MODERN DYNAMO ELECTRIC MACHINERY.*

BY

A L E X A N D E R GRAY, M. Sc., Professor of Electrical Engineering, Cornell University, Ithaca, N. Y.

T U R B O - A L T E R N A T O R S .

The history ~7 of the development of the turbo-alternator begins about ~889, when a 75-kilowatt machine was built by the Parsons Company to operate at 48oo revolutions per minute. Some of the other early turbo-alternators developed by this con> 1;any w e r e " ~s

Date Kilowatts R.P.M. Poles Remarks

1889 7.~ 4800 2 Rotating armature type with sur- face ventilation.

1891 150 4800 2 Rotating armature with radial ducts fed by axial holes.

1894 35o 3coo 4 Rotating armature with axial ducts and a fan at one end.

It is of interest to note that, as early as ~894, Parsons had used b~)th radial and axial ventilation; even to-day there is nmch controversy as to which is the most satisfactory for modern turbo-alternators.

By i899 the output had becoine so large and the voltage so high that it was no. longer considered desirable to use the rotating armature type of machine, and so the rotating field type gradually won favor. The Brown Boveri Company, from the start in I9OI, have built the smooth drum type of rotor, shown in Fig. 57, their purpose being " to subdivide the field winding into a num- ber of slots with a view to distributing the centrifugal force of the winding, the whole of the field iron being thereby brought into close proximity with the armature surface and the space between the usual salient polepieces being utilized.

Up until I9O 5 two methods were used to cool these machines. In the older method standard slow-speed practice was followed,

* Continued from page a26, August issue. ~7 ,, The Engineering Evolution of Electrical Apparatus," Lamme, Electric

Jour~lal, vol. if, p. 221, I914. ~8 ,, H igh - speed Electr ical Machinery ," Storey and Law, Journal of lhe

h~.stitution of Electrical E~tgineers. vol. 4L P. _086, 19o8. 409

4 I O ALEXANDER GRAY. [J. F. I.

the air being stirred up around the machine and the exposed surface made as large as possible. In modern practice, however, the machine is totally enclosed so that it will operate quietly while air is forced through the machine and directed on the surfaces to be cooled.

In those days, also, there was much discussion as to the best type of turbo-rotor. 19 Some had salient poles as shown in Fig. 56 , others were of the cylindrical drum type as shown in

FIG. 56.

\ \ / \ / X I

\ / \ /

High-speed rotor with salient poles.

Figs. 57 and 58 , some were built up out of laminations and others of 2}/2-inch disks mounted on a shaft, some were solid forgings and others were of cast steel. There was much argu- ment as to the relative merits o.f the salient pole and of the cylindrical type of construction, and this subject was finally thrashed out in the discussion of a paper y0 by Miles Walker in I9IO , by Which time' it had become apparent to designers of turbo- alternators of large output that the cylindrical type of rotor was

19,, Steam Turbine Dynamos ," Ellis, Ibid.. vol. 37, P. 3o5, I9o6. ' ~" Design: of Turbo Field Magnets for AlternateCcurrent Generators,"

Walker, Ibid., vol. 45, P. 319, I9IO.

Sept . , I9~7.] ~ { O D E R N D Y N A M O E L E C T R I C ~ A C H I N E R Y . 4 I I

FIG. 57.

/ ' , \

/ ' ' \ g

CyI[ndrical type of rotor wigh parallel slots. In this type the end connections a are well sup- ported, but the bending stress at b becomes excessive at high speeds.

FIG. 58.

/ Y i . . . . . . . . . . . . . .

Cylindrical type of rotor with radial slots. There is no bending force on the teeth, but the support for the end connections is not as good as in Fig. 57.

FIG. 58A

Part of the rotor of a I 2 5 o - k v . a . , 36oo-r.p.m. generator.

4 1 2 A L E X A N D E R GR AY. [J. F. I.

the more satisfactory because of the better utilization of the space in the rotor bore. Many of the different types of con- struction have been weeded out until we now have in this country only two that are used for lnachines of large output : in one case, shown in Figs. 59 and 6o, the ro.tor is forged out of a solid block of steel which is carefully worked and annealed; in the other case it is built up out of plates that are rabbited into one another and held together by through bolts as shown in Fig. 6I. 21

Fro. 59.

