[1]Patented May 13, 1947 2 @ 4 2 0 , 5 6 4 UNITED STATES PATENT O@FFIC@@E 2,420,564 'PROCESS OF PRODUCING-FERROSILICON Percy B. Royster, Montclair, N. J. No Drawing. Application April 27,1942, Serial No. 440,703 7 Claims. (Cl. 75-133.5) Th-'Ls invention Y61ates to the production of iron-silico-n alloys,4of silicon content higher than 20% by weight, in a,fuel-fired shaft furnace, e. g., a blast furnace. It relates to the production of stg,nda,,@d ferro-silicon of 50% and 70% grades 5 (silicon content) and also to the production of sv-b,sto,ntially -pure silicon (termed herein "ferrosilicon of 100% grade"). Iron-silicon alloys containing less than 5% to 7% silim-q are generally classed as "pig iron." 10 Allo,ys cc@,it-aining from 7V,, to 16% and 18% silico-ki are variously termeti "silvery iron," "Besser,-.ter'ferrosilicoii," "Di)r@tron," elue. In the following description, the term ferrosilicon unless otherwise defined, will refer to metal containing 15 not less tl,,..avi 20% silic<)r-,, vvith relatively low content of alloying elements (Mn, Cr, P, S, etc.) other than iron. "Silvery iroii" c6ntaining -from about 14% to ?.bout 17% of silico-n haq been prodiiced for many 20 years in blast furnaces, ,ind it @4s ino+, improbable tha,t at infreq-Lient , ilitervals of Qperatinn silvery ,Iron of 18 % to 19 % silicoti has been so prodliced. So ffir as I have been able to a@seertain. Petal contairiing as much as 20% silicon has n.ot been @5 rnade in the blast furnace blown with atrfiospheric air, and most metallurgists had con,,3idered 20% to be an upper limit for blast furna,ce .gilicc,n ec)ntent: higher silico-l- a-Iloys heretofore @,;c-,re considered to fall within the exclusive do,- 10 ,na,in of tbe electric furnace. T ron-,,@ilicide has the comnosition Fe3Si and or-irtains 14.32% by weight of silicon and 85.68% of iron. @Vhen t-@ie silicon content of an irovisiller,,-i ,illoy is carried above 14.32% 'the - molecu- 35 la.r com-oosition of the m.etal coy-)sists of silicon d,ssolved in iron-silicide. For exar@,iple, a 20% @ 'con inetal, which ig taken herein as 9, dividsill ing lipe laetween "silvery iron" and "ferrosilicon," has the compgsition .33.2 molar per- 40 cert silicon and 66.8 molgr percerit iron-silicide. Th-@s acceptp-ci. up.Per iimit of blast furnace silicon co-i,responds to one-third silicon dissolved in twothirds Pe3Si. it may be th6ught to corresporid to P. hypothetical silicide Fe2Si (20.1.% silicon by 45 iveight) although experimen,'al evidence of the existence of such a compound is not c6ncltisiiie. Silico,.q is produced in any furiiace (electric or blasil) by the carbon reduction of silica. in tl-ie electric fltrnace process, silica, iron as metallic 50 scrap, aid carb(,@n as coke, are charged, without f u@,, and the reduction of SiO2 is effected by carbon to produce ii-ietal with concurrent production of very littl,, sl,,ig. in rny present process silica is 19uxed with slag@forming constituents and the re- ri, 5 2 duction reaction takes place according to the fbllowingequation: SiO2 (solution) + 2C.(solid):z-4- Si (solution) +2CO (gas) (1) At 25' C. the 8tandard entr6py change ASO of this ree@ction is 86.32 E. U. (entropy units, cal/mol' C.), and the endothetmic heat of reactio-Ti ' (entror)y change) is 148,550 cal ol (9540 B. t, u. per pound of silicbn). When the reaction is carried out at 1557' C. (2834' P.), AI-1, is increased several per cent to 155,900 cal/mol (10,000 B. t. u. per pound of sili@on) in the @circumstance that the reacti6n takes place between C and SiO2 in undiluted liquid state, the silicon produced is in undiluted liquid State elnd the CO producad is a-t the pres8ure of one atmosphere. In these circtimstonees, the entropy change for Reaction I is decreased slightly to 85.30 E, U. (instead of 86.32 E. U. at '25' C.). In the operation of my invention, the r6attants and products are usually diluted ond the CO is at other thaii atmospheric pressure. Wheri dilutio-@i a-@ld pr6ssure changes occur, the following three symbols in @parenthesis, viz., (Si), (SiO2) and (CO), are used to represent the molar conceiitration of silicon iii the alloy, the molar concentration of the silica in the slag, and the pressure of the furnace hearth (in a@tmospheres absolute), respectively. The oquilibrium constant of Equation 'l is given by the expression: k=L- (CO)2 (2) (siO2) when k is a function of the reduction temperature Tc, -determined by the free energy equation -RTc loge k=AH'-TcAS' (3)' ang ng loge to logio, i roducing tho values 15-5 000, 85.30 and L989 for AH', AS' and R, respe@tivelyj Equation 3 becomes: 34,100 T@- 18@70 - loglo 10 (4) Equat@ions 2 and 4 @are suffloient to describe tlie conduct of the reduction -reaction under any conditions in which the reaction may take place in the blast furnace. Ferrosilicon of high analysis, e. g., 50% ferrosilicon, requires a higher redliction terffperature than silvery iron. Fifty per cent ferrosilicon consists of 84.25 molar per cent silicon, and 16.74 'ragler per cent Fe3Si. The value of (Si) in Equation 2 for 50% ferrosilicon is 0.842 instead of 0.332 for 200/,, "silv6ry iron." By Equation 4@ To
[2]3 is increased 401 F. in raising the alloy grade from 20% to 50%. While significant, this increase in Te is less than had generally been supposed. A much greater increase in To is caused by increase in hearth pressure. At the point in the furnace where silicon reduction takes place, i. e., where liquid SiO2 contacts cokecarbon, CO is produced at the carbon-liquid interface, and this CO is undiluted and at hearth pressure. The term (CO) appearing in F-quation 2 refers, therefore, to hearth pressure (measured in atmospheres). The equilibrium constant k varies as the square of the hearth pressure, and any increase in hearth pressure greatly raise's thb temperature of reduction, Te. When a modem American blast furnace is operating with regular stock descent, without "ha.nging" and "slipping," the pressure of the bl9,st measured in the hot blast main averages 13 to 16 lbs./sq. in. gauge (i. e., abovebarometric). The drop in pressure from hot blast main to the furnace iiiterior averages about 2 lbs./sq. in., so that the hydrostatic pre@@sure inside the furnace hearth ranges from 11 to 14 lbs./sq. in. gauge (i. e., 25.7 to 28.7 lbs./sq. in. absolute). At 12.5 lbs./sq. in. gauge (1.85 atmos.) the value of (CO) 2 is 3.43 atmos.2. Bysuitable alteration of furnace design and of blowing procedure, the hearth pressure may be reduced to 2 lbs./sq. in. gauge; the value of (CO) 2 then is decreased from 3.43 to 1.29 atmos.5, (SiO2) and To will remain constant, and (Si) will vary inversely as (CO)2. A blast furnace making 20% silvery iron with a hearth pressure of 12.5 lbs./sq. in. gauge will "equally well" make 41% ferrosilicon at 2 lbs./sq. in. gauge hearth pressure. The phrase "equally well" here means at the same temperature of reduction Te, with the same thermal efficiency and with the same coke consumption (in pounds of coke per ton of silicon produced). A further decrease in hearth pressure to I lb./sq. in. gauge (1.068 atmos. abs.) will raise the "grade" of the ailoy from 41 to 56% without increasing the temperature of reduction T.. By the installation of an "induced draft" blower 'm the exhaust-gas main of the -blast furnace (in the "down comer" or "gas main"), it is possible to lower the pressures of the hearth under that of the surrounding atmosphere, 1. e., in the "vacuum" range. Wherever the term "minimum hearth pressure" is used herein, it is not to be supposed that barometric pressure (i. e., zer(> lbs./sq. in. gauge) is meant to be the "minimum" operation of a furnace with hearth pressure belo the barometric is specifically contemplated in many applications of my present invention. An alternate method of raising the silicon content bf ferrosilicon without increasing Tc, which does not involve decreasing hearth pressure, consists in increasing (SiO2), the molar concentration of silica in the slag. It is important here to define exactly this term "(SiO2)." Slags are classified and controlled, by all -blast furnace opT erators of my acqilaintance, by reference to their "acid" contents@ by which the operators invari_ Pbly mean the "arithmetic sum of the weight per cent of A1203 and SiO2 in the slag." The tel,m "acid slag" as used by such operators ineans "AI203PIUSSiO2greaterthan5O%." Thismethod of thinking is obviously founded on the assumption that bases and acids combine in such a fashion that one pound of either SiO2 or A1203 joins with one pound of CaO or MgO. A slag contain,:ng 25% each of CaO, MgO, A1203 and 2,420,564 4 S!02 in blast furnace parlance is termed "neutral"; it is thought of as being neutral. The blast furnace is operated as if the slag actually were "neutral." The molar percentages of the constituents, considered as a mutual solution of the four uncombined oxides, of course is 23.94% SiO2, 14.35% A1203, 25.82% CaO and 35.9-0% MgO and the mo-lar percentage of acids is only 38.29%, not 50%. It has been shown, 10 however (Feild and Royster U. S. Bureau of Mnes Tech. p,%per, 1919, pages 187 and 189), that in a molten slag the component oxides are not "free" but are combined into silicates, aluminates and alumino-silicates. The molecular composition of i5 this so-called "neuilral" slag therefore is: Molar per cent Forsterite (Mg2SiO4) --------------------- 58 C-.ehlenite (Ca2Al2SiO7) ------------------- 23 Spinel (MgAl2O4) ------------------------ 5.5 20 Calcium aluminate (CaAI204) ------------- 6.8 5.3 calcium aluminate (CasAI6014) --------- 6.7 'fhe slag i@s, in fact, strongly basic, being 81 molar per cent ortlaosilicate of magnesium plus the basic 25 alum,.no-,,,ilicate of lime, aiid gehlenite which functionally is an orthosilicate. Silicon can be reduced from such a slag as that just referred to, but such a reduction does not proceed according to Equation 1. The re30 duction of silicon from gehlenite takes pla-ce in two successive steps. First the reaction Ca2Al2SiO7T-:'@CaAl2O4+CaSiO3 (5), occurs, i. e., gehlenite dissociates into the alu35 minate and metasilicate of lime. This dissociation involves changes in entropy AH' and entropy AS' which are small compared with the values 155,000 cal/mol and 85.3 E. U., respectively, Equation 1, and can be ignored without serious error. 