REESE LIBRARY OK THK UNIVERSITY OF CALIFORNIA. ,190 on No. 82893 Class A POWER DISTRIBUTION FOR ELECTRIC RAILROADS BY LOUIS BELL, PH.D. THIRD EDITION. REVISED AND ENLARGED NEW YORK : STREET RAILWAY PUBLISHING COMPANY 1 20 LIBERTY STREET COPYRIGHTj 1900. BY THE STREET RAILWAY PUBUSHING COMPANY, NEW YORK. PREFACE. This little book is written in the hope that it may be of service to those whose daily work is concerned with the art of transportation, in which electrical traction is to-day so potent a factor. The part it may play to-morrow only the prophet can say. The author has endeavored to set forth the general principles of the distribution of electrical energy to mov- ing motors, to describe the methods which experience has shown to be desirable in such work, and to point out the ways in which these principles and methods can be co-or- dinated in everyday practice. The art of correctly design- ing systems of distribution requires, more than anything else, skilled judgment and infinite finesse-, it cannot be reduced to formulae in which these terms do not enter as variables. The most that can be done is to sketch the lines of thought that, followed cautiously and shrewdly, lead to good results. For the most part apparatus is too mutable to describe exhaustively, unless one is writing history. The reader will therefore find little of such detail, save in the frontier region which lies between established tramway practice and that greater field that stretches toward unknown bounds. Along that frontier experiment has blazed paths here and there, and we must note them carefully. We can see whither they lead, but dare not say how far. The best advice that can be given to the engineer is to keep his eyes and ears open and never to let himself get caught out of sight of experimental facts. 82893 PREFACE TO THE THIRD EDITION. The past three years have been marked by nothing particularly startling in the way of innovations. Electric railways are more numerous and longer, and apparatus is on a bigger scale now than then, but the changes in ' methods have been few and trifling. The most notable line of growth has been in the increasing use of substa- tions with rotary converters, sometimes with admirable results, sometimes with more energy than discretion. Stor- age battery auxiliaries have also been on the increase with varying economic effect. On the other hand, little progress has been made in heavy railway work, save in the multi- plication of roads, and the problem of economical distribu- tion for such work has not advanced toward solution. The substation idea has not yet been advanced beyond the crude conception of substituting motor for engine in an auxiliary station without improvement in the feeding system of the motors. This volume deals with principles and methods, and consequently, while it is pleasing to note that the motor has practically driven the locomotive off elevated roads, that third rail surface traction for heavy service is on the increase, and that long distance surbur- ban roads have developed with splendid rapidity, one must regretfully admit that the examples of each given in the first edition are, save in unimportant details, as typical of .current practice in 1900 as they were in 1896. Improve- ments d/e, however, already overdue, for we are still a long way from the filial development of electric traction. CONTENTS. CHAPTER I. FUNDAMENTAL, PRINCIPLES. Classes of distribution Fundamental formulae of distri- bution Plate I, showing size of conductors for various losses Losses with distributed and moving loads Simple and branched lines Irregular distribution of loads Center of gravity of load Location of centers of distribution Net- works and their computation Variations of load, their nature and magnitude. 1-27 CHAPTER II. THE RETURN CIRCUIT. Nature of the return circuit Conductivity of rails Systems of bonding Arrangement of bonds Resistance of bonds and bond contacts Earth resistance Leakage of current from the rails to earth and other conductors- Elec- trolysis, its distribution and amount Remedies Supple- mentary wires Continuous track Welded and cast joints The double trolley Energy lost by bad bonding Net re- sistance of bonded track Track constants Formula for complete circuit 28-59 CHAPTER III. DIRECT FEEDING SYSTEMS. Systems of arranging feeders Maximum and average drop Load factor of the station Wandering load Compu- tation of a feeder system Extent of lines Average load n lines Center of distribution Maximum loads Trolley wire and track return General feeding system Reinforce- ment at special points Plate II, chart for feeder computa- tion Cost of wasted energy 60-87 VI CONTENTS. CHAPTER IV. SPECIAL METHODS OF DISTRIBUTION. Boosters, their proper and improper use Three wire system Methods of balancing Self contained three wire system Advantage of high voltage Motors in series In- crease of working voltage Arrangement of feeders Com- posite systems 88-108 CHAPTER V. SUBSTATIONS. Auxiliary stations Distributed stations Substations with power transmission Cost of power in stations of vari- ous sizes Single station vs. distributed stations A typical auxiliary station Conditions of economy Typical dis- tributed stations Typical transmission station Compari- son of substation methods Relative costs 109-143 CHAPTER VI. TRANSMISSION OF POWER FOR SUBSTATIONS. Generation of alternating current Modern alternators Necessity of high voltage Inner pole machines Trans- formers Transmission lines Insulators Synchronous motors Polyphase generators Substations with synchro- nous motors Motor generators Rotary converters Lag- ging and leading current Computation of an alternating line of a three phase line Operation of the machines. . . 144-177 CHAPTER VII. ALTERNATING MOTORS FOR RAILWAY WORK. Varieties of alternating motors Synchronous motors with commutating start with inductive start Properties of synchronous motors Polyphase induction motors their structure and properties Regulation of speed Torque Tests of polyphase induction motors for railway service their weak points The first polyphase railway Induction motors on monophase circuits Methods of working poly- phase motors from monophase circuits Monophase induc- tion motors and their properties Application of alternating transmission to railway work Its economic relation to other methods of working .' 178-220 CONTENTS. Vll CHAPTER VIII. INTERURBAN AND CROSS COUNTRY WORK. Conditions on interurban roads Power required Com- putation of the feeding system Economy of various methods of distribution Light electric roads for country districts Narrow gauge roads-cost of construction and op- eration Bicycle and saddle-back roads-cost of construction and operation 221-250 CHAPTER IX. FAST AND HEAVY RAILWAY SERVICE. Kinds of work for which electric power is best suited Suburban passenger traffic Conditions of competition be- tween steam and electricity Power required for operating electrically Special trolley systems Computation of the feeders Cost of overhead system The third rail system The Nantasket road Cost of electric power for suburban work Very high speeds Air resistance Other resistances Tractive power required Track for extreme speeds Methods of operation The braking problem Computation of the feeding system Cost of power Electric elevated roads The Metropolitan road of Chicago The Lake Street road of Chicago. Distribution of the power Special heavy service The Baltimore & Ohio tunnel Results already at- tained Electric passenger locomotive Methods of power distribution 251-303 CHAPTER I. FUNDAMENTAL, PRINCIPLES. The distribution of electrical energy for use in propel- ling railway cars is, by nature, a special problem. It deals with magnitudes and distances greater than are usual in other branches of electrical engineering, and, in addition, with the difficulties of a load that constantly shifts in amount and position. Consequently, the design of a dis- tributing system is of singular difficulty. In computing the area of conductors, one ordinarily assumes the load to be the only independent variable, but in this case the distance of transmission must be so consid- ered, and both quantities are of the most erratic character. The general equations can therefore only be solved within limits, except in special cases, and even then only by very judicious assumptions. It is therefore worth while to investigate these limits, their extent and the causes which impose them. The conducting system of an electric railway, large or small, consists of three somewhat distinct parts the work- ing conductor, the return circuit and the feeders. By the first is meant that part of the total circuit from which the moving contact, carried by the car, immediately derives its current. Physically it is a wire or bar, uninsulated as re- spects the moving contact, and supported in any position overhead, on the ground or under the ground that cir- cumstances may require. The return circuit is, in a large proportion of cases, that which receives current from the wheels of the car, and is composed, partly or wholly, of the rails. In certain cases, conduit roads, double trolley roads and telpher sys- tems, the working and return conductors are alike and of 2 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. equal resistance. They may therefore be treated alike as parts of the working circuit. The ordinary return circuit calls for special investigation, because it is a heterogeneous conductor, unequal in resistance to the working conductor, and involving unusual complications. The feeding system in railway work serves the double purpose of reinforcing the conductiyity of the working con- FIG. i. ductor and equalizing the voltage at various parts of the system. It therefore must be deferred as a practical matter until the working system, which it supplements, has been considered. Three classes of working systems are common, making the classification according to the nature of the distribution. ID FIG. 2. The first class is illustrated by the linear system, shown in Fig. i. Ideally it is a straight line, A B, near some point at which the power station is generally situated. It may be modified by bends or curves, as in A 1 "B l and FUNDAMENTAL, PRINCIPLES. A 2 B 3 , but whether it be a small tramway line along a single street, or a long interurban road, it retains as its main characteristic a single working line, not generally re-curved on itself, and subject throughout its length to fairly uni- form conditions of traffic. The second class is illustrated by the branched type, represented in Fig. 2. As shown, it consists of a main line, A B, into which run two branches, C D and K F. The branched distribution is the one most commonly met with in electric street railways of moderate size, and may assume an infinite variety of forms. It is the legitimate re- suit of growth from the linear type, and, through all its modifications, is noteworthy in consisting of several lines which are neither interlinked, although often overlapping, nor subject to the same traffic conditions. Its conducting system is therefore essentially complex. Finally, we have the meshed system, Fig. 3. Ideally, it is, as shown, a simple network composed of parallel lines crossing each other at right angles and at nearly equal in- tervals, and under fairly uniform conditions. Practically, the various lines composing the network cross at all sorts of angles and intervals, and are subject to all sorts of condi- tions of traffic. All networks however have this property, tnat they are composed of interconnected lines, so that the conducting system of any line can reinforce, andean be- re- inforced by, other systems. Fig. 4 shows that portion of 4 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. the Boston network which lies within a mile radius from the Post Office as a center. It conveys an idea, better than any words, of the sort 'of network that occurs in practice. It differs totally from the networks usually met with in elec- tric lighting, in that it is without any pretense of sym- metry, either in configuration or load. In all large installations one is likely to find all three types of distribution, usually a network in the center, and branched and linear distribution in the outlying districts. In laying out the system as a whole, each type must con- form, as far as practicable, to its own conditions of economy, while the general feeding system must consider them all. The starting point in any discussion of a* con- ducting system for any purpose is Ohm's law in its simplest form C=- R FIG. 4. In problems of distri- bution such as we are considering, the term in- volving R is usually the quantity sought, since the current and loss of potential are generally known or assumed. It is therefore desirable to transform this simple equation into some form which allows the ready substitution of the known quantities to determine the un- known. The resistance of any conductor may be written R = K-^-, in which A is the cross section, L the length A and K a constant depending on the material considered and the units in which L and A are measured. If L is in feet and A in square inches the constant is obviously different from what it would be if I, were taken in miles. The con- stant is, in practice, so taken that R will be in ohms when FUNDAMENTAL PRINCIPLES. I, ?.nd A are in convenient units. In English-speaking countries it is usual to take L in feet and A in circular mils, i.e., circles T oVo of atl incn ' m diameter. The con- stant connecting L, in feet and A in circular mils with the resistance in ohms, for copper wire of ordinary quality at ordinary temperatures, is u. This is approximately the resistance in ohms of a commercial copper wire one foot long and yoVir f an i ncn * n diameter. The exact figure is a trifle less, but the ordinary contingencies of temperature, joints, etc. , make it desirable to take 1 1. Substituting now this value of R in Ohm's law it be- comes, reckoning the area in circular mils, Tf C = - or, transposing, c.m. nCL (i) c .m= This is the fundamental equation of electrical distribu- tion. It is like the original form of Ohm's law, strictly a linear equation, so that all the quantities are connected by simple proportions. Doubling K, for example, halves c.m. y while doubling I, doubles c.m. A convenient transposed form is II ly which determines the current which a particular line will carry without exceeding a given loss, and another, ( 3 )E = 2ik c.m. is convenient in figuring the actual fall of voltage. Throughout these equations K represents the fall in volts through the conductor under consideration, and I/ is al- ways the total length of the wire, i.e., double the length of the circuit, assuming a uniform return wire. For grounded circuits the equations give correct results for so much of the circuit as is exclusively copper the grounded portion in- volves a different constant and must be taken up as a sep- arate problem. 6 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. It is often convenient to have some simple expression connecting the area of a wire with its wt ight, so that the latter may be readily taken into account. By a fortunate chance, a copper wire, 1000 c. m. in section weighs almost exactly three pounds per 1000 ft. So if, in equation (i), we multiply the constant by three, and reckon L, in thous- ands of feet, we obtain directly the weight of conductor per 1000 ft. Putting Lm for the length, to distinguish it from the former I, reckoned in feet, we have (4) W m = 3 Thus, if we wish to transmit 100 amperes through 7000 ft. of conductor at a loss of 50 volts, the conductor must weigh 33 X 7 =462 Ibs. per 1,000 ft. The total weight 50 of conductor is evidently W m I, m , and since a simple way of getting the total weight, without reference to wire tables, is often desirable, we may re- write (4), as follows: which gives the total weight directly. These weight form- ulae are very easy to remember and apply, and are accurate to about one per cent. The diagrams of Plate I. put equations (i), (3), (4) in graphic form for ready reference. Four different values of K are assumed, and the unit of power is taken as 100 amperes. The chart is therefore independent of the initial pressure, and serves for transmission at any ordinary voltage. Distances on the horizontal axis repre- sent length of circuit, i. e. , half the total length of con- ductor. To find area or weight per 1000 ft. of conductor required for a certain distance, take an ordinate at the re- quired point on the distance scale and follow it up until it intersects the oblique line representing the assumed loss of voltage. The area of the necessary wire can then be read off on the left hand scale, and the weight per 1000 ft. on the right. The corresponding sizes of the B. & S. gauge wires are annexed to the former scale. In a similar FUNDAMENTAL PRINCIPLES. t -r L i t / 2100 7 -I -/ 1 ~? f 1950 1 _ ; N > H; \ 1 1 / / 1QQQ. 600 - ^ 1 1 / / 1 1650 \ / 1 / \. / 1 / 1 / I f 15W 1 / / / g- 2 1 / / g / / g / f f / Ol b / i -2 ) \ > H f f w 1 / ' / o is 1 1 / 1200 JS g b *>/ / . > ^ / 5 ^7 / / u 2 / / / / Ot. / 900 / { o \ / p 03 \ / P+ 1 / f 1 / / 750 / f / / / T^ / f N . ,xi ' f / 1 ^ fi(Y) / ) ~i ^ OX ) / / PLATE T. AREA AND WEIGHT O required for transmitting 100 tances of 25000 feet and less, at and 100 volts. F COPPER amperes for losses of 25, 5C / / / / l / No A; 1 ^ f / f Jis- f / ,75 100 ~ N )' / j i/ i f / \ T 1 / No > / / f / / i / / Jj// k ^ E 5 10 15 Distance in units of 1000 feet. 20 25 Street Railway Journal 8 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. way the distance for which a given wire will carry 100 amperes at a given loss can be found, while the loss for a given wire and distance can be rapidly approximated by estimating the position of the intersection of the area and the distance co-ordinates with reference to the oblique lines. By noting that the area of conductor varies inverse- ly with K, one can extend the working range of the chart. Halving the area shown for B = 75 gives, for instance, the area for K = 150, and so on. Taking up now the case of linear distribution, it has already been shown that the fall in voltage in any con- ductor is directly proportional to the load and the resist- ance. If, now, a uniform line, A B, Fig. 5, be loaded at B, the voltage evidently decreases uniformly throughout its length. To make the example more concrete, the length A B is taken as 20,000 ft., and the voltage kept constant at A, e. g. , 500. Now, if the drop at B under the given load be 100 volts, a straight line drawn from C to D shows the state of the voltage at every point of the line. An ordinate erected at any point of A B and extended to C D shows the voltage of the line at the point selected, and that part of the extended ordinate cut off between C D and C F shows the loss in volts. If the load be transferred from B to some intermediate point of the line, an ordinate there erected will show the drop and the residual voltage at the new point. C K similarly shows the conditions for a terminal drop of 200 volts. The average drop is evidently half .the maximum in each case, since the minimum drop is o, and the voltage varies uniformly. Now suppose one has to deal with a load moving uni- formly back and forth along A B. If the maximum drop be i oo volts, the voltage evidently moves uniformly along C D, and the average voltage is 450, since half the time the voltage is above this, and the other half an exactly equal amount below. This case corresponds to a line traversed on a uniform schedule by a single car. Such however is not the usual FUNDAMENTAL, PRINCIPLES. condition of things. The normal condition of an electric road of any kind is a plurality of cars. This means that current is taken from the working conductor at a certain limited number of points. In general, these points repre- sent approximately equal loads and, so long as the time table is maintained, are approximately equidistant. In Fig. 6, the uniform straight conductor, A B, is loaded, not, 400 2 300 200 100 3 10 15 Distances 1000 ft. units. FIG. 5. Street Ry. Journal as in Fig. 5, at one point, but at ten equidistant points, the loads being assumed equal, as they would be quite nearly if each load were a car on a level track. Here the conditions of fall in voltage are radically dif- ferent from the conditions of Fig. 5. At the power sta- tion, A, the full current for the entire load is supposed to be delivered at a uniform pressure of 500 volts. Assume the total current to be 200 amperes, and the resistance of 10 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. each of the uniform sections to be 0.05 ohm. The first section carries the whole 200 amperes, and the drop C R is 10 volts. The second section carries but 180 amperes, and the loss is 9 volts, and so on, until the tenth section carries 20 amperes, and the loss has diminished to i volt. Mapping these successive falls of potential on Fig. 6, the curved line, C D, is formed, showing the consecutive Points on line FIG. 6. 9 10 Street Ry. Journal values of the potential on A B. C E, a prolongation of the drop in the first section, shows the result of concen- trating the whole load at B. In such a uniformly loaded line the drop is found as follows : If C is the total current and there are n sections P in the line, then is the current taken off for each sec- n FUNDAMENTAL PRINCIPLES. H p tion, and r is the drop due to that current, where r is the resistance of each section. The drop in the first C C section from A is 10 - r, in the second section 9 - r n n and so on ; i. e. , for the whole n sections the total drop must be (6) E=-^r(i + 2-j- 3 . . .) n But the sum of this series of integers is well known, being - - -. Hence, substituting and reducing, we have 2V This gives the total drop produced by n uniform loads uniformly spaced and aggregating C amperes. It is generally convenient to have working formulae give the cross section of conductor directly, since that is most frequently the quantity to be determined. Equa- tion (3) can readily be transformed for this purpose as follows: c. m. But since the R here concerned is the total resistance, and not the resistance per section r, as in (7), we may write, __ ii L Then substituting this value of rin (7) and reducing, we have 2 \ n This equation gives the area of conductor required for C amperes supplying a line of known length equally loaded at n points at any required terminal drop. For a large number of sections n r I approach- es unity, so that, for a given current in amperes and a 12 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. given terminal drop, the copper necessary for a uni- formly distributed load is one-half that required for the same load concentrated at the end of the line. As the number of sections increases, too, the likelihood of ob- taining a disarrangement of load sufficient to disturb the terminal voltage much, decreases. The effect of a uni- form motion of all the loads on the terminal voltage is small. So long as the schedule is uniform and is ad- hered to, the worst that can happen is a transformation of the system into half the original number of sections. Suppose in Fig. 6 all the load points of odd numbers to be moving to the right and all those of even number to the left, at uniform speed. Then after each point had moved half a section, there would be five sections each loaded with a pair of coincident loads. Applying (7) to the data of Fig. 6, K = 60, assuming the sections uniform. As, however, the first section would be but three- fourths the length of the others, the real loss would be 55 as before. Another equal movement and the ten sections ap- pear in their original relation. Another and we have the five sections, but with an initial section one-fourth the length of the others and total loss of 45 volts. Next would come a ten-section arrangement, but with the first load at A, and K = 45, and so on. The upshot is that while the terminal voltage oscillates through a range equal to the drop in the first section, the final effect on the aver- age drop of uniformly moving the loads is the same as load- ing each section at the middle point or increasing n in- definitely. Hence, in a line with uniformly spaced and uni- formly moving loads, we may assume n+ * 1 = i in (9) and write or, transposing, L ii C c,m.=-. _. That is, the area of the line can be calculated for average FUNDAMENTAL PRINCIPLES. 1$ terminal drop just as if the load were concentrated at its middle point. Hence, for all practical purposes, by making this assumption, equations (i), (4), (5) can be used in calculating the line. To keep the voltage approximately uniform over a linear system of distribution is comparatively easy. In the most favorable case, a number of uniform loads moving uni- formly, the drop is half that met with in the most unfavor- able distortion of the load, z. e. , bunching at the end of the line. This latter condition brings the worst possible load upon the station, barring short circuits. Although long stretches of uniform conductor often occur in railway prac- tice it is usual to reinforce the working conductor by feed- ers variously arranged, as will be shown later. Such feed- ers were very necessary in the early days when trolley wire as small as No. 4 was used, but now, when No. oo is very FIG. 7. commonly employed, elaborate feeding systems are less necessary for linear working. The most important linear distributions are likely to come in long inter urban roads, which often demand special methods of feeding. What- ever these may be, the uniform working conductor is of sufficient importance in every system to warrant this dis- cussion of its general properties. As a corollary to this general investigation, it is evi- dent that in dealing with any linear system such as A B, Fig. 6, the best point for the power station is at the middle point of the line, since under the conditions of uniform load supposed, this point would give the smallest average drop. Since Iy in such case is one-half of its value when the whole line is fed from A, the total copper by equation (5) is re- duced to one-fourth the amount for the same loss. Considering now the branched type of distribution, shown in Fig. 2, it is best to take it up in the simplest available form. This, Fig. 7, shows a main line, A B D, 14 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. with a branch, B C, which is straightened and made paral- lel to the main in order to more clearly show their rela- tions. Unless the branch is of such magnitude and posi- tion as to require special feeders, it is supplied with current from the main linear system. In a few cases the service on a branch is from B to C and back. More gen- erally it is from C to A and back, a part of the cars being devoted to a through branch route. On the section A B, the load is the sum of those due to each line of cars. Be- yond B there are two independent linear systems. If there are m cars on the route A D, and n cars on the route A C, then the load on A B, due to both lines, will be and the loads on B C and B D respectively will be n Wc B~D Z5 and m 15' Consequently, if the section A B is computed for this load according to (10) we shall get the proper conductor for the assumed loss E. The lines B C and B D can then be com- puted for losses E t and E 3 . The values o E, E x , E 2 are usually taken with the condition imposed that E+E^ B + E 3 shall be less than a certain specified maximum. A more general method is that of Fig. 8. Here there is a line, A B, with branches running to C, D, E, F. The loads are /, m, n, o, p, amperes respectively. A B, A C, A D, A B, A F, are now considered as separate, each subject FUNDAMENTAL PRINCIPLES. 15 to its own conditions. Taking now a drop for each line, according to the dictates of economy or convenience, and fig- uring the conductors from ( 10) with the respective currents, an area is found for the conductor belonging to each line. Then the cross section of copper required from A to the first branch is [c m\ \ -f- [c m~] m -f- . . . . That from the. first to the second branch is [c m~\ m -\- \c m~\ n -\- ..... and so on. In practice the conductors would be installed sof the nearest convenient size, neglecting small variations of B from the calculated amount at the termini of the various lines. The same procedure applies to all sorts of independent lines radiating and fed from a common center, whether or not these lines have any sections in common. s We have thus far assumed all lines to be uniformly loaded all along their lengths. It, of ten happens how- B c *IG. 9. ever that for some cause a line is loaded unequally. In the long run, grades partially compensate themselves, since as many cars run down by gravity as go up by the expenditure of extra power, so that their effect shows more in the variations of power required than in the total amount. Not infrequently, how r ever, from the effect of grades, curves or local cars in an extended system, there is a regular demand for extra power at some point of the line. This is shown in Fig. 9. Here the line, A B, is divided into ten sections, each equally loaded, except that at 8 the load is three times the normal. Now it has just been shown that a uniformly (distributed load is the same in effect as if it were concentrated at the middle point of the loaded line; that is, the electrical loads, like mechanical ones, act as if concentrated at their center of 1 6 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. gravity. Hence we may represent the above case by A' B', Fig. 9. If c be ^the normal load of each section, then a load of 10 c will be concentrated at C while a load of 2 c is at D. Hence, following out the principle of center of gravity, the system requires for a fixed value of terminal drop the same extra area of copper as if the whole load, 12 c, were concentrated at K, a point chosen so that 2 cl'= 10 cl. The same result is reached in many cases more simply by figuring the normal uniform load as if concentrated at C, and then treating the load 2^at D as if it were on a separate line, as in computing branches. This is the best procedure when grades and other extra loads are superimposed on normal and regular traffic. FIG. 10. FIG. II. But the principle of center of gravity has another and a broader application. In any case of scattered load the center of gravity of the system is the proper point from which to distribute the power, at least in so far as this point gives the mini- mum weight of copper for a given loss. For instance, in the line of Fig. 9, E is the point from which the power should be supplied, whether direct from a generator or from a feeder, if A' B' is but a single part of a large system. The center of gravity of two points on a line is found by the ordinary balancing principle, as in Fig. 9. The center of gravity of any number of points in a plane is found by an extension of exactly the same method, as shown in Fig. 10. I^et there be, for example, five load FUNDAMENTAL PRINCIPLES. T/ points in value respectively i, 2, 3, 4, 5; required the center of gravity of the system. Take any two points, as 2 and 3, and find their mutual center of gravity, just as in Fig. 9. This will be located at a point at which the whole value, 5, of the 2-3 system may be assumed to be concentrated. Now find the center of gravity of this point and 5; this will be at a point at which the weight will be 10. Then taking i and 4, the resultant weight will be 5. Finally, balance these resultants and the center of gravity of the entire system is found at 15. The order in which the combinations are made is of no consequence, since a given system can have but one center of gravity. Now, suppose the points i, 2, 3, 4, 5, are supplied from a common source O, Fig. n, through lines / 1? / 2 , / 3 , / 4 , / 5 . Referring to equation (5) the total weight of copper in any line, as / 1? may be written W = K cl* , w r here K depends on the uniform drop assumed. For any number of load points thus con- nected to a center O 2 W = K 2 c / 2 . But this is directly proportional to the moment of inertia, 2ml 2 , of the loads considered as weights, about O as an axis. Now the moment of inertia of any body about any axis is composed of the sum of two terms, viz. , first, the moment of inertia of the parts of the body around its center of inertia and, second, the moment of inertia of the whole mass concen- trated at its center of inertia, about the axis chosen. Therefore, the minimum moment of inertia for a given set of loads is obtained when the axis coincides with the center of inertia, thereby causing the second term to dis- appear. Hence the total weight of copper required for supplying, at a given loss, any system of loads is a mini- mum when the system is fed from its center of gravity. And the penalty for disregarding this law is severe, as will presently be shown. For example, take the case of a circular area with an electric system made up of equally and uniformly loaded lines radiating from a power station at the center. It has already been shown that the cross section of copper needed iS POWER DISTRIBUTION FOR ELECTRIC RAILROADS. for a uniformly loaded line is the same as if the load were concentrated at the center. The weight is proportional tc the cross section multiplied by the length. In the circular distribution of Fig. 12, therefore, the area of the conduc- tors is proportional to j-r, the radius of the circle, while their lengths equal r. Hence, the weight of copper for such a distribution is directly proportional to the product of these factors and equals -J K r 2 . If, now, the system is fed from another point than O the center, such as A, the weight of copper will be propor- tional to the new moment of inertia, and, since this is made up of the sum of the terms mentioned, the copper w r ill be y doubled when d 2 =i r 2 , i. e. when d=7 . Itwillbemul- \/ 2 tipled by 3 when d 2 =r 3 and so on, rapidly increasing. The following table gives the relative weights of copper corresponding to a few values, of W. = 2 > " =3, rV$ " =4, rVZ 11 =5, >Vf =n, rVj In any sort of distribution the mechanical analogue furnishes a solution of the copper problem in the ways just indicated. It at once appears from these considerations that the cost of copper runs up with disastrous rapidity if the center of distribution is distant from the center of load. From the data given one can figure out readily the extra invest- ment in real estate that it will pay to make in order to put the station near the center of load. The facts set forth are a powerful argument for the economy of an alternating current distribution with higli tension feeders, if such a system can be rendered available for ordinary railway work. The main objection to locating FUNDAMENTAL PRINCIPLES. 19 a center of supply at or near the center of gravity of the load is the cost of site. For a regularly constituted generating station this cost is often prohibitive, so that it is far cheaper to endure the great increase of copper necessary for feed- ing from a distance. If the central plant be reduced to a substation for supplying an alternating current to the working conductors, the space taken up is so trivial that its cost is almost nominal.. The reducing transformers for a capacity of 1000 k. w., together with switch- board and all necessary station apparatus can easily be FIG. 12. accommodated in a room ten feet square, if compactness is necessary. Nor is there any need of extreme care in the matter of foundations, since there is no moving machinery, save motors for ventilation, in such a substation. Even if the day of alternating motors for railway service be delayed far longer than now seems probable, there are not a few cases in which substations with motor-generators are preferable in point of economy to an immense invest- ment in feeders. At present prices of apparatus such a condition will be met far oftener than would at first glance seem probable. In large cities, where there is a v strong and growing tendency to force all feed wires underground, 20 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. the cost of installing and keeping up conduits adds very materially to the disadvantage of elaborate feeding systems from a distant point. Another class of cases in which special attention to the location of power station is needed may be found in the interurban and cross country roads now becoming common. Generally the distribution is linear or branched, rather than a network. We should not, however, assume that the power station should lie at the middle, end or any other point on the line of the road. It very often happens that the center of gravity of the load, which is the most economical point for distribution, as we have just seen, is not on the line at all. For example, take the line shown in Fig. 13. It consists of three sections connecting, we FIG. 13. may suppose, four towns, A, B, C, D. The configuration of the system is here determined by the topography of the region, the amount of business at each point, and similar considerations familiar in the art of railway location. We may suppose the load of each section concentrated at its middle point as before, forming the load points, a, b, c. Suppose the loads to be as follows : a = 15, b = 10, c = 5. These loads may be taken in any convenient units pro- vided the same units are used throughout. Now, proceeding as before, draw b c and locate the center of gravity of the loads, b and c. This proves to be d, where the concentrated load is 15. Then drawing a d, the center of gravity of the system is found to be at O, quite off the line of the road, although not inconveniently distant from B. In other instances the center of gravity might very readily be as far from any of the towns, A, B, C, D,. FUNDAMENTAL PRINCIPLES. 21 as each is from its neighbor. This example, however, shows a common characteristic of long lines. The network type of distribution found in railway prac- tice is quite different in character and needs from a light- ing network. It is, save in a few instances, such as Fig, 4 (see page 4) , much less complex and is always much more irregular in load. In a well ordered central station for electric lighting, every street in the business district has its main, and the load, while far from regular, does not exhibit the extreme variations found in electric railway work. The general solution of even' a simple network, to find the current (and thence the drop) in each line due to one or more known load points, involves a most forbidding amount of tedious computation. But for the purpose in hand exact solutions are not needed so much as easy ap- proximations. Consider, for example, the simple network of conductors shown in Fig. 14. A is here the source of supply, either the station or the end of a feeder. The load is distributed along the lines, A D, A K, D E, D F, K F, D C, F B and C B. Such a circuit may be said to consist of three meshes, and it contains eight currents which we may call i^ , t 2 etc. In lighting practice it is necessary, knowing the load to be supplied by each line, to figure the conductors so as to maintain uniform voltage throughout the network. This involves algebraic processes too complex for convenient use; in fact the complete solution is a very pretty problem in determinants, which those interested may find elucidated in Maxwell's "Treatise on Klectricity and Magnetism," and somewhat simplified in a paper by Herzog and Stark, pub- lished in 1890. For railway work the conditions are, fortunately, simpler. We know, or can assume with suffi- cient accuracy, the normal distributed load on each of the lines. But we are absolved from any necessity for keeping closely uniform voltage throughout the system, since, even were it a matter of more importance than it ever is, it could only be accomplished by using an enormous excess of 22 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. copper, for a large part of the load is liable at any time to be concentrated on almost any part of the network. Two conditions must at all events be fulfilled. First, each one of the lines, A D, A K, etc., must be able to carry its own proper load without exceeding a standard drop; and, second, the sum of the distributed loads must be carried at certain points, which can be approximately pre- judged, without exceeding a certain maximum drop. It must be noted that the conducting system of a rail- way differs from that of a lighting plant in having a much greater proportion of feeders to mains. In fact the working FIG. 14. conductor of a railway is generally of quite limited carrying capacity. Practically, in laying out a network like that of Fig. 14, one has to cut loose from lighting precedents and deal with a special problem. Following the first of the conditions just named, a convenient first step is to compute the conductors as iso- lated lines, on the assumption that t lt z* 2 , z 3 , etc., are the currents due to the normal load on each line. This fur- nishes the skeleton, as it were, of the conducting system. This work can often be simplified by bearing in mind the main lines of traffic and treating as one their component conductors. For instance, in Fig. 14, if A be the station FUNDAMENTAL PRINCIPLE? 