Search Documents

By Chris Bennett, Glen Kosick

"Process modeling and equipment sizing procedures for SAG circuits has advanced significantly over the past two decades. However, no matter how exact the hardness tests, models and scale-up calculations, if the ore samples being used for design do not accurately represent the orebody then major design errors can be encountered. This paper will describe a hardness testing method for identiQing the power requirement profile of an orebody, to an extent that design errors due to poor sampling can be greatly reduced or avoided altogether. It will provide some insight into the magnitude of hardness variability that can be encountered in an orebody. Finally, the paper will preview how detailed hardness distribution data can be incorporated into mine block models, to optimize SAG circuit design for an orebody, both fi-om a capital investment and throughput point of view.INTRODUCTIONIt has long been known that many SAG/AG' circuits experience wide swings in throughput. These throughput variations are almost always of a magnitude not anticipated in the plant design; consequently, they are detrimental to the expected return on investment for the mining companies. It is clear that the throughput swings are mainly caused by changes in ore hardness. In 1993, in an attempt to address this problem, MinnovEX began investigating the possibility of using a small scale SAG test (2 kg charge) to predict SAG power requirements from drill core samples. The objectives of the small scale test were to enable mining companies to do many more SAG hardness tests than currently undertaken for a particular orebody. To make this possible, the SAG tests had to use a minimal amount of sample, be cost effective, and give acceptable accuracy in the prediction of SAG power. If this small scale SAG test proved accurate, then the high sampling frequency would yield results that approach the true hardness distribution of an orebody. This hardness distribution could then be used to derive the distribution and variability of specific energy required throughout the complete orebody. This specific energy (kWh/t) can be thought of as the power required to grind one tonne of ore per hour (kW/tph) to a specified size. Once the true distribution of power requirement throughout the whole orebody is understood, the grinding circuit design team can make an informed decision regarding the hardness value that is most appropriate for sizing the SAG equipment. Throughput losses due to hardness variation could be deliberately planned for or avoided altogether (a capital cost decision)."

Jan 1, 1999

By Ray Patterson

A planned modification to a cement kiln at the Holnam Portland Cement Plant near Florence, Colorado required the removal of a ring of cast-in-place (CIP) refractory lining. The CIP lining was approximately 14 inches wide, 12 inches thick and 38 feet in circumference.

Jan 1, 1999

By A. J. Lynch

Mineral processing models have been discussed for many years but it was not until digital computers became available about 1960 that serious attempts at circuit simulation could be carried out. Even then the work was primarily academic until PCs could be used by engineers in plants and design offices in the mid-1980s. Since then simulation has moved quickly and it is now at the point that most processes can be simulated reasonably accurately for purposes of design and optimization. There are still some significant deficiencies, notably in dry grinding and separation, and in the chemical interactions in flotation. Future work will involve: ?improvements of existing models ?dynamic simulation for automatic control and, in particular ,for training which is a weak link ?using advanced simulation models to predict liberation and separation from data acquired from SEM-based liberation analysers. Eventually mineral processes models will be part of total simulation systems which will incorporate mining and downstream processes including all environmental impacts.

Jan 1, 1999

By Gupta R. C

Syria has about 500 million ton deposit of minerals having 34 per cent iron content with alumina and silica as major impurities. Efforts were made to see the feasibility of concentration by four different methods involving washing, high intensity magnetic separation, reduction roasting and magnetic separation and acid leaching technique. It was observed that the reduction roasting with magnetic separation yields a concentrate with high iron (62.3 per cent) content without requiring water which is a scarce raw material in the region. The paper concludes that the detailed study may be made following this route to optimise the process Parameters.

Jan 1, 1999

Australia's iron ore industry is entering a challenging period with the development of new orebodies, the introduction of new products and the push for further downstream processing. In addition, the recent price reductions for iron ore exports have intensified the focus on reducing capital and operating costs. To assist in these areas, CSIRO is expanding its research capabilities, including the incorporation into CSIRO of the Mineral Processing Laboratories in Perth, which were previously managed by the WA Department of Minerals and Energy. In addition, a large part of the $24.4 million expansion of CSIRO's Pinjarra Hills laboratories in Brisbane will be devoted to enhancing iron ore processing research, providing purpose built laboratories and large process bays.

