Laboratory Procedures for Mining Pelletizing Characterization of Iron Ore Concentration

8/13/2019 Laboratory Procedures for Mining Pelletizing Characterization of Iron Ore Concentration

1/9

Laboratory Procedures For

Determining Pelletizing Characterist ics

f lro n O r e C o nc en tr ate s

by T E Ban and L J Erck

A discussion of laboratory procedures used to determine pelletquality and to simulate handling and firing conditions. Strength-temperature relationships in pelletizing; effe ct of chemical addi-tives on agglomeration; and crushing, abrasion, and impact re-sistance of the pelletized products are discussed with respect tofive different iron ore concentrates.

N most routine experim ental work on products andprocesses, method:; of operation an d product eval -

uation are established from accurately controlledlaboratory tests. With particular emphasis on thepelletization of fine iron ore concentrates, st and ard -ized procedures of lab orato ry production a nd producttesting have been established at Cleveland-CliffsIron Co. Research Laboratory at Ishpeming, Mich.A program of this natu re was used for investigatingthe agglomeration characteristics of various iron oreconcentrates. The factors which control the processwere studied and subsequently controlled for theseinvestigations.

Pelletizing as app.lied to iron ore agglomerationinvolves th e prepara tion of sp heroidal masses froma fine moist concentrate, followed by high-tempera-ture firing to produce a final product. The prepara-tion of moist sphercrids is genera lly desig nated a sballing This produces the so-called green pellets orballs. After, these ha ve been fired they ar e calledpellets Balling is acc'omplished by rolling moist con-centrates in a revolving drum until small spheroids

form by a snow-balling action. These balls are thenheated in a shaft fur nace , or a modification of thisdevice, to a tem pera ture high enough to cause ag-glomeration within the individual balls while theydescend countercurrent to hot oxidizing gases. Bythis process the lour strength balls are convertedthrough various chemical and physical-chemicalprocesses into hard pellets. With correct balling andfiring, the pellets are strong, resist abrasive actionand handling treatments, and provide a suitableblast furna ce and open hearth feed..

T. E. BAN, Member AI, ME, is Research Metal lurgist and L. JERCK is Chief Met allurgi st a t Clevela nd-Clif fs lron Co., Ishpeming,Mich.

Discussion on this paper, TP 355 28, may be sent 2 copies) toAl ME before Oct. 31, 1953. Manuscript Oct. 24, 1952 . ChicagoMeeting, September 1952.

In 1 9 5 Tigerschiold and Ilmoni developed a work-ing hypothesis to explain t he formation an d stabilityof green pellets.' Their work was presented on thebasis of geologists' discoveries co ncer ning the soil-water system and C. V. Firth's early postulates ofgree n pellet formation.' Fund ame ntal rea ctions oc-curing within fired pellets have been described byCooke and Ban, who showed the microstructurechanges caused by different firing temperatures. Allthese investigators have provided scientific clarifica-tions of agglome ration by th e pelletizing process an dhave opened avenues for greater development.

Mineral grain grow th and ceramic bond formationare the tw o basic phenomena imparting coherencyto particle s w ithi n a fired pellet of iron ore . Each ofthese is established by the tra nsfer of t herm al energyto the mineral particles. Grain growth mechanismsare chiefly physical reactions wherein smaller par-ticles consolidate into larger ones, with resulting de-crease in surface energy conforming with th e secondlaw of the rmodynam ics. Ceramic bond formation isa chemical reaction usually between basic and acid

metallic oxides of th e gangue and or e minerals. CaO,MgO, A1,0,, an d FeO of F e30, , eact in v aryin g pro-portions with SiO, or Fe,O,, to fo rm a grea t se ries ofcompounds or am orphous solid solutions. These prod-uct bond iron mineral particles together by a bridg-ing action or by direct reaction with th e surfaces ofthe particles. The result of the grain growth-ceramicbond formations is a consolidated, coherent compactof gangue and iron minerals fo rme rly held togetherby much weaker bonds of interstitial water andforces common to dry, unconsolidated particle^.^With respect to iron ore agglomeration, these hightemperature bonding reactions have been describedby Schwartz, who referred to the bonds in sinters,and by Cooke and Ban, who referred to the bondsin fired pellets.

