The benefit of separate milling of silicate and chromite particles for chromite-rich UG-2 ores

Transcription

1 The benefit of separate milling of silicate and chromite particles for chromite-rich UG-2 ores L. Maharaj *, B.K. Loveday, J. Pocock School of Chemical Engineering, University of KwaZulu-Natal, Durban, South Africa ABSTRACT Extraction of platinum group elements (PGE) is a major source of revenue in South Africa and the reserves represent about 75 per cent of world reserves. Most of the remaining PGE reserves are located in the UG-2 chromitite layer of the Bushveld Igneous Complex. Platinum concentrators experience significant losses of valuable PGE in their secondary milling circuits due to insufficient liberation of platinum-bearing particles. The chromium oxide (Cr 2 O 3 ) content in UG-2 concentrates is typically 3 %, which results in severe operational problems in the downstream smelting process. This paper follows up from the work done in 2010 which indicated that spiral concentrators could be incorporated into the secondary milling circuit design for UG-2 platinum concentrators for separate grinding of platinum-bearing silicate particles and chromite-rich particles. A detailed mineralogical analysis was conducted on a plant audit sample from the secondary rougher tailings (SRT) stream for comparison with the laboratory test work to assess the degree of potential losses of PGEs experienced with the conventional milling circuit.the effect of the size-by-assay analyses of the laboratory milled products and flotation tailings for a low grade UG-2 ore sample from the hydrocyclone (densifier) underflow were compared to the standard process. The effects on the recovery of PGEs, and the entrainment of Cr 2 O 3 were measured in combined batch rougher flotation tests. Mineralogical evaluations of the SRT plant sample showed losses of about 37 % of the PGEs in that stream to the coarse fractions (+75 μm). The chromite entrainment was significantly reduced (>30 %) with a 3.6 % increase in the PGE recovery. This paper confirms the benefit of separate ball milling in the secondary milling circuit for a chromite-rich (>50 %) UG-2 platinum ore. Keywords: separate milling; grinding; platinum flotation; UG-2 ore; mineralogy * Corresponding Author: maharajl3@ukzn.ac.za; Tel: ; Fax: South African Journal of Chemical Engineering, Vol. 16, No. 1 1

2 1. INTRODUCTION Platinum is a major resource to the South African economy as it generates large revenues which contributes to the GDP, results in job creation, infrastructure, skills development and also addresses the poverty crisis in the country. In the Platinum Group Minerals (PGM) industry within South Africa, the main exploitable reserves of PGMs are UG-2 (Upper Group 2) deposits. Since UG-2 ore has a high chromite (FeO.Cr 2 O 3 ) content, the reduction of the chromium oxide (Cr 2 O 3 ) recovery in the final flotation concentrate is crucial to the smelting process, as the smelter furnace feed can operate effectively with an upper limit of about 2.5 % Cr 2 O 3 (Hay and Roy, 2010). This paper focuses on confirming the benefit of implementing separate milling of chromite and silicate particles in the secondary circuit of the UG-2 concentrator to platinum producers, in view of the high Cr 2 O 3 entrainment noted after secondary flotation. A detailed mineralogical analysis together with size-by-assay data were compared to the standard milling circuit design for a low grade UG-2 ore sample from the hydrocyclone (densifier) underflow to quantify the benefits of separate milling on the PGE recovery and Cr 2 O 3 entrainment. 2. THEORY 2.1. UG-2 platinum ore The UG-2 (Upper Group 2) chromitite layer in the upper Critical Zone is probably the largest PGE resource on Earth (Vermaak, 1985). Chromitite is a rock containing chromite (FeO.Cr 2 O 3 ) as its main constituent, consisting of % Cr 2 O 3, and is usually analysed as a percentage of chromium oxide (Cr 2 O 3 ) (Hay and Roy, 2010). The average grain size of the PGEs in the UG-2 ore were found to be 9.3 m in diameter (Lee, 1996). This indicates that it requires a fine grind to liberate the PGEs. Lidell et al. (1986) noted that the majority of the PGEs (70% by mass) in the Merensky ore were found to be between 150 and 38 m and therefore could be liberated with a much coarser grind as compared to the UG-2 ore Smelting of UG-2 ore and the effect of chromite South African Journal of Chemical Engineering, Vol. 16, No. 1 2

