Somefactorsinfluencing goldrecovery bygravity concentration · 2009. 8. 26. · African gold-recovery practice in 1961, pointing out that two-thirds of the major plants incorporated

Some factors influencing gold recovery by gravityconcentration

SYNOPSIS

by M. D. BATH,. M.A.Se. (Met.) British Columbia (Visitor)A. J. DUNCAN,. B.Sc. (Hons.) Rhodes, M.S.A.C.I. (Visitor)E. R. RUDOLPH,* M.Se. (Eng.) Rand (Member)

Recent studies have shown that the overall extraction of gold from Witwatersrand ores is increased as the pro-portion of gold reporting to the amalgamation circuit is increased. Testwork has therefore been carried out with aview to improving recovery by gravity concentration.

In this paper, the history of gold recovery by gravity concentration is reviewed, and methods and machines atpresent in use are briefly described. The results of plant and laboratory tests are presented, and the limitation ofexisting concentrators is discussed.

A pilot plant is being built for the testing of improvements to the existing concentrators and of several newdevices, particularly the Reichert-cone concentrator.

SINOPSIS

Onlangse studies het getoon dOltdie totale ekstraksie van goud uit die Witwatersrandse ertse toeneem namate diehoeveelheid goud in die amalgamasiekring toeneem. Daar is dus toetse uitgevoer met die oog op die verbeteringvan die herwinning deur swaartekragkonsentrasie.

Hierdie verhandeling gee 'n oorsig oor die geskiedenis van goudherwinning deur swaartekragkonsentrasie enbeskryf kortliks die metodes en masjiene wat tans in gebruik is. Die resultate van aanleg- en laboratoriumtoetseword uiteengesit en die beperkings van die bestaande konsentreerders bespreek.

Daar is 'n proefaanleg in aanbou om die verbeterings aan die bestaande konsentreerders en verskeie nuwe toe-stelle, veral die Reichert-keelkonsentreerder, te toets.

HISTORICALGravity concentration of gold has

been used since antiquityl but itsimportance has decreased since theturn of the century, when the cy-anide process was developed. In theearly days of the Witwatersrand2,before the introduction of cyanid-ation, the ores were stamp-milledand the gold recovered by amal-gamation. This practice became ob-solete as the oxidized ores becameexhausted. The introduction of tubemills with corduroy-blanket strakeson the discharge to collect the coarsegold, which was recovered by amal-gamation later, gave rise to thecyanide plant that we know today.

The vastly more efficient cyanideprocess has caused many plants todispense with gravity concentration,although they tacitly use longerresidence times in the cyanide solu-tion to compensate for the lack ofconcentration. Doff and Bosqui3emphasize the importance of theearly removal of gold from themilling circuit and advocate gravityconcentration, especially for thoseores in which a significant propor-tion of the gold is associated withbase metal sulphides.

The history of modern gravity

...Anglo American Research Laboratories,Anglo American Corporation of SouthAfrica Limited

concentration on the Rand minesgoes back to the corduroy blankets,which were highly labour intensive.As a result, the endless-belt con-centrator, the Johnson drum con-centrator4, the plane table, andjigs came into use. Jigs have notgained wide acceptance, but thetwo concentrators are still in currentuse on Anglo American Corporationmines. An additional bonus in theform of osmiridium is realized onplants employing gravity-concen-tration circuits, and this often out-weighs the cost of the treatment.

Douglas and Moir5 reviewed SouthAfrican gold-recovery practice in1961, pointing out that two-thirdsof the major plants incorporatedgravity concentration followed bycyanidation of the bulk ore, up to73 per cent of the gold being re-covered by gravity concentrationwhen the gold was relatively coarse.

In this paper we shall attempt toput the argument for gravity con-centration in a more balanced light,especially since computer correlationstudies have shown it to have asignificant effect on the overallplant recovery.

GRAVITY CONCENTRATORS

In 1961, the following methods ofgravity concentration were in use:(1) corduroy-blanket strakes,

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

(2) Johnson concentrators,(3) plane tables,(4) endless-belt concentrators, and(5) jigs.

Corduroy-blanket strakesThese, or rubber-blanket strakes,

were apparently still in use on 24plants in 1961. They were placed atthe tube-mill outlets, and weremanually washed at intervals torecover the concentrates. Themethod was highly labour-intensiveand constituted a security risk and,as far as is known, has now beenentirely replaced by mechanical con-centrators. This change has beenaccompanied by circuit modificationsto provide selective grinding and bythe installation of as few machinesas possible.

J ohnson concentratorThis mechanical concentrator6 was

developed in 1925 and has sincegained wide acceptance. It consistsof a continuously rotating cylindricaldrum, having its axis inclined atbetween 2,5 and 5 degrees to thehorizontal, and lined internally withriffled rubber. Pulp flows down thelength of the drum, and the heavyminerals settle in the riffles and arecarried out of the pulp stream to thetop, where they are washed into alaunder by sprays. In 1961, minesusing these machines in conjunction

JUNE 1873 383

c"nc to

belt

.1amalgamation

oncentrator

I

tail to

high grode .econdary

cyclone .teady head

with endless. belt concentrators forupgrading the drum concentratereported an average gold recoveryof 38 per cent by concentration andamalgamation, the range being from30 to 52 per cent.

Plane tableThis device, developed at Rand

Leases, consists of riffled rubbercovering a series of inclined smoothsurfaces, with steps between them.The pulp flows down the length ofthe riffles, and the riffles collect thegold concentrate, which is drawn offthrough slots at the steps. The faster.flowing pulp is carried across theslots and is collected at the end ofthe table. At Rand Leases, the plane-table concentrates were regroundand re concentrated on secondaryplane tables, followed by a finalconcentration stage using Jamestables. In this circuit, 51 per centof the gold was recovered by amal.gamation.

Endless-belt concentratorThe belt concentrator was de-

veloped at Brakpan in 1949 to re-place corduroy blankets. The unitconsists of a flat, endless rubber belt150 cm wide, with 300 cm betweenhead and tail pulleys, and inclinedat a slope of 3 degrees to the hori-zontal. The upper surface of the belt

,'Hue t from

primary lIIi!lint

cyoniclation

364 JUNE 1973

has saw-tooth riffles, 5 mm deep witha pitch of lOmm, running across thebelt. The belt moves continuouslyagainst the pulp stream at a speed of30 cm per minute. A spray water-pipe washes concentrate from theriffles after the belt has passed overthe head pulley. The unit has beenused as a primary concentrator and,more usually, as a cleaner concen.trator for the retreatment of theproduct from J ohnson drums.

JigsThese machines have never gained

wide acceptance in the South Africangold industry, probably owing tothe large bulk of concentrate. AtDurban Roodepoort Deep and West.ern Areas, Yuba.Richards jigs areused. At Western Areas, the jigstreat secondary mill discharge, pro.ducing a concentrate that is de.watered and reground before re.concentration in secondary jigs.Tailings from these jigs are returnedto the regrind-mill cyclones, andconcentrates are cleaned on endless.belt concentrators and a Jamestable. The overall plant recovery isof the order of 30 per cent by gravityconcentration.

ANGLO AMERICAN PRACTICEOf the Group's 13. plants, 8

practise gravity concentration. Goldrecoveries by amalgamation rangefrom 21 to 52 per cent.

Only two types of concentratorsare in use, in all cases the Johnsondrums acting as roughers and theendless belts as cleaners. Flowsheetsare similar in all plants, differencesbeing in the number of concentra.tors, pulp feed rates, and the ratioof Johnson drums to endless belts(Figs. la and 1b). Table I givesmechanical details and operationaldata for the concentrators in thevarious plants.

Typically, secondary tube.millproduct is pumped to tertiary cy.clones whose underflows are distri-buted among several Johnson drumconcentrators. The tailings are re-turned to the mill circuit, while theconcentrates are retreated on end.less- belt concentrators, there usuallybeing one belt for every 4 to 8Johnson drums. The belt concentrateconstitutes the feed to amalgamationwhile the belt tailings are returnedto the mill circuit.

As part of a major project aimedat increasing gold recovery in AngloAmerican plants, improvements ingravity concentration are being in.vestigated both in the laboratory andon the plants. The project is divided

.econdory

mill.

Fig. la-Concentration with secondary milling

pump

JOURNAL. OF THE SOUTH AFRICAN INSTITUTE OF MINING AND MET AL.L.URGY

primary product

cyanidatian

into two phases: the first is shortterm, with its object the optimi-zation and improvement of existingmachines and circuits, and thesecond is longer term-the develop-ment of alternative machines and/orcircuits. The motivation for thisinvestigation was the finding, duringa computer correlation study ofplant operating data, that the overallgold recovery is increased as theproportion of gold recovered byconcentration and amalgamation isincreased. Without going into detailsthis is obviously due to the fact thatrecovery by. amalgamation is moreefficient than that by cyanidation.

GRINDING AND STRAKING

Grinding tests were carried out todetermine liberation grinds for oresfrom several of the Group's mines.The term "liberation" as used inthis paper is best defined as thatparticle size which maximizes profitfor the subsequent process. Hence,for cyanidation or any leachingoperation, the mineral grain needbe only partly exposed, whereas,for gravity concentration, it must bealmost completely separated fromthe gangue.

In an ore as unhomogeneous as thegold-bearing reefs, there can ob-

secondary

mills

Fig. Ib-Concentration with tertiary milling

viously be no single liberation size-the larger particles of gold areliberated at a much coarser grindthan the finer particles, for examplethose enclosed within pyrite grains.