The extent of the bonding phenomena and theconseque nt stre ngt h of fired pellets is governed by

TRANSACTIONS AlME AUGUST 1953, MINING ENGINEERING-803


2/9

some o f t h e following fil.ing conditions and prop-erti es of' the unfired pelle t: l--mine ~'alogical com-position of the pe llet, 2---particle packing within thepellet. 3-particle size range of the min eral s com-prising the green pellet, $-maximum tem per atur eto which the minerals are exposed and the retentionperiod at this tem pera ture , 5-the rate of attain men tof an d rate of recession from the maxim um tem pera-tu re , 6-the gas composition of the atm osph ere inwhich the pellet particles are fired, and 7-the hightem pera ture chemistry of th e minera l particles andthe ad ditives of the pellets.

The minera logical composition of th e con cent ratedenotes the quality and q uant ity of therma lly activecompounds. Acid and basic oxides when present incrit ical amounts will materially promote interparticleconsolidation by a ceramic bond formation , and ironminerals have grain growth properties peculiar tothe ore. If t he gangue constituents are present inabnormal quantities, pronounced slagging will occurwithin the fired pellet, and iron particles will beweakly bonded by excessive glassy constituentswhich inhibit the iron particles from interlockingby grain growth.

Thermal barriers must be overcome before manyceramic reactions occur. The free energy of a chem-ical reaction must b e of a negative val ue before thereaction occurs, and the greater the negative valuethe greater the reaction tendency becomes. At con-stant entropies and enthalpies, a negative free energyof reaction becomes more prom inent w ith increasingtemperature. A higher temperature generally in-creases a chemical reaction tendency and rate. Thissame higher temperature generally increases par-ticle mobility and solid diffusive characteristics con-tributing to the grain growth mechanism.

A rapid r ate of temper ature increase lessens thecompaction of a gre en pellet, owing to th e rapid

expulsion of air or bound w ater . This causes minera lparticles to become less intimate for the bondingreactions. A rapid temperature increase also causesa duplex structure to result within a green pelletof ma gnet ite. If sufficient heat is transfe rred t o thecent er of a pellet prior to oxygen diffusion a nd mag-neti te oxidatio n, the core will consolidate to coarse-grained magnetite, while the shell or outside surfaceoxidizes to hematit e. The coarse-grained ma gnetitecore becomes less susceptible to subsequent oxidationby vir tue of the coarse structu re and the presenceof slag coating s caused by t he rea ction of FeO withSiO,. The duplex-stru ctured pellets, hem atite shellsomewhat separated from a magnetite core, have a

lower strength index than pellets containing anhomogeneous netw ork of iron oxide. Rapid coolingof heated iro n ore pellets causes bond and latti cestrains from differen tial ther mal contraction. Thisis common to quenched vitreous materials.

If the atmosp here of t he fired iron ore particlesis not of homogeneous composition, differential ag-glomeration will come about within the pellet be-cause of particle ch emistry differences. Hem atite,mag neti te, and the lowe r oxides of iron definitelypossess different properties of agglomeration. If someof th e iron minerals a re contiguous to reducing sub -stances during firing, the agglomeration character-istics will differ from particles fired in an am bientatmosphere that is oxidizing in nature.

In a generic sense, ideal agglomeration is affordedby the following conditions: l-the pre senc e of anoptimu m quanti ty of slag-for ming constitue ntswithin the concentra te, 2-the use of an optimu m

firing tem per atur e, 3-a slow rat e of temp er atu rt~increase duri ng firing. 4-a slow ra te of temp er atu redecrease afte r firing, 5-a long retention time at theoptimu m firing tem pe rat ur e, 6-a high degree ofcompa ction, an d 7-an homo geneou s oxidizing am -bient atmosphere during the firing period.

The Cleveland-Cliffs Research Laboratory staff hasdeveloped an apparatus for evaluating the relativeresponse of various fine concentrates toward pellet-ization an d for enabling a study of the variou s addi-tives or factors which aid the process. This appar atuswas designed and oper ated to simulate th e conditionsimposed by the shaf t fur nac e firing of iron ore pellets.For thi s purpose, some of the ope ratin g conditionsof a pilot plant shaft furn ace were obtained and in-corporated within th e laboratory device.

Sha ft furnac e firing of green iron or e pellets iscarr ied ou t by th e placing of w et pellets on the topof a s haft h ear th con taining p ellets in th e process ofbeing fired. As the charge descends into t he shaf t bygravity, the wet pellets are fired by a countercurrentstream of combustion and heat tr ansf er gases. Vary-ing temperature zones within t he fu rnac e performthe following consecutive operations on the descend-ing we t pell ets: l-drying, 2-preheating, 3-firingto a maximum temperature, and 4-cooling.'