3 The processing of UG-2 chromites yields significant savings in the total energy and electrical power required, both per unit of ferrochrome produced and per chromium content (Barnes et al., 1983). Barnes and Newall (2006) noted that chromite spinel causes operational problems such as difficulties in obtaining clean separation of matte and slag, high operating temperatures, a shortened furnace life due to a reduced capacity, and high maintenance and operational costs with lengthy repair periods. Hay and Roy (2010) noted that although the gap in the percentage Cr 2 O 3 between the furnace practical limit and the UG-2 concentrator product is being closed, certain concentrators can produce concentrates with less than 2.5 % due to the difference in the mineralogy of the ore Typical UG-2 secondary ball milling grinding circuits In South Africa, UG-2 plants commonly consist of an MF2 (Mill-Float-Mill-Float) where primary grind is in the region of % -75 m followed by a finer secondary grind of 80 % - 75 m (Hay and Schroeder, 2005). Each grind stage is followed by its own flotation circuit. Most secondary circuits treating UG-2 platinum ores adopt a single stage ball-mill for the grinding of the primary flotation tails, followed by secondary flotation. A conventional hydrocyclone or densifier is used to remove the ultra-fine material prior to secondary milling and the hydrocyclone underflow is the feed to the secondary ball mill which is operated in open circuit. The feed to the secondary ball mill usually comprises of large amounts of liberated chromite and silicate-rich particles containing the bulk of the PGE-bearing minerals in a locked state. In order to minimise PGE losses and energy expenditure, there is considerable incentive to find ways of milling the silicate-rich particles selectively. Anglo Platinum have adopted separate milling and flotation in their milling circuit design which has been successfully implemented on most of their major concentrators (Rule and Anyimadu, 2007). This circuit consisted of separate silicate and chromite re-grind stages with silica scavengers and chromite flash flotation using a conventional hydrocyclone as the separating device. Recent test work (Maharaj et al., 2011) has shown that a full-scale MG-type spiral concentrator was able to achieve a good separation of silicate and chromite particles in a single stage gravity separation process on a UG-2 ore hydrocyclone underflow sample. Separate laboratory batch ball milling tests on the spiral products followed by combined rougher flotation showed positive results with a 40% reduction in the Cr 2 O 3 recoveries with similar PGE recoveries (< 1 %) as South African Journal of Chemical Engineering, Vol. 16, No. 1 3

4 compared to the standard process under laboratory conditions. The total energy input to the laboratory batch ball milling circuit was re-distributed with 88 % of the milling energy input to the silicates ball mill and 12% of the total milling energy distributed to the chromite ball mill. 3. EXPERIMENTAL PROCEDURE The milling circuit configuration is shown in Figure 1. Lights Mill (Silicate-Rich) Densifier Underflow Spiral Concentrator Heavies Mill (Chromite- Rich) Secondary Rougher Flotation Figure 1: Diagram of UG-2ore secondary milling circuit configuration (Maharaj et al., 2011) 3.1. Mineralogical investigations of a secondary rougher tailings plant sample A plant audit was undertaken at a UG-2 platinum concentrator. A representative sample of the secondary rougher tailings (SRT) was taken in addition to various other samples from the secondary milling and flotation circuit. A bulk densifier underflow sample (feed to secondary mill) was collected for the laboratory test work campaign. A representative 200g sample of the SRT was submitted to MINTEK for a PGM search and a bulk modal analysis. The mineralogical test work was conducted to provide the following information: PGM species and size distribution PGM liberation and floatability characteristics Minerals associated with the PGMs Bulk modal results, i.e. minerals identified and their relative proportions. Mineral association- i.e. the manner in which the various phases are associated with each other Experimental procedure for the separate milling tests Bulk rougher spiral concentrator runs were conducted on the low grade UG-2 secondary densifier underflow sample based on the previous laboratory conditions (Maharaj et al., 2011). South African Journal of Chemical Engineering, Vol. 16, No. 1 4