Jan 1, 1999

By Sugie Prawono

Jacked pile is the only piling system in which the load capacity of the pile, at any penetration depth, could be monitored directly during its installation process. Each installed pile could have a graph of jacking load versus depth of penetration. For safety purposes, the maximum applied jacking load during installation of a pile could be assumed as the minimum load capacity of the pile. This load capacity will increase with time since the drop in soil strength due to the installation process will recover later on. In order to determine accurately this jacking load, a research was carried out by comparing the load capacity of a pile at certain depths to the jacking load recorded during the installation process. The load capacity could be calculated using cone resistance, qc, and local friction, fs, from Dutch Cone Penetration Test. A graph could be drawn of the load capacity against the depth. By applying a correction factor, a, to the data of local friction, the calculated graph could be matched to the graph of recorded jacking load. The a factor is found to depend on the soil type, which is in this case, represented by the magnitude of cone resistance. The soil type is divided into 3 categories according to qc values (qc < 500 Kpa; 500 Kpa < qc < 2000 Kpa and qc > 2000 Kpa). Graphs of the a factor versus depth and a sample of calculation are presented in this paper for easy application.

Jan 1, 1999

Iron production in one year exceeds that of all other metals combined in ten years. It is a huge industry that started from a modest furnace producing one - two kilograms metal per day to a gigantic reactor producing 10 000 tons per day. The major milestones of these developments are reviewed.

Jan 1, 1999

By Mike Rylatt

"The BHP Diamonds Inc. and Blackwater Group (Dia Met Minerals Ltd., Charles Fipke and Stewart Blusson) joint venture began production at Canada=s first diamond mine during the last quarter of 1998.Construction of the processing plant began in mid 1997. During the 1998 winter season, the heavy process plant equipment was delivered to the Ekati site, 300 km NE of Yellowknife, by the ice road. Towards mid 1998 the construction activities were complete and commissioning began. The mine officially opened in October 1998.Initial production of 9000 tpd kimberlite will be from the Panda pipe grading 1.09 ct/t at an average value of US$13O/ct.Processing starts with a gyratory crusher reducing the ROM kimberlite to 90% passing 15Omm before further reduction to minus 75mm in a WaterFlush crusher. The crusher product discharges into a 4m diameter x 7m long primary scrubber. After screening, the minus 75 +1 mm kimberlite is conveyed to a 250 ft feed bin ahead of the 1.7 m diameter x 1.1 m wide high pressure grinding roll. The HPGR produces 50% minus lmm in a single pass, the compacted flake product is dis-agglomerated in a secondary scrubber. The scrubber product is screened into a minus 25 +lmm heavy medium separation plant feed and conveyed to a 375t HMS feed bin ahead of two 150 HMS modules. Milled 270D ferrosilicon at a slurry density of 2.6 g/cc and HMS feed are pumped to three 420mm diameter cyclones. Heavy mineral concentrates, approximately 1 % of the plant headfeed, are pneumatically transported to the Recovery plant for final diamond recovery using wet high intensity magnetic separators and high intensity (attrition) milling prior to three stages of x-ray recovery. At full production, Ekati will produce approximately 2 kg/day of diamonds.This paper details the later commissioning phase and the first three months Oof operational activities during the build up to full production at the Ekati processing plant."

Jan 1, 1999

By A C. Edwards

The Mongol-Okhotsk belt consists of detached fragments of the Circum Pacific fold belt which have been trapped within the Mongol-Okhotsk fold belt. In its western part, the structures of the belt form an east-facing loop that is not connected with the other orogenic structures of Central Asia. In the eastern part of the belt (the Mongol-Okhotsk suture zone) the terranes form a zone between the Siberian craton to the north and the Bureya composite terrane to the south, that is linked with the Circum Pacific belt. The accretionary terranes, together with their Precambrian backstops, form a series of consistently-stacked, paired units that are bounded by northwest and north trending sinistral strike-slip faults. The ages of the rocks forming these accretionary complexes vary from Late Precambrian - Carboniferous in the west to Cambrian - Jurassic in the east. The terranes developed due to a progressive westward stacking of the accretionary complexes during the process of accretionary growth. The magmatic rocks young eastward indicating that migration of the magmatic front of the arc followed the eastward growth of the accretionary complexes. The accretionary terranes may be differentiated by their Paleozoic lithological assemblages. Only the Selemdzha, Niman, and Kerbi terranes located in the eastern part of the Mongol-Okhotsk suture zone contain significant amounts of black shale, and it is these terranes that host the bulk of the hardrock and placer gold mineralisation that is present within the suture zone. Gold mineralisation is present both as vein style and disseminated mineralisation within structural duplexes.