40 The reduction of silicon from calcium metasilicate follows the reaction . CaSiO3+2Cz--@@CaO+Si+2CO (6) exhibiting an entropy change AH' of 192,000 cal/mol (approximately 24% higher than AH' in 45 Equation 1). For the reduction of silicon from calcium metasilicate (wollastonite), T,@ is given by the free energy relation: 50 T,= 42,100 (7) 18.70-logiolc Te from Equatioll 7 is 700' F. higher than Te from Equation 4, i. e., for the reduction of silica when in excess of that combined to form meta55 silicates (including anoithite, which functionally is a metasilicate). It seems necessary to emphasize that the expression "(SiO2) " signifies the "molar concentration of uncombii'led SiO2 in the slag," and is not 60 related to the total silica content of the slag whether the total (SiO2) is expressed in percentage by weight or in molar percentage. Otherwise, the present invention is difficult to explain to the furnace operator who seemingly cannot 65 divorce his iiotions of slag control from his arithmetically simple assumption that onb pound of any kind of base "neutralizes" one pound of any kind of acid, from his theorem that "acid" slags promote silicon reduction, and from his further 70 hypothesis that a slag is "acid" when its weight percentage of A1203 PIUS SiO2 iS greater than onehalf the total. In order to distinguish clearly between the present invention which teaches that (Si) is lin75 early proportional to (SiO2) when To and (CO)
[3]5 are coristant, from presently accepted theories, the following examples are given: ,Slag A Slag B Slag C - Per cent by Weight ofcao ---------------------------- 29.80 33.25 35.85 Algo---@ ------ ---------------- 8.40 4' 20 1.20 A]203 ----- -- -------- ------- 11.50 6' 50 2..50 sio2 ------ -------------------- 49.85 54.85 58.85 S--, ---------------- ------------ 0. 90 2.10 3-20 100.45 101.20 101.60 Lers oxygen equivalent of sulplaul,- -0.45 -1.20 -1.60 100.00 100.00 loo. Go Each of these three slags shows 61.35% "acids" (SiO2+AI203), but the values of "(SiO2)" in the meaning of Equation 2 for A, B and C are quite different. This may be seen immediately from the molecular composition of these slags: Slag A Slag 8 Slag C - Molar Per cent ofWollast,onite (CaSiO3) --- 52.92 48.95 48 56 Enstatite (MgSiO3)--. -------- 28.06 11.26 2'81 Anorthite (CaAl2Si2O8) --------- 15.23 6.89 2.36 Calcium sulphide (CaS) -------- 3.79 8.13 9.52 Silica (SiO2) -------------------- 0.00 24. 78 36. 76 Slag A contains no uncombined silica, (SiO2) =0, its composition consisting solely of calcium and magnesium metasilicates, anorthite, and calcium sulphide. Slag B has a molar concentration (SiO2) =0.2478 (one-fourth "free" silica), and slag C has a molar concentration (SiO2) =0.3676 (one-third "free" silic,,),). These three slags which react so differently with respect to silicon reduction wolild, by the conventionally accepted definition of "acidity," be considered equally acid." The above two methods of producing ferrosilicon by the present invention are, in actual fact, identical because both operate to increase (Si) ot constant Te. For e!,ample, at constant reductioll temperature, the molar concentratio'l of silicon (Si) is given (by rewriting Equation 2) as: (-3;0 @ 2) (9i)=k-cco-)2 (Si) is increased by decreasing (CO)2 (since it appears in the denominator) or by increasing (SiO2.) (since it appears in the numerator). Each is properly described as a meth6d of "increasing the ratio of (Sio2)-to- (CO) 2.11 In carrying ou-t my present invention, I generally prefer to utilize both of these tv;o expedients, i@ e., to increase (SiO2) and to decrease (CO) 2 as far as the litnits of pra&-,ical operation permit. I am seldoin cont----qt, hmvever, to col'ifine the reductio,ii of silicon solely, toi the improvement wliith results ilrom iiiereg,,qi@-ig the ratio of (SiO2) to-(CO)2. Greater intei-isity of silicon reduction rnay be realized with ecgnomic success by increasing Te, i. e., by operating the furnace at higher he,%rth temperatures. It is probably true that ferrosilicon could be made in a blast furn-ace with such b@last temperattires as are now enploved- in rriodern practice, biit since suel'i alloy productio-@i wouid necessitate the consumption of iineconolniP-ally large amounts of coke, no practi.esl importance will attach to this fact@ @Vhen attempting the productio@ti of ferrosilicon in a fuel-fired furnace in corqpetition wi+.h alloy made in an elee!@,420,564 6 mils/k. w. h., it is necessary to attain an approach to fuel economy. This can be realized only by the use of higher blast temperatures than are at present employed, and higher indeed than operators generally have believed could successfully be employed. Before the development of the "Debble" type stove in 1933, knovvn blast heating devices were incapable of heating the blast to temperatures 10 higher than 1500 to 16000 P. Many opezetors in fact were reluctanl, to pue@'n stove heats above 14@50' F. blast teinperature, because of the probable danger of dq-maging the brick work in the stoves. By thp- use of pebble stoves of the type 15 describ,,d in my U. S. Reissue Pa@.e,,it No. 19,757, little d.if-ficuls'Y is encou,- itered in raising blast @@ei---iperatures to, 2600 ard 2800' 7.1 ,vhen the stove is built with ordi-Tiari7 grade fire brick: by consluriict@@'iig the pebble stoves with the (somewhat 20 expensive) high-duty refractories now available iil the rqarket and composed variously o-f MgO, Cr2O3, A1203, ZrO2, ZrSiO4, etc., blast temperatures as high as 3600 and O@800' F. are readily realized. Blast fumace or@era-tion in the "high" 25 hlt-blast i:anr.,e, e. g., froi-n 1600 to 3600' F. is not a problem only of p-,-oducing those blastluemperatures. Difficulties arise which threaten the continuatiori of the entire furnace operation whenever blast temperatures are carried above an 10 i-tppe@, liri-lit which will vary somewhat witi-i the iiidividual furnace and,@i7ith the T)articular smelti-i-ig ot)crotion uildertaken. .A-t hi.-h blast temperature, every blast furnace with ,,7h,@ch I a,-a familiar will "ha-ng," i. e., the uniforr@i descent of the charge column in the shaft L-,ecories irf@psirel, ,'.'id frequeiitly is completely halted. This phe-ioireiion is familiar to, a.11 operators. 1,s cause is readily understood. aiid it is a-@'i esseitio.1 feature of a.11 blast furnp . ce ooerations whenever rqetp.Iltirgical coke is ei@,,.r-loyed as fi:iel. At 1000 to 1200' -.-7. bl@ast terqperatures, the lum,!is (Yf c,)I@e lying in the neigbborhood of the blast entrance, i. e., 'm the combustion zor)-e, hove been found to exhibit ar. aier2,.ge temperature of 45 305n' P. to 33.00- P. as measured with an optical pyrometer (cf. Poyster a,-@,,id Josenli, Traps. -Ai-n. Tn.Qt. Min. Eng., voluime "Pyrometry" (1920), page 554), vrhen the niean "bearti@, temperatiiie" (viz., average of the mean sl,,ig anci roean metal tem50 peratures) @v,-,s 2750' F. Tn si.,,ch operations as v@.7ere exer@iiii.ed by Ro@7ster and Jo-seph, "banging," although not completely abseiit, wos not seriou,@,. It is obvioi2s frop-i reference to. Efiu@-tions I., 2 a-nd a., si@,i3l.al that silica i-n the coke ash, b,eing 55 intim,,"'ely dispersed tbrolighr,,ut the coke lump Pnr3 ill r.-a-etive conts-,ct v,,ith coke carbon, is reduced to metalli.c silico,@i at temperatures far belo7,,71 ,,he 3050' F. and 31.0f)' F. obtaiiiing a,t tlie nose of the tuy6,@-es. The metallic silicon thus produced 60 is above the melting point of silicon (2600' P.) but belov7 its boiling point (4148' F,). The liqillid silico,nproduced (alloyed with Fe also in the, coke ash) remains harrqlessly inside tl..te coke lump. 65 When "high" blast temperatures are employed, the tepner,%tlre of the coke lumps is raised. The vapor Pressure of silicon increases as its b,)iling point is approacheo. V7ith the blast temperatur,@ at 1600, F. t( 3.8001 F., the temperature of the 70 coke lurdps in the co-,nbustion zone regches the boili'n-g point of silicon. The silicon distils from the interior of the coke Itimps, cor-yimiigles with the gas stream, is swept up the furiiace shaft with th.e ga,s, i-s cooled during ascent, its temperature tr-ic fumace with low cost power@ e. g., 25 to 35 75 falls beloiv the temperature of silicon r(-,ductiop,