5 The bulk rougher spiral lights (silicates-rich) and heavies (chromite-rich) products were then filtered, dried and stored for further milling and flotation test work using the previous laboratory conditions(maharaj et al., 2011) Size-by-assay analyses Samples were sized at 106, 75, 53, and 38 m to compare their size distributions and effect on particle breakage. Samples were first wet screened at 38 m and the +38 m fraction was dried and screened to determine an accurate size distribution. Selected milling samples, i.e. standard ball mill test, separate milling with equal energy input (50%) to the lights (silicates- rich) and heavies (chromite-rich) mill, and the 88 % energy input to the lights mill and 12% to the heavies mill were submitted for chemical analyses for 4E and Cr 2 O 3 to determine the metal distributions in each size fraction. Flotation tailings samples for the standard process and the separate milling option with 88 % energy re-distribution to the lights mill were sized at 106, 75, 53, 38, and -38 m. These samples were submitted for 4E and Cr 2 O 3 analyses at MINTEK, to determine the amount of coarse and fine PGEs that were liberated and floated as a result of the separate milling design, as compared to current process. This would indicate whether the separate milling was effective in reducing the amount of fine Cr 2 O 3, which would impact on the Cr 2 O 3 entrainment to the secondary rougher flotation concentrate. 4. RESULTS AND DISCUSSION 4.1. Size-by assay and mineralogy results Size-by-assay results of plant audit samples The size-by-assay data for the secondary UG-2 concentrator SRT plant audit sample is discussed in this section. This aids in benchmarking the current secondary milling and flotation concentrator circuit. The mass, platinum 4E and Cr 2 O 3 distributions for the SRT are shown in Table 1. South African Journal of Chemical Engineering, Vol. 16, No. 1 5

6 Table 1: Size-by-assay of the plant audit SRT sample Secondary Rougher Tails (SRT) Distribution (%) Screen Size ( m) Mass (%) 4E Cr 2 O Total The SRT is the major component of the final tailings (discard) from the process together with the tailings from the secondary cleaner circuit. The SRT stream contained close to 33 % coarse material (+75 m) and also a large proportion of fines (51% of -38 m). This will be related to mineralogy data in the subsequent section to provide details on the PGE and Cr 2 O 3 distributions in the circuit. About 37 % of PGEs in that stream occur in the coarse fraction (+75 m). A size-byassay on the densifier underflow stream indicated that about 68% of the PGEs were in the coarse fractions (+75 m) before secondary milling and flotation which indicates that the current secondary milling process may not be effective enough to reduce the amount of coarse PGEs present in the UG-2 ore. Majority of the remaining PGEs (45.6 %) lost to the SRT are largely distributed in the ultrafines (-38 m). The SRT stream contained 51 % Cr 2 O 3 in the ultrafines (-38 m). A size-by-assay analyses of the densifier underflow indicated that only about 1 % Cr 2 O 3 was present in this fraction prior to secondary grinding which suggests that significant over-grinding of Cr 2 O 3 is occurring in the secondary ball mill when milling the entire stream to the target grind Results on mineralogy of the secondary rougher tails (SRT) sample A 200 g un-sized sample of the SRT from the plant audit was sent to the Process Mineralogy Group at MINTEK for a PGM search and a bulk modal analysis. Since this was a low grade sample of below 1 g/t 4E, only 61 PGM grains were found and analysed. This may affect the South African Journal of Chemical Engineering, Vol. 16, No. 1 6