Tests were carried out to deter-mine what percentage of the goldcould be recovered at various degreesof grinding. Ore samples were milledfor various periods and the productspassed over a small vibrating strakecovered with riffled rubber. Curvesfor recover'y versus grinding areshown in Fig. 2 for three differentsamples of ore. Recoveries in therange 75-80 per cent were achievedat relatively coarse grinds. The VaalReefs ore, in particular, yielded agold recovery of 75 per cent withoutany milling at all (crushed to pass1,5 mm). As milling was continued,recovery increased slightly to 79per cent at 93 per cent minus 74 /km.The curves for Western Deep Levelshigh-grade and low-grade ores showincreasing recovery up to about 75per cent minus 74 /km, after whichthere is a marked decrease in re-covery with finer grinding. Thiseffect is thought to be due to aflattening of the liberated goldparticles as the result of overgrind-ing. The flattened particles settle farless easily than spherical grains, and


amalgamation

tertiary

mills

are more likely to be lost to thetailings. The fact that the sameeffect was not observed with theVaal Reefs material is probably dueto the much coarser gold in this ore-even when flattened, the grains arestill effectively caught in the riffles.

The results of these tests suggestthat gravity concentration could beeffectively incorporated much earlierin the circuit than it is at present,and that high recoveries can beobtained with simple equipmentvery similar in principle to that al-ready in use. Early recovery ofliberated gold has long been ad-vocated but somehow seems to havebeen forgotten over the years. Thedeleterious effects of overgrinding-not only flattening but also com-minution-of the gold, would thusbe avoided.

TESTWORK ON THE JOHNSONDRUM CONCENTRATORS

A series of tests was carried outon one of the low-grade Johnsondrum concentrators at Western DeepLevels Mine. This drum had beenmodified to facilitate sampling, anda variable-speed drive was installed.

Composite samples of the feed,concentrate, and tailing were takenunder various conditions of solids

JUNE 1973 365

-.....

~0

::'" 20

~«

feedrate, feed pulp density, anddrum speed. The least varied was thepulp density, which was adjustedonly to ensure that the pulp did notsand out in the drum, the highestpossible pulp density and hence thelongest residence time possible beingthus achieved. The gold content ofthe feed was an interfering variable,ranging from 91,5 to 197,6 g/t, butno correlation was found betweenthis and the concentrate grade,which varied from 920 to 5468 g/t.The mass of concentrate was 0,44t/h at the lower figure and 0,07 t/hat the higher. The solids feedratewas varied from 5,1 to 26,4 t/h.The percentage recovery was foundto be inversely proportional to thesolids feedrate (Fig. 3), while the

100

,,

80

.~DD

...

...:0III.~c.~~

60

...c:D

.!

El....i]~

~

0

10

Feedrate

100

...Z..,VIll:..,A.

VRWICS

WDLIHG

PERCENT MINUS 74 ~m

Fig. 2-Strake tests-recovery versus grind

8

actual yield of gold (kg/h) wasroughly constant (Fig. 4).

Fig. 3 is a least-squares fit ofpercentage gold versus feedrateand shows gold recovery to increasehyperbolically with decreasing ton-nage. This is to be expected sincethe residence time of the pulp in thedrum is increased when the tonnageis decreased. What is not so ex-pected is that the actual mass ofgold recovered per hour remainsconstant, as can be gauged fromFig. 4. Here the masses of goldrecovered per unit time are plottedon normal probability paper inorder of increasing magnitudeagainst the cumulative observedfrequency. The fact that a reasonablystraight line is obtained indicatesthat the differences observed aredue to experimental error and thevalues are not distinguishable fromthe mean. The explanation for theconstant mass recovery is to befound in the fact that, although theresidence time is increased with de-creased tonnage, the mass of heavyminerals, including gold, is reducedproportionately.

>-Ill:..,

~V..,Ill:

:t~.

0

0

Cb

tonnes/hour

0 00

0

20 30

Fig. 3-Johnson drum concentrator-recovery versus feed rate

366 JUNE 1973 JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

2 .

5

10

20

~ 30

>- 40~e~ SO1

60

e..; 700

1 60u

85 0

118

88400 600

Au Reco.ery

600 1000

0

latter for a small-scale belt con-centrator specially made for thepurpose.

Plant machines

The two plant machines are eachfed by the combined concentratesof three high-grade and three low-grade Johnson drum concentrators.The results represent day-shiftsamples taken over a period of sixmonths and are hence inherentlymore reliable than those obtainedin the laboratory. The average goldrecovery was 63 per cent, iron re-covery (including tramp iron) was29 per cent, and uranium recoverywas 9 per cent. The concentratemass was 3,8 per cent.

The uranium recovery is con-sidered to be of interest since someof this uranium comes from thelow -grade circuit via the J ohnsondrum concentrators, and an im-provement in this could be of con-siderable economic benefit. Doublingthe speed of one of the belts resultedin an overall increase in total goldrecovery by amalgamation of 2,25per cent, but at the expense of in-

gromsthour

Fig. 4-Johnson drum concentrators-Gold recovery versus cumulative frequency,showing the recovery to be normally distributed

The constant mass recovery ofgold per unit time is also a featureof the high-grade drums, whereabout 1 kg of gold was recovered perhour as compared with the 0,55 kgof gold per hour on the low-gradedrums. This is attributed to themuch higher heavy-mineral contentof the high-grade circuit, the mass ofgold being assumed negligible inparts per million.

The Johnson drum concentrator isa cheap, easily run machine. Al-though it is not very efficient as aunit, it nonetheless, because of alarge re circulating load, recovers upto 70 per cent of the gold enteringthe plant. Screen analyses showedthe recovery of gold to be poorestin the minus 43 /Lm fraction andbest for the coarse sizes. This is incommon with all gravity-concen-tration machines and is due to thelower settling rates of the fineparticles.

TESTWORK ON THE ENDLESS-BELT CONCENTRATORS

Both plant7 and laboratory resultshave been obtained for these con-centrators, the former for two ma-chines in routine service and the

1.!'0.cc!

4 x 1.1t no.1

0 1.lt no.2I

....J:I"'V!.:.0! 1::t

oC

2 3 5 64

F..d to b.1t Au kgltFig. 5-Endless-belt concentrators-recovery as a function of feed grade

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY JUNE 1973 367

creasing the concentrate mass by If:>nearly 50 per cent. '1A -<I~. ,=:2;

0was pre- ~....::

percentage recovery, the latter pass- ;- ~~~~'".Q"';:...oO re: Conclusion ~+++1 -+0 bv~>.15 -+0 '0" ---

~§x .;

b()~ca

From the combined results of theQ)

88 ~0>00 -+0 ~00 > .~~------ 03 b() ~0>~plant and the laboratory, it seems ..c: .8 8 ... .8 >:: 0> ca

~....15 1e -+0-+0 88 '0 ~~> 6 03

-+0b() ~Q)03 ca .t: ...

gthat, given a lower feedrate at a >:: ... ~00 -+0~~~5b > b() g ~g '"higher grade, it would be possible, 'S