Hydraulically forged rotor core and shaft for a 625o-kv.a., 36oo-r.p.m. turbo-alternator.

A cylindrical rotor with radial slots is shown in Fig. 62 in process of winding. One of the most difficult problems with this type of construction is to support the rotor end connections. They are generally held against centrifugal force by a retaining ring as shown in Fig. 58 . The stress in such a ring rotating with a peripheral velocity of 23,ooo feet per minute at ~8oo revo- lutions per minute is I5,ooo pounds per square inch, due to its

21,, S o m e Difficulties o f Des i gn o f H i g h - s p e e d Gene ra to r s , " Field, Ibid., vol. 54, P- 65, 1915 .

Sept., I917.] MODER N D Y N A M O E L E C T R IC ~ ' IACHINERY. 4 1 3

own weight, and, in order that the ring may support the end connections at 2o per cent. overspeed, it must be of high-grade steel, which material may be obtained in the form of a nickel chrome steel weldless ring with the following characteristics:

U l t i m a t e s t r e s s ~ I2o,ooo pounds per square inch. E las t i c l imit ~ mo,ooo pounds per square inch. E l o n g a t i o n ~ I8 per cent. on 2 inches. R e d u c t i o n in a rea m_ 5o per cent.

FIG. 60.

Rotor core of a 625o-kv.a., 36oo-r.p.m. turbo-alternator.

it is true that by using steel for this purpose a magnetic shunt is placed between the poles, and the resulting leakage flux is limited only by the saturation of the material, so that the leakage factor would be large were it not for the fact that the poles are hmg and the useful flux per pole is large.

l_imitations in Modern Tltrbo-Altcrnators.

XVe have seen that the cylindrical type of rotor has proved by experience to be the most desirable for turbo-alternators of large output and high speed, and, in ~rder to tmderstand the develop-

4 1 4 ALEXANDER GRAy. [J. F. I.

~i ! ~

FIG. 6 L

Cylindrical rotor built up out of plates.

Fro. 62.

Rotor of a I2 ,5oo-kv .a . , ISoo-r.p.m., 6o-cycle turbo-alternator.

Sept.,I9IT.] MODERN DYNAMO ELECTRIC MACHINERY. 415

ment that has taken place, we shall find the dimensions of a 60- cvcle machine with an output of 30,000 kilovolt-amp&res at 18oo revolutions per minute, and then show by how much the output would have to be reduced if the regulation and the permissible rise of temperature were as specified a few years ago.

Maximum Permissible Diameter o/ Rotor.--The subject of stresses in rotating disks that are loaded on the external periphery is very complex and is beyond the scope of this article. 22 Turbo- rotors have been built to run at 18oo revolutions per minute with a peripheral velocity of 24,000 feet per minute without the use of special steel, and, taking this as a suitable value, the diameter of the rotor for such a machine is 51 inches.

Width of Rotor Winding Belt.--It may be seen from Fig. 62

FIG. 63.

X & CONSTANT -5

FIG. 64.

.4 f f ..J O 0-.3 /

°-. 2 /-

~.1 /

.I -2 -3 -4- "S -6 .7 -8 .9 I'0 WINDING BELT/POLE PITCH

Effect of the width of the winding belt on the flux per pole.

== " Design and Construction of Steam Turbines," by H. M. Martin, pub- lished by Longmans, Green & Co., in 1913.

4r6 ALEXANDER GRAY. [J. F. I.

that the whole rotor surface is not utilized to carry the rotor winding; there is still room for two more slots per pole at p and q and for one more coil. It will be noted, however, that this coil would span only a small part of the pole: pitch, so that the addi- tional flux produced per pole would not be worth the cost of the additional copper.