7 statistical validity of results; however, this data still provides sufficient understanding of where the majority of losses of platinum minerals arise. In order to interpret the mineralogy data it is important to understand how PGM grains occur in plant samples. Table 2 defines the common terminologies used when discussing such data. Table 2: Table of mineralogical terms (MINTEK Process Mineralogy Group) L Liberated PGMs SL PGMs associated with liberated BMS (Base Metal Sulfides) AG PGMs attached to Silicate or Oxide gangue particles SAG PGMs associated with BMS attached to Silicate or Oxide gangue particles SG PGMs associated with BMS locked in Silicate or Oxide gangue particles G PGMs locked within Silicate or Oxide gangue particles A modal analysis based on the 61 PGM grains revealed that the main minerals in the secondary rougher tailings (SRT) were chromite (65 %), pyroxene (16%) and feldspar (14 %) by mass, with all other minerals such as pyrite, pentlandite, chalcopyrite and pyrrhotite less than one percent by mass. The PGM grain mode of occurrence is shown in the Table 3. Table 3: PGM grain mode of occurrence (MINTEK Analysis) PGM grain mode of occurrence PGM No of No of PGM grains Vol % PGM grains smaller than 3 µm ECD L SL AG SAG SG G It can be seen from the above table that out of the 61 PGM grains that were selected, 15 grains (24 %) are attached to silicate or oxide gangue particles, 11grains (12.8 %) are associated with base metal sulphides and 16 grains (25.2 %) are locked within silicate or oxide gangue minerals. Most notably of the 25.2 % locked PGMs, almost half of those PGM grains (7) are less than 3 m which indicates that only a very fine grind will liberate such particles. About 13 % of the grains are fully liberated (9 grains) but still lost to the flotation tailings. This is also due to the fact that almost fifty percent of them were also extremely fine (<3 m) and were lost due to South African Journal of Chemical Engineering, Vol. 16, No. 1 7

8 sliming effects or poor flotation techniques available to float ultra-fine liberated particles. Figure 2 shows the volume percent of PGMs relating to the grain mode of occurrence. Figure 2: MLA analysis of SRT sample (MINTEK Analysis) The distributions of the different PGM types are shown in Table 4 below. Table 4: Distribution of PGM types in SRT sample (MINTEK Analysis) Distribution of PGM types PGM PGM No of Average Type Vol % PGMs ECD (µm) PtFe PtS PtRhCuS PtPdS Ru(Os,Ir)S PtPdBiTe PdS PdPb PtPdSb PtAs PtAsS PtPbCuNiS PdBiTe AuAg This data confirmed the results of the modal analysis which indicated that pyroxene (PtS) and feldspar (PtFe) were two of the main minerals found in the sample. These two minerals South African Journal of Chemical Engineering, Vol. 16, No. 1 8

9 accounted for the major PGM losses as there were 17 grains of PtS, 9 grains of PtPdS and 12 grains of PtFe found in the SRT. Other PGM types are in small proportions and it may not be statistically relevant to consider them as there are less than 5 grains of these PGM types. From the mineralogy of the SRT plant audit sample, it is evident that a large proportion of PGMs that were lost to the final flotation tailings after secondary grinding and rougher flotation are due to poor liberation, as most PGM grains were either attached or locked with silicate or oxide gangue (silicate or chromite), or with base metal sulphides. This is a definite indication that the PGM circuit does require an improvement to the secondary ball milling process Separate milling results on a low grade PGE sample Table 5 confirmed that the 88% energy input to the lights ball mill and 12 % energy input to the heavies ball mill produced the best overall results. An average PGE recovery of 54.9 % was achieved with 1.5 % Cr 2 O 3 entrained to the secondary rougher concentrate. Table 5: Flotation results on the separate milling on the low grade feed sample Energy distribution to the lights and heavies mill (%) Masspull (%) Recovery (%) Lights Heavies Concentrate 4E Cr 2 O 3 Water Duplicate tests using the standard milling circuit (single ball mill) resulted in a 51.3 % PGE average recovery with 2.3 % Cr 2 O 3 entrainment. This indicated that the separate milling circuit resulted in a 3.6 % higher PGE recovery as compared to the standard test. The Cr 2 O 3 entrainment was reduced from 2.3 % to 1.5 %, which was a 35 % Cr 2 O 3 reduction overall. These results compare quite well with the findings with the high grade UG-2 sample (Maharaj et al., 2011) and suggested that this process route has significant benefit in reducing Cr 2 O 3 entrainment and substantially increasing the PGE recovery. The results of this investigation were plotted to South African Journal of Chemical Engineering, Vol. 16, No. 1 9

