Nov., ~
T H E JOL’R.Y.4L OF I-YDL-.STRIAL
1911
as before. Below is given a log sheet covering a 24hour period of the test. 24-HOKE LOG O F TESTO F 26TH T O
SOOX, A P R I L
cu. ft. Time. 12.00 3.00 6 00 9 00 12.00 3.00 6.00 9.00 12.00
Integrating meter reading. 2498.0 2603.6 2714 3 2831.7 2965.0 3086.7 3215.0 3375.0 3523.2 Totals
BRAVE,P A . 2iTH. 1911.
’rHO.?IAS l f E T E R ,
TemperaCu. f t . Pressure. ture. Pitot .--7 _*_ tube Brave. Uula. Brave. Bula.
Integrating meter.
C u . ft. chart.
...
...
1056000 1107000 1174000 1333000 1217000 1283000 1600000 1482000
1030800 1073400 1134000 1321200 1230600 1302900 1631400 1479000
1056539 1115176 1160802 1311671 1226774 1310577 1614811 1480000
10252000
10203300
10276350
...
NOON, APRIL
... . ...
..
..
140 161 160 -167 166 124 125 143
47 47 47 47 47 47 47 47
46 43 43 43 42 40 40 46
134 157 156 162 162 114 113 135
- - -
The flow during the test varied from 90,000 to 640,000 cubic feet per hour, the pressure from 4 j to 185 pounds gauge and the temperature between 45 degrees and 65 degrees F. The slight variations in readings between the two meters can be accounted for by the distance between them, these errors averaging out for long periods. A summary of the test shows these results : 22-Hour test, April 9th and roth, 1911: Total standard cubic feet of gas b y Pitot tube for 2 2 hours, 8,784>800;total standard cubic feet of gas by electric meter for 2 2 hours, 8,764,000. 8,784,800 - 8,764,000 ____ - 0 . 2 per cent. difference. 8,784,800 . . Service test, April 17th to June 3rd, 1 9 1 1 : Total standard cubic feet of gas b y Pitot tube, 337,546,182 ; total standard cubic feet of gas b y electric meter, 336,732,018.
After the test a n analysis of the gas gave a composition quite different from that assumed for the design, but its specific heat was nearly identical with the specific heat assumed. An inspection afterward showed the meter in perfect running condition. These tests show conclusively that the theoretical grounds on which the meter design is based are correct ; t h a t the meter will operate satisfactprily on air, artificial gas and natural gas with widely fluctuating temperatures, pressures and gas composition giving readings directly in standard cubic feet ; that the accuracy of the meter is maintained for extended periods of time. With the further advantages of ease of checking for error, small space occupied even for large capacities, small maintenance cost and freedom from clogging ’with dirty gas, this meter gives an extremely satisfactory instrument for both experimental and commercial use. LABORATORY OF
THE
CUTLER-HAMMER COMPAXY, MILWAUKEE
------
DESIGN OF A 30-TON INDUCTION ELECTRIC FURNACE.’ B y ALBERT HIORTH. Received Sept. 23, 1911.
The design of this furnace has been based upon the data and results obtained with the original 5-ton Presented a t the twentieth general meeting of the American Electrochemical Society. in Toronto, Canada, September 21-23, 1911.
furnace in Jossingfjord, Korway, as shown in the daily report sheets. As the step from 5 tons to 3 0 tons is a rather large one, i t was necessary to carefully study beforehand the different factors determining the design. The two most important things to be determined were the power-factor, and the energy required t o keep the charge a t the working temperature. The power-factor depends almost exclusively upon the design of the furnace and the periodicity of the supply. K i t h a given resistance in the bath, a certain current is necessary to produce the required temperature. As the bath consists of only one turn round the magnet, the secondary current is equal to the primary current multiplied b y the number of primary turns per leg. The voltage required t o pass this requisite current through tkie primary winding may be divided into two component parts, v i a , one required t o overcome the ohmic resistance (“The Resistance”), and the other required t o overcome the inductive resistance (“The Inductance”). These two factors determine the power-factor of the furnace. With a given inductance the powerfactor increases together with the resistance, and with a given resistance the power-factor decreases as the inductance increases. The total resistance map again be divided into three parts : First, the primary resistance which represents the loss in the primary coils; second, a very small part representing the loss in the iron-core ; third, the secondary resistance representing the resistance of the bath, and which may be termed the useful resistance. This useful resistunce depends upon the following three factors: The conductivity of the, iron a t the working temperature of the furnace; the skin effect; the dimensions of the bath. The first of these factors will be dealt with later; the second, the skin effect, can be neglected for our purpose, as i t may safely be assumed that its effect upon the total resistance will be the same, even if the sectional area of the bath is altered considerably. The third factor, the dimensions of the bath, is of course, the main one. If the skin effect is neglected, the resistance increases proportionally with the length of the bath, and decreases proportionally with the sectional ‘area. With a given periodicity the inductance depends upon two factors: First, the magnetic resistance, viz., the “reluctance” of the iron-core ; second, the magnetic leakage between the primary windings and the bath. The first of these may be neglected, as the magnetizing current of the furnace without any charge is only very small. The second one, the magnetic leakage, depends upon the total flux which passes round any one of the windings (the primary winding and the bath) without also passing round the other; in other words, upon the total flux which escapes through the annular spaces between the two windings.
