Real-Time Floc Monitoring Optimizes Dewatering of Fine Copper Tailings

A laboratory study of fine copper tailings from the Akka mine in southern Morocco found that pairing anionic polyacrylamide flocculant selection with real-time focused beam reflectance measurement can identify operating conditions that improve settling, underflow density and recycled-water clarity at the same time. Response surface models identified the high-molecular-weight CA39 polymer as the best of four candidates, with a predicted optimum of 26.46 grams per tonne of dry solids at 6.70 wt.% feed solids. Triplicate confirmation tests measured an average settling velocity of 27.15 metres per hour, underflow solids content of 45.62 wt.% and supernatant turbidity of 56.1 nephelometric turbidity units, all within 1.5% of model predictions. These laboratory results offer a testable starting point for pilot-scale tailings thickening rather than a directly transferable industrial set point.

The tailings presented a difficult dewatering challenge. Laser diffraction showed a median particle size of 27.9 micrometres, while 21% of the material was smaller than 5 micrometres, 44% was below 20 micrometres and 61% was below 40 micrometres. Their specific gravity was 2.86 grams per cubic centimetre and their Brunauer–Emmett–Teller specific surface area was 7.04 square metres per gram. X-ray diffraction and automated mineral analysis identified quartz, feldspar and carbonate as the main components, with approximately 19.68 wt.% clay, chiefly chamosite and clinochlore with minor kaolinite. Most clay occurred as liberated fine particles or in association with feldspar, quartz and carbonate, a structure expected to impede dispersion control and sediment consolidation. The sample also contained 50.4 wt.% silica, 11.11 wt.% alumina, 8.36 wt.% magnesium oxide, 6.94 wt.% calcium oxide and 6.84 wt.% ferric oxide.

Initial screening showed why both polymer chemistry and slurry conditions had to be considered together. At 30 wt.% solids, a dosage of 30 grams per tonne produced no appreciable settling improvement, because the dense particle network restricted floc mobility and masked polymer performance. Screening was therefore conducted at 10 wt.% solids and a natural pH of 8–9, with dosages of 15 and 30 grams per tonne. Anionic polyacrylamides outperformed polyethylene oxide and the cationic polymer polydiallyldimethylammonium chloride. Among anionic products with similar molecular weights, increasing charge density generally improved aggregation: at 10 wt.% solids, residual turbidity was 21.36, 30, 26.07 and 15.7 units for polymers with 0%, 6%, 18% and 50% charge density, respectively. Underflow solids ranged from 39.95 to 44.55 wt.% and water recovery from 79% to 87% across the tested polymers, compared with 27.34 wt.% and 73% without flocculant.

Focused beam reflectance measurement, or FBRM, captured changes that endpoint settling tests could not reveal. The instrument tracked particle counts and mean square-weighted chord length through polymer addition, shear and reflocculation. Three derived indices described initial aggregation, resistance to breakage and recovery after shear. At 5 wt.% solids, AX21 and CA39 generated mean square-weighted chord lengths above 300 micrometres at the highest dosage, while CAR26 and M345 rarely exceeded 250 micrometres. Raising solids content to 12.5 wt.% constrained growth: AX21 and CA39 produced aggregates of about 250 micrometres at 30 grams per tonne, compared with more than 300 micrometres at 5 wt.% solids. CA39 combined large flocs with high shear resistance, whereas the highly charged M345 formed smaller, more compact flocs and achieved the highest reflocculation index at 5 wt.% solids. These differences show that maximum size alone is not an adequate measure of process performance.

For CA39, empirical regressions linked FBRM descriptors to settling velocity across the tested range. Mean square-weighted chord length explained the change in settling velocity with an R-squared of 1.000 at 5 wt.% solids and 0.9996 at 12.5 wt.%, while the flocculation index produced R-squared values of 0.994 and 0.997. The minimum target of 16 metres per hour corresponded to a chord length of about 243 micrometres and a flocculation index of about 9.2 at 5 wt.% solids. At 12.5 wt.%, the corresponding requirements rose to approximately 296 micrometres and 12.0. Clarification followed a different pattern: turbidity reduction was associated more closely with the decline in dispersed particle count during conditioning than with the final aggregate size, indicating that capture of fine particles and rapid growth are related but distinct processes. The FBRM thresholds remain empirical because stirred flocs are not identical to those formed during quiescent settling.

Slurry chemistry further controlled the balance between aggregation and floc stability. AX21 and CA39 were tested at pH 5, 9 and 13. Moderately alkaline conditions produced the largest and most persistent flocs, while acidic conditions caused weak, short-lived aggregation. At 5 wt.% solids, AX21’s flocculation index rose from 3.64 at pH 5 to 13.59 at pH 9 before declining to 11.86 at pH 13. CA39’s index increased from 1.54 at pH 5 to 11.13 at pH 9 and 12.28 at pH 13. At 12.5 wt.% solids, both polymers declined at pH 13. The natural tailings pH of 8–9 therefore already occupied the favorable operating range. Reflocculation indices required cautious interpretation: high values at pH 5 partly reflected very little initial aggregation and therefore a small mathematical denominator, rather than strong recovery of substantial flocs.