10 compare the trend with the initial findings of the high grade sample. Figures 3 and 4 illustrate the recovery profiles for the repeat runs on the low grade PGE sample with the initial runs on the high grade PGE sample (Maharaj et al., 2011). The variances in the masspull, 4E, Cr 2 O 3, and water recoveries for the repeated standard flotation tests on the low grade ore were 0.1 %, 4.5 %, 0.1 % and 0.2 % respectively. The variances in the masspull, and 4E recoveries for the repeated flotation tests with an 88 % energy re-distribution to the lights mill were 0.2 % and 5.5 %, with a 0 % for both the Cr 2 O 3, and water recoveries. The graphs in Figure 3 indicate that although the 4E recoveries are of the order of 20 % lower in platinum mineral content for the low grade repeat sample, these results are encouraging as they result in a similar PGE recovery profile as in the case of the high grade material. Figure 4 shows the closeness of the Cr 2 O 3 graphs for the Cr 2 O 3 entrainment curves for the initial and repeat results. The curves in Figure 4 emphasise the importance of energy re-distribution in the secondary ball milling circuit for UG-2 ores and provides evidence that energy re-distribution through separate milling of the silicates and chromite particles is a vital method of reducing Cr 2 O 3 recoveries to the secondary rougher concentrate. The effect of water recovery on chromite entrainment was also investigated and the resulting relationship is shown in Figure 5. The data point for the energy distribution water recoveries would fit a linear profile for four out of the five measured points. The lowest water recovery of about 14 % was noted for the 88 % energy distribution to the lights mills. This data is in agreement with entrainment theory which indicates that an increase in water recovery to the froth corresponds to an increase in gangue mineral entrainment. A comparison of the water recovery of the conventional process route shows that the Cr 2 O 3 entrainment is much higher for a similar water recovery for the conventional process (2.3 % Cr 2 O 3 ) as compared to the 88% energy distribution run (1.5 % Cr 2 O 3 ). This indicates that the recovery of the higher amount of Cr 2 O 3 in the conventional process may be due to the secondary flotation feed containing a higher proportion of liberated Cr 2 O 3 as compared to separate milling tests. Since only 12 % of the total milling energy was distributed to the chromite-rich ball mill, the reduced Cr 2 O 3 recovery may be possibly due to a lower amount of liberated chromite. The conventional process may contain more liberated Cr 2 O 3 and thus that the recovery may be due to pure flotation rather than entrainment. This was considered when comparing the size by assay analyses for the milled products and is discussed below. South African Journal of Chemical Engineering, Vol. 16, No. 1 10

11 70 65 ) (% 60 e r y v e c o R E 55 4 Separate milling - High Grade Sample Separate milling - Low Grade Sample % Energy input to the Lights Mill Figure 3: Graph of 4E recoveries as for the high and low grade samples ) 2.4 (% 2.2 e r y v e c o 2 R O r 2 C 1.6 Separately milling - High Grade Sample Separate milling -Low Grade Sample % Energy input to the Lights Mill Figure 4: Graph of Cr 2 O 3 recoveries for the high and low grade samples South African Journal of Chemical Engineering, Vol. 16, No. 1 11

12 2.6 ) (% v e r y e c o R O 3 r 2 C Energy Re-distribution Conventional Process Water Recovery (%) Figure 5: Graph of Cr 2 O 3 recovery versus water recovery for the low grade repeat tests Size-by-assay analyses of the milled products were compared for the standard process and for two separate milling options, i.e. a 50 % energy distribution (8 minutes milling each) to the lights and heavies mill, and for the 88 % energy distribution to the lights mill (14 minutes milling) and 12% energy distribution to heavies mill (4 minutes milling) to determine the cumulative product size distributions and the 4E and Cr 2 O 3 mineral deportment per size fraction. This information may provide more scientific evidence to link the flotation results to possible mineral liberation for the platinum and chromite minerals. Table 6 shows the cumulative product size distributions for the milling products with the associated flotation recoveries. Table 6: Cumulative % passing each screen size for assayed laboratory milled products Avg. 4E Recovery (%) Avg. Cr 2 O 3 Recovery (%) Screen Size ( m) Standard (100 % Energy) Lights Mill (88 % Energy) Heavies Mill (12 % Energy) Lights Mill (50 % Energy) Heavies Mill (50 % Energy) South African Journal of Chemical Engineering, Vol. 16, No. 1 12