[4]7 and the reverse of Equation I takes place, vjz,: Si (vapor) +2CO (gas) - SiO2(fUme)+2C(soot) (9) that is to say, silicon vapor is reoxidized by CO, forming finely divided silica furne and carbon soot. A fraction of this fine!Y divided suspended matter becomes lodged on the materials, coke, stone, etc., descending the shaft, bliilding up accretions oii the surfaces of the char,,e particles, obstructing the interstitial channels between the solids through which the furnace gases must flonv, and creating a "tight" charge colum@n, requirirl-g increased pressure to force the gaqes throiigh the latter. TI-iis clogging of the cbonne!s in the charge raises the hearth presst7,re progressively. Ultimately the force of the ga,s strea-..,r. acting upv@rardly on the charge column exceeds the force of @ravity acting dov@nwardly ovi it, and descent c-@f thecha@r.-eisbalterl. WhenthechergecOlum',' is "held up from bel@ow" by the blast pressure@ the "furnace" is said to "hang." It is necessery theii momentarily to interrupt the :ffow of blast into the fuimace, thereby removing the up,,vard foreo of the gas stream and permitting the charge colurnt-l to descend freely under gravity. Ir. the operator's p,- 4rlance, this interruption of the flonv of air i-,ito the furnace is called a "pull" or a "check," i. e., the c@-Derator "checks the blast" or ,,-pulls the furnace." Vlhen the charge eglumn descends under gravity, the "furnace" is said to "slip." In the practice of my invention it is desirable to abandon the age-old blast furnace procedure of permitting pressure to build up until the furnace "hatigs" before checking the blast to cause a slip. As the hearth pressure rises due to building up of fume in the gas channels, (CO) 2 in Equation 2 iri creases, raising T. to a value higher than the temperature of the reduction zone. The reduction of silicon is impaired, the grade of metal falls, and the production of ferrosilicon of desired grade ceases. Consequently, as soon as hearth pressure begins to rise above normal, I find it desir@,Lble to "check" the blast and tliereby to dislodge the fume and reduce the back pressure from the charge colunin. This novel procedure requires more frequent "checks," and in the operator's present opinion accentuates the irregularity of stock movement. Operators have observed that with a "hanging" furnace, when frequent checking of the blast is necessary, the quality of the metal is impaired. Unfortunately, operators have ignored the thermodynamic significance of fhe rise in hearth pressure (which is not now recognized as a harmful factor in all blast furr.ace operation) and have attributed the - noor resul@s observed i)nder hanging-checking-slipping conditions to t-he "irregularity" of the stock descent. The "smooth working" of a furnace is ijsually judged by the frequent slips; good practi,@e calls for maintaining full blast until after the hearth pressure has built up to the point i@rhere the stock descent is completely halted before interrupting the flow of blast into the furnace. In usual practice when a furnace "hangs" a period of waiting (frequently twenty minutes) is permitted, during which period the operator hopes the furnace will "slip itself," i. e., that the charge column will slide down the furnace shaft without blast interruption, a phenomenon which may, or may not, take place. Vvhile checking is postponed, high hearth pressures prevail, and silicon reduction is diminished if not halted. A 2,420,564 hib@it a "normal" hearth pressure of 12 lbs./sq. in. gauge, when not hanging will build up to 24 lbs.Isq. in. gauge when hanging. If the furnace is "pulled" at the instant the hang occurs, the r, time-average hearth pressure is 1/2 (12+24) or 18 lbs./sq. in. gauge, with an average value of (CO)2 of 4.92 atmos.2. If the pressure build-up takes 20 minutes, and if the operator following standard practice waits 20 minutes before check10 ing the blast, the time-average pressure will be 1/2 (18+24) or 21 lbs./sq, in. corresponding to an average value of (CO)2 of !5.88 atmos.2. I-f the operator follows the teachings of the present invention and checks the blast 10 minutes after 15 the pressure build-ula starts, i. e., when it has attained a pressure increase of only one-half of its rise from 12 lbs./sq. in. (normal) to 24 lbs./sq. in. ("hanging"), the pressure will have attained a maximum of only 18 lbs./sq. in. when the blast 20 is checked and "normal" pressure restored. By following this procedure the timeaverage hearth pressure is 1/2 (12+18) or 15 lbs./sq. in. and the average value of (CO)2 will be 4.07 atmos.2. It is seen that there is a difference of 48 % (5.83 as 25 against 4.07) in the value of (CO)2 in Equation 2, caused by the difference between the two methods of controlling the blast. It is emphasized that the present process premises the operation of a blast furnace with a oo continuously rising hearth pressure, and my invention specifically includes the device of dislodging the accumulating fume in the charge column by intentionally enhanced irregularity in blowing rate, with the requirement of interrupt15 ing blast flow as frequently as necessary and desirable to maintain hearth pressure at a minirium. In particular, I recommend checking the blast before the hearth pressure reaches the point where the furnace hangs. 4 C, The loss of grade of the alloy produced thus by the failure of the operator to comply with the princ-iple of the present invention which teaches him to adopt a blowi-ng routine to maintain hearth pressure at a minimum time-average, is 45 not the most serious operating difflculty encountered. If the operator has followed the present instructions, he will be operating with a siliceous slag, the term "siliceous" being defined herein to refer to a slag containing a relatively high molar r)o concentration of "free" silica (i. e., uncombined with CaO, MgO, A1203, F203, MnO, etc.). V,7hen normal silicon reduction is suppressed as the hearth pressure is permitted to rise, the SiO2 in the charge, which is not reduced, remains in the 55 slag, increasing the slag volume and also increasing the silica content of the slag. The resultant increase in (SiO2), although favorable to silicon reduction, renders the slag so highly viscous that its free-running character is impaired, and the 6o loss of slag fluidity may (and frequently does) cause a choking of the smelting zone, combustion zone and hearth with a mass of sluggish, viscous slag, and causes a further increase in hearth pressure. 65 The heretofore-accepted fi-irnace practice in such circumstance requires the lowerip@,, of the blast temper,-@ture s(-@veral huy-idre(, de_grees, to remedy thp- h@iiigi)i!7 of the fum, ace. The Dresent invention requires Pn inc3:eps(- in bis,@t tempera70 ttire, to iiierease the available hearth heat, rertore silicon reduct4@o-vi, diminish the qil,@lnti@.y of s2a,-, lov,7er its SiO2 co-Tli.-ent, @-limi-nate its high viscosity, and bring the grade of alloy back to standard. In this respect, again, the process typical modern blast furnace which would ex- 75 of the present inventign is diametrically opposed
[5]9 io heretofore-accepted blast furnace procedure. In producing ferrosilicon according to my invention, it is an essential feature that the highest practical blast teniperature be employed, in order to maintain the coke coiisumption at a satisfac- 5 tory minimiim. 'Ln almost every actual case, the i necessary blast temperature will produce a combustion zone teriaperature bigh enough to cause reduction Of SiO2 in the coke ash, and contaminate the furnace ga@,es with silicon vapor and 10 reoxidized fume. In maliy cases, I contemplate operating with blast ternperatures as high as 20001 to 2400, P. in wlaich circumstances the combustion zone temperature is high enough to cause reducti.on of the A1203 ill the coke ash. I @-). When I heat the blast to between 2500' and 28001 P., not only is reduction Of A-1203 in- the coke ash complete but also the temperature is above the boiling point of aluminum (3723, P.) although below the boiling point of A1203 (5390' P.). At 20 blast temperatures from 3000' F. to 3300o F., the col@e ash is completely reduced by coke-carbon, including the S' '@02, A1203, CaO and M90. Although the iron in the coke ash is reduced at the lowest teniperature, it rerriaiiis undistilled in the coke lump longest, the boilir@,v point of iron (4955' P.) beiiig higher than the other reduced metals Si-, Al, Ca and Mg. At the highest blast temperatur e here conternplated, the coke ash is completeiy reduced and distilled. In operatiiag a given furnace to produce ferrosilicon, it is desirable not only that the timeaverage of the hearth pressure be maintained at a minirnuxn value, i. e., prevent a pressure rise above nor@-,2al, but also thet this "normal" hearth i!