850
T H E J O U R Z i A L OF I - Y D U S T R I A L A S D E S G I S E E R I - Y G C H E X I S T R Y .
If the radial width of both the bath and the primary coils remains the same, this flux is practically proportional to the total area of the space between the
.
.
-.
Fig. 1.
winding and the bath, and if the distance between these two be not altered, the flux is proportional to - t h e circumference of the bath. Consequently the inductance will increase with the diameter of the coils; if a t the same time the radial width of the bath be increased, the inductance will be somewhat reduced again, but not nearly proportional to the increase in the width. The inductance is also, of course, proportional t o the periodicity, and may be lowered by using a lower periodicity. The influence of this will be more fully discussed later on. The design of the present furnace is shown in Fig. I . The data for the furnace are the following: Mean diameter of the annular space occupied b y the bath, 2.1 meters. Width of this space, 2 0 cm. Depth of the bath with a 5-ton charge, 2 7 cm. (specific gravity of the iron assumed to be 7 ) . The primary coils have 15 turns on each leg, 8 below and 7 above the bath. Each turn has a sectional area of approximately 1,000sq. mm. The resistance of each leg at a temperature of approximately 180’ C. is 0.003 ohm, The weight of copper per leg, 840 kg. The core is approximately 40 b y 5 0 cm., giving a n effective sectional area of 1,800 sq. cm.
Nov., 1911
The weight of the core is approximately 15 tons. The periodicity of the supply is I 2 . 5 complete periods, corresponding t o 2 5 alternations per second. From the daily report sheets recording volts, amperes and kilowatts, taken at, intervals of about half an hour and extending from the beginning of July until the middle of October, 1910, we have selected six a t random, and from these plotted diagrams showing the readings as ordinates, with the time as abscissas. Three of these diagrams are shown in Figs. 2-4. These are for the charges No. 45 (which was the charge witnessed by Professor JOS. W. Richardsx), No. 5 0 and No. 63. During the first two, pig iron was melted and converted into steel, whereas the last one was run for the purpose of refining low-grade steel. In scrutinizing these, curves, it will be found that the electrical conditions have varied considerably, and in fact, more so than one is accustomed t o expect in an induction furnace. This is partly due to the fact that the curves are plotted from actual readings on indicating instruments instead of being traced by recording instruments. I n this way the peaks in the curves are naturally more pointed than would otherwise have been the case, and partly caused by the voltage of the generator being varied a good deal in order to alter the temperature of the bath. Some of the peaks are evidently caused by the
lifting of the upper primary coils in order t o gain access t o the bath, either for inspecting purposes or in order t o work the slag or t o add material to the bath. An 1
These Transactions. 18, 200 (1910).
Nov., 1911
T H E J O U R L V A L OF I A Y D C S T R I A L A.YD Eh'Gf.\-EERI.YG
CHE311STR17.
