some voltage in the range between 2080 and 2600, as in step-down transformers a constant ratio of transformation, of a multiple of 10 -f- i, is always used. In the following, therefore, when speaking of no, 220 or 2200 volts in distribution systems, always one of the voltages within the range of the lamp voltages is understood. In (this country, no volt lamps are used almost exclu- sively, while in England, for instance, the 220 volt lamps is generally used, in a three-wire distribution system with 440 volts between the outside conductors. The amount of copper required in the distribution system, with the same loss of power in the distributing conductors, is inversely proportional to the square of the voltage. That is, at twice the voltage, twice the voltage drop can be allowed for the same distribution efficiency; and as at double voltage the current is one-half, for the same load twice the voltage drop at half the current gives four times the resistance, that is, one-quarter the conductor material. By the change from the 220 volt distribution with no volt lamps, to the 440 volt distribution with 220 volt GENERAL REVIEW 15 lamps, the amount of copper in the distributing conductor, and thereby the cost of investment can be greatly reduced, and current supplied over greater distances, so that from the point of view of the economical supply of current at the customers' terminals, the higher voltage is preferable. However, in the usual sizes, from 50 to 60 watts power consump- tion and so 16 candle power with the carbon filament, and correspondingly higher candle power with the more efficient metallized carbon and metal filaments, the 220 volt lamp is from 10 to 15% less efficient, that is, requires from 10 to 15% more power than the no volt lamp, when producing the same amount of light at the same useful life. This differ- ence is inherent in the incandescent lamp, and is due to the far greater length and smaller section of the 220 volt filament, compared with the no volt filament, and therefore no possibil- ity of overcoming it exists ; if it should be possible to build a 220 volt 1 6 candle power lamp as efficient at the same useful life of 500 hours as the present no volt lamp, this would simply mean, that by the same improvement the efficiency of the 110 volt lamp could also be increased from 10 to 15%, and the difference would remain. For smaller units than 16 candle power, the difference in efficiency is still greater. This loss of efficiency of 10 to 15%, resulting from the use of the 220 volt lamp, is far greater than the saving in power and in cost of investment in the supply mains ; and the 220 volt system with no volt lamps is therefore more efficient, in the amount of light produced in the customer's lamps, than the 440 volt system with 220 volt lamps. In this country, since the early days, the illuminating companies have accepted the responsibility up to the output in light at the customer's lamps, by supplying and renewing the lamps free of charge, and the system using no volt lamps is therefore universally 1 6 GENERAL LECTURES employed while the 220 volt lamp has no right to existence; while abroad, where the supply company considers its responsi- bility ended at the customer's meter, and the customer is left to supply his own lamps, the supply company saves by the use of 440 volt systems at the expense of a waste of power in the customer's 220 volt lamps, far more than the saving effected by the supply company. In considering distribution systems, it therefore is unnecessary to consider any other lamp voltage than no volts (that is, the range of voltage represented thereby). In direct current distribution systems, as used in most large cities, the 220 volt network is fed from a direct current generating station, or as now more frequently is the case from a converter substation, which receives ks power as three-phase alternating, usually 25 cycles, from the main generating station, or long distance transmission line. In alternating current distribution, the 220 volt distribution cir- cuits are fed by step-down transformers from the 2200 volt primary distribution system. In the latter case, where con- siderable motor load has to be considered, some arrangement of polyphase supply is desirable, as the single-phase motor is inferior to the polyphase motor, and so the later is preferable for large and moderate sizes. COMPARISON OF ALTERNATING CURRENT AND DIRECT CURRENT At the low distribution voltage of 220, current can economically be supplfed from a moderate distance only, rarely exceeding from i to 2 miles. In a direct current system, the current must be supplied from a generating station or a converter substation, that is, a station containing revolv- ing machinery. As such a station requires continuous atten- GENERAL REVIEW 17 tion, its operation would hardly be economical if not of a capacity of at least some hundred kilowatts. The direct cur- rent distribution system therefore can be used economically only if a sufficient demand exists, within a radius of i to 2 miles, to load a good sized generator or converter substation. The use of direct current is therefore restricted to those places where a fairly concentrated load exists, as in large cities; while in the suburbs, and in small cities and villages, where the load is too scattered to reach from one low tension supply point, sufficient customers to load a substation, the alternating current must be used, as it requires merely a step- down transformer which needs no attention. In the interior of large cities, the alternating current system is at a disadvantage, because in addition to the voltage consumed by resistance, an additional drop of vokage occurs by self-induction, or by reactance; and with the large conduc- tors required for the distribution of a large low (tension current, the drop of voltage by self-induction is far greater than that by resistance, and the regulation of the system therefore is serious- ly impaired, or at least the voltage regulation becomes far more difficult than with direct current. A second disadvantage of the alternating current for distribution in large cities is, that a considerable part of the motor load is elevator motors, and the alternating current elevator motor is inferior to the direct current motor. Elevator service essentially consists in starting at heavy torque, and rapid acceleration, and in both of these features the direct current motor with compound field winding is superior, and easier to control. Where therefore direct current can be used in low tension distribution, it is preferable to use it, and to relegate alternat- ing current low tension distribution to those cases where direct 1 8 GENERAL LECTURES current cannot be used, that is, where the load is not sufficiently concentrated to economically operate converter substations. The loss of power in the low tension direct current system is merely the i 2 r loss in the conductors, which is zero at no load, and increases with the load; the only constant loss in a direct current distribution system is the loss of power in the potential coils of the integrating wattmeters on the customer's premises. In the direct current system therefore, the efficiency of distribution is highest at light load, and decreases with increasing load. In an alternating current distribution system, with a 2200 volt primary distribution, feeding secondary low tension cir- cuits by step-down transformers, the i*r loss in the conductors usually is far smaller than in the direct current system, but a considerable constant, or "no load", loss exists; the core- loss in the transformers, and the efficiency of an alternating current distribution is usually lowest at light load, but increases with increase of load, since with increasing load the transformer coreloss becomes a lesser and lesser percentage of the total power. The i 2 r loss in alternating current systems must be far lower than in direct current systems: 1. Because it is not the only loss, and the existence of the "no load" or transformer coreloss requires to reduce the load loss or i 2 r loss, if an equally good efficiency is desired. With an alternating current system, each low tension main requires only a step-down transformer, which needs no atten- tion ; therefore many more transformers can be used than rotary converter substations in a direct current system, and the i 2 r loss is then reduced by the greatly reduced distance of second- ary distribution. 2. In the alternating current system, the drop of voltage in the conductors is greater by the self -inductive drop than the GENERAL REVIEW 19 ir drop ; the ir drop is therefore only a part of the total voltage drop ; and with the same voltage drop and therefore the same regulation as a direct current system, the i 2 r loss in the alternat- ing current system would be smaller -than in the direct current system. 3. Due to the self-inductive drop, smaller and therefore more numerous low tension distribution circuits must be used with alternating current than with direct current, and a separ- ate and independent voltage regulation of each low tension cir- cuit that is each transformer, therefore usually becomes im- practicable. This means -that the total voltage drop, resistance and inductance, in the alternating current low tension distribu- tion circuits must be kept within a few percent., that is, within the range permissible by the incandescent lamp. As a result thereof, the voltage regulation of an alternating current low tension distribution is usually inferior to that of the direct cur- rent distribution in many cases to such an extent as to require the use of incandescent lamps of lower efficiency. While there- fore in direct current distribution 3.1 watt lamps are always used, in many alternating current systems 3.5 watt lamps have to be used, as the voltage regulation is not suffiiently good to get a satisfactory life from the 3.1 watt lamps. SECOND LECTURE GENERAL DISTRIBUTION DIRECT CURRENT DISTRIBUTION HE TYPICAL direct current distribution is the system of feeders and mains, as devised by Edison, and since used in all direct current distributions. It is shown diagrammatically in Fig. 2. The conductors are usually under- iii 24 GENERAL LECTURES ground, as direct current systems are used only in large cities. A system of three-wire conductors, called the "mains" is laid down in the streets of the city, shown diagrammatically by the heavily drawn lines. Commonly, conductors of one million circular mill section (that is, a copper section which as solid round conductor would have a dia*neter of i") are used for the outside conductors, the "positive" and the "negative" con- ductor; and a conductor of half this size for the middle or "neutral" conductor. The latter is usually grounded, as pro- tection against fire risk, etc. Conductors of more than one million circular mills are not used, but when the load exceeds the capacity of such conductors, a second main is laid down in the same street. A number of feeders, shown by dotted lines in Fig. 2, radiate from the generating station or converter substations, and tap into the mains at numerous points ; potential wires run back from the mains to the stations, and so allow of measuring, in the station, the voltage at the different points of the distribution system. All the customers are connected to the mains, but none to the feeders. The mains and feeders are arranged so that no appreciable voltage drop takes place in the mains, but all drop of voltage occurs in the feeders ; and as no customers connect to the feeders, the only limit to the voltage drop in the feeders is efficiency of distribution. The voltage at the feeding points into the mains is kept constant by varying the voltage supply to the feeders with the changes of the load on the mains. This is done by having a number of outside bus bars in the station, as shown diagrammaitically in Fig. 3, differing from each other in voltage, and connecting feeders over from bus bar to bus bar, with the change of load. For instance, in a 2 x 120 voltage distribution, the station may have, in addition to the neutral bus bar zero, three positive GENERAL DISTRIBUTION bus bars i, i', i", and three negative bus bars 2, 2', 2", differing respectively from the neutral bus by 120, 130 and 140 volts, as shown in Fig. 3. At light load, when the drop of voltage in the feeders is negligible, the feeders connect to the busses i, o, 2 of 1 20 volts. When .the load increases, some of the feeders are shifted over, by transfer bus bars, to the 130 volt busbars i' and 2'; with still further increase of load, more feeders are connected over to 130 volts; then some feeders are connected to the 140 volt bus bars, i" and 2", and so, by varying z # 2* td Fife. 3 the voltage supply to the feeders, the voltage at the mains can be maintained constant with an accuracy depending on the number of bus bars. It is obvious that a shift of a feeder from one voltage to another does not mean a corresponding voltage change on the main supplied by it, but rather a shift of load between the feeders, and so a readjustment of the total voltage in the territory near the supply point of the feeder. For instance, if by the potential wires a drop of voltage below 120 volts is registered in the main at the connection point of feeder A in Fig. 2, and this feeder then shifted from the supply 26 GENERAL LECTURES voltage 130 to 140, the current in the main near A, which before flowed towards A as minimum voltage point, reverses in direction, flows away from A, the load on feeder A and there- fore increases, and the drop of voltage in A increases, while the load on the adjacent feeders decreases, and thereby their drop of voltage decreases, with the result of bringing up the voltage in the mains at the feeder A and all adjacent feeders. This inter- linkage of feeders therefore allows a regulation of voltage in the mains, far closer than the number of voltages available in the station. The different bus bars in the station are supplied with their voltage by having different generators or converters in the sta- tion operate at different voltages, and with increasing load on the station, and consequent increasing demand of higher volt- age by the feeders ; shift machines from lower to higher voltage bus bars, inversely with decreasing load ; or the different bus bars are operated through boosters, or by connection with the storage battery reserve, etc. In addition to feeders and mains, tie feeders usually con- nect the generating station or substation with adjacent stations, so that during periods of light load, or in case of breakdown, a station may be shut down altogether and supplied from adjacent stations by tie feeders. Such tie feeders also permit most stations to operate without storage battery reserve, that is, to concentrate the storage batteries in a few stations, from which in case of a breakdown of the system, the other stations are supplied over the tie feeders. ALTERNATING CURRENT DISTRIBUTION The system of feeders and mains allows the most perfect voltage regulation in the distributing mains. It is however applicable only to direct current distribution in a territory of GENERAL DISTRIBUTION 27 very concentrated load, as in the interior of a large city, since the independent voltage regulation of each one of numerous feeders is economically permissible only where each feeder represents a large amount of power; with alternating cur- rent systems, the inductive drop forbids the concentration of such large currents in a single conductor. That is, conductors of one million circular mills cannot be used economically in an alternating current system. The resistance of a conductor is inversely proportional to the size or section of the conductor, hence decreases rapidly with increasing current: a conductor of one million circular mills is one-tenth the resistance of a conductor of 100,000 circular mills, and so can carry ten times the direct current with the same voltage drop. The reactance of a conductor, however, and so the voltage consumed by self-induction, de- creases only very little with the increasing size of a conductor, as seen from the table of resistances and reactances of conductors. A wire No. ooo B & S G is ten times the section of a wire No. 7, and therefore one-tenth the resistance ; but the wire No. ooo has a reactance of .109 ohms per 1000 feet, the wire No. 7 has a reactance of .133 oms, or only 1.22 times as large. Hence, while in the wire No. 7, the reactance, at 60 cycles, is only .266 times the resistance and therefore not of serious importance, in a wire No. ooo the reactance is 1.76 times the resistance, and the latter conductor is likely to give a voltage drop far in excess of the ohmic resistance drop. The ratio of reactance and resistance therefore rapidly increases with increasing size of conductor, and for alternating currents, large conductors cannot therefore be used economically where close voltage regulation is required. With alternating currents it therefore is preferable to use several smaller conductors in multiple : two conductors of 28 GENERAL LECTURES No. i in multiple have the same resistance as one conductor of No. ooo; but the reactance of one conductor No. ooo is .109 ohms, and so 1.88 times as great as the reactance of two con- ductors of No. i in multiple, which latter is half that of one conductor No. i, or .058 ohms, provided that the two con- ductors are used as separate circuits. In alternating current low tension distribution, the size of the conductor and so the current per conductor, is limited by the self -inductive drop, and alternating current low tension networks are therefore of necessity of smaller size than those of direct current distribution. As regards economy of distribution, this is not a serious objection, as the alternating current transformer and primary distribution permits the use of numerous secondary circuits. In alternating current systems, a primary distribution system of 2200 volts is used, feeding step-down transformers. The different arrangements are a. A separate transformer for each customer. This is necessary in those cases where the customers are so far apart from each other that they cannot be reached by the same low tension or secondary circuit ; every alternating current system therefore has at least a number of instances where individual transformers are used. This is the most uneconomical arrangement. It requires the use of small transformers, which are necessarily less efficient and more expensive per kilowatt, than large trans- formers. The transformer must be built to carry, within its overload capacity, all the lamps installed by the customer, since all the lamps may be used occasionally. Usually, however, only a small part of the lamps are in use, and those only for a small part of the day ; so that the average load on the transformer is a very small part of its capacity. GENERAL DISTRIBUTION 29 As the coreloss in the transformer continues whether the transformer is loaded or not, but is not paid for by the cus- tomer, the economy of the arrangement is very low ; and so it can be understood that in the early days, where this arrange- ment was generally used, the financial results of most alternat- ing current distributions were very discouraging. Assuming as an instance a connected load of 20 16 candle power lamps low efficiency lamps, of 60 watts per lamp, since the voltage regulation cannot be very perfect allowing then in cases of all lamps being used, an overload of 100%, which is rather beyond safe limits, and permissible only on the assumption that this load will occur very rarely, and for a short time the transformer would have 600 watt rating. Assuming a coreloss of 4%, this gives a continuous power consumption of 24 watts. Usually probably only one or two lamps will be burning, and these only a few hours per day, so that the use of two lamps, at an average summer and winter of three hours per day, would probably be a fair example of many such cases. Two lamps or 120 watts, for three hours per day, give an average power of 15 watts, which is paid for by the customer, while the continuous loss in the transformer is 24 watts ; so that the all year efficiency, or the ratio of the power paid for by the customer, to the power con- 15 sumed by the transformer, is only or 38%. 15 + 24 By connecting several adjacent customers to the same transformer, the conditions immediately become far more favorable. It is extremely improbable that all the customers will burn all their lamps at the same time, the more so, the greater the number of customers is, which are supplied from the same transformer. It therefore becomes unnecessary to 30 GENERAL LECTURES allow a transformer capacity capable of operating all the con- nected load. The larger transformer also has a higher effiicency. Assuming therefore as an instance, four customers of 20 lamps connected load each. The average load would be about 8 lamps. Assuming even one customer burning all 20 lamps, it is not probable that the other customers together would at this time burn more than 10 to 15 lamps, and a trans- former carrying 30 to 35 lamps at overload would probably be sufficient. A 1500 watt transformer would therefore be larger than necessary. At 3% coreloss, this gives a constant loss of 45 watts, while an average load of 8 lamps for 3 hours per day gives a useful output of 60 watts, or an all year efficiency of nearly 60%, while a 1000 watt transformer would give an all year efficiency of 67%. This also illustrates that in smaller transformers a low coreloss is of utmost importance, while the i 2 r loss is of very secondary importance, since it is appreciable only at heavy load, and therefore affects the all year efficiency very little. When it becomes possible to connect a large number of customers to a secondary main fed from one large trans- former the connected load ceases to be of moment in the trans- former capacity ; the transformer capacity is determined by the average load, with a safe margin for overloads; in this case, good all year efficiencies can be reached. Economical alternating current distribution therefore re- quires the use of secondary distribution mains of as large an extent as possible, fed by large transformers. The distance, however, to which a transformer can supply secondary current, is rather limited by the inductive drop of voltage ; therefore, for supplying secondary mains, transformers of larger size than 30 kw. are rarely used, but rather several transformers are em- ployed, to feed in the same main at different points. GENERAL DISTRIBUTION 31 Extending the secondary mains still further by the use of several transformers feeding into the same mains, or, as it may be considered, inter-connecting the secondary mains of the different transformers, we arrive at a system somewhat similar to the direct current system : a low tension distribution system of 220 volts three-wire mains, with a system of feeders tapping into it at a number of points, as shown in Fig. 4. These feeders Fl&. 4. Alternating Current Distribution with Secondary Mains and Primary Feeders. are primary feeders of 2200 volts, connecting to the mains through step-down transformers. In such a system, by vary- ing the voltage impressed upon the primary feeders, a voltage regulation of the system similar to that of direct current dis- tribution becomes feasible. Such an arrangement has these advantages over the direct current system: the drop in the feeders is very much lower, due to their higher voltage; and 32 GENERAL LECTURES that the feeder voltage can be regulated by alternating current feeder regulators or compensators, that is, stationary structures similar to the transformer. It has, however, the disadvantage that, due to the self-induction of the mains, each feeding point can supply current over a far shorter distance than with direct current, and the interchange of current between feeders, by which the load can be shifted and apportioned between the feeders, is far less. As a result, it is difficult to reach as good voltage regu- lation with the same attention to the system; and since this arrangement has the disadvantage -that any break- down in the secondary system or in a transformer may involve the entire system, this system of inter-connected secondary mains is rarely used for alternating current distribution, but the secondary mains are usually kept separate. That is, as shown diagrammadcally in Fig. 5, a number of separate secondary mains are fed by large trans- formers from primary feeders, and usually each primary feeder connects to a number of transformers. Where the distances are considerable, and the voltage drop in the primary feeders appreciable, voltage regulation of the feeders becomes necessary ; and in this case, to get good voltage regulation in the system, attention must be given to the arrangements of the feeders and mains. That is, all the transformers on the same feeder should 1 be at about the same distance from the station, so that the voltage drop between the transformers on the same feeder is negligible; and the nature of the load on the secondary mains fed by the same feeder should be about as nearly the same as feasible, so that all the mains on the same feeder are about equally loaded. It would therefore be undesirable for voltage regulation, to connect, for instance, a main feeding a GENERAL DISTRIBUTION 33 residential section to the same feeder as a main feeding a business district or an office building. ~i i_ Fi. 5. Typical Alternating Current Distribution. In a well designed alternating current distribution system, that is, a system using secondary distribution mains as far as feasible, the all year efficiency is about the same as with the direct current system. In such an alternating current system, 34 GENERAL LECTURES the efficiency at heavy load is higher, and at light load lower, than in the direct current system ; in this respect the alternating current system has the advantage over the direct current system, since at the time of heavy load the power is more valuable than at light load. THIRD LECTURE LIGHT AND POWER DISTRIBUTION N A DIRECT current distribution system, the motor load is connected to the outside mains at 220 volts, and only very small motors, as fan motors, between outside mains and neutral ; since the latter connection, with a large motor, would locally unbalance a system. The effect of a motor on the system depends upon its size and starting current, and with the large mains and feeders, which are gener- ally used, even the starting of large elevator motors has no appreciable effect, and the supply of power to electric elevators represents a very important use of direct current distribution. In alternating current distribution systems, the effect on the voltage regulation, when starting a motor, is far more severe; since alternating current motors in starting usually take a larger current than direct current motors starting with the same torque on the same voltage; and the current of the alternating current motor is lagging, the voltage drop caused by it in the reactance is therefore far greater than would be caused by the same current taken by a non-inductive load, as lamps. Furthermore, alternating current supply mains usually are of far smaller capacity, and therefore more affected in voltage. Large motors are therefore rarely connected to the lighting mains of an alternating current system, but separate transformers and frequently separate feeders are used for the motors, and very large motors commonly built for the primary distribution voltage of 2200, are connected to these mains. For use in an alternating current distribution system, the synchronous motor hardly comes into consideration, since the synchronous type is suitable mainly for large powers, where it is operated on a separate circuit. 38 GENERAL LECTURES The alternating current motor mostly used in small and moderate sizes such as come into consideration for power distribution from a general supply system is the induction motor. The single-phase induction motor, however, is so inferior to the polyphase induction motor, -that single-phase motors are used only in small sizes; for medium and larger sizes the three-phase or two-phase motor is preferred. This however, introduces a complication in the distribution system, and the three-wire single-phase system therefore is less suited for motor supply, but additional conductors have to be added to give a polyphase power supply to the motor. As the result thereof, motors are not used in alternating current systems to the same extent as in direct current systems. In the alternat- ing current system, however, the motor load is, if anything, more important than in the direct current system, to increase the load factor of the system ; since the efficiency of the alter- nating current system decreases with decrease of load, while that of a direct current system increases. Compared with the direct current motor, the polyphase induction motor has the disadvantage of being less flexible: its speed cannot be varied economically, as that of a direct current motor by varying the field excitation. Speed variation of the induction motor produced by a rheostat in the armature or secondary circuit, in the so-called form "M" motor is accomplished by wasting power : the power input of an induc- tion motor always corresponds to full speed; if the speed is reduced by running on the rheostat, the difference in power between that which the motor actually gives, and that which it would give, with the same torque, at full speed, is consumed in the rheostat. Where therefore different motor speeds are required, pro- visions are made in the induction motor to change the number LIGHT AND POWER DISTRIBUTION 39 of poles; thereby a number of different definite speeds are available, at which the motor operates economically as "multi- speed" motor. The starting torque of the polyphase induction motor with starting rheostat in the armature (Form L, motor) is the same as the running torque at the same current input, just as in the case of the direct current shunt motor with constant field excitation. In the squirrel cage induction motor, how- ever, (form K motor) the starting torque is far less than the running torque at the same current input; or inversely, to produce the same starting torque, a greater starting current is required. In starting torque or current, the squirrel cage induction motor has the disadvantage against the direct current motor. It has, however, an enormous advantage over it in its greater simplicity and reliability, due to the absence of commutator and brushes, and the use of a squirrel cage armature. The advantage of simplicity and reliability of the squir- rel cage induction motor sufficiently compensates for the disadvantage of the large starting current, to make the motor most commonly used. In an alternating current distribution system, however, great care has to be taken to avoid the use of such larger motors at places where their heavy lagging starting currents may affect the voltage regulation; in such places, separate transformers and even separate primary feeders are desirable. The single-phase induction motor is not desirable in larger sizes in a distribution system, since its starting current is still larger; in small sizes, however, it is extensively used, since it requires no special conductors, but can be operated from a single-phase lighting main. 40 GENERAL LECTURES The alternating current commutator motor is a single- phase motor which has all the advantages of the different types of direct current motors; it can be built as constant speed motor of the shunt type, or as motor with the charac- teristics of the direct current series motor : very high starting torque with moderate starting current. It has, however, also the disadvantages of the direct current motor: commutator and brushes; and so requires more attention than the squirrel cage induction motor. Alternating current generators now are almost always used as polyphase machines, three-phase or two-phase, and transmission lines are always three-phase, though in transform- ing down, the system can be changed to two-phase. The power supply in an alternating current system therefore is practically always polyphase; and since a motor load, which is very desir- able for economical operation, also requires polyphase currents, alternating current distribution systems always start from poly- phase power. The problem of alternating current distribution therefore is to supply, from a polyphase generating system, single-phase current to the incandescent lamps, and polyphase current to the induction motors. PRIMARY DISTRIBUTION SYSTEMS i. Two conductors of the three-phase generating or transmission system are used to supply a 2200 single-phase system for lighting by step-down transformers and three-wire secondary mains ; the third conductor is carried to those places where motors are used and three-phase motors are operated by separate step-down transformers. In the lighting feeders, the voltage is then controlled by feeder regulators, or, in a smaller system, the generator excitation is varied so as to main- LIGHT AND POWER DISTRIBUTION 41 tain the proper voltage on the lighting phase. At load, the three-phase triangle then more or less unbalances, but induction motors are very little sensitive to unbalancing of the voltage, and by their regulation by taking more current from the phase of higher, less from the phase of lower voltage tend to restore the balance. For smaller motors, frequently two transformers are used, arranged in "open delta" connection. 