In Fig. 6 3 let the rotor be wound uniformly over a belt of width .v and let the air-gap clearance be uniform; then the curve of flux distribution will be as shown, and

the flux density over the surfaceef = kx

the total flux per pole = a constant [,r ( I - : ¢ ) + x2]

= a constant [st ( I - x)]

This is the equation to the curve in Fig. 64, from which curve it may be seen that, for equal increments in the width of the winding belt, the corresponding increments in the flux per po.le become gradually smaller, and that the addition of the last 3 ° per cent. of the rotor copper causes an increase of only 9 per cent. in the flux per pole. There is evidently a point, therefo,re, beyond which the gain in the flux per pole is not worth the price of the additional copper, and, judging from present practice, it would appear to be uneconomical to make the width of the belt greater than 7 ° per cent. of the pole pitch.

Ma.vi.mum Depth of Slot.--Each tooth has to be strong enough to carry the centrifugal force due to its own weight and to that of the contents of one slot, so that if the depth of slot is increased its width will have to be decreased, as shown in Fig. 65, in order to leave enough metal at the root. For one particular value of the slot depth, the slot area (s x d) will have the largest value.

The stress at t, Fig. 65 , is given by the formula: Stress

in pounds per square inch - (rI3--r23) × ( t + s ) × 36o where 2 1 . 5 X I o 6 X / 2~" r2

the dimensions are in inches. If the diameter of the rotor is 51 inches, the speed is I8oo revolutions per minute, and the tooth stress is limited to I4,ooo pounds per square inch; then the maxi- mum slot width for each value of slot depth is as shown in Fig. 66; the variation of permissible slot area with slot depth is also shown, and in this particular case there is nothing to gain by using slots that are deeper than 7 inches.

Sept.,I917.] ~IODERN DYNAMO ELECTRIC ~'IACHI,N'ERY. 417

Number o/ Slots per Pole.--Given that the rotor diameter is 5 [ inches, the winding belt covers 7 ° per cent. of the pole pitch, and the slot depth is 7 inches, let us find the number of slots

FIG. 65.

' I

I

t

71 "( Z~~/!/t '/ 1

FIG. 66.

50. IN.

0.6 -r- 3 '0

I- o,~-~ / ~ZO-4~ a.o Z ~o o.~ / ~, 0.2 < 1.o > S

\

\

2 4- 6 8 I0 d IN INCHES

\ \

\

12 ]4- 16

Variation of the slot area per pole with the depth of the slot.

418 ALEXANDER GRAY. [J. F . I.

that will give the largest excitation, the rise in temperature of the rotor copper being limited to IOO ° C.

The maximum slot area per pole may be obtained from Fig. 66, and this area is independent of the number o.f slots per pole. The part of this space that is occupied by insulating material increases with the number of slots per pole, because the thick- ness of the cell does not change; one can indeed conceive of the case where the number of slots is so large that the slots are not wide enough to hold more than the insulating cell, and the space for copper is zero. On the other hand, the copper loss per slot decreases as the number of slots is increased, and, as the heat generated has to pass through the insulation before it can be dissi- pated from the external surface of the rotor, the temperature

&

M

t E

. . . . . 1

D 1

. . L - -

d c

FIG. 67.

151

drop through the insulation, being proportional to the heat gen- erated per slot, decreases as the winding is subdivided, so that the current density can be increased. There is a point, however, beyond which the effect of the reduction in copper space more than compensates for the increase in the current density, and the problem is to find the number o,f slots per pole for which the amp&e-turns per pole has the largest value.

A short digression will be necessary here to derive a formula for the temperature rise of the rotor copper. The assumption made in this discussion is that all of the heat generated in the part a b c d of the rotor, Fig. 67, is dissipated from the surface ~rD1, or that each part of the rotor gets rid of its own heat and there is no conduction axially along the winding. That this assumption is not unreasonable may be shown as follows:

If it is assumed that, since the copper is a good conductor of

Sept., I917.] MODERN D Y N A M O ELECTRIC M A C H I N E R Y . 4 1 9

heat and is surrounded by insulating material which is a poor conductor, the heat generated in the copper is all conducted to and dissipated from the ends, then the difference in temperature between m and n, Fig. 67, is equal to

f L/2 ;

where i is the current per conductor. 0 is the resistance of a conductor I inch long and I square

inch in cross-section = o.8 x IO -~ ohms at normal operating temperatures.

k is the thermal conductivity of a conductor I inch long and I square inch in cross-section = I I watts per I o C.

therefore p 1 ~ L 2 .

