1. Arsenic in feed (Jul+Aug 2009) is up at 51%, higher than 30 months average (Jan 2007 June2009), 1 153ppm vs 765ppm. 2. Arsenic in cons is down at 59% of the corresponding value for identical periods in 1 above, i.e. 1 449ppm vs 2 473ppm….August is running at 1 029ppm, the lowest ever in the history of the plant, first time sub 1 000ppm was achieved on 20th and 21st, month to date!!!!! Figure 1. Rejection of arsenopyrite is definitely taking place at the cleaning stage because rougher concentrate is still running with a tenor of about 3 650ppm arsenic.

Figure 1: Arsenic in concentrate since January 2007 3. We are definitely recovering less arsenic in cons compared to before, 8.2% vs 23.9% for the corresponding periods shown in 1 above. Figure 2.


Figure 2: Arsenic recovered in concentrate since January 2007 4. On the job training continues to get the operators to recognise forth texture, sheen and colour and its implications on the quality of the concentrate produced. Figures 3, 4 and 5.

Figure 3: Guelb’s froth on 29th December 2008


Figure 4: Best ever froth produced at Guelb, August 2009

Figure 5: Sketch indicating how froth column is to be cut to minimise arsenic in cons


5. Starting last Tuesday, we are critically looking at collector (SEX) dosage. Preliminary work concentrated on the physical aspects of flotation, we are now going to focus on the more difficult, chemical aspects. It will be refined as data is acquired and will form the backbone of a process control system, linking collector dosage to feed grade and tonnage. We expect: • Steadier operation • Better concentrate quality • Results consistent with plant observations, i.e., fewer surprises from lab results!!

After arsenic contamination of the concentrate, high graphite content of the ore and its associated treatment problems has been the second biggest problem at the mine. It is felt that since the arsenic problem has been solved, attention should now be focused on the graphite problem, as follows: It is proposed that a Davcra type flotation cell (Figure 7) be used to clean the rougher concentrate and skim graphite, prior to cleaning of the scavenger concentrate in the current cleaner circuits. By its hydrophobic nature, most of the graphite should be floating in the first two cells on the rougher/scavenger bank, i.e. the roughers. Sending this graphitic concentrate to the conventional cleaner banks, as we do currently, is inviting disaster, for the following reasons: • It is a viscous froth, so cleaning efficiency will deteriorate rapidly as a result, mainly because of entrained insols. A viscous froth does not offer selectivity, so arsenopyrite entrainment could also be a problem. • Such froth, if mechanically agitated, becomes “whipped”, like cream. So it is best treated in a cell with a more quiescent regime, like a column cell, a Davcra cell, or to a lesser extent, the Jameson or Maxwell cells. • The surface active fine sulphides, Table 1, and finer graphite combine using van der Walls adhesion to form a strong unbreakable froth which water sprays fail to break. It is such froth which builds up in the pump sumps and forces us to reduce tonnage, Figure 6.

Figure 6: Persistent unbreakable froth, indication of van der Walls attraction


A column type cell generates finer bubbles, 300-500 microns approximately, compared to 1 000 microns for an agitated cells. The operators floating talc, oxides and molybdenum have found these types of cells to be extremely useful. In theory, smaller bubbles generate less turbulence behind a rising air bubble, so the fine mineral particles (graphite/talc/molybdenite) can get attached to that bubble, or, cannot get dislodged easily in its greatly reduced wake. 1. On the other hand, skimming material from rougher cells 1 and 2 to a direct cleaning section should also help the froth texture produced by scavenger concentrate in the existing cleaner cells and total plant recovery, for the following reasons: 2. The concentrate from cells 3 to 6 is different in granulometry, significantly coarser than the concentrate from cells 1 and 2, as shown in Table 1 below for p80 size in microns. This should lead to a more brittle froth. Such froth should not build up in pump sumps, and the process of breaking and reforming should be beneficial also to concentrate grade, as insols and the slower floating sulphides (pyrite, arsenopyrite and pyrrhotite) will be given a chance to report as tailings during the froth collapse phase. Leading to copper enhancement and arsenic reduction. Table 1: Results from plant survey of 24th April 2007 and 16th May 2007 Cell 1 62.1 57.8 Cell 2 55.1 57.9 Cell 3 83.6 70.4 Cell 4 82.6 67.7 Cell 5 73.7 70.1 Cell 6 74.9 74.8 Sample 24th April 16th May

