Effect of pulp rheology on flotation of nickel sulphide ore with fibrous gangue particles A. Merve Genc1.2, I. Kilickaplan2,3 and J. S. Laskowski*2 Tests are carried out with suspensions of fibrous chrysotile particles and nickel sulphide ore containing chrysotile as a gangue. They corroborate published results, and indicate a very strong effect of particle shape on slurry rheology. Such suspensions exhibit time dependent properties. The flotation tests revealed that the viscosity of this system strongly affects flotation performance. At a solids concentration of 15-20%(w/w), the concentrate yield increases and the concentrate grade dramatically declines. This deterioration is observed when the Casson yield stress exceeds 1.5-2 Pa. The tests imply that the rheological properties of the investigated system are mainly determined by fibrous components. However, because of the presence of anisotropic mineral particles, the effect of particle-particle interactions cannot be separated from the effect of particle shape in this system. On effectue des essais avec des suspensions de particules fibreuses de serpentine et de minerai de sulfure de nickel contenant de la serpentine comme gangue. Ces essais confirment les resultats publies et indiquent un effet tres prononce de la forme des particules sur la rheologie de la boue. De telles suspensions exhibent des proprietes dependantes du temps. Les essais de flottation ont revele que la viscosite de ce systeme affecte grandement le rendement de la flottation. A une concentration en solides de 15 a 20%(poids/poids), la production de concentre augmente et la qualite du concentre decline dramatiquement. Cette deterioration est observee lorsque la contrainte de Casson excede 1.5 a 2 Pa. Les essais impliquent que les proprietes rheologiques du systeme etudié sont determinees principalement par les composantes fibreuses. Cependant, a cause de la presence de particules minerales anisotropes, on ne peut pas separer l'effet des interactions particule-particule de l'effet de la forme des particules dans ce systeme. Keywords: Flotation, Pentlandite, Serpentinite, Chrysotile, Fibrous minerals, Fine particle flotation, Rheology Introduction major constituent), after undergoing serpentinisation, are characterised by a high content of serpentine Nickel appears in two types of ores: sulphide ores and minerals(e.g. antigorite, chrysotile), the fibrous minerals lateritic ores. The major sulphides in the former are also referred to as asbestos minerals.12The presence of pentlandite, pyrrhotite and chalcopyrite. The ease with such minerals in the pulp changes the pulp properties which these ores are processed by flotation is determined entirely, and very strongly affects the flotation process. not only by mineralisation of the sulphides but also by Since deposits of high grade nickel sulphide ores are the rock hosting to the nickel sulphides. The basic being rapidly depleted, these ultramafic nickel resources (mafic) rocks such as peridotite (with olivine being a are becoming extremely important. A better under- standing of the fundamental properties of such ores is urgently needed. 'Metallurgical and Materials Engineering, Middle East Technical Rheological measurements are very useful in studying University, Ankara, Turkey the properties of fine particle systems. The rheology of fine 2NB Keevil Institute of Mining Engineering, University of British Columbia Vancouver,BC,Cana particle systems depends on many factors: particle size, Department of Mining Engineering, Eskisehir Osmangazi University, shape and solids concentration. In general, rheological Eskisehir, Turkey behaviour becomes more non-Newtonian as the particle Correspondingauthor,emailjsl@apsc.ubc.ca size decreases. Another factor that strongly influences 2012 Canadian Institute of Mining, Metallurgy and Petroleum Publlshed by Maney on behalf of the Institute Recelved 22 July 2011； accepted 3 February 2012 368 DOI10.1179/1879139512Y.0000000006 Canadian Metallurgical Quarterly 2012 VOL 51 NO4