Tmn = ~ ~2 ~- oeg. Cent.

A reasonable value for I/A, the amperes per square inch, is 195o (65o circular mils per amp6re), and for L is 6o inches, and for these conditions T . . . . . 125 ° C.

If, on the other hand, we assume that all of the heat is con- ducted through the insulation, then the heat generated per inch length of slot is equal to

conductors per slot X (current per conductor) 2 N 0~ cross-section of conductor

amp6re conductors per slot = circular mils per amp6re × I.O

since 01, the resistance of a conductor I inch long and I circular nail in cross-section, is equal to I ohm approximately at normal operating temperatures.

The temperature drop through the insulation is equal to

ampbre conductors per slot thickness of insulation I circular mils per ampbre × 2 X slot depth X

where K, the thermal conductivity of insulating material in inch units, is o.oo 3 watt per degree Centigrade.

For a turbo-rotor with 5I,OOO amp6re turns per pole, IO slots per pole, 65o circular mils per amp6re, o.I thickness of insulation in inches, 5-75 effective slot depth in inches,

the temperature drop through the insulation is 45 ° C.

4 2 0 ALEXANDER GRAY. [J. F. I.

The difference in temperature, Tr, between the rotor copper and the air which enters the machine is equal to T. + T~ + Tc where T~ = the difference in temperature between the copper and the iron, which, as already shown, is equal to

ampere conductors per slot thickness of insulation I circular mils per ampbre X 2 X effective slot depth X o.oo3

The thickness of insulation is generally o.I inch. Tb = the difference in temperature between the iron and the air

at the rotor surface is equal to

watts loss per inch length of slot X total slots rr D X K r

where K,, is the temperature rise per watt per square inch of the rotor surface and has a value of about 1o degrees Centigrade for surface velocities of the order of 2o,ooo feet per minute, therefore

amp6re conductors per slot Io X slots per pole Tb = - circular mils per-ampere X pole . . . . . . p i tch

T,. = the difference in temperature between the air at the rotor surface and that entering the machine, a reasonable value for which is 15 ° C.

therefore T,. = amp6re conductors per pole [ 33 circular mils per amp6re (2d) slots p e r pole

io ] q p o l e p i t c h + 15

We are now in a position to find the most suitable number of slots per pole given that the rotor diameter is 51 inches, the winding belt is 0. 7 times the pole pitch, or 28 inches, the slot depth is 7 inches, the stress in the rotor teeth is 14,ooo pounds per square inch, and the permissible value of Tr is IOO ° C.

Copper section per Slots Slot s, F ig . 65 t, Fig. 65 D e p t h of s lo t = (d-f :o .3)

pe r pole dep th wedge = f s -o .a)Xo.85 Inches Inches Inches Inches Squa re inches

4 7 3-34 1.76 2.6 IO.90 6 7 2.2I I.I7 1.75 8.46 8 7 1.66 0.88 1.3 6.70

IO 7 1.33 0.70 1.25 5.22 12 7 I.I I 0-59 1.25 4.22 16 7 0.83 0.44 1.25 2.92 20 7 0.66 0.36 1.25 2.13

Sept.,19~7.] MODERN DYNAMO ELECTRIC MACHINERY. 421

io

Pole p i tch +

0.25 + o.25 + 0.25 + 0.25 + 0.25 + 0.25 + 0.25 +

Am pbre

33

2d X slots per pole

0.94 = I.I9 0.52 ~--- 0 .77

0 .36 ~--- o .6 I

0 .29 ~ 0.54

0 .24 ~ 0.49

o.I8 ~--- 0.43 o . I 4 4 ~ 0 .394

Circular mils turns per pole, Fig. 68 per ampere

31400 . . . . . . . . . . . . . . . . . . . . . . . . . . . 880 4 2 o o o . . . . . . . . . . . . . . . . . . . . . . . . . . . 76o