23 it may be convenient to take A D C B as a single conductor carrying a load z 1 + z 6 -f i 8 , and A E F B as another loaded with 2*2+ z* 4 + z 7 . DEandDFmay then be taken sepa- rately. Now, this skeleton must be padded with reference to the second condition mentioned. Suppose that traffic is liable to be congested at or near B. This point is fed by the two main lines in multiple. If the drop chosen for these in making the skeleton would mean a drop at B sufficient to seriously impede traffic, enough copper must be added to relieve this condition. Just where this addi- tion should be made requires the exercise of considerable discretion. If F is a point where congestion is also to be feared the line, A D F, should be strengthened, being the nearest route. If C be threatened, ADC should be rein- forced. In either case the addition should be sufficient to put B out of danger. In any case z 3 and z 5 should be con- sidered with reference to the lines, A D and A K, and the drops in D K and D F so taken as to keep them at good working pressure in spite of any excessive demands near the terminus of the system. In other words, for railway work it is nearly always possible to split up a network into a combination of linear systems and branches, since the loads are, or may be, so uncertain that fine discrimina- tion in minor lines is out of the question. A good development of this splitting principle may be found in Fig. 15, which is a network of three meshes com- posed of two parallel lines, A and B, cross tied by the lines, C D, E F, G H, I J. Let A be a feeder and B the trolley wire and we have the well known "ladder" system of feeding in. As, in practice, CD, E F, etc. , are very short compared with C E, E G, etc. , the system may be regarded as composed of A and B in parallel, the only qualification being due consideration of the possible drop in B between a load point and the two nearest feeding points. But we may suppose A and B to run in adjacent streets and the former to be connected* to another trolley wire on its own street, then a track to run along GH, and so on until 24 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. the full network is developed. At each stage of compli- cation the system may be considered as composed of one or more mains with branches, without sensible error, the inaccuracy of the assumption being negligible compared with the uncertainty produced by the irregular load. The variations of load in an electric railway system are so prodigious as to render the most careful calculations only roughly approximate. They are, in general, of three kinds. First, the momentary variations due to accidental changes of load incident to the nature of the service. Second, periodic general variation of the aggregate load caused by the varying conditions of service throughout the day. Third, shifting of the load to various points of the FIG. 15. system, concurrent with the daily variations in total load, but bearing to them no simple relation. The momentary variations are constantly occurring from minute to minute, almost from second to second. They are most considerable in street railway systems oper- ating but few cars, and their amplitude may then be equal even to the maximum total load, and occur in a fraction of a minute. Such a condition may easily exist in a plant operating eight or ten cars. As the number of cars in- creases, the chance of so great variations diminishes, although somewhat slowly. In very large systems, the ex- treme amplitude of these oscillations of load may be re- duced to twenty or twenty-five per cent, of the total load, but they can never disappear entirely. Their effect on the design of the conducting system is but small, for the volt- age does not have to be kept closely uniform, and the con- FUNDAMENTAL PRINCIPLES. 25 ductors will be laid out for the average load based on the average consumption of energy per car. With a normal drop so computed and with care taken to allow a reason- able margin for maximum loads, these variations of the first class need not constitute a serious embarrassment. The diurnal changes of load based on average readings in which the minor oscillations are suppressed, are great in amount and of much interest. They are due to the habits and occupations of the community served, and often exhibit very curious peculiarities. Further, they are almost as strongly marked in very large systems as in quite small ones and serve to determine the relation of average to max- imum load, which in turn determines the allowance which must be made for drop at extreme loads. Even under very favorable circumstances the difference between average and maximum load is great. This is very forcibly shown in Fig. 1 6, which gives the load line on one of the largest electric railway systems for a December day, just before the holidays. The minimum load is quite uniform from 2 A. M. until 5 A. M. and is only about six per cent of the maximum. At about 5 A. M. the load comes on quite suddenly and con- tinues to rise until about 9 A. M. , when it begins to fall, and keeps diminishing until about 2 p. M. Then it rises, slowly at first and then more rapidly until it reaches a second maximum, about equal to the first, at 6 p. M. Then it falls somewhat irregularly until only the night cars are left. The average load for the twenty-four hours is about six-tenths of that at the two maxima. This difference is what must be kept in mind in providing a due factor of safety in the conductors. The load line is not, of course, invariable, being subject both to accidental and yearly varia- tions, but, in spite of these, it preserves its characteristics and the value of its'* load factor" with remarkable uni- formity. In small systems there are practically no night cars, the service being generally about eighteen hours. Were such the case in Fig. 16, the "load factor " would be 26 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. materially improved, rising in fact to about three-fourths on this supposition. But in small plants the day mimi- mum is relatively smaller than in Fig. 16, so that the load factor is worse. Indeed it only too frequently falls to one- quarter or one-third in roads operating five to ten cars. Any value of load factor over one-half may be considered good in any but the largest plants. In long roads operating a few large cars or trains at 20,000 6 12 P.M. Mid. Street Railway Journal FIG. 16. high speed, the load is subject to smaller casual variations,, but the load factor is apt to be low by reason of the great change made by the stopping or starting of a single load unit. The load during the period of acceleration is likely to be about double the running load even with carefully handled motors, and as this period is often several minutes, there is an excellent chance for the superposition of several such loads. More serious than any others are the variations in the location of load, since these may cause a heavy call for FUNDAMENTAL PRINCIPLES. 27 power at some distant part of the system. Such shifting of the load occurs in nearly all cases of linear distribution, and has already been noted, but it also occurs on all sorts of systems, and is the more serious as it is less to be regu- larly expected. A single blockade may fill a limited dis- trict with stalled cars, and when at last it is broken the call for power is of a most abnormal kind. It does not appear strongly marked on the load line, but shows in the shifting of load from one feeder to another. On systems of moderate size this shifting of load may be very serious. For example, through the baseball season many roads will find nearly their full output demanded at the ball park once or twice a week. The next maximum output may be at the other end of the system, to accommodate some special celebration. Even in a large network, at certain hours, during, and just before, maximum load, the bulk of the load will be within a small district, and within the same district only when the same causes produce the shifting. This wandering of the main load over the system is one of the most exasperating factors in the design of the conductors. It may easily amount to the concentration of a quarter or third of the total load at some quite unexpected point. It can be dealt with only by a minute study of the local conditions, which generally will furnish some clue to the probable magnitude and position of such wandering loads. Whatever may be the general conditions of drop, the conductors must be so distributed as to prevent the sys- tem breaking down when loaded in some abnormal man- ner at some unusual point. No theory can take account of such occurrences; their ill effects can be obviated only by good judgment, which is of more value than many theories. CHAPTER II. THE RETURN CIRCUIT. The outgoing circuit of an electric railway has just been discussed in its more general relations. Before invest- igating the proportioning of the working conductors it is necessary to look into the return circuit. Up to this point it has been assumed that this is similar to the outgoing system as it is in the case of motor systems in general. In nearly all electric railway practice it has been the custom to employ the rails and earth as the return circuit, since the former are good conductors and necessarily in contact with the car wheels, and the latter is as necessarily in contact with the rails. In some cases two running contacts are employed as in the double trolley system, conduit roads, some recent elevated roads, and the like, but in most instances the total circuit of any railroad consists of the outgoing system of copper conductors and a return circuit consisting of the rails and their environment. Now the conductivity of an iron or steel rail is com- puted with tolerable ease, but the rest of this heterogeneous system is most uncertain. It consists, near the surface, of bond copper, tarnished surfaces, iron rust, rock, dirt, dirty water, mud, wet wood and promiscuous filth, and deeper down of all sorts of earthy material, and in cities various sorts of pipes for gas, water, etc. In the early days of electric railroading the resistance of this strange assortment was assumed to be zero on the theory that the earth was the conductor concerned and was practically of infinite cross section. This was shockingly far from the truth and although data are rather scarce, we THE RETURN CIRCUIT. 29 may properly take up the return circuit piecemeal and see what the actual state of things may be. First as to the rails. Mild rail steel is a very fair conductor. Weight for weight it is, comparing the com- mercial metals, just about one-seventh as good a conductor as copper. Now a copper wire weighing one pound per yard has an area of about 110,000 c. m.; hence -an iron bar weighing one pound per yard is equivalent to about 16,000 c. m. of copper, very nearly equal to No. 8 B & S gauge. This enables us at once to get the equivalent conductivity of any rail neglecting the joints. The resistance of a copper wire of 16.000 c. m. is roughly six-tenths of an ohm per thousand feet. Hence the resistance of any single rail in ohms is, per thousand feet R = where W is the weight per yard. Or since two rails form the track R = ~- That is, if the rail used weighs sixty pounds per yard the track resistance is approximately -$$-$ ohm per thousand feet. For convenience the relation between weight of rail and equivalent copper is plotted in Fig. 17. The maximum figure is for mild rail steel. Of late there has been a tend- ency to use a harder steel rather high in manganese. This lowers the conductivity by no small amount, sometimes to one-tenth that of copper, for which the minimum in Fig. 17 is arranged. In close figuring the- conductivity should be measured, and specified in ordering rails. These relations enable one to figure the drop in the track, neglecting joints i by the formulae already given. For this purpose the distance in the formula should be, of course, the actual length of track, not the double length as when a return circuit of copper is figured. Thus one would separate the outgoing and return circuits and com- pute the drop in them separately. For simplicity it is however desirable to make allowance if possible for the return circuit, incorporating it in the constant of the original formula so as to make but a single calculation. 30 POWKR DISTRIBUTION FOR EI^CTRIC RAILROADS. The figures just given emphasize with tremendous force the need of thorough bonding of the track in order to take advantage of its immense conductivity. In the early elec- tric railways this was terribly neglected, the bond wires sometimes being as small as No. 6 and even of galvanized iron. Bonding is of very various character. Its most rudimentary form is shown in Fig. 18. In this case the Circular Mils / Min. > / / f / . / ^ / x / / / ^ / X x*^ / / x^" fiOO 000 jS / s X / f ^ / X t / / ^ > \/ / / jr . s^ J x* 1 C 25 50 75 10a Pounds per Yard MG. 17. bouds merely united the ends of adjacent rails, each line of rails being bonded separately. The improvement of Fig, 19 is quite obvious, for in Fig. 18 a single break compelled one rail to carry the return load. The cross bonding of Fig. 19 adds somewhat to the weight of copper required,' but ties the rails together so that no single break can be serious and nothing save a break from both rails on the same side of the same joint can really interrupt the circuit. A very large amount of track has been so bonded, al- though at present the usual construction is shown in Fig. THE RETURN CIRCUIT. $1 20. The supplementary wire effectively prevents ' ' dead rails.*' In modern practice the bond wires are often as heavy as No. oooo, and are generally tinned to prevent corrosion. All joints in the wire are soldered and the rail contacts made as perfect as possible. It is perfectly clear that the supplementary wire is of little value as a con- FIG. 18. FIG. 19. FIG. 20. FIG. 21. ductor compared with the rails, but it is of service in miti- gating the effects of bad joints. In a few cases this supple- mentary wire is reinforced by a heavy copper conductor laid alongside the track and connected at intervals to the sup- plementary wire as shown in Fig. 21. If the joints made by the bonds and rail are very bad this extra copper may be of service, but good joints render it quite unnecessary. The value of the rails as conductors is so great that every effort should be made to utilize them to the fullest possible extent. 32 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. The seriousness of the joint question may be seen by a moment's reflection upon the data already given. There are about thirty -three joints per thousand feet of rail. This means sixty-six contacts per thousand feet between rail and bond, in addition to the resistance of the bond wire itself. Now, the resistance of a sixty pound rail per thousand feet is, as we have seen, only y^ ohm, in decimals o.oi. If there should be even one-ten-thousandth of an ohm resist- ance in each joint between bond and rail, the total resistance would rise to 0.016 ohm per thousand feet. Add to this, the actual resistance of, say, sixty feet of bonding wire No. o, and the total foots up to 0.022 ohm, more than doubling the original resistance. If the joints were here and there quite imperfect, as generally happens, the rail resistance might easily be increased far more. One would be thought lacking in common sense who needlessly doubled the resistance of an overhead circuit, but in the rail circuit far more atrocious blunders are only too common. A few years ago it was frequent enough to find bond wire simply driven through a hole in the web of the rail and headed on the outside. Fortunately, the need of care here is now better realized and in the last few years the name of the rail bond is legion. Most of the contacts are modified rivets, not infrequently supplied with some sort of wedging device to ensure a tight contact. They are, most of them, good enough if properly applied, but a careless workman can easily destroy the usefulness of even the best bonds. The bond contact proper is often quite distinct from the bond wire and is generally given a greater cross section than the latter, to ensure an ample contact with the rail. Figs. 22 to 25, inclusive, show some of the best current forms of bonds. Fig. 25, the "plastic" bond, is composed cf a layer of a species of amalgam re- tained by an outer wall of cork and squeezed into intimate contact with rail and channel plate. It gives a singularly low resistance contact. As to the real resistance of a bonding contact, experi- ments, as might be expected, vary enormously. The re- RETURN CIRCUIT. 33 sistance of the bonding wire is, of course, determinate, but that of the contact is most irregular, varying with every FIG. 22. COLUMBIA BOND. o o Q 1 FIG. 23. CROWN STRANDED BOND. FIG. 23A. CROWN BOND. FIG. 24. BROWN PLASTIC BOND. FIG. 25. BRYAN BOND kind and size of bond and with the thoroughness with which the mechanical work is done. No part of electric railway construction deserves more careful attention. Cull- 34 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. ing the values of bond resistances from experiments on the bonds shown we get the following table : BOND RESISTANCES. Fig. 22 = 0.000131 ohm. " 23 = o.oooi " 23 A = O.OOO247 " 24 = 0.00006 " 25 0.000175 These resistances are for the complete bonds newly set. As nearly as may be judged, the resistance of a single contact, carefully made, can be counted on to be consider- ably less than .001 ohm. With bond plugs of large surface well set, it would seem safe to count upon a resistance not exceeding .0002 ohm. per contact. The bonding wires should be as short as can be con- veniently handled. The advantage of lessened length appears strongly in the results from Fig. 23. Such bonds under the fish plates are more dificult to apply than bonds around the fish plates, but are of low resistance and well protected. As to size, there is little reason for using any- thing smaller than No. ooo or No. oooo. With about a foot of No. oooo at each joint, and thorough contacts carefully made, the resistance of bonds ought to be about as follows per thousand feet. 66 bond contacts = .0132 33 ft. oooo wire = .00165 Total 0.0148 ohm. This is about one and a half times the resistance of a thou- sand feet of sixty pound rail and corresponds well with actual tests of well bonded track. It is quite near the truth to assume that under average circumstances of good con- struction the bond wire and contact resistance may aggregate about twice the resistance of the rails themselves. As regards the earth there is great misconception both as to its conducting power and the part it takes in modify- ing the rail and bond resistance which we have just been considering. Outside of the metals there are no sub- stances that have even fair conducting properties. That THE RETURN CIRCUIT. 35 is, all other so-called conductors are very bad compared even with a relatively poor conductor like iron. For example, carbon in the form of graphite or gas coke, is usually consid- ered a very fair conductor, yet it has several hundred times the resistance of iron, while nitric acid and dilute sulphuric- acid, the best conductors among electrolytes, have many thousand times the resistance of iron. The acid last men- tioned has a specific resistance of about 0.4 ohm for a cubic centimeter, while the resistance of a cubic centimeter of iron is only o.ooooi ohm. Water, even when dirty as it is found in the streets, would show a specific resistance of 1000 ohms or more. Earth, rock and other miscellaneous components of the ground are even worse, so that it is at once fairly evident that it would take an enormous con- ducting mass even of water to approximate the conductiv- ity of a line of rails. Even in theory the mass of earth really available for conducting purposes is somewhat limited, for if a current be passed between two earth plates, the current density de- creases very rapidly as the lines of flow depart from the direct path between the plates. It has long ago been shown, too, that when such a current is established be- tween, let us say, a pair of metallic balls sunk in the earth, the resistance of the circuit does not vary much with the distance apart of the terminals, but depends greatly on the surface of the ground connections. Numerous experiments, too, have shown that the earth is so heterogeneous, so broken up into strata of varying conductivity, that the current flow takes place mainly along special lines, the general mass taking very little part in the action. If, for example, a ledge of rock is in the line between earth plates, save for possible crevices filled with water, it is practically a non-conductor. At various times and places the value of a true earth return for railway and similar work has been thoroughly tried and has generally been found to be practically nil. In two cases the ground plates were sunk in considerable rivers which formed return circuits for lines in each case 36 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. about four miles in length. The ground plates themselves were of ample area, in one experiment several hundred square feet, and gave every opportunity for good contact with the water. The applied voltage in each set of ex- periments was 500 to 550. The resulting currents were insignificant and the resistance of the earth return proved in one case to be about 85 ohms, in the other but a few ohms less. In another more recent experiment the terminal sta- tions were about 3000 ft. apart. An attempt had been made to use an earth return for a motor circuit, with the usual result, and the failure led to investigation. The experi- ment was arranged as in Fig. 26. At A and B were care- fully arranged ground plates in duplicate. One of each pair was sunk in a well, the other imbedded in a mass of iron filings in damp earth. At i, 2, 3, 4, 5, stations 500 ft. apart, grounds were made by driving large iron bars deep into the earth. The voltages employed were vari- ous, from 60 to 150 volts direct current, and alternating current from a small induction coil. The results were nearly coincident in all the sets of experiments and showed the following curious state of affairs: Stations. "Res. ohms. A . . B 92.4 Ground plates alone. A. .B 121. o Well plates alone. A . . B 66. 8 Both well and ground plates. A . . I 201.6 A . . 2 374.0 A . . 3 92. A . . 4 506.3 A . . 5 180.0 The resistance is evidently not a function of the distance nor of anything else that is at all obvious. The only feature that is what might be expected, is the tolerably regular effect of putting both sets of earth plates in parallel as exhibited in the first three lines of the table. The re- sistances at the intermediate stations show how hopeless it is to predicate anything of earth resistance except that THE RETURN CIRCUIT. 37 it is too high to be of any practical use save for trivial currents such as are employed in telegraphy. Imagine the stations A and B, Fig. 26, to be con- nected by a track consisting of a pair of sixty pound rails thoroughly connected and put in parallel with the circuit via the earth connections. At best this has a resistance of 66.8 ohms while that of the track should, be at worst only a few tenths of an ohm. Following the ordinary law of derived circuits, it is clear that the current returning via A~_! *. 1 i 1 B TflG. 26. the earth is only a minute fraction of one per cent of the whole. If the track could be continuously in good con- tact with the earth throughout its length somewhat more current might be coaxed into the earth return by taking ad- vantage of all the fairly conducting streaks and strata. In rare instances the earth under the track has been found in such condition as to have a material amount of conductiv- ity, enough to lessen the drop through the rails very per- ceptibly. Such cases, while well authenticated, are so uncommon as to be of small value save in showing the enormous irregularity of earth resistance, and the utter lack of any well defined laws governing it. And in prac- tice, track is so laid that it is not in good electrical con- tact with the earth as a whole. Fig. 27 shows in section a type of track construction which has been very widely used. The rail is laid upon a longitudinal stringer tim- ber to which it is spiked firmly. The stringer is secured to the cross ties by angle irons. The ties are well tamped with clean sharp gravel which is packed around them and the stringer, and forms a foundation for paving of block granite set closely in upon the rail. Here the material in contact with the rail and surrounding it for some space is very badly conducting except when the track is flooded. Fig. 28 shows another track construction, which would appear to give even worse conduction between rail and 38 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. earth than Fig. 27. The rails are here supported at each tie by cast iron chairs, without an intermediate stringer, and the ties are set in concrete, while rail and chair are surrounded by coarse gravel on which the paving is laid. In no modern track is the rail in contact with better con- ductors than hard wood, gravel or stone. Consequently there is very little tendency for current to be shunted from rails to earth, unless the former are very badly bonded, FIG. 27. for the paths in derivation are bad and there is little differ- ence of potential between any two points of the track to impel branch currents of any kind. Of course, if one at- tempts to use the two rails as outgoing and return leads, the condition is wholly changed, for the full difference of potential then exists between two neighboring rails and there must be a very large amount of leakage. In fact, if there is any considerable difference of potential be- tween the rails or between them and any other conductor, there will be a perceptible flow of current, even through as bad a conductor as "damp gravel, if the path be not too long. THE RETURN CIRCUIT. 39 Thus it is that while ground plates along the track according to early usage are insignificant in modifying the conductivity of the return circuit, there may be, if tht rails are poorly connected, very perceptible flux of cur- rent from the track to, for instance, a water main running parallel to it and but a few feet away. Fig. 29 shows this state of things. Let A B be the track and C D a water main half a dozen feet below the level of the track. The resistance between any particular points of A B and C D is at all times large, owing to the high specific resistance of the material between them, but the area between A B and CD in a long stretch of track is so great that if the FIG. 28. fall in potential in A B is not very slight indeed, there will be a considerable flow of current into and along C D. To take a concrete example, let A B be twenty rods long, and suppose C D to be a foot in diameter and six feet distant from AB. The total area of material in direct circuit would probably be a strip 100 metres long and not less than a metre wide. Such a strip would contain a million square centimetres area and we then have to compute the resist- ance of a block of bad conductor a million square centi- metres in section and perhaps averaging 200 cm. long. This we can regard as built up of a million strips, each one centimeter square and 200 cm. long, connected in parallel. The total resistance would then be the resistance of one such strip divided by 1,000,000. In fact the resistances of these elements would be very various and the currents- would flow in all sorts of irregular lines, but we are deal 40 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. ing here only with the average result. Suppose the ma- terial has a specific resistance of a thousand ohms per cubic centimetre, then the resistance of one element would be 200,000 ohms, but the whole mass would have a resistance of only one-fifth of an ohm; hence if there should be between track and pipe an average difference of potential of ten volts, an amount sometimes exceeded in real cases, there would be within the distance considered a flow of fifty amperes be- tween track and pipe. As large pipes may weigh several hundred pounds per yard, it is clear that their conductivity cannot be neglected, although in most cases it has no noticeable effect on the resistance of the system. In any case, these extra- neous metallic conductors cannot properly be counted as a B e^yc:'{^;;V^:V;.^ " " Street Railway* Journal FIG. 29. part of the circuit, except under very unusual conditions, since flow of current to them is highly objectionable, as will presently be shown. To sum up the matter of earth return, properly so called, the earth, so far from being a body of high con- ductivity, useful for eking out the carrying power of the rail return, is, for most useful purposes, to be regarded al- most as a non-conductor. Its specific resistance is so high and irregular that it is of no value as part of the return circuit, while its conducting power in great areas comes into play only in an unpleasant and troublesome way. The conduction which occurs is very irregularly distributed and varies greatly from time to time. For all long lines of railroad and for many small street railway systems, the earth may be left entirely out of account, and in large street railway systems it is generally a source of anxiety. In the early days of electric railroading quite the opposite view was often held and roads were constructed accord- THE RETURN CIRCUIT. 41 ingly. In reality the bonding was then so generally ineffi- cient, that probably even the earth may have -improved the r general conductivity. Experience has shown that the view here presented is generally the correct one, and the realization of.it has done-much to improve general prac- tice. Possibly interference with telephone circuits did much to prolong faith in the earth as a conductor, but the .telephone deals with millionths of amperes, which are quite insufficient for operating street cars. . Recurring to Fig. 2Q,:and granting the conditions to be such that a current flows from track to pipe at some point in the system, that current must leave the pipe and either pass back to a part of the track having a lower po- tential or to some other conductor by which it may work its way back towards the station. Now wherever an electric current leaves a metallic con- ductor for one which owes its conductivity, as does the earth, to the presence of liquid, the surface of the former is corroded gnaw r ed away by the chemical action set up by the current. Hence the pipe under consideration w r ould soon show a surface pitted with rust, and eventually the corrosion would extend through to the inner surface of the pipe and start a leak. Similarly the rails are corroded from the exit of the current, but the result is not of much consequence. This matter of, electrolytic corrosion of water pipes, gas pipes and other buried conductors is serious in very many electric railway systems, so serious that it is worth detailed study as one of the gravest factors bearing on the design of the return circuit. One would naturally suppose that the actual amount of damage done by the compara- tively small currents distributed over a large space, would be rather slight. So it would be if it were intermittent, but when the electrolytic process goes steadily on week after week and month after month, the aggregate result is somewhat formidable. One ampere flowing steadily from an iron surface will eat away very nearly twenty pounds of metal per year. So, in the case of conduction to a pipe 42 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. just investigated, the resulting corrosion would amount to half a ton per year. This destruction would be done in the surfaces of exit from the pipe and if the conditions were such as to limit these surfaces to a comparatively small area the local damage would be very serious. Klectrolytic corrosion of underground conductors by stray currents was first noticed in the case of lead covered telephone cables in Boston by I. H. Farnham, to whose researches much of our knowledge of the subject is due. Lead is attacked at the rate of about seventy-five pounds pet ampere per year, so that the result is extremely LEAD CABLE FIG. 30. marked. Fig. 30 gives a diagrammatic view of the circuit through such a cable. Part of the current used on the railway circuit passes from the rails to the cable and thence along it to the neighborhood of the motors, where it passes back to the track and the moving cars. The mischief is done at this point and not while the current is flowing in the cable. The effect produced is a severe corrosion of the leacj covering of the cable taking place irregularly upon the surface and forming pits, which may penetrate the sheath and destroy the insulation of the cable. Investigation showed the state of things on the Boston system to be very interesting. At the time, the positive poles of the dynamos in the power station were connected with the rails so that the current passed into them and RETURN CIRCUIT. 43 thence to the pipes and cables, emerging from them at vari- ous points in the system. The corrosion was thus widely distributed, but from local conditions of conductivity was most apparent in spots. Careful measurements of the potential between the track and the cables were made in a large number of places with the result shown in the map (Fig. 31). Near the power stations the flow was from track to cables, but over the main area of the city it was from cables to track, giving a large area in which corrosion might be expected. Differences of potential as high as EAST BOSTON FIG 31. five volts were observed, while experiments in other cities have shown as much as twenty-five volts. It is interest- ing to note that one of the first experiments tried to re- lieve this' electrolytic action was to sink in the earth ground plates connected to the cables in the hope that the current flow would take place mainly through them. The potential differences even at points quite near these plates were practically unchanged, showing very plainly the intense badness of the earth as a conductor, which has already been pointed out. The method of treatment which proved most effective in reducing the electrolytic effects, was first to locate the 44 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. trouble as nearly as practicable in definite areas and then to check it in these areas. In the first place the dynamo connections were reversed so that the stray current would enter pipes and cables over the most of the system/but would leave them en route for the negative terminal of the dynamo only in the districts immediately surrounding the power houses. Thus it would be certain that the damage would be limited to known areas which could be attacked locally with success, instead of being scattered where the trouble would be hard to locate and -harder to remedy. Kven within these areas conduction and consequent elec- trolysis is likely to be very irregularly distributed, so that serious trouble may occur at one point when points near by are apparently unaffected. FIG. 32. Fig. 32 shows the result of this change. The ' ' danger areas ' ' shown here as before by shading on the map, are comparatively small, although within them the differences of potential were quite as great as before. Now the prob- lem was to lead the current back to the dynamo without compelling it to leave the cables, and corrode them at the points of exit. To this end, large copper conductors were extended through the danger area and thoroughly con- nected at intervals to the telephone cables. The result THE RETURN CIRCUIT. 45 was excellent, since the stray currents, instead of passing from the cables through the earth to the track, took the easier path through the supplementary conductors. A measurement of the current thus collected from the telephone cables into a main ground wire from the station showed over 500 amperes capable, if flowing continuously, of eating away 37,500 Ibs. of lead per year. And as this current did not include that which found its way to waiter and gas pipes, the real amount of current which left the rails and wandered home through underground conductors was considerably larger than the figure mentioned, prob- ably several times as great. The distribution of this cur- rent is so irregular from place to place, as indicated on the map, that it would be very hard indeed to estimate the total proportion it bears to the whole current on the system. So far as data are available however they indicate that we would not be wide of the truth in saying that ten to twenty per cent of the current on the system may follow other paths than that through the rails and bonds. Even more than this may appear in occasional instances. So while the earth helps the return circuit directly but little, buried conductors may help very materially, perhaps to their own serious detriment. It should be remembered that the elec- trolytic action is not necessarily proportional to the differ- ences of potential such as are noted on the maps. The places most injured depend on local conductivity and some of the worst instances recorded have occurred where the measured potential difference was only one or two volts. Figs. 33 and 34 give a graphic idea of the kind of damage that is done to pipes by electrolysis from stray currents. Fig. 33 shows the effect of corrosion on an iron gas pipe, and Fig. 34 that on a lead water pipe. Both are from photographs of the ' ' horrible examples. ' ' As the action tends to become concentrated in spots, a pipe may be perforated in a rather short time. Iron water pipe has some- times been riddled in five to eight months. That this is easily possible may be readily seen, for suppose that con- ditions are such as to get in a certain spot a flow of half 46 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. an ampere in a space of one square foDt. Suppose the pipe to be $/% in. thick, therefore weighing about twenty-five pounds per square foot of surface. If the electrolytic action were perfectly uniform the pipe would be reduced to an unsub- stantial shell in a single year, and since the corrosion al- ways shows irregular pits the pipe would almost infallibly be perforated in six months. Very curious differences FIG. 33, exist between electrolytic actions in various situations, depending on the chemical conditions in the soil. Some- times the action produces a thick dense coating of ordinary rust which almost suspends the electrolytic process, while FIG. 34, elsewhere the products of decomposition are more soluble, and the work goes on until the iron is eaten away leaving a mere shell of the contained carbon and electrolytic debris generally. It is worth while to note that surface protection of pipes by painting with asphalt and the like has been shown by the Boston experience to be practically worthless, as the corrosion seems to work under the film, which can never be made really insulating to any useful extent. THE RETURN CIRCUIT. 47 In spite of the quite perceptible assistance that may be rendered by underground pipes to the general conduc- tivity of the return system, every effort should be made to avoid it. For, even if the various lines of pipe are protect- ed by the supplementary wire method described, there may be electrical differences at the joints of the pipes quite sufficient to cause local corrosion in serious amount. Joints in water pipe are better mechanically than electrically and the currents flowing through them may, as we have seen, be rather heavy. Take for example Fig. 35. Suppose that owing to oxidized and dirty surface of contact the joint A has a resistance of .005 ohm and that a current of one hundred amperes is flowing through it in the direction indicated by the arrow. The fall of potential through the joint would then be .5 volt, lines of current flow would be set up as shown by the dotted lines and a ring of corrosion B C would be set up on the positive side of the joint. Half a volt is quite enough to do the work, and though the action might be slow it would be sure. In point ///-" of fact the lead calked joints used in w r ater pipe may readily show a resistance ten or twenty times that just assumed, sometimes even an ohm or more, a ^^y case still more serious. p IG> 35< Therefore all conduction by pipes ought to be avoided as faf as possible unless they are electrically continuous. Even if they are, protection by supplementary wires is somewhat risky since while it may relieve trouble in the conductors so connected it may enhance the danger to neighboring pipes not thus protected. Joints between pipes of different materials are espe- cially dangerous, for instance between cast iron and cement lined sheet iron. Under exceptionally unfavorable condi- tions joints have been eaten out in as short a time as six weeks. 48 POWER DISTRIBUTION FOR ELECTRIC RAILROADS Liberal use of supplementary wires has great use as an emergency measure, applied to systems already existing, but here, as generally, an ounce of prevention is worth a pound of cure. The proper return circuit of the railway should be made so good that the stray currents shall be quite negligible, and all methods of palliating their evil effects should be considered secondary in importance and to be shunned rather than courted. It must not be un- derstood that these methods are condemned, for they may be of much use, but they should be employed only to deal with the residual currents after they have been reduced to the lowest practicable terms by means of improving the track circuit. The main point of such improvement lies in the con- nections between rail and rail. If the resistance of the bonds and their contacts were negligible there would be very trifling stray currents. For example, if we are dealing with a double track of ninety- pound rail, the resistance is about -g^ ohm per thousand feet or .0087 ohm per mile. Such a structure could carry 1000 amperes with a loss of but 8.7 volts per mile and should reduce the stray currents to a very minute percentage since the resistance is not only very small compared with any probable value of the earth resistance between track and pipes, but also very small compared with the resistance of the pipes themselves in- cluding their bad joints. With, say, one per cent of the current in the earth conductors the electrolytic action, while not absolutely suppressed, would be so slow and so trifling as to be scarcely worth considering save at a few points which could be protected if necessary. All this points to the necessity of the most perfect bonding, as before pointed out. All sorts of devices have been tried. Two of the most ingenious, aside from those already referred to, consist respectively of a plastic con- ducting film squeezed between the bond surface and the rail surface, and of a heavy copper dowel pin driven into a hole in the end of one rail and the other rail forced upon it and THE RETURN CIRCUIT. 