~..r;; 00 -+0 '0 ;§>::

~~~00'0 '"0

'"0

'"~.0 ~03 §

'">:: g

by adding wash-water to the belt '"~0 ~"O

~Q) .01 ~000 Z 113 en en ;S ... ~~U ~0

~368.JUNE 1973 .JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY

... behind the feed point, to Increase...'E< 10, H ......""

0> """

>'IAx ......

'"0 a Johnson drum concentrator; and01

>'I 00 000> ...

'" '"0001 X""

0 """' 00""

' 10' ..:..;..:....; M M~P

f'iII> ...... ......

"""00 0>

0~~'"

0> a Reichert-cone, which is to be in- O! grade concentrate will be recycled.

~8~oj

§ ... CYCLONES ASa:>t)"'"

.;::CONCENTRATORS. 's

a:>

'"bi5 ... 6D I> bII g a:> 1'1 bII..0 ..a ", " ;§ I> a:> rather than classifiers. This is par-a:> "'" "C O!8 p.. a:> ~'" "C

t) 0 t)a:> a:> '0 a:>

oj1'1

t)1'1 "'"'"

., 0 p.. "O a:> a:> '(;j 0 a:> 0 0 ticularly true of the Western Deep;n Z U3 00 00 ~~p:; E-< Q ~Q E-<

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY JUNE 1973c 369

Solids as - eed-Test feed water rate Belt Conc. Mass Rp- Calc.no. % flowrate relative speed ratio % covery head

I/min units cm/min % Au g/t--- --- ~-

I 23 1,8 4,31 40 8,2 9,1 74,4 365,02 23 1,0 4,31 40 2,5 37,7 92,5 513,73 30 1,6 5,91 40 10,3 7,4 76,5 708,84 29 1,6 5,67 40 7,9 10,5 82,8 858,95 35 1,6 7,18 40 9,3 7,.~ 70,0 362,46 6 . 0 2,00 40 2,1 45,4 93,5 1311,77 35 1,1 7,18 40 6,6 1l,7 76,8 365,38 25 1,4 4,75 40 6,2 12,7 79,3 533,79 15 1,6 2,66 40 5,5 12,2 67,4 204,5

10 6 0 2,00 40 4,3 18,8 81,1 1149,211 20 1,6 3,67 40 6,1 10,9 66,1 207,112 30 1,4 5,91 80 5,2 14,9 77,9 247,413 25 1,8 4,75 80 5,7 14,4 81,8 282,5

20

10.!!!

".!!iie

iv5 -

0 weight

. recovery

t. 1

perunt

Fig. 6-Endless-belt concentrator-recovery and mass fraction of concentrate asa function of concentration ratio

TABLE IIRESULTS OF TESTWORK ON THE LABORATORY BELT CONCENTRATOR

W h F

concentrate gradeConcentration ratio =

feed grade

Computer result of regression analysis:

Concentration ratio = 0,555 (feedrate) (wash-water flowrate) +2,578

Standard error of concentration ratio = 1,489

Standard error of variable = 0,123

= 4,50t-test

370 JUNE 1973

Levels circuit, where there is notertiary milling. In other circuits(Fig. lb), the tertiary cyclones arealso concentrators for the feed tothe Johnson drums, but the Johnsondrum tailings are dewatered inanother cyclone before being milledin the tertiary mills, thus receivingsome classification. It is not possiblefor a given cyclone to perform bothfunctions simultaneously at maxi-mum efficiency - an increase inclassification efficiency automaticallyresults in a decrease in concentrationefficiency. This correlation betweenconcentration inefficiency and classi-fication efficiency is understandablewhen one considers that, for highconcentration efficiency, the cycloneis expected to make a separationin which coarse, light particlesreport to the overflow, and fine,heavy particles (gold-pyrite) reportto the spigot. This is contrary tothe normal tendency of the cycloneto act basically as a classifier. Thispoint will be discussed in moredetail later.

If the above is accepted, a de-cision must be made on which of thetwo functions-classification or con-centration-is the more importantfor the tertiary cyclones. Since thetertiary overflow is reclassified in thesecondary cyclones, low classifyingefficiency in the former is notserious. Any coarse, light gangue,(possibly containing unliberatedgold) that may overflow the ter-tiaries will be returned to thesecondary mills via the secondarycyclone underflow. It would seemreasonable therefore, to aim at in-creased concentration efficiency inthe tertiary cyclones at the expenseof classification efficiency. The aimshould be maximum recovery ofgold in the cyclone underflow atmaximum grade. Obviously, oncethis is accepted, it would theoreti-cally be possible to replace thecyclones with any other more effi-cient concentrator. However, thecyclones offer so many advantagesfrom a maintenance and operatingviewpoint that it seems logical toretain them, and to attempt toimprove their concentrating effi-ciency. As far as is known, noserious attempt has been made toimprove the operation of the tertiarycyclones as concentrators, and the


present investigations were moti-vated by the considerable scope forimprovement in this section of theplants.

Recent years have seen the de-velopment of various eyclones speci-fically for concentrating purposes.Their physical design is such thatclassification effects are suppressedand the influence of particle specificgravity is maximized. These are notheavy-medium cyclones, which havebeen in use in the mineral-processingindustry for some years, but truehydrocyclones. They are charact-erized by blunt cones (i.e., largeincluded apex angles) in the range60 to 180 degrees and long, large-diameter vortex finders. These de-velopments are largely a result ofwork in the coal industry8-12, wherecyclones operating with a watermedium are now in wide use for theupgrading of fine coal. Investigationshave also been carried out into theuse of cyclones for the beneficiationof cassiterite13, 14and iron ores, and arecent Russian paper15 describes theconcentration of gold from milledconglomerate ores by use of a short-cone hydrocyclone.

As mentioned above, classificationcyclones are of slender design, witha narrow cone angle-usually 15 to30 degrees. In these cyclones, theparticles must discharge eithm'through the vortex finder or theapex orifice, and the basic require-ment for separation is the achieve-ment of balance between the centri-fugal forces accelerating the particlesradially outward, and the centripetaldrag forces. For this purpose, rc-latively large vortex-finder cleal-ances are common in classificationcyclones, because this assists inproviding the time for particles inthe central flow region to reachtheir terminal free-settling velocityrelative to the entraining water.

In contrast, concentrating cy-clones are of squat design, with awide cone angle-in the range 60to 180 degrees-and are operatedto suppress classification phenomenain favour of gravity-concentrationeffects. Briefly, they are based ontwo hypotheseslO; (a) a particle bedstratifies according to density alongthe conical wall of the lower cyclonesection, and (b) particles enteringthe central spiral flow separate

during the initial phase of theiracceleration in a radial direction.The former hypothesis has led to theselection of wide-cone angles forconcentrators, because particlesstratify more reliably along theshallower wall of a wide cone thanwould be possible along the steepwall of a slender cone. The latterhypothesis has led to the preferencefor relatively short vortex-finderclearances in concentrators, becausethis design assists in separatingparticles that swirl towards the

Feed.. plantSplitt.,

box

reie"

'ydone

1/3.

00

~pump

.a!"pl'~

vortex-finder before the centrifugalmass and the centripetal drag forcesacting on these particles attainequilibrium. Since it is the dragforce that increases the relativeinfluence of particle size on separ-ation, this force is to be suppressedin the operation of cyclones asconcentrators.

The Compound Water Cyclone

The Compound Water Cyclone(CWC) was invented by J. Vismanand developed by him at the Mines

instrument

lank

TO

sp!>fJe, TONNAGE BOX

-ox OR

REJECT

....

RlJEC T


Fig. 7-Pilot plant no. I

JUNE 1913 371

FEED FROM PLANT

- 25'-to lont

",een 0 lompl;ngU/S

TO

MOOULE 2

cyclone feedpump

FROM10 plonl

MODULEmiddl;ng relurn

from

Re;chorl cono

Ro;cherl

IAna]ysis Distribution

%Ore/Test Fraction Mass ' 1--% Au, g/t S, % Au S

-Vaa] Reefs West-2° mill Conc. 10,2 467,6 37,4 88,4 92,6

discharge-high-grade Tai]ing 77,7 6,8 0,18 9,8 3,4Reichert-cone simulation S]imes 12,1 8,25 1,36 1,8 4,0

Ca]c. head 100,0 54,0 4,]2 100,0 100,0-

Vaa] Reefs West-2° mill Conc. 10,0 442,0 39,1 88,7 89,4discharge-low -grade Tai]ing 68,4 5,9 0,25 8,1 3,9Reichert-cone simulation Slimes 21,6 7,45 1,35 3,2 6,7

Ca]c. head 100,0 49,8 4,37 100,0 100,0-

Western Deep Leve]s-2° mill Conc. 4,7 581,3 20,9 66,9 88,8discharge-high-grade Tai]ing 87,2 14,35 0,1 30,6 7,9Reichert-cone simulation S]imes 8,1 12,65 0,45 2,5 3,3

Ca]c. head 100,0 41,00 1,10 100,0 100,0-

Western Deep Leve]s-2° mill Conc. 3,8 409,3 23,5 64,9 89,6discharge-low -grade Tai]ing 89,4 9,0 0,09 33,4 8,1Reichert-cone simulation S]imes 6,8 7,1 0,34 2,0 2,3

Ca]c. head 100,0 24,1 1,00 100,0 100,0-

Western Deep Leve]s-be]t- Conc. 0,9 30198,6 66,2concentrator tailing-one Midd]ings 7,6 1419,25 26,3pass On table Tai]ing 91,5 33,8 7,5

Ca]c. head 100,0 410,6 100,0Conc. & 8,5 4469,9 92,5mid. I

TABLE HIREICHERT-CONE SIMULATION AND OTHER TESTWORK

Branch, Department of Mines andTechnical Surveys, Canada1°, 16. Itconsists of a rather short cylinder,fitted with one of three alternativecompound cones (Fig. 9). Eachcompound cone has three sections,with included angles of 120, 75, and20 degrees. In each case, the widestangle is nearest the cylinder andthe smallest at the spigot, but thethree types differ in the relativeareas of their inner cone surfaces.The inlet is conventional, but thevortex finder is rather larger andlonger than that of an ordinaryclassifying cyclone.

Visman16 explains the concentrat-ing action taking place in his cycloneas follows (see Fig. 10).