4 8 7 0 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . 7oo

5IOOO . . . . . . . . . . . . . . . . . . . . . . . . . . . 650 52700 . . . . . . . . . . . . . . . . . . . . . . . . . . . 610 54000 . . . . . . . . . . . . . . . . . . . . . . . . . . . 550 54000 . . . . . . . . . . . . . . . . . . . . . . . . . . . 500

SQAN. 100

,,J so00o g 5o

tA 5 ~- "1-0.000 40

30.000 3 0 P

~0.000 20

< I0.000 I 0

Amp(~re conductors per pole

Circular mils per amp6re

71.5 I IO

139 157 I 73

I 9 7

215

FIG. 68.

.SLOT SEC'I ION MI .LED ER P( LE

• ~,~ ~RE Tt RNS ff R POlE

/ / °ER p,,Le " /

2 4- 6 ~ 10 12 14. 16 ]5 ,SLOTS PER POLE

Effect of the number of rotor slots per pole on the m a x i m u m excitat ion.

The depth of the wedge can be decreased as the slot width is decreased, but in this part icular machine it cannot be made less than 1.2 5 inches, because the project ing ro tor coils have to be supported by a retaining ring whose thickness y, Fig. 5 8, must be sufficient to support the total weight of end-connection copper, and this weight is ahnost independent of the number of slots per pole.

4 2 2 A L E X A N D E R GRAY. [J. F. I.

The copper section per slot was found by assuming that the cell of slot insulation is o.I inch thick and that the insulation between layers occupies 15 per cent. of the available space.

Armature AmpOre Turns per Pole at Full Load.--Fig. 69 shows typical characteristic curves of a modern alternator of large capacity. The air-gap clearance has such a value that the field excitation required to give normal voltage on no-load (namely,

FIG. 69. 1.4

/ 1.3 /

1-2 /

1-i / /

/ 1.0

,.~ 0.9 ",9 ~ 0-8 >

~0"7

g 0-6 F- z

uJ 0-5 r

~1.-- F--O~ 4

o 6 j ~ 1.2 0.3

~. o 0 . 8 0 . a gg • - z 0,÷ O.l g o

/ /

/

x .

I

o3

J / ,., oc

v f

2-2 2.4X K - o b . 2 .4 -6 -8 I.op. 1.2 ] .4 ]-o 1.~ 2-0 EXCITING AMPERE TURN-% PER POLE

Characteristic curves of a modern alternator.

/

op) is just sufficient to cause full-load current to circulate in the armature on short circuit.

The leakage reactance drop = ab. = 14 per cent. of normal voltage.

The demagnetizing amp+re turns per pole = bc. = o .48 ( o m ) . = o.48 (maximum

field a m p fi r e 'turns per pole).

Sept., 1917.] M O D E R N D Y N A M O E L E C T R I C ~ { A C H I N E R Y . 423

This quantity is also"S = o.41 (armature ampere turns per pole x 2) therefore armature ampere turns per pole

= o.59 (maxinmnl ampere turns per pole). = o.59 x 51,ooo with Io slots per pole, Fig. 68. = 30,000.

The corresponding value of ampere conductors per inch of stator periphery is approximately 144 o.

Generated Voltage per Conductof.--If the rotor and the stator teeth are not highly saturated then, since the air-gap clearance is large, the flux distribution over the pole pitch is as shown in Fig. 63, and, for a winding belt of 7o per cent., the average gap density is o.65 times the maximum value. If, then, the stator tooth density is limited to too,ooo lines per square inch, the flux per pole

t 4, = ioo,ooo X i~2- s X pole p i tch X L~ Xo.65

where L, is the net length of iron in the core, t is the width of stator tooth tip, s is the width of stator slot.

The value of t/s will not exceed l.IO for a I3,2oo-volt machine.

The effective electromotive force per conductor is Ec = 2.22 X 4'X f requency X I0 -s

I.I )~ = 2.22 X (IOO,OOO X ~.I xrr 5~X Ln M 0.65) M 60 XlO -8

= 1.88 L n for this par t icular machine .