8jI
example of this last-named disturbance IS shown abnormally low, it is safer t o assume that the readings, In being a t the lower part of the scale, are somewhat very clearly a t 2.30 P M during charge K O 45 this case the voltage has been kept constant and the inaccurate inductance has been suddenly increased, due to the If we consider only the resistance a t the end of a lifting of the upper coils. This causes a drop in the drop run or at the beginning, before fresh material is added, amperage, the power-factor and the energy, although we obtain the following figures for the resistance the voltage was not altered These diaqrams also .from the reports analyzed. E ~ u l dx e n t Charge h'o 8 8 29 29
41 41 45 45 50 50 63
show the variations in the power-factor, the resistance and the inductance These last named have been arrived a t b y splitting the voltage up into its two component parts, as explained on page 849,.and dividing these by the current The results thus obtained represent directly the resistance and the inductance, as defined o n page 849. As has already been explained, the resistance and (with a given periodicity) the inductance are constants which depend upon the dimensions of the bath, the weight of the charge (which depends upon the sectional area o f the bath) and upon the disposition of the primary coils relative t o the bath. Considering the resistance first, this should, in accordance with the above theory, remain the same as long as the weight of the charge is the same. Apart from the variations caused by the adding of fresh material to the bath, there is, however, a marked decrease of the resistance as the process progresses; whether this is due to the alterations in the chemical composition, or whether the conductivity, after a certain temperature has been reached, begins to increase again with a further increase of temperature cannot be stated definitely without further investigations. This decrease of the resistance in the course of the run is shown very clearly on the diagram (Fig. 2 ) and also toward the end of the diagram (Fig. 3 ) . In Fig. 4 this phenomenon is not observed, perhaps because this charge was only for refining purposes. I t appears from the diagram as if the resistance was somewhat higher during the night, when the temperature was lower, but as the inductance a t the same time appears
N-eight, kg
1475 5032 2532 5878 3600 5764 2775 5890 3569 4497 2905
'xctional area sq cm
160
545 273 635 390 623 300 630 386
485 314
Total resistance Ohm 0 0 0 0 0 0 0 0 0 0 0
390 094 128 066 124 070 125 070 125 088 140
useful resistance Ohm 0 0 0 0 0 0 0 0 0 0 0
380 086 120 058 116 062 117 062 ili 080 132
The last column in this table gives the equivalent useful resistance, vk., the total resistance of the bath (twice the resistance of one ring) multiplied by the square of the number of primary turns In order to obtaln this equivalent useful reslstance, the total resistance has to be reduced by an amount corresponding t o the resistance of the primary colls and to the losses in the iron-core. As stated already, the resistance of the primary coils is o 003 per leg = 0.006 for the whole furnace. The iron losses mal; be taken to represent the resistance equal to 0 . 0 0 2 ,
corresponding t o an iron loss of approximately 1-2 per cent., and the .total of these will thus be 0.008. Subtracting these figures from the total resistance, we obtain the equivalent useful resistance given' in the last column. As the resistance varies in inverse proportion t o the sectional area, the curve showing the relation between
852
F
T H E J O U R S A L OF IATDD'STRIAL AA-D EiYGINEERlA'G C H E - W I S T R Y .
the sectional area and the inverse value of the resistance should be a straight line. I n the diagram (Fig. 5 ) the points obtained are marked as dots, and the straight line drawn so as to represent the average of the readings. I n figuring the actual resistance from this curve, one obtains a specific resistance of the molten iron equal t o 1.4 ohms per sq. mm. cross-section and m. length. I n handbooks the value given is somewhat higher, and even as high as 1 . 7 . The discrepancy may be due to inaccuracies, which are quite unavoidable, considering the difficult nature of the measurements, or may be due t o the fact that our case the section of the bath is not absolutely uniform and that, for instance, the central portion of the bath, where the two rings meet, represents a somewhat lower equivalent resistance than the rest. The value
1.40,i t must be remembered, includes all the irregularities in the shape of the bath, and also the skin effect, as far as this influences the resistance. I t can only be used for figuring the resistance of furnaces of similar shape and disposition of the bath. As mentioned on page 849, the inductance is independent of the weight of the charge, and remains constant a t all times. I t will be seen from the diagrams presented that the inductance varies between 0 . 0 9 and 0.10,except when the upper coil is lifted, and during the night (as in case of charge No. 6 3 ) ; b u t as the readings in this case, as already pointed out, are probably inaccurate, we* may neglect this part of the diagram. All the other diagrams correspond to those shown, and the value of the inductance for the old furnace may therefore be taken as 0.095. Using the figures thus established for the resistance and the inductance, we can figure out theoretically
Nov.,
1911
the power-factor of the present furnace with different charges, and obtain the following table: Wlth 3 With 4 With 5 With 6
-
tons cos. $0 = 0 . 8 1 tons Cos. 0.73 tons Cos. tons Cos.