13 Table 6 indicates that the standard test results in 80% of the milled material - 75 m after 16 minutes of batch laboratory ball milling. This milled sample resulted in a 51.3 % 4E recovery with a 2.3 % Cr 2 O 3 entrainment to the rougher concentrate for the low grade sample. When implementing the separate milling circuit which comprised of separation with a spiral concentrator followed by separate ball milling of the lights and heavies, it can be seen that an equal energy distribution to both streams results in 77 % of the lights (silicates) material -75 m and 83.3 % of the heavies (chromite) material -75 m. This indicates that for an equal mass loading of 250 g to the mill and with an equal energy input to each mill, the chromite particles were found to have a marginally lower work index than the silicates. This is possibly due to the composite nature of the silicate particles that may result in their being slightly harder. The flotation results for the 50:50 energy distributions showed a 51.1 % flotation recovery with a 2.5 % chromite recovery. The sizing data above indicates that the high Cr 2 O 3 content is due to overgrinding of the chromite, as 83.3 % of the material is also ground -75 m, with more than 50 % of this material also -38 m. This possibly liberated more Cr 2 O 3 as the 12 % energy distribution showed a sizing of only 46.4 % of the material -75 m with only 22.7 % of the material -38 m with the resulting flotation recovery at 54.9 % 4E, and most significantly a 35 % drop in the Cr 2 O 3 content to 1.5 % in the rougher concentrate. This reduction in Cr 2 O 3 is strong evidence that supports the implementation of separating grinding in the secondary milling circuit as it clearly shows the effect on chromite liberation whilst also significantly improving the PGE recoveries. The 88% energy distribution to the lights mill resulted in 91.3 % of the silicates being ground -75 m as compared with 80% for the standard case, and 77 % for the equal energy distribution with 62.2 % of the material also below 38 m. A fine grind in excess of 90% -75 m is typically achieved with alternative milling technologies such as SMD's and IsaMills, which are added as tertiary milling devices. These units may also contribute to additional energy costs whereas the implementation of a low cost gravity separation device, such as the spiral concentrator, a good split in silicates and chromite can be achieved. This can be incorporated with two separate ball mills to achieve much finer grinds on silicates and coarser grinds on the chromite-rich fraction Comparison of flotation tailings samples The PGE and Cr 2 O 3 losses in each size fraction of the flotation tailings for the standard process and the separate ball milling circuit were compared. The results are shown in Table 7. South African Journal of Chemical Engineering, Vol. 16, No. 1 13

14 Table 7: Results of the size assay of the flotation tailings Avg. 4E Rougher Recovery (%) Avg. Cr 2 O 3 Rougher Recovery (%) Avg. Concentrate Mass % Size-by-Assay Standard rougher tailings Separate ball milling circuit design rougher tailings Screen Size ( m) Mass (%) 4E (%) Cr 2 O 3 (%) Mass (%) 4E (%) Cr 2 O 3 (%) Total in Rougher Tailings (%) A comparison of the rougher tailings of the standard process showed that it contained 18% of the total PGEs from densifier underflow that were lost to the m fraction. These fractions contain about 18 % of the total mass. The separate ball milling circuit design resulted in only an 8 % total PGE loss in the +106 and m fractions (10 % coarse 4E reduction) in a larger mass of about 31%. About 2 % of the PGEs were lost to the +106 m in the separate milling rougher tailings as compared to 0.8 % 4E for the standard rougher tailings. The corresponding Cr 2 O 3 minerals present in the +106 m was 1.3 % in the optimal test; however, the separate milling circuit showed an 18 % increase in coarse Cr 2 O 3 as compared with the standard tailings. This showed the benefit of separate milling as the over-grinding of chromite can be significantly reduced. This was confirmed by comparing the m and -38 m fractions, where the separate milling process resulted in about a 16 % lower Cr 2 O 3 content. The marginally higher PGE loss in the +106 m may also indicate that a proportion of the PGEs are associated with the chromite. When comparing the plant audit results of the secondary rougher tailings from the concentrator (Table 1) with that of the separate milling circuit tailings, it is evident that separate milling significantly reduces the coarse PGE losses (37.1 % PGEs in SRT) as compared to 17 % PGEs for separate milling circuit. The SRT also compares quite well with the laboratory tailings of the standard circuit which indicates that the milling and flotation characteristics in the laboratory are a good indication of the plant performance. An additional 10 % of the Cr 2 O 3 are not over South African Journal of Chemical Engineering, Vol. 16, No. 1 14