2. Two-phase generators are used, or in the step-down transformers of a three-phase transmission line, the voltage is changed from three-phase to two-phase; the lighting feeders are distributed between the two phases and controlled by poten- tial regulators so that the distribution for lighting is single- phase, by three-wire secondary mains. For motors, both phases are brought together, and the voltage stepped down for use on two-phase motors. This requires four, or at least three, prim- ary wires to motor loads. 3. From three-phase generators or transmission lines, three separate single-phase systems are operated for lighting; that is the lighting feeders are distributed between the three hases, and all three primary wires are brought to the step-down transformers for motors. This arrangement, by distributing the lighting feeders between the three phases, would require more care in exactly balancing the load between all three phases than two, but a much greater unbalancing can be allowed without affecting the voltage. 4. Four-wire three-phase primary distribution with grounded neutral, and 2200 volts between outside conductors and neutral. The lighting feeders are distributed between * O LIGHT AND POWER DISTRIBUTION 43 2. THREE- WIRE DIRECT CURRENT OR SINGLE-PHASE no- 220 VOLTS. Fig. 7. Neutral one-half size of the -two outside conductors. The two outside conductors require one-quarter the copper of the two wires of a no volt system; since at twice the voltage and one-half the current, four times the resistance or one-quarter //tf/K /TO* JUO* Jo* the copper is sufficient for the same loss (the amount of con- ductor material varying with the square of the voltage). Adding then one-quarter for the neutral of half-size, gives V 4 x V 4 = V 16 or altogether V* + V 16 = Vi of the conductor material required by the two- wire no volt system. That is, the copper economy is 5 / 16 . This is the most commonly used system, since it is very economical, and requires only three conductors. It is, however, a single-phase 44 GENERAL LECTURES system, and therefore not suitable for operating polyphase in- duction motors. Cu. V 3. FOUR- WIRE: QUARTER-PHASE: (TWO-PHASE). Fig. 8. Two separate two-wire single-phase circuits, therefore no saving in copper over two-wire systems. That is, the cop- per economy is : Cu. i. o u t o r i 4. THREE- WIRE QUARTER-PHASE. Fig. 9. Common return of both phases, therefore saves one wire or one-quarter of the copper; hence has the copper economy: Cu. 8 A. / /*2OO volts and more recently, even 22,000 volts is most common for shorter distances, as 10 to 20 miles, since this is about the highest voltage for which generators can be built; its use therefore saves the step-up transformers, that is, the generator feeds directly into the line and to the step- down transformers for the regular load. The next step is 30,000 volts; that is, 33,000 volts at the generator, 30,000 at the receiving end of the line. No inter- mediate voltages between this and the voltage for which generators can be wound is used, as 30,000 volts does not yet offer any insulator troubles ; but line insulators can be built at moderate cost for this voltage, and as step-up transformers 64 GENERAL LECTURES have to be used, it is not worth while to consider any lower voltage than 33,000 volts. This voltage transmits economically up to distances of 50 to 60 miles. 40,000 to 44,000 volts is the next step ; it is used for high power transmission lines of greater distance, where reliability of operation is of importance and the use of a conservative voltage therefore preferable to the attempt at economizing by the use of extra high voltages. A number of 60,000 volt systems are in more or less successful operation, and systems of 80,000 to 110,000 volts are in construction and a few in operation. Where the dis- tances are very great, power valuable, and continuity of ser- vice not of such foremost importance, such voltages are justi- fied in the present state of the art. In such very high voltage systems, the transformers are occasionally wound so that they can be connected for half voltage, for operating the line at half voltage, until the load has sufficiently increased to require full voltage; or the transformers are built for star or Y connection at full voltage, and at first operated in ring or delta connection, i at = 57% of full voltage. V3 The cost of a long distance transmission line depends on the voltage used. The cost of line conductors decreases with the square of the voltage. At twice the voltage, twice the line drop can be allowed with the same loss; at twice the voltage the current is only half for the same power, and twice the drop with half the current gives four times the resistance, that is, one-quarter the conductor section and cost. LONG DISTANCE TRANSMISSION 65 The cost of line insulators increases with increase of voltage. The cost of pole line increases with increase of voltage, since greater distance between the conductors is necessary and so longer poles, longer cross arms, and heavier construction, and not so many circuits can be carried on the same pole line. The lower the voltage, the greater in general is the reli- ability of operation, since a larger margin of safety can be allowed. Since a part of the cost of the transmission line decreases, another part increases with the voltage, a certain voltage will be most economical. Lower voltage increases the cost of the conductor, higher voltage increases the cost of insulators and line construction, and decreases the reliability. The most economical voltage of a transmission line varies with the cost of copper. When copper is very high, higher voltages are more economical than when copper is low. The same applies to aluminum, since the price of aluminum has been varied with that of copper. Aluminum generally is used as stranded conductor. In the early days single wire gave much trouble by flaws in the wire. Aluminum expands more than copper with temperature changes, and so when installing the line in summer, a greater sag must be allowed than with copper, otherwise it stretches so -tight in winter that it may tear apart. Aluminum also is more difficult to join together, since it cannot be welded. For the same conductivity an aluminum line has about twice the size, but one-half of the weight of a copper conductor, and costs 10% less; but copper has a permanent value, while the price of aluminum may sometime drop altogether, as the metal has no intrinsic value, being one of >the most common 66 GENERAL LECTURES constituents of the surface of the earth, and its cost is merely that of its separation or reduction. LOSSES IN LINE DUE TO HIGH VOLTAGE The loss in the line by brush discharge or corona effect is nothing up to a certain voltage, but at a certain voltage it begins and very rapidly increases. The voltage at which a loss by corona effect begins is where the air at the surface of the conductor breaks down, becomes conducting and thus luminous. This occurs at a potential gradient of 100,000 to 120,000 volts per inch. The potential gradient is highest at the surface of the conductor. In Fig. 18 let R = radius of conductor. 2 d = distance between conductor centres. At a point x from the centre O the potential is : C C 2 C X Tf __________ __________ d - - x d + x d 2 x 2 f or : x = d R that is, at the conductor surface, it is : i = e LONG DISTANCE TRANSMISSION 67 Substituting this in the equation, gives : c e = R hence : c = eR therefore the potential at point x is : 2 R X f = - e d 2 x 2 and the potential gradient: d f 2 R(d 2 + x 2 ) d x (d 2 x 2 ) 2 e hence for : x = d R or the conductor surface : g = R If this potential gradient becomes greater than the break- down strength of air, or 100,000 volts per inch, corona effects and energy losses take place: e = 100,000 R gives: e = 100,000 R or E = 100,000 D, as the voltage where the corona begins, and : e E R = - or D = - is the smallest radius 100,000 100,000 which can be used, at voltage E, where D is the conductor diameter = 2 R, and E is the voltage between the conductor = 2 e. For instance, wire No. oooo D = .46" ; corona effects begin at the voltage E = 100,000 D = 46,000. 68 GENERAL LECTURES For 100,000 volts (the smallest diameter for which no corona effects occur is : E D = = i" 100,000 In high potential transformers in the coils no corona effects may occur, because the diameter of the coil or the thick- ness is large enough, but the leads connecting the coils with each other and with the outside, if not chosen very large in diameter, may give corona effects and so break down. In a line or transformer, if one side is grounded, the other side has full voltage against ground, and so may give corona effects and break down ; while if not grounded, both sides have half voltage against ground and so give no corona effect. In the first case, the line or transformer so may break down, although the potential differences between the terminals are no greater than in the second case. For instance, in a 100,000 volt transformer or line, from each terminal to ground are 50,000 volts, and if the conductor diameter is V 2 ", no corona effects occur. If now one terminal is grounded, the other terminal has 100,000 volts to ground and so at Y 2 " diameter gives corona effects, thait is, glow and streamers which may destroy the insulating material or produce high frequency oscillations. At very high voltages it is therefore necessary to have the system statically balanced or symmetrical, that is, have the same potential differences from all the conductors to the ground. Any electric circuit, and so also the transmission line, contains inductance and capacity, and therefore stores energy as electromagnetic energy in the magnetic field due to the cur- rent, and as electrostatic energy, or electrostatic charge, due to the voltage. LONG DISTANCE TRANSMISSION 69 If: e = voltage, C = capacity, i = current, L = inductance, the electrostatic energy is : e 2 C 2 and the electromagnetic energy : i 2 L In a high potential transmission line both energies are of about the same magnitude, and the energy can therefore see- saw between the two forms and thereby produce oscillations and surges resulting in the production of high voltages, which are not liable to occur in circuits in which one of the forms of stored energy is small compared with the other. In distribution systems up to 2200 volts and even some- what higher, the electrostatic energy is still negligible and only the electromagnetic energy appreciable. In static machines the electrostatic energy is appreciable, but the electromagnetic energy negligible. LINES AND TRANSFORMERS At voltages above 2$,ooo step-up and step-down trans- formers are always used, which are therefore a part of the high potential circuit. Three-phase is always used in the transmission line. Some of the available transformer connections are given in Figs. 19 and 20. Grounding the neutral of the system has the advantage of maintaining static balance and so avoiding oscillations and disturbances in case of an accidental static unbalancing, as for 7 GENERAL LECTURES aeir/1-Y instance, the grounding of one line. It has the disadvantage that a ground on one circuit is a short circuit and so shuts down the circuit. LONG DISTANCE TRANSMISSION 71 In connections i, 4 and 6 no neutral is available for grounding and so three separate transformers have to be installed in Y connection for getting the neutral. In connections 2 and 3 the neutral can be brought out from the transformer neutral. In the T connection 5 and 7, the neutral is brought out from a point at one-third of the teaser transformer winding. Assuming the line properly installed and insulated, break- downs may occur, either from mechanical accidents or by high voltages appearing in the line. 7 i GENERAL LECTURES HIGH VOLTAGE DISTURBANCES IN TRANSMISSION LINES These may be: A. Of fundamental frequency, that is, the same frequency as the alternating current machine circuit. B. Some higher harmonic of the generator wave, that is, some odd multiple of the generator frequency. C. Of frequencies entirely independent of the generator, or of a frequency which originates in the circuit, that is, high frequency oscillations as arcing grounds, etc. If a capacity is in series with an inductance, as the line capacity and the line inductance, the capacity reactance and the inductive reactance are opposed to each other ; if they hap- pened to be equal they would neutralize each other, the current would depend on the resistance only and therefore be very large, and with this very large current passing through the inductance and capacity, the voltage at the inductance and at the capacity would be very high. For instance, if we have 20,000 volts supplied to a circuit having a resistance of 10 ohms and a capacity reactance of 1000 ohms, then the total impedance of the circuit is io 2 + iooo 2 = 1000 and the current in the circuit 20,000 = 20 amperes. iooo If now in addition to the io ohms resistance and iooo ohms capacity reactance, the circuit contains iooo ohms inductive reactance, the total reactance of the circuit is iooo iooo = o ohms, and the impedance is the same as e e the resistance, or io ohms. The current therefore = = z r LONG DISTANCE TRANSMISSION 73 2000 amperes, and the voltage at the capacity therefore is: capacity reactance times amperes = 2,000,000 volts, and the same voltage exists at the inductive reactance. These voltages are far beyond destruction. That is, if in a circuit of low resistance and high capacity reactance, a high inductive reactance is put in series with the capacity reactance, excessive voltages are produced. In a .transmission line the capacity of the line consumes for instance 10% of full load current; that is, full load voltage sends only 10% of full load current through the capacity. To send full load current through the capacity so would require 10 times full load voltage. With a line reactance of 20%, 20% or 1 / 5 of full load voltage sends full load current through the inductive reactance, while 10 times full load voltage is required by the capacity reactance; the capacity reactance therefore is about 50 times larger than the inductive reactance at the generator frequency and therefore cannot build up with it to excessive voltages ; but to get resonance with the fundamental frequency requires an inductive reactance about 50 times greater than the line reactance. The only reactance in the system which is large enough to build up with the capacity reactance is the open circuit reactance of the transformers. This is of about the same size as the capacity reactance, since a transformer at open circuit and full voltage takes about 10% of full load current, and the capacity reactance also takes about 10% of full load current. If therefore a high potential coil of a transformer at open secondary circuit is connected in series with a transmission line, destructive voltages may be produced, by the reactance of the transformer building up with the line capacity. In those transformer connections in which several high 74 GENERAL LECTURES potential coils of different transformers are connected between the transmission wires, this may occur if the low tension coil of one of the transformers accidentally opens and the high potential coil of this transformer then acts as inductive react- ance in series with the line capacity in the circuit of the other transformer. 1 This may occur for insitance in transformer connection 2, Fig. 19, if as shown in Fig. 21, the low tension coil c opens. Then the high tension coil C is an inductive reactance in series c /> 1 -L * . 2 ooooooo- 1 p- L O Fig. 28 circuit that is, the short circuit for a part of the half wave of a resistanceless discharge is a severe shock on the system, which must be avoided wherever permissible. This type of lightning arrester takes care of single dis- charges and of multiple discharges, no matter how frequently 144 GENERAL LECTURES they occur or how rapidly they follow each other, with the mini- mum possible shock on the system. It cannot take care, how- ever, of continuous lightning those disturbances, mainly originating in the system, where the voltage remains exces- sive continuously (or rather rises thousands of times per second to excessive values), and for long times. With such a recurring surge, the multi-gap arrester would discharge con- tinuously in protecting the system, until it destroys itself by the excessive power of the continuously succeeding discharges. Where such continuous lightning may occur frequently, as in large high power systems, and the system requires pro- tection against them, a type of lightning arrester which can discharge continuously, at least for a considerable time, with- out self-destruction, is necessary. The only lightning arrester which is capable of doing this, is the electrolytic, or aluminum arrester. In its usual form (cone or disc type) it comprises a series of cone-shaped aluminum cells, connected between line and ground through a spark gap. As soon as the voltage of the system rises above normal, by the value for which the spark gap is set, a discharge takes place through the aluminum cells, over a path of practically no resistance; but the volume of the discharge which passes, is not that given by the voltage on the system, but is merely that due to the excess voltage over the normal, since the normal voltage is held back by the counter e. m. f. of the aluminum cells. As a result with strokes following each other, thousands per second, that is, with a recurrent surge the aluminum arrester discharges continu- ously; but it can stand the continuous discharge for half an hour or more without damage, since it does not carry the short circuit current of the system, but merely the short circuit current of the excess voltage, and so protects the circuit LIGHTNING PROTECTION 145 against continuous lightning for a sufficiently long time, until the cause of the high voltage can be found and eliminated. Even the cone type of aluminum arrester discharges with a slight shock on the system, as the voltage must rise to the value of the spark gap, before the discharge begins, and in systems, in which even a small voltage shock is objectionable, as mainly in large underground cable systems, and also in cases where it is necessary to take care of recurrent surges for an indefinite time, the no-gap aluminum arrester becomes necessary. In principle, this type is the same as the cone type, but the aluminum cells are connected between the conductors and the ground without any spark gap, that is, are continu- ously in circuit, taking a small current. For this reason, the cells are made larger, and of different construction, so as to radiate the heat of the current which, while small, would still give a harmful temperature rise when allowed to accumulate. Being continuously in circuit, a no-gap aluminum arrester allows no sudden voltage rise whatever, however small it may be, that is, it acts just like a flywheel on the engine: while it allows gradual changes of voltages, any sudden change of voltage is anticipated and cut off, just as any sudden change of speed by the flywheel. The no-gap aluminum cell so can hardly be called a lightning arrester, but rather fulfills the duty of a shock absorber, an electrical flywheel on the voltage of the system, and as such finds its proper place on the bus bars of the station or substation, as "surge protector." The three types of apparatus : the no-gap aluminum cell, the aluminum cone arrester, and the multi-gap lightning arrester, then are not different types of apparatus intended for the same purpose, but their operation and proper field of use- fulness is different : the multi-gap arrester protects the system against atmospheric lightning and similar phenomena; the 146 GENERAL LECTURES aluminum cone arrester adds hereto protection against recur- rent surges, where such surges may occur and the system re- quires protection against them, and thus finds its field, but at the same time requiring somewhat more attention than the multi- gap arrester; and the no-gap aluminum cell should be installed as electrical flywheel at the bus bars of the station, and in cable systems, usually in addition to other protection on lines and feeders; it requires, however, occasional attention, and continuously consumes a small amount of power. Of other forms of lightning arresters, the magnetic blow- out 500 volt railway arrester is still in use to a large extent, but is beginning to be superseded by the aluminum cell. The multi-gap, being based on the non-arcing or rectifying prop- erty of the metal cylinders which exists only with alternating current, is not suitable for direct current circuits. In arc light circuits, that is, constant current circuits, horn gap arresters with series resistance are generally used, especially on direct current arc circuits, in which the multi-gap is not permissible. In such circuits of limited current, and very high inductance, the series resistance is not objectionable. Other- wise the horn gap arrester is still occasionally used outdoors as emergency arrester on transmission lines, set for a much higher discharge voltage than the station arrester, and then preferably without series resistance. TWELFTH LECTURE ELECTRIC RAILWAY TRAIN CHARACTERISTICS The performance of a railway consists of acceleration, motion and retardation, that is, starting, running and stopping. The characteristics of the railway motor are: 1. Reliability. 2. Limited available space, which permits less margin in the design, so that the railway motor runs at a higher temp- erature, and has a shorter life, than other electrical apparatus. The rating of a railway motor is therefore entirely determined by its heating. That is, the rating of a railway motor is that output which it can carry without its temperature exceeding the danger limit. The highest possible efficiency is therefore aimed at, not so much for the purpose of saving a few percent, of power, but because the power lost produces heat and so reduces the motor output. 3. Very variable demands in speed. That is, the motor must give a wide range of torque and speed at high efficiency. This excludes from ordinary railway work the shunt motor and the induction motor. The power consumed in acceleration usually is many times greater than when running at constant speed, and where acceleration is very frequent, as in rapid transit service, the efficiency of acceleration is therefore of foremost importance, while in cases of infrequent stops, as in long distance and inter- urban lines, the time of acceleration is so small a part of the total running time, that the power consumed during accelera- tion is a small part of the total power consumption, and high efficiency of acceleration is therefore of less importance. 150 GENERAL LECTURES Typical classes of railway service are : 1. Rapid transit, as elevated and subway roads in large cities. Characteristics are high speeds and frequent stops. 2. City surface lines, that is, the ordinary trolley car in the streets of a city or town. Moderate speeds, frequent stops, and running at vari- able speeds, and frequently even at very low speeds, are char- acteristic. 3. Suburban and interurban lines. That is, lines leading from cities into suburbs and to adjacent cities, through less densely populated districts. Characteristics are less frequent stops, varying speeds, and the ability to run at fairly high speeds as well as low speeds. 4. Long distance and trunk line railroading. Characteristics are: infrequent stops, high speeds, and a speed varying with the load, that is, with the profile of the road. 5. Special classes of service, as mountain roads and ele- vators. Characteristics are fairly constant and usually moderate speed; a constant heavy load, so that the power of accelera- tion is not so much in excess of that of free running; and usually frequent stops. This is the class of work which can well be accomplished by a constant speed motor, as the three- phase induction motor. The rate of acceleration and rate of retardation is limited only by the comfort of the passengers, which in this country permits as high values as 2 to 2 1 / 2 miles per hour per second, that is, during every second of acceleration, the speed increases at the rate of 2 to 2 1 / 2 miles per hour, so that one second after starting a speed of 2 to 2 a / 2 miles per hour, 5 seconds ELECTRIC RAILWAY after starting a speed of 5 x 2 to 2V 2 = 10 to i2*/ 2 miles per hour, etc., is reached. Steam trains give accelerations of 1 / 2 mile per hour per second and less with heavy trains, due to the lesser maximum power of the steam locomotive. SPEED TIME CURVES In rapid transit, and all service where stops are so fre- quent that the power consumed during acceleration is a large part of the total power, the speed time curves are of foremost importance, that is, curves of the car run, plotted with the time as abscissae, and the speed as ordinate. Choose for instance, a maximum acceleration and maxi- mum braking of two miles per hour per second, and assuming a retardation of one-quarter mile per hour per second by fric- tion (that is, assuming that the car slows down one-quarter mile per second, when running light on a level track) ; if then the time of one complete run between two stations is given equal to A B in Fig. 29, the simplest type of run consists of constant acceleration, from A to C, on the line A a, drawn 152 GENERAL LECTURES under a slope of two miles per hour per second; at C the power is shut off and the car coasts on the slope C D, of one- quarter mile per hour per second, until at D, where the coast- ing line cuts the braking line bB, (which also is drawn at the slope of two miles per hour per second), the brakes are applied and the car comes to rest, at B. As the distance traveled is speed times time, the area A C D B so represents >the distance traveled, that is, the distance between the two stations, and all speed time curves of the same type therefore must give the same area. During acceleration, energy is put into the car, and stored by its momentum, which is proportional to the weight of the car and the square of the speed. It is therefore at a maximum at C. A part of the energy represented by the car speed is con- sumed during coasting in overcoming the friction ; the rest is destroyed by the brakes. Assuming, as approximation, con- stant friction, the energy consumed by the car friction on the track, for runs of the same distance, is constant, and the energy destroyed by the brakes is represented by the speed at the point B, where the brakes are applied. The lower therefore this point B is, the less power is destroyed by the brakes, and the more efficient is the run. More accurately, by pro- longing C D to E so that area D E G = B F G, the area A C E F also is the distance between the stations, and E F so would be the speed at which the car arrives at the next station, if no brakes were applied, and the energy correspond- ing thereto has to be destroyed by the brakes ; that is, represents the energy lost during the run, and should be made as small as possible, to secure efficiency. The ratio of the energy used for carrying the car across the distance between the stations that is, energy consumed by track friction, (plus energy consumed in climbing grades, where such exist) to the total energy input, that is, track fric- ELECTRIC RAILWAY tion plus energy consumed in the brakes, is the operation efficiency of the run. As an illustration, a number of such runs, for constant time of the run, of 130 seconds, and constant distance between the stations, that is, constant area of the speed time diagram, are plotted in Figs. 29 to 37. i. Constant acceleration of two miles per hour per second, coasting at one-quarter mile per hour per secondhand braking at two miles per hour per second. Here the energy consumed by the brakes is given by the speed E F = 34. 5 miles per hour, while the maximum speed reached is 60 miles per hour. j. c?c ") i ""^ ~-~, / --i ^ D f. 1 f * ~^ \ \ f **! , -S 1 s I i \ 1 j 1 V f 5 / \ / i \ / i \ I ! - i \ A F B 2,. Acceleration and retardation at two miles per hour per second. Constant speed running between. Fig. 30. Compared with i, (which is shown in 30 in dotted lines), the maximum speed is slightly reduced, e. g., to 51 miles per hour, but the speed of application of the brakes, and therefore the energy lost in the brakes, is increased. That is, running at constant speed, between acceleration and braking, is less efficient than coasting 154 GENERAL LECTURES with decreasing speed. Besides this, at the low power required for constant speed running, the motor efficiency usually is aready lower. It therefore is uneconomical to keep the power on the motors after acceleration, and more economical to con- tinue to accelerate until a sufficient speed is reached to coast until the brakes have to be applied for the next station. Obviously, this is not possible where the distance between the stations is so great, that in coasting the speed would decrease too much to make the time, and so applies only to the case of runs with frequent stops, as rapid -transit. 3. Constant acceleration of one mile per hour per second, braking at two miles, coasting one-quarter mile. Dia- 7 gram i is shown in the same figure 31, for comparison. As seen, with the lower rate of acceleration, the maximum speed is greater, and the lost speed, or speed E F, which is destroyed by the brakes, is greater, that is, the efficiency of the run is lower. 