Jan 1, 1999

By G. Biles, K. Loughran, R. Anand

The Eskay Creek Mine, a high-grade gold and silver underground mine in northwestern British Columbia started operations in late 1994. For the first three years of operation, ore was crushed to minus 2 inch; blended to satisfy smelter terms and direct shipped to smelters in Japan and Canada. Before commencing the above operations, preliminary evaluations had indicated that an extremely complex metallurgical facility, requiring a large capital investment, would be required to recover the precious and other economic metals. Subsequently markets for raw ore were identified and underground mining and ore transportation operations, at a nominal production rate of 300 tonnes per day, were commenced. Exploration diamond drilling in 1995 and 1996 discovered additional lower grade mineralised zones, which are less economic or uneconomic as direct shipping ore. Test work demonstrated that these lower grade zones could be upgraded to marketable concentrates in a simple flotation mill. The mill was built in 1997 and has been in operation since late 1997. The mining rate now has increased to 400 tonnes per day including 250 tonnes per day of the direct shipping ore and 150tonnes per day of the mill ore. Prime Resources Group Inc. wholly owns the Eskay Creek Mine. Homestake Canada Inc. owns 50.6% of Prime’s shares and manages and operates the mine on Prime’s behalf. This paper discusses the mining and milling operations as well as ore shipment and smelter contracts&apos; administration.

Jan 1, 1999

The 102nd Annual General Meeting of The South African Institute of Mining and Metallurgy was held in the Gold Room, Transvaal Automobile Club, 60 5th Street, Lower Houghton, Johannesburg on Wednesday, 18th August, 1999. Obituaries The President announced the death, during the year, of the following members: Honorary Life Fellows and Past Presidents D.D. Howat Life Fellows O. Deane, J.S. van Zijl Fellow A.A. Hejja Retired Fellows A.L. Forder, C.B. Parker, A.C. Petersen, F.W.J. Ross Members J.S. Stanko, D.H. Gray Retired Member P.K. Bredell Retired Associate I.S. van Eyssen In memory of the deceased and in sympathy with the bereaved, all rose and observed a moment of silence.

Jan 1, 1999

By Nelson R. Shaffer

Implementation of the Clean Air Act Amendments of 1990 is negatively affecting production of coal used for generating electric power in Indiana and elsewhere in the Midwest. Required reduction of sulfur compounds from gas streams, or flue gas desulfurization (FGD), as mandated by these laws has encouraged installation of wet limestone scrubbers at many generating plants. Increasingly stringent requirements that will begin in 2000 are expected to necessitate even more extensive use of such scrubbers. Indiana has large deposits of limestone-the major consumable alkaline raw material for scrubbers-that can meet an expected demand of several million tons per year. Limestones are not, however, all the same. Different limestone units behave quite differently when used as scrubbing agents under differing conditions in actual practice. Decisions about FGD scrubber details would be more meaningful if made with thorough knowledge of which rock units contain the most efficient FGD limestones and which sites of suitable stone are located nearest to power plants. Even though abundant carbonate rocks occur in Indiana, their suitability in scrubbing systems has received little attention, and few attempts have been made to match details of stone resources to the needs of various utility scrubbing systems. Optimization of the fit between geologic characteristics of the existing limestones and prospective scrubber systems should allow for greatly improved planning and operation of FGD systems with attendant lower overall costs. Electric utilities have developed computer models to predict FGD system operations (i.e., the FGD-PRISM model of the Electric Power Research Institute). These models require geologic data such as reactivity that are commonly unavailable. We generated data needed for the model during a cooperative effort among the Indiana Department of Commerce, the Indiana Geological Survey (IGS), The Indianapolis Power & Light Company (IP&L), and the Indiana Mineral Aggregates Association. Our goal was to provide missing geologic details of Indiana scrubber-stone raw materials and apply them to the FGD-PRISM model to provide general guidance to determine which geologic units can provide stone for most efficient sulfur removal. Results of the model for selected limestones and parameters specific to one IP&L generating plant were calculated. Geologic data generated may be applied to other generating operations, and our examples should help other utilities optimize their FGD strategies. Several hundred samples from more than 30 active quarries were collected and their physical properties determined. Grindability values ranged from 7.2 to 21.7 for 50 selected samples. Acid-insoluble residues of less than 1% to 13% were found. Dissolution rates as an index of reactivity varied over two orders of magnitude for the selected set of samples. These data plus chemical analyses and Ca/Mg ratios were entered into a generic scrubber model to determine rates of stone use and scrubber efficiency for various stone. Removal of SO2 was projected by this model to range from 89% to 98% with stone utilization rates from 91% to 93%. Excellent scrubber stone sources were found in Mississippian rocks in central and southern Indiana. Rocks of the Ste. Genevieve and Salem Limestones were especially efficient. Even small amounts of dolomite, clay minerals, and quartz diminished a stone&apos;s usefulness for FGD scrubbing.