The regions of the hot zones depend upon th e tem-pe rat ur e of the hot gases, the chem istry of th e pellets,and th e rate of descent an d ascent of t he pellets andgases respectively. Ideal shaft furnac e operation re-quires isothermal temperature zones transverse toa path of u niform gas flow and charge descent. It isvery difficult to obtain these conditions because ofpellet particle segregation and inter-p ellet consolida-tion, both of which dist urb idea l gas flow and tem-perature patterns.

The regions of different temperature zones withinthe fu rnace shaft can be varied by operational con-trols, but a specific zonal region can also be a func-tion of the high tempe ratu re chem istry of t he orepellets. If pellets contain an intrinsic fuel, magn etiteor an additive which can exothermally oxidize withthe gases, a hot firing zone can literally be flo tedwithin the shaft when relatively low temperaturegases are allowed to ignite the pellet burden andmaintain continual combustion within the descend-ing charge. In this ma nne r descending pellets ofmagnetite can be ignited with oxidizing gases of1000C and a zone of burn ing pellets can be estab-lished t hat is 1300C in temp erat ure. The positionand mai nten ance of this zone gives rise to a longi-tudinal temperature gradient within the furnace

shaft, which consequently describes the changes ofpellet tempe rature with time.The locations and tem pera ture s of various zones

in a pilot plant shaft furnace were determined withprobing and stationary thermocouples during manypelletizing tests. From these data and from thegeometry of th e furna ce sha ft, the ra te of pellettempera ture change, Fig. 1 was computed. The tw ocur ve s of Fig . 1 show the results of these com puta-tions from furna ce tests using pellets of h ema titeand magnetite. The hematite pellets contained nofuel and the maximum temperature was impartedentirely from the tem per atur e of the hot gases fromthe combustion chamber. For this reason hematitepellets had a relatively low rate of temperat ure in-crease. Magnetite pellets, on the oth er h and, burn edin a zone above the entran ce of the hot gases andgave rise to a rapid temperature increase. The rateof tem pera ture decrease was approx imately th e same,

804-MINING ENGINEERING AUGUST 953 TRANSACTIONS AlME


3/9

Fig. 1-Time-tempe rature relationships during pilot pla nt andlaboratory firing of iron ore pellets.

however, regardless of the kin d of o re comprisingthe green pellet.

Air and fuel oil combustion products were theprincipal gases used -lor the pellet heat trans fer andfiring medium respectively. The heat transfer gaseswere admitted to the pellets from the furnace bottomand ascended countercurrent to the shaft burden ina quantity which was adjusted to the specific heatcapacities of or e and air . Thes e ar e 0.262 to 0.268:usually, therefore, 1 lb of a ir was used to abs tractth e heat from 1 lb of ore pellets. The combustiongases were transversely admitted to the pellets inthe upper portion sha.ft in a quantity adjusted to thethermal req uirements of the pellets. This quantitynaturally varied from test to test, but averaged about0.8 Ib of co mbu stio n pro duc ts to 1 Ib of p ellet s. Th e

composition of the combustion prod ucts in tu rn wassubject to variation in th at different quan titie s ofexcess air were used to provide different tempera-ture s. Combustion produc ts analyzed on th e orde r of6.42 pct CO,, 9.4 pct I,, 7.5 pct N and 6.7 pct H,O.Both the heat transfer gas and combustion gas mixednear the pelletizing hot zone, so that a n overall 1.8 lbof gas we re adm itt ed to 1 lb of o re; the gas analyz edabout 3.0 pct COI, 15.3 pct O,, 78.6 pct N,, and 3.1 pctH,O.

Laboratory Procedure for Firing Pellets

Shaft furnace pellets undergo a temperaturechange with time during th e transition from th e wetstate to the fired s tate in accordance with t he curvesof Fig. 1, an d about 1.8 lb of gas containi ng 15.3 pct0 ar e admitt ed for each pound of pellet c harge. Thisknowled ge was applied to th e operation of t he lab-oratory apparatus, see Fig. 2 used to simulate theshaft furnace firing conditions. Apparatus consistedof a n electrical, globar combustion tube furn aceequippe d with provisions for passing gas through t hetube at controlled flow rates. Nichrome boats whichserved as the pellet-clharging devices we re adm ittedthrough th e open end of t he combustion tube, passedcountercurrent to the gas stream, and removedthro ugh th e upstream end of t he tube. The boatchar ge consisted of five wet pellets, each of wh ichwas abou t 35 g in wei.ght and 28 mm in diam. These

were formed in a laboratory balling drum by useof procedures described by Tigerschiold and Ilmoni.The dru m products were stored in sealed containersand retained for the subsequent firing tests. To min-

Fig. 2-Apparatus for firing iron ore pellets.

imize pre-drying effects, green pellets representinga specific balling drum product were taken from thecontainers and placed on the boat approximately 1min before the actual firing operation.