4. Constant acceleration and braking of one mile per hour per second, coasting at one-quarter mile. In this case, ELECTRIC RAILWAY the run between the stations cannot be made in 130 seconds. For comparison, i is shown dotted in Fig. 32. Here the maxi- N \\ mum speed and the lost speed are still greater, that is, the efficiency of the run still lower, and at least 145 seconds are required. That is, the higher the rate of acceleration and of braking, the less is the maximum speed required, and the higher the operation efficiency. With constant acceleration up to the maximum speed, the operation therefore is the more efficient the higher a rate of acceleration and of braking is used. While very rapid acceleration requires more power developed by the motor and put into the car, the time during which the power is developed is so much shorter, that the energy put into the car, or power times time of power applica- tion, is less than with the lower rate of acceleration. The highest operation efficiency, in the case of frequent stops, therefore is produced by constant acceleration at the highest permissible rate, coasting without power, and then braking at the highest permissible rate, as given by i. 156 GENERAL LECTURES During acceleration at constant rate, from A to C, the motor however runs on the rheostat. That is, at all speeds below the maximum, to produce the same pull as at the maxi- mum speed C, the motor consumes the same current and so the same power; while the power which it puts into the train is proportional ito the speed, and therefore is very low at low speeds. Or in other words, the motor during constant acceler- ation, consumes power corresponding to maximum speed, while the useful power corresponds to the average speed, which during A C is only half the maximum; and so only half the available power is put into the car, the other half being wasted in the resistance, and the motor efficiency during constant acceleration therefore must be less than 50%. Constant acceleration up to maximum speed, while giving the best operation efficiency, so gives a very poor motor efficiency and thereby low total efficiency, (the total efficiency being the ratio of the useful energy to the total energy put into the motors, that is, is operation efficiency times motor efficiency). This is the arrangement necessary for a constant speed motor, as the induction motor; but it does not give the best total efficiency, but a better total efficiency is produced by accelerating partly on the motor curve, that is, at a decreasing rate. This sacrifices some operation efficiency, but increases the motor efficiency greatly, and so, if not carried too far, increases the total efficiency. The speed time curves of the motor are shown in Fig. 33, and the current consumption is also plotted in this figure. Acceleration is constant from A to M, on the rheostat, and at constant current consumption, from M, onwards, the accelera- tion decreases, first slightly, then faster, but the current also decreases, first rapidly, and then more slowly; and the ELECTRIC RAILWAY efficiency, plotted in Fig. 33, rises from O at A, to go% at M, and then remains approximately constant, while -the speed still increases. AlLttl 6. This gives the speed time curve of the car, Fig. 34, with acceleration on the motor curve and with maximum values of acceleration and braking 2, the coasting value one-quarter; that is, the same as i, and I is shown in dotted lines in the same figure. The acceleration is constant, on the rheostat, from A to M ; at M the rheostat is cut out, and the acceleration continues on the motor curve, at a gradually decreasing rate, until at C the power is shut off and the car coasts until the brakes are applied. The area A M C D B, representing the distance between the stations, is the same as in i ; the opera- tion efficiency is somewhat lower, but the total current con- sumption, as shown by the curves of current, shown together with the speed time curves, is much less, and the power con- sumption therefore is less ; that is, the total efficiency is higher. i 5 8 GENERAL LECTURES 7. Fig. 35 gives another speed time curve in which, however, the motor is geared for too low a speed ; so the motor curve is reached too early, and the power has to be kept on for too long a time, to make the run in time. As seen from the current curves, here the loss in car efficiency by the decreased fiK \ M ELECTRIC RAILWAY acceleration on the motor curve is greater than the saving in motor efficiency, and the power consumption by the motor is greater than that without running on the motor curve. That is, the total efficiency of operation is increased by doing some of the accelerating on the motor curve, but may be impaired again by carrying this too far. Usually the rheostat is all cut out and the acceleration continues on the motor curve, from about half speed onwards. 8. During the first half of the acceleration on the rheo- stat, 6, when more than half the voltage is consumed in the rheostat, half the current can be saved by connecting two motors in series; that is, by series parallel control on the motors, as shown in Fig. 36. If, however, the series connec- tion of motors is maintained too long, as shown in Fig. 37, so that the part of the curve S P gets -too long, the average rate of acceleration, and so the operation efficiency, is greatly reduced. That is, the lost area becomes so large, that the speed at application of the brakes, and so the power lost in i6o GENERAL LECTURES brakes, is greatly increased. Series connection of motors, for efficient acceleration, therefore should not be maintained for any length of time after the rheostat has been cut out. S 4 In series parallel control, as shown in Figs. 36 and 37, some acceleration occurs on Ihe motor curve in series connec- tion. That is, A S is acceleration on the rheostat, in series connection, S P acceleration on the motor curve ; P M on the rheostat in parallel connection, and M P on the motor curve in parallel connection. Compared with i, which is shown dotted in 9, the area A S P M H d is lost; and so the equal area H C D D , has to be gained, giving a higher speed of application of the brakes D, but gaining power more than the increased power consumption in the brakes, by the higher motor efficiency. CONCLUSION In short distance runs the efficiency is highest in running on series parallel control as much as possible on the motor ELECTRIC RAILWAY 161 curve, with as high a rate of average acceleration and retarda- tion as possible, and coasting between acceleration and retarda- tion ; that is, not keeping the power on longer than necessary. The longer the distance, the less important is high rate of acceleration and retardation, and for long distance running the rate of acceleration and retardation is of little importance. Therefore speed time curves are specially important in rapid transit service, and in general, in running with frequent stops. The heating of the motor at high acceleration, that is, with large current, is less than with low acceleration, that is, smaller current, because the current is on a much shorter time. Feeding back in the line by using the motors as genera- tors is rarely used ; because with an efficient speed time curve, using coasting, the speed when putting on the brakes is already so low that usually not enough power can be saved to compen- sate for the complication and the increased heating of the motors, when carrying current also in stopping. The motors are occasionally used as brakes, operating as generators on the rheostat. This, however, puts an additional heating on the motors ; and is therefore not much used in this country, where the highest speed which the motor equipment can give is desired. With induction motors, feeding back in the line is simplest, because induction motors become generators above synchronism, and so feed back when running down a long hill. Therefore on mountain railways, induction motors have the advantage. In an induction motor there is no running on the motor curve, and so the efficiency of acceleration is lower. Objection to the series motor is the unlimited speed ; that is, when running light, it runs away. In railroading this is no 1 62 GENERAL LECTURES objection, because the motor is never running light and some- body is always in control. In elevator work the series motor is objectionable, due to the unlimited speed ; therefore a limited speed motor is neces- sary. In elevators frequent stops, and so efficient acceleration are necessary; therefore a compound motor is best, that is, a motor having a shunt field to limit the speed and a series field (which is ctit out after starting) to give efficient acceleration. THIRTEENTH LECTURE ELECTRIC RAILWAY: MOTOR CHARACTERISTICS HE economy of operation of a railway system, station, lines, etc., decreases, and the amount of apparatus, line copper, etc., which is required, increases with increas- ing fluctuations of load ; the best economy of an electric system therefore requires as small a power fluctuation as possible. The pull required of the railway motor during accelera- tion, on heavy grades, etc., is, however, many times greater than in free running. In a constant speed motor, as a direct current shunt motor or an alternating current induction motor, the power consumption is approximately proportional to the torque of the motor and thus to the draw bar pull that is given by it. With such motors, the fluctuation of power consump- tion would thus be as great as the fluctuation of pull required. In a varying speed motor, as the series motor, the pull increases with decreasing speed; and the power consumption, which is approximately proportional to pull times speed, varies less than the pull of the motor. The fluctuation of load produced in the circuit by a series motor therefore is far less than that produced by a shunt or induction motor the former economiz- ing power at high pull by a decrease of speed ; the series motor thus gives a more economical utilization of apparatus and lines than the shunt or induction motor, and is therefore almost ex- clusively used. The torque, and so the pull produced by a motor, is approximately proportional to -the field magnetism and the armature current; that is, neglecting the losses in the motor, or assuming 100% efficiency, the torque is proportional to the product of magnetic field strength and armature current. 1 66 GENERAL LECTURES In a shunt motor, at constant supply voltage e, the field exciting current, and thus the field strength, is constant; and the torque, when neglecting losses, is thus proportional to the armature current, as shown by the curve T in Fig. 38. From this torque is subtracted the torque consumed by friction losses, coreloss, etc. (which, at approximately constant speed and field strength, is approximately constant and is shown by the curve TI) thus giving as net torque of the motor, the curve T. Neg- lecting losses, the speed of the motor would be constant, as given by line S 9 ; since at constant field strength, to consume the same supply voltage e , the armature has to revolve at the same speed. As, however, with increasing load and therefore in- creasing current, the voltage available for the rotation of the armature decreases by the ir drop in the armature, as shown by the curve e at constant field strength, the speed decreases in the same proportion, as shown by the curve Si. The field strength, however, does not remain perfectly constant, but with MOTOR CHARACTERISTICS 167 increasing load the field magnetism slightly changes: it de- creases by field distortion and demagnetization, and the speed therefore increases in the same proportion, to the curve S. The current used as abscissae in Fig. 38 is the armature current. The total current consumed by the motor is, however, slightly greater, namely, by the exciting current i ; and, plotted for the total current of the motor as abscissae, all the curves in Fig. 38 are therefore shifted to the right, by the amount of i , as shown in Fig. 39. me. If in the shunt motor, the supply voltage changes, the field strength, which depends upon ithe supply voltage, also changes ; it decreases with a decrease of the supply voltage, and the current required to produce the same torque therefore in- creases in the same proportion. If the magnetic field is below saturation, the field strength decreases in proportion to the de- crease of supply voltage, and the current thus increases in pro- portion to the decrease of supply voltage, while the speed re- 1 68 GENERAL LECTURES mains the same, the armature produces the lower voltage by revolving in the lower field at the same speed. If the magnetic field is highly over-saturated and does not therefore appreciably change with a moderate change of supply voltage and so of field current, the armature current required to produce the same torque also does not appreciably change with a moder- ate drop of supply voltage, but the speed decreases, since the armature must now consume less voltage in the same field strength. Depending on the magnetic saturation of the field : with a decrease of the supply voltage the current consumed by the shunt motor to produce the same torque, therefore increases the more, the lower the saturation, and the speed decreases the more, the higher the saturation. In general, a drop of voltage in the resistance of lines and feeders does not much affect the speed of the shunt motor, but increases the current consumption, thus still further increasing the drop of voltage ; so that in a shunt motor system, lines and feeders must be designed for a lower drop in voltage than is permissible for a series motor. The three-phase induction motor in its characteristics cor- responds to a shunt motor with under-saturated field, except that the effect of a drop of voltage is still more severe ; as not only the amount, but usually the lag of current also increases, thus causing more drop in voltage ; and the maximum torque of the motor is limited, and decreases with the square of the voltage. Hence, while in a series motor system the lines and feeders are designed for the average load or average voltage drop (and practically no limit exists to the permissible maxi- mum voltage drop), with an induction motor, the maximum permissible voltage drop is limited by the danger of stalling the motors. MOTOR CHARACTERISTICS 169 In the series motor, the armature current passes through the field, and with increasing load and thus increasing current, the field strength also increases ; the torque of the motor there- fore increases in a greater proportion than the current. Neg- 1 70 GENERAL LECTURES lecting losses and saturation, the field strength is proportional to the current; the torque being proportional to the current times field strength, therefore is proportional to the square of the current, as shown by the curve T in Fig. 40. The supply voltage, however, has no direct effect on the torque; but with the same current consumption, the motor gives the same torque, regardless of the supply voltage. The speed, at con- stant supply voltage, changes with the field strength and thus with the current : the higher the field strength, the lower is the speed at which the armature consumes the voltage. Since the field strength neglecting losses and saturation is propor- tional to the current, the speed of the series motor would be inversely proportional to the current, as shown by the curve S in Fig. 40. As the voltage available for the armature rotation decreases with increasing current, from e to e, by the ir drop in the field and armature, the speed decreases in the same pro- portion, from the curve S to the curve Si. In reality, however, the field strength, as shown by the curve MO, is proportional to the current only at low currents ; but for higher currents the field strength drops below, by mag- netic saturation, as shown by the curve M ; and ultimately, at very high currents, it becomes nearly constant. In the same ratio as the field strength drops below proportionality with the current, the speed increases and the torque decreases. The actual speed curve is therefore derived from the curve Si by in- creasing the values of the curve Si in the proportion, M to M, and is given by the curve S; and in the same proportion the torque is decreased to the curve TI. From this torque curve the lost torque is now subtracted ; that is, the torque represent- ing the power consumed in friction and gear losses, hysteresis and eddy currents, etc. Some of the losses of power are MOTOR CHARACTERISTICS 171 approximately constant ; others are approximately proportional to the square of the current; and the lost -torque, being equal to the power loss divided by the speed, can therefore be assumed as approximately constant : somewhat higher at low and high speeds, as shown by curve F. The net torque then is given by the curve T. As seen, it is approximately a straight line, pass- ing through a point I , which is the "running light current," and its corresponding speed, the "free running speed" of the motor. At this current i , the speed is highest ; with increase of current it drops first very rapidly, and then more slowly; and the higher the saturation of the motor field is, the slower becomes the drop of speed at high currents. The single-phase alternating current motors are either directly or inductively series motors, and so give the same general characteristics as the direct current series motor. In the alternating current motors, however, in addition to the ir drop an ix drop exists ; -that is, in addition to the voltage con- sumed by the resistance, still further voltage is consumed by self-induction; and the voltage e available for the armature rotation thus drops still further, as seen in Fig. 41. Since the self-induction consumes voltage in quadrature with the cur- rent, the inductive drop is not proportional to the current, but is small at low currents, and greater at high currents ; e therefore is not a straight line, but curves downwards at higher currents. The speed, Si, is dropped still further by the inductive drop of voltage, to the curve Si, and then raised to the curve S by saturation. The effect of saturation in the alternating current motor usually is far less, since the magnetic field is alternating, and good power factor requires a low field excitation, and therefore high saturation cannot well be reached. The torque curves are the same as in the direct current motor, except that the effect of saturation is less marked. T 7 2 GENERAL LECTURES In efficiency, the shunt or induction motor, and the series motor are about equal ; and both give high values of efficiency over a wide range of current. A wide range of current, how- ever, represents a wide range of speed in the series motor, and MOTOR CHARACTERISTICS 173 nearly constant speed in the shunt motor; therefore while the series motor can operate at high efficiency over a wide range of speed, the shunt motor shows high efficiency only at its proper speed. In regard to the effect of a change of supply voltage, as is caused, for instance, by a drop of voltage in feeders and mains, /the series motor reacts on a change of voltage by a correspond- ing change of speed, but without change of current ; while the shunt motor and induction motor reacts on a change of supply voltage by a change of current, with little or no change of speed. As the limitation of a system usually is the current, at excessive overloads on the system, resulting in heavy voltage drop, the series motors run slower, but continue to move ; while the induction motor is liable to be stalled. FOURTEENTH LECTURE ALTERNATING CURRENT RAILWAY MOTOR. N a direct current motor, whether a shunt or a series motor, the motor still revolves in the same direction, if the impressed e. m. f. be reversed, as field and arma- ture both reverse. Since a reversal of voltage does not change the operation of the motor, such a direct current motor there- fore can operate also on alternating current. With an alter- nating voltage supply, the field magnetism of the motor also alternates ; the motor field must therefore be laminated, to avoid excessive energy losses and heating by eddy currents (cur- rents produced in the field iron by the alternation of the mag- netism) just as in the direct current motor the armature must be laminated. In the shunt motor in which the supply current divides between field and armature when built for alternating voltage, arrangements must be made to have the current in the field (or rather the field magnetism) and the current in the armature, reverse simultaneously. In the series motor, in which the same current traverses field and armature, the field magnetism and the armature current are necessarily in phase with each other, or nearly so. Only the series or varying speed type of alternating current commutator motor has so far become of industrial importance. In the alternating current motor in addition to the voltage consumed by the resistance of the motor circuit and that con- sumed by the armature rotation, voltage is also consumed by self-induction; that is, by the alternation of the magnetism. The voltage consumed by the resistance represents loss of power, and heating, and is made as small as possible in any 178 GENERAL LECTURES motor. The voltage consumed by the rotation of the arma- ture, or "e. m. f. of rotation," is that doing (the useful work of the motor, and so is an energy voltage, or voltage in phase with the current; just as the voltage consumed by the resist- ance is in phase with the current. The voltage consumed by self-induction, due to the alternation of the magnetism, or "e. m. f. of alternation", is in quadrature with the current, or wattless; that is, it consumes no power, but causes the current to lag, and so lowers the power factor of the motor; that is, causes the motor to take more volt-amperes than corresponds to its output, and so is objectionable. The useful voltage, or e. m. f. of rotation of the motor, is proportional to the speed ; or rather the "frequency of rota- tion", NO, is proportional to the field strength F, and to the number of armature turns m. The wattless voltage, or self- induction of the field, is proportional to the frequency N, to the field strength F, and the number of field turns n. The ratio of the useful voltage to the wattless voltage therefore is mN -T- nN, and to make the useful voltage high and the wattless voltage low, therefore requires as high a frequency of rotation N and as low a frequency of supply N, as possible. Thus the commutator motors of more than 25 cycles give poor power factors; and for a given number of revolutions NO, which is number of revolutions per second times number of pairs of poles, therefore is the higher, the more poles the motor has. Hence a greater number of poles are generally used in an alternating current -than in a direct current motor. Good direct current motor design requires a strong field and weak armature, to get little field distortion and therefore good commutation ; that is high n and low m. But such pro- portions, even at low supply frequency N and high frequency of rotation N , would give a hopelessly bad power factor, and ALTERNATING CURRENT MOTOR 179 thus a commercially impractical motor. In the alternating cur- rent commutator motor, it is therefore essential to use as strong an armature and as weak a field (that is, as large a number of armature turns m and as low a number of field turns n) as pos- sible. Very soon, however, a limit is reached in this direction, even if the greater field distortion and the resultant bad com- mutation were not to be considered : the armature also has a self-induction ; ; that is, .the alternating magnetism produced by the current in the armature turns consumes a wattless e. m. f. This magnetism is small in a direct current motor, but with many armature turns and few field turns it becomes quite con- siderable ; and so, while a further decrease of the field turns and increase of the armature turns reduces the self-induction of the field which varies with the square of the field turns it increases the self-induction of the armature which varies with the square of the armature turns. There is thus a best proportion between armature turns and field turns, which gives ithe lowest total self-induction. This is about in this propor- tion : armature turns m to field turns n = 2 -=- i ; and at this proportion the power factor of the motor, especially at low and moderate speeds, is still very poor. In alternating current commutator motors it is therefore essential to apply means to neutralize the armature self-induc- tion and armature reaction, so as to be able to increase the proportion of armature turns to field turns sufficiently to get good power factors. This is done by surrounding the arma- ture with a stationary "compensating winding" closely adja- cent to the armature conductors, located in the field pole faces, and traversed by a current opposite in direction to the current in the armature, and of the same number of ampere turns ; so that the armature ampere turns and the ampere turns of the compensating winding neutralize each other, and the armature i8o GENERAL LECTURES reaction, that is, the magnetic flux produced by the armature current, and the self-induction caused by it, disappear. This compensating winding for neutralizing the armature self-induction was introduced by R. Eickemeyer in the early days of the alternating current commutator motor, and since then all alternating current commutator motors have it ; so that the electric circuits of all alternating current commutator motors comprise an armature winding A, a field winding F, and a compensating winding C. Since the compensating winding cannot be identically at the same place as the armature winding (the one being located in slots in the pole faces, the other in slots in the armature face) there still exists a small magnetic flux produced by the armature winding : the "leakage flux", analogous to the leakage flux of the induction motor ; and the number of armature turns cannot be increased indefinitely, otherwise the armature self- induction, due to this leakage flux, would become appreciable, and the power factor would decrease again. The minimum total self-induction of the motor with compensating winding occurs at a number of armature turns equal to 3 to 5 times the field turns ; at this proportion, the power factor is already very good at low speeds, and the motor is industrially satisfactory in this regard. For best results, that is, complete compensation and there- fore zero magnetic field of armature reaction, it is, however, necessary not only to have the same number of ampere turns in the compensating winding as on the armature, but also to have these ampere turns distributed in the same manner around the circumference. With the usual armature winding this is not the case, but the armature conductors cover the whole circum- ference; while the compensating coil conductors cover only the pole arc, as the space between the poles is taken up by the CM ; 1 82 GENERAL LECTURES field winding. That is, the magnetic distribution around the armature circumference is as shown developed in Fig. 42: the field gives a flat topped distribution, the armature a peaked, and the compensating winding has a small flat top and with the total ampere turns of the compensating winding equal to those of the armature, the compensating winding preponder- ates in front of the field poles, the armature between the field poles, or at the brushes, and there is thus a small magnetic field of armature reaction remaining at the brushes, just where it is objectionable for commutation. As it is not feasible to distribute the compensating wind- ing over the whole circumference of the stator, the armature winding is arranged so that its ampere turns cover only the pole arcs. This is done by using fractional pitch in the arma- ture; that is, the spread of the armature coil or the space between its two conductors, is made, not equal to the pitch of the pole, as shown in Fig. 43, but only to the pitch of the pole arcs, as shown in Fig. 44. With such fractional pitch winding, the currents in the upper and the lower layer of the armature conductors, in the space between the poles, flow in opposite dir- ections, and so neutralize, leaving only that par/t of the armature winding in front of the pole arcs as magnetizing. Hereby the distribution of the armature ampere turns is made the same as that of the compensating winding, and so complete compensa- tion is realized. The compensating winding may be energized by the main current, and so connected in series with the field and armature ; or the compensating winding may be short circuited upon itself, and so energized by an induced current acting as a secondary of a transformer to the armature as primary ; and as in a transformer, primary and secondary current have the same number of ampere turns (practically) and flow in opposite ALTERNATING CURRENT MOTOR 183 directions, such "inductive compensation" is just as complete compensation as the "conductive compensation" produced by passing the main current through the compensating winding. Fig. 43 Vice versa, the armature may be short circuited and so used as secondary of a transformer, with the compensating winding acting as primary. In either of these motor types, Fig. 44 which comprise primary and secondary circuits, that is, in which armature and compensating winding are not connected directly in series, but inductively, the field may be energized 184 GENERAL LECTURES by the primary or supply current, or by the secondary or induced current. In such a motor embodying a transformer feature, instead of impressing the supply voltage upon one circuit as primary, while the other is closed upon itself as secondary, the supply voltage may be divided in any propor- tion between primary and secondary. As primary and secondary current of a transformer are proportional to each other, it is immaterial, regarding the varia- tion of the current in the different circuits with the load and speed, whether the circuits are directly in series, or by trans- formation; that is, all these motors have the same speed torque current characteristics, as discussed in the preceding lecture, and differ only in secondary effects, mainly regarding commutation. The use of the transformer feature also permits, without change of supply voltage, to get the effect of a changed supply voltage, or a changed number of field turns, by shifting a cir- cuit over from primary to secondary or vice versa. For in- stance, if the armature is wound with half as many turns, that is, for half the voltage and twice the current, as the compen- sating winding, by changing the fitdd from series connection with the compensating winding to series connection with the armature, the current in the field and thus the field strength, is doubled; that is, the same effect is produced as would be by doubling (the number of field turns. According to the relative connection of the three circuits, armature A, compensating circuit C, and field F, alternating current commutator motors of the series type can be divided into the classes shown diagramatically in Fig. 45 : 10 : ^^-^ 1 86 GENERAL LECTURES Primary : Secondary : A + C + F Conductively Compensated Series Motor. (2). A + F C Inductively Compensating Series Motor. (3). A C + F Inductively Compensating Series Motor with Second- ary Excitation, or In- verted Repulsion Motor. (4). C + F A Repulsion Motor. (5). C A + F Repulsion Motor with Secondary Excitation (6). C & A + F Series Repulsion Motor A. C + F & A Series Repulsion Motor B. (8). The main difference between these types of motors is found in .their commutation. In a direct current motor, with the brushes set at the neutral; that is, midway between the field poles (as is custom- ary in a reversible motor like the series motor), the armature turn, which is shorted circuited under the brush during the commutation, encloses all the lines of magnetic force of the field; so during this moment it does not cut any lines of force by its rotation, and thus no e. m. f . is induced in this turn ; that is, no current is produced, if the armature reaction is com- pensated for, or is otherwise negligible. If the motor has a considerable armature reaction, and thus a magnetic field at the brushes, this magnetic field of armature reaction induces an e. m. f. in the short circuited turn under the brush, and so ALTERNATING CURRENT MOTOR 187 causes sparking. Hence high armature reaction impairs the commutation of the motor. In an alternating current series motor the armature reac- tion is neutralized by the compensating winding, and therefore no magnetic field of armature reaction exists ; hence no e. m. f . is induced in the turn short circuited under the brush by its rotation through the magnetic field. As this field, however, is alternating, an e. m. f. is induced in the short circuited (turn by the alternations of the lines of magnetic force enclosed by it, and causes a short circuit current and in that way, sparking. This e. m. f., being due to the alternation of the enclosed field flux, is independent of the speed of rotation ; it also exists with the motor at a standstill, and is a maximum in the armature turn under the brush, as this encloses the total