49 held with the fishplate. The uncertain point about these as about many other bonds is their ability to endure jarring and corrosion. Bonds are sometimes subject to the sam? sort of electrolytic action just mentioned in connection witb pipe joints. Lately many bonds have been electrically brazed to the rails by a process closely akin to electric welding. The amount of power required is only 15 to 20 K. w. and in point of low resistance and permanence the result is exceedingly good. The most radical cure for joint resistance of rails may be found in the two now familiar processes for making continuous rails. That a continuous rail is entirely feasi- ble mechanically now admits of no dispute. Expansion does not and cannot take place longitudinally when rails are firmly embedded in paving, even under the extremes of temperature encountered. Whatever yielding there is, is lateral, and the track is not thrown out of line. The electrically welded joint when carefully made is strong and reliable and of almost infinitely small resistance. The contact is non-corrodible, of great surface and so in- timate as not sensibly to increase the resistance of the track. It is as far superior to a bond contact as the latter is to the contacts made through rusty fishplates. A track so excellent mechanically and electrically needs no com- mendation here, more than to reiterate the value of a com- plete and permanent connection between rails. Unfortu- nately the simplest form of joint which has shown ample strength is the butt welded form which requires energy to the amount of 200 H. p. or more, a quantity not often readily attainable. Recently a very good and reliable form of joint has been made by welding on a pair of fish plates at each joint the union not being over the whole surface, but at three large and heavy bosses so distributed as to make a solid and rigid joint This form of weld takes much less current than a butt weld and is amply strong. The ' ' cast welded ' ' joint has now come into very con- siderable use Mechanically it is superior, but electrically 50 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. it is scarcely the equivalent of the welded joint. Between these two rival continuous rail processes it is difficult to choose Certainly both afford at once the solution for the joint alignment and the bonding difficulties. The ' ' cast welded ' joint is by far the more widely used on account of its great mechanical strength and the ease with which it is made. Both are likely to come into very extensive use in large city roads where the electrolytic troubles are usually most noticable, although small roads are not exempt from them. The resistance of a cast welded joint, although not uniformly negligible, is about the same as that of the very best bonded joints and is quite as permanent. It has often been urged that a double trolley system should be employed to avert danger of electrolytic action. Experience has shown that the double trolley is not likely to become a favorite with street railway men. It can be worked successfully with proper care, but the mechanical difficulties in the way of installing and keeping up the overhead system of frogs, crossings and the like are some- what formidable. On a straightaway road with no branches or few the task is easier, but for the purpose in hand such roads are not the ones requiring the most serious consideration. The troubles belong especially to compli- cated city systems in which the difficulties of a double trolley system are something terrific. Inasmuch as every electric railway company has to pay for what can be made a magnificent return circuit, it seems totally needless to throw away the rails and operate a double metallic circuit overhead. Especially is this true in view of the fact, that considerations of track stability and durability point to the use of the continuous rail which minimizes at the same time the electrical difficulties. It must be remembered that in long distance lines such as are found in interurban and similar work, the use of continuous rails is liable to cause trouble from insufficient resistance to expansion, as such roads generally are exposed THE RETURN CIRCUIT. 51 to more violent changes of temperature. On the other hand, in the case of such roads trouble from- electrolytic action is usually relatively small or entirely absent, so that bonding is sufficient. Also as will be explained later, in these roads for heavy service and rather high speed there may sometimes be good reason for using two trolleys, quite aside from all questions of good return. Of course, when the alternating current motor is thor- oughly developed for railway service much of the danger of electrolysis will be escaped, whatever the character of the return circuit, but there will still exist every reason for making the rail return as perfect as possible from motives of economy alone. For when bad bonding can increase the total resistance of the track circuit ten or a dozen times, as has happened many times, the waste of energy due to the increased drop in the circuit is burdensome. For example, take a single track of ninety pound rail 10,000 ft. long. With continuous rails the resistance per thousand feet would be ^ J 7 of an ohm and for the whole distance .033. With 200 amperes flowing, the drop would be 6.6 volts and the loss of energy more than one kilowatt. Now suppose each bond contact with its half of the bond wire to have a resistance of .001 ohm. On each line of rail there would be 660 of these so that the total bond re- sistance of the track would be .33 ohm and the drop due to this bond resistance with a current of 200 amperes would be 66 volts. The corresponding loss of energy would be 13.2 k. w. more than enough to operate an extra car. At the cost of power generally found this waste would represent in the vicinity of $1000 per year net loss, a pretty high price to pay for the privilege of having a poorly connected track, liable to cause serious trouble from stray currents. And this instance represents not at all an ex- tremely bad case, but a very common one. The moral of all this is that just as much care should be spent on the joints underground as on those overhead, in fact more, since the latter are but slightly liable to cor- rcsion while the former run great risk of it. For this 52 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. reason the continuous rail is doubly desirable since it not only avoids constant loss of energy in the rail joints, but averts a rather heavy cost of maintenance. With continu- ous rails some cross bonding may be desirable to give se- curity against breaks, but it comes into use only in emer- gencies. -Next to the continuous rail the best construction employs rails of some of the recent deep sections, rolled in 60 ft. lengths. These are laid with long fish plates at the joints secured with twelve heavy bolts, and are double bonded at each joint. A track so constructed has only half the usual number of joints, thus halving the usual resistance due to the bonding. These long rails are rather un wieldly as they weigh 1800 to 2000 Ibs. each, but their use is very advantageous. To prevent electrolytic destruction of neighboring con- ductors by stray current from the rails the best simple ad- vice that can be given is as folV>ws: 1 . Use the continuous rail system ; or 2. Bond very thoroughly; put the positive pole of the dynamo on the overhead line; join the negative directly to the track without intentional earth connection, and 3. In any case investigate the potential between track and buried conductors and run supplementary wires from these conductors to the dynamo if necessary. This applies to small systems as well as large. The only cases which may be fairly excepted are electric roads running through country where there are no buried con- ductors near, and elevated roads which are really a special case of the double trolley system. As electric railways have become more common and more thoroughly under- stood the conditions of the return circuit have been much ameliorated, but sins against Ohm's law are still distress- ingly common. A feeling still seems to be rife that what is concealed from the eye may be scamped, as when the guileful wiring contractor runs underwriters' wire through the ceilings and puts okonite at the joints. It is bad enough for a dishonest contractor to do that sort of thing, but what shall we say of a man who cheats himself by RETURN CIRCUIT. 53 doing poor work on his return circuit without even the ex- cuse of economy. We are now in a position to determine the quantity which was the ultimate object of this investigation into the details of the return circuit; i.e., its total net value as a conductor compared with the outgoing circuit. This is obviously not a fixed quantity in either abso- lute or relative value, for even neglecting joint resistances there is far less difference between the weights of the rail used in various systems than between the weights of over- head copper. An ordinary electric road uses perhaps a rail of seventy pounds per yard. A single track so constituted is, neglecting joints, of conductivity equal to 2,200,000 c. m. of copper. If the rails were continuous it is clear enough that in a road of small or moderate size they would be perhaps ten times as good a conductor as the overhead system. This would allow for a No. o trolley wire and a No. oo main feeder on the average all over the line. On the other hand, taking the resistance of bonds and joints as double that of the rail itself, the equivalent of the rail in copper falls to, say, 733,000 c. m., which is less than four times the overhead system just assumed. If this system averaged a No. ooo feeder, plus the trolley wire, it would have almost exactly three times the resistance of the track circuit. In large systems the rails often run as high as ninety pounds per yard, so that a single track would be equal to 3,000,000 c. m. of copper. With continuous rails this full equivalent could be taken, but the feeder area plus a No. oo trolley wire would hardly be less than 750,- ooo c. m., so that the resistance of the overhead wiring would be about four times that of the track. More com- monly, making the same allowance for bonds as before, the track equivalent would be 1,200,000 c. m. and the trolley and feeder copper would have only about one and a half times the track resistance. Not infrequently the bonding is imperfect enough to reduce the track equivalent to 900,- ooo c. m., which would frequently be equaled or ex- 54 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. ceeded by the trolley and feeder copper, raising the ratio to equality. A double track, of course, improves matters. We may tabulate these results somewhat as follows, calling R 1 the track resistance and R the overhead resistance. R 1 = .1 to .2 R. Exceedingly good track and very light load. Ri = .2 to Ri == .2 to 3 R. Good track and moderate load. ,6 R. Fair track, moderate load. 3 R. Exceptional track and large sys- tem. Ri == .3 to .7 R. RI = .7 to i.o R. Good track, large system. Poor track, large system. In cases now somewhat exceptional the track resist- ance may exceed the overhead resistance considerably. The Drop-Volts 8 "v ^ *X ^v ^ Vs "V,, X, o t , *s ^ tf ^. ^ f ^ X -V ^ ^N ^ ^X V^ Return circuit FIG. 36. Street Railway Journal assumption now frequently made, that the track resistance is one- quarter that of the overhead system really repre- sents a better state of things than usually exists. To justify it requires the combination of continuous rail or exceptionally perfect bonding, with conditions of load that do not require large feeder capacity. Under the ordinary conditions R 1 = .4Ris probably nearer the truth. The proportion between R and R 1 has, of course, a very im- portant bearing on the design of the overhead system. If the return circuit had no resistance then the entire drop THE RETURN CIRCUIT. 55 \vould take place in the overhead conductors and we could calculate the line for any given drop by the simple formula e . m . = with D for the linear single distance. Bearing in mind however the resistance of the return circuit, it is evident that for a given total loss in volts more copper must be placed overhead than would be necessary if the return cir- cuit were of zero resistance. In other words, if we are confronted by a considerable loss in this return circuit it is necessary to have proportionately less elsewhere in the 100 treet Railway Journal 37- circuit. With no resistance in the return circuit the drop in voltage may be represented graphically by Fig. 36. Here the whole drop is in the outgoing circuit which can consequently be rather small. If, on the other hand, we take the actual case in which the return circuit has a very perceptible resistance, the distribution of the drop will be as in Fig. 37, which is given by R 1 = .43 R. This means that to preserve the same conditions of total loss in the circuit the overhead copper must be increased by forty- three per cent, since of the total 100 volts to be lost it is now permissible to lose but 70+ in the outgoing circuit. Hence to take account of loss in the return circuit the formula just given must be Altered by changing the co-i- 56 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. slant in accordance with the new conditions, which are there actually found in practice. The proper amount of increase in the constant is a little uncertain as is indicated by the table just given. For R 1 = .4 R however the con- stant is 14.4 so that we may rewrite the copper formula as follows: 14.4 CD c. m. = I r - . In the vast majority of cases the constant will lie be- tween 14 and 15. The exact value to be assumed depends on the conditions as to track circuit and load in the par- ticular case considered, and can be judged approximately from the table. It may sometimes be desirable to make a few trial calculations with different constants in order to get a clear idea of the possible amount of copper. It is, of course, possible to determine a condition for minimum cost of the conducting system, taking account of the cost of copper, rails and bonding, but, generally speak- ing, the rail is fixed by purely mechanical considerations while there are, as has been shown, good reasons for making the track circuit thoroughly good. In applying the above formula, as we shall in the next chapter, it should be re- membered that in extensive systems the constant may have to be modified in passing from one locality to another, for the rail conditions will probably vary and the load condi- tions most assuredly will change. In cases where the track return is not used, as in double trolley and conduit roads, the outgoing and return leads may or may not be duplicates of each other. If the total drop were equally divided between them the feeder formula would of course become the familiar E and the return would^ have the same area and total weight as the feeder system thus determined. Ordinarily there would be little advantage in making the two sides of the circuit equal and the designer would be guided mainly by THE RETURN CIRCUIT. 57 2000 1200 600 200 REA OF COPPER REQUIRED FOR DELIVERING 100 AMPERES UP TO 30000' ) 1000 AMPERES UP TO 3000') AT 10 TO 50 VOLTS LOSS. UPPER SCALE EADS TO 3000'BY 50 r PER SQUARE. LOWER READS TO 30000 BY 500'PER SQUARE 15 DISTANCE PLATE II. 58 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. convenience. Often the easiest procedure mechanically is to install one or more heavy return cables for o predeterm- ined fraction of the total drop and to compute the feeder system precisely as if dealing with a track return. For the greatest economy in copper a particularly careful study of the probable distribution of the load should be made. Of course, one may divide the drop between feeder system and return in almost any convenient way, subject to the limitation imposed by danger of overheating, without COPPER EQUIVALENT OF STEEL RAILS FOR SPECIFIC CONDUCTIVITIES FROM 9 TO 15?b OF THAT OF COPPER POUNDS PER YARD PLATE III. .120 affecting the economy of the distribution, but when one deals with a track return of uniform section which must be installed and paid for anyhow, there is less need of refinement than in using costly cables. Probably the best method of design in these cases is to follow the general procedure to be found in the subse- quent chapters, but with close attention to the limits and variations of load in the various sections of the line, not adhering closely to anything like a track constant, but taking the data for feeders and return out of Plate II with such division of the total drop between them as seems expedient from the standpoint of simplicity in overhead or THE RETURN CIRCUIT. 59 conduit construction. Plate II is merely an extension of Plate I, p. 7, arranged with reference to heavy work of this class, the abscissae being the total lengths of the wire under consideration and not the lengths of the circuits as in Plate I. In case the working conductors are of other material than copper they should be reduced to the equivalent section of copper. For this purpose Plate III, developed from Fig. 17, p. 30, will be found convenient in all com- putations involving rails or other iron or steel conductors. Third rail systems with ordinary track return may or may not involve supplementary feeders. Plates II and III will enable these cases to be easily computed, once the loads are determined. CHAPTER III. DIRECT FEEDING SYSTEMS. By direct feeding is meant the supply of current to the working system of conductors from a single central sta- tion, without any intermediary apparatus. It is the system employed on most present electric street railroads, save a few of the largest size. It is ordinarily used on interurban lines and would be universally applied were there not many cases in which the distribution of power from a single station becomes uneconomical at any practicable voltage on account of the great distances involved. Nearly all interurban lines, and especially the systems which are likely to result from the conversion of steam into electric lines, can be best operated by other means which will be described in subsequent chapters. Indeed a care- ful examination of very many existing electric railways will disclose the fact that direct feeding is being worked far beyond its proper limits of application and is the cause of serious pecuniary loss, both in interest on a huge invest- ment in copper and in power needlessly lost on the line. Direct feeding however is properly applied in most in- stances, and must be ultimately applied as the distributing system almost universally, since even where substations are employed the lines proceeding from them are often a case of direct feeding and must be treated as such. Electric railway feeding systems are akin in principle to those employed in simple cases of distribution for light- ing, and yet in practice differ from them very radically in certain particulars. Railway feeders are not generally de- signed to preserve uniform voltage within the area fed, but to hold the voltage, admittedly variable, within certain rather wide, but fixed limits. Lighting feeders must be de- signed with reference to a load varying in the same area DIRECT FEEDING SYSTEMS. 6 T from time to time, but yet closely confined to that area; railway feeders must be so designed as to meet not only a load variable in amount from second to second, but shifting from place to place obedient to causes that follow no definite law. On the other hand not only are railway feeders absolved from the necessity of holding the voltage closely uniform, but by virtue of this they can the more easily be arranged to meet extreme shifting of the load. In early electric railways the trolley wire proper was rather small and the feeding was often relatively quite as complex as that in large modern systems. The conditions which must be met in planning a direct feeding system are roughly as follows: 1. The maximum fall in voltage at any point in the system under all working conditions must not exceed a fixed amount. 2. The average drop throughout the system under normal conditions must equal a certain predetermined amount. 3. The feeders must be so connected that accidents to the working conductors shall interfere with traffic to as small an extent as possible. To meet these various conditions a large number of arrangements of feeders have been devised, many of which are in extensive use. The following are some of the most usual, which have stood the test of experience. i . The so-called ladder system shown in Fig. 38. Here one pole of the dynamo is earthed as usual and the other is connected to the trolley wire C D, and also to the feeder A B. These are connected at intervals of a few hundred feet by subfeeders a, b, c, d, e,f, etc., which are generally hardly more than tie wires uniting the principal feeder to the trolley wire. This arrangement was very common in early electric roads. It made possible the use of a very slender trolley wire merely large enough to carry conven- iently the current for cars running between the subfeeders, and made the system tolerably free from interruption by accidents to the trolley wire, which from its small size was 02 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. rather prone to break. Both the trolley wire and the principal feeder are continuous and of uniform cross sec- tion. This continuity is useful in case of the crowding of cars at one or more points on the line since it brings to the rescue the full conductivity of the system. It is bad how- ever in case of short circuits in that the main circuit breaker at the station is quite likely to open and stop every car on the line. As a real feeding system it hardly deserves the name, since electrically it is nothing more than a continuous working conductor of uniform area. The properties of such a conductor have already been fully considered in Chap. i. The only additional fact that has to be taken into account in the ladder system is the limited conductivity of the trolley wire between the subfeeders. The drop in voltage at a car located at any point is practically the drop Street Ry. Journal D FIG. 38. in the principal feeder up to that point plus the drop in the trolley wire from the car to the nearest subfeeders, which are virtually in parallel, inasmuch as current flows into the trolley in both directions along the trolley wire. 2. A system similar in some respects to Fig. 38 is shown in Fig. 39. Here there is as before a principal feeder A B. The trolley wire C D is not however contin- uous, but is broken by insulating joints into separate sections of approximately equal length each with its own subfeeder a , b, c, etc. The added conductivity of the con- tinuous trolley wire is, of course, sacrificed by this arrange- ment. Both the trolley and feeder are generally of uniform area throughout their respective lengths and the system is electrically, to all intents and purposes, a uniform linear conductor save for the abrupt change in conductivity in passing from the principal feeder to any subfeeder and its section of trolley wire. As regards a load at any poin- DIRECT FEEDING SYSTEMS. 63 the total drop is that in the principal feeder up to the sub- feeder controlling the section in question plus the drop in the subf eeder and the trolley wire up to the load. The advantage gained by cutting the trolley wire into short, independent sections is a certain amount of immunity from breakdowns. The subfeeders a y b, c y etc., are usually provided with fuses or switches or both, so that while in case of a break in the trolley wire the cars on the adjacent sections are not deprived of current any more than in the ladder system, there is no longer the danger of stopping traffic by blowing fuses at the station, since the subf eeder fuse immediately acts to stop an excessive flow of current. In addition, in case of fire or flood affecting any part of the system, the disturbed region can be very promptly isolated by opening the circuit at the subfeeders. In cities o Street Ry. Journal FIG. 39. where fires are of frequent occurrence such an arrangement is highly necessary, although it is generally desirable to use a far more complete feeding system in connection with it. Both the arrangements just shown are entirely without special provisions for holding up the voltage at distant parts of the line, depending practically on the conductivity of the principal feeder. 3. A true feeding system corresponding in a general way with Fig. 38 is shown in Fig. 40. Here A JB is the trolley wire while in multiple with it are feed wires tapped into the trolley wire at a, b and c. These feeders are generally quite independent of each other up to their respective junctions with the trolley wire. A load at any point, as d, receives its current in both directions through the trolley wire, which in turn draws^current from the ad- jacent feeders. The conductivity available at the load d is that of the trolley wire from A to d, reinforced by the feed- ers a and b\ in parallel with that of the trolley wire section 64 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. from^to c and the feeder c. With the arrangement of Fig. 40 it is quite possible to hold the voltage fairly uniform by giving sufficient area to the longer feeders. As a matter of convenience, to avoid the undue multiplication of wires, the distances A a, ad, etc., between feeders are made consider- ably longer than in the ladder system: hence the trolley wire is generally larger. Of course, it must be large enough to avoid excessive drop in the sections b d and c d when load is applied at d. As a rule the distances A a, ad, etc., are several thousand feet except where the traffic is very heavy. With No. o or No. oo trolley wire the distance named is not generally excessive. As compared with the ladder distribution this one has the great advantage of giv- ing a fairly uniform voltage, and can be more readily ar- ranged to handle abnormal loads at distant parts of the -B , Street Ry. Journal FIG. 40. line. It has also the same convenient property of giving current to each car from two directions so as to minimize the effect of breaks in the trolley wire. It is however ex- posed to trouble in case of serious short circuits, and is in- convenient in the matter of cutting out portions to execute considerable changes in wiring or to avert accident. 4. An obvious modification of the arrangement just mentioned is that shown in Fig. 41. This bears the same relation to (3) that (2) does to (i). It shares with (3) the advantage of maintaining fairly constant voltage under normal conditions, though it is somewhat at a disadvantage in case of a heavy load on a distant section, since that sec- tion must depend on its own feeder alone without assist- ance from adjacent sections. The feeders a, b, c, etc., are provided with individual switches and cut-outs at the station so that if a short circuit occurs nothing worse can happen DIRECT FEEDING SYSTEMS. 65 than the temporary disabling of that particular section, while if necessity demands any section can be promptly cut out of circuit in case of fire along the line or any other sufficient cause. (4) is very well adapted for use on long lines with fairly regular traffic. L,ike (3) it requires a rather heavy trolley wire for the best results. A load at any point is supplied by the feeder for that section in series with the trolley wire between the load and the feeder junction, so that the drop under any given conditions is very readily computed. In both (3) and (4) it is sometimes convenient to tie two or more feeders together, as shown by the dotted line at d (Fig. 41). This procedure reinforces the conduc- tivity with reference to the section thus connected, as b } and while it may lower the voltage of sections beyond the 1 d f . r 1 A A a b c Street Ry. Journal Fic. 41. link, is very useful when a particular section is exposed to severe loads from grades or massing of cars, particularly since such linking can be applied at any time that the service may require it. In very many cases it is advantageous to install a com- posite feeding system which can be made in a considerable measure to unite the advantages of those already described. A very useful combination is that shown in Fig. 42. Here the trolley wire, AB, is cut into sections of vary- ing length, short where considerable danger of interruption of service exists, long where longer sections can be more con- veniently utilized. C is a principal feeder as in the ladder system connected at a and b to a continuous trolley line, and at c, d and e to trolley sections. This principal feeder is reinforced by feeders K and F to equalize the voltage more perfectly in the region of dense traffic, while the inde- 66 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. pendent feeeders,G and H, supply the long isolated sections, /and^-. G and H are moreover linked at f if the condi- tions of service require. Fig. 42 represents the actual arrangement of an extensive feeding system much more closely than any of the simpler arrangements shown. As a matter of fact such a complex system is generally the out- growth of the conditions which develop in service rather than the result of deliberate forethought. Nevertheless, good engineering often demands the adoption of such ap- parently complex methods. In general, independent feeders are necessary to pre- serve good working pressure in outlying districts where comparatively independent lines are worked, while in re- L Street Ry. J< I urnal | c 1 1 6 ' ' J 1 1 1 C (f. e . j V a 6 FIG. 42. gions of dense traffic the tendency is to link together the principal feeders of neighboring lines into a network rein- forced by special feeders wherever necessary. The trolley wire is sectionalized only in so far as danger from fires and electrical troubles require. Although a continuous trolley wire is now far less necessary than formerly on account of improved methods of construction, on the other hand an extensive subdivision into sections hinders the full use of all the copper installed and increases the danger of local stoppage of traffic. On any railway system, street or other, continuity of service is of the first importance, both by reason of the direct loss from suspension of traffic and the indirect, but far more serious, loss of public confidence and 'goodwill. Consequently it is often advisable to take chances in order to keep running, and linking feeders and trolley into a continuous system to drive through a time of short cir- DIRECT FEEDING SYSTEMS. 67 cuit if possible rather than shut down part of the system. The present tendency is to make the various sections of feeders and trolley wire separable rather than separate, so that they can be cut apart when absolutely necessary, but not long before that crisis. IvOng lines, interurban and the like, may often be best treated indirectly through substations, but when direct feeding is employed, it is ordinarily best to use a very sub- stantial trolley wire, not smaller than No. oo, installed in separable but not disconnected sections, and supplied with current by separate feeders, which may be linked if local conditions require. If large power units are to be em- ployed, requiring large currents, it is better to use a very large trolley wire than to install a principal feeder, since with large currents the larger the contact surface of the working conductor the better, and the conductivity of the trolley wire can be relieved if insufficient by connecting each section to its feeder in several places instead of one. There is no reasori however why, on large work such as is found in converting* steam roads to electric, the working conductor may not have a cross section equivalent to No. oooo wire or more which enables comparatively long sections between feeders to be employed with advantage. For ex- ample, suppose a No. oooo trolley wire carrying a current of 200 amperes per section received equally from the two adjacent feeders. This condition would be met by a train requiring one hundred kilowatts to drive and located mid- way between two feeders. Allowing no more than two per cent loss, i. e. , about ten volts in the trolley wire be- tween feeder junction and load and substituting the above values in the fundamental equation c. m. II , the distance between feeders should be about 4000 ft. Inas- much as the average drop produced by the moving train, with a maximum of two per cent midway between feeders, would be but one per cent, it would generally be advisable to increase this amount. Allowing an average drop of two per cent in the trolley wire, i. e. , a maximum of four per 68 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. cent, the proper distance between feeders would be virtu- ally doubled, rising to about 8000 ft. a mile and a half. For long roads, then, one may use with advantage such an arrangement of feeders as is shown in Fir;. 43. Here a continuous heavy trolley wire is divided into sections of, say, a mile to a mile and a half in length, each with a junction to the feeding system. This, as shown, consists of three main feeders, each supplying two sections of trolley wire. The number of these main feeders and the number of sections each supplies is regulated by con- venience and local conditions, as is too the length of each section. The sketch (Fig. 43) shows merely the principle, which is well suited to roads up to a dozen miles in length fed from somewhere near the middle. Such roads are apt Street Ry. Journal FIG. 43- * to require rather large units of loads, due to well loaded trains and high speed, but the number of trains to be oper- ated at any one time is usually small. A rather nice question sometimes arises as to the relative cross section of copper to be put in the trolley wire and in the feeders. In the large work that we are just now considering, the trolley wire must be in any event large enough to give sufficient contact with the trolley. And this is apt to indicate about as large a working conductor as can conveniently and se- curely be supported. Therefore the feeders will be rela- tively smaller than in ordinary street railway practice, and it is not advantageous to separate permanently the sections of trolley wire, thus throwing away the conductivity of its large cross section. Whenever double tracks are used it goes quite without saying that the whole system of con- ductors should be united, each trolley wire serving as a feeder to the other. DIRECT FEEDING SYSTEMS. 69 Occasionally, too, on single track roads with frequent turnouts, two trolley wires are strung" ten or twelve inches apart, each to accommodate the cars running in one direc- tion, so as to entirely avoid overhead switches of any kind. This arrangement is shown in Fig. 44, and while it is not now very widely used, it is exceedingly convenient in cer- tain cases. In Fig. 44 the track at a turnout is shown by the solid lines and the two trolley wires by dotted lines. The trolley wire, A B, would naturally be used by cars run- ning from right to left as indicated by the arrow, while C D would be used by cars running from left to right. Each car keeps to its own trolley wire throughout the track, un- less it is necessary to change over in backing around a turnout. This double trolley device enables long exten- sions to be handled without feeders. - o Street Ry. Journal FIG. 44. Before passing to the actual computation of a trolley and feeder system, we must go back to our two f funda- mental propositions and inquire into the permissible maxi- mum drop and what we mean by average drop. Suppose that ten per cent average drop has been de- cided upon in a given case, What is really meant by this? There has been considerable confusion on this point. Are we to understand that this average drop is that determined from the effect of the maximum working load throughout the system, or is it the average loss on the parts of the sys- tem considered separately irrespective of their relative amounts. Is it the drop produced by the average load or the average of the drops produced by the simultaneous loads at some particular time? To reduce the matter to a common basis with other cases of the electrical transmission of energy, we are at lib- 70 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. erty to put but one interpretation upon average drop. By it we should mean in every case that a certain specified proportion of the energy delivered to the line during a par- ticular period is to be lost in the transmission. On this basis we can design the system for conditions of maximum econ- omy, knowing approximately the probable cost of energy per kilowatt hour and the price of copper. Starting with this definition, we can then intelligently work out the re- lation of this average energy loss to the loss in volts at the various parts of the system. It is necessary however to bear in mind, first, that the same conditions of economy with respect to loss in transmission do not necessarily hold for all parts of a given system, and second, the question of economy in transmission is quite subordinate to that of successful operation. As regards the former consideration, the average energy delivered to an electric railway system is a very different thing from either the maximum energy or the average energy during the hours of heavy load. The load factor, i. e. , the ratio between average and maximum output on a railway system is generally rather unsatisfactory, as has al- ready been indicated. It ranges in general from .3 to .6, varying greatly with the size of the system, the character of the service and the habits of the people who ride. In cities many interesting facts appear from the load curve of an electric railway the movements of workingmen, the crowd of shoppers going downtown in the forenoon, the migration in the early afternoon, the homegoing at six and the theatre crowd an hour and a half later. All these factors of load operate with varying force, not only in different places, but in different parts of the same system. The changes from day to day are considerable, but on the whole the same line preserves its character remarkably well. The result of a varying load factor is a necessary limitation in the permissible loss of energy. For if we have a load factor of .3, the average loss of energy, what- ever economy of transmission may indicate must not be enough to cause at maximum load a drop in voltage suffi- DIRECT FEEDING SYSTEMS. 71 cient to interfere with the proper operation of the cars. If we write for the maximum permissible drop, V, v for the drop corresponding to the loss of energy for greatest econ- omy of transmission, for the load factor, I,, and for the drop assumed, V 1 , we have the following inequality which sets a limit of drop which must not be exceeded V^<^LV Very fortunately it usually happens that s;000 i bs , IOOO This amount is the same for both the two phase and single phase circuits, in the way usually employed for operating two phase circuits, i.e., a complete and independent circuit for each phase. Sometimes the two phases have a common wire which modifies the amount of copper required, but this method of interconnection is seldom used on a large scale, since on long lines and at high voltages it involves serious practical difficulties. The three phase system requires a special, though very simple, calculation for the line. As ordinarily installed the three phases are mutually interconnected, so that the line consists of only three wires. This combination of cir- cuits so utilizes the wire that for a given amount of energy delivered with a given maximum voltage between lines and at a given loss, the copper required is just seventy-five per cent of that necessary for an equivalent single phase line. This means that since the three phase line consists of three equal wires stretching from station to station, each of these wires must be of half the cross section needed for a single phase line wire under similar circumstances. If the single phase line consists of two wires each weighing 1000 Ibs. per mile, the three phase line will consist, for the same loss, of three wires each weighing 500 Ibs. per mile. There are, of course, divers ways of taking ac- count of this saving in the formulae, but the author has 176 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. found the following to be the most convenient and direct. Write in (i) for tions, for L. Then Write in (i) for C, and D, the distance between sta ' ... --- _ K being as before the loss in volts, while W is watts de- livered and V voltage of delivery. Applying this formula to the example just given we have ft m. = "X 50 X 100,000 = 1000 This is the area of each of the three wires. Similarly for the total weight we may modify (5) and multiply by 3, giv- ing for a close approximation the exceedingly simple form 100 Applying this to the case in hand we have w = 100X50X10,000 = lb 1000 A very simple formula for approximate cost is B wherein P is the total cost in dollars and p the current price of bare copper in cents per pound. These formulae for alternating transmission -circuits enable the economics of the matter to be investigated very rapidly. In the final design of the line it will usually be found, as in the case given, that the size of wire will fall be- tween two standard sizes. In this case, as a rule, select the nearest size and figure out the final amount of copper from the actual weight of this wire. If the excitation of the motor or rotary converter fields is properly adjusted no account need be taken of in TRANSMISSION OF POWER FOR SUBSTATIONS. 