During operation of th e laboratory appar atuspellet-laden boats were adm itted one by one into thecombustion tube and moved periodically through th etube so that, pellets could approach the consecutivetem perat ure changes of sh aft furnace pellets. Forthis purpose a standardized firing schedule was de-veloped fro m a knowledge of the furn ace tempera-ture gradient curv e of Fig. 2 and from the heatingra te curv es of Fig. 1. When fired by this sched ule,individual pellets were periodically changed in tem-perature in accordance with the circular points ofFig. 1, while a co unt ercu rrent stre am of air was

passed over the pellets at a rat e of 3.50 liters per min.This air flow corresponded to 1.23 mols of 0 permol of Fe,O,, which in t ur n was rough ly equiv alentto the oxygen supply of iron o re pellets during theshaft furn ace firing ope ration.

The gra ph of Fig. 1 denoted several instanceswhere the laboratory firing schedule did not exactlyduplicate th e firing schedule of the sh aft furnace.The laboratory cooling rat e was more rapid, theheating rate was a mean of the pilot plant rates forhematite an d magnetite, and the heating ra te variedslightly from pellet to pellet.

These discrepancies were tolerable. It would havebeen impractical to allow laboratory pellets to coolfor 10 hr , and the slight variations of th e heatingrates were of a small order when contrasted to t henormal variations of heating rates within a shaftfurn ace. The results of th e laboratory tests were tobe relative rather t han absolute. For this reason t hesha ft furna ce firing schedule was approached bu twas not attained.

Pellet Quality DeterminationsThe quality of fired pellets can be determined by

different mate rials tests to d enote th e toughness,hardness, an d compressive strength . Each of theserelates a meas ure of pellet particl e coherency. Ironore pellets should be able to withstand the crushingforces imposed by furnace and stockpile stacks and

the impact and abrasive forces imposed by conven-tional iron or e handling and shipping. The physicalproperties that mak e it possible f or pellets to endu rethese forces are ge nerally rela ted, althoug h not d ef-

T R A N SA C T IO N S A I M A U G U S T 19 5 3 M I N I N G E N G IN E E R IN G - 8 05


4/9

PERCENT MINUS 1 M E S H MATERIALY

6 25 12 5 25 50 100

QU NTITY D R O P P E D - LBS.

Fig. 3-Drop t e s t r e l a t ionsh ip be twe en pe l l e t quan t i ty andsubdivision.

initely. Brit tle pellets can conceivably withstandgreat compressing loads, yet will shatter remarkablyfrom impacts. Tough pellets, likewise, can be com-prised of coarse particles which read ily abrad e fromthe surfaces. For these reasons, each specific stren gthproperty was individually tested by the use of s tand -ardized materials test procedures. These were thecrush, impact, and abrasion test s used to denotepellet stre ngth , toughness, and hardness, respectively.

Str eng th is a measure of abil ity to endure stress.In th e case of slow diamet ric compression of sph er-ical material, stress is a function of sphere size andcompressing force. When like-size spherical pelletsare subjected to a slow diametric force, the strengthand maximum unit stress are directly proportional

to the crushing load. Consequently the pellet strengt hdesignated herein was approximated from a meas-urem ent of th e maximu m load imposed on an indi-vidual pellet, or the ext ent of the compressing loadwhich caused the pellet to crush.

The tests were performed by compressing a pelletbetween a hydraulic piston and a fixed steel plate.The hydraulic pressure required to crush the speci-mens was indicated by a Bourdon gage. The pressurewas converted to pounds of load, in t ur n recorded

a s the pellet-crushing stren gth. Pellets representinga specific test invariably yielded strength valueswhich we re not identical. This was influenced by :1-pellet size differences, 2-pellet sha pe differences,and 3-inadequate crushing techniques wherein thecrushing force was not applied at a constant rate.Laboratory procedures only minimized the effect ofthese factors and the differences in crushing loadsfor pellets representing a specific batch were conse-

quently minimized. Therefore, the average crushingload of 20 individ uall y tested pellet s was used torepresent the pellet-crushing s tren gth.