Particles of different sizes andspecific gravity form a hinderedsettling bed in Section I of thecompound cone. Light, coarse par-ticles are prevented from penetratingthe lower strata of this bed by thecoarse, heavy fractions and by thefine particles filling the interstices ofthe bed. Consequently, the waterpassing from the periphery of thecyclone chamber towards its mainoutlet (the vortex finder) erodes thetop of the stratified bed and sub-stantially removes the light, coarseparticles via the "central current"

around the air core.The remainder of the bed is forced

into the second conical section (II)by new feed entering the cyclone,substantially without losing its stra-tified character. Here, the centralcurrent is much stronger and erodesthe top of the bed, where themiddlings are now exposed. Thelight middlings are swept up anddischarged through the vortex finder.The heavy middlings that spiralupward in the central current maybypass the orifice of the lowervortex finder owing to their higherspecific gravity. Consequently, thecoarse heavy -middlings fractiontends to recirculate to the stratifiedbed and finally enters the thirdconical section. In this last section(Ill), the bed is finally destroyed ascoarse particles fan out along thecyclone wall in a single layer, ex-posing the small particles that sofar have been protected from beingwashed out. The central current inSection III is relatively weak, as ithas nearly spent itself in the pre-vious sections. The upward currentthat remains separates the smallparticles from the remainder of thematerial, with preference for thoseof low specific gravity. Thus thefine, light particles are finally dis-


charged through the vortex finderby a process of elutriation. Theheavy particles, fine as well ascoarse, are discharged through theapex. The separation thus takes placein three steps, as shown diagram-matically in Fig. 10.

TESTWORK AND RESULTS

Tests were carried out using boththe CWC 2M (50 mm) unit and theCWC4 (100 mm) unit. The testcyclone was fed from a centrifugalpump, and the overflow and under-flow were returned direct to thepump feed tank. Sampling wascarried out by cutting both productstreams simultaneously, generallyfor a period of 5 seconds. The cy-clone was operated in a verticalposition for all tests.

All tests were performed on aslurry of Western Deep Levelssecondary mill discharge, and theoperating conditions included vari-ations in feed pulp density, feedpressure, cone type, and vortexfinder clearance.

The results are presented in TableIV, which shows the experimentalconditions, gold assays of overflowand underflow, and the distributionof gold in these products. In addition,the table includes calculated valuesof the Gaudin "Selectivity Index"(S.1.)17. This is defined as "the geo-metric mean of the relative re-coveries and relative rejections oftwo minerals, metals, or groups ofminerals or metals"

i.e., S.l. = j RaJb .RbJaIn the present case,Ra = recovery of gold in underflow,Rb=recovery of gangue in under-

flow,J a=rejection of gangue to over-

flow,J b=rejection of gold to overflow,

all values being percentages.The index is used here as a con-

venient way of comparing the resultsof tests under various conditions.

In Fig. 11, the value of S.L isplotted against inlet pressure for thethree different cone types CWC 2unit. The curves all pass through amaximum at about 0,6 kgfcm2, andthe value of the S.L increases withcone type in the order M

in any of the tests-is mainly dueto the very low underflow mass of3, () per cent.

In Fig. 12, similar data have beenplotted for the ewe 4 (100 mm)cyclone, with the extra variable"vortex finder clearance" included(see Fig. 10). It appears that the

pressures employed were not highenough to show the maximum ob-served in tests on the ewe 2Mcyclone, and which presumablyexists for the larger unit. However,the L-type cone again gives highervalues of the S.l. than does the Mtype. This is true regardless of the

"I

TYPr: s

TY,. L

TYPE M

Fig. 9- The types of cone used in the Compound Water Cyclone

374 JUNE 1973

inlet pressure. An increase in vortex-finder clearance also increases thevalue of S.l.

Fig. 13 shows that, in the ewe2M (50 mm) cyclone, the value ofS.l. decreases as the vortex-finderclearance is increased, at pressuresabove 0,5 kgfcm2. Below this pres-sure, the value of S.l. at first in-creases and then decreases as thevortex-finder clearance is decreased.

Screen analyses and gold assaysof the fractions were carried out onthe products from selected tests.The results are shown in Table V,together with the gold recovery ineach size range. It is apparent thatrecovery is lowest in the finest(minus 44 /Lm) fraction, and, sincethe major proportion of the goldexists in this size range, it is thisrecovery that determines the overallgold recovery to the spigot product.

DiscussionAlthough the Selectivity Index

is a useful criterion for comparisonof the concentrating efficiencies ofvarious machines and experimentalprocedures, it is necessary to lookat the results of the present testsrather more critically, bearing inmind the potential applications toexisting milling and concentratingcircuits. A high S.I. value can resultfrom a low mass of underflow athigh grade and low recovery, or,alternatively, from high recovery in alarge underflow mass at low grade.Neither of these conditions is idealfrom the viewpoint of the presentinvestigation, the highest values ofboth grade and recovery beingdesirable.

Examination of the results showsthat gold recovery to the underflowof 80 per cent is easily achieved witheither cyclone and under a variety ofoperating conditions. If this is takenas the minimum acceptable re-covery, the "best" result is thatwhich yields this recovery in thelowest underflow mass. In three testswith the ewe 2M unit, this recoverywas achieved in underflow masses of37 per cent. With the ewe 4 unit,recoveries in excess of 80 per centwere obtained in underflow massesof 45 per cent. The L- type cone wasused in the tests on the ewe 2 unit,and the M type in the tests on theewe 4 unit. However, a test on the


Test Cyclone Coneno. diam. type

---1 50 M2 50 M3 50 M4 50 S5 50 S6 50 S7 50 L8 50 L9 50 L

10 100 M11 100 M12 100 M13 100 L14 100 L

Vortex FEfinder

clearance Pressure Solidsmm kg/cm" %

13 0,60 2513 0,35 2538 0,15 2513 0,40 2013 0,20 2038 0,25 3013 0,05 2013 0,60 3038 0,20 3035 1,30 1255 1,30 1290 1,30 1235 1,05 1255 1,05 12

UNDERFLOW

Au SoJids Solids Mass Cone. Re-Calc. % g/I pulp % Ratio covery S.l.

g/t Au %

108,1 62,0 1017 17,4 3,39 59,0 2,6108,6 60,4 975 23,4 2,68 62,8 2,3

81,3 44,6 620 54,8 1,63 83,1 2,099,2 54,5 831 3,6 10,20 36,7 4,098,1 44,7 622 17,4 2,95 51,3 2,2

133,5 63,1 1048 25,5 2,53 64,6 2,371,1 26,9 324 57,1 1,37 78,4 1,690,9 62,1 1018 36,6 2,26 82,6 2,9

133,0 45,8 644 57,0 1,51 86,3 2,290,5 43,5 599 26,2 1,95 51,1 1,787,3 61,9 1015 37,5 2,05 76,4 2,498,8 64,9 1098 45,7 1,85 85,3 2,691,1 61,7 1011 27,4 2,31 63,3 2,1

105,6 61,7 1011 33,5 2,33 78,0 2,6

VORTEX FINDER-

l

-

CLEARANCE

I

I

I

I

I

~I-et.--

~Iet.

~I1-.-~I-

~.

II

Ii ,u.,][III

ewe 4 unit with the L-type conegave a recovery of 78 per cent in anunderflow mass of 34 per cent.

The recoveries achieved appear tobe higher than those obtained in thetertiary cyclones in Anglo Americanplants. However, data on the latterare rather scanty, lacking tonnagemeasurements in particular. In ad-dition, the test cyclones are smallunits, operated under ideal con-ditions at low feed pulp densities.Further laboratory tests that couldbe carried out on these units includeinvestigation of the effect of feedpulp density and the determinationoftheir dsopointIS for gold, which wasprecluded in the present investi-gation because of lack of size databelow 44 /Lm. Furthermore, thesecyclones are at a disadvantage undertypical gold-plant conditions, be-cause the small size dictates anunrealistic level of operator atten-tion. Therefore, despite the in-complete nature of the laboratoryinvestigation, it is felt that furthertests should be carried out on ewe8 (200 mm) and/or ewe 12 (300 mm)cyclones under plant or pilot-plantconditions. The gravity-concen-tration pilot plant at present beingcommissioned provides ideal con-ditions for these tests.

A blunt-cone cyclone is alreadyoperating at Western Holdings Limi-ted with apparent success. The unitwas designed by the Anglo AmericanResearch Laboratories and is shownin Fig. 14. It has a cone included

wCl:0U

Cl:

:cc

LEGEND

.fJ0

GANGUE

LIGHT MIDDLlNGS

.HEAVY MIDDLINGS

C HEAVY CONCENTRATE

Fig. ID-Separating process in a Compound Water Cyclone (after Visman)

TABLE IV.SELECTED RESULTS FROM CONCENTRATION TESTWORK USING COMPOUND WATER CYCLONES

underflow gradeConcentration ratio =

calc. feed grade

ED

Solidsg/I pulp

297297297230230370230370370130130130130130

JOURNAL OF THE SOUTH AFRICAN INSTITUTE OF MINING AND METALLURGY JUNE 1973 375

4

3

)(

CD

"C

>--:~ 2-"..!CD

VI

CWC2M

13mm VF clearance

.0

cone type M

"

@ "

Pressure

L

s

0.5

Fig. I I-Selectivity Index versus inlet pressure for the Compound Water Cyclone50 mm unit

kg /cm2

angle of 120 degrees and is fed withtertiary cyclone underflow, thus re-placing a Johnson drum concentrator.The inlet pressure is of the order of0,2 kgfcm2. Recoveries of 70 per centin some 30 per cent by mass arecurrently being obtained.

Close liaison is being kept with afirm in Germany that is experi-menting in the field of cyclones usedas concentrators. This firm makescyclones from cast polyurethane,which has definite maintenance ad-vantages. They are currently build-ing concentrating cyclones withwide-angle cones for test purposes

376 JUNE 1973