177 ductive drop, since the widest departure of the power factor from unity will not in any practical case be great enough to disturb the working voltage seriously. The only time at which inductance is much in evi- dence is during the periods of starting the motors or ro- tary converters. For the best results the generators should have good inherent regulation so that lagging current will not reduce the voltage seriously and it is well to raise the initial voltage a little at the time of starting. Rotary converters when thrown into action may assume either polarity, but a few tentative touches of the switch with small current will secure an K. M. F. in the right direc- tion or better, one may start them from direct current. Both, polyphase generators and rotary converters oper- ate well in parallel, behaving, in fact, much like continu- ous current generators, when they are once in adjustment. The process of throwing alternators in parallel is very simple if one remembers that the currents must be in phase as well as of the same voltage at the moment of con- nection. The former condition is determined by phase lamps, the latter by the voltmeters. For general transmission for railway work the voltage should generally be from 5000 to 10,000, more often the latter. In favorable climates even higher pressures may be safely employed. The best field for such power trans- mission is in cases of distribution over distances of fifteen miles and upwards under circumstances in which a specially favorable spot can be selected for the main generating station. When alternating motors can be conveniently em- ployed on the cars, transmission from a central station at high voltage may become the rule instead of the exception, for with power delivered to the working conductors from static transformers requiring no attention there will be less excuse for long and heavy feeders. In the next chapter we will consider the application of alternating motors to service on cars and the relation of this practice to the development of long distance electric lines. CHAPTER VII. ALTERNATING MOTORS FOR RAILWAY WORK. Avast amount of money, time, and ingenuity, has been spent in attempts to develop motors for alternating current good enough to replace continuous current motors in all their varied uses. These attempts have led to many fail- ures, but through them all we have come at the present time to a very gratifying measure of success. But rail- way service is on the whole the severest work to which any motor can be put, for it involves severe strains in starting, heavy loads on grades, constant and severe shocks and jarring, and exposure, usually, to dust and moisture. Beyond this a railway motor must be easily reversible, and must be able to work week in and week out without close attention or frequent overhauling. Until very recently these difficulties have deterred engineers from any serious attempts to put into use alter- nating motors, but the development of electric railway sys- tems into conditions that demand the methods and appar- atus of long distance power transmission has forced the alternating motor into consideration. We have just seen the nature of substation distribution for continuous current railway motors, and to tell the truth it leaves much to be desired. The losses of energy incurred in passing from alter- nating to continuous current are at best rather serious, the apparatus for the purpose is a very considerable item of expense and, what is worse, a substation with rotary con- verters requires constant attention, so that the cost of at- tendance, to say nothing of repairs and depreciation on sub- station equipment, makes transmitted power so expensive as to bar it from the general use which it finds when not necessarily distributed in the form of continuous current. ALTERNATING MOTORS FOR RAILWAY WORK. 179 Power transmission to rotary converter stations is there- fore under existing conditions of limited applicability, for purely financial reasons. With an available alternating motor for use on the cars the matter puts on a very different aspect. Reducing transformers would be placed at suitable intervals along the line, supplied with energy from high tension feeders and feeding the working conductors directly from their sec- ondaries. The rotary converters or equivalent machines, with the accompanying apparatus, the substation itself and all the attendance would be dispensed with. In addition, the energy lost in conversion to continuous current from ten to twenty per cent of the whole would be saved. As- suming one hundred kilowatts average output in the sub- station, working twenty hours per day, the actual saving would amount to not less than half a cent per kilowatt hour, $36.50 per kilowatt per year. The abolition of this charge for the conversion of energy to continuous current would make power distribution from a central station pay in a large number of cases where boosters or separate generating stations are now the most economical methods available. Furthermore it would make it possible to employ water power far more freely than is at present worth the while, and would give a particular impetus to long interurban and cross country lines now hampered by the heavy cost of transmitting the necessary power. Admirable as is this outlook we must not for a moment lose sight of the fact that before entering this promised land we must have an alternating motor substantially as efficient and durable as the present standard railway motors. It is not, however, necessary that there should be any striking similarity in appearance or in methods of opera- tion between the two types of motor, or even that the al- ternating motor should be suited to all conditions under which continuous current motors are now worked. Alter- nating and continuous currents have found for themselves 180 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. distinct fields of usefulness in electric lighting why not also in electric railroading ? Out of the motley throng of alternating motors four types are fairly possible for application to railway practice. Each is characterized by a combination of good and bad qualities somewhat difficult to evaluate in the present state of our knowledge of alternating railway work. We may tabulate the types in question as follows: I. Synchronous motors started by commutation. II. Synchronous motors started as induction motors. III. Asynchronous polyphase motors. IV, Asynchronous monophase motors. The first two classes have exceedingly valuable properties for certain purposes, but are not suited for railway work requiring very frequent stopping and starting or constant variation of speed. The third class can meet all requirements as to starting torque and speed variation, and can be made substantially as efficient and durable as continuous current motors, but requires a somewhat troublesome system of working con- ductors. The fourth class starts moderately well, is somewhat weak at present in the matter of speed variation, but can be operated on existing systems of working conductors. I. It is a well known fact that a series wound motor with fields laminated to check eddy currents will start and run fairly well on an alternating circuit, particularly if the frequency is low. The late Mr. Kickemeyer produced a motor of this class which gave admirable starting torque and ran with a good degree of efficiency. The practical difficulty that has hindered the commercial development of such motors is rather severe sparking, which seems to be irremediable and if long continued does serious damage to the commutator. If, however, the sparking only occurs during the pro- cess of starting it is not a difficult matter to avert injury to the commutator, so that if such a motor can be worked normally as a synchronous alternating machine, and as a OF TVIK UNIVERSITY ALTERNATING MOTORS FOR RAILWAY WORK. l8l series commutating motor only at starting, it become cap- able of doing excellent work. There are divers other means of starting an alternat- ing motor by means of. a commutator. A commutated field in shunt to the armature can be made to give a power of starting sufficient to bring an unloaded motor up to syn- chronous speed, and in fact, an ordinary compound wound alternator can be made self starting by means of its com- pounding commutator. These devices do not permit of starting under anything much exceeding friction load and, hence, are inferior for severe work to the series starting device just mentioned and various modifications of the same idea. II. Synchronous motors of the polyphase type are capable of starting fairly well as induction motors, the field poles serving as armature. When the starting torque is obtained merely from eddy currents in the pole pieces, as in most synchronous motors and rotary converters, the torque is weak and the starting current abnormally large. To secure a quarter of the full load running torque, fully twice the full load current would be ordinarily required, or proportionally less if the motor is starting under merely friction load. It is quite possible, however, to construct a specialized field with inductive windings in the pole faces, so that the the motor will give its full normal torque at starting on a current not greatly in excess of its full load current, and will be capable of shifting over to synchronous run- ning when up to speed. In a similar way a monophase motor could be arranged to be self starting as an induction motor and then trans- formed to the synchronous type. For starting under load these forms are probably in- ferior to those starting as series motors by commutation, out they are simpler and sufficient for starting unloaded. To ordinary street railway service with constant stop- ping and starting under all sorts of unfavorable conditions, these essentially synchronous motors are inapplicable, 1 82 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. since they do not start well enough and are incapable of speed variation when running in synchronism. Neverthe- less, they are not to be despised for certain classes of rail- way work to which we must look forward. For long lines with stops only at stated stations such motors can even now be made available. If starting by clutch be considered inadvisable there is now no serious difficulty in the way of a commutating start quite good enough to bring a train up to speed. Once in synchronism the motors would drive steadily ahead up grade and down at a uniform speed until the next station was reached. The longer the line and the fewer the stops the better would be the operation of the system. The great advantage in synchronous motors for such work lies in their freedom from lagging current, and their insensitiveness to changes of voltage. A power factor approaching unity such as can readily be obtained from large synchronous motors reduces the difficulties of trans- mission very materially, and particularly it diminishes the necessary capacity in the generating station and in the line. In general transmission plants for a mixed load of lights, synchronous and induction motors, the power factor can be kept fairly high, with careful operation prob- ably up to .85 or .90. This power factor means that for operation at a given voltage ten to fifteen per cent more current must be generated and transmitted than cor- responds to the energy delivered. In addition a similar amount of reserve voltage must be available to compensate for the inductive drop in the line and the reaction of the lagging current in the generators. The total net effect then, of even this power factor is to call for not less than twenty-five per cent extra capacity in the generating plant. Were it not for the fact that polyphase generators have a high output compared with continuous current generators, even this increase would be serious as it is it is annoying. In plants operating induction motors only, the increased capacity necessary by reason of lagging current may be very much more serious, ALTERNATING MOTORS FOR RAILWAY WORK. 183 and makes the synchronous motor a thing not lightly to be put aside as impracticable. III. Although the asynchronous polyphase motor is now not unfamiliar and its theory is fairly well known to most engineers, its practical characteristics are not widely understood. We may best regard it as an alternating motor in which the current is led into the armature by induction as FIG. 93. in an ordinary transformer instead of by brush contacts. Its field and armature windings are so organized that the currents in them bear to each other the relation necessary to secure effective torque, as in any other motor. Whether the windings which deliver current to the armature are used alternately for this purpose and for establishing a field with which the induced current can react, or whether in- ducing and field windings are specialized; whether the structure is so disposed that there is a true resultant rotary magnetization or whether there exists a rotary pole only in the sense in which the poles rotate in a continuous current 184 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. armature all these are questions which have but a trivial bearing on the actual properties of the machines. As a matter of fact induction motors are much closer in prin- ciples and properties to continuous current motors than is generally supposed. Like shunt motors they tend to run at a constant speed and when the load changes they speed up or slow down just enough to permit enough armature cur- rent to flow to adjust the motor to the new conditions of load. Like shunt motors too, they require at starting a re- sistance in the armature circuit to keep the starting cur- rent within bounds. Their general properties are very little influenced by the number of phases for which they are wound. There is supposed to be a slight increase of output with increase in the number of phases, but as in the case of multipolar continuous current machines the increased output is more a matter of finesse in design than it is dependent on any theoretical considerations. At the present time all polyphase induction motors are strikingly alike in structural features. With very few ex- ceptions they consist of two concentric annular masses of laminated iron, of which the inner one is supported on the shaft and is free to rotate, while the outer one is carried by the frame of the machine. The outer face of the inner ring and the inner face of the outer ring are provided with slots or holes to receive the windings. Fig. 93 shows the character and relation of these rings. The slots or holes are various in number and shape, but those in the two members are different in number to keep the magnetic re- lations constant irrespective of the position of the rotating member. The teeth are very seldom developed into any- thing approaching projecting pole pieces, unless in small motors, as it is desirable to distribute the windings as uni- formly as possible. In American motors, the slots are usu- ally open, in Kuropean types the} 7 are frequently closed as shown. Both rings are supported in a suitable frame. In one set of slots is wound the primary inducing winding, in the other the secondary or induced current winding. Some- ALTERNATING MOTORS FOR RAILWAY WORK. 1 85 times one winding rotates, sometimes the other. Conven- tionally we call the primary member the field and the sec- ondary member the armature. Fig. 94 shows a fifty horse power, two phase induction motor of a recent design and gives an admirable idea of the way in which such a machine is constructed. In this case the field revolves, while the armature is stationary. The working current is led into the field through the three col- lecting rings just outside the bearing, the two phases being given a lead in common at the motor. This revolving field FIG. 94 construction has several well marked advantages. The primary element in which the heaviest hysteretic loss occurs is reduced to the smallest practicable dimensions. The secondary being stationary can have resistance put in series with it through ordinary binding posts, sometimes a great convenience, and since the secondary winding is, as shown, very simple, the armature can be split like the field of a dynamo and the upper half lifted off to permit inspection or removal of the revolving field. As the clearance in iu- )86 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. duction motors is usually very small, }$ in. or less, such an arrangement is very convenient. Fig. 95 shows a fine three phase motor of 125 h. p., in which the armature revolves while the field is stationary. The main leads are taken to the connection board on the top of the motor, and there are no moving contacts what- ever. The resistance used in starting the motor is stowed inside the armature ring and its terminals brought out to three contacts secured to the armature spider. When the FIG. 95. motor is up to speed these are short circuited by a solid ring slipped a couple of inches along the shaft by the small handle shown alongside the bearing. Sometimes this re- sistance is in two or more sections, successively short cir- cuited by a similar motion of the ring. This arrange- ment does away once for all with all moving contacts. The field, being stationary, can be safely wound for higher voltages than if it were rotating and suffers less mechani- ALTERNATING MOTORS FOR RAILWAY WORK. 187 cal strain at all voltages. The machine thus requires very little attention, and besides is quite free from all danger of sparking, sometimes a very undesirable possibility. Both the constructions shown have merits for special purposes. The revolving armature arrangement gives a simpler and safer machine for most ordinary purposes, and especially for high voltage work without transformers. The revolving field is the better for very large motors and for all work requiring considerable and variable resistance in the armature circuit. It is therefore, particularly well adapted for railway work at varying speed, hoisting and similar severe service. In general properties, efficiency, power factor, regulation and so forth the two construc- tions are indistinguishable. For effectively meeting the demands of railway service a motor must be simple, durable and easy to inspect and repair; it must also be capable of regulation in speed with- in rather wide limits, must have great initial torque, and must have a good efficiency. The first three mechanical qualifications the induction motor is amply able to meet. The simplicity of the structure has already been set forth. The nature of the field winding is well shown in Fig. 96, the field of a slow speed, two phase motor of one hundred horse power output, and the winding is for 2000 volts. In ordinary American practice the field coils are in open slots so that they can be the more readily repaired or replaced. The armature winding is usually of massive bars with heavy end connections and is well exhibited in Fig. 94. The matter of durability is best settled by experience. During the past seven years there have been put in opera- tion in this country polyphase induction motors aggregat- ing more than 5o,oooh. p. in output; and from the author's own personal knowledge it may be said that the repairs upon these have been almost negligible, far smaller than in any other class of moving electrical machinery. This is a strong statement, but it is fully borne out by the facts. Speed regulation in polyphase induction motors is ef- fected by means not unlike those used for continuous cur- 1 88 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. rent motors. A common shunt motor may have its speed varied in two very simple ways. First, the field strength may be changed; second, the armature current may be cut down by a rheostat. A series wound motor may be simi- larly governed by changing the field strength or changing the voltage. FIG. 96. In an induction motor the same devices are used in a somewhat different way. Weakening the field of such a motor by reducing the voltage of supply causes the arma- ture to run slower, but since the armature current is sup- plied by the field as a transformer the armature is also greatly weakened and, hence, the torque falls off very rap- idly as the voltage is lowered. Modifying the armature ALTERNATING MOTORS FOR RAILWAY WORK. 189 strength by a rheostat in circuit, however, cuts down the speed until the added transformer effect of the field sup- plies current enough to handle the load at the new rate of speed. By varying the resistance in the armature circuit the speed can be varied to any desired extent, the torque remaining constant throughout. Fig. 97 shows the speed 10 100- I 680- 4+7-O Q 2SO--S 13 14 456 7 8 9 1O 11 Speed in 100 r-p.m. FIG. 97. variation characteristics of a fifteen horse power induction motor with a rheostat in the armature circuit. Starting at full output and speed, the speed was gradually lowered from i4oor. p. m. to 150 r. p. m. The torque remained uniform, so that the output was almost exactly proportion- ate to the speed. The relation between them is shown in curve A. The input meanwhile remained nearly con- stant. B gives the variation of the power factor and C shows the slight and gradual diminution of the input. Altogether this motor behaved almost exactly like an ordinary railway motor with rheostatic control, regulating quite as well and with closely similar inefficiency. 1 90 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. At full load this motor had about the efficiency of a fifteen horse power motor of the ordinary kind, but sub- stantially all the reduction in output by lowering the speed represented loss of efficiency as is the case with a series wound, continuous current railway motor with rheo- static control. The power factor in this case was notably high at all speeds, high enough to cut very little figure in the operation of the system. A car equipped with motors like the one under con- sideration would handle very easily as regards speed varia- 240 220 200 3 fiiso 20 40 60 80 100 120 140 Pounds 1 Foot Radius 160 180 200 220 Street Ry.Jourual FIG. 98. tion and would give quite as good efficiency as hundreds of cars now in operation. For interurban and similar work in which running at reduced speed is the exception, the efficiency would be all that can reasonably be desired. As regards starting torque, which for railway motors is a consideration of prime importance, the modern two or three phase motor leaves little to be desired. Not only will it start with very great torque, but it will give this torque with relatively less current than will a series con- tinuous current motor. That such must be the case is obvious from the fact that while the fields of an ordinary /* I/TERN ATING MOTORS FOR RAILWAY WORK. IQI railway motor are nearly saturated at all working loads, the fields of an induction motor are normally worked at low saturation to avoid hysteretic loss, so that since the torque of a motor is proportional to the product of arma- ture current and field strength, doubling the input in an induction motor nearly quadruples the torque. This is well shown in Fig. 98, which gives the relation between current and starting torque in the motor referred to in Fig. 97. The maximum torque was obtained with a very small resistance in the armature circuit, which resistance was gradually raised to obtain the other points in the curve. The torque was truly static and the power factor of the machine under this condition was lower than when running normally, as shown by the larger current than in Fig. 97. The maximum starting torque, more than four times the full load running torque, was obtained at normal volt- age by the use of about 2 ^ times the normal full load current. Four times the normal drawbar pull is enough for ordinary starting purposes even in severe street railway service, but even this can be still further increased if nec- essary, by raising the voltage. The torque, so long as the field is unsaturated, then increases nearly in proportion to the square of the applied voltage. Thus, if the field coils of the motor are in the star connection for normal opera- tion, and are thrown over to the mesh connection as an ex- treme measure, the applied K. M. F. per coil is increased in the ratio of 1.73:1, and the resulting torque is three times the normal. This in combination with the changes of armature resistance indicated in Fig. 98, is enough to in- crease the torque enormously in spite of increasing satura- tion of the field. In fact one can obtain from an induction motor more starting torque than is ever called for in prac- tical work. Fig. 99 shows the results obtained in testing a pair of three phase induction motors specially arranged for rail- way work. Each motor was designed to produce a normal 192 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. drawbar pull of 800 Ibs. , equivalent at full car speed to about twenty-five horse-power. These machines were wound for no volts between lines, weighed substantially the same as standard railway motors of the same output and were coupled up to a special controller, designed to vary both the armature resistances and the field connections. These connections were threefold, the mesh or A for ex- treme torque, star or Y for normal full speed running, and ' ' concatenated ' ' for half speeds. The latter was a quasi - series connection giving much the same result as reducing the primary voltage, without calling for special appliances. It consisted, practically, of using the secondary current of one motor as the primary current of the other, and of course suffered through adding the inductances of the two. The A curves refer to current, the B curves to effi- ciency, and the C curves to speed of car. Although the concatenated connection was decidedly inferior in efficiency , ALTERNATING MOTORS FOR RAILWAY WORK. 193 both real and apparent, to the others, it still gave half speed very smoothly and with an efficiency reasonably high. The drawbar pulls registered were amply great for any service conditions, and the net commercial efficiency given, which includes all the gearing losses, compares not unfa- vorably with ordinary continuous currents. The running of the motors was as good as could be desired, and the abolition of the commutator is a very material gain, since collecting rings give decidedly less trouble. Change of armature resistance gave opportunity to pass smoothly from one field connection to another without jerking the car. As appears from curves C II and C III, apparatus of this kind has the very considerable advantage of fairly constant speed over a wide range of drawbar pull. Al^ though polyphase induction motors are termed asynchron ous they have so strong a tendency to run near synchronous speed that they have the power of driving ahead regard- less of grades unless grossly overloaded. None of the methods of regulation as yet devised is quite the equivalent of the series parallel control so exten- sively used in continuous current practice, so far as effi- ciency is concerned. It is possible to get, however, as complete control of the speed and nearly as good efficiency at all except the lowest speeds. In the line of work for which alternating motors are most needed, i.e., long inter- urban and similar lines the need of highly efficient control at very low speeds is not so great as in ordinary street railway work, since by far the largest aggregate output is at the higher speeds. In spite of the extent to which induction motors have been used in the past seven years, no important apparatus is less generally understood. Even engineers who are well posted in other matters are apt to be dismally ignorant of the practical properties of induction motors. They have too often derived their scant information from scholastic papers on the subject full of solemn inanities on the gen- eral theory of rotating magnetic poles, fortified by emi- nently respectable equations which are valuable only to those who know the limiting conditions. 194 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. In point of fact the induction motor is a most simple and reliable machine much closer in its properties to con- tinuous current motors than is generally supposed. Its adaptation to railway work is beset with fewer difficulties than confronted the continuous current motor a dozen years ago. The nature of those arising from speed regulation we have just considered. The state of the case is about as follows : Wherever rheostatic control of railway motors is sufficient, the use of induction motors presents no special difficulty, giving the same power of speed reg- ulation upon the same terms as in continuous current practice. This clears the way at once for much long dis- tance and interurban work. In urban and suburban work upon a large scale something more is necessary. Several methods are available, as has already been indicated, con- catenation and the passage from mesh to star connections being the most advantageous yet tried. The application of these methods is now in the tentative stage, with the chances good for a favorable result ^when the work is seriously attempted. It should not be forgotten that series-parallel control of regular railway motors was tried and abandoned on account of forbidding complications several years before it was taken up again and pushed through to definite success. Another interesting suggestion for speed variation is varying the number of motor poles. As an induction motor has no salient poles this is a possible procedure, but it does not promise very good properties at the lower speeds at ordinary frequencies. Varying the impressed E. M. F. by reactance in the primary circuit suggests itself as the simplest method of control. Practically, however, it leads to lower efficiency at all speeds than the rheostatic control and at low speed the power factor is infamously bad. All these things will have to be threshed out experimentally as the present railway apparatus has been. The question of actual armature speed deserves con- sideration in this conection. As a starting point it will be convenient to remember that in ordinary practice one mile ALTERNATING MOTORS FOR RAILWAY WORK. 195 per hour means very nearly ten wheel revolutions per minute. The usual gear reductions found in standard railway motors range from about 1:4.8 to 1:3.5. Hence for a normal speed of 10 miles per hour, one may say roughly that the armature speed should be not over 500 r. p. m. and would not probably be below 400 r. p. m. An 8 pole motor at 30 ^ would give at load say 425 r. p. m. , hence at speeds below 10 miles per hour some form of con- troller would have to be used. This is somewhat awkward for urban work, although it does not differ materially from everyday street railway practice. It simply means the same sort of inefficiency at low speeds to which we have long been accustomed. There is this difference, however, that the polyphase motor could not at that frequency run above 10 miles per hour, which would be a bit awkward in suburban running. Probably a frequency of 40 *> would prove a convenient compromise, or a 6 pole motor at the lower frequency. The conditions just mentioned are compatible with a thoroughly good motor in other respects. The peripheral speed of the armature would probably be somewhat higher than is usual in continuous current railway motors rising to 2500 feet per minute as against 2000 or thereabouts for a common railway motor under similar conditions. In fast suburban work these speeds might be doubled. There is no possible objection to such surface velocities for either kind of motor. An utterly foolish opinion is just now abroad that induction motors demand enormous peripheral speeds, which is not at all the case, as the above will show. When worked at 60 r. or more it is convenient to run the arma- tures at a surface velocity of 5000 feet or so, but at low frequency it is quite unnecessary. For interurban work one would probably choose a 4 pole design, giving say, 850 r. p. m. of the armature at 30 % with a full speed of about 20 miles per hour. The power to pass from 8 to 4 poles or the reverse would ob- viously be very convenient, but the electric railway was an established success on a large scale long before any 196 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. better means of regulation than the rheostat was in use. It is perfectly feasible then to work three-phase motors in a precisely similar way for interurban or even urban work, when conditions make it desirable. 100 40 .20 20 H. P. 3 PHASE MOTOR 4 POLE, 25 ro SPEED 715 R. P. M OH.P. 10 20 FIG. 100. 30 40 We have already discussed the properties of such motors somewhat, but a further examination of the attain- able qualities in a practical motor for street railway service may be worth the while. A good idea of the performance of a well designed three-phase induction motor of about the vSize and speed required for railway work is given in Fig. loo. ALTERNATING MOTORS FOR RAILWAY WORK. 197 The curves are from a 4 pole, 25 ~ machine having a normal rating of 20 h. p. and capable of working up to double that power. The full load speed is 715 r. p. m. 100 t50 200 250 300 POUNDS TORQUE AT 1 FOOT RADIUS FIG. 101. which corresponds to a car speed of about 15 miles per hour. The current (at 220 volts) for full load is 49.7 am- peres, and for 40 h. p., 130 amperes, while the current running light is 13.55 amperes, the input in watts being 198 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. 960. The power factor and commercial efficiency are high all along the line, the former passing 80 per cent at about YZ load, rising to 92 at a little above full load and holding up to 88 even at double output. The latter stays above 80 per cent from 6 h. p. to 34 h. p. rising to 86 near full load. This motor heated but little, the field conductors rising but 4oC, and the armature conductors only 22C after a full load running test of 3 h. 30 m. Results even better than these can be attained, as is shown in Fig. 101. This is from an 8 pole 60 motor of 50 h. p. , having a full load speed of 850 r. p. m. This latter motor, however, is operated with a very small arma- ture clearance, while the former had a clearance of about A- in. For railway work a little greater clearance would be desirable, about }i in. The effect of this would be to lower the full load power factor to about 90 per cent in each case, with no material change in the efficiency. The long and short of the matter is that by careful design it is perfectly feasible to produce an induction motor having an efficiency as good as that of the usual railway motor, and a power factor good enough to dispose forever of the bug- aboo of " false current," as a practical factor in the situa- tion. The motor of Fig. 100 fitted for railway use with gears and gear casings weighs about 2000 Ibs. at an outside estimate, which is not at all bad for a motor of that capac- ity. One must remember that while some street car motors Would show considerably less weight per horse power than this, they are, as a rule allowed pretty stiff heating at Iheir rated load, and as a class have been industriously skinned in the matter of weight for the last ten years. Street railway motors, less gears and cases, usually run vVom 50 to 70 Ibs. per h. p. according to rating, while standard induction motors for stationary service weigh on ^n average from 65 to 70 Ibs. per horse power, sometimes v*own to 60 Ibs. or less. The effect of rheostatic speed regulation on the effi- ^iency of induction motors is worth a brief examination. As regards power factor, Fig. 97 gives the facts in the ALTERNATING MOTORS FOR RAILWAY WORK. 199 case, showing that on the lower speeds the power factor is quite as good as at full speed and load. The efficiency as already stated falls with the speed, in fact almost di- rectly as the speed. The motor of Fig. 100, giving a max- imum efficiency of 86 should show about 43 per cent at half speed and 22 per cent at quarter speed. These facts are set forth not for the purpose of recom- mending induction motors for indiscriminate use on elec- tric railways, but to point out that induction motors are to-day better developed for such work than were the con- tinuous current railway motors that built up the railway business eight or ten years ago. It is not too much to say that at the present time it is practicable to build poly- phase induction motors quite good enough for the entirely succesful operation of the long interurban lines for which they are most needed, and that their use would secure certain advantages not otherwise to be obtained, in the economical distribution of power. The weak points of polyphase induction motors for railway work are as follows: I. Necessity for at least two trolley wires. II. Lagging current. Inasmuch as all true polyphase systems require at least three working conductors, the best that can be done in supplying polyphase current is to utilize the rails for one conductor and provide separate trolley wires for the other two. In rare instances it might be possible to use a third rail and a single trolley wire or even to utilize the two track rails as separate conductors, but such cases are likely always to be exceptional. In conduit work, of course, two working conductors are available without much difficulty, but for general purposes the burden of two trolleys is difficult to avoid. Most street railway men strongly dislike the dou- ble trolley in any form, and beyond question it compli- cates the overhead work, where crossings and turnouts are frequent, in the most frightful manner. Nevertheless even for city work it can be made steadily operative,. 200 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. as the experience of some years in Cincinnati has shown. The principal advantages of the continuous current, double trolley system which are its only excuse for existence, viz., the independence of track condition as regards motive power, lessened interference with other circuits, and ab- sence of electrolysis, do not apply with the same force to a double trolley polyphase system. One branch of the circuit is still grounded and bad track contact is bound to be felt in the operation of the motors under some con- ditions. In. short the double trolley for polyphase work is a disagreeable necessity and nothing better. Again, however, comes to the rescue the fortunate circumstance that in much of the long distance work for which alternating motors are desirable a double 'trolley wire is less objectionable than elsewhere. The matter of lagging current is more serious. Were all induction motors possessed of as good power factors as the one shown in Fig. 97 there would be no trouble, for the lagging current is too small to influence much either the capacity of the plant or its regulation. But armature clearance is a potent factor in varying the power factor, and the motor in question being intended for hoisting had a clearance but little over ^ in. This is too small for the rough and tumble work of electric railroading, and with double this clearance, as in case of the motors of Fig. 99, the power factor is not nearly so favorable. A good power factor of .85 to .90 is very hard to obtain in motors of moderate size and speed such as would be used in street railway practice, and a poor power factor means mischief. Take for example a power factor of .75. This means that a third more current must be generated and distrib- uted than is indicated by the energy and the voltage of supply. Hence the Caving in copper effected by the three phase or two phase three-wire circuits is more than wiped out at once. Moreover, the large inductance of such a circuit involves both a heavy inductive drop and a very unfavorable armature reaction in the generator. Between these and the extra current the station capacity required ALTERNATING MOTORS FOR RAILWAY WORK. 2OI would not be less than i ^ times that needed to supply the same effective energy by continuous current. With large induction motors intended for rather high Gerso f? N St.Sa1vatore Cable Railway Scale 1:25000 Street Railway Journal FIG. 102. speed it is practicable to keep the power factor well up, high enough to render this trouble quite insignificant. All these facts point to the desirability of developing polyphase work in the direction of fast interurban service and heavy long distance work rather than toward ordinary street railway equipment. In the former the polyphase 202 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. system is at its best, its many good features are thoroughly available and its disadvantages are minimized. Nevertheless the abolition of the commutator is so desirable that there is a strong tendency to work polyphase apparatus for ordinary purposes, and it is noteworthy that the first polyphase electric road to be put in operation belongs distinctively to the class of street railways. This very important piece of pioneering work was carried out in 1896 by the famous firm of Brown, Boveri & Company, at Lugano, Italy. I/ugano is a fine prosperous town situated on the lake of the same name at the foot of the Italian Alps. A water- fall a little more than seven miles away furnishes power for lighting the town, and is now utilized for the railway as well. The road runs for the most part along the lake front on each side of the town. It has a total length of al- most exactly three miles, and its general situation is shown on the sketch map (Fig. 102). There are only moderate grades of about three per cent except for three short pitches of six per cent. At the power station is a 300 h. p., horizontal turbine direct connected by a flexible coupling to a 150 k. w., three phase generator. This machine is of the inductor station- ary armature type generally advantageous for high volt- ages and is wound to give directly 5000 volts between lines at 40 ~. The exciter armature is carried directly on the main shaft so that the generator is quite self contained. Its speed is 600 r. p. m. The line is of three wires each about No. 46. & S. gauge, and leads at present to a single transformer station on the southern edge of the town not far from the middle of the line. The three phase transformer here located re- duces the voltage to 400 volts which is the working press- ure between the conductors. The conducting system consists of the track which is thoroughly bonded, as one lead, and two trolley wires, each about No. 3 B. & S. gauge. Bracket construction is employed and the two trolley wires are carried side by side ALTERNATING MOTORS FOR RAILWAY WORK. 203 about ten inches apart. The general character of the over- head structure is well shown in Fig. 103. The current is taken off as there shown by two distinct trolley poles set one behind the other about forty inches apart. This separ- ation of the trolleys, by the way, has been found to be the best arrangement when using a double trolley continuous current system. The trolleys themselves are very similar to those generally used in this country. FIG. 103. Four motor cars are now in use, each of them having a twenty horse power induction motor geared, with a speed reduction of i to 4 to one of the axles. The arrangement of the motor and its suspension from the truck is shown in Fig. 104. The motor itself (Fig. 105) is of the iron clad type with revolving armature furnished with three collect- ing rings. These rings permit the insertion of a three-part resistance in the secondary circuit for the purpose of speed regulation. The function and practical effect of such a rheostat has already been described. In this case it has 204 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. been found to permit perfect control of the speed, as might be anticipated, but with poor efficiency at low speeds. The normal car speed is between nine and ten miles per hour. The starting torque of the motors has proved to be ample, quite sufficient to start a very heavily loaded car from rest on the steepest grade on the line, and the per- FIG. 104. formance of the cars has been on the whole very good. The two trolleys perform well, and, what is rather extra- ordinary, the heavy alternating currents have not given so much trouble as might be expected to the telephone system of the town. There must be a strong element of FIG. 105. good luck in this matter, for under ordinary circumstances induction would be at least quite perceptible, although the leakage difficulties, of course, are practically suppressed as are also most electrolytic troubles. In some recent two-motor car equipments made by Brown, Boveri & Company, a quasi-series connection has been employed for low speeds, the induced current from one motor serving as the inducing current in the other as ALTERNATING MOTORS FOR RAILWAY WORK. 205 in the ' ' concatenated ' ' arrangement already mentioned. Although such devices are, as indicated already, useful in