Jan 1, 1999

By Ray H. S, Galgali R. K

It is well accepted that the blast furnace will continue as the main route for iron-making for many more years. However, its limitations have motivated researchers to look for alternative routes, particularly, those where non-coking coal is used for liquid iron production. Smelting reduction (SR), where coal and oxygen are used to produce liquid iron and power is now receiving serious attention. Several smelting reduction reactors are now operating in pilot plant scale and one or two have reached commercial status.

Jan 1, 1999

By Michael F. Barber

"Highway slopes are intended to remain stable over a long period of time. Unexpected slope failures can be costly in terms of human life and property damage or loss. Cleanup of failures can prove to beextremely expensive."

Jan 1, 1999

Deep level block caving as a mining method is becoming more popular due to the fact that it is the lowest cost per ton underground extraction system. However, many block caves do not achieve their full potential due to draw horizon instability. In this paper five design rules are given to overcome this problem. The design rules described are applicable for slusher and trackless block caves where the operational depths exceed 500 metres.

Jan 1, 1999

By P. K. Sharma

The acidophilic and chemolithotrophic Thiobacillus ferrooxidans bacteria grown in the presence of soluble ferrous ions and solid substrates (sulfur, pyrite and chalcopyrite) as their energy source have been used in the investigations. The cell surfaces are characterised by zeta-potential, contact angle, FT-IR and FT-Raman studies. The adsorption isotherms of the cells on pyrite and chalcopyrite indicate that the adsorption begins on pyrite at a much lower equilibrium cell population than on chalcopyrite. The isotherms tend to level off at a cell adsorption density corresponding to a horizontal monolayer coverage when the geometrical dimensions of the cells are considered. In the presence of cells, the xanthate flotation of pyrite completely depressed, whereas chalcopyrite flotation is unaffected. The results suggest that chalcopyrite can be selectively floated from pyrite, also at a neutral pH region, in the presence of T. ferrooxidans cells.

Jan 1, 1999

By N M. Caira

The North Arm of Sulawesi hosts numerous gold and base metal occurrences. Compilation and analysis of geological, tectonic, geochemical and isotopic age date data collected over the past six years on the central part of this region have resulted in the definition of 184 prospects and mineral occurrences of various styles occurring in diverse geological settings representing varying tectonic environments and erosional levels. Mineralisation of the following styles have been recognised: porphyry Cu-Au; porphyry-related gold and base metal veins; high-sulphidation Cu-Au-As; low-sulphidation epithermal Au; hydrothermal breccias; and sediment-hosted Au mineralisation. Clusters of occurrences/prospects form mineralised districts representative of magmatic systems with characteristics determined by the tectonic environment prevailing during formation and the subsequent level of exposure by erosion. Distinctly different Miocene and Pliocene porphyry systems are now recognised whilst most of the remaining epithermal mineralisation is thought to be Pliocene or later. The central North Arm of Sulawesi is composed of a Miocene magmatic island arc that is built on an Early Tertiary basement of predominantly oceanic basaltic rocks and overprinted by a Pliocene magmatic arc. A structural fabric dominated by SSE arc-normal and ESE arc-parallel faults was established during the Miocene under a dextral wrench regime and intersections of these major fault sets are favoured sites for Miocene porphyry Cu-Au mineralisation. Strong dextral tension caused by the collision of the Sula Platform, at the close of the Miocene, led to arc-parallel rifting. Sinistral reactivation of the major Miocene structures in the Pliocene led to further rifting, and ENE-dilation. Pliocene intrusions and related mineralisation exploit these dilatant settings. Miocene porphyry Cu-Au mineralisation occurs where late-Miocene quartz diorite stocks have intruded the margins of earlier Miocene plutons of intermediate composition. Porphyry centres have associated peripheral base metal veins, and skarns. Overprinting of porphyry vein sets by epithermal veins is common. Sediment-hosted gold mineralisation occurs in calcareous sediments associated with the Miocene arc. Remnants of the Pliocene volcanic arc, comprised of dacite volcanics, intrusions and voluminous pyroclastics, dominate the south coast and host porphyry Cu-Au, high sulphidation Cu-Au-As deposits, low-sulphidation epithermal Au veins and disseminated gold mineralisation. In some prospects, mineralised sinters are preserved adjacent to vein systems and deeper mineralised stockworks. Diatremes are common and large caldera structures have been identified. Pliocene stocks and porphyry dyke swarms penetrate the main central and northern parts of the Miocene- arc introducing additional porphyry Cu-Au, base metal hydrothermal breccias and low-sulphidation gold veins. Rapid uplift and erosion has resulted in a highly variable preservation of styles of mineralisation adding further diversity within the mineralised districts and telescoping within deposits.