The abrasion test was developed to find a meansof expressi ng the relati ve resistance to abrasive actionoffered by th e surface s of pell ets. This abrasive actio nwas caused by inter-pellet rubbing induced by pelletcontact and movement in a small rotating d rum. Thedrum consisted of a short cylinder mou nted a xiallyon a variable speed motor. The dimensions were

6 in. diam by 6 in. long. Exactl y 1000 g of p ellets,about 35 in number, were placed into the d rum andthe drum was rotated for 200 revolutions at 24 rpm.This rotatio nal velocity was about 50 pct of t hecritical velocity so that the pellet movement waslargely a cascading and rolling action. When th e 200revolutions were completed, the pellets and the finedust produced from the test were removed andweighed separately. The weight of the abraded dust,which was all 65 mesh material, was used to ex-press the results of the abrasion test. The int actabraded pellets and th e fine dust produced from thesurface were the only products of the test.

The impact test was developed for determ iningthe toughness of the pellets or th e relative resistanceto subdivision from impact forces. These forces we reimparted by dropping the pellets successively frompredetermined height to a solid steel plate surface.

Exa ctly 1000 g of lab orato ry pellet s, abou t 35 in

number, were measured for size and placed into acontainer mounted 1/ 3 f t above a %-in. steelplate surface. The container was fixed with a trap-door exit from which th e pellets were allowed todrop through a 4-in. pipe guide onto the plate. Thepellets and fragments were successively droppedthree times and a screen analysis through 28 meshwas made of the final product.

A stren gth inde x, or th e percent of the originalparticle size, was computed from the screen analysis

Ta ble I. Determ ination of Average Particle Size by Screen Analysis by Coghill s Me tho d

Original Product.. .

Dropped Product

Screen Size

Mesh In.

Screen Ordinal Screen OrdinalA n a ly s i s , N o .* A r b i - A n a lys i s , N 0 . t A r b i -

Wt, Pct trary) Wt, Pct Wt, Pet trary) Wt, Pct

34.7 = 1.347 100.0 134.7 413.0 = 4.130 413.0

00 100

Th is o rd in a l n u mb er i s b e tween 1 an d -in. scree n size. f Th is o rd in a l n u mb er i s b e tween 6 i n . an d meshB y in terpolat ion the average s ize equals : s c reen size. By in terpolat ion, the average s ize equals :1.050 0.347 1.050 0.742) = 0.943 0.371 0.130 0.371 0.263) = 0.357

Average s ize after impa ct 0.357Th e p erce n t o r ig in a l s i ze -st ren g th in d ex = 37.9 p c t

Average s ize before impact 0.943

8 0 6 -M I N IN G E N G I N E E R I N G A U G U S T1953 T R A N S A C T I O N S A l M E


5/9

Fig. 40-Intact pellets before testing. Pellets of hem atiteconcentrate, 1000.0 g, containing 2 pct limestone and fireda t 1 300 C in laboratory apparatus.

of th e pellets before dropp ing and t he screen analysisof t he pellets and fragmen ts after th ree drops.Coghill's method of mean mesh determination wasadapted for th is purpose.' This provided a means ofestablishing the av erage size of th e particles directlyfrom t he screen analysis. An example of this deter-mination is shown in Table I.

This index, along with t he percen t of fine mate-rial ( pct -10 mesh) rela ted the toughness of th epellets, or the ability to endu re the shock of impactforces. In eff ect this w as provided from t he ratioof the averag e particle diamet er after testing to theaverage particle diameter before testing.

Significance of the Product TestsThe drop tests were carried out on 50-lb samples

of pellets produced from th e pilot plant s haft fur -nace and on 1000-g batches of pellets produced inthe laboratory furnace. With the percent originalsize as a streng th index, the severi ty of th e test ap-peared to be a functio n of the qu anti ty of pelletsused for a specific test. Wh en 50 lb of pellets w er edropped one at a time, a very small numb er of pel-lets were exposed to the greatest impact; that is,only a few pellets from each 50-lb batch receivedth e primar y impact because of the relatively smallprojected area directly under the drop test guide.Pellets directly on the bottom of a load received t hesteel plate impact, while pellets in the center or topof t he load wer e subject to hindered falling and aresulting cushioning due to the mass effect. Thisdropping condition did not exist for the 1000-g droptest because nearly every pellet and fragment wassubjected to an impact force which was a functiononly of th e particle mass.