and achieving similar results to ourown.

CONCLUSIONS

Any improvement in the overallrecovery of gold by gravity con-centration and subsequent amal-gamation relies on the efficiency ofeach concentration stage in theremoval of gold from the pulp.Since no machine can be 100 percent efficient, one must aim for thehighest unit efficiency in terms ofthe presented pulp, as well as thehighest unit efficiency in terms of the

1.0

plant as a whole. On those plantshaving both gravity concentrationand cyanide circuits, the gravity-concentration circuit must comple-ment the cyanide circuit in that thegold entering the cyanide circuitmust be of such a nature as torespond with maximum efficiencyto cyanidation. Feather and Koen19point out that the majority of thegold in the residues is free but iscoated with iron oxides and hasgangue minerals pressed into thesurface. These grains are thosewhich have bypassed the gravity-concentration circuit for one reasonor another, presumably "floating"into the secondary-cyclone over-flows after being flattened by re-peated passes through the mill.

The essence, then, of efficientgravity concentration lies in selec-tive classification of the gold to thegravity-concentration circuit, accom-panied by a reduction of the re-circulating load of gold in the mills.It is felt that blunt-cone cycloneswill prove to be of considerableefficacy here, provided dilution ofthe mill circuit is not experienced.At the same time, considerablyimproved recovery of the gold by theconcentrators is called for. This canbe achieved in two ways: firstly, anincrease in the concentrator capacityto relieve the overloaded conditionsthat exist (this is currently beingdone on most Anglo Americanplants); and, secondly, the install-ation of novel machines that areinherently more efficient than theexisting concentrators, which arepossibly obsolete in terms of modernpractice. The work currently beingundertaken, coupled with the vastlysuperior analytical tools availabletoday (such as the electron micro-probe and high-speed computers),should elucidate the problems in-volved and indicate the most prac-tical method of attaining optimumand efficient operations. It is hopedthat the two pilot plants will bevery useful in optimizing the existingconcentrators and improving theconcentrator capacity. This appliesespecially to the Reichert-cone con-centrator plant, which uses theprinciple of the pinched flume with-out introducing wall effects andovercomes the problem of choked


CWC4cone type VF clearance

mm0 M

cD

Ml

(i lX M

Overflow Underflow AuTest Size recoveryno. fraction Au Au .to l!/f

JLm Mass Au distrn. Mass Au distrn. III sIze% g/t % % g/t % %

-- --297 3,7 120,60 3,5

2,9 92210 22,64 0,5 4,1 67,60 2,1149 10,4 7,84 0,6 6,5 52,64 2,7 82105 15,6 9,52 1,2 4,4 98,92 3,4 74

6 74 12,0 11,88 1,1 2,9 178,20 4,0 7853 12,0 18,06 1,7 1,9 325,56 4,8 7444 5,6 41,52 1,8 0,9 834,80 5,8 76

-44 16,0 254,86 31,6 1,1 4 138,60 35,2 53

Original 74,5 66,95 38,5 25,5 316,36 61,5

105 3,5 14,05 0,6 8,6 29,3 2,9 8374 9,8 5,85 0,7 8,3 32,30 3,0 81

13 53 14,4 6,30 1,1 5,8 64,80 4,4 8044 8,3 11,30 1,1 2,6 186,10 5,7 84

-44 36,6 78,30 33,5 2,1 1 914,50 47,0 58

Original 72,6 42,47 37,0 27,4 197,57 63,0

riffles by removing the concentratesthrough a slot.

The ultimate result of this workwill be to reduce the inventory ofgold held in the mill circuit and toreduce cyanide losses by lessening

the probability that gold will becomecoated while in the mill circuit.

ACKNOWLEDGEMENTS

We much appreciate the invalu-able assistance and co-operation of

"

3

le..".!:~:!¥J

2

~---0-

0.5 1.0Prenure kg/cm2

Fig. 12-Selectlvlty Index versus inlet pressure for the Compound Water Cyclone100 mm unit

TABLE VSCREEN ANALYSES AND GOLD DISTRIBUTION OF PRODUCTS FROM SELECTED TESTS


the various mines whose operationswe investigated, and of those com-parties who lent us equipment anddid testwork on our behalf.

References

0

1. ADAMSON, R. J., ed. Gold metallurgyin South Africa. Johannesburg, Cham-ber of Mines of South Africa, 1972.

2. WARTENWEILER, F. Development ofmilling and cyanidation on theWitwatersrand. Am. Inst. Min. Met.Engrs Trans, vol. 112 Milling Methods.1935. pp. 760-784.

3. DoRR, J. V. N., and BOSQUI, F. L.Cyanidation and concentration ofgold and silver ores, 2nd edition.New York, McGraw Hill, 1950.

4. ROSE, T. K., and NEwMAN, W. A. C.The metallurgy of gold. 7th edition.London, Griffin, 1937.

5. DoUGLAs, J. K. E., and MOIR, A. T.A review of South African gold re-covery practice. Trans. 7th Common-wealth Min. and Met. Congress, 1961.Johannesburg, South African In-stitute of Mining and Metallurgy.Vol. Ill. pp. 171-1003.

6. JOHNSON, E. H. Concentration andselective regrinding, J.Chem Met.and Min. Soc. S. Afr., Apr. 1927.pp. 215-220.

7. Plant report from Western DeepLevels Limited, 1972.

8. WEYHER, L., and LOVELL, H. L. Theresponse of parameter variation inthe hydrocyclone processing of finecoa!. AIME Trans., vo!. 235. Dec.1966. pp. 333-340.

9. FONTEIN, F. J. Separation by cycloneaccording to specific gravity. Cy-clones in industry, K. Rietenia andC. G. Verver, eds. Amsterdam,Elsevier Publishing Company, 1961.pp. 118-134.

10. VISMAN, J. The cleaning of highlyfriable coals by water cyclones.Fourth International Coal PreparationCongress, Harrogate, 1962. Paper C2,pp. 155-162.

11. WEYHER, L. H. E., and LOVELL, H. L.Hydrocyclone washing of fine coal.AIME Trans, vol. 244. 1969. pp.191-203.

12. FALcoNER, R. A., and LOVELL, H. L.The response of varying hydrocyclonecone angles in fine coal cleaning.AIME Trans, Dec. 1967. pp. 346-354.

13. SHEAHAN,P. M. Hydrocyclones. Fed-eration of Malaya, Department ofMines Research Division, BulletinNo. 6. 1961.

14. SHEAHAN, P. M. A proposed dualcyclone system for Malayan dredges.Min. .T., Lond., Feb. 10, 1961. pp.146-147.

15. LOPATIN, V. R., and DYESHCHITS,V. S. Gravity beneficiation of gold-containing conglomerates in short-cone hydrocyclones. Tr. Tsent. Nauch.Issled. Gornorazved. Inst. Tsvet. Redk.Blagorod. Metall., no. 97. 1971.pp. 97-103. (Translation availablefrom AARL).

16. VISMAN, J. Bulk processing of finematerials by Compound Water Cy-clones. Can. Min. Met. Bull. Mar.1966. pp. 333-346.

JUNE 1973 377

CWC2M

1 cone type M

V F clearance mm

0 6-4. 13X 38 A

17, GAUDIN, A. M. Selectivity Index: Ayardstick of the segregation accom-plished by concentrating ope"ationsTrans. Am. Inst. Min. Met. Engrs,vol. 87. 1930. pp. 483-487.

18. DAHLSTROM, D. A. Fundamentalsand applications of the liquid cyclone.Chem. Eng. Progr., Symp. SeriesNo. 15, vol. 50. 1954. pp. 41-61.

19. FEATHER, C. E., and KOEN, G. M.The significance of the mineralogicaland surface characteristics of goldgrains in the recovery process. .T.S.Afr. Inst. Min. Metall., vol. 73,no. 7 Feb. 1973. pp. 223-234.

A

DISCUSSION

1. S. K. DE KOK*, B.Sc. (Eng.)Rand (Fellow)

I wish to comment on the subjectof the recovery of gold beforecyanidation, the beneficial effect ofwhich, to my mind, has emerged asthe most significant single findingof the Anglo American Researchteam. Their conclusion supports the

*Assistant Consulting Metallurgist, Anglo-Transvaal Consolidated Investment Com-pany Limited

4.0

3.0

)(

."c

>-..'>......'"

2.0

0.5'

Pre..ure kg/cm2

Fig. 13-Compound Water Cyclone 50 mm unit-effect of variation in the vortex-finder clearance

r-I

AIIL..-

cone apex angle1200

r ,I ~I I

@ section A Ascale 1:100

Fig. I4-Blunt-cone cyclone

long-held convictions of the Anglo-vaal Group, which have found ex-pression in the inclusion of gravityconcentration within the grindingcircuits and in the employment offlotation, either in the grindingcircuit or in the milled-pulp streamon its way to the cyanide plant.This contribution is made in thehope that our experience in the fieldof concentration may be of somevalue to the Anglo American in-vestigators in their further efforts.

On reading between the lines ofthe paper by Messrs Bath, Duncan,and Rudolph and also those of therelated papers by other members ofthe team, I get the impression thatthey are possibly on the verge of

1.0 repeating the error of omission thatwas made some twenty years agowhen we failed to realize the fullpotential of the plane table as a


concentrator of refractory, as wellas of free, gold. To substantiate thisview, I should first like to elaborateon the application of this device ona few mines within our group andthen attempt to draw a parallelwith the authors' present work.

The plane table, functioningessentially on the flowing-film prin-ciple, is an exceptionally high-capacity roughing concentrator,which is a decided advantage inview of the trend towards largegrinding units and high circulatingloads. At the usual pulp dilution, ithas a throughput of some 1500 ton-nes of solids per day per metre width,which means that it does not lenditself easily to laboratory or evenpilot-plant investigation, althoughI understand that NIM has hadsome success in this regard. It isextremely simple to operate, butrequires careful initial adjustmentsto get it functioning; once working,it never stops. It was invented atRand Leases in 1949 by D. McLeanfor the sole purpose of replacingcorduroy strakes that were installedin the secondary grinding circuits1.Now the corduroy cloth is probablythe most selective concentrator offree-milling gold that has ever beenused because it produces a smallbulk of very high-grade concentratethat is easily dressed on a shakingtable before being charged into anamalgam barrel. The plane table,on the other hand, produces a verymuch greater bulk of lower-gradepyritic concentrate, which, to a fargreater extent requires reconcen-tration so that the quantity is re-duced to that which can be con-veniently handled in the barrel. Butthe removal of total gold from thegrinding circuit is far higher thanthat achieved by corduroys. AtRand Leases, the plane tables re-covered some 90 per cent of the goldbeing fed to the plant, which wasdouble that previously recovered inamalgam form by the cloth. How-ever, because of the constraint im-posed on the system by the smallcapacity of the amalgam barrelsand because not all the gold re-covered by this primary or "rough-ing" device was amalgamable, some50 per cent of the value removed wasreturned to the mill in the middlingsand tailings products from the