Jan 1, 1999

By P Mitchell, J Proffett, I Munajat

Batu Hijau is an island arc porphyry copper-gold deposit located in the southwest corner of Sumbawa Island in the Sunda-Banda archipelago of Indonesia. Southwest Sumbawa Island is underlain by Early Tertiary, low-K calc-alkaline to weakly alkaline volcanic and intrusive rocks. Within the mine area the oldest exposed rocks are medium- to thick-bedded, medium to coarse andesite volcanic lithic breccia, thin bedded, fine-grained volcaniclastic sandstones and mudstones and hypabyssal porphyritic andesite. In the eastern portion of the mine area there is a thick sequence of quartz diorite that postdates the andesites. A small stock of Equigranualar Quartz Diorite predates the Batu Hijau deposit and occupies the contact between the andesitic volcaniclastic sequence of rocks and the Porphyritic Quartz Diorite. Batu Hijau is a world-class copper-gold deposit that has a drill indicated mineable reserve of 914 million metric tonnes grading 0.525 per cent copper (10.58 billion pounds copper) and 0.403 g/t gold (11.8 million ounces gold). Mineralised material not in reserve accounts for an additional 580 million metric tonnes at 0.328 per cent Cu (4.19 billion pounds copper) and 0.171 g Au/T (3.19 million ounces gold), and including an oxide copper resource of 57 million tonnes at 0.37 per cent Cu and 0.12 g Au/T. Copper and gold mineralisation is directly related to quartz veining and wall rock alteration that developed with multiple tonalite porphyry intrusions. The tonalites were emplaced along the contact between the andesitic volcaniclastic rocks and quartz diorite intrusives. Copper and gold are zoned around the porphyry intrusions, which are designated by their contact-age relationships and other features as Old Tonalite, Intermediate Tonalite(s), and Young Tonalite(s). The Old Tonalite was the first copper-bearing intrusion, and although it is the smallest intrusion it has the highest quartz vein density, the strongest alteration and the highest grade. At least two tonalite porphyry intrusions followed the emplacement of the Old Tonalite. Each of these intrusions was associated with less quartz veining, weaker hydrothermal alteration and lower grade copper and gold than the previous intrusion. The intrusive cycle culminated with the emplacement of the Young Tonalite, which contains relatively few quartz veins, is weakly altered and mineralised. Hypogene hydrothermal alteration and sulphide mineralisation developed in five temporally and spatially overlapping stages. These stages consist of Early, Transitional, Late, Very Late, and I Zeolite alteration. Early alteration and sulphide mineralisation accompanied each tonalite porphyry intrusion, but with diminished intensity in each subsequent intrusion. Transitional and later stages of alteration and mineralisation postdate the tonalite porphyries and overprint Early stage alteration and mineralisation. Early alteration and mineralisation consisted of biotisation, secondary magnetite and plagioclase development, fine porphyry-style stockwork veining and ore mineralisation. Magnetite-biotite????quartz stringers and dark micaceous (biotite-sericite) veinlets were developed. The veinlets were accompanied by biotisation of magmatic mafic minerals and groundmass. Thin, irregular and discontinuous æAÆ type quartz veinlets and fine-grained chalcocite, digenite and bornite mineralisation accompanies the Early alteration. It is hypothesized that most of the copper mineralisation was originally deposited as bornite, digenite and chalcocite. Secondary plagioclase was added along the selvages of the early quartz veinlets. Early alteration is pervasive within and proximal to the tonalite porphyries. In the distal portions of the deposit, andalusite and anthophyllite formed at intermediate to shallow depths, and in the deeper portions of the deposit, respectively. The andalusite and anthophyllite probably developed at the same time as the early magnetite-biotite-quartz alteration in the core of the deposit. Transitional-age alteration consists of oligioclase/albite-sericite-chlorite- quartz-¦vermiculite and is associated with æBÆ quartz veins. Through-going planar æBÆ veins are thicker than æAÆ and æA-familyÆ veinlets and have chalcopyrite????bornite along vuggy centrelines and narrow oligioclase????albite alteration envelopes. Rare æCÆ chalcopyite¦quartz veinlets were also formed during Transitional alteration. During Transitional alteration early copper sulphide minerals (chalcocite, digenite and bornite) are converted to bornite and chalcopyrite. The majority of the copper (50 - 70 per cent) in the deposit is contained in ores in which chalcopyrite is the predominant sulphide. Late alteration consists of feldspar destruction and replacement by sericite and other minerals. The late alteration is accompanied by the development of æDÆ sulphide veinlets and veins. These veins consist of pyrite and quartz????chalcopyrite. Within the halos of the late pyritic veinlets biotite is replaced by sericite, quartz-sericite, quartz-andalusite, or quartz-pyrophyllite+kaolinite replace plagioclase, and magnetite is converted to specularite and/or pyrite. In the periphery of the deposit, the feldspar destructive alteration, together with outer zones of Transitional-age alteration, constitutes what is commonly referred to as a æpropylitic zoneÆ. Very Late hydrothermal alteration is also characterised by feldspar destruction, but differs from Late alteration in that feldspar is replaced by smectite in association with sericite and chlorite. Sulphide mineralisation in this stage consists of sphalerite, galena, tennantite, pyrite and chalcopyrite. The last stage or I Zeolite alteration consists of low-temperature, open space filling consisting of stilbite or laumontite¦calcite. Supergene oxidisation of the orebody extends to a maximum depth of 210m. Isolated deep oxidation occurs along open fracture zones to depths of ~400 metres along æDÆ vein filled fracture zones. Weak supergene copper enrichment occurs in a thin horizon (15 to 60 m thick) that blankets the top of the deposit. The main supergene copper mineral is secondary æsootyÆ chalcocite replacement of sulphide grains and as veins.