Several tests were conducted to determine themass effect of t he dro p test and to determine th ereliability of results procured from a specific test.Approximately 700 lb of pellets of a uniform qualitywere thoroughly mixed for these tests and varyingquantities were cut out from the entire sample.These qua ntit ies consisted of t wo 100-lb portions,two 50-lb, two 25-lb, two 12.5-lb, two 6.25-lb, andeight 1000-g portions. Each portion was droppeden masse with th e exception of f our 1000-g portio nswhich were dropped in a manner to conform withsingle particle impact. This was accomplished byindividually drop ping each pellet and pellet fra g-ment onto th e cleaned plane surface area. This typeof dr op test natura lly afforded the most drastic im-pact, or the most favorable conditions for particlesubdivision by impact. Th e graph in Fig. showsthe dro p test results fr om using different quantitiesof pellets fo r a specific test batch. I t was observedthat th e percentage of -10 mesh material producedfrom the impacts was independent of the qu anti ty

Fig. 4b-Broken pellets aft er drop testing, 33 1/3-f t dropsonto a -in. ste el plate surface. Drap test results: 3.5 pc t-10 mesh material; 59.4 pct original size.

dropped, and th e percent original size decreased w ithdecreasing quantities of dropped mat erial. This wasunderstandable because the greater impact forceswere impar ted to a greater percentage of the pelletcharges. The produ ction of fines ( pc t -10 mesh mat e-rial) was caused by both abrasion and impact, abra-sion being more severe with the larger quantities.Th e fines produced from t he impact of droppingsmaller quanti ties of pellets w ere compensated forby th e norma l abrasion of d roppin g larg e quantities.The net res ult of both thes e effects yielded an inde-pendent value to the fines production with droppedpellet quantities.

Four 1000-g batches of -ll/g 1 in. pellets werehan d picked for uniformity of qu ality and wer egiven an identical drop test to determine the reli-ability or degr ee of reprod ucibility of t he test. Theresults as indicated by t he fines production and per-cent original size did not vary from an average bymore t han 30 or 7 pct respectively. This discrepancywas tolerable and of s mall enough magnit ude towar rant the drop test procedure as a reliable meas-urement of pellet toughness.

Figs. 4a and 4b show 1000-g batches of iron orepellets before and after the three 33 1/3-ft drop tests.These pellets we re composed of fine hema tite con-cen tra te conta inin g 21/2 p ct lime sto ne and w er e firedto 1300C in the laboratory app aratus. The drop testresults indicated that these pellets had a relativelyhigh impact resistance for iron ore pellets, 3.5 pct-10 mesh and 59.4 pct original size. Som e direct-mined ore specimens subjected to the same droptest, -1 1 in. specimens, 1000 g in weight, 3drops of 33 1/3 f t, had a greater impact resistancethan the pellets. Below are listed some drop testresults on severa l kinds of iron ore from th e Cleve-land-Cliffs Iron Co.:

I l e m 10 M e s h P c l Original Size Pot

Cliffs shaft l u m p o r e 2.9 79.4Tilden siliceous ore 1 7 70.9Cliffs group ore 9.2 28.8

Some of these relatively high values may be gr eate rthan requisite values. For 'this reason the drop testresults from the iron ore specimens were used forcomparative purposes rather than as standards foriron ore pellets.

The most expedient method of measuring the de-gree of parti cle coherency within an iron ore pelletis the crushing test. A specific test carried out as ameasu rement of v alue, such as the crushing stren gthof fired pellets, is only as good as t he de gree of re -producibility of th e test. If values a re assigned ascrushing strengths of pellets treated in a standardmanner, t he values should be attained with reason-

T R A N S A C T I O N S A l M E A U G U ST1953, M I N I N G E N G IN E E RI N G- -8 0 7


6/9

mean value, and the num ber of measuremen ts.In effect, this signifies that the probability is one totwo that t he mean value will have a residual de-noted by the probable deviation.

Table I1 gives the individual crushing values of15 kind s of iron or e pellets fired by t he laborator yfiring schedule to a maxim um tem per atu re of 1300C.Th e significance applied to a crushing val ue for aseries of pellets is indicated by t he deter mina tion ofth e probab le deviation of t he mean of 20 individuallycrushed pel le ts . The f ig~~res ndicate the averagecrushing values, the probable deviations of the

zoo average values, the best values, and the probabledeviations of the best v alues sta ted as percentages.

T C Y P C I A T V I E - . C It was observed that the probable deviations of thebest values we re no larger than 12 pct and no lesstha n 2.3 pct for t he 15 sample measurements. Thesesmall deviations indicated that the laboratory firingand crus hing t,ests wer e reasonably r eproducible.