field flux. The position of the armature turn during commutation, which in a direct current motor is the position of zero induced e. m. f., is therefore in an alternating current motor, the position of maximum induced e. m. f., but induced not by the rotation of the turn, but by the alternation of the magnetism. That is, this turn is in the position of a short circuited secondary to the field coil of the motor as primary of a transformer ; and as primary and secondary ampere turns in a transformer are approximately equal, the current in -the armature turn during commutation is very large; if not limited by the resistance or reactance of the coil, it is as many times greater than the full load current, as the field coil has turns. This causes serious sparking, if not taken care of. One way of mitigating the effect of this short circuit cur- rent is to reduce it by interposing resistance or reactance ; that is by making the leads between the armature turns and the commutator bars of high resistance or high reactance. Obvi- ously this arrangement can merely somewhat reduce the spark- 1 88 GENERAL LECTURES ing by reducing the current in the short circuited coil, but can not eliminate it ; and it has the disadvantage, that in the moment of starting, if the motor does not start at once, the resistance lead is liable to be burned out by excessive heating ; while when running, each lead is in circuit only a very small part of the time: during the moment when the armature turn to which it connects, is under a commutator brush. As the resistance of the lead must be very much greater than that of the arma- ture coil, and as the space available for it is very much smaller, if remaining in circuit for any length of time, it is destroyed by heat. In direct current motors, commutation may be controlled by an interpole or commutating pole; that is, by producing a magnetic field at the brush, in direction opposite to the field of armature reaction, and by this field inducing in the arma- ture turn during commutation, an e. m. f. of rotation which reverses the current. Such a commutating pole, connected in series into a circuit, would, in the alternating current motor, induce an e. m. f. in the short circuited turn, by its rotation; but this e. m. f . would be in phase with the field of the commu- tating pole, and thus with the current, that is, with the main field of the motor. Therefore it could not neutralize the e. m. f. induced in the short circuited turn by the alternation of the main field through it, since this latter e. m. f. is in quadrature with the main field, and thus with the current; but would simply add itself to it, and so make the sparking worse. A series commutating pole, while effective in a direct current motor, is therefore ineffective in an alternating current motor, due to its wrong phase. To neutralize the e. m. f. induced by the alternation of the main field through the armature turn during commutation, by an opposite e. m. f. induced in this turn by its rotation ALTERNATING CURRENT MOTOR 189 through a quadrature field or commutating field, this field must therefore have the proper phase. The e. m. f . of alternation of the main field through the short circuited turn is proportional to the main field F and frequency N, and is in quadrature with the main field. The e. m. f. induced in the short circuited turn by its rotation (through the commutating field is proportional to the frequency of rotation or speed N , and to the commutating field F , and in phase therewith ; to be in opposition and equal to the e. m. f . of alternation, the commutating field ; must there- fore be in quadrature with the main field, and frequency times main field must equal speed times commutating field. That is : N F = No Fo or in other words, the commutating field must be : N F = F No or equal to the main field times the ratio of frequency to speed, and in quadrature therewith. Hence, at synchronism: N = N, the commutating field must be equal to the main field ; at half synchronism : I NO = N, it must be twice ; at double synchronism : 2 NO = 2 N, it must be one-half the main field. The problem of controlling the commutation of the alter- nating current motor therefore requires the production of a commutating field of proper strength, in quadrature phase with the main field of the motor, and thus with the current. In a transformer, on non-inductive or nearly non-induc- tive secondary load, the magnetism is approximately in quad- rature behind the primary, and ahead of the secondary current ; transformation between compensating winding and arma- 1 90 GENERAL LECTURES ture thus offers a means of producing a quadrature field in the alternating current motor for compensation. In the conductively compensated series motor, at perfect compensation, no quadrature field exists; while with over or under compensation, a quadrature field exists, in phase with the current, and therefore not effective as commutating field. In the inductively compensated series motor, the quad- rature field, which transforms current from the armature to the compensating winding, is of negligible intensity, as the compensating winding is short circuited, and thus consumes very little voltage. A quadrature field, however, appears in those motors in which the compensating winding is primary, and the armature secondary, that is in repulsion motors ; since in the armature the induced or transformer e. m. f. is opposed by the e. m. f. of rotation; so a considerable e. m. f. is induced, and therefore a considerable transformer flux exists. Therefore, when impressing the supply voltage on the compensating winding, and short circuiting the armature upon itself, that is, in the repulsion motor, the voltage is supplied to the armature by transformation from the compensating wind- ing, and the magnetic flux of this transformer is in quadrature with the supply current ; that is, it has the proper phase as com- mutating flux. The repulsion motor thus has in addition to the main field, in phase with the current, a transformer field, in quadrature with the main field in space and in time, and so in the proper direction and phase as commutating field ; thus giving perfect commutation if this transformer field has the intensity required for commutation, as discussed above. As in the repulsion motor, the armature is short circuited upon itself, the voltage supplied to it by transformation from ALTERNATING CURRENT MOTOR 191 the compensating winding equals the voltage consumed in it by the rotation through the main field. The former voltage is proportional to the frequency N and to the transformer field F 1 , the latter to the speed N and to the main field F, and it so is : N F 1 = No F, that is, the transformer field is : No F 1 = _F N or equal to the main field times the ratio of speed to frequency. Comparing this value of the transformer field of the repul- sion motor, F 1 , with the required commutating field F , it is seen that at synchronism N = N, F 1 F ; that is, the trans- former field of the repulsion motor has the proper value as commutating field, so that no short circuit current is produced in the armature turn under the brush, but the commutation is as good as in a direct current motor with negligible armature reaction. i At half synchronism, N = N, the transformer field of 2 I the repulsion motor : F 1 = F, is only one quarter as large as 2 the commutating field required F = 2 F, and the short cir- cuit current is reduced by 25% below the value which it has in the series motor ; and the commutation, while it is better, is not yet perfect At double synchronism : N = 2 N, the transformer field is F 1 = 2 F, while the commutation field should be: i F = F, and the transformer field thus is four times larger 2 than it should be for commutation ; so that only one-quarter of i 9 2 GENERAL LECTURES the transformer field is used to neutralize the e. m. f. of alter- nation in the short circuited turn; the other three-quarters induces an e. m. f., thus causing a short circuit current three times as large as it would be in a series motor. That is, the short circuit current under the brush, and thus the sparking, in the repulsion motor at double synchronism is very much worse than in the series motor, and the repulsion motor at these high speeds is practically inoperative. Hence, as regards commutation, a repulsion motor is equal to the series motor at standstill where no compensation of (the short circuit current is possible but becomes better with increasing speed : as good as a good direct current series motor at synchronism; and then again becomes worse by over com- pensation, until at some speed, at 40% above synchronism, it again becomes as poor as the alternating current series motor ; above this speed, it becomes rapidly inferior -to the series motor. To produce right intensity of the transformer field, to act as commutating field, it is therefore necessary above syn- chronism to reduce the transformer field below the value which it would have when transforming the total supply voltage from compensating winding to armature. This means, that above synchronism, only a part of the supply voltage must be trans- formed from! compensating winding to armature, the rest directly impressed upon the armature. Thus at double syn- chronism, where the transformer field of the repulsion motor is four times as strong as is required for commutation, to re- duce it to one-quarter, only one-quarter of the supply voltage must be impressed upon the compensating winding, three- quarters directly on the armature. To get zero short circuit current in the armature turn under the brush, below synchronism more than the full supply ALTERNATING CURRENT MOTOR 193 voltage would have to be impressed upon the compensating winding, which usually cannot conveniently be done. At syn- chronism the full supply voltage is impressed upon the com- pensating winding, while the armature is short circuited as repulsion motor ; and with increasing speed above synchronism, more and more of the supply voltage is shifted over from com- pensating winding to armature; that is, the voltage impressed upon the compensating winding is reduced, from full voltage at synchronism, while the voltage impressed upon the armature is increased, from zero at synchronism, to about three-quarters of the supply voltage at double synchronism. Such a motor, in which the transformer field is varied in accordance with the requirement of commutation, is called a "series repulsion motor." The arrangement described here eliminates the short cir- cuit current induced in the commutated armature turn by the alternation of the main field, and that completely above syn- chronism, so that during commutation, no current is induced in the armature turn. This, however, is not sufficient for per- fect commutation: during the passage of the armature turn under the brush, the current in the turn should reverse ; so that in the moment in which the turn leaves the brush, the current has already reversed. For sparkless commutation, it therefore is necessary, in addition to the neutralizing e. m. f . of the trans- former field, to induce an e. m. f. which reverses the current. This e. m. f., and thus the magnetic flux which induces it by the rotation, must be in phase with the current. That is, in addi- tion to the "neutralizing" component of the commutating field (which is in quadrature with the current), to reverse the cur- rent, a second component of the commutating field must exist, in phase with the current; this component so may be called the "reversing field". The total commutating field required i94 GENERAL LECTURES to eliminate the short circuit current due to the alternating main field by the "neutralizing" flux, and to reverse the arma- ture current by the "reversing flux", must therefore be some- what less than 90 lagging behind the main field and thus the main current. While in a transformer with non-inductive load on the secondary, the magnetic flux lags nearly 90 behind the primary current, in a transformer with inductive load on (the secondary, the magnetic flux lags less than 90 behind the primary current; and the more so the higher the inductivity of the secondary load. Therefore, by putting a reactance into the armature circuit of the motor, and so making the armature circuit inductive, the transformer flux is made to lag less than 90 behind the current, and act not only as neutralizing but also as reversing flux; and so, if it be of proper intensity, it gives perfect commu- tation. An additional reactance would in general be objectional, in lowering the power factor of the motor. The motor, how- ever, contains a reactance: its field circuit, which has to be excited, can be used as reactance for the armature circuit. That is, by connecting the field coils into the armature cir- cuit, or in other words, using secondary excitation, the trans- former flux of .the motor is given the lead ahead of quad- rature position with the main field, which is required to act as reversing field. In this manner, it is possible in the alternating current commutator motor, to get at all speeds from synchronism upwards, the same perfect commutation as in a direct current motor with commutating poles, by varying the distribution of supply voltage between compensating winding and armature, and exciting the field in series with the armature circuit ; that 196 GENERAL LECTURES is, in the series repulsion motor B of the preceding table. Obviously, this distribution of voltage would for all practical purposes be carried out sufficiently by using a number of steps, as shown diagrammatically by the arrangement in Fig. 46 : T is the supply circuit, F the field winding, A the arma- ture, and C the compensating winding. Closing switch i, and leaving all others open, the motor is a repulsion motor. Closing switch 2, and leaving I and all others open, the motor is a repulsion motor with secondary excitation. Closing switch 3, or 4, 5 . . . and leaving all others open, the motor is a series repulsion motor B, with gradually increasing armature voltage and decreasing voltage on the compensating winding. By winding the armature for half the voltage and twice the current of the compensating winding, when changing from position i, the field in the compensating circuit, to the next position, with the field in the armature circuit, (the field current and the field strength becomes double the value it had in start- ing, where no compensation exists, and which it would have to maintain in a series motor; and thus a correspondingly greater motor output is secured, than would be possible in a motor in which the commutation is not controlled. FIFTEENTH LECTURE ELECTROCHEMISTRY LECTROCHEMISTRY is one of the most important applications of electric power, and possibly even more power is used for electrochemical work than for rail- roading. In electrochemical industries the most expensive part is electric power; material and labor are usually much less. Such industries therefore are located at water powers, where the cost of power is very low. The main classes of electrochemical work are : A. Electrolytic. B. Electrometallurgical. A. ELECTROLYTIC WORK. The chemical action of the current is used, by electrolyz- ing either solutions of salts or fused salts or compounds. Electrolysis of solutions in water is possible only with such metals which have less chemical affinity than hydrogen. For instance, Cu, Fe, and Zn can be deposited from salt solu- tions in water, but not Al, Mg, Na, etc. Electrolyzing, for instance, NaCl (salt solution) the sodium (Na) which appears at the negative terminal immediately dissociates the water and gives Na + H 2 O = NaOH + H, or: sodium plus water = caustic soda plus hydrogen. It takes 1.4 volts to electrolyze water; any metal requiring more than 1.4 volts for separation therefore is not separated, but hydrogen is produced. Therefore the highest voltage used is an electrolytic cell containing water is 1.4 + the ir drop in the resistance of the 200 GENERAL LECTURES cell; which latter, for reasons of economy, is made as low as possible. Even fused salts require fairly low voltage, at the highest from 3 to 4 volts. Since the voltage required per cell is very low, a large number of cells are connected in series, and even then large low voltage machines are required. Some of the important applications of electrolysis are : Electroplating; that is, covering with copper, nickel, silver, gold, etc. Electro typing; that is, making of copies, usually of cop- per; and especially Metal refining. A very large part of all the copper used is electrically refined. The crude copper as cast plate is used as anode or positive, and a thin plate of refined copper is used as cathode, or negative (terminal in a copper sulphate solution. The anode is dissolved by the current and the fine copper is deposited on the cathode ; while silver and gold go down into the mud, lead goes into the mud as sulphate, tin as oxide; sulphur, selenium and tellurium, arsenic and other impurities also go in the mud ; and zinc and iron remain in solution as sulphates if the current density is sufficiently low. If the current density is high, some zinc and iron may deposit : zinc and iron have a greater chemi- cal activity than copper, since they precipitate copper from solution. Therefore it takes more power, that is, more voltage, to deposit zinc and iron, than it takes to deposit copper. If the current density is low, the voltage required to deposit the copper plus the ir drop, that is, the total voltage of the cell, is less than the voltage required to deposit zinc or iron, and they do not deposit, but dissolve at the anode and remain in solution. ELECTROCHEMISTRY 201 At higher current density the ir drop in the cell is higher ; thus the total voltage of the cell is higher, and may become high enough to deposit iron or even zinc. If the anode is crude copper, the cathode pure copper, the voltage at the anode is higher than at the cathode and the cell takes some voltage. The voltage required for copper refining is the higher, the more impure the copper is; but is always very low, usually a fraction of a volt, and therefore very many cells are run in series. The solution gradually becomes impure and has to be replaced. Other metals are occasionally refined electrolytically, but only to a small extent. Metal Reduction. Metals are reduced from their ores electrolytically, especially such metals which have so high chemical affinity that they are not reduced by heating with carbon. In this way aluminum, magnesium, sodium, calcium, etc., are made electro- lytically. Since their chemical affinity is greater than that of hydrogen, they cannot be deposited from solutions in water, but only from fused salts, or solutions in fused salts. So cal- cium is produced now by electrolyzing fused calcium chloride, CaCl 2 . Aluminum is made by electrolyzing a solution of alumina in melted cryolithe (sodium aluminum fluoride). SECONDARY PRODUCTS. Frequently electrolysis is used to produce not the sub- stances which are directly deposited, but substances produced by the reaction of these deposits on the solutions. For instance, electrolyzing a solution of salt, NaCl, in water, we get sodium, Na, at the negative, chlorine, Cl, at the positive terminal. 202 GENERAL LECTURES If we use mercury, Hg, as negative electrode, it dissolves the sodium and so we get sodium amalgam. Otherwise the sodium dees not deposit but immediately acts upon the water and forms sodium hydrate or caustic soda, NaOH. The chlorine, Cl, at the anode also reacts on the water, one chlorine atom taking up one hydrogen and another chlorine atom the remaining OH of the water H 2 O; that is, we get 2C1 + H 2 O = C1H + C1OH, that is, hydrochloric + hypo- chlorous acid. With the sodium hydrate from the other cathode these acids form NaCl and ClONa, that is sodium chloride and hypochlorite, or bleaching soda. If the solution is hot, the reaction goes further and we get 6C1 + 3H 2 O = 5C1H +C1O 3 H, that is hydrochloric and chloric acid, and with the sodium hydrate from the other side (these form NaCl and ClO 3 Na, that is, sodium chloride and sodium chlorate. In this way considerable industries have developed, pro- ducing electrolytically caustic soda, bleaching soda, and chlorates. Alternating current is used very little for electrolytic work, as with organic compounds to produce oxidation and reduction at the same time; that is, act on the compound in rapid succession by oxygen and hydrogen, the one during the one, the other during the next half wave of current. Very active metals like manganese and silicon dissolve by alternating current; that is, one-half wave dissolves, but the other does not deposit again. Very inert metals like platinum are deposited by alternat- ing current; that is, the negative half wave deposits by alter- nating current, but the positive half wave does not dissolve. ELECTROCHEMISTRY 203 B. ELECTROMETALLURGICAL WORK. In electrometallurgical work the heat is used to produce the chemical action; thus it is immaterial whether alternating or direct current is used. The voltage required is still low but not as low as in elec- trolytic work : The carborundum furnace takes from 250 to 90, mostly about loo volts; that is, it starts cold with 250 volts. While heating up the resistance drops, and the voltage decreases down to 100 volts when the furnace is hot and remains there until towards the end. Then the inner layer of carborundum begins to change to graphite and the resistance, and therefore the voltage falls. The carbide furnace and arc furnaces in general take from 50 to 100 volts ; the graphite furnace takes from 10 to 20 volts. To get very high temperatures a very large amount of energy has to be concentrated in one furnace; and with the moderate voltage used, this requires very large currents, thousands of amperes. Alternating currents are almost exclu- sively used, since it is easier to produce very large alternating currents by transformers, and since it is easier to control alter- nating than direct currents. Electric heat necessarily is very much more expensive than heat produced by burning coal, and so the electric furnace is used mainly: ist. Where very perfect control of the temperatures and freedom from impurities is essential. 2nd. Where temperatures higher than can be produced by combustion are required. i. Very accurate temperature regulation and freedom from impurities, for instance, are important in making and 204 GENERAL LECTURES annealing high grade tool steels, etc. By using coal or oil as fuel, contamination by the gases of combustion, and by the metal taking up carbon or (if an excess of air is used, oxygen) is difficult to avoid. By electric heating, by resistance at lower temperature and by induction furnace at higher temperature, contamination can be perfectly avoided and even the air can be excluded. 2. The temperature of combustion is limited. Four-fifths of air is nitrogen which does not take part in the combustion, but which has to be heated, thus greatly lower- ing the temperature; therefore combustion in air, even if the air is preheated, gives a lower temperature than when using oxygen. But even the temperature of the oxy-hydrogen, or the oxy-acetylene flame is only just able to melt platinum. The temperature which can be reached by combustion, is limited, since at very high temperature the chemical affinity of oxygen for hydrogen and carbon ceases : water dissociates, that is, spontaneously splits up in hydrogen and oxygen at 2000 degrees Centigrade and no temperature higher than 2000 can therefore be reached by the oxy-hydrogen flame; carbon dioxide, CO 2 , already dissociates at about I5OOC into carbon monoxide, CO, and oxygen, O. Carbon monoxide, CO, splits up into carbon and oxygen not much above 2000 C. (In all high temperature reactions of carbon, as in the formation of carbides, CO therefore always forms and not CO 2 , since CO 2 cannot exist at a very high temperature ; and 1he CO when leaving the furnace then burns to CO 2 with blue flame) . Higher temperatures than those generated by the com- bustion of carbon and hydrogen can be produced by the com- bustion of those elements whose oxides are stable at very high temperatures , as aluminum and calcium. In this way, many metals, as chromium and manganese, which cannot be reduced ELECTROCHEMISTRY 205 from the oxides by carbon (due to the lower temperature of carbon combustion) can be reduced by aluminum in the "ther- mite" process. That is, their oxides are mixed with powdered aluminum and then ignited : the aluminum burns in taking up the oxygen of the metal, and so produces an extremely high temperature, which melts the metal and the alumina (corun- dum) which is produced. Since, however, all the aluminum is made electrolytically, the thermite process still requires the use of electric power. The temperature of combustion of aluminum, however, is still far below that of the electric carbon arc, since in the car- bon arc, alumina boils. For temperatures above 2000 to 25ooC, and up to the arc temperature or about 35OOC, electric energy is therefore necessary. Electric furnaces are of two classes : Arc Furnaces and Resistance Furnaces. In the resistance furnace any temperature can be produced up to the point of destruction of the resistance material, that is, up to 35OOC, when using carbon. The arc furnace gives the arc temperature of 35ooC, but allows the concentration of much more energy in a small space and thus produces reactions requiring the very highest temper- atures. Some of the electrometallurgical industries are : (a). Calcium carbide production. Arc furnaces are used and the reaction is CaO + 3C = CaC 2 + CO. A mixture of coke and quick lime is used in the process. (b). Carborundum production. A resistance furnace is used, containing a carbon core about 24 feet long, around which the material is placed and heated by the current passing 2 o6 GENERAL LECTURES through the core. The furnace takes 1000 HP and the reac- tion is : SiO 2 + 3C = SiC + 2CO. The material is a mixture of sand, coke, sawdust and salt. (c). Graphite furnace. A resistance furnace somewhat similar to the carborundum furnace is used, but with lower voltage and larger currents ; the material is coke or anthracite, which by the high temperature is converted into graphite, probably passing through an intermediate stage as a metal car- bide. (d). Silicon furnace. Either arc or resistance furnace is used ; the reaction is : Si0 2 + 2C = Si + 2CO. or, Si0 2 + 2SiC = 3 Si + 2CO. (e). Titanium carbide furnace. Arc or resistance fur- nace is used which requires a very high temperature; that is, a greater temperature than that of the calcium carbide furnace. Ti0 2 + 3 C = TiC + 2CO. Other products of the electric furnace are siloxicon, sili- con monoxide, etc., and numerous alloys of refractory metals, mainly with iron; as of vanadium, tungsten, molybdenum, titanium, etc., which are used in steel manufacture. The use of the electric arc for the production of nitric acid and nitrate fertilizers ; of the high potential glow discharge for the production of ozone for water purification, etc., also are applications of electric power, which are of rapidly increas- ing industrial importance. SIXTEENTH LECTURE THE INCANDESCENT LAMP HE two main types of electric illuminants are the in- candescent lamp and the arc. In the incandescent lamp the current flows through a solid conductor, usually in a vacuum, and (the heat produced in the resistance of the conductor makes it incandescent, thus giving the light. Incandescent lamps in an electric circuit therefore act as non-inductive ohmic resistance and can there- fore be operated equally well on constant potential as on con- stant current. As electric distribution systems are always constant potential, most incandescent lamps are operated on constant potential ; and only for outdoor lighting, that is, for street lighting in cases where the arc lamp is too large and too expensive a unit of light for the requirements, incandescent lamps are used on a constant, direct or alternating current cir- cuit ; they are then usually built for the standard arc circuits, and thus for low voltage. For general convenience the efficiency of incandescent lamps is given in watts power consumption per horizontal candle power, when operating on such a voltage, that the candle power of the lamp decreases by 20% in 500 hours run- ning; and the time, in which the candle power decreases by 20% that is, 500 hours with the present efficiency rating is called the useful life ; since experience has shown, that after a decrease of candle power of 20%, with the carbon filament lamp, under average conditions, it is more economical to replace the lamp with a new lamp, than to continue its use ; as then the increased cost of light due to 'the lower efficiency is greater than the cost of the lamp, when distributed over 500 hours. 210 GENERAL LECTURES In discussing incandescent lamp efficiencies, it is therefore essential to make sure that the efficiency is given at the useful life of 500 hours; since obviously any efficiency can be pro- duced in any lamp, by running it at higher voltage, but the life is greatly shortened thereby. Therefore efficiency compari- sons have a meaning only when based on the same length of useful life, as 500 hours. Obviously, for other types of lamps, the economic life may be greater (as for more expensive lamps) or less than 500 hours. Illummants are measured and compared by the total flux of light which they give. Usually, however, this is expressed in "mean spherical candle power"; that is, the candle power which would be given by the illuminant if this light were dis- tributed uniformly throughout. Since the object of a lamp is to give light, obviously the only logical way of measuring it is by the total amount of light which it gives, and so by the mean spherical candle power ; this therefore is standard. The conventional rating of the incandescent lamp, in hori- zontal candle power, therefore has to be multiplied by a reduc- tion factor, to give the mean spherical candle power. With the carbon filament lamp, this reduction factor is usually .79; a 1 6 candle power so has a mean spherical candle power of 16 x .79 = 12.6 c. p., and at an efficiency of 3.1 watts per 3-i horizontal candle power, it has an efficiency of == 3.92 79 watts per mean spherical candle power. The carbonized bamboo fibre used in the very early days was very soon replaced by filaments made of structureless cellulose, squirted from a cellulose solution, and then carbon- ized. By "treating" these filaments, that is, heating them in THE INCANDESCENT LAMP 211 gasolene vapor and therefrom depositing a thin shell of car- bon on them, a considerable increase in efficiency became pos- sible; their efficiency was thus greatly increased, from 5 to 6 watts per candle power in the early days, to 3.5 and 3.1 watts per candle power. Of these two types, the 3.5 watt lamp is used in systems of poor voltage regulation, in which (the more efficient 3.1 watt lamp would have too short a life; with the improvements in the voltage regulation of systems, the less efficient 3.5 watt lamp is thus coming out of use. By exposing these "treated" filaments to the highest temperature of the electric furnace, their stability at high temperature is greatly improved ; so that in these "metallized"* filament lamps an efficiency of 2.5 to 2.6 watts per candle power is reached. Whether a still further increase of efficiency of the carbon filament will occur, as is quite possible, or whether the carbon filament will be replaced by the metal fila- ments, remains for the future to decide. In the last years, metal filament lamps giving efficiencies far higher than has so far been possible to reach with the car- bon filament, have been developed. First came the os- mium lamp, of 1.5 watts per candle power. As the total supply of osmium available on (the earth is far less than would be required for one year's production of incandescent lamps, the osmium lamp never could hope for more than a very limited use. The tantalum lamp, which was developed next, and is now quite extensively used, gives an efficiency of about 2 watts per candle power; that is, it is not quite as efficient as -the osmium lamp, since tantalum is somewhat more fusible than osmium. As tantalum is a metal which can be drawn into wire, the tantalum filament is of drawn wire ; while * The name "metallized" is given to the form of carbon produced in these filaments by the elec- tric furnace temperature, since it has metallic resistance characteristics-: a positive temperature coefficient of resistance, while the other forms of carbon have a negative temperature coefficient GENERAL LECTURES all the other metals which are used for lamp filaments are not ductile, and the filaments have to be made by some squirting process, similar to the carbon filament. The highest efficiency was reached by the tungsten (wolfram) lamp, of i to iV 4 watts per candle power; that is, tungsten (or rather wolfram metal, since tungsten is the name of the ore of the metal), has the highest melting point of all known metals, and so can be run at the highest temperature, that is, highest efficiency. All these metals melt far below the temperature where carbon melts or boils, but carbon has the great disadvantage of evaporating considerably below its melting point, while these metals evaporate very little, and so can be run at a temperature fairly close to their melting point; while the carbon filament has to be operated at a temperature very far below the melting point. The great difficulty with all these metal filaments is, that the metals are very much better conductors than carbon; to get the same filament resistance, so as to consume the same current, at the same voltage, the metal filaments must be very much longer and very much thinner than the carbon filament. As the efficiency of the metal filament is far higher, *to produce the same candle power at the same voltage, less current and therefore a higher resistance is required, which makes these metal filaments still thinner ; as a result, although the metals are mechanically stronger than carbon, the metal filaments are far more frail, due to their exceeding thinness, and it is very diffi- cult to produce lamps of as low candle power, as is feasible with carbon filaments. For larger units, however, and for larger current low voltage lamps, for series lighting, the metal fila- ments are specially suited. For general use, the 16 candle power lamp has proved the moat convenient unit of light. The limitation of voltage, for THE INCANDESCENT LAMP 213 which efficient incandescent lamps of such size can be built, has been the cause of the general use of 1 10 volt distribution. 