Merve Genc etal.Effect of pulp rheology on flotation of nickel sulphide oreWETVGRINDINGCONDITIONING5-30%SOLIDS80%below1125μmFLOTATIONCLASSIFICATIONCOARSE-25μmFINECONC.TAILSFRACTIONFRACTIONaoedebineeFlow diagram oftests2safety regulations, could be used to handle asbestosScanning electron micrographofchrysotile particlesminerals.showing fibrous structureExperimentalMaterialsrheological behaviour are particle-particle interactions.The nickel sulphide ore (0.5%Ni) used in this studyWhile well dispersed systems of spherical particles atcontains pentlandite, millerite and violarite as the mainlow to moderate solid concentrations (<40%) exhibitbu joeasurueipeueonickel bearing minerals. A few per cents of sulphideNewtonian behaviour,the aggregating systems (thoseminerals in this ore are hosted in serpentinised ultra-with net attraction energy)are non-Newtonian.Themafic rock that is mostlylizardite,together with significantrheologyof the fine suspensions also depends on thequantities of antigorite and chrysotile.The ore typicallyshape of the particles;'the effect of the solids concentra-assays 40%MgO.9Themineral of interest in this study istion is much morepronounced if the particles are notchrysotile [a magnesium silicate MgsSi2Os(OH)4], due tospherical.its unique needle-like structure.In the field of dense medium separation, it is wellIn order to evaluate the effect of chrysotile concentra-known that the suspensions of spherical particles oftion,pure chrysotile suspensions were also prepared.Pureferrosilicon exhibit lower viscosities,for a given slurrychrysotile samples were obtained from Wards Chemicalsdensity and particle size, than the irregular particles ofand suspensions were prepared at different solids conground magnetite.4centrations. The grinding was carried out in a ceramicIncreased viscosity has an effect not only on the(o)ball mill to preserve purity. The grinding process for thisrheological properties but also on flotation performance.uuamineral was extremelydifficultand after a prolonged timeKlassen and Mokrousov claim that increased viscosityresulted in very thin and long fibres.Scanning electronof the pulp hinders aeration and the rise of mineralisedmicroscopy analysis (Fig.1)shows fibres that areaboutbubbles into the froth. Schubert indicates that many0-1μm thick and 100μmlong.researchers who reported beneficial effect of dispersingagents on flotation attributed these results entirely to theRuenkapausiendFlotation testsremoval of slime coatings. According to Schubert, theyBatches of 1kg of the precrushed ore (<3mm)wereneglected the reduction of the apparent viscosity.Everyplaced with1-5Lof the processwater(48-2gL-1ofeffect caused by thedispersants also involvesreduction ofNaCl,24-4gL-1of MgSO4.7H2Oandsomeminorthe pulp viscosity, thus directly increasing the collisionquantities of potassium and calcium chlorides) in a steelrate and improving flotation kinetics. In addition to this,rod mill. The duration of grinding was adjusted such thatturbulence intensity is reduced when viscosity increases80% (w/w)ofthegrindingproduct was<125μm.Washingand so, as also pointed out by Schubert,6 when theout with another 1.5 L of the process water resulted in aparticle size decreases or solids concentration increases,pulpwitha25%solids content (w/w).Theproductsoffourthe apparent viscosity of the pulpincreases.This results ingrindings combined together were classified using a C700a sharperdecrease inthe size of theturbulent zone aroundMozley cyclone rig at a cut point of 25 μm.The yields ofthe impeller.Thisturbulence damping'effect is reducedunderflowand overflowwereabout45and55%(w/w)when the pulp viscosity is lowered.respectively.The overflow was left todecant,and after 4 h,In this project the rheological properties of the sus-a clear supernatant was siphoned off. The sediment afterpensions of serpentinised ultramafic nickel sulphide oredecantation contained~30% solids (w/w).This solidscontaining chrysotile as gangue are investigated to find aconcentration is changed by either adding or diluting thepossible correlation between pulp viscosity and flotation.pulp as necessary in order to see the effect of solidsThe results reported in this paper constitute a part of aconcentration (Fig. 2).larger project supported by BHP Billiton.7.8 The projectFlex-41 xanthate (equivalent of isobutyl xanthate) ofrequired building a special lab that, in accordance with180 g t-1 was used as a collector (40% was added to theCanadian Metallurgical Quarterly2012VOL51NO 4369
MerveGenc et al.Effectofpulprheology onflotationof nickel sulphideore310+5% sold→10%solld9815%solld20%sold2,5(ed)ssans seas72a654T3Bingham Model1000000000000000000090,5Casson Model-Data5010015020000ineeaeShear Rate (1/s)050100150200250Shear Rate (1/s)4 Effect of solids concentration (ww%) on rheologicalcurves obtainedforslurries of studied nickel sulphideore3Example of rheogram of slurry with ground nickel-sul-phideore.fittedwithCassonmodelandBinahammodel:The Bingham yield stress values,which can beobtainedas this example demonstrates, obtained t and tc valuesas the intercept of the linear portion of the experimentalare not identicalcurve extrapolated to the shear rate of zero,mayovershootthereal yield