giving a fair efficiency at low speeds, they can hardly be regarded as the full equivalent of the series parallel con- troller now so generally and successfully used with con- tinuous current motors. L,ast July a notable example of three-phase railway work was put into operation by the same enterprising firm. This is a true inter urban road 25 miles long between FIG. 106. the towns of Burgdorf and Thun, Switzerland, running through several smaller towns on the way and connecting at each end with a steam line. The power station is on the Kander River at Spies, about 6 miles beyond Thun and the end of the railway. Here are installed three phase generators direct coupled to turbines. From the raising transformers current is delivered at 16,000 volts, 30 ro, to the railway transmission line. This line consists, of three 5 m. m. (about No. 4 B. & S.) bare wires sup- ported on porcelain insulators. The line follows the road in general, with occasional short cuts across curves, and 206 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. feeds 14 transformer stations which reduce the pressure to 750 volts for the working conductors. These latter are of 8 m. m. (about No. o B. & S.) wire carried on cross suspensions, the two conductors being about 43 ins. apart and about 1 6 ft. above the track. The road is single track with turnouts, of 5 ft. gauge, and is laid with a plain T rail weighing about 73 Ibs. per yard, on steel ties placed a little over 30 ins. between centers. The maximum grade is 2.5 per cent. ALTERNATING MOTORS FOR RAILWAY WORK. 207 <$> 208 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. The electrical equipment of the road consists of two locomotives, each driven by two 150 h. p. motors, and six motor cars each with four 55 h. p. motors. Current is taken from the working conductors by four bow trolleys in pairs side by side at each end of the car. Fig. 106 gives a good idea of the motor car, trolleys and working con- ductors. The motors are arranged much as is usual in four motor car equipments, two on each truck, single geared to the axles. Their full load speed is 586 r. p. m. and they are controlled by rheostatic resistances in the revolving secondaries. Fig. 107 shows the general ar- rangement of the trucks. The wheels are 40 ins. in diam- eter corresponding to railway rather than tramway condi- tions. In fact the whole line is worked on the block system and follows railway practice throughout, the cars being equipped with air brakes and run in short trains as in ordinary suburban work, at a normal full speed of 22.5 m. p. h. "The brakes are worked by an automatic motor compressor and the cars are lighted and heated electric- ally. In general the whole equipment is that of a thor- oughly up-to-date heavy interurban electric road, developed for polyphase transmission. Fig. 108 gives the electrical diagram of one of the motor cars showing the various connections. It should be noted that the rheostats are worked by a rod from the controller instead of being elec- trically connected to it. The motor cars accommodate 68 passengers and weigh fully equipped 32 tons. The loco- motives are intended largely for freight service and have a capacity sufficient to haul easily a 100 ton train up the 2.5 per cent grade at a little better than 11 m. p. h. They have, however, change gearing to enable them to be speeded up for passenger traffic when needful. Fig. 109 shows one of the fourteen transformer sta- tions. Each of these consists of a 450 k. w. three-phase oil transformer in a sort of gigantic metallic sentry box on a concrete foundation. Above and in the rear are the cut-outs, fuse boxes and lightning arresters. The trans- former stations are located close to the way stations, and ALTERNATING MOTORS FOR RAILWAY WORK. 209 are under the charge of the station masters. The whole system is a practical demonstation of the applicability of the three-phase system to the working of interurban lines on a large scale. The most startling thing about the road is the singularly small amount of copper required. Aside from the high tension line there is no feeding system, all FIG. 109. the current being distributed over the working conductors. The total coppe'r in the system is only about 145,000 Ibs. worth at the basic price of 15 cts., $21,750, less than $1000 per mile of line. This for a system designed for motors of more than 1800 h. p. total capacity is sufficiently re- markable to afford considerable food for contemplation. American engineers have fought shy of undertaking this line of work, preferring the easier but more costly and 210 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. less efficient method of transmitting to rotary converters. In the logical development of railway work, however, the polyphase motor certainly has a legitimate place and it is unwise to make a fetish of uniformity to the extent of barring the way to progress. The problem is being worked out for us abroad in roads like the one just described, and in due time we may profit by it as we profited nearly a decade ago by the I,auffen- Frankfort experiment in polyphase transmission. IV. Motors of the asynchronous type working on a monophase circuit are not as yet far enough developed to be immediately available for railway purposes, although they have come abroad into considerable use for general motor work in connection with lighting service. They may be divided into two classes, rather distinct from each other in method of operation, although closely similar to each other in principle and in practical qualities. First may be mentioned those motors which are oper- ated as true polyphase motors by derived polyphase cur- rents obtained by splitting up a monophase current. In this case the actual motor is a true polyphase machine with all the properties thereto belonging, and the real novelty of the system lies in the special methods of transformation adopted in breaking up an ordinary alternating current into symmetrical components. Systems of this sort have been brought forward in this country by C. S. Bradley and abroad by M. Desire Korda. They are somewhat complicated, but are nevertheless operative, and may find a field even in electric traction, particularly in special problems in railroading. The apparatus of Mr. Bradley is shown in diagram in Fig. no. The process employed consists essentially of two operations the splitting up of the original current into two components, differing in phase by 90 degs. , and, second, the combination of these to obtain a three phase resultant system. In the diagram, A is the generator, B one section of the transformer primary system, D a condenser which acts in conjunction with the inductance ALTERNATING MOTORS FOR RAILWAY WORK. 211 of the compound section of the transformer system to pro- duce the requisite 90 deg. phase difference, n and /, the parts of the compound transformer, and g h i j k the segments of the secondary windings. Once given the two phase current, the shifting over to three phase is easy. The coii, z, furnishes one phase, the resultant of g and k a second, and the resultant of h and/ the third, all of which are connected in the ordinary way to the motor, M. The result of this very ingenious combination is a very close FIG. no. approximation to a true three phase relation throughout a considerable range of load, both in starting and running. The use of three resultant phases tends to preserve a more uniform phase relation than would be obtained by utilizing the original two derived phases. The employment of a condenser, while it adds to the complication, tends to annul the inductance of the main circuit. At all events it can be made to give a very high power factor, better than that given by ordinary poly- phase motors. On the other hand, the condenser is an element of weakness in that it is of somewhat uncertain life, and un- less exposed to high voltage and used at rather high fre- quency, is both bulky and expensive. Its use involves 212 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. difficulties in the way of maintenance that, while probably surmountable, are serious in its application to railway con- ditions. M. Korda's device dispenses with the condenser and initially splits up the primary monophase current into two components 60 degs. apart by inductance alone and re- combines these so as to give three phase resultants. It gives a somewhat less stable phase relation and power fac- tor than the method just described employing a con- denser. Second in the list of motors for monophase circuits comes that class which employs a split phase current at starting to obtain a simultaneous transformer and motor action, but in running is purely monophase. Motors of this kind have been considerably developed abroad, but are only used tentatively in this country. As at present made they all start either with very poor torque, or if with better torque demand an enormous starting current, which lags badly. When once up to speed, however, they perform w r ell although never with as high output as a polyphase motor of the same dimensions and efficiency. There are a large number of w r ays of getting the phase difference at starting, some of them requiring modifications of the motor structure, others merely special connections. A consider- able variety of phase splitting devices were devised by Tesla as corollaries to his pioneer polyphase work and di- vers others have been added to the list. Variations of cap- acity and inductance in branches of the main circuit exter- nal to the motor are most often used. In construction and appearance these monophase motors are closely similar to the polyphase ones already described. Indeed most polyphase motors can be worked as monophase motors with very trifling changes. When carefully designed, these machines give a high efficiency and a high power factor when once at speed. Fig. in gives the curves of efficiency and power factor for a fifteen horse power, Brown, monophase, asynchronous motor de- signed for a speed of about 850 r. p. m. at 40^. ALTERNATING MOTORS FOR RAILWAY WORK. 213 These results are nearly as good as can be obtained from a polyphase motor of similar output, but since most of these monophase motors are built with exceedingly small clearance for the armature, down to less than ^ in., there is little likelihood of approximating closely the fig- ures just given with a motor fit for railway work. Nor is it possible to get effective speed regulation in monophase motors by a resistance in the secondary or any other simple means. Summing up the present state of the art, we find that kfre only alternating motors yet constructed, of properties 100 70 10 H. P. out put FIG. III. Street Ry.Journal immediately suitable for railway service, are the polyphase induction motors, which while often weak in power factor, are of sufficient efficiency and general excellence to replace existing continuous current motors. It is certain too, that the lag factor trouble can be overcome by careful design particularly if the frequency is kept low, say, 30^ to 40^. The synchronous motors, both monophase and poly- phase, have excellent properties when up to speed, but do not start will except at the cost of considerable complica- tion. Tjie commutating start appears to give the best torque, but this is not comparable with the best that can be done by polyphase induction motors. The whole 214 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. class are liable to poor power factors when starting, though when running the power factors are uniformly high. The induction motors for monophase circuits are still in an early stage of development as regards application to such severe service as is necessary on electric railways. The most promising of them are those supplied with de- rived polyphase currents even if this advantage involves the use of condensers, since they can be made to give high starting torque and a good power factor. The starting de- vices applied to all existing strictly monophase motors are entirely insufficient for railway purposes unless a clutch connection is used in which rather unmechanical case syn- chronous motors would be generally preferable. With derived polyphase circuits, at least for starting, it is, in the author's opinion, entirely practicable to produce even now a motor for monophase circuits entirely capable of doing certain railway work successfully. It does not follow from this that all classes of electric railway work can now, or ever, be accomplished best by the use of alternating motors of any sort. But the same logic of circumstances that has brought alternating sys- tems into increasing use for lighting and general power purposes applies to railway work with ominous force. It is altogether probable that for a vast amount of strictly street railway work the continuous current motor is here to stay. In its present state of development it is, at least as a motor, as good as any alternating current motor is likely to be. But the question of voltage presses hard, and as the distances to be reached continually grow the time comes when a distribution that can be used for con- tinuous current motors becomes outrageously costly in ma- terial or in loss of energy. The economic value of alter- nating motors depends on their adaptation to a very economical method of distribution. In many cases they not only meet this condition, but can be applied with ad- vantage irrespective of the distribution system. For urban work they possess few intrinsic advantages over continuous current motors. For much interurban ALTERNATING MOTORS FOR RAILWAY WORK. 215 and long distance work they are not only important as a part of the distribution, but have some material points of superiority. In such work, with infrequent stops at stated intervals, their tendency to run at a uniform speed irres- pective of grade and load must be very useful in main- taining the running schedule. The maintenance of speed in spite of moderate variations in voltage is also useful in working long feeders at variable load, and the possibility of working at high voltages greatly simplifies the problem of drawing large amounts of energy from the working conductors. The alternating motor is then fortunately best adapted to that class of work in which the exigencies of distribu- tion make it most necessary. In high speed and long dis- tance work lies its chief strength, and when this kind of railroading is attempted in earnest it is quite safe to say that alternating motors will be used. For light railways running considerable distances across country also, the alternating motor is peculiarly adapted. In no way can the importance of this branch of work be exhibited more forcibly than by computing the initial and operating expense of a road under assumed condi- tions; first, utilizing continuous currents; second, employing transmission to substations with rotary transformers, and finally, using an alternating distribution with alternating motors. It is, of course, quite impossible to select a case that will be exactly equally fair to all three methods, but we can, perhaps, approximate to a fair general case. Let us assume an electric road thirty miles in length running through a series of villages with two cities of moderate size as termini. For simplicity we will assume that the cost of fuel and labor is uniform throughout the line so that the location of the station is uninfluenced to any extent by local conditions. The train service we will assume to be conducted on a twenty minute headway, the actual running time being two hours, including stops. This would keep twelve cars in service. We will also as- 2l6 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. sume the grades to be moderate so that the power required would be fairly uniform throughout the line. The cars stop at fixed points only, with good opportunity for clear running over a large part of the system. With ordinary conditions of load the use of two twenty-five horse power or thirty horse power motors per car would be sufficient and the normal current demanded should not exceed fifty amperes per car or one hundred amperes per car, at 500 volts as a maximum for the system. The total output to be delivered to the cars may then be taken at 300 k. w. average and 600 k. w. maximum. For such a line as this four methods of supply would be worth investigation: I, direct supply from two sym- metrically placed stations; II, supply from a single station with boosters; III, supply from one station with a rotary transformer substation; IV, supply from a single station by alternating currents and static transformers. For sim- plicity, we will assume the cost of track and overhead structure to be the same for all four. So, in fact, it would be for the first three methods, and the extra working volt- age readily obtained with the alternating system at least compensates for lagging current in the trolley wire or the extra expense of stringing and maintaining two trolley wires, if the polyphase system is used. We will compare the systems on the basis of the same loss of energy reckoned from generator to motor, since the efficiency of generators and motor is substantially the same through- out, and for simplicity will not figure out close details of distribution, but reckon the copper required in the simplest possible manner. The permissible loss of energy from gen- erator to working conductor, we will take as fifteen per cent at maximum load, allowing five per cent loss in the trolley wire. We have already seen that if maximum load is taken care of , the average load will look out for itself . In supplying current from two separate power houses these would naturally be placed 15 miles apart and 7^ miles from each end of the line. Each power house would then feed half the line, 7^ miles on each side of its ALTERNATING MOTORS FOR RAILWAY WORK. 217 location. The average distance of transmission would then be 3^ miles, quite nearly 17,000 ft. The maximum voltage for standard generators may be taken as about 600, giving with fifteen percent loss 510 volts at the motors. Each station would have to be able to deliver 600 amperes at a distance of 17,000 ft., with a less of ninety volts. Falling back on our stock formula w = 42X600X289 = 80)920 lbs> 90 At current prices (fifteen cents per pound) this would mean the expenditure of $24,276 for feeder copper for the two stations. The annual output for both stations would be about 2,000,000 k. w. hours. The operating expense of two stations each of 300 k. w. maximum output would, of course, be decidedly more than if the output were concentrated in a single station. The extra expense due to this cause can be esti- mated with fair accuracy. With coal at about $3 per ton it would probably amount to 0.25 cents per kilowatt hour, the difference between, say, 1.5 cents per kilowatt hour with a single station and about i . 75 cents with the two stations. The total extra expense would be then about $5000 per year. With a booster system the principal gain would be the ready use of the extra working voltage on the line. The motors could with advantage be run at 575 to 600 volts giving, say, 700 volts for transmission. The dis- tance of transmission would, however, be doubled, as the best situation for the station would be the center of the line. Taking now the average distance as 34,000 ft. the current, reduced by the extra voltage, as 525 and the per- missible volts drop as 105, we have as before w = 42X525X11,56 = 242>76o lbs< 105 for the transmission in each direction, giving a total of double this amount costing at 15 cents per pound $72,628. The boosting apparatus would probably add $2500 to the cost of the station, and the cost per kilowatt hour generated 2l8 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. would be as above, about 1.5 cents for 2,000,000 k. w. h, per year. Now coming to the transmission systems proper, with a substation and rotary transformers the cost of the funda- mental station, with double-ended generators would be about the same as for an ordinary continuous station. For the transmission there must be added a set of raising trans- formers of about 300 k. w. costing, say, $10 per kilowatt, and extra switchboards and subsidiary apparatus amounting to, say, $1000. The line will have the advantage of high voltage, but the drop will have to be small since the loss in transformers and the rotary transformer must come out of the fifteen per cent allowed as total. With the best efficien- cies that can be expected from these the line loss must not exceed four per cent. The voltage of transmission may be taken as 10,000, hence the drop would be 400 volts. The copper must, of course, be figured as a complete metallic circuit, and the formula will become in this case the current may be taken as thirty-five to make allowance for residual lag and L m is about sixty-eight. We get, therefore, for the transmission line w= 44X3X35X H56X4 81b 400 in all costing, at 15 cents per pound, $8012. At the sub- station there will be switchboards, transformers and rotary transformers for 300 k. w. , which with the house may be lumped at $10,000. Beyond these costs of transmission is the distribution system of feeders, which will cost the same as in Case I, together with the maintenance and depreciation of the transmission plant and labor at the substation, in all, say, $4000 per year. And even after this comes the fact that although the voltage on the working lines can be bold within the fifteen per cent limit of loss, we still have ihe energy loss in the distribution system of feeders. With alternating motors the case is very different.. ALTERNATING MOTORS FOR RAILWAY WORK. 219 The station generating apparatus has the same cost as be- fore. The reducing transformers may be taken at $4000. The whole feeder system would be at high tension, and there would be no need for raising transformers, since the fairly large station generators could well give 5000 volts and be overcompounded for, say, ten per cent loss in the line. The cost of machines for such voltage might be slightly higher, perhaps $1000 on the plant. A like amount should be added for high tension switchboard and extra appliances. Now the total energy in this case is transmitted an average distance of 34,000 ft., as in the booster distribution. Using the same formula as in the preceding case we have, allowing ten per cent line loss, and fifteen per cent extra current to compensate for lag, 44X3X 138X1156 W = ~ ~^o~~~ " = 4 2 >? 8 Ibs. of copper, costing $6312. It is but fair at present to as- sume an extra cost of fifty per cent for the car equipments, say, $500 per car for fifteen cars, in all $7500. We may now gather these data as follows : Case. Cost of copper. Cost of Extra apparatus. Cost of 2,000,000 kwh. 10 % on copper. 10 % on extra app. Extra labor for working. Sum of these anuual charges. I. $2 4) 276 $35,ooo $*> 427 $37,427 II. 72, 628 $2 ,500 30,000 7, 262 $ 250 37,512 III. 32, 288 14 ,OOO 30,000 3> 228 I ,400 $2,500 37,128 IV. 6, 312 13 ,500 30,000 631 I ,350 31,981 These figures speak for themselves. In reality III is, under the assigned conditions, decidedly inferior to the others in efficiency, as already indicated. It can only be used economically under rather rare conditions, and then only in default of a proper alternating motor system. I and II are almost exactly equivalent, and very small differ- ences in cost of power generation would throw the advan- tage one way or the other. IV is easily the best, and would still hold its position of superiority in the face of a considerably larger allowance for lagging current tljan that here made. With a smaller permissible loss of energy than fifteen per cent, the booster system would drop rapidly to 220 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. third place in desirability, and IV would have even greater advantage than at present, while III would be out of the question. Any increase in the price of copper or decrease in the cost of apparatus would give a still further advant- age to the alternating motor system. It should be noted that the tabulated figures do not in any case include the working conductors. At all distances and losses a good alternating system would be in the front rank, and excepting at very moderate distances, would easily lead. One fact, however, must be remembered. An alternating current does not penetrate far into the substance of an iron conductor, hence in using an alternating system the rails cannot be counted on for their full conductivity. This would be very serious even at 25 no if it were not that the magnetizing force due to the current in the rails would under ordinary circum- stances be so low that the permeability of the steel would be small, not over 200 to 300. At 25 *\/ the equivalent conductivity of rails of the usual sections- cannot safely be taken at over 0.5 the usual value, perhaps as low as 0.3. Fortunately, in the interurbaii and long distance lines for which alternating motors are most . needed, the current density in the rails is likely to be so low that the permea- bility is kept down and the rails are still fairly good con- ductors. CHAPTER VIII. INTERURBAN AND CROSS COUNTRY WORK. The most important class of electric roads at present is that composed of tramways that have outgrown and reached beyond their urban starting points and serve to interlink cities and villages. These lines are important and interesting to the engineer, since they are often sub- ject to unusual conditions and require special treatment, and they are of immense value and importance to the pub- lic, because they tend to break down the industrial barriers that have been artificially established between city and country, and give to both some of the advantages now peculiar to each. There is nothing in the nation's growth more menac- ing to good government and the healthy growth of industry than the rapid concentration of population and enterprise at a small number of overcrowded spots. The opening of easy channels of communication through the country at large, increases enormously the areas available for profitable manufacture and decent habi- tation. Much has already been accomplished by the in- terurban and suburban electric railway systems already installed, and much more can be done by the extension of these lines and the building of new lines through regions that are now isolated. Fig. 112, showing the connected system , of which Boston is the center, gives a vivid idea of the extent of country covered and the thoroughness with which the work of in- terconnection is done in certain regions. Still, large dis- tricts are left untouched, giving ample room for further extensions: The districts already interlaced, however, have an aggregate population of very nearly 1,250,000 in- 222 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. habitants. And all this, with few exceptions, is the result of extension of strictly urban systems and not of independ- ent effort at new avenues of intercommunication. This character of growth is attested by the fact that of the entire network only the road from Lowell to Nashua, N. H., and the isolated Nantasket Beach road, differ in engineering features from the general practice on purely urban roads. Practically all the work is done in the ordi- nary way at about 500 volts. Of course, the tout ensemble is a shocking example of inefficient and costly distribu- INTERURBAN AND CROSS COUNTRY WORK. 22$ tion the necessary result, however, of its manner oi growth. Some of the component systems, of course, are beautifully designed. Most existing roads of the interurban class have in similar fashion been the result of extensions, but recently there has been a tendency toward systems intended delib- erately for interurban work, and designed with this in view. Such is the system about Cleveland, O., described in a former chapter, the recently opened line between L,os Angeles and Santa Monica, Cal. , and divers others. These lines are rapidly increasing in numbers and form the con- necting link between street railways with thefr suburban extensions on the one hand, and electric systems replacing ;steam railroads on the other. The distinction between these classes is somewhat ar^ tificial, but none the less real. We shall consider only those roads that are prepared to operate capacious trains at speeds of thirty miles per hour and upwards as really entering upon the functions of ordinary railroads. The strictly interurban roads have a function of their own, and a most important one, in linking together urban systems and opening up direct service between points previously connected very indirectly. A glance at Fig. 112 will show that the latter f unction is even now very imperfectly fulfilled. There are still left great areas in which there is no intercommunication except by paying a double tariff into and out of one of the larger cities. The cross country roads, as yet but little used in this country are destined to play a very important part in the development of our country. They should serve as feeders both for steam roads and interurban electric roads, form- ing the capillaries, as it were, of the industrial circulation. They are naturally allied to interurban systems, but owing to the necessity for cheap construction and the compara- tive unimportance of high speed, must be separated from them in engineering details and particularly in equipment. The interurban road proper differs from the ordinary 224 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. street railway in several very important particulars. First s the speed is on the average very much higher; second, the stops are relatively much less frequent; third, the average distance between generator and motors is far greater; and fourth, the average power per car is considerably more in amount. As regards the first count, the actual speed on all elec- tric roads is apt to be overestimated. Most cars on street railways have an average speed, including stops, nearer five miles an hour than ten, as can readily be figured from the hours of running and the average daily mileage. Now FIG. 113. for runs between town and town much greater speed than this is desirable and can be readily reached in the absence of traffic obstructions. The interurban line should be able to make at least double the average speed of the street railway proper, and this means from twelve to eighteen miles per hour includ- ing ordinary stops. The maximum speed corresponding to this is likely to be from twenty to thirty miles per hour, seldom, however, the latter figure. The general running speed is likely to be between fifteen and twenty miles per hour, seldom the latter figure. These speeds call at once for modifications of standard cars and trucks. Under such conditions the common single truck is positively unsafe on ordinary track, and recourse INTKRURBAN AND CROSS COUNTRY WORK. 225 must be taken to double truck cars. The importance of this has been emphasized by several serious accidents from attempting high speeds with single trucks. So in the natural course of evolution a fine type of double truck car, similar to that used on many large urban systems has come to be used for most interurban service. Such a car is well shown in Fig. 113. It is, save in size, closely similar to an ordinary railroad car, having the same FIG. 114. general interior arrangement. It is forty feet long, ves- tibuled at one end, and is provided with special air brakes. Another recent interurban car partly open and partly closed (a favorite construction on the Pacific Coast) is shown in Fig. 114.. This is rather lighter and five feet shorter than Fig. 113, and like it is provided with air brakes. At interurban speeds, electric or air brakes are almost a necessity and on the later roads are quite generally pro- vided. As .a rule too, the wheels are larger than the thirty-three inch size now standard on most street rail- ways, thirty-six and forty-two inch wheels not being in- 226 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. frequent. These sizes give more room for the larger motors required and are better adapted for the cars. As to track, careful laying and good ballasting are the essential points. The rails themselves are what would be used for a light steam railroad, forty to sixty pound T being the rule, although at the termini the usual girder rails often have to be employed. It should be remembered that a city track gets far more wear and tear than the average interurban track and must be, accordingly, even more substantial. The rather infrequent stops in interurban work pro- duce on the whole a tendency toward uniform distribution of load that operates favorably on the necessary distribu- tion of power. The service is less liable to blockades, it is easier to hold to a regular schedule and there is less of the troublesome shifting of the load, than in street railway practice. Consequently it is somewhat easier to plan the feeder system. On the other hand, the average distance to which power has to be transmitted is considerable, so that the aggregate amount of feeder copper is great, and it is ag- gravated by the frequent attempts to transmit power un- reasonably long distances at 500 volts to avoid distributed stations or other appropriate methods. The absolute amount of power required per car is, for an approximation, nearly double that required for a stand- ard double truck car in street railway work. The speed of the interurban car is nearly double, and the car itself is often heavier. On the other hand the average live load is likely to be smaller and the power wasted in stopping and starting is less. On the ordinary urban railway twenty to twenty- five amperes per car is not far from the aver- age power required through the day; on a busy interurban line forty to forty-five are likely to be required, or thirty to forty if the traffic be moderate. Consequently heavier motors are often employed than on street railways, although for many cases they are un- necessary. If the traffic is likely to be large or if the speed TNTERURBAN AND CROSS COUNTRY WORK. 22/ is to be carried toward the higher limits mentioned extra large motors should always be used. Figs. 115 and 116 show motors especially planned for interurban and similar work. They are of the usual Gen- eral Electric and Westinghouse types respectively and may be classified as of forty to fifty horsepower. They are fully up to the speeds and loads needed for heavy interurban serv- ice and are coming into extensive use for this purpose. In general construction and arrangement they are closely sim- ilar to the standard street car motors of the same makes, and are habitually worked with series parallel control, FIG. 115. which may properly be considered a necessity for economi- cal operation. The saving by such control in interurban work is, of course, less than usual, since the motors are in parallel most of the time, but the device is very necessary to bring the speed within reasonable limits in running through towns. Except for the unusual size of the motors and the gen- eral use of power brakes there is little peculiar in the car equipment necessary for interurban work. The trolley and its connections are quite as usual and the method of oper- ation is unchanged. The trolley wire, too, is of the same character and sus- pended in the same way as for ordinary street railway work. 228 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. It is advisable, however, to use a larger trolley wire than usual, not at all to secure larger area of contact with the trolley, for this is needless, but to simplify the feeding system. The larger the trolley wire, the 'easier it is to equalize the voltage along the line. Nothing smaller than No. oo should be used and No. ooo or No. oooo may often be useful. These larger sizes require special precautions in suspension, but sometimes are worth the trouble. In most interurban work the bracket suspension can be freely used and is advisable, being cheaper and easier to keep up than the crosswire suspension. FIG Il6. The supply of power to an interurban line can best be illustrated by working out the details of a concrete case. We may take for this the hypothetical line discussed at the end of the last chapter, selecting for discussion the first case, using two stations for the line. Fig. 1 17 is a diagram of the system. Here A and B are the termini, C and D intermediate towns which may have an influence on the distribution of the power and E and F the points selected for the power stations in Chap. VII. The track is, as will usually be the case in such roads, a single track with turnouts. The distance from A to B is thirty miles, from A to C about five miles and from D to B twelve INTKRURBAN AND CROSS COUNTRY WORK. 2 29 miles. In the previous discussion the load was assumed uniform along the line. Obviously it is unlikely to be so, and we must accordingly modify the simple arrangement there shown. We may still allow the fifty amperes per car as before. The exact effect of C and D on the inter- urban traffic cannot possibly be foretold, and indeed it will constantly be subject to some variation, nevertheless cer- tain things can be safely predicted. The local traffic between C and A will have the effect of shifting the load centre of the section, E A, toward A. Similarly the traffic between D and B will shift the load on the right hand side of the section, K F, somewhat toward F. If the towns, C and D, are of nearly the same size, the two halves of the line will be about equally loaded, so that the stations will be of the same size. E D will assuredly be the most lightly loaded section of the line. H- T-! Di< --- M --- >iF -12-m.- FIG. 117. Now what conclusion as to the distribution of power are we justified in making? Although some data are avail- able as to purely urban traffic in cities of known size, there are as yet no data for predicting the probable actual travel on an interurban line. The assumption as to current re- quired is as close a guess as one would be justified in mak- ing. Any change in the distribution of feeder copper, due to assumed differences of load in different parts of the line, is somewhat hazardous, and about the only change author- ized by the evidence is a change of position of the station, E. A situation at or near C is certainly an improvement. It might be advantageous to make F equidistant from D and B, but in view of the shift in E it probably would not be desirable to further increase the distance between sta- tions. Throughout we assume that the real local traffic over our line in A and B is small, owing to local street railways. 230 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. We may now lay out the system as shown in Fig. 118 and plan the stations and feeders. We have two stations, E and F, of equal size, each supplying half the whole line, A B. The station, E, however, serves a line five miles in one direction (E A) and ten miles in the other (E G) to G, the center of the line. The station, F, feeds 7^ miles each way. Each of these stations must be able to furnish a maximum current of about 600 amperes and an average output ot about half this. The voltage should be as great as the standard generators can conveniently give, say, 600 volts as a maximum. If we are, as in the previous discussion, to allow fif- teen per cent loss at full load, ten per cent in the feeders and five per cent in the trolley wire, the generators may ,;;:;,,;_,_,; . <-._"] j j | ! j j \ ! j i ' '' \ 250000 C. TO.. |250000 C. IT l WOOO c."m. J ""5WOW cr^ '^pj ilii!^--!