The Maximum Output.--The maxinmm permissible output for a machine is

K V . A . ~- k b M Ec ampere conduc tors per pole M poles IO -a

where k,,, the winding distribution factor, is equal to 0.96 for a full-pitch, three-phase winding, therefore

K V . A . -~- 0.96 X (I.88 L~) X (30.000 X 2 ) N 4 X Io-~

432 per inch length of net iron.

The net core length for a 3o,ooo-kv.a. machine is therefore equal to 7 ° inches.

Stator I/entilation.--Many schemes have been suggested and used for the cooling of turbo-alternators. In modern practice

" Electr ical Machine Design," Gray, p. 289.

VOL. I84, NO. I IOI--30

4 2 4 A L E X A N D E R G R A Y . [ J . F . I .

the machine is totally enclosed and air is forced through and directed on to the surfaces to be cooled. About IOO cubic feet of air is required per minute per kilowatt of loss, and this will cause the air to rise 18 ° C. in passing through the machine.

A turbo-alternator with a capacity of 3o,ooo kv.a. will have a loss on full load at 85 per cent. power factor of about 76o kilowatts, distributed approximately as follows:

S t a t o r I2R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . = 37

C o r e l o s s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ 23o

S t r a y l o s s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . = IOO

367 R o t o r I ' R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . = 68

W i n d a g e , f r i c t i o n a n d f a n . . . . . . . . . . . . . . . = 325

T o t a l . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . = 760

This machine will therefore require about 76,ooo cubic feet of air per minute, which is the contents of a room IO feet cube emptied about I ~ times every second. This large volume of air should not be drawn from the power-house and then discharged

-

/4-"CENTRIFU6A k

o<.e#: ,o , o. .

FIG. 70.

I

i RIF..R , . g , ; , : . , ,

TUN'R[L FOR AtR

ITO~

Air-ducts for a turbo-alternator.

Sept.,I9~7.] MODERN ])YNAMO ELECTRIC 5'IACHINERY. 425

back again, because it rises I8°C. every time it passes through the machine. It is considered good practice to draw the air from outside of the building, pass it through air washers, and then dwough the machine, after which it is discharged outside of the

Ftc;. 7t.

Turbine room, river station, Buffalo General Electric Company.

building by a suitable duct. Such a scheme of ventilation is partly shown in I~'ig. 7 o. The appearance of the station is then as shown in Fig. 71 .

This large volume of air must be brought into close contact with the heated surfaces, and the two best-known methods of accomplishing this result are indicated in Fig. 72. In one case

426 A L E X A N D E R G R A Y . [J. F. I.

the air is forced through axial ducts across the edges of the laminations, and in the other case radial ducts are provided. In the former case the duct surface is ~5o,ooo square inches, and in the latter case 2oo,ooo square inches are provided, so that the watts of stator loss per square inch of stator duct surface are 2.45 and r.83 respectively; the higher value can be used with the axially ventilated machine, because the temperature drop through

FIG. 7 2 .

11T

Q ]HHI[IIIt HNIIt IIIItlIIHf It

Diagram showing axial and radial ventilation of a 3o,ooo-kv.a. turbo-alternator

Fro. 73.

Stator punching of an axially ventilated machine,

the iron part of the heat path is lower than in the machine with the radial ducts, because, in this latter case, the heat has to flow across the laminations.

Stator Construction.--Fig. 73 shows one of the punchings of an axially ventilated stator. The duct surface must be sufficient to limit the watts per square inch to about 2.45 and the duct

Sept., I9~7.] MODERN D Y N A M O ELECTRIC M A C I t I N E R Y . 427

section such as to limit the air velocity to 5ooo feet per minute, with all air supply of zoo cubic feet per minute per kilowatt of total loss.

Fig. 74 shows the stator assembled and ready to be wound; the air passes axially as shown in Fig. 72 and escapes through the central duct. Fig. 75 shows a machine of greater output, the volume of air being such that three central duets are re- quired for its escape. The stator winding, it will be noted, is

FI(;. 74.

Stator of an axially ventilated turbo-alternator.

thoroughly braced on the ends so that it will be able to with- stand the enormous mechanical forces that are exerted when the machine is suddenly short-circuited. 2.