Th e abrasion resistance of pellets is dependentupon both th e particle coherency and the frictionfactor of pellet su rfaces. Coarser or e particles tendto increase the friction factor, and these coarser p ar-ticles abrade m ore readily t han equally bonded fine

particles. This logic was substantiated from the com-paris on of abras ion test resu lts of pellets com prisedof t he original ore concentrates a nd pellets com-prised of reground ore concentrates , see Table 111

T E M P E R A T U R E -C

Figs. 5 to 8-Strength1 tempe rature relationships of iron ore Results of Laboratory Pelletizing Tests

pellets. The laboratory tests were carried out on five dif-feren t iron or e concentrates, designated a s M-1, M-2,

accuracy of the test A measure of M-3, and H-1, H-2. The and screen analysesthis reproducibili ty is denoted by the probable de- of these are presented in Table IV. M-l and M-3viati on of t he mean of a series of me asure ment s, a nd repr esen t two different concentrates of na tur al mag-the magnitu de of this deviation indicates th e pre- netite produced from magnetic concentration cir-cision of t he measurements. This is math emat ically cuits. H-2 and M-2 repre sent concentrates of art i-deter mine d by use of t he equation: ficial hematite and artificial magnetite produced by

I1 roasting iron sulphides under different atmosphericQ = 0.6712 N N 1 Zr conditions. H-1 represents a flotation concentrate of

specular he matite. Each of th ese ores possessed pecul-wherein Q equals th e probable deviation of the mean iar properties, a nd as such responded differentlyvalue, th e residual of the measured value and th e toward agglomeration by pelletizing.

Table II . Crushing Strength of 20 Randomly Selected Pellets Expressed with Probable Error of Average V al ueP e l l e t s 5 r e d a t 1300C

H 1 H 2 M 1 M 2 M 3

ddi tives , Lb Pe r Ton 10 b 10 b 10 b 10 b 10 10 b 10 b 10 b 10 b 10 b 10 b 10 b 10 b 10 b 10 b50 Is 50 1s 10 c 40 c 30 c 10 b x

Conc. Condition MR RG RG Su r- RG 50 P c tface RG

lx n 1460 730 670 1250 goo sno fioo 680 1250 600 fisn 9nn fino 79n

~..

1200 2500 820 620 620 630 820 1100 540 1250 300 53 0 fino 4zn nxn

Average 1773 1650 769 572 1224 1009 946 780 528 1450 765 689 927 653 882probable dev. - Q 2 1 2 9 k 1 38 k 3 0 k 1 3 k 5 4 k g 9 5 5 4 5 3 0 +-23 2 6 8 ~ k 8 8 -C32 53 29 k 3 8Best value 1800 1700 770 570 1220 1000 950 780 530 1450 760 690 650 880Probable dev., pct k 7 . 2 k 8 . 4 2 3 . 9 k 2 . 3 k 4 . 4 + 9. 9 k 5 . 7 k 3 . 8 k 4 . 3 k 4 . 7 k 1 1 .6 2 4 . 6 2 1 4 . 5 k 4 . 3

The additives were ad ded in pounds per ton of ore . s denot es pulverized lim estone.b denotes bentonite c lay. bx denotes borax.c denotes pulverized anthr acite coal. RG denotes a reground concentra te .MR denotes a magnetically roasted conc.

808-MINING ENGINEERING AUGUST1953 TRANSACTIONS AlME


7/9


8/9


9/9

1400-

ELLETS CONT INING XB OR X

- PELLETS CONT INING NO DDIT IVE

Fig. 14-Strengt h-tempe rature relationships of M-2 iron orepellets showing the effect of pulverized borax as an additive.

lets with lower impact and crushi ng resistance. Thesetwo reground ores, M-1 an d H-2, showed great tend-ency to spa11 while being fired, tha t is, the o re com-pacted too tightly in the wet stage, and the expulsionof balling wa ter wa s inhibited d urin g the firing proc-ess. App roxi mat ely 21 pct of t he pellets composed of

regro und M -1, 9 pct I-omposed of re grou nd H-2, a nd5.2 pct composed of 50 pct reg roun d min us 50 pctoriginal M-1 spalled during the drying stage of thefiring process. No spalling occurred duri ng th e firingtests of all oth er pellets.

Th e beneficial effect of 21/2 pct l imest one in H-1pellets and the less remarkable beneficial effect ofborax within M-2 pellets are shown by the strengthproperties given in Table 111. Visual examination ofthe cracke d fragm ents of the se pellets denoted con-choidal glassy fract ures which exemplified t he bond-ing action from the additives, especially so whencontrasted to the rough crystalline fractures of thesame pellets containing no additives.