@ ,@)resstu,e be itself made a minimum. The m6st effective method of attaining low hearth pressure is to deerea,.sp- the ratio of the fi)rna-.e height to its average cross-sectionsl area. If "h" indicates the height of the flirnace shaft from the blast -ti@ entrance (center line of the tuy@res) to the average stockli-re (in feet), and if "V" represents the voltime of the charge colu@-a7i (in --ubic feet), then the average 9,rea "A" (in square feet) is given by V/h. The ratio of furnace height to 45 average area A, hereinafter called its "caliber," is defined, therefore, as h divided by V/h or h2,/V (ir-i, feet-'). For example, for the furnace in which pressures we,. e measured by Kinney (U. S. Bureau of Mines T. P. No. 442. "The Blast Fur- ro nace Stock Column," S. P. Kl:nrley (1928), page 10, F-igure 2), 1 the flirnace volume V (between tuyt,.re plane and mean stock line) is 22,680 cu. ft., the hei.@ht h is 69 feet, th@ mean area A is 326 sq. ft. (V/h), and the "caliber" is 0.218 feet-. 55 The hearth pressure was measured as 141.75 lbs./sq. in. gauge (Figure 48, page 124). The back pressure at the stock line was 1.96 lbs./sq. in. E,'auge, the pressure drop through the furnace being 12.79 lbs./sq. in. (14.75 minus 1.96). The (;o pressure gradient, per foot of furnace height was 0.185 lb./sq. in./foot (12.79@69 feet h). If the height of this furnace be reduced to two-twrds its actual value, e. g., to 46 feet (f@-om stoelc line to ttiy6re), the pressure gradient would remain 65 unchanged (with the same blowilig rate) and the pressure drop through the furnace would be oil-ly 0.185X46 or 8.5 lbs./sq. in. With the same top pressure of 1.96 lbs./sq. in., the hearth pressure would be 10.46 lbs./sq. in. The reduction of 70 .Lurnace height h to two-thirds value (caliber reduced from 0.216 to 0.144) decreases the value of (CO) 2 in Equation 2 from 4.02 atmos.2 to 2.94 atmos.2. A 30% increase in average cross-seetional area A (froin 326 @q. ft. to 422 sq. ft. by 7(- increasing all diameters 14"/o) would decrease the pressure gradient from 0.165 lb./sq. in./ft. to 0.1092 lb@/sq. in./ft., due to the fact that the pressute gradient will vai@y as the square of the gas velocities. With a 46 foot height and a 30% .nerease in average area, the pressure drop through the furnace would be only 5.02 lbs./sq ii,,., cori-esponding to a I-earth pressure of 6.90 lbs./sq. in. (5.02 plus back pressure at furnace toP Of 1.96 lb,@./scl. i-n.). The value of (CO)2 iS deci,c-@,qsed to 2.17 with the 30% increase in A, th.@- cali-ber of tj".le furna6e is reduced to O@,111 foot-'. In designing a blast furnace for carryin.g out the process of the present invention, I prefer to decrease its active h-light at least to one-half (34'611) .f not tb one-third (22'0"), if its mean section area is to be left -Linaltered (at 326 sq. ft.), that is, I design the furnace with the unusually small "calibers" of 0.108 ond 0.072 respectively; I also enlarge the dimensions of the gas main, dust catchers, gas scrubbers, stove and boiler burners, i-n order t,o decrease the back pressure at the flixnace top to 6-8 i-,iches of water pressure (0.22 to 0.29 lb./sq. in.), thereby decreasing the h-earth pressure to 6.57 lbs./sq. in. (for the ha-If heigl-lt) and to 4,32 1-bs./sq' in. (for one-third hei-ght), correspondiug to vo,!Lies of (CO)2 m lo%v as 2.09 atmos.2 and 1.67 atiros.2, respectively. WI-iile tle present iaventiori is not limited in any respect to the details of furnace design and construction as such, it is true the geometric sliape of the fumae e and the relation of its height to its cross--section are factors in my present process to the extent that these dimensions and ratios QQntrol and affect the hearth pressure. Since hearth pressure is a very significant factor in the control of the temperature of silicon reduction Tel the design of the furnace, in an indirect way at least, has an iinportant influence ola the process of producing ferrosilicon. The term "critical tornperat-ure" as applied to the. blast furliace hearth was introduced into the discussion Gf the fuel econorny of the blast fui-nace by the late J. E. Johnsoii, Jr@ (cf. "Principles, Operation and Products of the Blast Flurnace"), and the general Principles of his "available" Iiearth heat theory were explained by him in considerable detail. The exact definition of the "critical temperature", as Johnson used it, was not very precise; he appears to have identified it with the "free flowing temperature" of tlAe slag without defining that term. On the production of ferrosilicor4 in the blast furnace, I have found that the temperature of sili-Con reduction To (de1-ined in Equation 3) is identical with Johnson's "critical temperature." I have fgund further that his method of compliting the "heat available above the critical temperature" by means af a hearth heat b.alance is substantially corre@-t as Johnso-n described it. In discussizig the produc@ tion of phosphorus in the blast furnace, Royster and Turrentine (Indusirial and Engineering Chemistry, vol. 24, Page 223, Feb.. 1932), illustrate the niethod of applying Johnson's available hearth heat principle to the reduction of phosphorus. In atteriapting to cany out the presentl3riliveiited process, the blast furnace operator will have little, chance of technical success unless he is prepared to calculate the available hearth heat for each furnace operatiqn he undertakes. In order to indicate the proper method of deterrriining available heat the following illustration of the production of ferrosilicon in the blast furnace is given:
[6]2,420,564- ILLUSTRATION 1 In produciilg 50% ferrosilicon by my invention I charge into a blast furnace, in the course of a 24-hour day, the following materials: Pounds Col,.e --------------------------------- 629,000 Ore ----------------------------------- 144,700 Sandstorie ---------------------------- 394,000 Limestone ---------------------------- 128,600 Cast iron scrap ------------------------ 20,600 Steel scrap --------------------------- 29,400 The furnace produces: Gross tons (2,240 lbs.) per dL%y .Verrosilicon --------------------------- 100 Slag ---------------------------------- 130 Dust ---------------------------------- 2.13 Fume --------------------------------- 14.35 The materials charged have the following analysis: Ore Sandstone Limestoiie Per ceat Per cent Per cent SiO2 -------------------------- 36.61 94.72 1. is A1203 ------------------------- 3.56 0.87 0.35 cao -------------------------- 0.36 0.22 50.40 mgo ------------------------- 0.08 0.14 3.30 Fe2O3 ------------------------- 53.76 1.97 0.54 MiaO2 ------------------------ 0.21 0.08 0.04 P205 -------------------------- 0.15 0.05 0.03 S03 --------------------------- 0.02 0.01 0.25 002 -------------------------- 0.19 0.32 42.72 NEL20 ------------------------- 0.05 0.17 0.06 X20 -------------------------- 0.02 0.11 0.04 H20 (<105' C.) -------------- 2.35 1.13 1.07 H20 (>105' C).) -------------- 2.08 0.21 0.02 100.00 100. 00 100.00 Cast Iron Steel Scrap Scrap Fe ------------------------- ----------- 94.48 98. 95 Mn ------------------------- ---------- 0.25 0.46 Si ------------------------------------- 1.32 0.12 p ------------------------------------- 6.45 0.06 s ----------------------------- 0.22 0.07 ---------------------------- 3.29 0.31 100 @ 00 100.00 Coke rroximate Analysis TJltimate Analysis Per cent Per cent Moisture ------- 1.75 H20 ------------- 1.75 sio2--- 5.34 Volatile --------- 1.15 Volatile --------- 1.15 A1203- 2,47 Fixed Carbon --- 87.29 Fixed Carbon --- 86.22 CaO-- 0.68 Asb ------------- 9.81 Fixed Ilydrogen0.23 MgO-- 0.23 - Nitrogen -------- 0.18 P205-. 0.04 100.00 S ulphur --------- 0.87 Na2O- 0.13 SuIphur -------- 0.87 Iron ------------- 0.62 l@:20 0.09 The ferrosilicon and slag produced have the following analysis as reported in weight percentage and as converted into rnolar percentage: Fe?lrosilicon By Weight Molecular Coraposition Per cent Per cent Si ------------------------ 50.37 Silicon (Si) 82.62 --------------- Mn ---------------------- 0.25 Iron Silicide (Fe3Si) ------ 15.76 --- 0.378 P ------------------------ 0.21 Iron Phosphide (Fe3P) s------------------------ 0.013 Iron Sulphide (FeS) ------ 0.23 ------------------------ 0.003 Manganese Carbide Fe ----------------------- 49.15 (Mn3O) ---------------- 0.014 - 1\f anganese (Mn) -------- 0.21 100.00 100.00 12 Slag By Weight Molecular Composition 5 Per cent Per cent SiO2 --------------------- 65. 63 Silica (SiO2) 56. 31 --------------- A1203 -------------------- 7.86 Wollastonite (CaSiO3) ---- 26. 25 CaO --------------------- 23. 29 Er)statite (AIgSiO3) ------ - 4. 