The quiet revolution inside grinding circuits: smarter cyclones, real‑time sizing

Across modern concentrators, closed‑circuit ball mills are almost universally paired with hydrocyclone classification (hydrocyclone: a centrifugal separator that splits slurry by size and density). The aim is simple: route fine particles to the next stage and send coarse ones back to the mill. The reality is messier. As one study put it, “hydrocyclones are relatively simple to operate, [yet] their classification performance is generally low, leading to significant fines recirculation… consequently overgrinding the product” (mdpi.com) (mdpi.com). Misplacement happens both ways, with fines short‑circuiting into the underflow and some coarse particles slipping into overflow (mdpi.com).

The cost is tangible. Reclassifying with high‑frequency screens instead of conventional cyclones in one zinc operation delivered a 13% increase in throughput and a 15% reduction in grinding energy (mdpi.com).

Hydrocyclone operating parameters

Operators tune the “cut” using a set of geometric and operating levers to control d50 (the size reporting 50% to underflow) and P80 (the product size at which 80% passes). Each parameter shifts the separation point and its sharpness.

Feed pressure/flow. Raising feed pressure (or pump speed) intensifies the centrifugal field and typically reduces the cut size, delivering a finer overflow. A practical guideline is 49–157 kPa (≈7–23 psi), with “increasing the feed pressure” known to “reduce the classifying granularity and increase the processing capacity” (xinhaimining.com) (xinhaimining.com). McLanahan’s plant guidance echoes the trend—higher pressure yields finer separation, and lowering pressure coarsens the cut (mclanahan.com) (mclanahan.com). The trade‑off is higher power draw and cyclone wear at high pressure (xinhaimining.com).

Diameter and number of cyclones. Larger units carry more throughput but deliver a coarser overflow; many smaller cyclones in parallel can maintain capacity with a finer cut—e.g., several 305–406 mm (12–16″) cyclones in parallel instead of a single 610 mm (24″) unit when targeting fine classification (xinhaimining.com).

Cylinder height. Increasing the height (retention time) tends to produce a finer cut; a cylinder height of roughly 0.6–1.0× its diameter is advised (xinhaimining.com). Longer residence “reduce[s] the separation granularity and improve[s] separation efficiency” (xinhaimining.com).

Cone angle. A smaller cone angle (a longer, narrower cone) slows the upward flow and yields finer overflow; larger angles (30–60°) coarsen the separation, while smaller angles (15–30°) suit desliming and fine cuts (xinhaimining.com).

Vortex finder geometry. The overflow pipe (vortex finder) diameter and insertion depth affect fines short‑circuiting and coarse carryover. Placing it roughly 50–80% of the cylinder height is recommended; too shallow increases “short‑circuit flow rate” and “coarse grain content in overflow,” too deep “reduces the classification efficiency” (xinhaimining.com) (xinhaimining.com).

Spigot (apex) diameter. The underflow aperture sets the flow split and density. Too small and the cyclone ropes; too large and “a lot more water – and consequently, more fines – will come out the underflow,” shifting the cut point (mclanahan.com) (mclanahan.com). To “reduce bypass material” (fines in underflow), operators typically reduce the aperture; if underflow ropes, they “install a larger apex” (mclanahan.com).

Feed solids and density. Denser ore (at a fixed solids%) promotes a finer cut; high solids concentration (viscous “mud”) coarsens it. Typical operation is 50–65% solids by weight. Entrained air or overly dilute feed worsens fines bypass and can trigger roping events (xinhaimining.com) (hydrocyclones.net) (xinhaimining.com).

In practice, multi‑cyclone banks with carefully balanced distributors and separate pumps keep each unit on point; operators may, for example, raise pressure and/or adjust vortex geometry to target a finer d50 when they need overflow at <106 μm, or coarsen the cut by reducing pressure or tightening the spigot if coarse shows in the overflow (mclanahan.com).

Coarse/fine split strategies

Beyond the tuning knobs, plant‑level strategies reinforce the split: secondary classifiers can re‑mill and reclassify underflow, and hybrid circuits are documented to sharpen the effective separation compared to single‑stage setups (mdpi.com).