-^ ^ ^ Street Ry.JounuJ FIG. Il8. well be given an overcompounding for ten per cent, thus taking care of the loss in the feeders. Each station should be equipped in duplicate, partially at least. The maximum continued output at any time we have assumed at 360 k. w. , 600 amperes at 600 volts. The average output is about 180 k. w. and, save at special times, the maximum output would be considerably less than that noted above. If at each station were installed two generators, of 225 to 250 k. w. output apiece, one of these would handle the whole load of the station during a considerable part of the day, and the second could be thrown in on the heavy hours or all day on special occa- sions. In case of accident to a dynamo or an engine the remaining one could keep up the service without serious interference with traffic, particularly if the feeders were arranged judiciously. As regards the arrangement of these units it is rather an open question between direct coupling and direct belt- INT3RURBAN AND CROSS COUNTRY WORK. 231 ing in a plant of this size, with, perhaps, the weight of ad- vantage rather in favor of the former alternative. At this output, however, one is quite near the point at which the construction of direct coupled machines becomes embarrass- ing on account of low speed, and it often happens that the belted plant is not only cheaper, but more efficient. It most emphatically does not pay to couple directly to a sim- ple, non-condensing engine, instead of belting to a Corliss type engine in a plant of this size. It always pays to con- dense, and it nearly always pays to use compound engines. The simple, single valve, non-condensing engine is an eco- nomic abomination in such a plant, and should not be con- sidered for a moment. The author's choice would be a compound condensing engine, with independent admission and exhaust valves, running not less than 120 to 150 r, p. m. Such an engine plant will produce power at about two-thirds the cost by ordinary simple engines, and very nearly as cheaply as the best that can be done under similar circumstances by the best triple expansion engines- which, in a plant of the size considered, are less suited to the conditions of variable load than compound engines. We may now take up the distribution step by step. For a constant we may safely take 14, as representing the- case of a good track and moderately heavy traffic. Beginning with the section E A, we may safely assume that about one-fourth the total load will be concentrated upon it, and uniformly distributed. We will assume a No. ooo trolley wire of 167,000 c. m., weighing 2677 Ibs. per mile. This wire will carry 100 amperes, the maximum current for a single car, over 4000 ft., with moderate loss of voltage. Three cars normally belong on the section, and we must meet the contingency of all three being at A and loaded, calling for, say, 200 amperes. In this con- tingency we should be justified in assuming as much as 100 volts drop in the feeders alone, and that not more than one other car will be fairly upon the section. At five miles distance the delivery of 200 amperes under these conditions calls for 728,000 c. m. This may be very ad- 232 POWER DISTRIBUTION" FOR ELECTRIC RAILROADS. vantageously approximated by a 600,000 c. m. cable plus the trolley wire. If this cable is tied into the trolley wire at frequent intervals, say, every 1000 ft., for a mile or two near A, the drop in the trolley wire as such becomes trifling, and the drop saved here may be transferred to the feeder account. Nearer K the taps need not be so fre- quent, and the trolley wire should be directly connected to the station. We may then arrange this section as shown in Fig. 118 by the dotted lines. Under this arrangement the drop at normal full load, with one car at A, a second nearly midway between K and A, and another near K, assuming for current 100 amperes per car would be pretty near the required fifteen per cent, although, as we have pre- viously seen, the conditions of extreme load must in the last resort determine the amount of feeder copper. A feeder of 600,000 c.m., uninsulated, weighs 1800 Ibs. per thousand feet (three times the circular mils in thous- ands, as we have already seen), hence we must write down against this section about 48,000 Ibs. of copper. Next comes the long section E G. On this three or four cars may normally be operated. About the worst that can be expected is a couple of cars near G calling for perhaps 150 amperes together and a similar pair fairly near K. As to drop we may here take rather extreme measures and allow, so far as station E is concerned, a maximum drop of 150 volts. This calls for 770,000 c. m. which we can again make up of a 600,000 c. m. cable plus the trolley, the two being frequently tied together. But even this does not properly take account of the second pair of cars. These at worst cannot be ex- pected to be more than five, miles from E. Hence under the same conditions of drop the total area of copper re- quired would be 385,000 c. m. In connection with the trolley wire a 250,000 c. m. cable would be rather more than enough to do the work. This feeder should be tied to the trolley perhaps every 1000 ft., and should cover the first half of E G. This pair of feeders, as shown in Fig. 1 1 8, complete the distribution system for the station E. The INTERURBAN AND CROSS COUNTRY WORK. 233 main feeder here would weigh about 96,000 Ibs. and the short feeder about 20,000. We may now take up the station F and its connec- tions. F is midway of its section and the only disturbing factor is a small one, the town D. Its tendency would be to move the load center of the section G F toward F since a town in such a situation would probably be tributary to B rather than A. The section G F would normally contain three or at most four cars. The worst concentration of load to be expected would be a pair at G with another pair between D and F. Allowing 150 amperes for each pair and a maximum of twenty-five per cent drop to G, we find about 6oo,oooc. m. required. But G can draw part of its current from E. Therefore we can take advantage of this fact and not only use less copper from F to G, but reduce that from K to G. Altogether we are unlikely to get more than three cars in the vicinity of G calling for, say, 200 amperes. This would call for only about 1,000,000 c. m. from both stations. Since it is desirable to give one station the ability to ex- tend some help to the other it is desirable not to cut down the copper too much. One of the most practical ways of doing this is that shown in the figure. Reducing the main feeder from K to G to 500,000 c. m. we run a similar feeder from F out to and beyond G, making the two feed- ers of the same length. This leaves on the section G F two cars unprovided for. As there may be an occasional call for extra conductivity toward D, this section may well be provided for by a 250,000 c. m. cable up to D. The 500,000 c. m. feeder weighs 1500 Ibs. per M feet, and there is ten miles of it, weighing, say, 30,000 Ibs., which is also the weight of the revised feeder from B to G. The 250,000 c. m. feeder weighs 750 Ibs. per M feet and its total weight is about 15,000 Ibs. We may now pass to the final section, F B. The con- ditions at B are similar to those at A. Allowing 200 am- peres possible demand near B, about 750,000 c. m. will do the work there. There may be, however, a car or two 234 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. elsewhere on the section at the same time, calling perhaps for one hundred amperes. The joint load can be best taken care of by a pair of feeders, one to B of, say, 500,000 c. m., the other out, say, four miles from F, of about 250,000 c. m. The function of the latter is to handle cars within that distance of F and also to improve the conditions at B. These are shown with the rest in Fig. 118. The weight of the main feeder here would be 60,000 Ibs., that of the small feeder, say, 15,000 Ibs. We can now take account of stock and find the total cost of copper for the feeding system. We may tabulate as follows; Section. Wt. A K 48,000 EG (main) 80,000 (adjunct) 20,000 G' F ( long main ) 80,000 adjunct 18,000 FB (main) 60,000 adjunct 15,000 321,000 This would cost at fifteen cents per pound about $48,000, a very different figure from that previously found by the assumption that the maximum load may be taken at the middle point of the proposed line to be fed. The existence of this excess and the causes that pro- duce it must be carefully examined. In the first place 20,000 Ibs. of copper, the section of feeder G'G belonging to station F, is directly chargeable to safety precautions, and is for the purpose of enabling the two stations to be of some material assistance to each other in case of accident to one of them. The large remaining discrepancy is almost wholly due to the fact that the load on an electric railway is a shifting one. Instead of being able to assume a uniform distribu- tion of the maximum load, it must be treated as a concen- trated load, perhaps even at the most unfavorable point of the line. In fact, it often happens that the maximum load INTKRURBAN AND CROSS COUNTRY WORK. 235 must be handled at the extreme end of the section, instead of the middle point. Since the copper required varies as the square of the distance, this extreme position would re- quire four times the copper called for under the original hypothesis, but on the other hand for this abnormal load much more than the average drop is allowable. In this as in all other railway work the real invest- ment in copper is determined not by the average loss of energy that may be desirable, but by the maximum drop permissible under the worst conditions of load. This con- dition weighs heavily on interurban lines, since where a network is possible, the various parts of it will not be loaded simultaneously and can help each other out, while on a straightaway line each section of conductor must act for the most part independently. In a long interurban road of this character the booster may often find a legiti- mate and important use. If we could depend on a fairly uniform schedule of traffic the practical arrange- ment of the feeding system would be much simplified. In cases like the one in hand there are likely, in spite of care- ful operation, to come times when cars are massed at one point on the line to an extent not contemplated in the design of the system. Suppose for example that occasionally extra cars must be run between A and B. A circus comes to the latter place or some special celebration takes place there and it becomes necessary to accommodate a very unusual number of passengers within a limited time. It may then be very important to deliver double the usual maximum current, say, as much as 400 amperes, at B while still retaining a good working voltage. This the existing feeders would be quite inadequate to maintain, since the drop would be in the vicinity of 300 volts. To bring the working pressure to about 500 volts, which would be highly desirable to meet this exigency, would require the installation of something like 75 tons of copper, costing about $22,500. The best alternative is to install a boosting dynamo 236 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. at the station F, to furnish about 200 additional volts on the long feeder, F B. The capacity of this booster should be about 100 k. w. and its cost, complete with motive power, and ready to run would not be more than $5,000. This would enable the voltage to be kept up at B under most trying conditions, and would moreover enable station F to be of great assistance to station E in case of accident to the latter. A similar booster at E would be able to render similar service at F and to take care of the most abnormal loads at A. Ordinarily these boosters would be in sendee only at infrequent intervals, a few hours per day for a day or two at a time, but they would be well worth installing merely as a precaution, insurance as it were. If the loads on the line become such as to require fre- quent aid from the boosters the economics of the case would have to be looked into as indicated in Chap. IV, and it might prove to be wise to install additional copper. In all roads of this class the local conditions must ulti- mately determine the character of the feeding system for the most economical results. In spite of this the copper required in the case in hand is not very formidable. Unless the road is operated on a regular schedule, still more copper would be required, since if the operation of the cars is careless and irregular, more may be massed at a single point than were allowed for in the estimate. For economy in copper the road must be intelligently operated as well as skillfully planned. The same uniform schedule that secures good and regular serv- ice throughout the line will facilitate good and economical distribution of power. The only reasons for unusual massing of cars at one point are accidents to the track or motors or very unusual demands for car service. In the former case, the service can be resumed on regular time without any extraordinary demand for power, and in the latter there will be no trouble if any extra cars that may be necessary are run in an orderly manner, as they would be on any well conducted railroad. Managers should bear in mind that the operators on an important interurban line INTERURBAN AND CROSS COUNTRY WORK. 237 should be picked men of more than usual skill and intelli- gence, and that it pays to get such men. They are worth the extra cost merely as a form of insurance. No general rule can be assigned for the increase in feeder copper due to the demands of heavy displaced loads. The amount and character of the displacement varies in different cases in a way that cannot be formulated. The only thing to be done is to take up each case as a special problem as we have just done. The effect of this extra copper on the relative economy of the various methods ot supply is easy to approximate. Recurring to the estimates at the end of the last chapter, it is evident that they need revision. The annual cost by method I will be increased by the interest and depreciation on the additional copper. Method II will suffer in almost the same ratio as method I, and hence the absolute increase in copper and the added annual expense will be greater, putting the booster method to very serious disadvantage, if used without undue loss of energy except as it may be adopted for emergencies, as just described. . Method III, which really consists in transmitting power at high voltage from K to F (Fig. 118), is affected to precisely the same absolute extent as method I, and there- fore has practically the same relative value as before. Method IV must take into account the same condi- tions of displaced load that influence the other cases, but in a somewhat different way. The trolley wire alone is unable to carry the current for a severe load any consider- able distance, hence it must be reinforced unless the line is to be supplied with an exaggerated transformer capacity and the transformers are placed very near together. To give a good practical distribution of power there must be sufficient feeder capacity to easily carry the current for the extreme loads already mentioned without demanding trans- formers at too frequent intervals. The net result of the conditions of load will be, first, to demand the installation of feeder copper to distribute the energy delivered from the transformers, and second, extra transformer capacity 238 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. enough to respond to the utmost calls of a displaced maxi- mum load In the case before us it would not be wise to subdivide the transformer capacity greatly, since the in- dividual units would be small and expensive. The maxi- mum load of 1 200 amperes at 500 volts will be increased somewhat by lagging current, and no part of the line can be left without transformer capacity enough to take care of the heaviest load to be met. Consequently the total capacity of the transformers will certainly be much greater than the nominal 600 k. w., probably at least one-half greater. Without going far into details, feeders of not less than 250,000 c. m. would be needed to reinforce the trolley wire for its entire length. These would cost about $18,000 and in addition the extra cost of transformers due to their moderate size and extra capacity could hardly ag- gregate less than $5000 more. Hence an annual charge of about $2300 must be added to the annual cost of powei obtained in the last chapter. This leaves the distribution by alternating current in the same relative position of ad- vantage as before, a position which is the stronger as the distances to be covered grow greater. Only when the serv- ice undertaken is exceedingly heavy can distributed sta- tions compete with a good alternating transmission, and the latter always has the possible use of water power or utilization of cheap coal to its credit. A very striking instance of the value of transmission to alternating motors is given by the Burgdorf road de- scribed in the last chapter. Here full advantage is taken not only of the economies of transmission but of the facil- ity with which polyphase motors may be worked at volt- ages considerably higher than 500 volts. The result is that in spite of a heavier load than is here assumed, no low tension feeders are employed and the total cost of cop- per is hardly more than we have here taken for the cost of auxiliary feeders alone. Allowance must be made, however, for the fact that in Switzerland copper is more expensive and transformers are materially cheaper than in this country. Here the INTERURBAN AND CROSS COUNTRY WORK. 239 legitimate tendency would be to use fewer transformer stations and more copper than in the foreign case. With 750 volt motors the weight of copper required for the distribution would come down to about one-half of that which we have assumed and the whole amount could be conveniently placed in the working conductors. As to number and location of transformer stations, the most beautiful feature of the whole method is that these can be placed, without any material variation of cost, just where they will do the most good. On the road that we have assumed for investigation the most advantageous number would probably be somewhere between six and ten. The natural locations would be near A, B, C and D respectively, between C and D, and between D and B. The distribution in number and position would be gov- erned by the distribution of the load. It may often be convenient too, to vary the size of the individual trans- former stations so as to best meet local conditions, and the system as a whole is wonderfully economical and flexible. Very different in character, but nevertheless allied in function to interurban roads are those which we have de- signated as cross country roads. It is surprising to realize how small a part of this or any other country is conveniently tributary to existing railway lines of any kind. A glance at the map of any well settled state will show many townships not touched by any railway and many more only reached in round- about ways. It is not uncommon to find a rich farming district almost without means of communication with neighboring cities and totally devoid of facilities for inter- communication betweens its parts save in the good old fashioned way. Nearly one-seventh of the towns in Mas- sachusetts are without railway stations. Within fifteen miles of Boston is one whole township untouched by a railway of any kind, steam or electric. In the less popu- lated states, there are many fine regions that are quite isolated. 240 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. Railroads have left these regions untouched because a route elsewhere would pay better, or would give pros- pects of traffic sufficient to float a heavier capitalization without creating undue suspicion. There are, of course, plenty of cases in which an ordi- nary railroad, even a branch, would not pay and conse- quently is not built, while the prospects of traffic are yet quite near the paying point. A railroad is a rather in- flexible thing at best. It requires a nearly level track, must avoid severe curves, has often to acquire an expen- sive right of way and is in general subject to restrictions and limitations in such wise as to render construction and operation somewhat too costly for many places that are yet in the aggregate of considerable importance. Espec- ially in the agricultural regions there is much rather scat- tered freight traffic which cannot be easily handled by an ordinary road at paying rates, but could be profitably gathered and increased by roads built with this specific object in view. Abroad much has been done in the way of building light railways especially for the purpose of developing agricultural districts. Most of them are narrow gauge, be- tween two and three feet, although a few conform to the existing standard gauges for convenience in exchanging and transmitting cars. In Belgium and Prussia especially this class of service is very considerable in amount, although there are roads of this kind all over the Continent and not a few in England and English colonies. Owing to foreign habits of railway construction most such lines are from our standpoint too expensive, costing in general from a minimum' of $7500 to $ 15,000 or more per mile to build and equip. In this country there was fifteen or twenty years ago an epidemic of narrow gauge construction, generally re- sulting in a change to standard gauge; The truth is that while these light, narrow gauge rail- roads can be built and equipped quite cheaply, often for half the cost of standard construction, they are seldom INTERURBAN AND CROSS COUNTRY WORK. 241 cheap enough to give much advantage when they attempt serious railway service. In competition with regular lines, they soon find themselves handicapped, and for purely local purposes they generally are too costly. The need in very many cases is for feeding lines to facilitate the movement of commodities and passengers now laboriously hauled over country roads. For this specific purpose the first consideration is cheapness; these lines would not come into competition with existing rail- roads, hence there is no need for more than very moderate speeds; there is no need of handling heavy trains; light passenger cars and freight skips are quite sufficient. The moment one attempts to use standard gauge and exchange cars with through lines heavy construction is necessary to stand the wear and tear, and the cost becomes too great for the purpose in hand. For this cross country service electric construction is singularly well suited. Grading, always an item of ex- pense to be feared, is much reduced with an electric road, for while two or three per cent grades are all that would be attempted in ordinary light railway construction, ten per cent is perfectly practicable for an electric car with a light trailer or two. A gain equally important is the weight of the motive power. Instead of a locomotive weighing six to ten tons, the dead weight of the motor need not much exceed half a ton, which, with all the load in the motor car, is available for securing adhesion. With this lessened weight to be carried the track construction can be lightened and cheap- ened correspondingly. In spite of the singular fitness of electric service for this particular and most useful purpose, little has as yet been done. Perhaps the reason is lack of popular apprecia- tion of the exact conditions to be met. The danger lies in trying to do too much, in building an ordinary cheap elec- tric railroad, instead of something little more elaborate than a telpher line; in trying for a speed of twenty miles per hour where ten is amply sufficient. 242 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. To meet the need of small places for transportation facilities one must cut his coat according to his cloth. A little hard common sense applied to the problem will result in the establishment of many a most useful line, giving greatly increased facilities for intercommunication, and yielding good returns on a small investment. A standard gauge (4 ft. 8^ ins. ) electric railway track can, if the grading is trivial and the route is generally easy, be put in position for a total cost of as little as $5000 per mile of single track, exclusive of bridges and other special construction and right of way, using ordinary cars and car equipments. This supposes T rail of forty to forty-five pounds per yard and economy everywhere. The cost of overhead wire, bonding, equipment and station per mile, of course, depends entirely on the service. For a road, say, ten miles in length, very economically equipped, $4000 to $5000 per mile may be enough. In other words, the cheapest feasible price for building and equipping a stand- ard gauge electric road is somewhere about $9000 to $10,000 per mile, anything under $10,000 being extraordi- narily low. Now for the work properly belonging to cross country roads that figure is often prohibitively high. In order to do the work at a less price, radical changes have to be made in the structure. For localities where grading is slight, and there is not likely to be much trouble from snow, light, narrow gauge roads meet the conditions fairly well. Foreign practice gives valuable data in this line. For a gauge of o. 6 metre frequently used abroad (practically two feet), a rail weighing about twelve kilos per metre (twenty-five pounds per yard) is freely used. The sub- structure can be light in proportion, for the rolling stock is also light, albeit the locomotives are decidedly heavier than a loaded motor car would usually be. We must re- member that with light cars, comparatively low speeds and rather infrequent service, a light rail can be safely used, and will give no more trouble than heavy track under ordi- nary street railway service. INTERURBAN AND CROSS COUNTRY WORK. 243 A twenty-four inch gauge track, laid with thirty pound rails, can be put down under favorable circum- stances for about $3500 per mile. Then comes the bond- ing and the erection of the overhead structure. The amount of wire required for such a line is comparatively small, for the power also is small. For a line ten miles in length, two trains in steady service, each consisting of a light motor car and a freight skip, would meet all ordinary requirements. The total weight, loaded, should not often exceed ten tons. To drag this load on a level track at eight miles per hour requires about seven horse power at the car wheels. As grades would naturally be taken at a somewhat lower speed, the power required would not increase very greatly, and an expenditure of fifteen horse power at the wheels would seldom have to be exceeded. FIG. 119. In reckoning the copper we should have to allow for the delivery of about thirty amperes to the train. With 600 volts initial pressure, and allowing one hundred volts drop at the end of the line, it appears that the copper re- quired is trifling. Using 1 3 as the constant in our stock formula, the wire, supposing the station to be at the cen- ter of the line, conies out No. o, which may conveniently be suspended as the trolley wire. For economy bracket construction should be used, unless circumstances require cross suspension, in which case the very neat diagonal suspension, due to J. C. Henry, is the cheapest and most convenient method for light work. This is shown in Fig. 119. Here A, B, C, D, K, etc. , are the poles set in the usual way, 100 to 125 ft. apart, but alternately on either side of the track. The suspension wire is strung from pole to pole, as shown, and the trolley 244 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. wire hung from it in the ordinary manner. It is a very neat and cheap arrangement where only a trolley wire of moderate size has to be carried. For either this or bracket construction fifty poles per mile are sufficient, costing, laid down, say, $125. Setting in country districts ought not to cost more than $1.50 per pole, bringing the pole line in place to about $200 per mile. The trolley wire would cost about $275 per mile, and suspension wire, insulators, brackets and so forth, about $200 additional, making about $675 per mile for overhead structure and material. For bonding the track and suspending the trolley wire, to- gether with incidental expenses, $300 to $400 per mile is sufficient. Bringing these items together we may say roundly that with rigid economy $900 to $1000 per mile will pro- vide the electrical structure and connections for such a road as that under consideration. Taking the larger figure we see that the electric narrow gauge track can be built complete ready for traffic for about $4500 per mile. For a ten mile road, the car equipment should be, say, two motor cars with an extra motor in reserve and four freight skips and a couple of freight cars of a larger size. The whole outfit should not cost over $5500 delivered and ready for action. Now for the station and other equipment. A genera- tor of, say, forty kilowatts, and a fifty horse power engine and boiler equipment is sufficient. Boiler and engine should be of the simplest kind and the whole plant as com- pactly arranged as possible, since it should ordinarily be operated by a single capable man. Engine, boiler and generator set up ready for operation should not cost in the aggregate more than $4000. This is enough to provide a thoroughly well built, durable equipment on which the re- pairs should be very light. -A combined power station and car house, with iron stack for the boiler, should cost complete not over $2000, and $500 more would provide waiting rooms and freight platforms at the ends of the line. INTERURBAN AND CROSS COUNTRY WORK. 245 Altogether these items of construction and equipment would aggregate $12,000, or $1200 per mile. Bringing together the various items reduced to the basis of cost per mile, we have for a ten mile road" Roadbed and track $35oo Electrical construction 1000 Rolling stock 550 Power station and buildings 650 Total l57oo An addition of $300, bringing up the total cost to $5ooo per mile, would provide for all normal contingencies of construction. It is safe to say that in most situations a good narrow gauge electric line can be built and equipped for this sum if right of way can, as would nearly always be the case, be obtained along the public road. This is a reduction of about $4000 per mile over sim- ilarly close figures for a cheap ordinary electric road, a dif- ference that would turn the scale from loss to profit in many country localities. The cost of operating such a road is correspondingly low. The hours of running need not be eighteen or twenty as in street railways, but could be so reduced that the work could be arranged for a single set of men with- out unreasonably long hours. A total force of six men could operate the line without difficulty. Of these two, the engineer and superintendent who should understand the motors and linework well, would probably have to be paid $75 per month each; the other four could be obtained in most country districts for about $45 per month each. Under ordinary circumstances the mechanical output at the station would not exceed, say, 250 h. p. hours per day. Counting five pounds of coal per horse power hour the daily fuel consumption would be a little over half a ton of coal per day costing, at $3 per ton, in round numbers $600 per year. $400 per year more should cover ordinary repairs and incidental expenses at the power station. An- other addition of $500 should cover taxes and miscellaneous 246 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. expenditures, making in all very nearly $5500 per year as the total expense account, irrespective of depreciation and interest. Roads such as we are considering have the advantage of being able to charge relatively more than urban lines, and with a tolerable passenger service, express and mail service and freight traffic should be able to pick up a very satisfactory living. The ten mile line in question must show gross earnings of about $9000 per year to pay a fair return on the investment and set aside a tolerable sinking fund practically $24 per day, or $12 per train per day. As each of the two trains should make six or eight single trips per day it appears that the road would pay on gross receipts of $2 per trip, twenty cents per train mile. FIG. 120. It is a lean region indeed that cannot furnish that amount of patronage. But this is by no means the last word on cheap cross country lines. It is quite certain that there are available constructions cheaper than the narrow gauge just described. At least two existing arrangements are capable of a lower minimum cost of construction than that mentioned. Cur- iously enough both of them have been zealously exploited for heavy high speed railway work for which they are not in the least needed, instead of being pushed into a most useful field to which they are well adapted and in which they have decided advantages. One of these is the well known ' ' Boynton Bicycle' ' road of which an excellent idea is given by Figs. 120 and 121, INTERURBAN AND CROSS COUNTRY WORK. 247 Fig. 1 20 shows the appearance of the construction across the country. Fig. 121 shows an end view of the narrow, pointed car in position on the single railed track. The upper bearing carried by the brackets extended from the FIG. 121. heavy side poles along the line is merely a steadying rail whose function it is to hold the car upright when at rest and guide it around curves when in motion. In normal running the pressure against this upper guide is trifling. All the weight is carried by the central double flanged wheels on the track rail. The cuts are from pho- 248 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. tographs of the experimental track on I,ong Island. The apparatus here was on a considerable scale, as high speed was attempted without conspicuous success, probably owing to a track too short for speed. Nevertheless, a glance at the cuts shows how readily and neatly the system can be applied to cross country roads with light cars operated at very moderate speeds. Under these circumstances the upper supporting structure having . 122. little strain upon it can be light and cheap, while a mere row of short posts rising just far enough from the ground to assist in the grading may serve to carry a light but rather deep girder rail , quite strong enough for the traffic. The rails would serve admirably as conductors since even a fifteen pound rail is far more than the equivalent of the copper required and the lower rail being off the ground would be little troubled by snow in winter. The supply of power is thus very easy and simple and the cost of grading is in great measure averted. Another construction which can be carried out very cheaply on the scale necessary for cross country roads is the saddleback railway. The Meigs elevated structure is INTERURBAN AND CROSS COUNTRY WORK. 249 s good specimen of this type, which has been successfully worked up as an electric line in the Beecher railway, an experimental section of which has been put in use near Waterport, N. Y. This arrangement is shown in Figs. 122 and 123 cf FIG. 123. which the first shows a car on the experimental track, and the second gives the essential details of the structure. It is, as shown in Fig. 123, a quasi-elevated road com- posed of posts or similar supports carrying longitudinal stringers which support the central bearing rail and the lateral guide rails. These latter serve to steady the car, but are under very little stress when the car is in motion. 250 POWER DISTRIBUTION OF ELECTRIC RAILROADS. As in the bicycle railway, very little grading is nec- essary none under favorable conditions the rails serve admirably as conductors, and the supporting structure is cheap and simple. In the experimental road storage bat- teries were used, thereby throwing away one of the essen- tial advantages in the conductivity of the rails and adding unnecessary weight to the car. For cross country work- ing such a system should be a decided success since it could be carried high enough to be out of the way at cross- ings and takes up singularly little room. Either the bicycle or the saddleback road can be in- stalled even more cheaply than the narrow gauge line just discussed, owing to the practical abolition of grading, free- dom from an overhead trolley construction and full utiliza- tion of the rails as conductors. These roads too are much less liable to trouble from snow and bad weather than the narrow gauge and are equally efficient for the purpose in hand. Under favorable conditions they could be built and equipped for the same service as the narrow gauge for a sum scarcely, if at all, exceeding $5000 per mile for a ten mile line. At a pinch any one of these roads could get along with one man per train exclusive of the men at the power house, thereby giving an electric railway, of which the necessary expenses would be hardly more than $4000 per year, and which would pay fairly on gross receipts as small as $7500 to $8000 per year. The possibilities of such roads for opening up the country are self evident. Throughout the estimates just given it will be noticed that nothing is included for franchise and right of way. This omission is for the very good reason that in the regions to be benefited by such roads, franchise and way would always gladly be given, with not infrequently a sub- stantial bonus in some form or other. Built for cash and operated for profit, such roads offer good prospects for excellent returns on the investment, and their economic value to the country can hardly be over- estimated. Almost nothing has yet been done in this line, but the field is a most promising one. CHAPTER IX. FAST AND HEAVY RAILWAY SERVICE. Up to the present time a large proportion of all elec- tric railway work has belonged strictly to street railway service, a few per cent can be classed as inter urban, and only here and there have there been any serious attempts to beat the locomotive on its own ground. The task is a serious one not to be undertaken without good cause. Our present locomotive is a wonderfully reliable and efficient machine, beautifully adapted for its work, and if it is to be replaced by the electric motor, there must be good cause for the change. The economic relation between the motor and the lo- comotive has been several times carefully investigated with the uniform result of showing, assuming the same condi- tions, no very considerable advantage on either side. It is in the variations in the conditions, the exigencies of traffic of different kinds, that positive economies in favor of electricity or of steam must be sought. Without taking up the application of electric power to universal railway work, there are three classes of service for which it is now admittedly highly desirable, irrespect- ive of any saving reckoned on the horse-power-hour basis, which does not completely tell the story of ultimate profits. In general these three classes have this in common, that in each of them electric power gives positive advan- tage in earning capacity, aside from the saving in operat- ing cost which certainly exists in two of them. The classes in question are as follows: 1 . Heavy local passenger traffic. 2. High speed interurban traffic. 3. Elevated roads, tunnels, and special service. 252 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. In the first case experience has aucady taught the mag- nitude of the inroads made on local passenger service by elec- tric railroads covering the same district. A striking ex- ample of this has recently come to the author's notice, in which a short steam road was actually deprived of more than ninety per cent of its traffic by the operation of a parallel electric system. Near every large city the effect of this competition is severely manifest and is doubly seri- ous by reason of the increasing network of electrics that serves its territory so effectively as to overbalance the ex- tra speed of the railway trains. That which decides the route of the suburban passen- ger, in the absence of any great inequality in fare, is ulti- mately the time taken to travel from his home to his place of business. Convenient termini offset superior running speed, and the electric cars consequently catch the greater part of the traffic. Then too, in the time of the journey must be included probable delays. The net result is that where electric cars and steam railways come into competition for suburban or similar business, the former gets the lion's share. To give good local service, the cars or trains must be frequent, the run- ning time fast and the passengers must be delivered some- where near where they wish to go. In most cases steam roads cannot meet the latter requirement, consequently they must compensate for its lack by fast and frequent service. This means short trains run on short headway, and right here the locomotive is at a serious disadvantage. In the first place the experience of railroads has shown that with increasing numbers of trains the cost per passenger mile increases. For a given amount of traffic carried in a certain territory, doubling the number of trains increases the cost per ton mile something like fifty per cent. That such must be the case is easily to be seen, since the number of passengers per train is halved while the labor per train remains substantially the same, the power per train is not very greatly decreased, and the investment FAST AND HEAVY RAILWAY SKRVICK, 253- and depreciation are increased by using more locomotives for the same service. In point of fact for passenger serv- ice alone the cost per passenger handled would be nearly doubled by doubling the number of trains. If at the same time the running time were to be quickened there would be a still further increase of cost. Largely increased total traffic gives the only opportunity of squaring accounts. In this heavy local work electric traction has very great advantages. The distances are usually moderate, so that all the power can be easily distributed from one or two power houses. The service too, is so dense that the station can be kept well loaded a large part of the time, and consequently working at a high plant efficiency. Hence the total efficiency of the power supply is great, while the abso- lute amount of power required is considerably less with elec- trics than with locomotives, since the former do not have to carry their power stations upon their backs. The re- sults of actual competition have shown the desirability of electric working for suburban passenger traffic, and the character of the service to be given is tolerably obvious. It is necessary for the railway company to take advantage of the weak points of its competitors. Electric street rail- ways have the advantage in the matter of termini and cover their field thoroughly. In speed, however, they are necessarily somewhat deficient and are liable to blockades causing very annoying delays. Hence it should be the object of a competing railway by running frequent trains at high speeds to gain enough time for its passengers amply to compensate them for the time lost in walking at the ends of their route. It is specially necessary to retain the advantage at moder- ate distances, say, up to five miles from the center of the city, for here the competition is the most severe. Fre- quent express trains, while very useful in extending the exterior service, cannot regain the traffic lost within the effective sphere of the street railway. The electrical problem is then to provide frequent trains capable of accommodating one or two hundred people 254 POWKR DISTRIBUTION FOR ELECTRIC RAILROADS. each, running at a speed of twenty-five to thirty-five miles per hour, including stops. In the present state of the art, this is not a serious matter. The only material difficulties that have been met in practice are those connected with the delivery of the necessary current to the moving car, and these are no' now of much moment. The actual amount of power used for such service is easy to compute. Taking for a unit a train composed of one long motor car and one trail car, capable together o* accommodating nearly two hundred people, we can derix the necessary power. The weight of the two cars complete would be about fifty tons of which about thirty tons would belong to the motor car and twenty to the trailef Allowing for ten tons live load the total weight of the loaded train is sixty tons. The tractive power per ton may be taken direct fro' railway practice since the roadbed and rails are, or alway should be, the same ordinarily used in steam railroading For such track and speed the tractive coefficient shou* never be more than 12 Ibs. to 15 Ibs. per ton. Taking tr ^ latter figure as covering all ordinary contingencies of curves etc. , the horizontal effort becomes 900 Ibs. ; to this must be added the air resistance, and whatever resistance may be due to grades. At thirty miles per hour the air resistance is between 3 Ibs. and 4 Ibs. per square foot of surface normal to the direction of motion. Allowing 200 Ibs. for this factor of the resistance we have a horizontal tractive effort of uoo Ibs. and there would be required at thirty miles per hour the expenditure of eighty-eight mechanical horse power. Maintaining this speed of thirty miles per hour on grades, the additional horse power required would be ninety-six for each per cent of grade, or dropping the speed to twenty miles per hour on the grades, sixty-four horse power for each per cent of grade. Allowing about eighty per cent net efficiency from the motor terminals to the wheels it appears that the elec- FAST AND HEAVY RAILWAY SERVICE. 