stress values.Forthediscussedmill, the remaining 60% was added in the flotation cellcase, the real yield stress values are much closer to thewith10min of subsequentconditioning.Teric 407frotherCasson yield stress values.However, in the case of pureof 30g t-was conditioned forI min in all thetests.Thechrysotile suspensions,the Bingham model, whichis muchBUUWJOfrothproduct was scraped off fromthe lipofthe cell everysimpler,adequatelydescribestheexperimentaldata5 s. Concentrates were collected in constant time inter-vals. Natural pH of 8-8.5 was used at all times and allStress decay testsflotation tests were performed multiple times for everyDanseirereoSince some rheological tests indicated that the apparentcondition to allow for statistical evaluation.viscosityofthetestedsuspensionsdependednotonlyonRheologicalmeasurementsthe shear rate but also on the time over which the systemhas been sheared (see for example Fig.4) shear decayRheological tests were conducted on the samples of thetests were also carried out. These tests were conducted toflotation pulp obtained from the cell just before aeratingcharacterise the time dependent properties of the suspen-the pulp to initiate flotation.The samples were immedi-sion by using the Haake Viscotester 550.For theatelytransferred to a controlled rate coaxial rheometerdetermination of the time dependent flow behaviour, aHaake 550 Viscotester in order to eliminate the effect of豆constant shear rate was set and the independent variablessettling.riinwere recorded as a function of time.Measurements wereAll measurements were carried out using an MV-1P bobcarried out in duplicate,For each run, a fresh sample wasand cup fixture (p"indicates a ribbed surface).The shearused.These tests involved the measurement of the shearstress versus shear rate data were recorded. Rheogramsstress over a 100 s period at predefined shear rates.were generated in the shearing rate range between O andEquilibriumflowcurvesweregenerated from stressdecay200 s-1for 1min to obtain complete curves.TheflowRauenepeusintestsbyplottingtheequilibriumstressversustheshearcurves were modelledby using either Bingham or Cassonrate.During the stress decay tests, at increasing shearmodels.The Bingham model was developed to describerates (10, 50 and 100 s-) the shear stress values werepseudoplastic behaviourrecorded after 100 s of shearing at each rate.(1)T=TB+npIYResultswhere t is the shear stress,t is the Bingham extrapolatedyield value, np is the plastic viscosity and yis the shear rate.Initial flotation tests were carried out to study the effectFor some systems the flow curves are better describedof pulp density on flotation. The experiments wereby the Casson modelperformed varying solids content in the flotation feed;the concentrates were collected after 2, 5, 10, 20 andt=[t +(ncn)/2/g(2)30 min of flotation.Every set of conditions was repeatedwhere t is the shear stress, tc is the Casson yield stress, nc isthree times and the mean nickel recovery values werethe Casson viscosity andis the shear rate.The equation ispresented as a function of time in Fig. 5.of the same form as the Bingham relation; however, theIt is observed that the recovery of nickel has similarexponentsaredifferentandtherelationshipisnotlinear.Itvalues at all solids content with the exception of thewas found that the Casson model describes better thehighest solids content (20%) tested. At 20% solidsexperimental data obtained with the suspensions of thecontent,recovery is obviously lower,as some turbulencenickel sulphide ore (Fig.3);for the suspensions of pureeffectswere visually observed.Gas dispersion within thechrysotile, the Bingham model provided a better fit.cell was significantly affected and a large number of large370Canadian Metallurgical Quarterly2012VOL51NO 4
Merve Genc et al.Effect of pulp rheology on flotation of nickel sulphide ore1with increasing pulp density, suggesting that an increase+5%Solids90in solids recovery translates into higher entrainment.