75 MgO -------------------- 2.04 Anortbite (CaAl2Si2O@) @--- 7.20 S------------------------ 1.79 Calcium Sulphide (CaS)-- 5.13 FeO ----------------- --- 0.07 Grilnerite (FeSiO3) -------- 0.09 M90 ------------------ 0.03 RhodoDite (MnSiO3) ------ 0. Or) 10 Na2O ------------ ------- 0-11 Sodium Silicate (Na2SiO3)- 0.17 K20 ------------------ 0.08 Potassiujn Silicate P205 --------------------- 0.004 (K2SiO3) ------------------- 0.08 100.90 100.00 Less oxygen equivalent of Bulphur -------- ------- -0. 90 15 100.00 The furnace used has the total height from stock line to tuy6res of 19 feet, its interior diameters being 12 feet at the stock line, 18 feet 6 20 inches at the bosh, and 14 feet at the bearth. The bosh height is 6 feet 6 inches, the bosh angle 71 degrees. TI-ie in-wall batter in the mantle is 3.3 inches per foot of height. The "active" volume of the furnace (between blast entrance and 25 stock line) is '.3770 cu. ft. (V), its average crosssectional area A is 198 sq. ft. (Vlh), and its -.aliber (hlA) is .096 feet-'. The furnace is blown w,th a normal blast flow of 23,700 cu. ft./min. of @tmospheric air meas30 ured at 60' F. and 14.7 lbs./sq. in. pressure, at an average blast-main pressure of 2.5 lbs./sq. i-,i. gauge, at the blast temperature of 2750' F. Tlae blast enters the furnace through six water-cooled tuy@res, 12 inches in diameter. The blast pres35 sure drops 0.63 lb./sq. in. in going from the blast main to the interior of the hearth, the pressure inside the hearth being 1.87 lbs./sq. i,,i. gauge (16.57 lbs./sq. in absolute, 1.127 atmospheres). Because of the frequent interruptions of th-. 40 blast due to "checking," there is a loss of 125 minutes per day in blowing time, the average flow of blast being only 21,600 cu. ft./min. (at 60' F. and 14.7 lbs./sq. in.). The temperature of the slag as it emerges from 45 the hearth averages 2950' P., and the rletal 2840' F. to 2875' P., when the optical pyrometer readings are corrected for emissivity. Due to flame, smoke, fume, and rapid sculling on the surface of the metal and slag streams, these temperatiO tures are difficult to determine ac.-urately. The furnace (under normal wind) produces 32,220 cu. ft. per min. of top gas (measured at 60' P.) which averages 29,720 cu. ft./min. when allowance is made for loss of blowing time due 55 to checks. This gas is extremely "dirty" in the sense that it carries a large volume of extremely fine silica fume. Although the we4@ght of fume is not great (4.8 grains/cu. ft.) it is difficult to remove the fume from the gas because of its fine60 ness. The loading of ordinary blast furnace "flue dust," caught in the dust catcher and prir,,lary scrubber, is small (0.72 grain/cu. ft.). The analyses of the gas, dust and fume are as follows: 65 Dry Top Gas (by Dust (by weight) Fuine (by w(@ight) volume) P'r ceiit CO ------------- 37.93 A1203 ---------- 5.15 A1203 ----------- 4.68 70 C02@ ------------- 2.13 SiO2 ----------- 59.50 SiO2 ------------ 66.50 Carbon -------- 20.16 R2 --------------- 1.96 Cao. ---- ----- 7.90 N2 -------------- 58-03 MgO ---------- @0.75 Na2CO3 --------- 5.57 FeO ----------- 16- 50 K2CO3 ---------- 3.14 Carbon ------- 10.20 I 75 ne gas discharges from the furnace top at a
[7]13 very high temperature, averaging 1675' F., making it necessary to water cool the top of the furnace where exposed to this hot gas, including the big bell and hopper. The stock lirie becomes white hot before each charge is dumped off the big bell@ A very hot, dirty top-gas is an essential feature of the present process of producing f errosilicion. It requires some care but no great ingenuity to design the gas main ard gas cleaning system for this service. An electrostatic precipitator is recommended for the final clean-up of gas heavily laden with silica fume' The very high top temperature of the furnace is due to the large amount of heat generated in the furnace in excess of the total thermal requi-rements of the process. This excess of thermal input over thermal requirements is due to the small fraction of the total heat which is "available" above the- "critical temperature" (Johnson), meaning tiae temperature 6f silicon reduction (Te of Equation 2), The value of T,: here is seen to be: heartli pressure, 1.87 lbs./sq. in. gauge (CO) @1.27 (CO)2=1.27(atMOS.2) Metal analysis-molar concentration of silicon (Si) =0.8362 Slag analysis-molar concentration of silica (SiO2) =0.5631 By Equation 2, k=1.88, logiok=0.274, and by Equation 4, T,,=1843' F. (1570, C@, or 2858, F.). By refere--qce to the carbon charged, the wind blown, aiid the top gas analysis, it is seen that 288 lbs. of coke-carbon per minute, or 4124 lbs carbon per ton of alloy, is oxidized by the 02 of the blast. The total tb-ermal input into the reduction zone or smelting zone of the furnace is the sum of (a) the heat of the blast eiitering eu blast temperatiire (2750' P.), (b) the heat Of the carbon entering the smelting zo-@ie, at Te (2858' F.), and (e) the heat generated by the reaction between coke-carbon and blast oxygen to form CO (3960 B. t. u./lb. carbon). The t6tal 14 Tdble 2.-Hearth heat balance, B: B. t. u.Itb. of I carbon reqttired to maintain hearth equitibrium at Tc=2858' F. B. t. U. r, Item 1, reduction of 0.269 lb. silicon -------- 2690 It em 2, reduction of 0.014 lb. iron ---------- 79 It em 3, reduction of 0.0085 lb. phosphorus-- 84 It eui 4, reduction of 0.002 lb. manganese---- 19 It em 5, reaction of 0.8 cu. ft. blast moisture 10 with carbon --------------------------- 121 Item 6, preheating 0.542 lb. metal in transit through zone -------- ------------------ 56 Item 7, pri@heating 0.730 lb. slag in transit through zone --------------------------- 64 15 Item 8, heat loss through walls and coolers-- 494 Item 9, reduction of 0.025 Si from coke ash- 260 Item 10, heat (unavailable) of gases leaving zone at T ------------------------------ 5905 20 Item 11, total thermal requirements ------- 9772 Item 10 shows that 60% of the total thermal input into the smelting zone must be lost to that zone. Comparison of items 1 and 10 of Table 2 25 shows that for every B. t. u. used in silicon reduction more than two B. t. u.'s are "unavailable." The total heat of reduction fo the alloy r (sum of items 1 to 4) is only 2872 B. t. u. and is less than 30% of the total thermal input to the Even this loxv "thermal efficiency" is at30 =.only with the very high hot blast i-ised. Referring to Table 1, item 2, it is seen that the hot blast brings into the furnace 7% more heat than is generated by the combustion of carbon. At lower blast temp6ratures the fraction of the total heat available for silicon reduction decreases rapidly@ which explains the rapid iiidrease in coke consuraption observed when @operating at lo@,v blast temperatures. It also explains why all pre40 vious attempts to produce silicon alloys of silicon content higher than ?,O % in the blast furnace have failed - My present invention consistg, in its simplest terms, in providing to the smelting zone a thermal heat generated in the smelting zone per lb. of - input with an available fraction E@dequate to meet co,rbon burned by 02 in the blast is 4o the thermal requirements of the smelt'mg reactions. The failure of previous atterapts to proTable l.-Hearth heat balance, A: Thermal in duce high-analysis ferrosilicon in the blast furp7tt B. t. u. per lb. carbon burned by thoo blast nace has been caused by a failure to meet this 02 at tlie tity6res, measured above 60, F. bdse ther modynamic requirement. Operators have atline 5( tempted to produce ferrosilicon in the blast furB. t. u. nace, and with excessive use of coke have introIte-m 1, combustion of I lb. C -------------- 3960 duce d more than enough total heat to carry out Item 2, heat of 75.67 cu. ft. of air (3.5 grams the hearth reactions many times over; ho,@vever, per cu. ft. rnoisture) ---------------- 4240 beca use of their failure to recognize the magniItem 3, heat of 1.275 lbs. C entering at T@ --- 1572 r)5 tude of the unavailable fraction of total heat (item 10, Table 2) their efforts failed, and most Itel.q 4, total thermal input --------------- 9772 oper ators have become convinced ti@at the blast furnace cannot produce ferrosilicon carrying The thermal input to the smeltiiig zone can be more than 20 % silicoli. compared with the thermal requirements of this 60 In applying the present ilivention to furnace zor-e, which lal@-ter include the endothermic chemoper ations other than that described iri the above ical reactions taking place therein, the heat lost exa mple (Illustration 1) it becomes necessary to tlirough the brick wa,lls and to the water cooling repe at the calculations for the changed condidevices (tuy@res, etc.) and (as the largest item tions . For example, the coke consumptioll for of all) the sensible heat of the gases leaving the 6r) 50% ferrosilicon with a blast temperature of smelting zone at Te. The slag and metal are free 2750' F. is 6290 lbs. coke per gross ton of ferroflowing at 26701 F. (controlled by the slag vissilico n as given above. If the blast temperature cosity) .ind pick up a small amount of heat in is altered, it is necessary to change the amount passing through the smelting zone (items 9 and of coke in each charge to provide "available" heat 10, Table 2). The greater part of the heat in the 70 in the smelting zone in an amount which is adehot rnetal a-,id slag (including heat of fusion) quat e to meet the thermal requirernents of the was absorbed in the furnace shaft before the slag hear th reactions 'With the altered blast temperaand metal entered the smeltilig zone. Table 2 folture. Elor each di erent hot blast temperature it lowing gives the thermal reqiiirements of the is necessary, of course, to recompute the hearth @melting zone: 75 heat requirements. To indicate the nature of the