The first method of cooling turbo-alternators by means of radial ducts was that covered by the Brown patents of IgoI and is shown diagrammatically in Fig. 76 . This method is still used for machines up to about IOOO kv.a. capacity. Air from the

'-'~ " Short-circuit ing of Large Electric Gene ra to r s and the R e s u l t i n g Fo rce s on the A r m a t u r e W i n d i n g s , " Wa lke r , Journal of the Institution of t~lectrical Engineers. vol. 45, p. 295, :9:o.

428 ALEXANDER @RAY. [J. F. I.

rotor fans is forced into the chamber B, whence it passes up both sides of the machine in parallel. In the case of larger lnaclaines, however, the length of path c becomes excessive and it becomes difficult to keep the top of the machine cool. This difficulty is overcome by introducing cool air at several points on the stator periphery, as at A, Fig. 77.

FIG. 75.

Stator of an axially ventilated turbo-alternator.

Fig. 78 shows the stator core for such a machine. Air from the rotor fans passes to the openings A and then through the radial vent ducts to the exits B. Fig. 79 shows the same machine supplied with its winding, which again is thorougly braced.

One other method of cooling used in this country makes use of the radial ducts, but these are fed from the air-gap as

Sept.. 1917.] MODERN DYxa~[o t~LECTRIC )dACHINERY. 429

shown in Fig. 80 rather than from the outer periphery Of the stator. \Vhen the ducts o.f a very long machine are fed in this wav the velocity of the air in the gap becomes very high, and

Fro. 76.

f Venti lat ion of small turbos.

F~(;. 77.

Turbo ven t i l a t ion by means of radial ducts.

when it reaches a value of about i2,ooo feet per minute it becomes desirable to feed the air-gap from the centre as well as from the ends of the machine, additional fans being added to the rotor for this purpose. Such a rotor is shown in Fig. 8I.

4 3 0 [j. F. I. A L E X A N D E R G R A Y .

FIG. 78 .

S ta tor core for a machine vent i ]a ted as shown in Fig. 77.

Fie;. 79.

S ta tor for a machine ven t i l a ted as shown in Fig. 77.

S e p t . , I Q [ 7 . ] ~ / I O D E R N ~ ) Y N A M O E L E C T R I C ~ I A C H I N E R Y . 431

In the 3o,ooo-kv.a. luachine under discussion the total loss on full load is 76o kilowatts, the volume of air required is 76,ooo cubic feet per minute, the air-gap clearance is 1.75 inches, and the area of the air-path along the gap is 2 square feet, so that, if the machine is fed only from the ends, the maximum air velocity will be I9,OOO feet per minute, but will have only half of this value if air is also fed into the gap from the centre of the machine. i f the air is uniformly distributed to the ducts the maximum air velocity in the ducts will be less than 3ooo feet per minute,

Fro. 8o. ,~ 14 ~5 iS 17 i~ 19 ~O 21

DIS(::-HARGE 2' ', AIRINLET DISCHARGE. AIRINLET I

Smal l t u rbo -a l t e rna to r wi th a i r -gap ven t i l a t ion . I . O u t b o a r d bearin 'g b racke t and cap. 2. Air shield. 3. Air-shield r ing, ou te r (gener-

a t o r end) . 4. :Brush holder s tud. 5. Oil deflector. 6. Oil fan. 7. Oil r ings. 8. Cooling w a t e r tube . 9. E n d cover , io . Lining. IZ. Sight-hole plug. I2. Collector r ings . I3. Air- shield r ings ( inner) . I4. F ie ld- re ta in ing r ing. I8. A r m a t u r e coils. I6. A r m a t u r e spider . z7. A r m a t u r e punch ings . I8. L a g g i n g for a r m a t u r e spider . I9. Field coils. 20. Ai r f an a n d end flange. 2I . M a i n bear ing , cap, and connect ion piece. 22. Air-shield r ing, ou te r ( tur- b ine end) . 23. Shaf t and revo lv ing fie]d. 24. Turb ine half coup l ing land shaf t .

except where the air passes through the ducts in the teeth. The velocity of the air is therefore high and the ventilation most effective in the air-gap and tooth ducts, and these are the places around which the losses are most concentrated.