220 volt 1 6 candle power carbon filament lamps can be built, but are of necessity less efficient, by about 15%, than no volt lamps : at 220 volts, half ithe current and so four times the resistance is required for the same power as at no volts; the filament therefore is about twice as long and half as thick, hence more breakable and more rapidly disintegrating ; so that there is no possibility of reaching the same efficiency in a 220 volt 1 6 candle power lamp, as in no volt lamp made with the same care. For the same reason, the 8 candle power no volt lamp must be less efficient than the 16 candle power no volt lamp. In an incandescent lamp are specified : the candle power, the efficiency, and the voltage. To produce lamps fulfilling simultaneously all three conditions, requires either to allow a large margin in either condition that is, gives a product inferior in uniformity or to get a uniform product, a large percentage is thrown out as defective, and the cost of the lamp is thus seriously increased. For this reason, in the manufacture a very close agreement is aimed at in candle power and in efficiency; the lamps are then assorted for voltages, and dif- ferent voltages are then assigned by the organization of illum- inating companies to the different companies, so as to tonsume the total lamp product. As a result hereof, a far more uniform product is derived than could be derived in any other way, and than is available in any other country. This is the reason, that in distribution systems not one and the same voltage, as no, is employed throughout ; but different cities use different voltages, between 105 and 130. The average incandescent lamp used in this country therefore is decidedly superior in uniformity and in efficiency to those used abroad. The ultimate cause hereof 2i 4 GENERAL LECTURES is, that since the earliest days the illuminating companies have followed the principle of supplying light, and not power ;* and 220 volt distribution, while being more efficient from the gen- erating station to the customer's meter, is decidedly inferior in efficiency from the generating station to the candle power pro- duced at the customer's lamps, as the saving in distribution losses does not make up for the lower efficiency of the 220 volt lamp. For this reason, 220 volt distribution has never found any entrance in this country. In gas lighting, an enormous increase of efficiency resulted from the development of the Welsbach gas mantle. In the same direction, that is, by using what may be called "heat luminescence" in electric lighting, the Nernst lamp was developed, using the same class of material : refractory metal- lic oxides, as in the Welsbach mantle. The "glower" of the Nernst lamp, however, is a non-conductor at ordinary tempera- ture, and requires some heating device, the "heater", to be made conducting. When conducting, it has a very high nega- tive temperature coefficient; that is, the voltage consumed by the glower decreases with the increase of current, just as in the arc, and it therefore requires a steadying resistance, called the "ballast". The lamp therefore requires some operating mechanism, to cut the heater out of circuit after the glower is started. The glower of the Nernst lamp is not operative in a vacuum, since air seems to be necessary for its heat lumines- cence. Fairly good efficiencies have been reached with these lamps, especially in larger units, as 3 to 6 glower lamps, but not of the same class as with the tungsten lamp. * Fora long time, the bills were even made out in 'Mamp hoars." and in the earlier dayi the machines rated in "lights" and not in kilowatts. SEVENTEENTH LECTURE ARC LIGHTING HILE incandescent lamps can be operated on constant potential as well as on constant current, the arc is essentially a constant current phenomenon. At con- stant length, the voltage consumed by the arc decreases with increase of current, as shown by curve I in Fig. 47. If, there- fore, an attempt is made to operate such an arc on constant potential, for instance on 80 volts which would correspond to 3.9 amperes on curve I then any tendency of the current to increase as by a momentary drop of the arc resistance would lower the required arc voltage, and so increase the cur- rent, at constant supply voltage, hence still further lower the arc voltage, etc., and a short circuit would result. Vice versa, a momentary decrease of arc current, by requiring more volt- age than is available, would still further decrease the current, increase the required voltage, etc., and the arc would extin- guish. Therefore only such apparatus are operative on constant potential, in which an increase of current requires an increase of voltage, and vice versa; and so limits itself. While therefore arcs can be operated on a constant cur- rent system, to run arc lamps on constant potential, some cur- rent limiting device is necessary in series with the arc, as a resistance; or, in an alternating current circuit, a reactance. The voltage consumed by the resistance is proportional to the current, and a resistance of 8 ohms inserted in series to the arc would thus consume the voltage shown in straight line II in Fig. 47. The voltage consumed by the arc plus .the resistance then is given by the curve III, derived by adding I and II. As 218 GENERAL LECTURES seen, below 3.35 amperes, the total required voltage still decreases with increase of current, and the arc is still unstable ; that is, the resistance is insufficient. Above this current, an increase of current requires an increase of voltage and so limits itself; that is, the arc is stable; with 8 ohms series resist- ance, 3.35 amperes therefore is the limit of stability of the arc; and attempting to operate it at lower current, as for instance at 2 amperes and 106 volts supply, the arc either goes out, or ARC LIGHTING 219 the current runs up to 5.5 amperes, where the arc becomes stable on 106 volts supply. With a higher series resistance, the arc remains stable to lower currents, and vice versa. It follows herefrom, that for the operation of an arc lamp on constant potential, a higher voltage is required than that consumed by the arc proper. At every value of series resistance therefore a point a in Fig. 47, is reached, at which for decreasing current the arc becomes unstable; and all these points, for different resistance values, give a curve IV, which is called the "stability curve" of the arc curve I. The supply voltage required to operate the arc represented by curve I must therefore be higher (than that given by the stability curve IV. For instance, at 4 amperes, the arc cannot be operated at less than 104 volts supply. At 104 volts supply the limit of stability is reached ; that is, a change of current does not require a change of voltage, but the arc voltage decreases as much as the resistance voltage increases and the current thus drifts; and for supply voltages higher than 104, the arc is stable, the more so, the higher the supply voltage is above 104. The difference in voltage between the supply voltage and the arc voltage thus is consumed by the "steadying resistance" of the arc. High reactance in series with the direct current arc retards the current fluctuations and so reduces them; so .that with reactance in series to the direct current arc, the arc can be operated by a supply voltage closer to the stability curve IV than without reactance ; reactance therefore is very essential in the steadying resistance of a direct current arc. Obviously, no series reactance can enable operation of the arc I on a supply voltage below that given by the stability curve IV. 220 GENERAL LECTURES The arc characteristic I is far steeper for low currents than for high currents, and is ithe steeper the greater the arc length. Low current arcs and long arcs therefore require, that on a constant potential supply, a greater part of the supply voltage is consumed by the steadying resistance (or steadying reactance with alternating arcs) than high current arcs, or short arcs; and are therefore less economical on constant potential supply. Constant potential arc lamps are necessarily less efficient than constant current arc lamps, due to the power con- sumed in the steadying resistance. A large part of this power is saved in alternating constant potential arc lamps, by using reactance instead of resistance, but the power factor is there- fore greatly lowered ; that is, the constant potential alternating arc lamp rarely has a power factor of over 70%. Where therefore high potential constant current circuits are permissible, as for outdoor or street lighting, arc lamps are usually operated on a constant current circuit, with series connection of from 50 to 100 lamps on one circuit. With the exception of a few of the larger cities, all the street lighting by arc lamps in this country is done by constant current systems, either direct current or alternating current. For direct current constant current supply, separate arc light machines have been built, and are still largely used. In these machines, inherent regulation for constant current is produced by using a very high armature reaction and relatively weak field excitation; that is, the armature ampere turns are nearly equal and opposite to the field ampere turns, and thus both very large compared with the difference, the resultant ampere turns, which produce the magnetic field. A moderate increase of current and consequent increase of armature ampere turns therefore greatly reduces the resultant ampere turns and ARC LIGHTING 221 so the field magnetism and the voltage, tthat is, the machine tends to regulate for constant current. Perfect constant current regulation then is secured by some governing device, as an auto- matic regulator varying a resistance shunted across the series field. It must, however, be understood .that the "regulator" of the arc machine does not give a constant current regula- tion, but the armature reaction of the machine does this, and the regulator merely makes it perfect; but even with the regu- lator disconnected, arc machines give fairly close constant cur- rent regulation. As the voltages produced by arc machines are very high 4,000 to 10,000 commutation of the current, with the ordinary commutator, which is limited to a maximum of 40 to 50 volts per segment -is not well suited, but rectification is used. The Brush arc machine therefore is a quarter-phase alternator with rectifying commutator. That is, the commuta- tor shifts the connection over from the phase of falling e. m. f. to that of rising e. m. f., and thereby is able to control as high as 3,000 volts per commutator ring. With the development of the mercury arc rectifier, which converts constant alternating current into constant direct cur- rent, arc machines are rapidly going out of use. The arc machine necessarily must be a small unit, since 100 to 150 lamps in series give already as high a voltage as is safe to use in arc circuits, but do not yet represent much power; and when supplying thousands of arc lamps a large number of small machine units are required, which are uneconomical in space, in attendance and in efficiency. The mercury arc rectifier in combination with the stationary constant current transformer enables us to derive the power from the alternating current con- stant potential supply system. 222 GENERAL LECTURES Constant alternating current is derived by a constant cur- rent transformer or constant current reactance. Diagram- matically, the constant current transformer is shown in Fig. 48. The primary coil P and the secondary coil S are movable with regard to each other (which of the two coils is movable, is immaterial, or rather, is determined by consideration of design). Fig. 48 shows the coil S suspended and its weight partially balanced by counter-weight W. With the secondary coil S close to the coil P, that is, in the lowest position, most of the magnetism produced by the primary coil P passes' through the secondary coil S, and the secondary voltage therefore is a maximum. The further the secondary coil moves away from the primary coil, the more of the magnetism passes between the coils, the less through ,the secondary coil, and the lower therefore is the secondary voltage, ARC LIGHTING 223 which becomes a minimum (or zero, if so desired), with the secondary coil at a maximum distance from the primary, that is, in the top position. Primary current and secondary current are proportional and in opposition to each other, and repel each other, and the repulsion is proportional to the product of the two currents; that is, proportional to the square of the secondary current. The weight of the secondary coil is balanced by the counter- weight W and -the repulsion from the primary coil, at normal secondary current. Any increase of secondary current by a decrease of load, increases the repulsion, in this way pushing the secondary coil further away from the primary and thereby reducing the secondary voltage and thus the current; and vice versa, a decrease of secondary current, by an increase of load, reduces the repulsion and so causes the secondary coil tc come nearer to the primary, that is, increases its voltage and so restores the current. Such an arrangement regulates for con- stant current between the voltage limits given by the two ex- treme positions of the movable coil. These usually are chosen from some margin above full load, down to about one-third load. The constant current reactance operates on the same principle : the two coils P and S are connected in series with each other into the arc circuit supplied from the constant potential source, and by separating or coming together, vary in reactance with the load, and thereby maintain constant current. While the alternating current arc lamp is less efficient, that is, gives less light for the same power, than the direct cur- rent arc lamp, the disadvantages of the use of numerous arc machines have led -to the extended adoption of alternating cur- rent series arc lighting before the development of the mercury 224 GENERAL LECTURES arc rectifier, which enabled the operation of direct current arc circuits from constant current transformers. While incandescent lamps give the same efficiency for all sizes except such small sizes where mechanical difficulties appear in the filament production, the efficiency of the arc decreases greatly with decrease of current; that is, the arc is at the greatest efficiency only for large units of light, but rather inefficient and not so well suited for small units of light. Even in large units, the efficiency of light production of the direct current carbon arc lamp is not superior to that of the tungsten incandescent lamp ; that of the alternating current carbon arc lamp is inferior to the tungsten lamp ; and the carbon arc lamp thus finds its field mainly where large units of light are required, especially as long as the cost of renewal of the metal filament lamps is still very great. Entirely different, however, are ithe conditions developed in the last years, with the luminous arcs, as the flame carbon arc, the mercury lamp, and the magnetite and titanium carbide arc. In these, efficiencies of light production have been reached which no incandescent lamp can hope .to approach. In the carbon arc, practically all the light comes from the incandescent tips of the carbons, very little from the arc flame. Then by using materials, which in the arc flame give an intense- ly luminous spectrum, the efficiency of die arc lamp has been vastly improved. j So far only three materials have been found, which in luminous arcs give efficiences vastly superior to incandescence : mercury, calcium (lime), and titanium. All three even in moderate sized units, give efficiencies of one-half watt or better per candle power. The mercury arc has the advantage of perfect steadiness, a long life requiring no attention for thousands of hours ARC LIGHTING 225 and high efficiency over a fairly wide range of candle powers ; but it is seriously handicapped for many purposes by its bluish- green color. In the flame carbon lamp carbons impregnated with cal- cium compounds, usually calcium fluoride (fluorspar) are used, and the arc then has an orange-yellow color. Other com- pounds which give red or white color to the arc are so much inferior in efficiency that they are used only to a very limited extent. The compounds, after coloring the arc and giving it efficiency, escape as smoke; the arc therefore must be an open arc. This, however, means short life of the carbons and fre- quent trimming. The open arc lamp, which was used formerly, has, how- ever, been almost entirely superseded by the enclosed carbon arc, in spite of the somewhat lower efficiency of the latter ; and the inconvenience of daily attendance required by an open arc, and the large consumption of carbons, makes a return to this type improbable. For this reason the flame carbon lamp has not proven suitable for general outdoor illumination, as street lighting, where the cost of carbons and trimming would usually far more than offset the gain in efficiency. Flame carbon lamps, however, have found a field for decora- tive lighting, for advertising purposes, etc., for which the glare of light and its color makes them very suitable. They are generally used on constant potential circuits with two or three lamps in series. To eliminate the objections of short life and consequent frequent trimming and high cost of carbons, and thereby make the luminous arc able to enter the field of general outdoor il- lumination, carbon had to be eliminated altogether as electrode material, and its place was taken by magnetite, while titanium compounds give the high efficiency. This lead to the long 226 GENERAL LECTURES burning luminous arc of the white color of the titanium-iron spectrum as represented by the magnetite arc, the metallic oxide arc, and other types still in development. In all these long burning luminous arcs, some efficiency had to be sacrificed in developing sufficiently small units for general illumination. While the substitution of the flame car- bon in the open arc has quadrupled the light at the same power consumption, and the substitution of the magnetite electrode for carbon at the same power consumption would in the same manner increase the light, for street illumination the main problem was, -to decrease the power consumption rather than increase the amount of light given ; and so in the long burning luminous arcs, which are now beginning to replace the carbon arcs of old, the power consumption has been reduced by from 30 to 60% with a sufficient increase of light to be marked. In the arc lamp, the current is carried across the gap between the terminals by a stream of vapor of the electrodes ; thus the electrodes consume more or less rapidly. Some feeding mechanism is therefore required to move the electrodes towards each other during their consumption. This arc lamp mechanism may be operated by the current, or by the voltage, or by both; this gives the three different types of lamps: the series lamp, the shunt lamp, and the differential lamp. In the series lamp, an electromagnet energized by the lamp current, and balanced against a weight or a spring, moves the carbons towards each other when by their burning off, the arc lengthens and the current decreases. Obviously, this lamp cannot be used on constant current circuits, or with several lamps in series, but only as single lamp on constant potential circuits, and therefore has practically disappeared. In the shunt lamp, the controlling magnet is shunted across the arc, and with increasing arc length and consequent ARC LIGHTING 227 arc voltage, moves the electrodes towards each other. In con- stant current circuits, this lamp tends towards hunting, and therefore requires a very high reactance in series; it thereby gives a lower power factor in alternating current circuits, and has therefore been superseded by the differential lamp. It has, however, the advantage of not being sensitive to changes of current. In the differential lamp, an electromagnet in series with the arc opposes an electromagnet in shunt to the arc, and the lamp regulates for constant arc resistance. It is the lamp now universally used in constant potential and constant cur- rent systems, is most stable in its operation; but in constant current systems, it requires that the current be constant within close limits : if the current is low, the arc is too short, and the lamp gives very little light, and if the current is high, the arc becomes so long as to endanger the lamp. From the operating mechanism the motion is usually transmitted to the electrode by a clutch, which releases and lets the electrodes slip together. In the carbon arc lamp, the mechanism is "floating" ; that is, the upper carbon, held by the opposing forces of shunt and series magnets, moves with every variation of the arc resist- ance, and so maintains very closely constant voltage on the arc. In the long burning luminous arc, as the magnetite lamp, which the light comes from the arc flame, and thus constant length of arc flame is required for constant light production. The floating mechanism, which constantly varies the arc length with the variation of the arc resistance, has therefore been superseded by a mechanism which sets the arc at fixed length, and leaves it there until with the consumption of the electrodes the arc has sufficiently lengthened to cause the shunt coil to 228 GENERAL LECTURES operate and to reset the arc length. Thus in some respects, these lamps are shunt lamps. During the early days of the open carbon arc lamp, 9.6, 6.6 and 4 amperes were the currents used in direct current arc circuits, with about 40 volts per lamp. The 4 ampere arc very soon disappeared, as giving practically no light. In the enclosed arc lamp, the carbons are surrounded by a nearly air tight globe, which restricts the admission of air and thus the combustion of the carbon, and so increases ithe life of the carbons from 8 or 10 hours to 70 to 120 hours. In these lamps, lower currents and higher arc voltages, that is, longer arcs, are used : in direct current circuits, 6.6 amperes and 5 amperes, with 70 to 75 volts per lamp; in alternating current circuits, 7.5 and 6.6 amperes are used with the same arc voltage. In the direct current magnetite arc lamp, 4 amperes and 75 to 80 volts per lamp are used ; in the alternating current titan- ium carbide arc lamp, only 2.5 amperes and 80 to 85 volts per lamp are used. APPENDIX I LIGHT AND ILLUMINATION Paper read before the Illuminating Engineering Society, December 14, 1906. REVISED TO DATE. I. OMPARED with other branches of engineering, as the transformation of electrical power into mechanical power in the electric motor, or the transformation of chemical into mechanical energy in the steam engine, we are at the disadvantage when dealing with light and illumination, that we have not to do any more strictly with a problem of physics, but that we are on .the borderland between applied physics that is engineering, and physiology. Light is not a physical quantity, but it is the physiological effect exerted upon the human eye by certain radiations. There are different forms of energy, all convertible into each other, as magnetic energy, electric energy, heat energy, mechanical momentum, radiating energy, etc. The latter, radi- ating energy, is a vibratory motion of a hypothetical medium, the ether, which vibration is transmitted or propagated at a velocity of about 188,000 miles per second; and it is a transverse vibration, differing from the vibratory energy of sound in this respect, that the sound waves are longitudinal, that is, the vibration is in the direction of the beam, while the vibration of radiation is transverse. Radiating energy can be derived from other forms of energy, for instance, from heat energy by raising a body to a 2 3 o GENERAL LECTURES high temperature. Then the heat energy is converted into radi- ation and issues from the heated body, as for instance an incan- descent lamp filament, as a mass of radiations of different wave lengths, that is, different frequencies. All kinds of frequencies appear : from very low frequencies, that is only a few millions of millions of cycles per second, up to many times higher frequencies. We can get, if we desire, still very much lower fre- quencies, as electromagnetic waves, such as the radiation sent out by an oscillating current or an alternating current ; but the radiations which we get from heated bodies are all of extremely high frequency, compared with the customary frequencies of electric currents. At the same time .they cover a very wide range of frequencies, many octaves, and from all this mass of radiations, from all the frequencies of radiating energy, some- what less than one octave can be perceived by the human eye as light. Light, therefore is the physiological effect exerted upon the human eye by a certain narrow range of frequencies of radiation. Frequencies lower than those visible to the eye, and frequencies higher than those visible to the eye, are again invisible. We frequently speak of those frequencies which are lower than the visible ones, as radiating heat, and of those frequencies higher than the visible ones as chemical rays. This, however, is misleading, and there is no distinction in character between radiations of different frequency. There are no heat rays differing from light rays of chemical rays. Any form of energy when destroyed gives rise to an exactly equivalent amount of some other form of energy. If there- fore we destroy radiating energy by intercepting the beam of radiation by interposing an opaque body in its path, then the energy of radiation is converted into some other form of LIGHT AND ILLUMINATION energy, usually into heat. That means that any radiation when absorbed produces heat and the amount of heat pro- duced merely represents the amount of energy which was con- tained in the radiation. If the radiation contains a very large amount of energy, the heat evolved by intercepting it may be sufficient to be felt by putting your hand in the beam. If the amount of energy is less, it may not be possible to feel it, though with a sensitive instrument, as a bolometer, we may still be able to measure the heat. All radiations therefore are convertible into heat: the visible light waves as well as the invisible ultraviolet rays, and the usually more powerful long ultrared waves ; but none of the radiations can be called heat, no more than the mechanical momentum of a flywheel is heat, because when destroyed, it produces heat. If we consider the infinite range of radiation issuing from heated bodies, we find that those rays which are of lower fre- quency than the visible rays will be felt as heat, because they contain a very large amount of energy. The rays which are visible represent very little energy and therefore they do not give as much heat. For instance, in the case of a hot steam boiler, although we get no light, we can feel the radia- tion from it by the heat which it produces when intercepted by our hand held near it. We do not feel the radiation as heat which issues from the green light of the mercury lamp, merely because the energy of radiation in the latter is less than the amount of energy in the radiation .f rom a hot steam boiler ; but while it is less in the former case, it happens to be of that fre- quency which affects the eye and is visible. As a consequence, when we speak of cold light, this does not mean that it is different from hot light from the light, for instance, given by a hot coal fire, where we feel the radia- tion as heat; it merely means that what is usually called cold 232 GENERAL LECTURES light (as the light of the firefly is supposed to be) is radia- tion containing to a very large extent rays of the visible frequencies and not much energy outside of the visible range ; i. e., containing very little total energy, so that the energy when destroyed, that is, converted into heat, cannot be felt easily, but requires more delicate methods of determination; while a very inefficient light, as a coal fire for instance, which gives most of its energy as invisible radiation of low frequency, very little as visible radiation, can be felt by the heat pro- duced by the interception of the rays, mainly the energetic low frequency rays. As stated, then, there is no essential differ- ence between so-called heat waves and light waves, but any radiation can be converted into other forms of energy, .the so- called chemical rays of ultraviolet light, the X-ray, as well as the ultrared and the visible rays, and when converted into heat can be noticed as such. Now it just happens that most of our means of producing radiating energy give high intensi- ties of radiation only for very low frequencies, invisible ultra- red rays, but we are not able to produce anywhere near the same intensities of radiation for higher frequencies. So also, when we speak of ultraviolet, or short, high frequency waves, as chemical waves, that does not mean that they have a distinctive character in producing chemical action any form of energy, naturally, can be converted if we know how, into chemical energy, the long ultrared waves just as well as the short ultraviolet waves. It just happens that those chemical compounds which are easily split up by radiating energy, are silver salts or salts of gold and platinum ; they are especially affected by the ultraviolet and violet rays. We observe, then, the chemical action of these rays, but do not observe so well the chemical action of other rays. There