3. Since pump sumps will not be overflowing, and with an enhanced concentrate grade obtained, operators will be able to pull the scavenger cells harder, as scavengers should be run. This practice will lead to higher copper and gold recoveries. • Use a tank cell to remove the faster floating fines. Referring to the comprehensive plant survey conducted on 16th May 2007, almost 64% of float feed is -53 microns, Picture 1, leaving the conventional cleaner cells to cope with the slower floating fine sulphides. These fines should float early; Picture 2 confirms this, p80 approx 58 microns from a floatn feed of 96.5, while the last cell in the bank, cell 6, floated the slower floating coarser sulphides and locked middlings. Picture 3, p80 of almost 75 microns! This was the material which needed regrinding, or in the case of the old circuit, one stage of upgrading and that concentrate sent for regrinding.


Float Feed
Size - µm +212 +180 +150 +106 +75 +53 +45 -45 Total Grade Weight % Dist Cu Dist As 1.87 1.84 2.76 9.60 12.87 12.07 5.35 53.63 1.0 1.0 1.5 6.9 12.4 12.8 6.0 58.4 1.62 0.6 0.2 0.0 5.2 9.4 15.4 6.3 62.8 638 Dist Au 1.2 2.2 3.0 9.0 13.5 14.2 7.5 49.4 1.99

Rougher Cell 1 Concentrate
Size - µm Weight % Dist Cu Dist As Dist Au +106 +75 +53 +45 -45 Total Grade Total 3.93 7.79 10.60 5.11 72.57 4.0 9.1 12.9 5.6 68.4 20.95 P80 (µm) 4.8 17.3 25.8 12.1 39.9 3737 2.4 1.3 9.0 7.5 79.9 14.26

P80 (µm) 96.54 Total n Picture 1: Float feed p80 96 microns


Picture 2: First rougher cell p80 58 microns

Rougher Cell 6 Concentrate
Size - µm +106 +75 +53 +45 -45 Total Grade Total

Weight % Dist Cu Dist As Dist Au 7.30 12.59 16.14 6.07 57.90 14.0 11.2 10.2 3.0 61.6 1.70 P80 (µm) 0.0 0.0 4.1 2.4 93.4 1283 10.6 7.1 10.8 3.9 67.6 3.30


Picture 3: Last rougher cell p80 75 microns In addition, with the imminent commissioning of the ball mill and the anticipated concomitant increase in throughput, removal material early to the concentrate thickener will soon become an imperative. My prognosis is that the proposed installed Davcra cell will be used full time, even during the treatment of less refractory low graphitic feed! Skimming graphite off as early as possible to the filter plant will also have downstream benefit in the CIL circuit, the amount of graphite present to act as “preg robber” will be greatly reduced thus enhancing gold recovery.


The Davcra cell is non-agitated, so maintenance cost is minimal, on average the feed nozzle is changed every 4-6 weeks. Instrumentation is basic, level controller (often a DP cell, tails auto valve, air flow meter and a small control panel to link these). The major cost to be associated with this project will be the pump required to pump the concentrate to the cons thickener. Tailings will gravity flow into the 1st cleaner tails sump. A mains compressed air pressure of 6 bars minimum is required for the air sparger. There is no need to rubber line the tank.

Figure 7: Sketch of proposed Davcra cell

RAMOUTAR SEECHARRAN Plant Superintendent 24 August 2009