255 trical energy to be delivered to our unit train to maintain a uniform speed of thirty miles per hour is about eighty to eighty-five kilowatts per train on a level track. To maintain a thirty mile per hour schedule under ordinary conditions, including stops and the net effect of such casual grades as might generally be met in suburban work, might require 100 k. w., but the mean daily output per train in service would hardly rise above the original figure of eighty to eighty-five kilowatts. During crowded hours an extra trailer would often have to be carried. This would add about twenty-five tons 700 500 |400 S300 200 100 Street Ry.Journal FIG. 124. to the weight of the trains and would call for about thirty- six additional horse power, bringing the total kilowatts for the train up to nearly 120. This estimate of power, based on known data as to the weights and speed, is fully borne out by experiments on trains in actual operation. Figs. 124 and 125, give the actual power taken to drive trains of five and three cars over a substantially level track at approximately thirty miles per hour. No continuous records of speed were taken, but the averages were about as stated, sufficiently near for a fair comparison. Fig. 124 is the record of a run with a train consisting of a motor car and four trailers weighing, with a moderate load of pas- sengers, very nearly 122.5 tons, a trifle more than double the weight of our assumed standard train. The average 256 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. voltage was 530 and the average amperes are very nearly 290, or 1 54k. w. Since the air resistance for four cars is but a trifle more than for two, the close agreement of this run with our estimate is obvious. Fig. 125 shows a run with one motor car and two trailers weighing with the passengers 89. 5 tons; speed about the same as in Fig. 124 and average voltage 475. The average current appears to be about 230 amperes, giving 109 k. w. total output, which again, reduced to a two-car, sixty-ton basis gives in the vicinity of eighty kilowatts for the nor- mal train. Another run over the same track as in Fig. 125 coo 500 400 0,300 e ^200 100 n h \ i\ K Xt * \j \ \ N N \ ^ ' \ /N I FIG. 125. with a three-car train two tons lighter, and in the opposite direction showed an average power consumption of 125 k. w. The same motor car was used in all three tests. The sudden increases in the current were mainly due to sudden changes at the controlling apparatus causing rapid accelera- tion. These very large momentary currents are, of course, undesirable and can be much reduced by careful handling and better adjustment of the controller to its work. The normal average current for such a train at 500 volts would then be not far from 1 60 amperes. With a working voltage of about 600 at the motors, which is a desirable arrangement, about 135 amperes would be re- quired. One would not go far wrong, then, in taking for ordinary cases 200 amperes as about the largest average which would be called for by any one train, allowing the use of two trailers when convenient. The ordinary loaded FAST AND HEAVY RAILWAY SERVICE. 257 train would average 135 to 160 amperes, according to the voltage. The experience of the past two years on the Nantas- i 1 * ket Beach line has added materially to our knowledge of electric railway work of the larger sort. Fig. 126 shows in detail the result of one of the experimental runs over the entire length of the road, 10.5 miles. The train con- 258 IOWER DISTRIBUTION FOR ELECTRIC RAILROADS. sisted of a motor car weighing 32 tons and a trailer weigh- ing 28 tons. Both were mounted on double trucks with 36 in. wheels, and the former was equipped with two 125 h. p. motors, geared, with a speed reduction of 1.45 to i, to the axles of the same truck. The diagram shows the speed, amperes, volts, watt-hours and time, together with the curves and profile of the road. The power required at an average speed of 29.6 miles per hour was 87.89 k.w. equivalent, taking account of the motor efficiency at this particular output, to about 90 mechanical horse power. A service run with the same train weighing with its passengers 64 tons, at an average speed of 1 7 miles per hour, including twelve stops at intermediate stations, re- quited an average output of 65.2 k.w. In this case the severe work of acceleration, due to the numerous stops, is very evident. In these and many other runs on which careful meas- urements were made, one singular fact regarding the train resistances was noted, which has an important bearing on railway work. The output required for a motor car alone was not greatly increased, for the same speed, by the addition of a trailer. Even two or three trailers produced a dispropor- tionately small increase. The apparent decrease of the tonnage coefficient on long trians has been well known in general railway work, but the ease of exact measurements makes it particularly striking in the case in hand. Of course even here the varying conditions of track, load, speed, acceleration and wind produce somewhat divergent results, but the same general fact is apparent throughout. The apparent power required per ton is much greater for the motor car than for those forming the train. The following table shows the approximate results obtained at several different speeds and under various conditions, in kilowatts per ton. The reduction for air resistance is made by the data from Fig. 135. The value of this resistance has been so thoroughly determined for these very moderate speeds FAST AND HEAVY RAILWAY SERVICE. 259 SPEED MILES PER HOUR. 14* 17** 30 40 2.8 1.44 2.2 2.O Motor car less air pressure. 2-5 1.14 I.4I I.O One trailer 0.71 0-55 0.46 0-5 Two trailers 0.85 0.66 Pour trailers. . . 0-53 * Heavy acceleration. ** Service run. that the above results can hardly be materially in error. Even taking into account the uncertainty introduced by the wind and gear friction, the above results, particularly those at low speeds, show clearly enough that we are here dealing with differences of resistance other than those pro- duced by air pressure. It is altogether probable that the tractive resistance of the driving wheels is materially greater than the pure rolling friction of the other trucks. This is assuredly the case if there is any tendency to slip, and near the limit of adhesion the effect must become very noticeable, which would produce an apparent increase of tonnage coefficient with trains above a certain length and weight. Under even ordinary conditions there must exist a certain grind- ing friction of the driving wheels, much larger, to judge from the data at hand, than ordinary rolling friction, per- haps twice as great. A similar condition is thought by many engineers to hold with respect to locomotive driving wheels. The matter is important since it indicates great ad- vantage in employing trains rather than the single cars which have often been advocated for electric service on a large scale. 260 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. The Nantasket experiments also afford valuable data with respect to the power required for acceleration. They are fairly concordant and show, for a 60 ton two car train, that acceleration from rest to 25 miles per hour in one minute, requires an aggregate expenditure of just about 2 k. w. h., /'. e. about 120 k. w. average output. This gives an experimental basis for computing the power required by a train making frequent stops, as in suburban service on steam railways. We shall calculate a case of this kind later. The important fact to note con- cerning such service is the very severe acceleration due to the short runs between stations and the high maximum speeds that must be reached. With stops every mile or mile and a half this maximum has to be something like twice the average speed including stops, and can only be maintained for a fraction of a minute, after which the brakes have to be applied. For example, if the stations are a mile and a half apart and the running speed is to be 30 miles per hour, the maximum speed must be 50 to 60 miles per hour and it must be reached in one minute or less. This would, for a 60 ton train, demand during ac- celeration an average of 250 to 300 k. w. The nearer together the stations the more severe becomes the work of acceleration due to a given schedule of speed. With stations as near together as they have to be in some ele- vated service the tremendous drawbar pulls are likely to come near to the limit of adhesion if a single motor car be used and there is then considerable to be said in favor of making every car a motor car, in spite of the loss of efficiency. Where certain work must be done as part of the necessary traffic scheme the method that accom- plishes it need not be too closely scrutinized. Single cars or trains with many motors quite certainly take considerably more power to drive at speed than ordinary trains of the same capacity and the advisability of using one or the other is purely a question of local traffic con- ditions. Each case of this kind must stand on its own merits, with the presumption rather in favor of ordinary FAST AND HEAVY RAILWAY SERVICE. 26 1 train practice until it is shown to be inadequate under the conditions imposed. On the other hand if one is compelled to resort to very extreme work of acceleration it is more economical of power to accelerate very quickly and then coast than to accelerate more slowly and cut off current only to put on the brakes. In starting, during certain periods of acceleration and on grades, much more current is required. From Fig. 125 we may judge that the current, even at 600 volts working pressure, might well rise to 400-500 amperes, while to maintain schedule on a grade of, say, two or three per cent would demand fully as much. Altogether the maximum working current per train must be taken as high as 500 amperes, although this amount would be seldom called for. FIG. 127. The supply of so great a current to the moving train is not altogether a simple matter, and has involved consider- able experimentation. The ordinary street car trolley burns badly with such currents, and special wheels arranged to secure extra large contact with the trolley wire are needful, while sometimes two independent trolleys have helped the matter. The trolley wire itself is necessarily of large cross sec- tion, so large as to involve some trouble in support, and several unusual shapes have been tried to improve the con- tact area and facilitate suspension. Fig. 127 shows three such forms, the simpler of which is in use on a portion of the Nantasket Beach electric road, the Cleveland & L,orain Railway and the Boston Subway. Neither shape of the right hand pair is unobjectionable, though both give a good 262 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. opportunity for gripping the wire firmly in the clamps with- out forming projections which would be likely to throw off the trolley when running at high speed. Both are likely to give trouble from twisting, so as to make poor contact with the trolley wheel. The more nearly circular the cross sec- tion of the wire can be made, while still permitting projec- tions or grooves for gripping, the more smoothly the trolley will run and the better for general contact. A plain round wire would be the best if it could be clamped so as not to produce projections to cause trouble at high speed. A grooved round wire with special clamps has recently been introduced with good results. It appears at the left in Fig. 127. Of the two pioneer heavy service roads, one, the Nantasket, uses the two-lobed trolley wire shown in Fig. 127, weighing one pound per linear foot, the other, the Mt. Holly branch of the Pennsylvania Railroad, took for its rather lighter service a No. oo plain wire. 900000 c. m. 900000 c. m. 750000 c. m. \ / 750000 c. m. 1 1 1 | 1 1 1 1 ^Y | 1 1 1 | 1 1 1 \ a b c d e [Cj / g h i j E Street Rj.Jourual FIG. 128. To get a clear idea of the power requirements on this class of road let us assume a fairly simple case and work out the feeder system. Let A B (Fig. 128) be a straight suburban system, 50,000 ft. (nearly 10 miles) in length, with no grades steeper than ^ per cent, double tracked throughout with stations, say, every 5000 ft. Let the power station be at C, the middle point, which would gen- erally be as convenient as anywhere. We will assume trains to be run on ten minute's headway, and to make the round trip in an hour. During the busy hours, 7-10 A.M. and 4-7 P.M. , the trains should consist of motor car and two trailers, at other times of motor car and a single trailer. Certain trains would probably have to carry three trailers. From 8 P.M. on, and before 7 A.M. twenty minute headway would be sufficient. During the busy hours there would FAST AND HEAVY RAILWAY SERVICE. 263 then be twelve trains in service, six of them heavily loaded, and each a three-car train. From the rush hours on the number of trains would be the same as before, until 8 P.M., after which six trains would suffice. From these data we may calculate the power which would have to be delivered. As in other railway work the feeding system is really determined by the conditions of maximum load. This would usually fall between 8 and 9 A.M. during which period six trains would be in service on each half of the line. Of these the outgoing trains would be nearly empty, but on the other hand all the in- going trains would be crowded, and one or two of them would carry an extra car. We must, therefore, allow for extra load, and a fair assumption would be to consider all the trains as three-car trains well loaded. This means not far from 120 k. w. per train, about 1440 k. w. for the full output of the station. The working voltage should be as high as feasible. Without any radical innovations it is quite practicable ta allow a normal voltage of 600 at the motors. This should not be much exceeded, while the pressure may without trouble be allowed to fall ten per cent below this at the termini during heavy loads. Let us first examine the ter- minal conditions. Two trains will ordinarily be handled in that region, requiring by our assumption 240 k. w. To allow for rapid acceleration of a heavy train, fully this amount of power may be temporarily required, but two trains will not have to start together. If, following Fig. 119, we allow 500 amperes available at the terminus we shall be safe so far as this point is concerned. As to drop, if we take ten per cent as average during the busy hours we shall not go far wrong, allowing twenty per cent at the termini during heavy loads. Even a little more would be safe if occasion demanded, so that if the dynamos gave about 600 volts overcompounded about ten per cent, say, to 670, the minimum pressure could be safely taken down 150 volts to 520. We must then have at the termini enough feeder capacity to give 500 amperes with- out dropping the voltage below 520. 264 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. Now for such a road as we are considering the track should be first class, rails not less than eighty pounds per yard, and most carefully bonded. Four lines of eighty pound rails give an equivalent conductivity of about 5, 120,- ooo c. m. Assuming that the bonding lowers the con- ductivity one- third, the track is equivalent to about 3,400,000 c. m. of copper. In spite of this the heavy service makes it necessary to take, say, 14 as the track constant. Now turning to Plate II (p. 81) we can find the feeder area. It is 700,000 c. m. per one hundred amperes for fifty volts drop. In our case then the feeder area is 700,000 X 5 = 1,166,000. 3 This feeder should supply the terminal sections of track, say, 5000 ft. long. For convenience we may divide the line into 5000 ft. sections lettered on Fig. 122. Sections # (and/) being thus disposed of, we may turn to sections b and c. treating them together. The average distance of transmission is 15,000 ft. and the maximum load may be taken as one train under full headway and one starting, say, 650 amperes. From Plate II the copper is 400.000 X 6 5==866>ooocm 3 Similarly, for sections d and e we have approximately 140,000 x 6. 5 = o Now for the working conductors and then to fine down the feeders. Using trolley wire such as is used on the Nantasket Beach road, we should have about 660,000 c. m. available at once in the two trolley wires. Much smaller trolley wire would be inadvisable on account of lack of contact surface and carrying power. Sections d and e will obviously take care of themselves and generally have large capacity to spare. Along b and c the trolley wire is available, and even if the maximum load were at the further end of b a 750,000 c. m. feeder extended from*: along these sections would give FAST AND HEAVY RAILWAY SERVICE. 265 sufficient conductivity. Now for the terminal sections. Throughout a the 660,000 c. m. of the trolley wires is available. Hence up to the beginning of the section 1,000,000 c. m. is sufficient without allowance for help from the other feeder. Just how much this help would be is hard to estimate. It should certainly not be less than 100,000 c. m. If then the long feeders are of 900,000 c. m., the maximum load conditions for the road as a whole will be properly met. We may now count up the copper as follows: Ft. Lbs. Trolley wire 100,000 100,000 750,000 c. m. 40,000 90,000 900,0000. m. 50,000 135,000 Total 325,000 This copper would cost in ronnd numbers $50,000, and in place, including the pole line, nearly or quite $60,000. At average load during busy hours, say, 1800 amperes total, the loss would not be far from ten per cent, while the aver- age loss for the all-day run would be considerably smaller. But this is not the last word on the working conductor question by any means. A daring and apparently highly successful experiment has been carried out on a new section of the Nantasket Beach line 3^ miles long, which promises good results on a larger scale. It consists of the applica- tion of third rail supply to the service track of a steam railroad. The line thus changed was that section of the Plymouth division of the New York, New Haven & Hart- ford Railroad, lying between East Wey mouth and Nan- tasket Junction. An insulated steel rail was placed mid- way between the track rails and made to serve as the working conductor. Current is taken from this rail by means of a soft cast iron shoe carried beneath each of the trucks. The third rail is laid in thirty foot lengths, each supported by four ash blocks, saturated with insulating compound by treatment in vacuum pans. These blocks are so let into the ties that the surface of the third rail is one inch above the track rail. The third rail is bonded J66 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. with two heavy copper bonds at each joint, and where there are crossings the rail is omitted and the cars pass over on momentum. The rail is made continuous electri- cally over the crossings by a buried lead covered cable, and a sloping leading-block of hard wood is spiked to the ties at each side of the crossing to prevent shock to the shoes. The arrangement of the third rail and the contact shoe is shown in section by Fig. 129, and an elevation of a single shoe in Fig. 130. The supply rail weighs ninety-four pounds per yard FIGS. 129 AND 130. and is of rather odd shape, to secure sufficient weight without making the rail too high, and to shelter the in- sulating blocks. The shoes are a little more than one rail length apart, and are supported, as shown in Fig. 130, by a double toggle joint having a rather limited play. The weight of the shoe, about twenty pounds, is enough to give good contact. ' The return circuit is, of course, through the track rails, which weigh about ninety pounds per yard, and are thoroughly bonded with short lengths of copper cable. As a matter of fact, during some weeks of successful operation FAST AND HKAYY RAILWAY SERVICE. 267 the bonding was incomplete, and contact was furnished by the fishplates at many of the joints. The system has now been working several seasons with entire success. The cars, which run over the entire Nantasket Beach road, are, of course, equipped with an overhead trolley as well as with the contact shoes, and from Nantasket Junction to the FIG. 131. Pemberton terminus the overhead trolley line is in use. The character of the overhead structure in this part of the line is well shown in Fig. 131. The greater neatness and simplicity of the third rail arrangement is obvious. Until this experiment fear of serious leakage has deterred engineers from using such construction on ordinary road- beds. A regular railroad construction with rails carried on ties slightly above the surface of .the ground is very 268 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. much less liable to leakage than street railway construc- tion with nearly buried rails, particularly since in the former the third rail can be supported on adequate insu- lators. Besides this, an amount of leakage which would be formidable in street railway work may be relatively quite small in the heavy service of a suburban line. The third rail has about 700 insulators per mile, and if they are of tolerably good material, the leakage current must neces- sarily be small even in very wet weather. Tests show that this is so. In ordinary weather the leakage is not serious, and under the worst conditions it still leaves the system in good operative condition. This might be expected, for it is certainly a poor insulator that, even when damp on the surface, would let pass any considerable current under a pressure of 600 volts. If the track is not actually sub- merged, the insulation should remain fairly high if the insulators do not deteriorate. Snow is a rather good in- sulator, and if the roadbed is well drained, even melting snow will not cause much inconvenience. Such a third rail structure generally renders feeders quite needless. For a road such as we have been investi- gating a one hundred pound supply rail on each track would give, when well bonded, a total equivalent conductivity of just about 2,130,000 c.m., allowing one-third of the total resistance to be in the bonding. This is almost precisely the equivalent of the available copper shown in Fig. 128. On a longer road, or with heavier service, supplementary feeders would be necessary. The cost of this third rail system is decidedly low. A one-hundred pound rail weighs eighty-eight tons per mile, costing at present prices not far from $2300. Insulators, placing and bonding should not exceed $700 per mile addi- tional. On this basis the third rail system can be installed rather more cheaply than the overhead system and is far simpler to maintain and operate. A sectionalized third rail has been more than once suggested as a remedy for leakage. Whatever may be its FAST AND HEAVY RAILWAY SERVICE. 269 merits for street work, it is disadvantageous in that it virtually throws away the immense conductivity of the supply rail and thus greatly increases the first cost of the line. A fraction of the extra expense applied to careful drainage of the roadbed and good insulation would render sectionalization needless for this particular kind of work. A copper third rail deserves consideration in connec- tion with this class of work on account of its great con- venience in the matter of insulation, ease of placing, and elimination of the bonding difficulty. Its net cost is rather more than that of a steel third rail. The third rail section of the Nantasket line has now been installed about three years and although it has not been in operation during the winter, when the most trying weather conditions would have been encountered, the re- sults have on the whole been so satisfactory that the rail- way company has equipped another of its lines in a singular manner. This is a line extending from Hartford to Ber- lin, Conn., via New Britain, a distance of 12.3 miles. The arrangement of the conducting rail is substan- tially that shown in Fig. 130. The lower edges of the rail are little more than i^in. above the ties and the road is operated throughout the winter thus furnishing a crucial test of the insulation. Since the construction of this Hartford-New Britain line it has been extended beyond New Britain to Bristol a distance of 8.8 miles and be- tween New Britain and Berlin 2.5 miles. The Nantas- ket Beach road has also been extended from East Wey- mouth to Braintree 4.4 miles. These additions after a year or two of experience are strong evidence of the operative qualities of the third rail system, which is used throughout. Still another branch line of the N. Y., N. H. & H. R. R. has been transmuted into an electric road with others to follow. The branch referred to is that from Stamford to New Canaan, Conn., 8 miles long. Here the overhead trolley is used to facilitate exchange of traffic with the Stamford street railway if convenient. The trolley wire is No. ooo and No. oooo and there has oeen no trouble in 270 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. getting adequate contact. The bonding on this and simi- lar lines is worth noting as it seems to be especially good. At each joint are applied a pair of " Crown " bonds of the general type shown in Chap. II. both of leaf copper pro- portioned and arranged as shown in Fig. 132. This bond- ing is applied under the base of the rails, the terminals being forced upwards into holes drilled in the base and the drift pins being squeezed up into place by a special tool. Street Ry. Journal FIG. 132. The countersunk bond terminals are then set up with a hammer so that the pins cannot work loose. Another ingenious innovation in this road is found in the motor trucks. Each of these is fitted with two 175 h. p. motors which are carried by and suspended to a truck frame independent of the car truck proper which rests upon it at the boxes. Thus the whole upper part of the truck is removable leaving the working parts freely accessible. The wooden insulating blocks used at Nantasket have been, so far, fairly successful and have kept the leakage, under ordinary conditions, down to a rather small amount. The insulators, however, have in some cases shown marked deterioration and it is the writer's belief that the leakage will become serious if the use of wood is long continued. Insulators so short and presenting so great surface as FAST AND HEAVY RAILWAY SERVICE. 271 these should be of porcelain if they are to have and main- tain insulating properties such as the conditions demand. On these lines it was necessary to keep the third rail close to the ties in order to avoid striking the fireboxes of locomotives occasionally used on the same tracks, but this proximity is certain to produce some disagreeable results unless insulation is more carefully carried out than it is at present. Moreover the third rail is" in no way protected from accidental contact of any kind, and while the voltage em- ployed, 600 to 700 volts, cannot be condemned as highly dangerous to life it is yet certainly beyond the danger line, and can unquestionably produce grave shocks and death. At least one man has been killed on these circuits and others have been injured. It is not putting the facts too strongly to say that to continue the use of an un- guarded third rail on surface roads approaches criminal negligence. Aside from this question of danger short circuits are very easily produced on an unguarded third rail and consideration of public safety and private conven- ience alike demand the suppression of so dangerous a practice. The facts regarding the leakage encountered on these third rail systems have never been made public. That there is at times heavy leakage admits of little doubt, but the roads have continued operative under rather trying conditions in spite of it. The insulation used seems, however, inadequate and should be assiduously shunned in future work along this line. Nevertheless we have in the third rail system a very important addition to the methods of electrical traction, and one that is capable of being developed far beyond any point which has yet been attained. The use of proper insulators and the allowance of sufficient clearance under the third rail will lead to greatly improved results, and in the last resort careful construction and drainage of the roadbed will prove immensely helpful. When the method has been thoroughly worked out 272 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. it promises to be of very great value in many cases of heavy traction, for it is cheap, simple, easy to apply and gives what most other systems conspicuously lack, an adequate contact area on the working conductor. The greatest difficulties yet apparent in the third rail working are the troubles due to sleet and to failure of the bonding. In sleet and ice a coating of the rails is supplied that is, so far, almost impossible to cut through enough to get good contact, and under these circumstances the third rail lines are sometimes driven back to steam locomotives. It should not be impossible to devise means of cutting through the sleet successfully, but thus far the difficulty has been formidable. Even the thorough system of bonding employed gives considerable trouble from the gradual breaking of bonds under the stress of continual shocks at the joints. The center third rail is particularly bad in this respect, as it takes up the space in the middle of the track so as to make proper surfacing and tamping of the roadbed ex- tremely difficult. A side third rail would be far easier to keep up in this respect. As to bonds the writer is in- clined to think that very flexible cable bonds with the terminals electrically welded to the rail would give relief from the present situation. At all events it is well worth trying. One of the standard open cars of the Nantasket Beach line is shown in Fig. 133. Bach has sixteen seats and will accommodate fully one hundred passengers. Sixteen of these are fitted up as motor cars and similar cars are used as trailers. Each motor car is fitted with two G. E. 2000 motors of the type shown in Fig. 134 arranged for series- parallel control. The cars are fitted with air brakes and air whistle, the air being pumped into a tank by a motor auto- matically controlled by the pressure. The motors are good for over a hundred horse power each at full field, and on the straight level stretch in the middle of the Nantasket Beach line a speed of more than seventy miles per hour has been reached. At such a speed the motion is quite smooth and FAST AND HEAVY RAILWAY SERVICE. 273 the great speed cannot be realized except by timing the car. The normal speed is from twenty to thirty miles per hour in regular service, and the system has proved entirely reliable. For this heavy special or suburban service electric power is singularly well suited. It does the work well, at high efficiency and at moderate cost. Basing an estimate of cost on a normal two-car train, requiring eighty kilowatts FIG. 133. while running and allowing for this eighty kilowatts average output at the station, we can figure readily the cost of power per train mile. The train makes an average of about twenty miles per hour. It thus demands four kilo- watt, hours per train mile. The service is twenty hours per day, and the average load fairly high, probably more than half the maximum. On this basis the power per train mile should not cost, delivered on the line at the station, more than six cents, including station charges of every sort and kind. Two cents additional should cover all charges for the delivery of the power, including the motors. Even. 274 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. more unfavorable conditions than those assumed would generally leave the power charge per train mile at not over ten cents. This is, of course, relatively very much better than street railway practice, but the units are far larger, the service easier in every way, the grades smaller and the work far more controllable and regular. By far the most interesting line of advance in electric railway work is toward long distances at very high speeds. The idea of clipping the wings of Time by doubling our present railway speeds is a very attractive one, not lightly to be cast aside as chimerical. The problem naturally divides itself into three queries: FIG. 134. Can it be done ? How can it be done ? Will it pay ? As re- gards the first question we are now in a position to give a definitely affirmative answer. Suppose we set for our goal a schedule speed of one hundred miles per hour. Under the conditions which may be expected to obtain with ex- press service, the corresponding maximum speed would not have to be very high, probably not over 120 miles per hour. Obviously the attainment of such speed depends on only two things the delivery of sufficient power to the moving locomotive, and the mechanical security of track and rolling stock. In our present express service both FAST AND HEAVY RAILWAY SERVICE. 275 here and abroad trains have within the past few years repeatedly run on nearly level track at the rate of one hundred miles per hour and its immediate neighborhood. This speed has not been maintained for more than a few miles at a time, but it has been accompanied by no special phenomena in the way of vibration, strain on track and rolling stock or rapid increase of resistances. In fact the motion at these high speeds seems to be smooth and the track resistances, if anything, are less than at more mod- erate speeds. Air resistance, once much dreaded, is not very serious, for indicator cards from locomotives drawing trains at ninety miles per hour or thereabouts show a total tractive effort so low (even below ten pounds per ton in some cases) as to leave very little room for atmospheric resistance. Perhaps the easiest way to appreciate the facts is to calculate from the best attainable data the power required to drive a given train at one hundred miles per hour. We shall have to exterpolate with respect to some of our data, but so short a distance as to involve very little uncer- tainty. The normal resistances encountered by a moving train may be roughly classified as friction, grades and air re- sistance. The first mentioned, including all the ordinary tractive resistances, is usually ten or twelve pounds per ton of moving weight on good track. Anything below ten pounds is unusually good and few railway engineers would care to count on anything below eight pounds even under the most favorable circumstances, although lower results are probably now and then reached at high speeds. The atmospheric resistance used to be taken as varying with the square of the speed, but the work of Crosby and recent experiments with fast running trains have made it certain that up to speeds of fully 125 miles per hour the air resistance increases very little faster than the speed. Moreover it can be greatly lessened by shaping the front of the locomotive into a plane or parabolic wedge. Fig. 135 shows the results of Crosby's experiments with whirling 276 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. bodies in addition to several points approximately estab- lished by direct experiments on moving trains. The latter are somewhat uncertain owing to insufficient data concern- ing exposed surfaces, but the results given have been taken as large as the data permit, so that they are over rather than under the real resistances. The data for actual train resistances are in a very badly mixed state. A considerable number of formulae have been deduced from experiments, but as a rule they have not held far outside the experimental limits. Much of the confusion has arisen from trying to take account of 40 50 60 70 80 Miles per Hour 100 110 120 130- Street Ry .Journal FIG. 135. several complex variables in one simple formula. As just noted, the air resistance was at first assumed to vary as the square of the speed and the first efforts at formulae assumed a constant tractive resistance plus a term includ- ing the square of the speed. Now the law of squares assumes in general terms that at double speed, double the number of cubic feet of air are displaced per minute and at double velocity. Now an elastic fluid like air, pushed at a speed far below its velocity for compressional waves, obeys no such simple law. Experiments with pro- jectiles show that the variation of resistance with the ve- locity of the disturbing body changes enormously with that velocity. Of all the early workers Rankine alone FAST AND HEAVY RAILWAY SERVICE. 277 assumed a term in the first power of the speed only, quali- fying it by the assumption that each ton of engine should be reckoned as two tons in computing the weight of the train. A linear formula substantially takes it for granted that doubling the speed doubles the air displaced per minute, but leaves the velocity of displacement unchanged, or that both quantities are by no means doubled, which is probably nearer the truth. Crosby's experiments make it perfectly clear that, at least for bodies no more than one or two diameters long, the air resistance is certainly very close to a linear function of the velocity, and very far from being a function of the second power. For elongated rough bodies moving endwise, such as trains, Crosby's values are probably somewhat low. Not only are there powerful air eddies at the rear if it be blunt, but there are, as an ex- perimental fact, strong inward swirls dragging against the train. In general, formulae based on the second power of the speed give resistance values at high speeds much greater than are actually found, while those based on the first power give two little resistance at very low speeds. Those including both powers are more or less successful com- promises according to the data. Broadly speaking the facts seem to be about as fol- lows: i. Tractive resistances, i.e. journal and track fric- tions considered as a whole, tend to fall off at very high speeds very possibly showing a weak maximum at some moderate speed. 2. Air resistance is nearly a linear func- tion of the speed, with a slight tendency to rise. The combination of the two obviously leads to a shape which can be approximated by either a parabolic or hyperbolic function of the speed in either case of small curvature within the range taken. At high speeds either function approximates to a straight line, while at low speeds the curvature is more manifest. At the speeds with which we wish here to deal the best available formulae, i.e. those best confirmed by ex- periments at very high speeds, are those of Mr. Angus 278 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. Sinclair, Mr. David Barnes and Mr. Vauclain, which are respectively R 2 + .24 V R 4 + .16 V R 3 -f .166 V In which R is the total train resistance per short ton and V the speed in miles per hour. They are intended for' use between 40 and 75 m. p. h. Fig. 136 shows these equations graphically and with them a linear function based on Crosby's air resistances 70 80 Speed M.P.H. FIG. 136. and an assumed uniform tractive resistance of 8 Ibs. per ton at these high speeds, with the equivalent of 100 sq. ft. of normally exposed head surface. This is large enough to take account of eddies and lateral air resistance. All these formulae are for running in still air, and none of them are based on any exact theory of resistances, but merely fit closely the facts around which they have been built. It should be distinctly understood that these formulae apply to trains of moderate length and not to single motor cars such as have been used on electric railways. The different values of the first term in the various FAST AND HEAVY RAILWAY SERVICE. 279 formulae indicate the uncertainty as to the real values of the various forms of track resistances. If, as the writer believes, the sum of these gradually rises and then falls off at very nigh speed, a reason would appear for the ap- parent rapid rise of total resistance at medium speeds which furnished a basis for the large terms in V 2 in the earlier formulae derived for experiments at moderate speeds. All the experiments at 60 miles per hour and upwards show that if there be a term in V s its coefficient is very small. It is, of course, possible that the total air resistance including eddy effects passes through a maxi- mum, or one of a series of maxima as does the resistance of a ship, but only towing a train by a very long cable is likely to bring out the real facts of the case. For computing this and for estimating power at very high speeds it is best to recognize squarely the fact that one is dealing with two distinct classes of resistance, one depending 011 the weight of car or train and the other on head area, both being more or less mixed up with the length and number of cars. The process of calculation is by no means difficult and is probably more accurate than any moderately simple formulae. In all such calculations for high speed work it must be borne in mind that all the facts concerning resistances point to the use of a train rather than to a single car, driven by two or four large motors instead of more smaller ones. From these data we can calculate the power required to drive a given train at, say one hundred miles per hour. We will assume a three-car train, motor car and two reg- ular coaches, weighing complete with passengers 140 tons. This demands no special construction; in fact the less departure from the usual form and appearance of cars the better with respect to securing traffic. It is worth while, however, to give the locomotive a head in the form of a parabolic wedge, which is slightly better than the wedge of Fig. 135, to vestibule the cars snugly, and to build the cars as free from projection? as is consistent with usual models. 280 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. With these precautions the total equivalent sectional area could easily be kept within 100 sq. ft. Nearly all of this, too, can gain advantage from shaping. For the rela- tive resistance of wedge and plane Fig. 135 gives accurate values, while the close agreement of the experiments based on normal surface attests their general accuracy. Adding one-third to the wedge value for 100 miles per hour to take account of plane and irregular surfaces, we have a total atmospheric resistance of ten pounds per square foot. The track resistance we will assume at eight pounds per ton since this value is quite attainable at high speed on a good track, and furthermore was used in computing the points shown on Fig. 135 so that if eight pounds is too low, the air resistances are too high. We may now compute the total train resistance as follows: 140 tons at 81bs., ii2olbs. 100 sq. ft. at lolbs., 1000 Ibs. Total drawbar pull, 2120 Ibs. At 100 miles per hour, 8800 ft. per minute, this means 565 mechanical horse power developed by the motors. This power would be raised to about 1300 h. p. in taking a one per cent grade at the same speed. At 125 miles per hour, the assumed maximum, the air resistance would rise to about thirteen pounds per square foot and the horse power to 733. Even if through increased speed and headwind the air resistance were doubled, the necessary output would still- be below 1000 h. p. We may safely assume that with a nearly level track, 1000 h. p. would suffice for all service conditions, while the normal output would be between 500 h. p. and 600 h. p. Now this output can readily be reached with a power- ful locomotive, and except for the difficulties of firing, the speed mentioned could be maintained by a locomotive with a single car. The advantage of electricity lies with the removal of this difficulty and decrease of useless weight, together with what advantage can be gained from a very large and perfect power plant. That such an output can FAST AND HEAVY RAILWAY SERVICE. 