-0~ 10% SolidsThe grade-recovery curves in Fig. 7 also reveal that8015% Solidsconcentrate grade is affected dramatically when pulp20%5.70density is increased from 5to 20% (w/w).At this stageOANrheological experiments were carried out to providefurther insight into the observed flotation behaviour.4The flow curves obtained with the flotation feedvarying in solids content are given in Fig.4.The CassonYyield stress values obtained from theflowcurves are20showninFig.8.10Figure 8 shows the correlation between the calculated0Casson yield stress values and the solids concentration.S20250s10304As seen, the yield stress is highly dependent on the solidseoneTime (min)concentration. As the solids content increases, so does5 Nickel recovery as function of flotation time at varyingthe yield stress of theflotation pulpsolids contentThe concentrate grade is plotted versus the Cassonyield stress in Fig.9.The figure shows that the grade ofpuethe flotation concentrates clearly deteriorates when theCasson yield stress reaches a certain value. These criticalbubbles were evident.This phenomenon seems to beKJnyield stress values seem to be in the range of 1-5-2 Pa.similar to that described by Schubert.6Since Fig.7 seems to indicate the time dependenteeThedifferences in recovery are not very obvious inproperties of the tested suspensions (especially at highFig.5. However, when the concentrate yield (masssolids content),this was examined further.Theresults ofrecovery) is compared with the nickel recovery (Fig. 6),DuNthe stress decay tests are shown in Fig.10. These testsatendencyto shiftto the left with increasingpulpdensitywereconductedat20%(w/w)solids content,atais evident.Therefore, the same nickel recovery requiresconstant shear rate of 10, 50 and 100 s-1.Equilibriumhigher concentrate yields (mass recovery)when pulpoflow curves were generated by plotting the equilibriumdensity increases.nsreeeostress versus the shear rate. The apparent viscosity wasThe cleanest flotation conditions were apparentlyrecorded as afunction of time.created when the solids content in the pulp was verylow (5% solids)and over the first 2min when the contentDiscussionof the valuable sulphides was relatively high. WhilerecoveryincreaseswhentheconcentrateiscollectedoverTheflow curve shown inFig.3 istypical for a5 min, the grade clearly deteriorates (point B), and thepseudoplastic system.gradedeteriorates further for longerflotation times.ThisIt is also known as shear thinning and as this term(9)indicates that while true flotation is best at the beginningsuggests, it is characterised by a gradual decrease inof the flotation tests, entrainment increases with time.usianeuenapesienapparent viscosity with incereasing rate of shear.Figure7demonstrates that the concentrate grade declinesTherefore, this material behaves like an elastic solid for30+5%Solids.·-- 10% Solids2515%Solids*-20%Solids20ea5o020406080100Nickel Recovery (%)6Mass recovery-nickel recovery curves for the flotation tests carried out varylng pulp density after 2(A),5 (B),1o (C),20 (D)and 30 min (E)of flotation:10% solids content only has two points;one of them is for2min and the other isfor30minflotation;errorbarsrepresent95%confidenceintervalofmean2012VOL51NO4371Canadian Metallurgical Quarterly
Merve Genc etal.Effectof pulp rheologyon flotationofnickel sulphideore95%Solids8-0-10% Solids一0—15% Sollds1eaieninioao→*20%solldsB1naonereE030402050607080Nickel Recovery (%)7 Mean concentrate grade versus Ni recovery curves for flotation tests carried out at varying solids content after 2 (A),5 (B),10 (C),20 (D)and 30min of flotation (E):error bars represent 95% confidence interval of mean of repeatedexperiments(usually v/v%),provide the curve displayed in Fig.11.As this plot indicates, beyond a volume fraction OmaxSGuuwocalled the maximum packing fraction, the dispersediparticles lock into a rigid stucture and flow ceases.Thepacking fraction for hexagonally packed monodispersespheres is 0-74. As pointed out by Barnes et al.3 thesesvaluses are much smaller for other particles (only 0-3for1grains,and0-2forrods)The relative viscosity function depicted in Fig. 11 intreipeuethe dilute region is determined by the requirement thatAthe slope at the origin equal the intrinsic viscosity (2.5o+for spheres).The function is best decribed by the0510152025Krieger-Dougherty equation1o,11Sollds Concentration (w/w%)yBuusigna8Cassonyieldstresssolidsversusconcentration[n]max1relationship(3)nr-Tomasufficiently small stress but once some critical stress iswhere n, is the relative viscosity,n is the viscosity ofexceeded the material yields and the resulting flow maysuspension/colloid, n。is the viscosity of solvent, is theresemble Newtonian behaviour. Such measurements,volume fraction of the solid,max is the maximumENwhen viscosity is plotted versus solid concentrationpacking fraction and [n] is the intrinsic viscosity.apeshn6shear rateof10s+shearrateofS0s10.85sdeshearrateof100g-1泉l i peioeuouo广0.60.4020+00L23a56204060806100120Casson Yield Stress (Pa)Time (sec)9Effect of Casson yield stress on grade of flotation con-centrates:5% solid (→),10% solid (),15% solid (A)10Apparent viscosity versus time of shearing at differentand 20% solid ()shear rates at 20% solids content372Canadian Metallurgical Quarterly2012VOL51NO 4