Cyclone Classification Failure at 390% Circulating Load
A 36 ft SAG mill (8 MW installed, 12% ball charge, 75% critical speed) discharges through a pebble crusher to a 5.5 m × 9.1 m ball mill (30% ball charge, 6.5 MW installed). Ball mill discharge reports to ten 660 mm cyclones, eight in service; overflow goes to copper flotation and underflow returns to the ball mill. Design is 1,200 t/h at P80 150 µm. Last month: throughput 1,020 t/h, cyclone overflow P80 205 µm, ball mill draw 5.6 MW, circulating load 390%, cyclone feed density 58% solids at 110 kPa on the distributor, ball mill discharge 78% solids. Pebble crusher product (90 t/h, 16 mm CSS) returns to the SAG feed belt. Flotation recovery has fallen 2.5 points, flotation feed now carries 22% minus 10 µm (previously 14%), and reagent cost per tonne is climbing. The superintendent wants approval to raise ball charge to 35% and speed to 78% critical, arguing that "the mill is only at 86% of installed power, so more grinding energy will fix the grind".
1. Diagnose the most probable root cause of the coarse overflow together with the rising slimes fraction, and rank corrective actions. Support your ranking with a mass balance calculation.
Read the worked answer
Mass balance. At steady state, overflow = fresh feed = 1,020 t/h. CL = underflow/overflow = 3.90, so underflow = 3,978 t/h and cyclone feed = 4,998 t/h. With solids SG 2.8 at 58% w/w, pulp density = 1/(0.58/2.8 + 0.42/1.0) = 1.59 t/m³, giving 3,135 m³/h, or about 0.11 m³/s per cyclone across eight units - well inside the hydraulic capacity of a 660 mm cyclone at 110 kPa. The circuit is not volume-limited; it is density-limited.
Diagnosis. At 58% solids the pulp is viscous and classification efficiency collapses in both directions: coarse particles are entrained to the overflow (P80 205 µm) while fines are carried down with the water to the underflow and reground. That single mechanism explains all three symptoms - coarse product, 390% circulating load, and a doubling of the minus 10 µm fraction with the associated flotation recovery loss.
Ranked actions.
- Add water at the mill discharge pump box to reach 45–48% solids. Recalculated at 45% solids the pulp is 1.41 t/m³, so the same solids load needs 3,553 m³/h (0.12 m³/s per cyclone) - still within capacity, so no additional cyclones are required.
- Measure vortex finder and apex diameters against specification. A worn or oversized vortex finder coarsens the overflow directly; a worn apex ropes and sends fines to the underflow.
- Check per-cyclone pressure and roping/spray pattern to rule out a single rogue unit.
- Redirect pebble crusher product to the cyclone feed rather than the SAG feed belt; 90 t/h of 16 mm material is currently being re-crushed and reground unnecessarily.
- Revisit mill charge only after the classification cut point is corrected and re-surveyed.
2. Evaluate the superintendent's proposal to raise ball charge to 35% and speed to 78% critical.
Read the worked answer
Verdict: defer, and probably reject as stated.
The ball mill is at 5.6 MW of 6.5 MW installed, so there is headroom in the motor - but that is not the constraint. With a 390% circulating load, roughly 80% of the mill feed is material that has already passed the target size and been rejected by the cyclones. Additional grinding energy therefore reports preferentially to the finest particles, which is exactly how slimes are made. Raising charge and speed in this configuration would likely deepen the flotation problem while buying little throughput.
Quantify the risk. Mill power scales roughly with charge volume to the power ~0.5 and with speed to ~1.0, so 30→35% charge and 75→78% critical lifts draw to roughly 6.2–6.4 MW - 95–98% of nameplate. That leaves no margin for voltage dips, liner wear-induced charge slumping, or bearing temperature excursions, and it invites motor trips and gearbox fatigue.
Correct sequence. Fix the classification first, then re-survey. If, after dilution and cyclone repair, the circuit reaches design tonnage at P80 150 µm with the mill pinned at 5.6 MW, the mill has become power-limited and the charge increase is justified - then re-check motor service factor, drive torque, liner and trommel capacity before committing.
A defensible framing for the superintendent: "more energy per tonne of new feed is a valid lever only when the classification is efficient. Right now we are spending energy on material that is already fine."
3. What data would you collect in the next 48 hours to confirm the diagnosis, and what effect do you predict from redirecting pebble crusher product to the cyclone feed?
Read the worked answer
Survey and sampling (48 hours).
- Full circuit survey with timed samples of SAG discharge, pebble crusher feed and product, ball mill discharge, cyclone feed, overflow and underflow - each with PSD, % solids and Cu assay - then close the mass balance. Do not rely on the plant PSD alone; an unclosed balance is the usual reason these problems persist.
- Calculate classification efficiency directly: the proportion of minus 150 µm solids misreporting to the underflow. Expect a high number (well above 30%), confirming bypass.
- Physical cyclone inspection: vortex finder and apex diameters versus drawing, internal wear, blockages, and a per-cyclone flow split using a tracer or individual launder checks.
- Ball mill charge volume (camera or mill weigh system), power, and liner condition.
- Flotation feed size-by-size Cu assay and liberation to verify that the recovery loss tracks the minus 10 µm fraction rather than reagent or pH drift.
Predicted effect of rerouting pebble crusher product. Sending 90 t/h of 16 mm product to the cyclone feed removes it from the SAG recirculating load, which should relieve critical-size pebble accumulation and let SAG throughput rise. The crusher product is classified immediately instead of passing through the SAG mill and the ball mill, so the slime generation per tonne of new feed falls. Watch two side effects: cyclone feed density will shift (re-balance the dilution water), and the pebble crusher will see a slightly different duty cycle. Expect circulating load to fall measurably within one shift and overflow P80 to respond after the classification change takes effect.