Copper flotation grade-recovery trade-off with talc penalty
A copper flotation plant treats 12,000 t/day of ore assaying 0.85% Cu, mainly chalcopyrite, with 3.5% pyrite and 18% talc. The circuit is rougher-scavenger-cleaner, grinding to P80 90 µm, rougher pH 10.5, cleaner pH 11.8, sodium isopropyl xanthate at 25 g/t, and MIBC at 15 g/t. Cleaner tail returns to the rougher. Current concentrate assays 21% Cu, 5.2% MgO and 12% SiO2, at 87% copper recovery. The smelter applies a penalty above 4% MgO and above 10% SiO2. The plant wants to raise concentrate grade to at least 26% Cu without losing more than one percentage point of copper recovery. A proposal is to regrind cleaner feed to P80 30 µm; an alternative is to add a talc depressant and split the cleaner tail.
1. Calculate the current concentrate mass pull and the mass pull required at 26% Cu if recovery is held at 87%. What does the difference imply about the amount of gangue that must be rejected?
Read the worked answer
Take 100 t of feed. Copper feed = 0.85 t. At 87% recovery, copper in concentrate = 0.85 × 0.87 = 0.7395 t. At 21% Cu, concentrate mass = 0.7395 / 0.21 = 3.52 t, so mass pull is about 3.5%. At 26% Cu, concentrate mass = 0.7395 / 0.26 = 2.84 t, or 2.8% mass pull.
The difference is 3.52 - 2.84 = 0.68 t per 100 t feed, about 19% of the current concentrate mass. That is a large rejection requirement: the circuit must send roughly one-fifth of the current concentrate mass to tailings while losing no more than about one percentage point of copper recovery. A simple increase in cleaner pH is unlikely to achieve that without a liberation or depressant change, because the rejected material must be low enough in copper not to erode recovery.
2. Evaluate the regrind proposal versus the talc depressant and split-cleaner-tail proposal. Which would you test first, and what measurements would decide?
Read the worked answer
The regrind proposal addresses liberation: at P80 90 µm, chalcopyrite may still be locked with pyrite or talc, and regrinding to 30 µm could raise cleaner grade. However, talc is naturally floatable and regrinding increases slimes, which can worsen entrainment and froth stability. A 30 µm regrind is also a large capital and operating change, and it may not fix talc reporting by true flotation.
The talc depressant plus split cleaner tail is more directly aimed at the MgO penalty. A depressant such as guar or CMC can suppress talc, while splitting the cleaner tail prevents circulating talc and pyrite from building up in the rougher. I would test the depressant and split-tail option first because it is lower capital and directly targets the 5.2% MgO. Deciding measurements would be concentrate grade and MgO, copper recovery, cleaner tail copper, froth stability, and a mineralogical scan of the cleaner tail to see whether copper is lost as liberated particles or as locked composites.
3. If the regrind is installed and grade rises but copper recovery falls by two percentage points, what would you check and how would you recover performance?
Read the worked answer
First check whether the recovery loss is in the cleaner or the rougher. A two-point drop after regrinding often comes from overgrinding: slimes are harder to float, entrain more gangue, and can stabilize a brittle froth. Compare particle-size distributions and mineral liberation before and after regrind, and assay the cleaner tail and final tail by size class. If coarse liberated chalcopyrite is reporting to tailings, the regrind may be too fine or the collector addition may be inadequate for the new surface area.
Recovery actions include adjusting collector and frother dosage to the finer feed, reducing cleaner pH if it is over-depressing chalcopyrite, adding staged addition rather than all collector to the rougher, and controlling the split of cleaner tail so values are not over-circulated. If slimes are the problem, a split cleaner circuit or a flash flotation step for coarse values may be better than a single regrind. The decision should be based on size-by-size assays, not on overall grade alone.