Jan 1, 1999

By J. M. A. Geldenhuis

This work is concerned with studying possible reactions in the ilmenite furnace which may control the product quality of both slag and metal. Two different approaches were followed: comparing actual furnace slag compositions with the calculated equilibrium values and using commercial oxygen probes to measure the effective partial oxygen pressure in the slag and metal. It appears from a comparison of actual furnace slag compositions (in terms of the distribution coefficient of vanadium) with equilibrium values, that the TiO1.5-TiO2reaction may fix the effective partial oxygen pressure for the vanadium reaction. This possibility was tested using commercial oxygen probes (traditionally applied during steelmaking) to measure the effective oxygen partial pressure of both the slag and liquid iron in Iscor?s 3MVA pilot plant ilmenite furnace. From these measurements, it appears that commercial oxygen probes containing zirconia solid electrolytes can be used to measure the effective oxygen potential of both the slag and liquid iron. This oxygen activity correlates well with the FeO content of the slag. The results, furthermore, clearly illustrate that equilibrium conditions are not attained in the ilmenite furnace during the production of high titania slags, since the effective oxygen potential of slag and metal differ substantially.

Jan 1, 1999

By M. Pourkani, M. Karga, Samid Banis, Andre Laplante, G. Langari-Zadeh

"The Sarcheshmeh mine processes 40,000 t/d of ore grading 0.9% Cu and 0.035% Mo. Grinding is effected in single stage ball mills using 80 mm forged steel balls, with an average consumption of 850 g/t. One of its 5 m diameter, 8 m length ball mills was emptied to measure the ball size distribution. The broken balls were first separated from the total charge, and the remaining balls classified into five size intervals of +7O mm, -70 +60 mm, -60+50 mm, -50+40 mm, -40 mm.The charge was composed of 280,534 balls (excluding broken balls) with a total mass of 305,180 kg. It was found that 4.0 % of the balls were broken; the size distribution of the charge was also markedly different from the desired one, in particular for the small-sized balls. For instance, 9% of the balls were expected to measure 2.54 to 3.18 cm in diameter, but none was found. On the other hand, the amount of top size balls was 30% higher than expected. The radial hardness of balls, which was determined by cutting the balls in halves, showed that the hardness sharply drops when the balls become smaller than 65 mm. The unexpected ball size distribution was attributed to the rapid change in the balls radial hardness. Based on the measured ball size distribution, a new ball make-up charge which uses 80% of 80 mm and 20% of 60 mm balls instead of 100% of 80 mm was proposed and implemented. The effect of the new ball make-up procedure in terms of ball consumption and grinding efficiency is being investigated."

Jan 1, 1999