28 1 easily be reached by motors on the motor car admits of no question, since it has already been done by the Baltimore & Ohio tunnel locomotives under more trying conditions, i. e. , moderate speed and enormously heavier trains, thus robbing the motors of the advantage of high rotative speed. As regards track, the best is required and the curves should be very moderate, not less, perhaps, than 2000 ft. radius. But the speeds in question are quite feasible on a well laid and well ballasted track. Dr. P. H. Dudley, prob- ably the greatest living authority on track, designed sev- eral years ago a 105 Ib. rail section which he considered would give a perfectly safe track for speeds as high as 120 miles per hour, and his dynagraph records show, moreover, that for such a track there is a marked saving in power owing to much smaller deflections of the rails. A 140 ton electric train would give much less strain on the track than is now found in the case of fast express trains of approx- imately double that weight. Nor is the driving wheel speed dangerously high With good steel wheels the assumed speed would have to be doubled before the factor of safety would be seriously re- duced. Altogether, the evidence shows that a schedule speed of one hundred miles per hour is quite possible without calling for extraordinary power, unusual material of con- struction or great innovations of any kind. As to methods, divers are available. Ordinary con- tinuous current motors worked at, say, icooto 1500 volts are competent to do the work, but to facilitate distribution and keep down the working current, alternating motors are desirable, monophase preferred if practicable. With polyphase motors the work is not now difficult. The syn- chronous motor with special means for starting is the neat- est if stops are very few, but is impracticable for heavy work of acceleration. For the high rotative speed re- quired it is not difficult to design induction motors with both high power factor and great efficiency, amply capable of doing the work and doing it well. Probably such motors 282 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. offer on the whole the best available means for attaining- the end in view. Four motors of 1 50 nominal horse power each, capable of working up to 250 h. p. without much drop in speed would be fully equal to the work, and such motors can be readily produced at any time, as the size is nothing unusual, and the conditions quite easy to meet. The work- ing voltage should, of course, be kept high; 2500 volts is entirely practical, and this pressure would keep the cur- rent through the trolley contacts down to limits already passed in present practice. It is at least an open question whether under the conditions which would be found on a high speed road of this kind it would not be feasible and advisable to use the whole voltage of transmission 10,000 volts or more on the trolley wire and carry the trans- formers upon the locomotive. A bare wire would be used for the transmission in any event and there is no conclu- sive reason why it should not be carried over the track. Otherwise a large number of large transformers, aggregat- ing several times the capacity of the motors, would have to be distributed along the line. Unless the service is very heavy this is needlessly expensive and increases the items of labor and depreciation. Except for the added weight and complication a rotary transformer on the car, with contin- uous current motors, would prove a very practicable method, as has been more than once suggested. Of course, it might be desirable to use two motors in- stead of four and to vary the arrangement of parts in many ways, but such details have no place here, where merely the general scheme is under discussion. The problem of effective braking is a serious one, but not so serious as at first appears. A well protected clear right of way with no grade crossings is absolutely necessary for speeds like those considered, and ought to be insisted on even for present express speeds. With such a clear track and reduced speed, really running on momentum, in nearing termini, the braking effort required is by no means out of reach. The momentum of a 140 ton train at ,100 miles per hour is less than that of a 300 ton train FAST AND HEAVY RAILWAY SERVICE. 283 at 60 miles per hour and about the same as that of a 300 ton train at 50 miles per hour, and such trains are within the limits of present practice. To be sure, the number of wheels subject to braking would be much less in the elec- tric train, but on the other hand a powerful braking action can be obtained by throwing the motors into action as dynamos through a resistance. It is not difficult to figure the braking action. Assum- ing, from one hundred miles per hour to rest, a coefficient of friction of o. i between brake shoe and wheels, and a brake pressure of 5000 Ibs. per wheel, we have for twenty wheels, eight on motor car and twelve on trail car, a net average retarding effort of o. i X 5 X 20 = 10,000 Ibs. The air resistance would average from our previous computation 500 Ibs., and at least 2000 Ibs. could be counted on from the motors. The total retarding force would then be 10,000 + 500 + 2000 Ibs. = 12, 500 Ibs. The momentum of the train at one hundred miles per hour would be absorbed by this retardation in about 2500 yds. As a matter of fact 140 tons is needlessly heavy for a two- car train, and eventually high speed structures would be built much lighter than this. It is, however, perfectly pos- sible to get the speed without departing from ordinary rail- way construction and the average man at present prefers this to being enclosed in the species of sheet steel projectile that has been thought necessary in many projects for high speed service. We may now take up the line question. The simplest method is to make the working conductor the transmission line, as previously suggested. For the working condi- tions, monophase transmission gives quite as great econ- omy as polyphase, for the immense conductivity of the track gives nearly the equivalent of a perfectly grounded circuit. This statement holds approximately even if the conductivity is taken for alternating currents. The cross section of a pair of 100 Ib. rails is, roughly, 20 sq. ins. which leaves an ample margin even with the necessary reduction. 284 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. us assume a line one hundred miles long con- necting two cities, and six trains in regular service. Using transformers on the motor cars the whole trans- mission problem works out in a singularly simple manner. Using 12 as our track constant and taking 10,000 volts as terminal voltage with 2000 volts extreme drop, a single power station in the middle of the line would do the work very easily. Applying our usual formula for 1000 k. w. delivered 12 X 100 X 26s, ooo c. m. = - - = 159,000. 2OOO Hence a No. ooo wire over each track would do the work easily with not more than 7^ per cent average drop. The total amount of copper would then be about 270 tons, cost- ing, say, $75,000. The transformer capacity should beat a maximum about 1000 k. w. per train, normally not over 800 k. w. This would add a weight of not over eight to ten tons, which can easily be spared from the 140 allowed for. The copper for a polyphase system would probably be in excess of that just figured, but would not vary materi- ally for the purpose in hand. If the distribution were effected by delivering power to transformers along the line the cost of the conducting system would evidently be much increased, for the primary feeding line could not be decreased while retaining the same loss and the secondary working line would have to be of at least the same size to carry the necessary working current. For the same total loss the cost of copper would be more than doubled, and the transformer capacity when distrib- uted along the line would also have to be nearly or quite doubled. In point of total cost there is no comparison be- tween the systems, and it is likely that the maintenance of the former would also be consfderably less, thus giving a double advantage. Speaking broadly, one may at the present time say with certainty, that a maintained speed of FAST AND HEAVY RAILWAY SERVICE. 285 one hundred miles per hour is perfectly feasible as a matter of engineering. It requires no methods that are really un- tried, no apparatus that could not now be furnished by more than one manufacturer and no precautions that have not already been taken in the best steam railway practice. When there is a demand for such speed, that demand can be promptly met, be it for a road 100 or 1000 miles long. Increased length would simply mean a power station every hundred miles or so. Now as to the financial side of such an undertaking. It has been very judiciously pointed out by Dr. Louis Duncan in dealing with the . general question of utilizing electricity upon railroads that no existing road having less than four tracks could well undertake to operate an elec- tric express system, since two tracks must be reserved for local and freight service. While a local electric service and express service might be worked on two tracks the general traffic of a system would require more accommoda- tion. The time is not yet ripe for accomplishing all rail- way service electrically, although there are forerunning shadows of such a probability. For special high speed service, however, there is ample opportunity now. A road between two considerable centres of population with a schedule speed of one hundred miles per hour, would in a very short time either drive competing roads out of the through traffic or force them to the same methods. The longer the distance the more deadly would be the competition of fast service. Such speed would gather to itself much of the traffic if the ter- mini were but a hundred miles apart, but on a run like that between New York and Chicago it would almost monopolize it. In any given case the probability of financial success would turn on the amount of passenger and express traffic between the points concerned. The mere motive power expen.se of the high speed is not serious, nor are the items of repair and depreciation greatly to be feared. 286 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. Assuming an average output per train of 600 to 650 k. w. and a cost of power at the station of about 1.25 cents per kilowatt hour, which should be quite attainable in a station of 4000 to 5000 k. w. capacity, the energy itself should not cost above eight cents per train mile including all station charges. Repairs and depreciation on line, motors and rolling stock, and labor on trains should not more than double this figure, so that fifteen to sixteen cents per train mile should cover the regular charges aside from ad- ministrative expense. As regards cost of roadbed, it varies so with conditions as to amount of grading, number of crossings, cost of labor and so forth, as to defy close es- timation. The rails should not be lighter than ninety to one hundred pounds per yard, preferably the latter, and would cost $10,000 to $12,000 per mile of double track. The overhead electric structure, including the copper for high tension current and the track connections, should not cost more than $3000 per mile. The station complete with steam plant and all necessary electrical apparatus could be in- stalled ready for running for $350,000 to $400,000, perhaps less, for one hundred miles of road. The total cost would thus be for such a section probably not over $15,000 per mile, plus right of way and general construction of road- bed, etc. The total cost would thus be not much in excess of that of a first class steam road in the same situation, and with the volume of first class passenger, mail, special freight and express service to be expected between two im- portant termini, it would nearly always pay if built for cash and operated for profit. For elevated roads electric traction cannot be in the future treated as a luxury, but it must be considered a ne- cessity. Even were it notably more costly than traction by steam locomotives, instead of the reverse, public opinion from now on will compel its use in every new enterprise, and on existing roads will make abolition of the locomo- tive the price of the slightest municipal concession. Aside from this consideration the experience with the FAST AND HEAVY RAILWAY SERVICE}. 28 7 Intramural line during the World's Fair, and subsequent results on the Metropolitan & Lake Street elevated roads in Chicago and several others, have shown, what theoretical investigation had indicated, that for such service electric power is the cheapest available means of propulsion. Elevated service is of a rather trying nature on account of the frequent stops generally about every quarter of a mik and the large amount of power that has to be used MAP OF THE METROPOLITAN WEST SIDE ELEVATED R. R. Chicago, Illinois Street Ry Journal in acceleration. The experiments of Mr. Sprague made on the Third Avenue elevated road in New York established the facts very clearly. It was found that for the ordinary train, weighing from eighty to ninety tons, with a speed reaching, between stations, twenty to twenty-five miles per hour, the average indicated horse power of the locomotives during service was 70.3 reaching an occasional maximum of 185. These great inequalities almost vanished when the whole power for the line was considered. Sixty-three trains were in ordinary use and the mean power, smoothed out by the large number of units, varied little from 4500 in- 288 POWER DISTRIBUTION FOR EXECTRIC RAILROADS. dicated horse power. The coal used on the locomotives amounted to 6.2 Ibs. per horse power hour while in use. With these facts as to power required the electrical conditions are easy to find. The motors should together be able to work readily up to 200 h.p. with a good efficiency at half this output. The average power required is not far from that already computed for the case of suburban service at higher speed and with rather lighter trains. The load, however, is on the whole more uniform on the elevated line, although varying more as regards indi- vidual units. The cost of power should therefore be somewhat less. The first notable electric elevated road in service in this country was the Metropolitan line in Chicago. This road, which went into operation in the spring of 1895, was designed to furnish rapid transit on the west side of Chi- cago. Its general location is well shown on the map (Fig. 137). The portion now in operation consists of 1 3 ^ miles of double track with thirty-two stations. The structure is a substantial one of deep plate girders, admirable mechanic- ally, but very unsightly. The track is of ninety pound rail with massive guard timbers. The electrical equipment, with which only we are here concerned, involved divers excellent and novel features. In the first place the track rails are not bonded together in the usual way, but each rail is bonded in the middle to the supporting structure, thus giving an enormous iron con- ductor for the return circuit. If thoroughly carried out this arrangement is exceedingly effective, although it would be well to bond the track itself as a precaution against bad bonds in occasional rails. The working current is taken from a contact rail lo- cated a few inches outside of and above the track rail. This contact rail is supported about every six feet by blocks of paraffine-soaked wood to which it is secured by clips held in place by wood screws. This rail weighs forty-five pounds per yard and the insulated blocks are six inches FAST ANI> HEAVY RAILWAY SERVICE. 289 square and rest upon iron brackets about one inch high, thus raising the contact rail about seven inches above the general level of the track rails. A rail joint in position is shown in Fig. 138. At the FIG. 13?. joint the rails are held in line by a light fishplate secured by two bolts, and are thoroughly bonded. The bonds are formed of flexible copper strip about -fy in. thick and the full width of the foot of the rail, under which they are FIG. 139. placed and to which .they are secured by two large rivets, one on each side of the web of each rail. Current is taken off this contact rail by chilled cast iron shoes carried on the trucks. One of these is shown ira Fig. 139, which exhibits the arrangement of its parts and 290 POWER DISTRIBUTION FOR ELECTRIC RAILROADS the connecting cable. The general construction is that of a double toggle, and the weight of the shoe is sufficient to ensure contact, no spring being employed. There are four of these contact shoes on each motor car, one at each corner. All are ready for service. Two are normally in contact with the feed rail, and when, as at some switches, it is desirable for this rail to change sides, the correspond- ing shoes go into service. The device works admirably. The feeding system is extraordinary. It is composed of forty-five pound steel rails, like the contact rail, sup- ported on and insulated from the main structure and boxed with boards to keep them out of mischief. These rails are thoroughly bonded, cross bonded when feasible, and are connected to the supply rail about every 300 ft. , of tener at switches and sidings. On the section from the power house to the eastern terminus, about i ^ miles, the contact rails alone are suf- ficient, but westward from the power house run eighteen feeder rails, supplying various sections of the road, and each connected to the distribution board in the power house. On the eastern section current is taken under the Chicago River by lead covered cables laid in a trench dredged in the mud bottom. There are eighteen of these cables, four of 500,000 c. m. each, the others of 235,000. c. m. The motor cars are forty- eight feet long and, except for a steel sub-frame for extra strength, display no re- markable peculiarities. Each is equipped with two G. E. 2000 motors, like those shown in connection with the Nan- tasket road. At each end of the car, occupying half its width and projecting into the car and onto the platform, is a little cab containing the controlling apparatus, auto- matic motor pump for the air brakes, and other acces- sories. The motor car complete weighs about twenty-five tons. The operation of the whole system has been highly successful. The Lake Street elevated road, equipped about a year later, shows some useful modifications, although the general FAST AND HEAVY RAILWAY SERVICE. 2 9 I equipment is much the same. The working conductor is here a forty-eight pound T rail, located much as in the Metropolitan line, but supported on special insulators in- stead of wooden blocks. These insulators have a cast iron base and clamping top, united by a bolt sheathed like a Street Ry.Journal FIG. 140. trolley hanger in dense insulation. This bolt is screwed into the base and secured to the cap by a coarse thread cast in cap and insulation and packed with melted sulphur. Fig. 140 shows the arrangement of this standard with its rail and guard planks. These insulating chairs support the rail every six feet. The bonding of the third rail is with copper strips, secured to the rail with two rivets, Y^ in. in diameter, while the track rails are bonded to the main structure at their middle points. 2Q2 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. The feeders on this road are of copper cable, bare on the structure, but boxed over. They are supported every ten feet by vitrified clay insulators arranged as shown in Fig. 141. Every hundred feet this clay insulator is re- placed by an iron clamp provided with insulating bushing, The cables are of 1,000,000 and 1,500,000 c. m. section. The contact rail is well guarded, in this construction, a precaution that should be carried out on every such road and particularly when a contact rail is used as at Nan- iron supports 100 ft. centers * FIG. 141. tasket for a surface road. The earliest elevated road in Chicago, the so-called " Alley" line has been equipped for electric traction and the New York elevated roads are now taking a similar step. It is highly_probable that copper feeders are in the long run more economical than feeders composed of rails. When freshly bonded the rail feeders just described had about one-tenth the net conductivity of the same weight of copper. At present prices of new rails and copper the total cost of the feeding system is about the same by either method, with the maintenance and depreciation FAST AND HEAVY RAILWAY SERVICE. 293 greatly in favor of the copper. Even if old worn out rails were utilized for the feeders it is an open question whether the extra maintenance would not eventually more than eat up the saving in first cost over copper. Personally the author believes the centre rail con- struction used at Nantasket to be better than any side rail for elevated service, where it can readily be given ample insulation for all ordinary cases. It is quite as easy to put in place, and the great cross section of the. rail is ad- vantageous, since the bonding must be maintained in any case and the extra conductivity costs less than if it were secured by copper feeders. Whatever the construction, a third rail supply system must be protected against danger of accidental contacts, and the insulators must be kept free of conducting material brake shoe dust and the like. On a large system the electrical load is fairly constant and, except for the question of branches, can be considered as nearly uniformly distributed. If the schedule is pre- served, there is unlikely to be any very great massing of cars, so that less provision has to be made for wandering of the load than in street railway service or even suburban service. This simplifies the computation of the conducting system greatly. If the rails are thoroughly bonded to the structure, and preferably also to each other, the resistance of the return circuit is extremely low. A track constant of 1 2 should be quite enough to allow under these circum- stances, and the power demanded should not often average over seventy-five kilowatts per train at the power house. The work of rapid acceleration is the most severe contin- gency that must be taken into account, for elevated roads are practically level. This work will usually be not far from double the average work, at times perhaps consid- erably more. An elevated structure gives an admirable opportunity for the use of polyphase motors, since the three necessary working conductors can readily be provided, and such a system has been several times suggested. In long roads an alternating distribution at high voltage might be 294 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. advantageous, but intrinsically, under the necessary con- ditions of frequent stopping and starting, heavy loads of acceleration and large power units, there is very small reason for abandoning the continuous current motor. If the conditions of distribution demand high voltage feeders, however, a polyphase motor system can be made to meet fully the conditions of service. Aside from elevated service the most promising field for heavy electric traction is in those special cases where FIG. 142. the abolition of the steam locomotive is in itself desirable quite apart from the question of saving. Such cases are plentiful enough, particularly in tunnels and large terminal work in cities. The time is near when cities will defend themselves by legislative enactment against the well nigh intolerable nuisance of scores of smoking locomotives, polluting the air and distributing cinders with lavish pro- fusion. While there was no practical means of avoiding the trouble it was endured, but with the means at hand the people are likely at any time to " get up and biff you/' as the phenomenon was happily described by a certain distin- guished politician lately released from the penitentiary. FAST AND HEAVY RAILWAY SERVICE. 295 Terminal yards in the heart of a city are as at present operated simply an abominable nuisance. Tunnels in ad- dition are often more or less dangerous. Any one who has been through the St. Louis tunnel or the St. Clair tunnel at Port Huron realizes that stalling a train would be a very serious matter, with an unpleasantly good chance for asphyxiation. Ventilation is at best difficult and seldom well done. The now notable experiment of the Baltimore & Ohio Railroad in escaping from the tunnel difficulty has proved FIG. 14;. so successful as to leave no doubts as to the applicability of electric traction to this and all similar work. This tunnel runs through the heart of the city of Bal- timore. It is 7350 ft. long, 27 ft. maximum width and 22 ft. maximum height. Its relation to the transit through Baltimore is well shown in Fig. 142. The old route via the ferry caused continual delays and annoyance and was a constant stumbling block in the way of a fast through serv- ice to Washington. The completion of the tunnel has saved nearly twenty minutes in the time between New 296 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. York and Washington, besides facilitating the general serv- ice greatly. Unfortunately it was necessary to have a grade of nearly forty-three feet to the mile in the main tunnel, and this demanded so great power in hauling the heavy freight service as to make the smoke question exceedingly grave. In attempting to carry it on by steam locomotives just after the completion of the tunnel several men were asphyxiated, and freight service via the tunnel was dropped until the electric equipment was ready. The relatively light passenger service caused comparatively little trouble from smoke. The first electric locomotive went into regular sendee FIG. 144. on Aug. 4, 1895, and has operated with entire success since that date. The total length of the electric run, in- cluding the approaches to the tunnel, is about two miles. The locomotive complete is shown in Fig. 143. It is of standard gauge, 35 ft. long and 9 ft. 6% ins. extreme width, and weighs complete 96 tons. It is composed of two flexibly connected trucks, each having four driving wheels 62 ins. in diameter on a 6 ft. 10 in. wheel base. All the weight is, of course, on the drivers. On each of the four driving axles is mounted a six- pole, direct connected motor of 360 nominal horse power. These motors, shown unassembled in Fig. 144, are not placed directly upon the axle. The armature shaft is a sleeve thirteen inches in exterior diameter, concentric with FAST AND HEAVY RAILWAY SERVICE. 297 the axle, but with a clearance of about i ^ ins. On this armature shaft is carried a five-armed driving spider which bears on lugs on the driving wheels through intermediary rubber cushions. The axle is thus relieved of the direct weight of the armature and there is sufficient flexibility to take up vibration due to irregularities of track. The loco- motive is fitted with air brakes and air whistle, and a headlight at each end. The supply of the immense current demanded by such a locomotive at full load was a difficult matter and the FIG. 145. need was met by a most ingenious and effective, though from our present point of view too complicated and costly, arrangement. This was a species of reversion to the slotted tube used on some of the earliest foreign electric roads, from which current was taken by an interior brush some- thing like a gun cleaner. In this case, however, the tube was built up of two angle irons bolted to a covering strip and weighing about ninety pounds per yard. The channels thus formed are carried on trusses in the open and sus- pended from the arch within the tunnel. Fig. 146 shows the character of the hollow working conductor and the - , UNIVERSITY POWER DISTRIBUTION FOR ELECTRIC RAILROADS. method of supporting it in the tunnel. Current is taken off by a snug-fitting brass shuttle carried on a toggle joint trolley frame, and leading to the motors by a flexible cable. Fig. 136 gives a clear idea of this trolley structure, which in practice does its work exceedingly well. Save for occa- sional trouble before the conductors were cleared of rust and dirt, at the verv first, the arrangement has left little to be desired. The conductor in the tunnel is supported every fifteen feet, and outside the tunnel the spans are thirty to sixty feet. The trolley support has great lateral flexi- bility and the working conductor is normally alongside the locomotive rather than over it. FIG. 146. The power house is near the southern terminus of the electric system and current is taken from it to various points on the line over 1,000,000 c. m. feeder cables. The working conductor is carefully bonded at each joint by two No. oooo bond wires. Now as to operation. After continuous service for more than four years, the system has shown itself to be thoroughly efficient and reliable. Repairs have been light, the working conductors have been easily kept clean by running through a scraping shoe every two or three weeks, the leakage current in spite of the moisture of parts of the tunnel and very dirty insulators has been only about four amperes. FAST AND HKAVY RAILWAY SERVICE, 299 In a sample month of operation, locomotive No. I ran 5168 miles in regular service, hauled through the tunnel 375,000 tons in trains averaging a little over 1000 tons apiece, and did this at a total cost for labor, fuel, maintenance and incidentals, of $2186. This means a cost of 0.58 cent per ton actually hauled, or 42. 3 cent per engine mile. But with the three locomo- tives now in service, the labor expense at the power house is unincreased, while the other expenses increase with the Seconds FIG. 147. number of locomotives in service. The result is greatly to reduce the cost per engine mile, probably to between twenty-five and thirty cents. The cost per engine mile for the freight service of one of the large steam railroad sys- tems is stated to be on the average 26. i cents. , varying on the different sections between twenty- three and thirty-four cents, so that the electric traction does not differ notably in cost from steam haulage, in spite of the fact that the station is necessarily somewhat uneconomical from the frequent periods of light load. The coal consumption during the 300 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. month in question was 294 tons, which shows the unfavor- able load conditions very forcibly. Increased service would improve this notably. The average amperes per train during the same period were 986, showing an aver- age input, at the usual voltage of 625. of about 600 k. w. With a 500 ton train a speed of thirty-five to forty miles per hour could be reached, and on one occasion a 1900 ton train was taken through the tunnel up grade. The drawbar pull in this case reached 63,000 Ibs. Fig. 147 shows the current required for acceleration and running of a moderate train (875 tons) on the grade. The severe character of the work is sufficiently evident, and the effect on the economy of the power station of an intermittent load of this kind is obvious. The plant efficiency with the three locomotives is very materially increased. These General Electric Baltimore & Ohio locomotives were intended for very heavy service at moderate speeds about thirty miles per hour but on a spurt of the engine alone up the grade more than double this has been reached. A radically different type of locomotive intended for a different class of service is the Westinghouse-Baldwin machine shown in Fig. 148. It is built along the lines of a motor car, and is in fact a combined baggage car and loco- motive. It is thirty-eight feet long and eight feet wide and weighs complete eighty tons. The eight forty-two- inch driving wheels are mounted on two trucks with un- usually long wheel base. On each axle is a 250 h. p. geared motor. By this means lighter and cheaper motors can be used than with the direct coupled construction. The gearing is arranged for a full speed of seventy-five miles per hour, as the locomotive is designed for fast passenger service. As in the Baltimore & Ohio locomotives *he motors are arranged for series-parallel control. The problem of distributing power to units of so great capacity as these is serious. For tunnel work and perhaps for general work on special tracks, the centre rail distribu- tion used on the Nantasket road or a corresponding side 303 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. rail is to be preferred to anything as yet proposed for heavy currents. With very high voltage the overhead or side running trolley becomes necessary and with a trolley wire of large section and a pair of trolleys there is little difficulty in operating locomotives of moderate capacity even at 600 or 700 volts. The enormous capacity of the B. & O. locomotives leads to quite exceptional difficulty in taking current. At ordinary voltages the feeder section required at even moderate distances is formidable. To op- erate two locomotives of the B. & O. pattern on a two mile section with the power house at one terminus requires a capacity for delivering the equivalent of about 3000 am- peres at the end of the line. From Plate II, using 16 as track constant, since the conductivity of the track cannot safely be taken as more than twice that of the outgoing system, the feeder area required for a transmission of 10,000 ft. at i oo volts loss is 4, 800,000 c. m. Using loolb. center rails on a double track one gets about 2,200,000 c. m. equivalent conductivity, leaving 2,600,000 to be supplied by supplementary feeders. By allowing a little extra drop this could safely be reduced to, say, two 1,000,- ooo c. m. cables. It at once becomes evident that direct supply at ordi- nary voltages is out of the question, except for relatively very short distances. For more extensive work we are brought back either to high voltage supply with trans- formers and perhaps rotary converters on the locomotive or with a low voltage working conductor supplied from transformers or rotaries along the track. Direct current throughout is barred out by the conditions of practical working except in cases similar to that just described. For heavy special service in yards and tunnels the cen- ter rail is undoubtedly the simplest and most practical method of distribution yet tried, and for such service con- tinuous current motors at 600 to 1000 volts with series-paral- lel control, leave little to be desired. If in the course of development alternating long distance service has to be linked to heavy terminal traffic, a terminal power system at FAST AND HEAVY RAILWAY SERVICE. 303 moderate voltage and relatively low frequency meets the requirements. The growth of heavy electric traction in the past few years has been in the direction of rather long interurban roads worked at moderate speeds, and now and then involving freight haulage by good sized electric loco- motives. Of such practice there are many admirable instances without any material change in apparatus or methods of distribution. The period has been rich in minor improvements and much experience has been ac- quired within a somewhat limited range. The most nota- ble item of growth has been the complete demonstration of the success of the slotted conduit system, which how- ever, is beyond the scope of this work except in so far as it has been noted already. A little latter a new period of activity in methods, such as generally follows a season of standardization, may reasonably be expected. That fast electric trains over long distances are soon coming, no one who is conversant with the art of electric traction can seriously doubt. How extensive such service will be, how far it will supersede present methods, and what methods out of those which are now practicable will survive competitive trial these are questions for the prophet rather than the engineer. PUBLISHED BY The Street Railway Publishing Company, 12O Liberty Street, New York. The Standard flathority of the World ON STREET, ELEVATED, INTERURBAN, SUBURBAN RAILWAYS OPERATING BY ELECTRIC, CABLE, COMPRESSED AIR AND OTHER POWERS* ENGINEERING CONSTRUCTION-OPERATION INVEN- TIONMANUFACTURING-FINANCE ACCOUNTING LAW. SUBSCRIPTION, ----- $4.00 PER YEAR. LIST OF^ WORICS ELECTRICAL SCIENCE, PUBLISHED AND FOR SALE BY The Street Railway Publishing Co., 120 Liberty Street, New York. Hanchett's Electric Railway Motors. The various commercial types of electric railway motors are described in full detail, including mounting and control $2.00 American Street Railway Investments. Financial data of over 900 American city, suburban and interurban elec- tric railways, statistics of operation, details of plant and of equipment and names of officers. Published annually 5.00 Bell's Power Distribution for Electric Railways. Deals with the laying out and calculation of electric railway circuits. Sub-station distribution is also considered. Third Edition. Revised and enlarged 2.50 Electric Railway Motors. By Nelson W. Perry. 238 pages, 60 illustrations 1.00 Pratt & Alden's Street Railway Roadbed. A treatise on the construction of the roadbed, giving data as to rails, methods of track fastening and making joints, guard rails, curves, etc. 135 pages, 157 illustrations 2.00 Reference Book of Tables and Formula for Electric Street Rail- way Engineers. By E. A. Merrill. Second Edition. Flexible Morocco. 128 pages. Interleaved with blank pages 1.00 Shop and Road Testing of Dynamos and Motors. By E. C. Parham and J. C. Shedd. 520 pages, 163 illustrations 2.00 Bell's Electric Power Transmission : a Practical Treatise for Practical Men. The best treatise yet written on the practical and commercial side of electrical power transmission. Third Edition. Revised and enlarged. . . 3.00 Electro-Dynamic Machinery. By E. J. Houston, Ph. D., and A. E. Kennelly, D. Sc. A text-book on continuous-current dynamo-electric machinery for electric-engineering (students of all grades. 981 pages, 232 illustratiqns 2.50 Practical Calculation of Dynamo Electric Machines. A manual for electrical and mechanical engineers, and a text-book for students of electro-technics. By A. E. Wiener. 683 pages, 375 illustrations 2.50 The Theory and Calculation of Alternating Current Phenomena. By Charles Proteus Steinmetz. Third Revised and greatly enlarged Edition... . 4.09 ELECTRO-TECHNICAL SERIES. BT E. J. HOUSTON AND A. E. KENNELLT. Alternating Electric Currents. 225 pages, 77 illustrations $1.00 Electric Arc Lighting. 383 pages, 155 illastrations 1.00 Electric Heating. 290 pages, 86 illustrations 1.00 Electric Incandescent Lighting. 449 pages, 146 illustrations l.oo Electric Street Railways. 367 pages, 158 illustrations 1.00 Electric Telegraphy. 448 pages, 163 illustrations 1.00 Electro-Therapeutics. Second Edition. 452 pages, 147 illustrations 1.00 Magnetism. 294 pages, 94 illustrations 1.00 The Electric Motor. 377 pages, 122 illustrations 1.00 The Electric Telephone. 412 pages, 143 illustrations 1.00 Electrical Engineering Leaflets, Containing the Underlying Principles of Electricity and Magnetism. By Prof. E. J. Houston and A. E. Kennelly. ELEMENTARY GRADE suited to the study of Electrical Artisans, Wiremen and to Elementary Students. 299 pages, 149 illustrations. 8 vo. cloth . 1.50 INTERMEDIATE GRADE suited to Students in High Schools, or Col- leges, and those beginning the study of Electrical Engineering. 300 pages, 140 illustrations. 8 vo, cloth 1.50 ADVANCED GRADE suited to Students taking Technical Courses in Electrical Engineering. 280 pages, 121 illustrations. 8 vo. cloth. New York,1895 1.50 Electricity and Magnetism. Being a series of advanced primers. By Edwin J. Houston, Ph. D. Second Edition. 318 pages, 117 illustrations Electrical Measurements and Other Advanced Primers of Electricity. By Edwin J. Houston, Ph. D. 429 pages, 169 illustrations 1.00 The Electrical Transmission of Intelligence and Other Advanced Primers of Electricity. By Edwin. J. Houston, Ph. D. 330 pages, 88 illustrations 1.00 Electricity One Hundred Years Ago and To-day. By Edwin J. Houston, Ph. D. 199 pages, illustrated 1.00 Gerard's Electricity. With chapters by Dr. Louis Duncan, C. P. Steinmetz, A. E. Kennelly and Dr. Cary T. Hutchinson. Translated under the direction of Dr. Louis Duncan. 392 pages, 112 illustrations 2.50 Miller's American Telephone Practice. A comprehensive treatise, including descriptions of apparatus, line con- struction, exchange operation, etc. Third Edition. Revised and enlarged. 3.00 Dobbs, A. E. Practical features of Telephone Work. Illustrated. 12mo. cloth. New York, 1898 75 Emmet, William L. R. Alternating Current WMring and Distribution. Contains a very clear account of the principles of alternating currents from the practical point of view, and of their distribution and application to lighting and power. Polyphase currents, the load factor, the effects of capacity and induct- ance, and the various losses in distribution, are simply explained, no mathematics beyond arithmetic being used. 76 pages, 29 illustrations. Brno, cloth. New York, 1898 1.00 Universal Wiring Computer. For Determining the Sizes of Wires for Incandescent Electric Lamp Leads, and for Distribution in General without Calculation, with Some Notes on Wiring and a Set of Auxiliary Tables. By Carl Heriug. Third Thousand. 44 pages, 4 charts 1.00 Dictionary of Electrical Words, Terms and Phrases. By Edwin J. Houston, Ph. D. Fourth edition. Greatly enlarged. 900 double column octavo pages, 382 Illustrations 7.00 Pocket edition, 9-ft pages, cloth 2.50 Pocket edition, 945 pages, Ifeather 3.00 THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW AN INITIAL FINE OF 25 CENTS WILL BE ASSESSED FOR FAILURE TO RETURN THIS BOOK ON THE DATE DUE. THE PENALTY WILL INCREASE TO SO CENTS ON THE FOURTH DAY AND TO $1.OO ON THE SEVENTH DAY OVERDUE. LD21-100w-7,'3 . 3 , *'