The appa rent det rime ntal effect of in tern al car-bonaceous matter in iron ore pellets was evidentfrom the decreasing strength properti es of H-2 pel-lets containing increasing quantities of in tern al coal.Pellets containing 2 pct coal on the exterior, how-ever, appeared to have greater strength propertiestha n pellets containin g lesser amoun ts of coal in-ternally. This demorlstration indicated that withinpellets containing internal coal, gaseous diffusionlagged behind heat l.ransfer, and carbon abstractedoxygen from the ore rather than from the hot gases.This reaction caused direct reduction of th e iron or eand consequent defects in the pellet structure, suchas hollow cores or contracted, sintered volumes fromthe phase changes at the v ery high pelletizing tem-

perature. The reduced products reoxidized in someinstances to the higher oxid e stat es of iron, bu t leftevidence of past history as a poor pellet st ructu rewith a low materials strength.

Provisions were m ade for be tter gas-solid reactionsduring the pelletizing tests by the application ofcoal to the pellet exterior. This naturally preventeddisruption of t he pellet interior, and se rved as apractical means of conveying fuel into th e pellet-izing zone of the laboratory apparatus.

Pelle ts comprised of H-1 con vert ed to artificialmagnetite and mixed with 2% pct limestone werecompara ble in all s tren gth properties to pellets ofunconverted H-1 ore containing the sam e quantity

of lim estone. From this con version some of t hechemical advantages of ma gne tite were obtaine dwithout detriment to the pelletizing process. Thecommercial production costs of pelletizing hematitewill unquestionab ly be great er tha n tha t of pellet-izing magn etit e because of t he int rinsi c source ofthermal energy within magnetite. This cost differ-ence, if app recia ble, could dicta te th e necessity ofbeneficiating some fine-grained hematite ores by themagnetic roasting process.

SummaryThis paper describes the design and operation of

a laboratory apparatus for pelletizing iron ore bysimultatio n of shaft furn ace firing conditions.

Strength-tem perature relationships were determinedfor five different kin ds of iron ore c oncentrate s rep-resenting products from various regions planningto beneficiate iron ore by agglomeration. In additionto these relationsh ips, th e effects of vario us chemic aladditive s and various methods of concentra te prep-aration were investigated. All the investigationswere discussed with reference to the crushing, abra-sion, an d impact resis tanc e of t he pelletized products.

cknowledgementThe authors wish to express their appreciation to

th e Cleveland-Cliffs Iro n Co. for th e support of t hisinvestigation and also for permission to publish theresults.

References'M. Tigerschiold and P. A. Ilmoni: Fundamental

Factors Influencing the S tren gth of Green and BurnedPellets Made from Fine Magnetite Ore Concentrates.AIME Blast Furnace and Raw Materials Proceedings(1950) 9 p. 18-45.

C. V. Fir th: Acelomeration of Fine Iron Ore. AIMEBlast Furnace a n d ~ a w aterials Proceedings (1944)4, pp. 46-65.

.'S. R B. Cooke and T. E. Ban: Microstructures inIron Ore Pell ets. Trans. AIME (November 1952) 193pp. 1053-1058.

F. H. Norton: Refractorie s, 3rd ed., p. 165. M cGraw-Hill Book Co., Inc., New York, 1949.

.;G. M. Schwartz: Iron Ore Sinter. Trans. AIME(1929) 85, pp. 39-66.

T . E. Ban: The Agglomeration of Fine TaconiteConcen trates by Pelletizing. Unpublished Master'sthes is, University of Minnesota, 1951.

E. W. Davis and H. H. Wade: Agg lomeration of IronOre by the Pelle tizing Process. Inf. Circ. No. 6, MinesExperiment Station, University of M innesota , 1951.

W. H. Coghill: Evaluating Grinding Efficiency byGraphical Methods. Enginee ring and Mining Jour nal(1928) 126 p. 934.

R . B. Sosman and H. W. Merwin: Preliminary Re-port on the System, Lime: Ferri c Oxide. Jour nal of theWashington Academy of Science (1916) 6, pp. 532-537.

'S. R. B. Cooke and W. F. Stowasser: The Effect ofHeat Treatment and Certain Additives on the Strengthof Fired Magnet ite Pell ets. Trans. AIME (December1952) 193 pp. 1223-1230.

F. G. Donnan : The Agglomeration of Gra nula rMasses. Trans. Faraday Society (1919) 14 pp. 12-13.

J D. Zett erstrom: Oxidation of Magne tite Concen -tra tes . U. S. Bur. Mines R.I. 4728. 1950.

TRANSACTIONS IME AUGUST 1953 M I NI N G ENGINEERING-811

Laboratory Procedures for Mining Pelletizing Characterization of Iron Ore Concentration

Documents

Transcript of Laboratory Procedures for Mining Pelletizing Characterization of Iron Ore Concentration