All three of these solutions of the cooling problem are suc- cessful in practice. This is made certain by the insertion of thermocouples or resistance thermometers into the machine so that the tenaperature rise of the hottest part may be determined at any time. This temperature rise we have limited to IOO ° C. in the case of the rotor, and this value set a limit to the permissible

4 3 2 A L E X A N D E R G R A Y . [ J . F . L

rotor excitation and therefore to the stator amp6re turns per pole. Although the number of the stator conductors has therefore

been fixed, their cross-section depends on the permissible tem- perature rise of the stator. I f the coils are insulated with var- nished cloth and mica, the maximum temperature should not exceed ioo ° C. as measured by a thermocouple. If, oi1 the other hand, what is called a fire-proof insulation is used, this maxinlmn temperature may have a value of ~4 o° C., although operating engineers are not all prepared to run machines as hot as this.

Let us then compare two stators, one with a rise of ten>

Fro. 8t.

35,ooo-kw., I2oo-r.p.m. Curtis,~steam turbine. The inner fans feed the air-gap from the ends and the outer fans feed it f rom the centre of the core.

perature of 75 ° C. and the other with a temperature rise of Ioo ° C., which are reasonable figures if the air is passed through an air-washer and enters the machine with a temperature of

25 ° C. In the high-temperature stator the current density is 15 per

cent. greater, the slot is I 5 per cent. shallower, and the tempera- ture gradient through the insulation is 32 per cent. greater than in the low-temperature stator. The flux density in the core is increased Io per cent., the duct surface is decreased Io per cent., the core loss is increased 2o per cent., and the temperature drop at the duct surface is increased 32 per cent. The difference

Sept., 1917.] MODERN DYNAMO F~LECTRIC MACitlNERY. 433

between the two stators is therefore that the low-temperature machine has 15 per cent. more stator copper, Io per cent. mo.re stator iron, smaller loss, and higher efficiency, while the rotor is the same for each machine.

The effect of desired voltage regulation on the output was discussed in the previous article. The maxinmm exci,tation, ore, Fig. 69, is fixed in the case of a turbo-alternator, and the regnlation can be improved only by a reduction in the reactance drop ac and in the armature reaction bc, and this can be accom- plished only by a reduction in the armature ampere turns per pole and a corresponding reduction in ,the output. Fig. 7 2 shows the dimensions of a 6ooo-kv.a., iSoo-r.p.m, machine with a rotor hui!t for a maximum temperature rise of too ° C. with full-load current at zero power factor and for a regulation of 25 per cent. :*~ 85 per cent. power factor.

(?ntputs greater than those at present obtained from a given i rame are possible, provided the regulation is still fur ther sacri- riced and the temperature rise further increased.

(To be contMued.)

Car Nosing. S . A . P, UI~LOCl{. (Electric Railway Journal, vol. 5o, No. 4, P. 149, July 28, I917.) - -Car nosing, which is horizontal oscillation of the car body about instantaneous axes, occurs on tangent track and is usually negligible on curves. In passenger service a slight anaount of nosing is preferable to the rigid transverse shock of non-s~yinging bolsters, but when these oscillations become exces- sive, nosing becomes a question of serious moment. The initial cause of nosing is poor track: that is, low joints and irregular gauge, which force the wheels to slide horizontally, producing a hori- zontal force at the wheel. This horizontal blow is transmitted through the axle to the various members of the truck, and finally to the car body.

Theory and practice both prove that nosing is reduced with long distances from centre to centre of king-pins and with heavy passenger loads. Nosing can be minimized by the following expedi- ents: f t) Improvement of track conditions by levelling the low joints, keeping normal track gauge, and using proper weight of rail. (2) Reducing lateral play in the transverse moving parts by renewal of worn pieces, and breaking up synchronism of the move- ments hv variable damping of the transverse motions in the front and rear trucks. (3) Reducing the overhang of the car body by lengthening the distance, centre to centre, of king-pins and reducing the .overhang of the motors by placing the motors inside of the wheels.