Recirculating load is another lever. Coarsening the target P80 can lift throughput, but pushing too far invites roping. Studies show that optimal throughput often comes from relaxing grind just short of that constraint, with APC (advanced process control) managing the boundary (scielo.org.za) (scielo.org.za).

As one source summarized, “the primary objective is to maximize fines reporting to the overflow and coarses reporting to the underflow,” with spigot and feed conditions chosen accordingly (hydrocyclones.net). Reducing apex size curbs fines bypass (mclanahan.com), while CFD (computational fluid dynamics) shows that a longer vortex finder can raise the split water ratio and improve separation efficiency (mdpi.com).

Online particle‑size analysis

Ore variability can overwhelm even well‑tuned cyclones. That’s why mills are fitting online particle‑size analyzers (PSD: particle size distribution) that measure slurry PSD in real time and feed it to the control room, typically via DCS/APC (distributed control system/advanced process control). Laser diffraction units like Malvern Panalytical’s Insitec Wet cover 0.1–2500 μm continuously and are designed for 24/7 use (metso.com) (malvernpanalytical.com). Older ultrasonic attenuation units exist but are sensitive to air bubbles and viscosity changes and demand “time and labor consuming calibrations” (metso.com).

Image‑based systems also play a role. On‑belt analyzers can track crusher feed size and flag oversized pebbles before they choke downstream stages—one Metso case in Kiruna used an on‑conveyor analyzer to spot >400 mm pebbles (metso.com). Vision and laser‑camera units in the classification loop can process thousands of particle images per minute and compute PSD for control (advancedminings.com).

A third approach is acoustic sensing. CiDRA Minerals Processing’s CYCLONEtrac PST (particle size tracking) inserts a probe into each cyclone overflow; particle impacts generate acoustic signals that correlate with PSD. The PST offers near‑continuous overflow PSD without manual sampling and has demonstrated >99% uptime in copper plants (scielo.org.za) (scielo.org.za).

The value is in the feedback loop. Instead of waiting hours for grab‑sample assays, operators see actual overflow P80s and the full PSD in real time and adjust mill load, water addition, or cyclone pressure almost immediately. For many grinding APC strategies, PSD is the critical feedback variable (metso.com).

The same data protects the circuit from classification limits. One case study showed that as throughput rose, a single cyclone roped and dumped coarse material to flotation—“one wrong cyclone corrupts the whole battery of cyclones.” With PSD online (in that instance, PST data was available but not yet in automatic control), engineers could see how close they were to the roping boundary and, with APC, “relax” the grind (raise P80) to increase throughput without tripping the battery (scielo.org.za) (scielo.org.za).

Case data and economics

The screening example—13% throughput gain and 15% energy reduction—illustrates how classification drives profitability (mdpi.com). Other work indicates that operating cyclones ~15–20% above design load (coarser P80) can raise mill output, but only with real‑time sensing to maintain efficiency and avoid roping (scielo.org.za) (scielo.org.za). Published methodologies predict that each 1% relaxation in P80 can translate into a significant increase in net metal production (by Bond’s equation), assuming PSD is in closed‑loop control (scielo.org.za).

Online analyzers also improve consistency. By “line‑scanning thousands of particles per minute,” they steady product size despite daily shifts in feed hardness and grade (metso.com). In short, they turn grinder/classifier circuits from “feet‑on‑the‑bridge” oversight to tightly controlled processes, ensuring that fines go to flotation and coarse to regrind as intended.

Reference notes

Core sources quantify hydrocyclone inefficiencies and the upside from better classification (mdpi.com) (mdpi.com) and discuss improving separation via geometry and operating setpoints (xinhaimining.com) (xinhaimining.com). Practical control insights and case studies come from McLanahan, Metso, and CiDRA (mclanahan.com) (mclanahan.com) (metso.com) (scielo.org.za). CFD work demonstrates that extending the vortex finder can increase the split water ratio and improve separation efficiency (mdpi.com). The zinc‑circuit result—13% throughput and 15% energy—comes from high‑frequency screen classification (mdpi.com).