may, however, be some feature in the constitution of matter, which LIGHT AND ILLUMINATION 233 accounts for the high chemical action of the ultraviolet and violet rays. It is obvious that if we intercept and destroy radiations and so convert their energy into other forms of energy, if .the energy is only great enough, we get a high temperature, and thus a high chemical action, merely by the effect of temperature. But we may also get a chemical effect by what probably is some kind of a resonance phenomenon. The particles of a body, atoms or molecules, must have some rate of vibration of their own. If, then, a ray of radiation impinges upon them which is of a frequency of the same magnitude as the inherent rate of vibration of -the atom, by resonance this vibration of the atom must rapidly increase in intensity until the atom breaks away from the others, or the molecule breaks up, that is, the chemical combination is split up. The inherent frequency of oscillation of the atom seems to be of about of the same magnitude as the visible radia- tion, or rather of a little higher frequency; that is, if the atoms are left to vibrate freely as under the influence of an electric current in the arc, then we get radiations of the frequency inherent to the atom. The general tendency then is toward the violet or short wave end of the spectrum. If we assume -that the mass of the silver atom is such as to give a rate of vibration in the range of the violet and ultraviolet, it is easy to understand that radiation of this frequency splits up the silver salt by increasing -the vibration of the atom by resonance, and that shorter or longer waves have no effect, or much less effect. So it may be a mere incident that those chemical compounds on which we observe the chemical action of radiation just happen to be sensitive to the violet end of the spectrum. It is indeed a fact that other chemical changes brought about by radiating energy, as the formation of ozone 234 GENERAL LECTURES from oxygen, that is, the splitting up of the oxygen molecule and reforming of the ozone molecule from the atoms, do not take place in the violet or ultraviolet, but requires frequencies very much higher, about the highest frequencies which the mercury arc at low temperature gives. Possibly, since the oxygen atom is so much lighter than the silver atom, its fre- quency of vibration is much higher, which means that resonance effects and destruction of the molecules take place only with a much shorter wave length of radiation, or much higher frequency. ( Vice versa, it seems that these frequencies which are chemically active in organic life, which give the energy absorbed from radiation by plants, and so the chemical activity utilized in building up the growth of vegetation, are not at the violet, but at the red end of the spectrum. It appears that the red and ultrared rays produce growth of plants and the chemical activity which we call life, while the violet and ultraviolet rays kill. Even ithis we can well under- stand if we consider the chemical activity as a resonance phenomenon, because the metabolism of protoplasm which we call life, is based on the existence of unstable structures of carbon compounds. We have here not atoms combining with each other, but groups, chain and ring formations, which are of larger mass and therefore have a lower rate of vibration and so should be expected to respond .to lower frequencies or to red light, as indeed seems to be the case. The violet and ultraviolet light does not split up the organic matter into groups, which recombine to form complex bodies, and so represent the changes called life; but due to its higher fre- quency, resonance occurs with the atoms, that is, the organic compound splits up into atoms and so disintegrates, which means death. LIGHT AND ILLUMINATION 235 So it can be understood that the chemical activity of different radiations may be different; the chemical activity of long rays gives life to the vegetation and the short waves, death; one splits up into carbon groups and .the other carries destruction down to the atom. The popular distinction between heat waves, chemical waves and light waves, therefore is not a physical distinction, but all are radiating energy of the same character, differing merely in wave length, and the visible range is somewhat less than one octave, rather at the upper end, at the higher fre- quencies, which are difficult to produce. This makes the prob- lem of investigating and dealing with light difficult for the engi- neer, because it is not any more a physical quantity which can be measured accurately, as in .the case of power or velocity, but it is a physiological effect. We can, indeed, measure very accurately the total energy of radiation from a heated body, but the total energy of radiation is not light : only a very small part of it is visible. We can go further and split up the total radiation issuing from a hot body, as the incandescent lamp fila- ment, into its different wave lengths and different frequencies ; as for instance, we can resolve the (total radiation into the spectrum by using a prism to separate the different frequencies, and then collect the total of the radiation within the visible range, by a lens or other means, and measure all the energy of the visible radiation. Or, still simpler, although approximate, we may interpose in the beam of radiation some medium which absorbs the invisible long rays and invisible short rays, and which transmits, all or rather most, of the visible rays, as for instance glass and water. In this manner one could easily measure .the energy of the visible radiation, and compare the energy of the visible radiation with the total energy producing this radiation. That would give a physical measure of the 236 GENERAL LECTURES efficiency of producing visible radiation but it would not be a measure of the efficiency of producing light, since unfortunately the different wave lengths of visible radiation are very differ- ent in their physiological effect. The same amount of energy as visible radiation, giving the effect of green light, represents an entirely different amount of light, a many times greater physiological effect than the same amount of energy as red rays, that is, rays of the wave lengths which give the impres- sion of red light. That means, the physiological effect or light-equivalent of mechanical energy within the visible range is a function of the wave length and varies with the wave length, that is, with the color. That really is obvious, if you think of it: if you follow the range of frequency from a low frequency to a high frequency, you see that energy radiating at low frequency represents no light whatever, has no physiological equivalent, is invisible. When you come into the visible range it has a physiological effect. Therefore, when you pass from the invis- ible into the visible range, the physiological equivalent must pass from zero into a finite value and must necessarily pass continuously, tthat is, at the extreme end of the visible range; the light equivalent of energy must be extremely low, and the further you go into the visible range, the greater it is, reaching the maximum in the middle of the visible range, in the green and yellow, and decreasing again down to zero at the violet end of the visible rays ; beyond that, at still higher frequencies, the physiological equivalent of energy is zero again ; or, vice versa, if we consider the mechanical equivalent of light, it is a mini- mum in the middle of the visible range, where one candle power of light represents the lowest amount of energy, and increases toward the ends of the spectrum of the visible range, to infinity at the ends of the visible range. LIGHT AND ILLUMINATION 237 Now, that means, in plain language, that the efficiency of light production is a function of the wave length, that is of the color, and that it is at its maximum in the middle of the spectrum, where the same amount of power, measured in watts, gives the largest amount of light measured in candle power. So the efficiency of light production is a function of the wave length. Unfortunately, the physiological equivalent of power, or the physiological effect of light varies not only with the wave length, but also with the absolute intensity. Suppose we undertake to compare red, yellow and green lights, or any lights of different colors. First we meet great difficulties in comparing them. We want one candle power in light, as red, yellow, or green. You cannot compare different colors of light directly, since there is no physical measure of light. Lights are compared with a standard lamp, which has a cer- tain color, yellowish white. A light of the same color we can compare exactly; if the color is not much different, we still get an approximate comparison; but with widely different colors, we obviously can not get even an approximate com- parison, can not say when the two sides of the photometer screen, one illuminated by green light, the other by red light, are equal in intensity. There is thus no direct comparison of differently colored lights. You have then to go one step farther and consider that light is used for illumination, is used to see by, and this gives you a fair comparison : you observe at what distance from the two lights, red and green, you can read with the same convenience, read the same kind of print, or to meas- ure more exactly, get the maximum distance at which you can just read a certain size of print, by either light. At that distance the two illuminations are the same, and the two lights so have an intensity inverse proportional to the square of .these 238 GENERAL LECTURES distances. In this manner lights of different color can be com- pared. Necessarily, the comparison has not the accuracy of photometrical comparison. This cannot be expected, since you do not compare physical quantities, but only physiological effects on the eye, and different observers may have different personal contants. The eye of the one may be more sensitive to green, and the eye of the other may be more sensitive to red, and therefore the comparison may be different. However, these individual differences are not great, and different observers, even with widely different colors of light, do not give results differing much from each other, so that a com- parison of intensities of differently colored lights, and thereby a measurement of the intensity of differently colored lights in candle power is feasible, by some such method, that is, of observing the illumination produced by the different lights. You find, however, if you have a green light and a red light, which at a certain distance appear equally brilliant to the eye, then when you get nearer to the two lights the orange red light appears much brighter than the green, and when you go further away the green light appears brighter, and at still greater distances you still see the green light fairly brightly, while the red light is almost invisible. That is, the relative physiological effect of different wave lengths varies, not only with the wave lengths, but also with the absolute intensity of illumination, and while throughout the whole range the sensi- tivity of the eye for green light is much greater than for red light, the difference is far greater for low than for high illumination, that is, the ratio of sensitivity for green com- pared with that for red is greater for faint illumination than for intense illumination. If you desire to express lights of different colors in candle power it therefore seems necessary LIGHT AND ILLUMINATION 239 also to give the distance, or the intensity of illumination at which you have observed ; in other words, the light from the middle and the short wave end of the spectrum gives a better and more efficient illumination where the total intensity of illumination is low, while the long wave or low frequency of the red and orange and yellow light gives a much more bril- liant effect at high intensity than the same volume of light of shorter wave length. This is of importance for the illuminating 1 engineer, because where you desire to get high intensity effects, as in decorative lighting or in advertising, better results are given by the low frequency end of the spectrum, by orange and yellow light, whereas when you are satisfied with low intensity of illumination, as in street lighting, you get better results from the short wave end or the middle of the spectrum, from the greenish-yellow of the Wellsbach gas light and the bluish- green of the mercury lamp, and not from the orange-yellow of the old incandescent lamp. Therefore the white light of the carbon arc gives better results in street lighting than the yel- low of the incandescent lamp, even at equal intensity of illumination. These features have been of less importance until a few years ago, since the available sources of light were all approximately of the same color, varying from the orange- yellow, to yellow and yellow-white, to white; from the gas lamp, kerosene lamp an*d tallow candle of orange-yellow color, to .the yellow incandescent lamp and the yellowish-white arc, yellowish-white sunlight, to the white diffused daylight. This was a fairly limited range. It is only in the last few years that illuminants of high efficiency have been brought out, which give marked and decided color differences, and are available in units of suitable size and of high efficiency, as the greenish- yellow of the Wellsbach gas lamp, the bluish-green of the mer- 240 GENERAL LECTURES cury lamp, and the orange-yellow of the flaming arc, and hence these questions are increasing in importance. II. This brings us to the consideration of the methods of producing light. Until a few years ago, until the develop- ment of the Wellsbach gas mantle, practically all methods of producing light were based on incandescence. That is, by impressing energy on a solid body, either the chemical energy of combustion, or electric energy in the incandescent or car- bon arc lamp, the temperature is raised to such a high degree that amongst the total radiation issuing from the heated body a certain very small percentage appears within the fraction of an octave of visible light. With increasing temperature of the radiating body, the average wave length of radiation decreases, that is, the average frequency of radiation increases and so approaches nearer to the visible range, although still at the very highest temperature which can be produced the average wave length of radiation is very far below the visible. This means that the higher a temperature is reached by an incandescent body, the higher is the average frequency of radiation, and therefore the larger a percentage of the total energy of radiation is within the visible range, as light. The problem of efficient light production by incandescence therefore is the problem of reaching as high a temperature as possible in the luminous body. In the gas flame and the kerosene lamp, this temperature is the temperature of combustion, rather lim- ited. In the incandescent lamp it is limited also. In the latter case the temperature which can be reached is limited by self- destruction of the incandescent body. The highest temperature probable which can be reached is the boiling point of carbon ; it is reached in the crater of the LIGHT AND ILLUMINATION 241 carbon arc lamp, and therefore the carbon arc gives the most efficient incandescent light. It is incandescent light, because it conies from the incandescent crater, and the arc flame or the vapor conductor does not appreciably contribute to the amount of light issuing from the arc lamp. Very much lower, neces- sarily, is the temperature of the incandescent lamp, of the car- bon filament. The problem is to find materials which can stand very high temperatures, to increase the temperature of the gas flame as well as of the incandescent filament. We have increased the temperature of the gas flame by using a gas of higher chemi- cal energy, as acetylene. The acetylene flame is white; the ordinary gas flame is yellow. We have increased the tempera- ture of the carbon filament by replacing the carbon with some more refractory material, such as tantalum, osmium, tungsten, etc., and thus getting a higher efficiency. We can increase the temperature of the gas flame by increasing the rapidity of com- bustion. We can increase the temperature of the carbon fila- ment in the incandescent lamp by increasing the energy input, but if we increase the temperature of the carbon filament, it is more rapidly destroyed. If we increase the temperature of the gas flame by more rapid combustion to a certain extent we have done it already, by having the gas issuing not from a round hole, but from a flat slit, so as to give a larger surface to the flame; if we go still further and mix the gas with air, we get a still higher temperature, a more rapid combustion, but we loose the incandescent body, because the incandescence of the gas flame is the light given by carbon or heavy hydro- carbon particles, floating in the gases of combustion. We could increase the efficiency of the gas flame by mixing the gas with air, as in the Bunsen flame, but we have then to insert a luminous body of some other material, as no carbon is pro- 242 GENERAL LECTURES duced by the gas in its dissociation. We can do it by a skele- ton of platinum wire. In no case, however, can we reach very high efficiencies by incandescence, due to the temperature limit. We could, however, increase the efficiency of light pro- duction if we could find an incandescent body which would not radiate in the same manner as the carbon filament or the so- called black body, but which would give an abnormally low rad- iation in the low frequency range, or an abnormally high radia- tion in the high frequency or visible range. Such a body may be said to give selective radiation, because the distribution of ener- gy in the spectrum amongst the different frequencies of radi- ation is not the same as it would be with an ordinary black body of the same temperature. If we found a body which would give an abnormally low radiation in the visible range, or abnormally high radiation in the invisible range, this body would be an abnormally inefficient light producer. Vice versa, if we found a body giving abnormally high radiation of short wave lengths, in the visible range, or abnormally low radiation of long waves, of low frequency, this would give an abnormally efficient incandescent body. Such bodies exist and the enor- mous progress in gas lighting made by the introduction of the Wellsbach mantle is based on selective radiation, that is, the oxides do not radiate the same range and intensity of waves as a black body, the incandescent carbon, but give an abnorm- ally large amount of visible rays compared with invisible rays ; that is to say, they give a larger percentage of high frequency light rays compared with the low frequency invisible rays. Possibly and even probably some of these highly efficient fila- ments like the tungsten filament, also owe some of their high efficiency of one watt per candle power to selective radiation. When discussing selective radiation, we have first to come to an agreement on what we understand by selective LIGHT AND ILLUMINATION 243 radiation. The question whether an illuminant owes its high efficiency to selective radiation, depends largely on the defini- tion of the term "selective radiation". We have here a simi- lar case to that of the much discussed problem of the "counter electromotive force of the electric arc". Whether the electric arc has a counter e. m. f . or not, entirely depends on the defini- tion of counter e. m. f. In the same way, the decision on the question of selective radiation depends upon what you define as selective radiation. If you define as selective radiation any radiation in which the intensity of radiation is distributed through the total spectrum differently from that of the theoreti- cal black body, then the Wellsbach mantle has selective radia- tion. If, however, you define selective radiation as the radia- tion of a body which gives spectrum lines, or bands, or absorp- tion lines and bands, that is, sharply defined narrow ranges in the spectrum, of abnormally high or abnormally low intensity, then the Wellsbach mantle has no selective radiation. So all discussions and statements on selective radiation have rather little meaning, if the writer does not give his definition of selective radiation. In the following, I define as selective, any radiation which differs in the distribution of its intensity from the radiation of the theoretical black body. In an incandescent lamp filament we do not get a definite pitch, or definite frequency of vibration, but we get an infinite number of different waves. The reason is perhaps, that in a solid or liquid body the vibrating particles are so close together as to interfere with each other. If you could set a body in vibration, in which the vibrating particles, atoms or molecules, are so far apart as not to interfere with each other, as in a gas at low pressure, then they would execute their own periods of vibration, and then .the light from such a body would not be a radiation of all wave lengths, but we would get radiations of 244 GENERAL LECTURES only a few definite wave lengths, or a line spectrum. So in- candescent or luminous sodium vapor gives only one kind of light, a yellow spectrum line, and in addition thereto a number of utrared and ultraviolet rays. Since the spectrum light is based on the non-interference of the vibrating particles, it is easy to understand, that when you bring tthe atoms or molecules closely together as at atmospheric pressure interference may begin, and the lines of the spectrum become more confused and blurr into bands. Therefore, we see in the mercury arc spectrum, which is at low vapor pressure, a small number of definite, sharply definite lines. In the calcium spectrum of the flame carbon arc, we get a large number of lines blurring into each other to an almost continuous spectrum; so also in the white spectrum of the magnetite- titanium arc. When we set a gas or vapor in vibration, it vibrates at its own frequency, independent of the temperature, and it is merely a question of the character of the material whether a very large percentage of the total energy of radiation happens to be within the visible range or outside the visible range. Temperature does not come in as factor, because the frequency of radiation is no longer a function of the temperature, but independent of the temperature. Sodium vapor gives the same frequency of radiation, the same yellow line when the sodium vapor is at low temperature or at high temperature. Some spectrum lines may increase in intensity with an increase of temperature faster than others, and the color of light may change with the -temperature, change from yellow to white or blue, or from green toward white, and red, as the mercury light does with increasing temperature, but that is merely a characteristic feature of that particular body, and not a general character of the temperature effect; the possibility there- LIGHT AND ILLUMINATION 245 fore exists of finding materials which, when energized, as vapor or gas, give a spectrum with a large amount of energy in the visible range, thus giving an efficiency of light produc- tion far in excess of that available by incandescence. So far the only materials which give these characteristics are mercury, calcium and titanium. These three metals give spectra which contain such a large percentage of the total radiation in the visible range, that the amount of light meas- ured physiologically in candle power is far in excess of that which possibly could be produced by incandescence, even with the assistance of selective radiation. Their industrial appli- cations are represented by the mercury arc, the yellow flame carbon, and the white magnetite and titanium arc, and these are of such very high efficiency as to be of higher magnitude than any incandescent light. Even if we consider only these three illuminants, we have quite a color scale. From the orange-yellow of the flame carbon, which is about the longest wave length we could use, to yellow and yellow-white, in the acetylene flame and the tungsten filament. Then we have the greenish-yellow of the Wellsbach mantle, by selective radiation. We have the bluish- green in the mercury arc, and the yellowish-white of the car- bon arc, as well as the clear white of the titanium arc. Each of these can be modified. We can modify the titanium arc, giving all colors from yellow-white to bluish-white, by the addition of other materials which give either yellowish or bluish spectra. We can modify the yellow calcium arc, from the orange-yellow of calcium fluoride down to the yellowish- white of calcium borates. You can modify each color over a certain range, and you can get pretty nearly any color, with the exception perhaps of a clear blue and violet : no means have been found to get approximately the same efficiency in those 246 GENERAL LECTURES colors of very short wave length, as in the other colors of lights. This feature makes the effect of color which I discussed before, the variation of the physiological effect with the bril- liancy of illumination, of more importance now than years ago, when the only method of producing colored light was by the absorption of all other colors. III. After all, however, it is not light, that is wanted, but illumination ; it is not the amount of visible rays issuing from the source of light, the incandescent lamp or gas flame, which is of importance, but the amount of light which reaches the objects we desire to see, that is, the illumination produced by light. In this respect, I believe a mistake has been made by the gas industry as well as the electric lighting industry, for many years, by devoting all energy to the production of light, the development of the lamp, while they have almost entirely left out of consideration, that the production of an efficient light is not the only important problem, but that of the same import- ance is the arrangement of the light so as to get efficient illum- ination, that is, get the greatest benefit from the light produced ; and this feature has been usually left to the tender mercies of the architect or the decorator, who placed the lights where- ever he thought they would look artistic, regardless of the requirements of effective illumination. If you look around, you find cases everywhere of artificial illumination where the lights have been arranged, so that you get a very poor illumin- ation from a large amount of light. To overcome these defects, it is necessary to study the problems involved between the production of light and the physiological effect produced by the light upon the eye, and it requires a careful study, just as LIGHT AND ILLUMINATION 247 any other engineering problem. It is of importance to con- sider not only the amount of light issuing from the source, but the amount of light which reaches the object to be seen by the illumination. The demands of illumination are mainly of two classes, general illumination, and local, or concentrated illumination. Many cases require general illumination, as a meeting room, where it is desired to see equally well everywhere; that is, to get the same intensity of illumination throughout the whole illuminated area. So also a draughting room, a school room, the hall of a house and the streets of a city require general illumination, a uniform fairly high intensity in a draughting room or school room, a relatively low but as nearly as possible a uniform intensity in the streets of a city. It is true, street lighting is usually very far from uniform, but that merely means that the problem of proper street lighting is usually not solved in the most efficient and satisfactory manner. In other cases concentrated lighting is required, as in domestic lighting, in the dining room, the living room, etc., where light is desired on the table where we work, eat, read, etc. In such cases, the general illumination of the room is of lesser importance ; it is not needed to any extent, or is frequently undesirable, because a room with a very low intensity of general illumination fre- quently is considered more homelike, especially by the feminine part of the human race. In still other places general illumina- tion may be directly objectionable, as in a sick room. Most cases, however, require a general illumination of moderate intensity, and a far more intense local illumination, as over the desks in an office, or the reading tables in a library. In such cases merely a general illumination would be sufficient, if very intense, but -this is uneconomical and to some extent objectionable on account of the blinding glare, which is disa- 248 GENERAL LECTURES greeable; and so a combined general and local illumination is more efficient and more satisfactory. In producing illumination either direct lighting or indi- rect lighting may be used. That is, the rays issuing from the source of light may either pass directly to the illuminated objects, or they may pass to a reflecting surface, and be reflected from this surface to the object, or may pass through a refracting body, as the frosted incandescent lamp globe, or opal globe of the arc lamp, and so reach the illuminated object. In general, it is obvious that any method of indirect lighting by refraction or reflection wastes a considerable amount of light. That means, the total amount of light which reaches the illuminated object must necessarily be less with indirect light- ing, as compared with direct lighting, with the same amount of light. Indirect lighting can be done by reflection or refraction by some attachment to the lamp, as a reflector or a holophane or frosted globe, or by reflecting the light from the ceilings and walls of the room, on the objects -to be illuminated. In the latter case, it is obvious that white walls give the highest efficiency of reflected light. It is easy to observe that the same source of light in a room with white walls gives several times the intensity of illumination which it gives in a room with black or non-reflecting walls. That means that (the total amount of illumination is increased several-fold by reflection from white walls. So in a draughting room, or school room, by using as light walls as possible, we get the best efficiency of illumination. It is not always feasible to have light walls, especially w r hen you come to machine shops or foundries, and other places where the walls do not remain white, but change to some darker color. The question is, what color do these walls LIGHT AND ILLUMINATION 249 assume? The color of almost everything which is changed by age is due to either iron or carbon. In most cases of discolora- tion by age you see the reddish-brown color of iron and the brownish-yellow color of carbon. This is the color most sub- jects gradually assume. This color of age is in the long wave or low frequency end of the spectrum. To get the benefit of reflected light from walls which cannot be kept perfectly white, a source of light rich in the long low frequency waves, or of a yellowish tinge is therefore more efficient by giving more reflection from the walls than a source of light rich in short or high frequency waves, that is, bluish-white. This effect is very marked when you compare the mercury lamp with the flame carbon lamp. The illumination given by the mercury lamp in a draughting room is very satisfactory. The same illumination in a foundry or machine shop is far less satisfac- tory, and you notice there a marked absence of reflected light, that is, the walls and ceilings all gradually assume a color which is rich in red and yellow, and so reflect very little light of the violet end of the spectrum. Even -the black- begrimed walls of a blacksmith shop reflect a considerable amount of light with an orange-yellow source of light; prac- tically none with the bluish-green of the mercury lamp. Thus the shade of color of the illuminant may be very essential in getting efficient illumination. In the interior of a city, the walls usually have a reddish-yellow