= 0.65
= 0 60
As will be seen, this corresponds very well with the values given on the curves representing the different charges. We are now in possession of the necessary figures t o proceed with the design of the large furnace and t o base i t upon the experience with the old one. With regard t o this design, the first condition, of course, is that the space for the charge must be large enough t o hold 30 tons, but the actual dimensions and shape of the bath must be such as t o give the best possible power-factor. I n order to increase the capacity of the furnace, it is necessary to increase the sectional area of the bath more than the length, as otherwise the diameter of the bath, and consequently the dimensions of the furnace, would become too large. The resistance of the bath will thus be considerably decreased, and a t the same time the inductance will be increased, owing to the larger diameter which increases the area of the space between the primary and the secondary windings. . The power-factor will thus in any case be reduced with the increased capacity, and we have figured that a 30-ton furnace built on the same line as the present one, for single-phase current, with a two-leg magnet and a current supply with 2 5 alternations, would show a power-factor equal to 0 . 2 5 , and that the power-factor even with 1 5 alternations per second would still be only 0.38. It was therefore decided b y me that the 30-ton furnace should be built on the three-phase principle instead of the single-phase. I n this case the weight of the charge per ring in only I O tons instead of 15 tons with a single-phase furnace, and, as will be shown later, the reduction of the power-factor is not nearly so great. I n order to facilitate the tilting of the furnace, the three rings were disposed in one row. This arrangement is not strictly symmetrical, and it is to be expected t h a t the load on the middle phase will be somewhat different from the load on the two outer ones, on account of the difference both in the inductance and in the resistance. With a special supply this is, however, not a very' serious drawback, particularly as the difference between the phases in all probability will not be very great. The dimensions finally chosen for the new furnace are shown' in Fig. 6; from this we may now determiqe the resistance and the inductance in order t o arrive a t the power-factor. Using the figure 1.40, the useful resistance figures out t o be approximately O..OOOI. In order t o arrive a t the total resistance, this has to be increased approximately 7-10 per cent. on account of the losses in the primary winding and the ironcore, and the total equivalent resistance based on one I. winding becomes then 0.0001
T H E JOURiVAL OF ILVDUSTRIAL A N D EA'GI-YEERISG C H E - I T I S T R Y .
Nov., 1911
The inductance of the old furnace was 0.095,or, reduced, to one turn, O . O O O Z I . In the new furnace this will be increased in proportion to the diameter, t h a t is, 0.00021 x 3- _ 0.oooj.
853
ments. If the water-cooling is arranged in the same way and the amount of water increased correspondingly, it may therefore be assumed that the loss of heat, and consequently the energy required to keep the charge hot, will be in proportion to the surface of the furnace. From Fig. 6 we determine the surfaces of the old and new furnace, as follows:
2.1
These values of the resistance and the indixtance would give a power-factor equal to 0.34, which is still too slow. i
a
Inside circular surface (A) Outside surface (E) Upper sllrface (C) Lower surface (n)
Kew furnace.
Old furnace.
2 0 . 5 sq. m. 35 .O s q . m. 28 . O sq. m. 2 8 . 0 sq. m.
8 sq. m. 17 sq. m. 15 sq. m. 15 sq. m.
The surface of the new furnace is consequently between two and two and a half times as large as the old one. The amount i of energy required in the old one to keep the charge in a molten condition during the night was approximately 180 kw., and during the refining energy varied between The minimum zjo kw. and 3jo kw. energy required for the new furnace should consequently be z X 180 = 360, and the 1 maximum, 2 . 5 X g j o = 875. We have assumed 7 0 0 kw., but a t the same time the generator should be constructed in such a way t h a t , b y merely altering the voltage, it can give anything between 400 and 1,000 ---.I kw. With 7 0 0 kw. the current per ring becomes approximately 46,000 amperes, and the current density in the bath will be about 0.34 ampere per sq. mm. The further data of the furnace, based upon the figures arrived a t above, are: Iron-core, 1,200 sq. cm. Weight, about 23 tons. Primary windings; voltage, 2 3 0 volts. Number of turns per leg, 13: disposed 7 below the bath and 6 above. Sectional area, z x 2 0 cm.
-k 1
'
L.-,
. .~ -
- . /*e
-..-
..-
-*-----
-.
Fig. 6.
I n order t o increase the power-factor, it is necessary t o further decrease the inductance by using a lower periodicity. What can be obtained in this way is shown in the following table: Periodicity 25 alts 20 alts. 16 alts. 10 alts.
Power-factor. 0.34 0.41
0.50 0.67
From this table we select a periodicity of 16 alternations per second. To decrease the periodicity further still will hardly pay, seeing that the cost of the generator will thus be further increased, even if the output is reduced on account of the power-factor being improved. This, however, is a matter which should be gone into more closely later on, in connection with the actual cost of the generator. It remains to figure out the necessary energy required for the fumace, the dimensions of the iron-core and the number of primary turns, together with the sectional area of these. As the furnace will be used with a liquid charge, and no melting-down will take place, the amount of energy required is only the amount necessary to keep the bath a t the required temperature. Apart from the small losses in the primary coil and the iron-core, all the energy is converted into heat in the charge itself, and, as soon as a constant temperature is reached, finally conducted to the surface of the furnace and given out from this b y radiation, conduction and through the cooling arrange-
-
I
t
Fig. i .