shaking tables and in the barreltailings; the net recovery of gold inamalgam form was virtually un-changed from that achieved by thecorduroy cloths. Of the gold con-tained in the shaking-table middlingssome 30 per cent proved to be amal-gamable. In other words, it wasshown that the plane tables weremore effective than the corduroy inrecovering fine free gold, and thatthe shaking tables were not aseffective as the plane tables inretaining the flaky grains. The re-mainder of the gold returned to thecircuit was either encased in sul-phides or tarnished by iron oxidestains. The circuit finally adoptedwas a stratagem devised to reducethe load on the shaking tables andto subject the concentrate, grading36 per cent minus 200 mesh, to aseparate fine grind. The concentratesfrom ten tables installed in thesecondary grinding circuits were fedto a pebble mill-2 m by 6 m (6 ft6 in by 20 ft)-and re concentratedon a single plane table treating themill effluent. The concentrate fromthis table, approximately 15t/d, was delivered to the shakingtable and thence to the amalgambarrel, while the classifier overflow,amounting to 3,6 per cent of thetotal plant feed and grading 96per cent minus 200 mesh, joined thesecondary-mill products for deliveryto cyanidation. Following the re-duction in feed tonnage to theshaking table and the release of moregold by the finer grind, total recoveryby amalgamation increased by 5per cent to 50 per cent. The neteffect of the circuit was that 3,6per cent by mass of the ore contain-ing 40 per cent of the total gold,including the major portion of therefractory gold, was allowed toproceed to the cyanide plant, to-gether with the remaining 96,4 percent of the ore containing only 10per cent of the gold, without anypreliminary treatment other thanfiner grinding. The mistake I havereferred to lay in not subjectingthis small tonnage of high-gradematerial to separate treatment suchas more-intensive cyanidation orpretreatment with acid. It is notirrelevant to note that, in thetreatment of the far more refractoryores of the three mines of the


Eastern Transvaal ConsolidatedGroup, the plane table concentratesafter regrinding and removal of thefree gold and siliceous gangue, areadded to the flotation concentratesfor roasting and further grinding(in cyanide solution) prior to cyanid-ation.

Subsequent experience at Harte-beestfontein has confirmed the bene-fit of a preliminary acid treatment.On this plant, plane tables wereinitially installed in the secondary-grinding circuits, as at Rand Leases,but in this case only 60 to 70 percent of the gold in the mill feed wasremoved, the lower recovery almostcertainly being due to the assoc-iation of much of the gold with thelight mineral, thucholite, and to theextreme fineness of the free gold.After re-dressing of the plane-tableconcentrates on an endless-belt con-centrator and a shaking table, thenet gold recovery by amalgamationwas reduced to 11 per cent. Con-centration was consequently aban-doned. The cyanide residue valuewas far higher than desirable, evenat an extremely fine grind, but itwas subsequently reduced by 50per cent on the introduction of thereverse-leach process of uranium-gold recovery, which subjected thetotal pulp to acid treatment beforecyanidation2. It is, of course, notclaimed that the reverse.leach pro-cedure is the economic solution tothe problem in all cases, but it doespoint the way to the more effectivetreatment of concentrates containinggold that either is filmed withcoatings, for which expensive scrubb-ing or prolonged cyanidation maybe the only alternative answer, or isencased in acid-soluble minerals,which would not otherwise permitcyanide attack. After all, uraninitewas first detected in our ores froman examination of corduroy con-centrates.

I turn now to the authors' work onthe more effective application of con-centration in an attempt to improveoverall recovery. The Anglo Amer-ican plants employ the Johnsonconcentrator as the primary device;this is a fairly low-capacity, mech-anized form of the corduroy cloth,although it produces a greater bulkof concentrate. For cleaning of theJohnson concentrate, use is made of

JUNE 1973 379

endless-belt concentrators, in whichthe belt movement is against thedirection of pulp flow. For thispurpose, I think consideration shouldbe given to the use of the Gallaghertable, which is extensively used inRhodesia but is hardly known inthis country; in this device, thebelt moves transversely at rightangles to the pulp flow. I fullyagree with the authors' statementthat early removal of the gold ismost desirable to avoid the deleter-ious effects of overgrinding, butthe Anglo American grinding andconcentration circuits as at presentconstituted are not conducive toefficiency in this respect. Concen-tration is confined to the tertiarystage of classification/milling, atwhich stage much of the liberatedbut maltreated gold has alreadyescaped to the cyanide circuit viathe secondary cyclones. For an in-creased concentration recovery, re-liance must therefore be placed onthe concentrating effect of the cy-clones to deliver as much of the goldas possible to the Johnson con-centrators. As a consequence, theauthors are having to investigatewider-angled cyclones, which tendto suppress the classifying andaccentuate the concentrating effectsof settling. Earlier removal of thegold would seem to be the logicalmethod, but this would require theconcentrators to be placed within thecirculating loads ofthe primary- andsecondary-milling circuits. The lowcapacity of the standard-sized John-sons is not a factor in their favourfor this purpose, and the authors areinvestigating devices such as theblunt-cone cyclone and the Reichertcone, which is but a version of theplane table without its side-walleffects. Experiments have beencarried out at the Anglovaal Lab-oratory with laboratory-sized vesselslined with riffled rubber. These aresomewhat similar to the Johnsonconcentrator but are conical in shapeand run at super critical speeds;the centrifugal forces developed give,in open circuit, 90 per cent recoveryof the gold at high ratios of concen-tration, but, as with all centrifugalmachines of this type, continuousdischarge of the concentrate presentsdifficulties. The development of a con-centrator utilizing centrifugal forces

380 JUNE 1973

for the recovery of the finer sizes ofthe heavy minerals is, I feel, one ofthe fields of research in which our in-dustry should be actively engaged.

Arithmetical averages of the fig-ures presented by the authors, whichare sufficiently accurate for illus-tration purposes, indicate that theendless- belt concentrators recover70 to 75 per cent of the gold de-livered to them by the Johnsolls,and that the amalgamation of theconcentrate so produced results in anet recovery of 35 per cent of thegold fed to the plant. If, as stated,the Johnson concentrators recoversome 70 per cent of the total gold,it is obvious that the endless-belttailings contain some 15 per centof the gold and, in the circuit as atpresent composed, this is merely re-turned to the grinding circuit forfurther defacement and delivery tothe cyanide circuit; at least, atRand Leases this material was givena separate finer grind. After em-phasizing the importance of a highrecovery of gold before cyanidation,the authors have quite rightlyapplied themselves to improving theprimary recovery, but so far haveneglected to investigate the correcttreatment of the gold that is lost inthe process of reducing the bulk tosuit the amalgamation process. Theyadmit that doubling the speed ofthe endless belts increased totalgold recovery by amalgamation by2,5 per cent but at the expense ofincreasing concentrate mass by 50per cent. I submit that there is a fargreater potential pay-off in investi-gating the non-amalgamable gold,which the primary concentratorshave handed to them on a plate,than in an improvement in theamalgamation process itself. Indeed,the complete elimination of amal-gamation, besides removing the re-strictions imposed by the limitationsof the plant, would obviate the veryreal danger of mercury escaping withthe pulp and inhibiting cyanidation.

This prompts me to ask whetherthe improvement in total recoveryfollowing an improvement in amal-gamation is not in part due to areduction in mercury loss. Alter-native methods of treatment of thetotal primary concentrate could bedevised that would provide for therecovery of the osmiridium and free