color. In that case white or yellowish lights are superior. When outside of a city, the greenish-yellow of the Wellsbach lamp, the bluish- green of the mercury arc, gives a much larger amount of reflected light from vegetation than the yellow of the incandes- cent light and so a better illumination. Vegetation absorbs the long waves, the low frequency radiation; so with a yellow source of light there is practically no reflection from 2 5 o GENERAL LECTURES living vegetation, but only reflection from dead vegetation, and in the light of the incandescent lamp or the flame arc all vegetation appears very poorly, the dead parts are very promi- nent, while the reverse is the case where the light is deficient in the red and yellow, and rich in the green and blue, as with the mercury arc : the green shows prominently, while the dead leaves, etc., are not visible. IV. It is, however, not the amount of light which reaches the illuminated objects which is of importance, that is, not the physical intensity of illumination, but the amount of light which from these illuminated objects reaches the human eye. With the same intensity of illumination, the same amount of light reaching the illuminated object, of the same color, the amount of light entering the eye may vary widely with the opening or contraction of the pupil of the eye. The eye is automatically adjusting for intensity of light. This is the reason we see well at sun light and at the light of the full moon, although the former is many thousand times greater in inten- sity than the latter. The eye can accommodate itself to intens- ities varying over an enormous range. It does this partly by the fatigue of the nerves of vision, partly by the contraction or opening of the pupil. This is undoubtedly a protective device developed in the human race. It means that if we have in the field of vision a source of light of high intrinsic bril- liancy, the eye protects itself by contraction of the pupil and so it receives very much less light in the field of vision where we want to see objects, than if the source of light were taken out of the field of vision. By eliminating the source of light from the field of vision and eliminating the contraction of the pupils resulting from the high intrinsic brilliancy of the illuminating LIGHT AND ILLUMINATION 251 body, we get actually a much larger amount of light into the eye with the same amount of light striking the illuminated object ; that is, we get a higher physiological efficiency. Even with a much smaller amount of light reaching the illuminated objects, we still get more light in the eye. That means if we reduce the intrinsic brilliancy of the illuminant by indirect lighting, by diffusing the light, we may lose a considerable amount of light, actually get a considerably reduced quantity of light on the object which we desire to see, but still we get a larger amount of light from these objects into the eye, because the eye is open further and admits more light and is less fatigued. It follows from this that in efficient illumination, it is of foremost importance to arrange the illuminants so as not to have excessive intrinsic brilliancies in the field of vision when looking at the objects we desire to see. That means that the proper field for the illuminant is outside of the field of vision, or where you cannot get it out of the field of vision, that its intrinsic brilliancy should be reduced by diffusion : thereby we actually get a much higher physiological effect. This is the reason for indirect lighting. We may have a very large amount of light thrown on any object in a room, but if the eye is fatigued by seeing the source of light in the field of vision, we get very little light in the eye, while with a properly arranged indirect lighting, with a much lower amount of light reaching the object, we get a higher physiological effect, that is a better and more efficient illumination. It appears, however, that this automatic protective faculty of the eye was developed through the ages as a protection not against light, but against energy; apparently the eye is pro- tecting against the energy of radiation, not the physiological intensity, and since the energy of radiation is mainly in the 252 GENERAL LECTURES ultrared, in the long waves, the frequency which causes the protective reaction is the frequency of the long wave end of the spectrum, the red and yellow waves; they make the pupil contract. This action is much less for the green and blue rays. That is the reason the eye does not react on the mercury lamp to any great extent. It means a green light, like the mercury or Wellsbach lamp, can be in the field of vision to a much greater extent without causing the contraction of the pupil and so reducing the physiological effect. This is of importance in places where the light cannot well be taken out of the field of vision, as in street illumination. In this case, all the sources of light must be arranged along the street and so must be in the field of vision. By cutting off the red end of the spectrum you eliminate the contraction of the pupil, and get the full benefit of the light between the illuminants, while with a yellow source of light, as with the incandescent or arc lamp of old, you do not get the benefit, that is, the physiological effect of the illumination by a green illuminant in such cases is superior to that by a yellow illuminant, the illumination appears brighter and more uniform. A light devoid of red and yellow rays is at the same time the safest and most harmless, and also the most harmful. It is the safest and most harmless, and gives the most uniform illumination, if its intrinsic brilliancy is sufficiently low to be below the danger limit of energy of radiation, but it is harmful if above that, because the eye does not protect itself against it, probably because these lights have not existed throughout all the ages when this protective action of the eye was devel- oped, and sunlight and fire were the only sources of light, both rich in red rays. This accounts for the rather contradictory effect observed, that green or blue light, as the Wellsbach mantle or mercury lamp, is a very good light to work by, LIGHT AND ILLUMINATION 253 superior to the yellow kerosene lamp, and at the same time there is some suspicion that it is harmful to the eye. It may well be that where it is of very high intensity, the automatic protection of the eye is not sufficient to protect with such light. Where you use such sources of light you can get the benefit of the absence of the contraction of the pupil, but it devolves upon you to arrange the illumination so as not to get the harm- ful effects against which the automatic protection of the eye fails. That means all these lights are superior for illumination if they have low intrinsic brilliancies, but somewhat question- able if they have extremely high intensities. V. It is, however, not even the amount of the light which enters the eye which is of importance in illumination, but the difference in the amount of light. If in the illuminated area the light were of uniform intensity, and everything of the same color, we would see nothing but a glare of light. The seeing takes place by a difference in color, and difference in intensity. Difference in intensity includes shadows. Shadows are thus an essential feature in seeing things. Considering, then, the seeing by shadows and seeing by color differences, you observe that by this feature we can divide illumination into directed and diffused illumination. In diffused illumination light comes in all directions with approximate uniformity, and shadows do not exist : in directed illumination, shadows exist. In some cases shadows are objectionable, and in other cases shadows are necessary for clear distinction, and diffused illumination in such cases would not be satisfactory. As regard to seeing differences in color, it is obvious that where definite color distinctions are required, you can intensify 254 GENERAL LECTURES the sharpness of vision by selecting the color of your light best suited to bringing out the colors desired. Where the color conditions you want to distinguish are .those due to age, iron and carbon, then the light which is deficient in red and yellow, which therefore shows the colors given by iron and carbon, as black, gives a much sharper distinction, and the mercury lamp shows blemishes and dirt much more pronounced than the white light. Again, the sources of light which are very rich in red and yellow rays show these colors due to iron and car- bon very much less, and therefore show blemishes or a slight amount of dirt much less, soften them; and where the color distinctions are those due to these two most prominent ele- ments, in the yellow light their appearance will be greatly softened, and under the green light they will be made harsh and sharp. If you desire to soften effects, as in a ballroom, it would be fiendish to use mercury lamps, but where you want to search out a spot that is soiled, it would be very wrong to use a dull, yellow incandescent lamp or a gas flame, but rather to use the green Wellsbach light, or better still the bluish- green mercury arc, which gives in such case sharp distinction, where white light shows little, and yellow light nothing. Where you desire to see all colors in about the same relation as by daylight, you obviously desire a white light. It is therefore important for the illuminating engineer to select the shade or color of the light and study the require- ments of each case which comes into his charge. It would be just as wrong in one case to use an incandescent lamp, where the mercury lamp would be better, as to do the reverse. We have to distinguish then between general illumina- tion and local illumination, between direct illumination and indirect illumination, and between directed illumination and diffused illumination. These three different classes or distinc- LIGHT AND ILLUMINATION 255 tion to a certain extent overlap. It would be very wrong, how- ever, to mistake them, and a very serious mistake in the design of a system of illumination can very easily be made; for instance, by mistaking general illumination and 'diffused illumination for each other. The problem may be to get uni- form intensity all over. You can get that by distributing a large number of small units all around the cornices and reflect the light from white walls and ceilings and get a very diffused illumination, or you could get general illumination, where the intensity of illumination all over is the same, in a moderate sized room, from one source of light by using one of these sources of light as an incandescent lamp with a holophane reflector, which gives the proper distribution, or you could get light from any other source by controlling the distribu- tion curve of the light so as to get uniform distribution. The former arrangement gives diffused light, the latter directed light. You may get -the same intensity of illumination all over the room, in both cases; but in the former case no shadows, in the latter case absolutely black shadows. Probably in the former case for domestic use the lighting will be unsatisfac- tory and trying to the eyes, because you do not see well, you do not have any shadows, objects around you are not so dis- tinct, because you lose the distinguishing feature of the shadow. In the latter case with the directed lighting from one source, the lighting will be unsatisfactory because you get very dark shadows, and you do not see anything in the shadows, and the eyes will be made tired by trying to see in the very dark shadows. You have to consider how much directed light and how much diffused light you require. In some cases you may desire only diffused light. In the general lighting of a draughting room you do not want any directed light, since 256 GENERAL LECTURES you must have no shadows, because if the ruler casts a shadow, it is trying to the eyes to distinguish between the edge of the ruler and the edge of the shadow, and mistakes may be made. In this case, you see only by differences in color and in the intensity, and not by shadows. You therefore get satis- factory illumination from many small units, or by indirect lighting, reflected light from white walls and ceilings, but you get unsatisfactory illumination from a few units even when properly distributed so as to give uniform intensity all over, but giving little reflected light. In other cases, you may also require a general illumination equal in intensity all over, but you need directed illumination so as to see by the shadows. So for instance, a good draughting room illumination would not be suitable for a foundry. In a foundry, where all the objects assume more or less the same color, you require shadows to see by. Then you need a number of units of light to give directed illumination, but you must not go so far as to be unable to see in the shadows ; you must have some diffusion, or overlapping of the different beams of light. So if you take a satisfactory foundry illumination and put it in the draughting room, even if the intensity were satisfactory, it would be entirely unsatisfactory, and so would be the reverse. It is therefore not merely the distribution of the intensity of the light, which is essential, but also the character, whether diffused or directed light, or how divided between diffused and directed light. In the different lighting problems you therefore meet (the question of concentrated and general illumination, of directed and diffused illumination. In domestic lighting, by reflected light from white walls and ceilings we can get a high intensity, and can increase the illumination several-fold over that given directly from the source of light, such as the incandescent LIGHT AND ILLUMINATION 257 lamp or gas flame. Still the illumination would be unsatis- factory and tiring to the eyes. We all know that in the home a room with white walls is not as agreeable as one with darker walls. We say we have too much light. But we do not have too much light, because we do not have anywhere near the same amount of light as we get during the daytime out of doors. We have too large a percentage of diffused light. The intensity of diffused light is too great as compared with the directed light. We lose the shadows and that is tiring to the eyes. The problem of domestic lighting then is, to get sufficient directed and not too much diffused light- ing so as to get the best vision, that is, to get sufficient shadows to see by, but the shadows must not be so dark as to make seeing objects in the shadows tiring to the eyes. During the daytime we get directed light from the win- dows, diffused light reflected from the walls. To get the proper proportion between directed and diffused light, fixes the shade of the walls, and in general we have to use walls of somewhat darker color. When you come to lighting in the evening, with a source of light like the incandescent lamp or gas lamp, sending out light in all directions, the diffused light compared with the concentrated or directed light is a higher percentage than in the daytime for the same color of walls, partly due to the color of the light, which is yellow, and is more reflected from the walls, largely, however, because with the daylight through the window the directed light is a much larger percentage of the total light than in the lamp, where only a small part is concentrated light. It is not comfortable to have this strong diffused light, and so we put shades on which absorb three-quarters of the light, but which give us a more comfortable illumination in the room. That means waste, however, and you pay for light which you do not use. 258 GENERAL LECTURES The proper illuminating engineering then is to secure the cor- rect distribution curve of the source of light, so as to give the desired amount of concentrated lighting on the dining or read- ing table, and give only as much diffused lighting as is com- patible with the amount of direct light used, to see in the shadows. The problem of domestic lighting, from the illum- inating engineering point, is to determine the illumination over the entire area, and also the character of illumination, whether directed or diffused; how large an amount of light should be concentrated, and how large an amount should be directed ; then the question of colors and shades also comes in as an important factor, as was discussed before. Practically nothing has yet been done in this direction systematically and intelligently, but all has been done by trial which at the best usually means producing more light than necessary, and throw- ing away the excess of diffused light by absorption. APPENDIX II LIGHTNING AND LIGHTNING PROTECTION Paper read before the Annual Convention of the National Electric Light Association, 1907. Revised to date. I. LIGHTNING PHENOMENA IN THE CLOUDS. T HE first man who attacked the problem of lightning and lightning protection, a century and half ago, was our great citizen, Benjamin Franklin. He gave us the lightning rod, which is now universally recognized as the most effective and only protective device for isolated points, as steeples, chimneys, etc. The next step in advance was made by Faraday : he showed that in the interior of a perfectly con- ducting body no electric disturbances can be produced by out- side electric forces. This led to the most effective protection possible against lightning or electric disturbances, the use of a grounded metal cage, "Faraday's cage", enclosing the struc- ture which is to be protected, whether a building against lightning, or a delicate instrument against electric fields. In its simplest form, Faraday's cage, applied to a trans- mission line, is the ground wire above the line, and the pro- tection afforded by it is the more complete, the more the over- head ground wires represent the condition of an enclosing cage of perfect conductivity. That is, a system of wires above and on the sides of a transmission line is superior to a single wire, a wire of high conductivity superior to a small iron wire. Here I specially desire to draw attention to the second requirement of the Faraday cage, high conductivity. Thus it is not sufficient merely to have any kind of overhead 260 GENERAL LECTURES grounded wire regardless how small, but high conductivity of .the grounded conductor is essential in many cases of atmospheric disturbances. In the last ten years, transmission voltages have crept higher and higher, transformers have been built of consider- able size, of still higher voltages, so that exact data on the action of voltages up to 300,000 are now available, and approximate data for potentials above a million volts. It was found that air has a definite and fixed breakdown strength, that is, just as a beam breaks mechanically as soon as the stress in it exceeds a definite value, the breaking strength of the material, so air breaks down by a disruptive spark, as soon as the electric stress in the air, or the potential gradient, exceeds a certain value, which is about 100,000 volts per inch. The disruptive strength of air is, over a wide range, propor- tional to the pressure, that is, at a pressure of two atmospheres it is .twice as high, or 200,000 volts per inch; at one-quarter atmosphere it is one-quarter, or 25,000 volts per inch.* The striking distance in air between needle points has been investigated up to 300,000 volts, and found that for high voltages it is very nearly 10,000 volts per inch, that is, a dis- charge of 30" length between needle points requires 300,000 volts. If between two needle points the potential difference is gradually increased, already at relatively low voltages the dis- ruptive strength of the air at the needle points is exceeded, the air at the points breaks down and becomes conducting, and luminous, as "brush discharge", so that the terminals are not the needle points any more, but the whole space, of approxi- mately spherical shape, which is covered by the brush dis- charge. As result thereof, for high voltage, no appreciable difference exists in the striking distance between needle points * Only at very low pressures, where the distance between air molecules become appreciable, this law ceasei, and the disruptive strength increases again, and seems to become infinitely great in a perfect racnum. LIGHTNING AND LIGHTNING PROTECTION 261 and between spheres, the centers of which approximately coincide with the needle points, as long as the diameter of the spheres is small compared with their distance apart apart. With increasing potential difference between needle points, the brush discharges spread out against each other, until only about 40% of the space between the needle points is free, and then a dis- ruptive spark passes. Naturally, as soon as determinations of spark voltages became available, attempts were made to estimate the voltage of a lightning flash. Considering, in a lightning flash, the dis- charge as that in an ununiform field, similar to that between needle points, and so requiring about 10,000 volts per inch. In this case, a lightning flash of two miles, or about 10,000 feet length, would require a potential difference of about 1200 million volts. The existence of such voltages in the clouds does not appear possible: a potential difference of 1000 mil- lion volts would produce a brush discharge of about one-half mile in length, before the final lightning flash occurs. In the brush discharge the air is electrically broken down, and becomes conducting. But it is also mechanically and chemically broken down, that is, the molecules are dissociated and recombine after the discharge, in all possible combinations. That is, we get ozone and nitric acid, and a lightning flash produced by a thousand million volts would thus be followed by a deluge of nitric acid. This fortunately is not the case. An estimate of the voltage and the current in a lightning flash would not yet give the energy, if the duration of the dis- charge is not also known. We can, however, get an approxi- mate estimate of the magnitude of the energy of the lightning flash indirectly, from photometric considerations, and elimi- nate the consideration of the duration of the flash by the integrating feature of the human eye for impressions of very 262 GENERAL LECTURES short duration: an impression on the human eye persists for some time, about .1 seconds, and any phenomenon of shorter duration than . i seconds so appears to last . I seconds. Hence the effect on the eye by a lightning flash would be about the same whether the flash lasted .1 seconds, or if it were of a thousand times greater intensity but lasting a thousandth of the time. This means that the eye would see a lightning flash about in the same manner as if its light, and so probably its energy were spread uniformly over . i seconds. The illumination given by a brilliant lightning flash is about of the same magnitude as good artificial illumination, perhaps one foot candle, since at night time in a well lighted room, the light of a lightning flash is still quite appreciable. Estimating roughly one watt per candle foot, a lightning flash illuminating a space of two miles square or io 8 square feet, with one foot candle would consume io 8 watts, and as this is the illumination as averaged by the human eye over . i seconds, the energy is io 7 watt-seconds, or 10,000 K. W. seconds. The energy of a large lightning flash, estimated from its light, would thus be of the magnitude of 10,000 K. W. seconds. This value, while considerable when expressed in electric quanti- ties, is by no means so very great: reduced to heat measure, it only equals the latent heat of evaporation or condensation of about 9 Ibs. of water. As seen, an estimation of the voltage of the lightning flash from length and disruptive potential gradient of the air, does not give reasonable values, that is, the lightning flash cannot be a single discharge as that of a Leyden jar. An estimation of the voltage may then be attempted in a differ- ent manner. Lightning flashes usually occur within thunder clouds and only rarely from cloud to cloud or from cloud to ground. LIGHTNING AND LIGHTNING PROTECTION 263 They therefore seem to be rather due to equalization of potential differences within the cloud, than to dicharges between oppositely charged bodies. Lightning occurs mainly when rapid condensation of moisture takes place in the air and the electric phenomena seem ,to be the more intense, the greater the rapidity of condensation, or rain formation. Thus the atmospheric electric disturbances seem to be connected with the condensation of water vapor to clouds and rain. There exists normally a potential gradient in the air. That is, a potential difference exists between the air at different elevations, reaching sometimes several hundred volts per foot, so that we can estimate as a fair average, a natural potential gradient in the air, in vertical direction, of about 100 volts per foot. A point 100 feet above ground may show a potential difference of about 10,000 volts against ground. Usually the higher strata of the air are positive against the lower. The cause of this potential gradient, whether terrestrial or cosmic, is of no interest to us here, but merely its existence. It is of interest to investigate, what effect must be expected, from our well-known physical laws, from the con- densation of moisture, and agglomeration of the moisture particles to rain drops, in an atmosphere having such a poten- tial gradient. Assuming water vapor in a higher stratum of the atmosphere to condense to moisture particles, these moisture particles have the potential of the air in which they float, that is, have a considerable potential difference, perhaps hundred thousands of volts, against ground, and so contain an electric charge against ground. These moisture particles conglomer- ate with each other to larger moisture particles and ultimately rain drops. By the collection of n 3 particles into one, the diameter of the particle has increased n fold. Its capacity 264 GENERAL LECTURES has also increased n fold (the capacity of a sphere being pro- portional to the diameter). The particle contains, however, the accumulated charges of n 8 smaller particles, and n 3 times the charge, with n times the capacity, gives n 2 times the poten- tial. It follows herefrom that with the conglomeration of the water particles, their potential must increase rapidly, propor- tionately to the square of their diameter. The conglomeration of moisture particles in the clouds is, however, very uneven, due to the uneven distribution of moisture, as is plainly seen by looking at any cloud : dense or dark parts representing consid- erable condensation and so considerable moisture content, alternate with light parts, in which little or no condensation occurs. As a result thereof, starting with a uniform potential in the stratum of the air, where condensation begins, differ- ences of potential distribution by necessity result from the differences in the condensation of water vapor to moisture and the accumulation of the moisture particles to larger ones, that is, the denser portions of the cloud are at a higher potential than the lighter portions. Thus, starting with uniform poten- tial, and thus zero potential gradient in the air at the moment of the beginning of condensation, potential differences and thus potential gradients appear. Such potential differences in the clouds increase with increasing agglomeration of moisture particles to rain drops, and so the potential gradient rises. Assuming even as low a potential gradient as 100 volts per foot in the cloud at the beginning of agglomeration of moisture particles, the collec- tion of n 3 such particles to one rain drop of n times the diameter and so n times the capacity, but containing the static charge of n 3 particles, gives n 2 times the potential, and since the dis- tances between the particles are now n times as large, the potential gradient has increased n fold. That is, by conglom- LIGHTNING AND LIGHTNING PROTECTION 265 eration of water particles, the potential gradient rises propor- tionately to the diameter of the particles. Estimating then the average diameter of moisture particles as icr 4 inches at the be- ginning of agglomeration, when the potential gradient in the cloud is about 100 volts per foot, then the breakdown potential of the air, of between 100,000 and 200,000 volts per foot, would be reached when the drops have reached about .1 to .2 inches diameter, that is, the size of rain drops. Potential gradients in the cloud thus gradually rise, until somewhere the disruptive strength of the air is reached, and a discharge passes, equalizing the voltage at this spot. This, however, causes a greater potential gradient at the ends of the discharge, exceeding the breakdown strength of the air, and so causes a second discharge, following partly over the path of the first, then a third and so on, until all of the potential differences or inequalities of the potential distribution in the cloud, are leveled down by a series of successive discharges. The phenomenon thus is similar to that of a landslide, setting off another and another landslide. Or it can best be pictured by representing the unequal moisture distribution in the cloud by a relief map built of wet sand, the dense portion of the cloud, and therefore the portions of high potential, being repre- sented by the hills, the light or low potential portions of the cloud by the valleys of the relief map. As soon as (the sand dries, somewhere, where the declivity is very steep, that is, the potential gradient is very high, a slide occurs, this causes another slide and so on, until the whole configuration of sand settles down to a flat and smooth shape, the hills are leveled off and the valleys filled. The existence of such successive discharges, following each other after appreciable intervals of time in the same path, has been shown by the photographs of lightning flashes taken 266 GENERAL LECTURES with a rotating camera. In this case, by the motion of the camera the successive flashes are recorded side by side, and sometimes more than forty successive discharges have been counted, the whole phenomenon lasting about .6 seconds, that is, quite an appreciable time. Oscillographs of lightning discharges from (dead) trans- mission lines also showed the frequent occurrence of multiple strokes, or strokes following each other within a fraction of a second. It follows herefrom, that lightning flashes in the clouds, of several miles' length, occur without any considerable poten- tial difference between the ends of the flash, but result from the disruptive equalization of the unequal potential distribu- tion in the clouds, caused by unequal vapor density and so unequal condensation and conglomeration of moisture particles. This also explains the relatively small tendency to dis- charges between cloud and ground, across a space in which no condensation takes place and so no unequal potential distri- bution supplies the power of the discharge : although the dis- tance between cloud and ground is smaller than the distance traversed by a lightning flash in the clouds, and the average potential differance between cloud and ground probably is greater than -the potential differences in the clouds, a discharge to ground probably occurs in general only where by a heavy downpour of rain a range of high potential is carried bodily part ways down to ground. This also may explain, that light- ning discharges to the ground are usually followed by a heavy downpour of rain. The potential gradient in the air may rise to disruptive values in still another, slightly different manner, and lead to lightning discharges without being accompanied or followed by rain. By conglomeration of moisture particles the potential LIGHTNING AND LIGHTNING PROTECTION 267 gradient rises, as described above, but before the water drops have reached sufficient size to precipitate as rain, evaporation again sets in: for instance by the drops falling to a lower and warmer stratum of the air, or by intercepting the heat of the sun's rays, and the drops thus dwindle away. The decrease in size of the