[8]results when this has beeii dorie, tile following table is given: Table 3.--Coke combustion, and daily tonnage of 50% ferrosilicon in Illustration 1 for varying hot blast temperatures Coke Con-F sumption, Tonnage, tods Blast Temperattire, lbs./gross ton alloy/day alloy 920 ------------------------------------ Infinite none 1,000 ---------------------------------- 68,000 9 ilioo ---------------------------------- 33,000 18 1,200-@ -------------------------------- 25,000 24 1,500 ------ ---------------------------- 14,300 43 2,000 ------ -------- --------- --------- 8,850 71 2,600 ------ -------- ------------------- 6,760 9 3000 ------ ----------------------- ---- 5,500 111 3:500 ---- ----------------------------- 4,730 128 4,000 ------ -------- ------------------- 4,150 147 For a hearth pressure of 1.87 lbs./sq. in. gauge, and for slag of the composition given in Illustration 1, Table 3, of course, is a complete, sufficient description of my present invention, so far merely as the production of ferrosilicon is concerned' For the process to be useful, in the sense that it is economically practical, it is necessary to set an upper limit on the coke consumption. The present invention purports to be an improvement in the method of Producing ferrosilicon. But a process is only an improvement when the value of the product made exceeds the cost of the materials used in it@ production. It is a fact, of some academic interest perhaps, that 50% ferrosilicon can be made with low hearth pressure and with a Mgh "free" silica in the slag with any blast temperature above 9200 F. At 1000' F. the production of the alloy (9 tons per day) with a coke consumption of 68,000 lbs. is not a "process" in the sense intended in this description: at $4.50 per ton of coke, the cost of the coke alone is more than $150 per ton of alloy produced. Since the ferrosilicon.produced is worth less than $75, such an operation is not a "process" in any practical serise. In the appended claims where the phrase used is the "process of economically producing ferrosilicon," the term "economically" is herein defined as a specific disclaimerwhich is intended to confine the scope of the present invention to those furnace operations only iii which the coke consumption is sufflciently low so that the cost of the raw materials used, plus the cost of the furnace operation, does not exceed the commercial value of the. alloy produced. In the above table, for example, it is possible to state that, at present prevailing prices, both of the materials charged and of alloy produced, a coke consamptioii somewhat below 12,000 lbs./ton of alloy is necessary to permit the operation properly to be defined as the "economical,, production of ferrosilicon. A coke consumption of 12,000 lbs. or less corresponds to a blast temperature of 16401 F. or Mgher. Since the cost of the materials and the value of the product may change in a manner impossible to anticipate, it is not feasible to set a Precise and fixed limit either tb the maxi. mum coke consumption or the minimum blast temperature. It should be recalled that the relationship between coke consumption and blast temperature, as shown in Table 3, is applicable only to the hearth pressure of 1.87 lbs. gauge. Without great difficulty, it is possible to repeat the calculations for a hearth blast balance (as in Tables 1 and 2) for several hearth liressures. For example: 9 420,564 Table !.--Coke consu?nption ds control d 6z( varying hearth pressure and blast temperature@ in Illustration I 5 Blast Ilearth Pressuro, lbs./sq. in. (abov.e Find below) temperature, F. -5 -2 +0 +2 +5 +8 +10 - io 1,000 --------- 33,000 42,700 47,000 75,000 ------- -------- ------- 1,100 --------- 24,400 26, OIDO 29,800 36,200 22,000 -------- ------- 1,200 --------- 18,700 20,300 22,000 26,000 55,000 128,000 ------- 1,300 --------- 15,700 17,300 17,900 19,800 32,000 47,200 75,000 1,400 --------- 13,400 14,400 15,300 16,500 22,600 30,800 39,000 1,500 --------- 11,800 12,400 13,200 14,200 17,800 21,700 26,300 1600 ---- ---- 10,600 11,200 11,700 12,420 14,600 18,000 20,100 1:700 --------- 9,700 10,200 10,600 11,220 12,800 16,000 18,600 15 1,800 --------- 9,000 8,900 9,900 10,200 11,500 13,100 14,200 1,900 --------- 8,400 8,350 9,200 9,600 10,500 11,400 12,400 2,000 --------- 7,920 8,100 8,600 8,860 9,850 10,400 11,400 2,200 --------- 6,950 7,340 7,660 7,850 8,400 9,000 9,420 2,400 --------- 6,450 6,600 6,8901 711001 7,400 7,850 8,700 2,600 --------- 1 5,960 1 6,110 6,350 6,480 6,700 1 7,000 7,lGO 20 The above calculations are sufficient to illustrate the inter-relationship of the several factors involved in the process described here. In Table 4, for "Mgh" hearth pressures, 5, 8 aiid 10 lbs./sq. in., and at "low" blast temperature, 1000, 1100 25 and 1200' F., the blank spaces (in the coke consumption column) indicate that 50% ferrosilicon -.alinot be produced at all, i. e., an infinite coke consumption is inadequate. The hearth pressures "usual" in Ainerican blast furnace practice,, 30 i. e., 12 to 16 lbs./sq. in., are ornitted from the, table since the excessively high coke consump-. tioris make such operations lack practical inter-est, except with high blast temperatures. It is well to point out that the two important@ 35 factors which effect the fuel economy of the pro@@-. ess, viz., (1) hearth pressure and (2) blast temperature, are not independent of each other. At@ low blast temperatures, e. g., 13001 F., the coke! consumption' is increased five fold from 15,000 40 lbs./ton to 75,000 lbs./ton by raising hearth pressure from 5 lbs./sq. in. below barometric (vacuum operation) to 10 lbs./sq. in. above barometric, 1. e., more than $100 per ton of alloy change in fuel cost. At high blast temperature, however, 45 say 2600' F., the same change in hearth pressure increases the coke consump'Llion by only 20% from 5960 lbs./ton to 7160 lbs./ton, a change in fuel cost of about $3 per ton of alloy. The figures given in Table 4 are sufficient to guide the opera50 tor in carrying out my invention for the ca@se of 50% ferrosilicon and for the molar concentration shown in Illustration 1. These figures will not apply for values of (Si) and (SiO2) in Equation 2 different from the 0.836 and 0.563, respec65 tively. The permutatioi-is and combinations of coke consumption, blast temperature, hearth pressure, slag composition and metal composition which may be encountered in practice make it impossible to tabulate them all in detail. The 00 operator will need to carry out the above cal.-ulations as the furnace operation proceeds, and he may have to. repeat these calculations frequentls,. As an example, if a furnace making 50% alloy has been charged with a ratio of coke 65 to burden to operate with with 9600 lbs. coke per ton of alloy, while the furnace exhibits a hearth pressure of 2 lbs. per sq. in. gauge, the blast temperature required (Table 4) is 1900, P. Should the charge column "tighten uP" and the heart.4 70 pressure increase to 5 lbs./sq. -'An., the blast temperature must be increased to 2090, F. in order to maintain the grade of alloy. This rise in blast temperature may cause a further increase in the hearth pressure, e. g., to 10 lbs./sq. in., which 75 wiII necessitate a further increase in blast tern-