gold, possibly in the form of a"clean-cut" for direct smelting or byflotation. The reduction in cyanideresidue value resulting from im-proved amalgamation, noted by theauthors, is not, in my opinion, dueto an increase in amalgam recoveryper se, but rather to a reduction inthe quantity ofrefractory gold in thetailings from the amalgam barrelthat proceeds to cyanidation. Thereis undoubtedly far more of thismaterial present in the endless-belttailings than was ever present in thebarrel tailings before changes weremade in the techniques of amalgam-ation. Contrary to the authors'assertion that recovery by amal-gamation is inherently more efficientthan that by cyanidation, I believethat amalgamable gold is the goldthat is also most easily cyanided. Ifspecial treatment is required torender the gold amalgamable, thatsame treatment will render it moreamenable to cyanide. It is not im-portant that the gold recovery asamalgam be 30 per cent or 50 percent or any other figure; what isimportant is that, if concentrationhas been considered necessary torelieve the cyanide plant, advantagebe taken of the opportunity tocorrectly prepare for cyanidation thegold in the primary concentrate thathas not reached the barrel, as wellas the gold that amalgamation hasfailed to recover. In failing to makefull use of this opportunity, theauthors are perpetuating the myththat concentration is synonomouswith amalgamation, just as we diduntil we saw the light. It is notsurprising that, in the days whenconcentration was regarded merelyas a means to amalgamation, theproponents of all-cyanidation werenot convinced of the value of con-centration.

The advantages of separate treat-ment of the high-grade, more-re-fractory portion of the ore arehighlighted at three of our mineson which flotation plants were in-stalled for the recovery of pyriterequired for acid manufacture forthe uranium industry3. Anglovaalbroke with the traditional practiceof recovering this pyrite from theuranium-plant tailings by inter-posing the flotation plants betweengrinding and cyanidation. At one


of the Hartebeestfontein plants, theflotation cells are installed withinthe grinding circuits, whereas, atLoraine and the other Hartebeest-fontein plant (formerly Zandpan),flotation is carried out on the finalmill pulp. Gravity concentration inthe grinding circuit has been re-tained at Loraine, both for osmirid-ium recovery and because of ourexperience, verified on the EasternTransvaal Consolidated Group ofmines, that flotation is least success-ful in recovering the very finest andthe very coarsest fractions of thegold particles. On all the mines, therecovery of gold into the flotationconcentrates exceeds 80 per cent. Itis of interest to note that well over50 per cent of the total gold can befloated with the addition of frotheronly and without the addition ofxanthate or any other pyrite pro-moter, the main dilutent in thehigh-grade concentrate being finesiliceous gangue. The separate re-grinding and intensive cyanidationof the pyrite concentrates result in99 per cent recovery of the containedgold. The net results have beeneither increased overall recoveries orunchanged recoveries with increasedtonnage throughputs, the naturalconsequence of decreased depend-ence on fine overall grinding. AtHartebeestfontein, the combinationof reverse-leach treatment of flo-tation tailings, reground concen-trates, and calcines has resulted inan increase in recovery of some 2tper cent at a grind coarser by 18per cent minus 200 mesh.

In conclusion, I wish to congratu-late the authors on their stimulatingpaper. Of their overall philosophyand approach there can be nocriticism. They are tackling theproblem in a methodical sophisti-cated manner; I venture to predictthat the evidence they are still togather from their pilot-plant willsoon result in a revision of theirpriorities.

References

1. ZADKIN, T. The Rand leases planetable. .f. Chem. metall. min. Soc. S.Afr.,vol. 54. Feb. 1954. pp. 292-298.

2. BRITTEN, H., and DE KOK, S. K.Uranium in South Africa, vol. 1, p. 462.

3. BUSHELL, L. A. The flotation plantof the Anglovaal Group. J. S.Afr. Inst.Min. Metall., Jan. 1970.

2. J. LEVI N*, B.Sc. (Eng) Rand(Member)

The National Institute for Metal-lurgy is engaged on tests on theconcentration of uranium, and, be-cause gold and uranium are closelyassociated in the ores of the Wit-watersrand, some of the observationsthat have been made have a bearingon the problems of gold extraction.Currently, an ore from the Vaal Reefis being examined, and the infor-mation obtained so far points to thefollowing tentative conclusions, whichare based on the experimental resultsshown in the accompanying tables.(1) Gravity concentration for the

recovery of gold and uraniumcan be carried out advantage-ously in an open circuit, insteadof in the usual closed grinding-classification circuit.

(2) The proportion of concentratemade by gravity or flotationconcentration should be fairlyhigh-between 5 and 10 percent of the ore-so that themaximum amount of gold anduranium can be recovered.

(3) Extraction of gold by cyanid-ation is increased significantlywhen cyanidation follows acid-.leaching for the dissolution ofuranium; a significant increaseis observed even when thetailings are subjected to prioracid leaching.

(4) Satisfactory extraction of goldfrom gravity concentrates canbe obtained by cyanidation afterfine grinding and acid leaching,without resort to amalgamation.(Amalgamation would not beapplicable to the relatively largeproportion of concentrate beingrecommended. )

(5) Flotation effects a higher re-covery of gold than doesgravity concentration.

(6) When gravity concentration isused, the tailing is the largestsource of gold not extracted bycyanidation. When flotation isused, the first xanthate con-centrate is the largest source ofgold not extracted by cyani-dation.

Gravity concentrationWhen concentration in a closed

grinding-classification circuit is prac-

*Head, Ore-dressing Division, NationalInstitute for Metallurgy


tised, the quantity of material fedto the concentrate is generallyseveral times that of the new feedentering the circuit, and the pos-sibility of applying concentrationin an open circuit is worth con-sidering. To ascertain if recoveriesfrom an open circuit would besubstantially lower than from aclosed circuit, laboratory tablingtests were done on material groundto various extents, and the resultswere compared with those obtainedfrom a test in which closed-circuitgrinding and concentration weresimulated. Table I summarizes theresults obtained. The figures actuallysuggest that, if the optimum degreeof grinding for open-circuit con-centration is used, the recoveries inopen circuit will be higher than froma closed circuit, but it is preferredmerely to draw the conclusion thatthe recoveries of gold and uraniumfrom an open circuit should not bemarkedly lower than from a closedcircuit. Table I also shows thatsubstantially increased recoveries areeffected when the mass of concen-trate is increased from 5 to 10 percent of the feed.

Gold extraction from products ofgravity concentration

A batch of coarsely ground orewas concentrated by use of a Hum-phreys spiral and a shaking tableto provide three concentrates and atailing. The four products obtainedwere ground to various degrees, acidleached, and cyanided. Fig. 1 showshow the concentration was effectedand the results obtained. Table IIshows the results of cyanidation ofthe individual products of con-centration. The very high extractionsof gold obtained from the first con-centrate are noteworthy, and itappears that this concentrate doesnot include gold in a form that needsintensive treatment for its extractionby cyanidation. However, all theother products need to be finelyground and acid leached to enablesatisfactory extractions of the goldto be achieved. The high gold con-tent of the tailings should be noted.

Concentration by flotationFig. 2 illustrates the procedure

used in concentrating the ore byflotation, and Table III gives theresults of the pretreatment andcyanidation tests done on the flo-

JUNE 1973 381

Distbn. of gold, % of gold in:Gold in cyanide

residue g/t Prod. of conc. Original oreConcentra- Grindtion Product fLm Dir.* Rev.t Dir.* Rev.t Dir.* Rev.t

1st conc. -150fLm 7,0 3,1 0,30 0,20 0,09 0,06-100fLm 4,5 4,0 0,21 0,14 0,06 0,06

- 53 fLm 10,0 4,9 0,54 0,20 0,16 0,06

2nd conc. -150 fLm 8,4 3,6 9,0 3,8 2,49 1,05-100fLm 5,0 2,1 5,4 2,3 1,49 0,64

- 53 fLm 3,7 2,4 3,5 2,5 0,98 0,69

3rd conc. -150 fLm 10,5 3,9 9,1 3,4 I,ll 0,42-100 fLm 7,5 2,6 6,5 2,2 0,79 0,27

- 53 fLm 5,2 2,7 4,2 2,4 0,51 0,29

Tailing 53%-75fLm 0,84 0,52 13,5 8,4 4,04 2,5065%-75fLm 0,74 0,37 11,9 6,0 3,56 1,7876%-75fLm 0,47 0,25 7,6 4,0 2,26 1,2087%-75fLm 0,39 0,26 6,3 4,2 1,87 1,25

Gold in cyanideresidue g/t

GrindDir.* Rev.t

-150 fLm 17,8 4,7-100 fLm 13,7 2,1

- 53 fLm 11,1 3,1

-150 fLm 15,7 10,8-100 fLm 10,3 6,9

- 53 fLm

-150 fLm 11,4 8,5-100 fLm 7,0 4,4

- 53 fLm

57%-75fLm 0,40 0,2767%-75fLm 0,30 0,2275%-75fLm 0,31 0,1785%-75fLm 0,24 0,10

Distbn. of gold, %of gold in:

Prod. of flotn. Original ore

Dir.* Rev.t Dir.* Rev.t-- ~-

1,20 0,30 0,69 0,170,90 0,14 0,52 0,080,72 0,20 0,42 0,11

~-10,70 7,40 3,30 2,78

7,10 4,70 2,19 1,45

~-26,8 20,0 1,34 1,0016,5 10,4 0,82 0,52

29,0 19,7 1,91 1,3021,4 16,2 1,42 1,0522,6 12,2 1,49 0,8017,39 7,0 1,15 0,46