drops represents a decrease of capacity, the capacity being proportional to (the diameter, and as each drop retains the same charge, its potential increases with the decrease of size, without limit, and so also the potential gradient until its disruptive value is reached and the lightning discharge occurs. This phenomenon is frequently observed towards the evening of a hot summer day, and is called "heat lightning", and, being the result of evaporation, thus does not lead to rain. Estimating then as disruptive strength of air under dis- charge conditions in a non-uniform field, and at the reduced air pressure in the clouds, 100,000 volts per foot, the average potential gradient in the path of the lightning discharge through the clouds would be about 50,000 volts per foot. This gradient, however, would not be unidirectional, but the poten- tial would rise from a low, or even negative value at a light portion of the cloud, to a maximum value af i dense position, then decrease again, that is, give a gradient in opposite direc- tion, to a light position, etc., and the potential gradient would vary from nothing at a maximum potential point, to a maxi- mum, equal to the breakdown strength of air at the starting point of the discharge, to zero at a minimum potential point, etc. To estimate the current which discharges in the lightning flash, the conductivity of air in the path of the discharge, and the diameter of the discharge are required, and as both are unknown, any estimate must be very approximate only. The specific resistance of gases and vapors decreases with 268 GENERAL LECTURES increasing temperature and with decreasing- pressure. It is a few ohm centimeters at atmospheric pressure and the high temperature of the magnetite or carbon arc, and is also a few ohm centimeters at the low temperature and low pressure of a high current Geissler tube discharge. The mercury arc stream also gives a specific resistance of a few ohms. The temperature of the air in the lightning discharge probably is moderately high, but the pressure is also not far from atmos- pheric, so that 100 ohm centimeters may not be very far from the true magnitude of the resistance. Estimating one to two feet as the diameter of the discharge path, and 100 ohm centi- meters as the specific resistance, and allowing for the induct- ance, gives, with an average potential gradient of 50,000 volts per foot, a current of about 10,000 amperes. The heating effect and the magnetic effect of lightning strokes also point to the passage of currents of some thousand amperes. Assuming then the average potential gradient in the light- ning flash as 50,000 volts per foot, the current as 10,000 am- peres, a lightning flash of two miles' length would represent a power of 5 x io 9 K. W. Estimating the energy of the discharge, as approximated from the photometric consideration, as 10,000 K. W. seconds, the duration of the discharge would be: io*/5 x io 9 = 2 x io~ 6 sec., or two-millionths of a second. The discharge probably is oscillatory. In view of the high resistance of the discharge path, the damping effect must be very great, that is, a very large part or nearly all the energy is expended in the first half-wave ; that is, the discharge consists of only one or very few half- waves. With a duration of the discharge of 2 x io" 6 seconds, assuming two half- waves as an average, gives 500,000 cycles. LIGHTNING AND LIGHTNING PROTECTION 269 The frequency of oscillation of the lightning flash thus appears to be of the magnitude of half a million cycles. Since the velocity of propagation of electric disturbances is the velocity of light, or 188,000 miles per second, the wave 188,000 3 length of a discharge of 500,000 cycles is = miles, 500,000 8 or about 2000 feet. A wave length of 2000 feet means that the current in the discharge flows in one direction for 1000 feet, in the opposite direction, that is, with opposite potential gradient, in the next thousand feet, etc. That is, in our former discussion, the average distance through which the potential gradient has the same direction, or the distance between maximum and mini- mum, between densest and lightest parts of the cloud is about 1000 feet. This agrees fairly well with the appearance of the clouds to the eye, and it also agrees in magnitude with the dis- tance over which the wind velocity varies, in gusts, as shown by Prof. Langley in his investigation on the "internal energy of the wind". It appears herefrom, that the varying wind velocity as measured by Prof. Langley, that is, the gusty character of the air currents, results not only in an internal mechanical energy, which the bird utilizes for soaring, but also results in unequal moisture distribution, and so, when condensation occurs, in an "internal electrostatic energy" of the thunder cloud, which dis- charges as lightning. With an average length of the half -wave of 1000 feet, and 50,000 volts per foot as potential gradient, the potential differences in the clouds would be of the magnitude of fifty million volts. These are values which appear reasonable. Assuming that a lightning flash drains the electric energy of the cloud within a radius of about 100 to. 200 feet from the 270 GENERAL LECTURES path of the discharge, this affords a different method of esti- mating the magnitude of the energy of the lightning flash: assuming for instance saturated air at 40 C mixing with air at oC, condensation of a part of the moisture occurs which can easily be calculated. Assuming that this moisture has conglomerated to rain drops of .1" to .2" diameter, the num- ber of such drops in a space of two miles' length, and 200 to 400 feet diameter, can be calculated, and also their electro- static capacity. With a wave length of 2000 feet, and a potential gradient of 50,000 volts per foot, from the capacity follows the energy of the electrostatic charge, which dis- charges as lightning flash. This is found under the above assumption, as of the magnitude of 10,000 K. W. seconds, so agrees with the results derived from the photometric considera- tions. To conclude then, as approximate values of magnitude of the electric quantities in a lightning flash may be estimated : Average potential gradient: 50,000 volts per foot at the moment of discharge. Average potential difference between different points of the cloud : 50 million volts. Average current in the discharge 10,000 amperes. I Average duration of the discharge sec. 500.000 Average frequency of discharge : 500,000 cycles. Average energy of the discharge: 10,000 K. W. sec., or seven million foot pounds. II. LIGHTNING IN ELECTRIC CIRCUITS. Of greatest importance to an electrical engineer are the high potential phenomena produced in electric circuits by atmospheric lightning as well as by other causes, frequently LIGHTNING AND LIGHTNING PROTECTION 271 internal to (the circuit, which give the same or similar effects to such an extent, that it has become customary when dealing with electric circuits, to distinguish between external or atmospheric lightning, and internal lightning, as caused by electric circuit disturbances or defects, such as sudden changes of load, or arcing grounds, etc. While a very large amount of data on high potential phenomena in electric circuits has accumulated, the possible variety of phenomena is so great that an intelligent under- standing of the phenomena, as is required for effective pro- tection of the circuits, is feasible only by a theoretical investi- gation of the high potential phenomena which may be expected in electric circuits, and a comparison thereof with the observed effects. In general, the high potential phenomena possible in electric circuits are the same three classes of phenomena which can occur in any medium, as a body of water, which is the seat of energy. 1. Steady electrostatic stress, that is, a gradual rise of potential of the total circuit against ground, until a discharge occurs somewhere; just as in a body of water, as a river, the pressure, that is, the water level, may gradually rise, until it breaks through the embankment. 2. Impulses, or traveling waves, similar to the ocean waves rolling over the surface of the water. 3. Standing waves, or oscillations or surges, similar to the oscillation of a tuning fork, or a violin string. A more extended discussion on the three forms of electric disturbances, and their causes, is given in a paper read before the A. I. E. E.* *A. I. E, E. Transact. March, 1907: "Lightning Phenomena in Electric Circuits." 272 GENERAL LECTURES Steady electrostatic stress obviously can occur only where the circuit is very well insulated from the ground, but not in a grounded circuit, or a leaky circuit, as low voltage circuits usually are, and such static stresses can be eliminated by a permanent leak, that is, a high resistance connection between the circuit and the ground. As sources of impulses or traveling waves only two characteristic phenomena may be considered here: the light- ning flash, or induction by the clouds, as external, and the arc- ing ground as internal cause. Assuming a thunder cloud to pass over the line. The ground below the cloud then assumes an electrostatic charge, corresponding to the opposite charge of the cloud. The trans- mission line, as part of the ground, thus also assumes a static charge, higher than that of the ground, since it projects above it. . Any equalization of the potential distribution in the cloud by a lightning flash, as discussed in the preceding, requires a change in the electrostatic charge of the line, corresponding to the changed potential difference between ground and cloud above the ground, and the static charge thus set free on the line rushes as an impulse or wave along the line. The wave shape of such impulses induced by cloud discharges is in general not a smooth sine wave, but may be very irregular: during the equalization of the cloud potential by the lightning flash, the potential difference against ground, of the part of the cloud above the electric circuit, may vary in almost any conceivable manner, thus giving rise to very different wave shapes of the impulses. So some impulses may rise very rapidly, with extremely steep wave front, and slowly die down. Others may rise slowly, then suddenly fall and reverse, or a series of oscillations may occur in the impulse, etc. If the lightning flash is parallel with the line, simultaneous impulses of different LIGHTNING AND LIGHTNING PROTECTION 273 directions may be produced, corresponding to the different directions of the potential gradient in the different parts of the lightning flash, and these waves, of different directions, intensity and wave length, traveling over each other, then pro- duce a very complex system of phenomena. So for instance, by the intereference of two impulses of nearly equal wave length, moving in opposite directions, a high voltage point may be produced, traveling slowly along the line, and visible to the eye as a luminous streak. The frequencies of these impulses then are those corres- ponding to the frequencies of cloud discharge, that is, of the magnitude of hundred thousands of cycles per second. With the velocity of light, 188,000 miles per second, they travel along the line, until they gradually fade out by the dissipation of their energy, or are reflected at an open end of the line, or at the entrance to the station are broken up by partial reflection, in reactances, and interference between the reflected waves, the incoming waves and the waves passing over the reactances, and so give rise to systems of standing waves or oscillations, similarly as an ocean wave rolling on to a sloping beach breaks up into surf. Where a traveling wave is reflected, the combination of the reflected wave and the incoming wave produces a standing wave or oscillation, that is, a wave in which the voltage maxi- ma and the zero points or nodes have fixed positions on the line. By superposition of the wave maxima of incoming and reflected wave, the standing wave rises to a maximum double that of the traveling wave. Where different oscillations or standing waves superimpose upon each other, their maxima substract at some places and add at others, and thus again double the voltage, that is, a traveling wave or impulse, break- 274 GENERAL LECTURES ing up into systems of oscillations at a station, doubles and quadruples the potential; so that a traveling wave of moder- ate potential may cause dangerous voltages when breaking up into oscillations, just as in the ocean surf, the waves rise to far greater heights than in the on-rolling ocean wave before it reaches the beach. If we consider that the impulses traveling along the line are not sine waves, but of very irregular shape, that is, can be considered as consisting of a fundamental of some hundred thousand cycles, and numerous higher harmonics of still greater frequency, and each of the components when breaking up at the station gives rise to a set of oscillations at every inter- ference point, that is, at every reactance, the complexity of the phenomenon can be imagined. Since the equalization of cloud potential usually occurs by a series of successive discharges in short intervals, a small fraction of a second, and each discharge gives rise to an impulse in the line, and so a system of oscillations at the sta- tion, whatever protective device is used, must restore itself instantly after a discharge, so as to receive the next following discharge. Any device depending on mechanical motion to restore itself after a discharge to operative position, therefore fails to protect, when a series of discharges follow each other in very rapid succession, as discussed above. Traveling waves very similar in character to those due to induction from the clouds, but frequently of far greater volume, sometimes occur in an electric circuit from internal causes, as arcing grounds, or spark discharges. Let, for instance, a spark occur in an insulated under- ground cable system between one of the conductors and the grounded cable armor, through a weak spot in the insulation, as a faulty joint or a cable bell. Normally a potential difference LIGHTNING AND LIGHTNING PROTECTION 275 exists between the cable conductor and the ground, equal to the Y potential of the system, and so an electrostatic charge on the conductor corresponding thereto. A spark passing between conductor and ground, connects it to ground, and the charge of the conductor so passes over the spark as arc to ground. As soon, however, as the conductor is discharged and at ground potential, the arc between conductor and ground ceases, since there is no voltage left to maintain it, and so the conductor disconnects from ground. The conductor then charges itself again to its normal Y potential and during the in-rush of the charge, momentarily the potential builds up to double voltage. Thereby a spark again passes between con- ductor and ground, discharges it again, opens after discharge, again causes a spark to pass, etc. So a series of successive sparks occur between conductor and ground, discharging the conductor by currents which momentarily rise to very high values, the discharge current of the capacity of the conductor against ground, over a path of practically no resistance. Each spark discharge sends out an impulse or traveling wave, and thus a spark discharge between conductor and cable armor, or in the same manner an arcing ground on an overhead transmis- sion line, as is for instance caused by a broken insulator, pro- duces a continuous series of impulses or traveling waves, which follow each other with the rapidity of charge and discharge of the cable or the line, that is, many thousands per second, and so give what has been called a recurrent surge. In a long distance transmission line, the frequency of the recurrent surge usually is somewhat lower than in an underground cable system, but is still thousands of impulses per second. The frequency of oscillations occurring in electric cir- cuits varies over an enormous range: from low frequencies, very little above alternator frequency, up to hundreds of mil- 276 GENERAL LECTURES lions of cycles per second ; and the effect of the oscillations in the system therefore varies accordingly: from the relatively harmless static displays; brush discharges, streamers, sparks, etc., of extremely high frequencies, down to the disastrous high power low frequency short circuit oscillations, in which even in 10,000 volt system*, currents of many thousands of amperes may surge, which voltages approaching 100,000, and with which no protective device can cope, which does not have unlimited discharge capacity, that is, contains no resistance whatever in the discharge path. III. LIGHTNING PROTECTION OF ELECTRIC CIRCUITS. From the preceding considerations it follows that the problem of protecting electric circuits from lightning is two- fold: 1. To guard against high potential disturbances enter- ing the circuit from the outside or originating in the circuit. 2. To discharge harmlessly to ground, whatever high potential phenomena may appear in the circuit. From atmospheric electric disturbances, complete protec- tion can be secured by putting the circuit under ground, or, where this is not feasible, to put the ground over the electric circuit. This means the use of grounded overhead wires. The overhead ground wires so protect the circuit the more com- pletely, the more they realize a complete shield interposed between line and sky. While complete protection thus would require a system or network of grounded conductors above, beside, and also below the transmission line, very good protec- tion in most cases is secured by a single ground wire of good conductivity, installed well above the line; and in no place of LIGHTNING AND LIGHTNING PROTECTION 277 electric transmission systems can money be more efficiently spent, than in securing good overhead ground wire protection. To guard against the appearance of internal lightning requires constant watchfulness in the design, construction and operation of the system, to avoid all conditions which may lead to the formation of oscillating arcs. Thus poor contacts, loose joints, masses of insulated metal near high potential con- ductors, etc., should be carefully avoided. The disturbances which have to be taken care of by the lightning arresters proper, are steady accumulation of static pressure; impulses or traveling waves; oscillations or surges ; occurring singly or in groups, and of frequencies vary- ing between many millions of cycles and ordinary machine frequencies; and recurrent surges, that is, impulses and oscil- lations, usually of high frequency, following each other in very rapid succession, usually thousands per second. It is necessary that the discharge over the lightning arrester should occur with the least possible disturbance to the system, that is, the discharge current should be as small as per- missible without causing a voltage rise due to the resistance of the discharge path. At the same time, the protective devices must be able to discharge practically unlimited currents, that is, currents of the magnitude of the momentary short circuit current of the system. This obviously requires that the pro- tective devices should have no appreciable resistance in the discharge path. Any lightning arrester containing series resistance obviously fails to protect as soon as the discharge current is so large that the ohmic drop across the resistance becomes serious, and the maximum discharge current which may occur, is the short circuit current of the system, that is, extremely large. 278 GENERAL LECTURES Three types of protective devices are at present available. 1. The circuit is connected to ground by a single spark gap set for a voltage exceeding the normal operating voltage by a safe margin : the so-called horn gap, or goat horn light- ning arrester. As soon as the voltage rises beyond the value for which the spark gap is set, it discharges, and the system is short circuited to ground, until the arc rises and gradually blows itself out. As this requires an appreciable time, motors and converters have usually dropped out of step, and the gen- erators have broken synchronism, that is, the system is shut down and has to be started up again. This type of protection therefore is not particularly favored in systems which require reasonable continuity of service, but if used, it is considered rather as an emergency device in addition to other arresters and is then adjusted for much higher discharge vokage. A reduction of the current over the horn gap by series resistance is not per- missible, since it correspondingly reduces the protective value, as explained above, and the arrester ceases to protect against a high power surge. While such surges are relatively infre- quent, their destructiveness is such that protection against them is especially needed. Fuses in series with the horn gap, if they open slowly, would still shut down the system, and if opening very rapidly, the shock of the explosive opening of the fuse on the short circuit current of the system may be disastrous. Obviously, the use of series fuses require a multi- plicity of spark gaps to give continuity of protection. 2. The type of lightning arrester now almost universal- ly used is the multi-gap arrester, which short circuits the system for one-half wave only. It consists of a large number of spark gaps between metal cylinders, in series with each other. As now designed, different sections of the gaps are shunted with different resistances, for the purpose of affording equal LIGHTNING AND LIGHTNING PROTECTION 279 protection against all frequencies, and adjusting automatically the resistance of the discharge path to the volume of the dis- charge, as for instance, discharge slow accumulations of poten- tial over a very high resistance, short circuit surges over a path of zero resistance, and thus pass a discharge with the minimum shock on the system. The operation of the multi-gap which by the way is suitable only for alternating current systems depends on the non-arcing character of cer- tain metals. Metals of low boiling point, as mercury or zinc, cannot maintain an alternating current arc, but the arc goes out when at the end of the half wave, the current falls to zero, and a very much higher voltage is required to again start an arc for the next half-wave.* Alloys of such metals, usually zinc, with metals of high melting point, as copper, are therefore used as (terminals in the multi-gap arrester. A discharge over the multi-gap arrester short circuits the system for the rest of the half- wave during which the discharge occurs. At the end of the half- wave, the current falls to zero, and the reverse current cannot start, that is, the circuit of the arrester is opened. A short circuit on the system, for a fraction of a half- wave, does not interfere with the operation of synchronous apparatus, that is, the operation of the system is not affected by a discharge over the multi-gap arrester. In a large system, the short circuit current is very consid- erable ; its power, and thus the heating effect produced by it, is enormous. The energy, and thus the heat produced by the short circuit current during the fraction of the half -wave, which the discharge over the multi-gap arrester lasts, is moderate, due to its very short duration, and can easily be absorbed and radiated See paper A. I. E. E. Transact. 1906, p. 789. "Transformation of Electric Power into Light. ' ' 280 GENERAL LECTURES by the arrester; so that even if lightning discharges rapidly follow each other for some time, they can be taken care of by the arrester with moderate temperature rise : assuming a vicious thunder storm, in which lightning flashes succeed each other practically continuously, several per second. Each discharge causes a short circuit over the lightning arrester, varying in duration from nearly a half-wave if the discharge occurs at the beginning of a half- wave to practically nothing if the discharge takes place near the end of a half-wave that is, in average, for one-half of one-half wave, or 1/240 sec., in a 60 cycle system. Therefore from two to three lightning dis- charges per second would still short circuit the system over the multi-gap arrester only for i % of the total time, and the heat- ing effect, caused by a short circuit during i % of the time, can be taken care of by the arrester for a considerable period. Let us see, however, what happens to the multi-gap light- ning arrester in case of the appearance of a recurrent surge, as an arcing ground, that is, discharges following each other in rapid succession, thousands per second. The first discharge, passing over the lightning arrester, short circuits the system for the rest of the half-wave, and at the end of the half-wave, the arrester functionates properly, that is, opens the circuit. At the next moment, however, at the beginning of the next half- wave, the next oscillation of the recurrent surge again dis- charges over the arrester, and thus again short circuits. That is, with a recurrent surge, the multi-gap arrester at the end of every half-wave opens the circuit, at the beginning of the next half -wave, the next oscillation of the recurrent surge short circuits again. As far as the effect on the operation of the system, and the heating of the arrester is concerned, a recurrent surge causes a permanent short circuit on the system, except that at the beginning of every half-wave, for a LIGHTNING AND LIGHTNING PROTECTION 281 short period, the circuit is opened and free for the appearance of disruptive voltages elsewhere, and so apparently, simul- taneous with the short circuit, destructive high potentials may appear in the system. The heating effect of the short circuit current, which occurs at every half -wave, rapidly destroys the arrester. In such cases, to save the arrester, it has been cus- tomary to insert a series of auxiliary gaps, which are thrown in by the blowing of a fuse shunting them, and raise the dis- charge voltage of the arrester so that the recurrent surge does not pass over it. It is obvious, that in this case the arrester ceases to protect the system against the recurrent surge: but if left in circuit, the destruction of the arrester would put it out of operation anyway. It is obvious now, that no lightning arrester, which func- tionates by short circuiting the system for the rest of the half- wave, during which a discharge occurs, can take care of and protect against a recurrent surge, since the proper functionating of the arrester, with a recurrent surge, represents a permanent short circuit on the system over the arrester, and so a destruc- tion of the arrester, no matter whose make it may have been, and a shutdown of the system. 3. To take care of a recurrent surge, a protective device would thus be required, which does not short circuit the system even for one half- wave, but which never allows the normal voltage of the system to pass a current over the arrester, but acts as a short circuit for any excess voltage above the normal voltage. The possibility of such a device we can understand by considering the effect, which in a direct current circuit a storage battery would have, when shunted between the circuit and the ground. Assuming for instance in a 500 volt trolley circuit, a 500 volt storage battery of very high capacity, that is, negligible internal resistance, permanently connected 282 GENERAL LECTURES between line and ground. With the normal line potential of 500 volts, no current would pass over the battery to ground, except the very small current required to maintain the battery charged. No rise of voltage, however, could occur in the system by lightning or any other cause, since any voltage above 500 volts, the counter e. m. f. of the battery, would be short circuited to ground through the battery, and such a battery would thus give perfect protection against any high voltage dis- turbances in the system. In case of a recurrent surge, the cur- rent discharging over the battery would be the short circuit current of the excess voltage, that is, the surge potential, and the heating effect of this current is negligible, since high potential high frequency phenomena are of limited power and especially of limited current, as condenser discharges. A storage battery obviously is not suitable for alternating current and would not be practical in any case, as it requires a cell for every two volts. The same effect, however, is pro- duced at a much higher voltage, in an alternating current cir- cuit, by the aluminum cell. If such a cell, consisting of two aluminum plates in certain electrolytes, is exposed to an alter- nating voltage, a film forms on the aluminum plates, which holds back the impressed voltage, that is, acts like a counter e. m. f. equal to the impressed e. m. f., so that practically no current passes through the cell, or only the small current required to maintain the film, of a magnitude of about .01 amperes per square inch plate surface, while for any sudden rise of voltage the cell acts as a short circuit for the excess voltage. Over the storage battery, the aluminum cell has the advantage of higher voltage: a single cell can take care of 300 to 400 volts and even more, and also that it does not have a fixed counter e. m. f., but a counter e. m. f., which adjusts itself to equality with the impressed voltage, at any value up LIGHTNING AND LIGHTNING PROTECTION 283 to about 600 volts per cell. Assuming for instance an alumi- num cell connected across an alternating e. m. f. of 300 volts. With the film formed, a negligable current passes through the cell, for instance, 1/4 of an ampere, maintaining the integrity of the film. If now the voltage is suddenly raised to 330 volts, in the first moment the cell acts as a short circuit of the excess voltage, in this case, 30 volts, and for an instant a very large current, possibly hundreds of ampers if the supply source is capable of giving such a current, rushes through the cell. This current very rapidly decreases, by the film of the aluminum plates forming for higher voltage, so that in a few seconds the current is already small, and in a few minutes the normal current of 1-4 ampere again passes, but now at 330 volts impressed, and the film has formed to a counter e. m. f. equal to this higher voltage, probably has thickened. If now we again lower the voltage suddenly to 300, in the first moment the current in the cell practically disappears, and then graually rises again, and after a few minutes is again normal at 1/4 ampere, that is, the film has built down again to 300 volts. In this manner the aluminum cell adjusts its counter e. m. f. to changes of impressed voltage, by the film building up or build- ing down. This adjustment, for moderate voltage variation, as may be expected when varying the generator voltage of the system, is quite rapid, most of the change occurring within less than a second, but is still extremely slow compared with the rapidity of lightning phenomena, and for lightning phenomena .the aluminum cell therefore acts as a short circuit of the excess voltage above the normal machine voltage. Thus the recurrent surge, with a system of aluminum cells in series with each other connected directly across the circuit, cannot produce any rise of voltage, but the excess voltage over the normal, or the surge potential, is short circuited through the aluminum 284 GENERAL LECTURES cell, so causing a small increase of the current in the cells, by the superposition of the high frequency surge current over the normal leakage current of the cell, but no rise of voltage. Since the recurrent oscillations are intermittent, obviously the film of the aluminum cells cannot build up to their voltage, but remains corresponding to the machine voltage, that is, the aluminum cell can permanently discharge a recurrent surge without any short circuit of the main voltage, or any disturb- ance on the system. UNIVERSITY OF BERKELEY Return to desk from which borrowed. This book is DUE on the last date stamped below. C.1NV. 1 JUN 4 1952: a i J V i s V 1 iA 9 \ IV C- iu i ' ' * '' ' ' 1 ~* ,. $ 2. v '"'"'% & ^^ P 789562 UNIVERSITY OF CALIFORNIA LIBRARY ^^M^