[9]2142o,b64 17 perature to 2180' P. Since such a rise in hearth pressure may occu r within an hour's time, it is not possible suddenly to alter the amount of coke in the furnace charge, since the time of passage of a charge through the furnace will be 5 several hours. It is nbt even possible to ascertain that the silicon cantent of the alloy @as decreased since the pressure rise may occur before metal can be tapped from the furnace, and before a chemical analysis for silicqn can be re- 10 ported from the laboratory. For technical success, control of the furnace must be attained by adjustment of the availability of the thermal i-nput to correspond to changes in the value of Te occurring during the operation. 15 18 and increasing the reduetion temperature Te frc>m 28580 F. to 2864- F., a 6 rise, which is completely negligible. ILLUSTRATIOiq 3 @ Because of the high iron content of the coke, sandstone, and limestone, the maximum grade of alloy can be produced by using an ash-free petroleum coke for fuel, a high grade sandstone (e. g.., a "firestone" from northern Ohio) as a source of silica and marble chips for flux with materials of the following analysis: Sandst.one Marble Chips Fortunately, there is but a slight change in TSiO2 --------------------------- ---------------- 98-58 0.35 A120 ------------------------------------------- 1.21 6.13 caused by changes in tfie grade of the alloy, for CaO --------- ----------------- 7 ----------------- 0.03 55.00 high silicon contents. The following table will mgo --------- ----- ---------------------------- 0.02 0@ 56 illustrate this fact. Fe2o3 --------- ---------------------------------- 0.11 0.09 SO S --------------------------------------------- 0.01 0.03 P 05 ------------------------------------------- 0.007 0.013 20 C)2 Table 5.@--Molar concentration of silzcon in fer02 -- ------------------------------------------- 0.03 43.86 rosizicon of varytl?zg silico?z content The fu@rnace is charged with: IVeight percentage of silicon I4bg./day in the alloy ---------------- 50 80 90 100 25 Petroleum coke (98.53' P.,C.) ----------- 552 000 Molo,r percentage of silicon Sands tone --------------------------- 328,000 (Si) in the alloy ------------ 83. 30 95.84 98.09 100. 00 Marble chips -------------------------- 36,(00 Molar percentage of iron silicide (Fe3Si) in the alloy ---- 16.70 4.16 1. 91 00. 00 T4e same furnace and th,e same operating Logio (Si) ------------------- -0,080 -0,018 -0. 007 0.000 1. The Temperattire of silicon redticc,,nditi ons are used as in Illustration tion (T,) ------------- 'F- 2,858 2,868.7 2, 870- 7 21 S71. 9 30 furnace produces 51.5 gross tons/day of "alloy" I -- ---- - 5 gross tons/day of slag. The analysis of The increase in Te in raising the grade of the slag is: alloy from 50 to 100% is only 13.9' P., (7.7' C.) , a temperature change much too small to be as35 By Weight Mol ecular Composition certained in practice. The change in the available heat (in Tables I and 2) due to such a 14' P. Mol ar per cent Sio2 ------ --- ----------- 76.30 Silic a ----------------------- 71.36 change in Te is less than 0.8%, a quantity smaller Al 03 ----------- 3.96 Wol lastoiaite --------------- 25.10 or in determining the thermal quan19.64 Ano rthite--- than the err Ca ------ ----------------- -------------- 3.14 MgO ---------------- ------- 0.@20- Enstatite ---- : ------ -------- 0.4 tities involved in the heat balance. In attempt- 40 ing to produce "pure" silicon (100% ferrosilicon) no difficulty is encountered _due to any inability of the furnace to reduce silicon. The practical difficulty arises in -attempting to find charge materials low enough in iron content. The folIoNving will illustrate the problem Gf increasing the,grade of the alloy. ILLUSTRATioN 2 The operatio@n described in Illustration 1 is continued, unchanged in every respect except that the cast iron scrap and the steel scrap are omitted from the charge. The tonnage of alloy produced is thereupon reduced from 100 gross tons/day to 73.20 tons/day and the coke consumption per ton of alloy is increased from 6290 lbs./ton to 8000 lbs./ton. The analysis of the alloy produced is increased in grade from 50.37% silicon to 64.209'o silicon. The metal composition is as f ollows: Ferrosilicon Composition By and the a:lloy alialysis is: Si --@ -- ---------------- -------------- 99.744 Fe --------- --------------------------- 0.237 weight Molecular Composition less fuel, since n occurs, with @a corresponding saving both in carPer cent bon and in available heat. Because of the abSi -------------------------- 64.20 Si ---------------------- ---- 90.53 65 sence of silicon fume from the coke ash, sticking Fe - ----------------------- 35.38 Fe3Si --------- --------------- 9.'02 up of the charge column does not take place, Mn --------------------- --- 0.21 FeP ------------------ -------- 0.28 P --------------------------- 0.20 Mn ------------------------- 0.16 hanging does not occur, and the hearth pressure S ---------------------------- 0. fil FeS -------- ---------------- 0.014 can be maintained at an average of O@54 lb./sq. in C--------------------------- 0.00 100.00 gauge The equilibrium constant k for Illustra45 p------------------------------------- 0.017 s-------------------------------------- 0.002 The coke consumption for this exceedingly "pure" silicon is 10,700 lbs./ton, which may seem to be au increase over the coke consumption of 5o -6290 lbs./ton in Illustration I and 8000 lbs. in Illustration 2. -When @coke consumption is re,ported in terms , of "per tons of silicon produced," instead of "per ton of alloy," the coke consumptions are: Illustration 1, 6290 lbs. coke for 50.37 % 55 alloy or 12,420 lbs. coke per ton of silicon reduced; IllustTation 2, 8000 lbs. coke/ton for 6420% all6y or 12,400 lbs. per ton of silicon; and Illustration 3, 1.0,700 lbs./ton of petroleum coke (fixed carbon 98.53), equivalent to 12,200 6o lbs. of coke at 86@22% fixed carbon. There is a slightly lower coke consumption when using a@cho reduction of coke ash silica 100.00 To With the same heart-h pressure and slag analysis, (SiO2) and (CO) are unchanged and the equilibrium .constant is increased from 1.88, in Dluatration 1, to 2.04, increasina, logiok by 0.035 75 tion 3 is l@51 which is lower than the 1.88 value i-n Illustration 1. In tlus particular case, substantially @100% silicon metal is produced with a lower reduction temperatu're, and wi'Lh lower coke consumptibn, than the 50% ferrosilicon required. The gas is cleaner, - there is less silica fume pres-
[10]19 ent, the furnace pressures are lower, and the movement of the charge columri is improved. The major difl-iculty encountered in producing such pure silicon is caused by the lo-w density of the metal. Silicon is alm(>st as light as the slag and the separation of slag from metal both in the hearth and after tapping from the furnace is quite difficult. Success in the blast furnace smelting of ferrosilicon can be attained by supplying the smelting zone wtih sufficient heat "available for silicon reduction" to carry out the heatabsorb,.ng hearth reactions. In the electric furnace all of the heat derived from electric energy is "available" for all and any purpose. By comparison, the 30% "availability" of the thermal input in IlIustration I appears discouragingly low. An-d this is particularly true when it is remembered that the astounding"y high blast temperature of 2750' F. is required to realize an availability of this lOw figure. With the low blast temperatures used in present-day furnace practice, the fraction of the total thermal input which is available for hearth reactions is almost ridiculously low, and it must be emphasized that the 30% availability in IlluStration I is attained only by altering the furnace dimensions and the blowing procedure in order to operate with the low hearth pressure shown there, less than one-fifth of the gauge pressure which is customary in the typical blast furnace. With usual low blast temperatures and high hearth pressure, it is easy to understand why previous efforts to produce high analysis alloy have failed. It is to be observed that in carrying out my invention it is necessary to abandon many of the ba,sic principles which years of blast furnace experience have taught the furnace operator. My invention requ@ires raising the blast temperature to such a high value that the furnace is catised to hang. My process makes an exaggerated irregularity in blowing rate desirable, and causes a maximum irregularity in descent of the charge column. This method of furnace operation violates the operator's most respected tenet, that a .'smooth working" furnace is essential to good p@actice. My invention teaches that a great improvement in economy is attained by operating with a "short" furnace, e. g., 25, 30 or 40 feet high: also, that the hearth pressure should be maintained at a low value. These facts help to explain why an obvious extension of currently recognized blast furnace methods has not been successful in increasing silicon contents into the ferrosilicon range, whether the desired silicon was 25, 65 or 95 %. Attempts have been made to duplicate the electric furnace technic, by charging a blast furnace with coke, scrap a?id substantially just enough silica to produce the silicon in the alloy, so as to produce no greater slag volume than is formed by the minor impurities, A1203, CaO, MgO, etc., inevitably present in the chargematerials. Such attempts were unsuccessful. Since no explicit statement of the function of the slag in the blast furnace production of ferrosilicon appears in technicalliterature, it is important to explain here the principles involved, since operating difficulties will be encountered in carrying out my invention whenever the slag volume is insufficient or absent. in controlling the composition and amount of slag in my process, the following principles obtain: I. The slag should be as silicious as possible (i. e., exhibit a maximum molar concentration 2,420,564 20 of "free" silica), Two factors limit this niaximum: (a) the slag