tation products. It can be seen that57,6 per cent of the gold was re-covered by flotation with only afrother. Cyanidation of the frotherconcentrate was very effective, al-though acid leaching is essential ifmaximum extraction is to beachieved, and acid leaching in thisinstance appears to be more impor-tant than fine grinding. Cyanidationof the xanthate concentrate was farless satisfactory, and it is evidentthat these concentrates contain goldin a form that requires intensive treat-ment for its extraction. The residuesfrom the flotation tailing are mar-kedly lower than those obtainedfrom gravity concentration, buteven it benefits from acid leachingprior to cyanidation, the gold assayvalue being reduced by approxi-mately 0,14 g/t for all degrees ofgrinding tested.

Mineralogical examinationsApproximately half the gold ob-

served in the first two gravity con-centrates was in the form of freegrains, the remainder being assoc-iated in roughly equal amounts withsulphide or uraninite. In the thirdgravity concentrate, free gold is lessabundant, and gold associated withuraninite is more abundant; thisconcentrate also contains a markedproportion of grains in which thu-cholite is attached to uraninite.Although most of the gold grainsappeared to be discoloured or stainedto various extents, the considerableproportion of grains in which goldis associated with uraninite suggeststhat much of the benefit obtainedfrom acid leaching prior to cyani-dation can be ascribed to theassociation of gold and uranium.However, there is no definite evi-dence of the magnitude of thebenefit that might have resultedfrom the removal of coatings fromthe gold grains by the acid leachingstep.

The frother concentrate obtainedby flotation contained numerousfree grains of gold and numerousgrains in which the gold was asso-ciated with pyrite or uranium min-erals; the latter were predominantlythucholite, but there were alsonumerous grains in which uraninitewas associated with the thucholite,and some grains that appeared tobe wholly uraninite.

382 JUNE 1973

TABLE tCOMPARISON OF GRAVITY CONCENTRATION IN OPEN AND CLOSED CIRCUITS

Gold recovery, %at

Circuitsimulated

Grind, %-75fLm 10% mass of conc.5 % mass of conc.

OpenOpenOpenClosed

405061

3 stages;1st stage

40%-75fLm

43,051,048,245,5

63,862,065,458,0

TABLE 11CYANIDATION OF PRODUCTS OF GRAVITY CONCENTRATION

(See Fig. 1).

*Direct cyanidationtCyanidation after acid leaching

TABLE IIICYANIDATION OF PRODUCTS OF FLOTATION

(See Fig. 2).

Concentra-tion Product

Frother conc.

1st xanthateconc.

2nd xanthateconc.

Tailing

*Direct cyanidationtCyanidation after acid leaching


2005 92,5 116,0

30,4 27,7 12,2

4126 1258 1976

3,6 21,4 11,9

Batch grind to 52%-75/Lm

I

Float in stages

I

I I IFrother 1st xanthate 2nd xanthate

concentrate concentrate concentrate

Mass, % I I[

0,74 4,12 2,29

Gold, gjt 1520 146 42,6

Gold distribution, % 57,6 30,8 5,0

U.O., gjt 5284 1748 1945

U3O. distribution, % 11,1 20,4 12,6

xamination of the two xanthateconcentrates indicates that theycontained less free gold than the

[rother concentrate and. that moregold was associated with sulphidesthan with uraninite or thucholite.

3. E. A. D. RUBIOGE- (i=ellow)

Batch grind to 45 % - 75 /Lm

I

HumPh'!" "p"'"

I

Bulk concentrate

I

Bhok;.. tab,!,

Mass, %

I

Concentrate 1

I

0,29

I

Concentrate 2

I

5,71

I

Concentrate 3

I

2,01

Gold, gjt

Gold distribution, %

U30., gjt

U3O. distribution, %

Fig. I-Procedure and results for gravity concentration test

Fig. 2-Procedure and results for concentration by flotation

I

Tailing

T91,99

6,2

29,7

230

63,1

I

Tailing

I

I!!2,85

1,38

6,6

212

55,9


The paper reported many factorsrelating to the efficiencies attainablein amalgamation.

At one stage the paper defines theword "liberation" when applied togold particles and states that themineral grain "for gravity concen-tration . . . must be almost com-pletely separated from the gangue."In his address relating to this para-meter, Mr Rudolph remarked thathydrated iron oxides often coveredthe surfaces of the gold particles andprevented or delayed the process ofamalgamation.

It is on this point that I wish toreport on some experiments carriedout in the Rand Leases AssayDepartment in 1954. They dealtwith concentrates from the VirginiaMine. The metallurgist in chargefound that barrel tailings repeatedlycontained "floured" mercury, which,of course, carried much gold. Heasked that tests be carried out toindicate the cause of the troubleand, if possible, provide informationthat would lead to effective amal-gamation.

The material contained many par-ticles that were described as "rustygold", which was, of course, causedby the "hydrated iron oxide" re-ferred to in the paper under dis-cussion.

The line of investigation under-taken on that occasion was mainlydirected to conditioning of the con-centrate by chemical means beforeproceeding with the amalgamationtests.

Only five of the many significanttests will be reported here, with thepurpose of high-lighting the partplayed by the use of cyanide to cleanthe gold particles before the additionof mercury and rolling for a verybrief period without further grinding.

Weighed portions of concentrate,with water added, were dealt withas follows: (see table on page 384)

Note: In all these tests, no grind-ing balls were added. It was assumedthat the grinding that usually accom-panies amalgamation work adds tothe build-up of hydrated iron oxidefilm. This would not, of course,occur to any great extent in the

*Formerly Group Assayer, Anglo- Trans-vaal Consolidated Investment CompanyLimited

JUNE 1973 383

TestNo. Treatment given Observations

---- --- ----.Washed, CaO and KMnO. added, and rolled on lab-oratory rolls

la Washed out all CaO and KMnO. from No. 1 above,added CaO and KCN, and rolled for 30 minuteswith Hg

2 Washed three times with water, then added CaO andHg, and rolled for 15 minutes

3 Treated as in 2 above, added aaa, KCN, and Hg,and rolled for 15 minutes

5

Treated with CaO & KNOs and, without washing,added Hg and rolled for 15 minutes

Treated as in 4 above but added KCN before rollingfor 15 minutes X

4

Much "floured"mercury

98 % of gold found inamalgam

68,3 % of goldrecovered

92,6 % of gold inamalgam



short time of rolling reported in theabove tests. None-the-Iess, it isof importance to note that, withthe gold surfaces cleaned (and thus"almost completely separated") bycyanide, the amalgamation appearsto take place almost immediately.

Many other tests were carried out,and, wherever cyanide was omitted,the extraction was much poorer inthese short-time runs.

It will be seen that cyanide woulddissolve the very fine or flattenedparticles of gold (probably a desir-able feature!), and the transfer ofthis gold-bearing solution wouldhave to be attended to in anyplant designed to follow the sug-gested procedure.

The need for the further grindingof the resulting barrel tailing might,of course, be necessary if inspection

or assay proved the need for such amove, but returning the tailings tocircuit (or other further treatment)after more than 90 per cent of thecoarse gold has been removed wouldnot, I am sure, raise any anxieties.

Perhaps the authors of the paperwould find it useful to apply theirvery thorough skills to proving thevalue or:. ,otherwise of the abovefindings.

Company affiliatesThe following members have beenadmitted to the Institute as Com-pany affiliates.

AE & Cl Limited.AfroxjDowson and Dobson Limited.Amalgamated Collieries of S.A. Limi-

ted.Apex Mines Limited.Associated Manganese Mines of S.A.

Limited.Blackwood Hodge (SA) Limited.Blyvooruitzicht G.M. Co. Ltd.Boart & Hard Metal Products S.A.

Limited.Bracken Mines Limited.Buffelsfontein G.M. Co. Limited.Consolidated Murchison (Tvl) Gold-

fields & Development Co. Limited.Doornfontein G.M. Co. Limited.Durban Roodepoort Deep Limited.East Driefontein G.M. Co. Limited.East Rand Prop. Mines Limited.Free State SaaiplaasG.M. Co. Limited.Fraser & Chalmers SA (Pty) Limited.Goldfields of SA Limited.The Grootvlei (Pty) Mines Limited.Harmony Gold Mining Co. Limited.Hartebeesfontein G.M. Co. Limited.

Hewitt-Robins-Denver (Pty) Limited.Highveld Steel and Vanadium Corpo-

ration Limited.Impala Platinum Limited.James Sydney & Company (Pty)

Limited.Kinross Mines Limited.Kloof Gold Mining Co. Limited.Lennings Holdings Limited.Leslie G.M. Limited.Libanon G.M. Co. Limited.Lonrho S.A. Limited.Loraine Gold Mines Limited.Marievale Consolidated Mines Limit-

ed.Matte Smelters (Pty) Limited.Northern Lime Co. Limited.O'Okiep Copper Company Limited.Palabora Mining Co. LimitedPresident Stern G.M. Co. Limited.Pretoria Portland Cement Co. Limit-

ed.Prieska Copper Mines (Pty) Limited.Rand Mines Limited.Rooiberg Minerals Development Co.

Limited.Rustenburg Platinum Mines Limited

(Union Section).

Rustenburg Platinum Mines Limited(Rustenburg Section).

St. Helena Gold Mines Limited.Shaft Sinkers (Pty) Limited.S.A. Land Exploration Co. Limited.Stilfontein G.M. Co. Limited.The Griqualand Exploration and Fi-

nance Co. Limited.The Messina (Transvaal) Develop-

ment Co. Limited.Trans-Natal Coal Corporation Limit-

ed.

Tv!' .Cons. Land & Exploration Co.Tsumeb Corp. Limited.Union Corporation Limited.Vaal Reefs Exploration & Mining Co.

Limited.Venters post G.M. Co. Limited.Vergenoeg Mining Co. (Pty) Limited.Virginia O.F.S. Gold Mining Co.

Limited.Vlakfontein G.M. Co. Limited.Welkom Gold Mining Co. Ltd.West Driefontein G.M. Co. Limited.Western Deep Levels Limited.Western Holdings Limited.Winkelhaak Mines Limited.


Documents

Somefactorsinfluencing goldrecovery bygravity concentration · 2009. 8. 26. · African gold-recovery practice in 1961, pointing out that two-thirds of the major plants incorporated