854
T H E J O U R N A L OF I N D U S T R I A L A N D E N G I A T E E R I S G CHEMISTRY
Weight o f , copper for the whole furnace, about tons. As stated above, the theoretical number of turns is only 13,but in order t o be able t o run the furnace I 3.5
N o v , 1911
The lower coil is embedded in the masonry below the bath, and consists of copper tubes arranged for watercooling. The upper coil may be lifted in order t o inspect the bath
Fig 8
a t full charge. even if the inductance and resistance should prove t o be somewhat different from the values given, two turns should be added per leg, making the number of turns in the upper coils 8. The last
The inner surface of the furnace, marked “A” on Fig. 6, is also arranged for water-cooling. In order to keep the covers for the bath small, the channels were given the form shown in Figs. 6, 7 , 8
Fig. 9.
three turns in each of these coils should be provided with terminals, enabling one t o use a t will 12, 13, 14 or 1 5 turns per leg. In Figs. 7, 8 and g is shown a general arrangement of this furnace, designed on the same lines as the old one.
and 9 . The primary coils have been shaped accordingly t o follow the shape of the bath as closely as possible. The primary voltage has been fixed a t about 2 3 0 volts, as this low voltage greatly facilitates the construction and the insulation of the windings. To
h-ov.,
1yr1
T H E J O L.Ri1-A L OF I-YD I;S T R I A L An'D EL\-GIATEERIATG C H E M I S T R Y .
855
,embed the coil in the masonry of the furnace would, for instance, be impossible with a higher voltage, and also the upper coil would have to be built and cooled in quite a different manner if it had to carry high-tension current. The use of high-tension current
The generator should, according t o the foregoing, be designed for a normal output of 700 kw., 8 periods (corresponding to 16 alternations) per second, 230 volts, 3,540 amperes, and a power-factor of 0.5. At the same time it must be able to give 400 kw. a t 165
would therefore necessitate a considerably increased distance between the bath and the coils, and thus a considerably lower power-factor. The table below shows the principal data for the old furnace and the new one:
volts, and 1,000kw. a t 2 7 5 volts, with the same power-factor, and 2 9 0 and 1 2 5 kw., respectively, with a power-factor of 0.362. The maximum current will thus be 4,300 amperes. The engine or turbine driving this generator should be for a maximum output of 1,550 effective h. p., and run a t a practically constant speed a t all loads between 600 h. p. and 1,550 h. p. This plant should be quite sufficient t o supply energy to the 30-ton furnace, even if the constants of this should vary considerably from the figures given above. I n order to study this, we have prepared the table (Fig. IO) showing what would happen, and the maximum and minimum energy which could be supplied to the furnace, in case either the resistance or the inductance, or both should vary 2 5 per cent. from the assumed value, either up or down. This table shows the reason why the upper coils should be provided with connections, such as heretofore described, as these coils enable one t o compensate for quite considerable variations, both in the resistance and in the inductance.
Old furnace.
Total capacity.. . . . . . . . . . . . . . . . . . . . . 5 tons Capacity per leg. . . . 2 . 5 tons Diameter of b a t h . . . . 2 . 1 m. Width of b a t h . . . . . . . . . . . . . . . . . . . . . . 2 0 cm. Depth of bath 27 cm. Total surface of masonry.. . . . . . . . . . . . 55 sq. m. Total length of platform. . . . . . . . . . . . . m. Total width of platform.. , . m Sectional area of core per le 1800 s q . cm. Weight of core.. . . . . . . . . . . . . . . . 15 tons Number of primary turns per leg 15 1000 sq. m m . Sectional area of primary windin 0.875 tons Weight of copper per l e g . . . . . . . . . . . . . Total weight of copper. . . . . . . . . . . . . . . 1.75 tons Copper losses.. ..................... Iron losses.. . . . . . . . . . . . . . . . . . . . . . . . Power f Voltage Periodicity (I/? alterations per second) Amperes per phase.. . . . . . . . . . . . . . . . . Kind of current. ....................
2 5 0 kw. 12.3 kw. 18 kw. 0.65 250 121/2 1400 1 phase
New furnace. 30 tons 10 tons 3 m. 3 0 cm. 45 cm. 110 sq. m. 13 m. 6 . 5 m. 1200 sq. cm. 23 tons 13 4000 sq. min. ca. 4 . 5 tons c a . 13.5 tons 700 kw.
40 kw. 18 kw. 0.50 230 8 3540 3 phase
KRISTIANIA. NORWAY.