viscosity must not be so high (the viscosity depending on the slag temperature, cf. Feild and Royster, loc. cit. supra) that its free-flowing properties are impaired; (b) it must contain suffic-lent lime to combine with the sulphur in the coke, to produce CaS, which must be kept diluted to a sufficiently low molar concentration to maintain a satisfactorily low sulphur 10 content in the metal. It is true that at the high hearth temperatures prevailing in a ferrosilicon furnace the desulphurization reaction proceeds almost to completion, and that this is also favored by low hearth 15 pressure, but there is, in every case, a maximum concentration of calcium sulphide in the slag beyond which high sulphur metal wiII reswt. With 1600' F. hot blast, 8 lbs. hearth pressure, making 50% alloy (Table 4, Illustration 1) the 20 coke consumption is 18,000 lbs./ton of alloy. With coke containing 1.5% sulphur, the charge contains 270 lbs. sulphur/ton alloy. If this sulpliur were not absorbed in the slag, the sulphur content of the metal would be 12%. If a slag 25 volume of 1000 lbs./slag/ton metal (a typical slag volume in standard American blast furnace practice, cf. Royster and Joseph loc. cit. supra) were maintained, the sulphur content of the slag would be 27%, or 47.5% (by weight) calcium sulphide, 30 a prohibitively high figure. It is seen that the amount of slag customary in blast furnace practice is wholly insufficient in ferrosilicon practice, a fact which has been responsible for some of the previous failures. It is not safe to assume 35 (without the present statement) that the operator "would nattirally increase slag volume" in attempting to produce ferrosilicon, since the precepts of electric furnace practice, with no slag volume at all, might induce him to do exactly the 40 opposite. II. As shonvn by Equation 4, the reduction of silicon takes place at all temperatures above T,: (in Illustration 1, 2858' P.). Whenever the tem45 perature is below T. this reaction is reversed, i. e., Si is oxidized by CO. If the metal bath in the furnace hearth is not covered with a protective layer oL slag but is left exposed to CO, it cools below T. and its surface becomes covered with unfused SiO2 produced according to the equation: 50 Si(metal) +2CO(gas) @ SiO2(solid)+2C(solid) (11) The fi@st ferrosilicon flowing down on the hearth bottom after each cast spreads out into a thin 55 liquid layer. If exposed to furnace gases, its surface becomes coated with a thin layer of unfused silica (melting point Of SiO2 3092' P.). The successive additions of ferrosilicon trickling into the hearth spread out on top of the solid films of 60 SiO2 aiid the lower pal't of the hearth builds up with these lamellar layers of silicon and silica. When the furnace is "tapped" nothing flows out and the hearth is said to "freeze up" although its temperature is well above the melting point of 65 silicon (26001 F.). A bath of molten silicon can be maintained liquid, without a protective layer of slag interposed between its surface and the furnace gases when this "freezing up" of the hearth bottom has lifted the surface of the metal 70 bath up clo@e enough to the combustion zone to keep its temperature above To. III. When the difficulty from reoxidation disappears, a new difficulty is encountered due to vaporization of silicon from the surface of the 75 metal. When the metal temperature is below T,
[11]21, (and re-oxidation@ occurs), the vapor pressure of silicon is low-0;069- mm. Hg vapor 'aressure at T.-2858' V. (in Illustration 1). If the total volume@ of fur-nace gases were saturated with silicon va-por at this partial pressure, the loss of silicon from the alloy woulcl be less than Oi25%. Superheating the surface of the metal bath by exposure to gas from the combustion zone with its temperature at 3500P P. to 4500' F. (depending upon the blast temperature), loss of silicon by vaporization may amount to 20 to 40% of the silicqn@ reduced. The resultant loss of available heat is. not as serious as the clogging of the gas channels in the charge colunin, when the silicon vapor is reoxidized by carbon monoxide. The silica fume is added to the fume from the coke ash, silica, and intensifies the choking of the charge coluxnn. These two difficulties, due either to distillation of silicon from the bath (III) or to the formation of a silica erust on the surface of the metal bath (II), are both avoided by providing a suffleient volurie of slag tc) keep the metal covered at all times with a protecti_v-e bath of slag which prevents contact with CO; (causing oxidation) O_r siiper,heatin.@ (causing@ vaporization). It has seemed@ necessory to point out that the present invention provides for the production both of metal arl-d of slag (differing thus frorn present ferrosil,"con practice which does not produce slag) but tha-t this. provision cannot be complied with in any simple manner by charging the fumace with the "right. amount" of materials to produce some sele@c@ued sleg volume of some select ed composition. The amount of slag produced@ and its ce,mpositio,n are controlled not so much by the rel,.ttive amounts of the several materials charged as by the amount of silicon reduced, and this latter quantity depends primarily UPOL available hearth heat. The customary instruction to adjust@tlie relative amounts of coke, iron, silica and flux, to produce an "acid" slag of a specified slag vo-lume, is completely without meaning in the production of ferrosilicon. The slag produced may be "acid@" "b.asiell or "neutral" with a fixed charge depending upon how much of the Si(2 in the charge enters the metal as silicon or remains unreduced in the slag. If all the SiO2 is reduced tiae slag will comist Qnly of the CaO, MgO@ and A1203 in the charge and will not only not be "silicious" but will n6t even contain any silica a'u all. If none of the SiO2 is reduced, the slag will be so high in silica that its excessive viscosity will seribusly interfere with furnace performance. The bperator must knoiv, before selecting materials fbr the furnace charge, approximately how much SiO2 is going to be converted t6 silicon and how much is going to stay in the slag@ He can know this only if he knows what the available hearth heat will be. The available hearth heat depends upon Te, which depends upon (Sio2), which depends upon silicon reduction, which depends upon the available hearth heat. Since the available heat can be found by the method illustrated in Tables@ I and 2, the proper Pmount of each charge constituent, (coke, silica, flux, ore scrap, etc.) can be, determined without di@i%tultyWhen the value of the available heat has not been so determined, successful operation of the process is difficult to attain. The hearth heat balance (Tables I a-.Id 2) suggests that some advantage might be had, particularly at lower blast temperatures (including cold blast), from enriching the blast with oxygen. 2,420,064'. 22@ lying outside the scope of my inveneon, since known: methods of p@,-oducing an oxygen enriched blast are too@ expensive for operation with oxygen enric.@ied blast to be economical. The oxygen costs ir-ore tllan the fuel saving is worth. From '-;'ables I and 2 it is also obviou&that ridding the blast of its natural moisture i@@ functionally eqtiivalent to inereasing@ the blast temperature. To the extent that moisture present in atmospheric 10 air can be removed at low expense, increasing the availabl@e heat in this way is economical, and the qaving of coke may or may not pay for the cost of drying the blast. This depends on the method used for removing blast moisture. Within the 15 mean-ing o@f this explanation, the term "-blast of air" as used herein is sufficiently clear. The term "availability" of heat for carrying out smelting zone reactions is defined as the fr,,iction: itern (11) minus item (10) 20 Availability= item (11) in Table 2@ The ";available heat" (in B. t. u.) per lb. o'c- carbon is the sum of i'uems I to 9 in Table 2. ",Vhen the term "carbonaceous Luel" or "car2.5 boii," rr "solid fuel" is used herein, reference is made specif-cally to metallurgical coke, anthraei'@e, bi-tuminous coal (coltng and non-coking), ii.- rite, Peat (natural, dried, or carbonized), gasho7ose col@:e, petroleum coke, oil still residue, So briquettes of coal-, coke or sawdust, wood, charcolltl, etc., i. e., E),ny form of car on a s antial rart of wb-ich is non-volatile at high temperature. When the term "iron" is used as applied to a constituent of the furnace charge, reference is to 3r) any material containing iron, either as ore, or metallic scrap. Ore may be hematite, magnitite, mill scale, irori carbonate, iron silicate, roasted pyrites, etc. The sulphur content of the iron componenb of the charge may be higher than in 4( normal@ pig-iron practice. Where "silica" is designated as a charge component, it includes not only qliartz, quartzite, sandstone, shale and sand, but also silicates and alumino-silicates, hydrated or anhydrous, or any 45 material containing silica.