Central composite response surface modelling examined feed solids content, dosage and four anionic polyacrylamides. The 44-run design generated statistically significant models for settling velocity, turbidity and underflow solids, with predicted R-squared values of 0.982, 0.998 and 0.967 and no significant lack of fit. Three untreated controls and replicated centre points supported estimation of experimental error. Solids content had the largest influence on settling velocity, followed by dosage and their interaction. With CA39 at 30 grams per tonne, settling velocity reached 38.95 metres per hour at 5 wt.% solids, fell to 15.35 metres per hour at 12.5 wt.% and dropped to 2.04 metres per hour at 20 wt.% as crowding, hindered settling and less favorable floc structure became dominant. Turbidity generally improved with higher solids content and dosage, whereas underflow density depended on their interaction: excess polymer reduced compaction at low solids content but increased it at higher concentration.

Desirability-based optimization imposed process limits of at least 16 metres per hour for settling, at least 40 wt.% underflow solids and no more than 70 units of turbidity, while minimizing dosage. CA39 produced the highest overall desirability score of 0.343 at 6.70 wt.% solids and 26.46 grams per tonne, with predicted responses of 26.98 metres per hour, 45.88 wt.% underflow solids and 55.3 units of turbidity. M345 and AX21 also had feasible solutions, but both required higher dosages and produced inferior combinations of responses; CAR26 had no feasible solution. Independent triplicate tests at the selected CA39 condition gave relative errors of 0.63% for settling velocity, 1.45% for turbidity and 0.57% for underflow solids. The water discharged with underflow fell from 2.33 to 1.19 cubic metres per tonne of dry solids, a reduction of about 49%.

Mechanistic measurements indicated that the anionic polymers adsorbed mainly through physical bridging rather than charge neutralization or covalent attachment. Untreated particles were negatively charged across the measured pH range, with zeta potential shifting from about −3.4 millivolts at pH 4 to −41.3 millivolts at pH 10. Polymer conditioning made the measured potential still more negative; for example, M345 shifted it to approximately −50 millivolts at pH 10. Because charge reversal did not occur and the most negative potential did not correspond to the fastest settling, global charge neutralization was not the dominant process. Fourier-transform infrared spectra showed an approximately 58% increase in the 1,000–1,100 reciprocal-centimetre silicate band and about a 60% increase in lower-frequency metal–oxygen features without measurable peak shifts, supporting adsorption onto quartz and clay surfaces. At pH 9, extended carboxylate-bearing chains could bridge negatively charged particles, producing large but permeable structures whose settling, strength and water retention depended jointly on molecular weight, architecture and dosage.

The economic scenario, based on 800 tonnes of dry tailings per day, estimated daily CA39 consumption of about 21.2 kilograms and annual polymer expenditure of €23,179 to €38,632 at €3–5 per kilogram. At an assumed process-water cost of €0.30 per cubic metre, the modeled water saving was worth about €99,979 per year, producing a net annual benefit of €76,800 to €61,347. These figures exclude possible savings in pumping and storage, but they also do not include the effects of residual polymer on flotation or the long-term fate of acrylamide. The experiments used one composite tailings sample re-dispersed in filtered tap water, batch settling, four polymers, a natural pH of 8–9 and 25 °C. Continuous thickening in actual process water remains necessary, along as with underflow rheology, recycled-water flotation tests and environmental monitoring before the laboratory optimum can be deployed at an operating mine.

Subject of Research: Optimization of anionic polyacrylamide flocculation for dewatering fine copper tailings

Article Title: FBRM-Guided Optimization of Flocculation for Sustainable Dewatering and Water Recovery from Fine Copper Tailings.

Article References: Hamraoui, L., Taha, Y., Bergani, A., Aboulaich, A., Loutou, M., De La Fuente, E., Blanco, A., Benzaazoua, M., & Khalil, A. (2026). FBRM-Guided Optimization of Flocculation for Sustainable Dewatering and Water Recovery from Fine Copper Tailings.. Cleaner Engineering and Technology, Article 101338. https://doi.org/10.1016/j.clet.2026.101338

Image Credits: AI Generated

DOI: 10.1016/j.clet.2026.101338

Keywords: fine copper tailings, flocculation, anionic polyacrylamide, focused beam reflectance, dewatering, water recovery, response surface methodology, zeta potential, FTIR spectroscopy, mineral processing, FBRM-Guided, Optimization