15 ground when comparing the Cr 2 O 3 distribution in the coarse fractions for the SRT and the separate milling circuit. The above results are very encouraging and indicate the significant benefit of considering separate milling of the silicates and chromite particles in the secondary UG-2 circuit. This process may reduce the losses of coarse PGEs in the secondary tailings which are a major concern to platinum producers. 5. CONCLUSIONS Separate milling of the chromite and silicate particles of a low PGE grade densifier underflow sample showed an improvement of 3.6 % in the PGE rougher flotation recovery with a reduction in the Cr 2 O 3 entrainment by over 30 %. This paper confirmed previous research findings (Maharaj et al., 2011) that separate grinding of silicates and chromite particles is a beneficial option for processing chromite-rich UG-2 ores. The batch silicates laboratory ball mill was able to achieve much finer grinds in excess of 90% - 75 m which indicated that the move towards finer grinding technology for processing UG-2 ores may be overcome by considering gravity separation followed by separate milling in the secondary ball milling circuit design. REFERENCES Barnes, A.R., Finn, C.W.P and Algie, S.H The prereduction and smelting of chromite concentrate of low chromium concentrate of low chromium-to-iron ratio. The Journal of the South African Institute of Mining and Metallurgy, March 1983: pp Barnes, A.R., and Newall, A.S Spinel removal from PGM smelting furnaces, Southern African Pyrometallurgy, Edited by R.T. Jones, The Southern African Institute of Mining and Metallurgy, Johannesburg, 5-8 March, pp Hay, M.P. and Roy, R., 2010, A case study of optimising UG2 flotation performance, Part 1: bench, pilot and plant scale factors which influence Cr 2 O 3 entrainment in UG2 flotation. Minerals Engineering, Vol. 23, Issues 11-13, October 2010, pp Hay, M.P. and Schroeder, G., 2005, Use of the SUPASIM flotation model in optimizing Impala s UG2 circuit, Minerals Engineering, Vol.18, pp South African Journal of Chemical Engineering, Vol. 16, No. 1 15

16 Lee, C.A A Review of Mineralization in the Bushveld Complex and some other Layered Mafic Intrusions, In: Cawthorn, R.G., Layered Intrusions, Elsevier Science, pp Liddell, K.S Years of UG2 concentrators, Mintek 75 A celebration of Technology Conference, Randburg, 4-5 June 2009 Liddell, K.S., McRae, L.B and Dunne, R.C Process routes for beneficiation of noble metals from Merensky and UG-2 ores. Mintek Review No.4, pp Maharaj, L., Loveday, B.K., Pocock, J The effect of the design of a secondary grinding circuit on platinum flotation from a UG-2 ore, Minerals Engineering, Vol. 24, Issues 3-4, February-March 2011, pp Rule, C.M. and Anyimadu, A.K Flotation Cell Technology and Circuit Design An Anglo Platinum Perspective, Flotation Cell Technology in the 21st Century, The Southern Institute of Mining and Metallurgy. Vermaak, C.F The UG-2 layer-south Africa s slumbering giant. Chromium Review, Vol. 5, pp South African Journal of Chemical Engineering, Vol. 16, No. 1 16