Hammer mills may be the wrong tool for coarse gold in Suriname’s saprolite ores

A detailed field audit of two gold processing circuits at an artisanal mining operation in Suriname has found that a mechanized hammer mill and centrifuge system performed no better than a traditional sluice setup, and that both circuits lost large shares of their gold through basic operational errors rather than any inherent equipment weakness. The case study, conducted at the Antino operation in South-East Suriname within the Marowijne Greenstone Belt, compared a sluice-based circuit known locally as the IST system with a hammer mill-centrifuge circuit designated HS, both fed with the same run-of-mine saprolite ore during a six-day monitoring campaign.

The central finding concerns the character of the ore itself. Using fire assay data on size-fractionated samples, the researchers found that roughly 80 percent of the gold in the feed sat in coarse fractions larger than 100 micrometres: 80.2 percent for the IST feed sample and 77.3 percent for the HS feed sample. In several coarse fractions, a tiny share of the retained mass carried a disproportionate share of the gold value, a signature of discrete coarse nuggets rather than finely disseminated grains. This size distribution suggests the ore is naturally amenable to simple gravity recovery, since coarse, dense particles settle readily in sluices and centrifuges without any grinding at all.

That geological picture directly challenges a widespread assumption in artisanal and small-scale gold mining, where mechanized comminution is often adopted on the belief that applying mechanical energy universally improves gold liberation and recovery. The study’s authors argue that comminution should only be undertaken to the liberation size required by the downstream separation process, and that for a soft, clay-rich saprolite hosting coarse gold, intensive milling may be a metallurgical mismatch. Their deportment data gave the hammer mill little credit: while it did reduce the coarsest particles above 1000 micrometres, an overwhelming 69.9 percent of the gold lost by the HS circuit was concentrated in mid-size fractions between 250 and 1000 micrometres, suggesting the mill fractured the ore without fully detaching the gold from its quartz matrix.

The audit also raised a subtler risk. Drawing on established comminution literature, the authors suggest the mill may have acted as a nugget flattener rather than a grinder. Gold is malleable, and the impact action of hammer mills can flatten particles into discs or coins rather than shattering them. Such flattened particles have a high surface area relative to their mass, settle poorly, and are notoriously difficult for centrifugal concentrators to capture, a phenomenon described in the literature as the sail effect. The study notes this mechanism remains a plausible hypothesis pending morphological confirmation, and recommends optical microscopy measurements of particle shape factors in future campaigns to test it definitively.

Operational data revealed why both circuits underperformed in the field. The HS circuit’s H3 hammer mill, rated at 3 tonnes per hour, was consistently run above its design limit, with average feed rates of 3.4 to 4.8 tonnes per hour and a Day 6 load of 4.11 tonnes per hour, roughly 37 percent above capacity. In this choked condition, material is forced through the discharge screens by water flow rather than by comminution. At the same time, the feed to the HS centrifuge ran at only 11 to 17 percent solids by weight, far below the 30 to 50 percent typically required for efficient enhanced gravity concentration, meaning the slurry carried roughly three times the water needed for optimal separation. The extreme dilution was not accidental: operators flushed the overloaded system with massive water volumes to prevent clogging, creating surge velocities that swept gold particles into the tailings regardless of the centrifuge’s separating force.

The traditional IST circuit failed in the opposite direction. After a cautious startup on Day 1 at about 15 percent solids, operators ramped up throughput and the sluice ran at 30.9 to 77.3 percent solids, well above the 10 to 20 percent recommended for fine gold recovery in sluice boxes. The resulting viscous flow hindered gold settlement and caused hydrodynamic carry-over, dragging coarse particles over the discharge lip. Feed rates swung wildly from 11.6 to 55.8 tonnes per hour, driven by unmetered excavator feeding. On Day 3, the calculated proportion of gold lost to tailings, 66.8 percent, was double the proportion recovered.

Because coarse-gold deposits suffer from severe fundamental sampling error, where rare nuggets randomly miss small grab samples, direct daily mass balances frequently produced negative results, with assayed tailing grades exceeding assayed feed grades. The researchers retained these anomalous days to document the sampling volatility and instead used a reconstituted head grade methodology, comparing physically verified gold production against calculated tailings losses, as their primary performance metric. On positive balance days, the combined IST circuit achieved a calculated recovery of approximately 46.7 percent, with a broader range of 36.2 to 65.9 percent, while the HS circuit lost between 42.9 and 60.1 percent of its gold across verified days, equivalent to recoveries of roughly 40 to 57 percent. Despite its higher capital and operating costs, the mechanized circuit thus failed to consistently outperform the simple sluice.

The losses were economically striking. Average tailings grades were 2.60 grams per tonne of gold for the IST circuit and 2.20 grams per tonne for the HS circuit, grades higher than the run-of-mine feed of many large-scale open-pit mines. Using a first-order financial model at the September 2019 gold price of about US$1,510 per ounce, and assuming a theoretical full recovery of gold measured in tailings, the authors projected annual unrealized value of roughly US$6.9 million for the IST circuit and US$3.8 million for the HS circuit. Scaled to 2026 prices of about US$4,710 per ounce, those projections climb to over US$21.6 million and US$11.9 million respectively. Historical gold room amalgamation tailings from the same facility averaged about 48.8 grams per tonne, with fine fractions below 63 micrometres reaching nearly 100 grams per tonne, confirming that manual amalgamation is inherently inefficient at capturing fine gold.

The authors are careful to frame their conclusions as site-specific rather than universal. The campaign ran from September 2019 to February 2020 and was disrupted by the COVID-19 pandemic, the fractional analysis rested on only a handful of mathematically closed sample pairs, and no direct mineralogical liberation analysis was performed, so the coarse gold cannot be formally classified as fully gravity recoverable. External gravity recoverable gold testing at the Antino project by Founders Metals, reporting maximum recoveries of 62 percent on fresh rock composites, lends support to the picture of a gravity-amenable ore, and prior regional studies found over 70 percent of gold lost by mill-centrifuge circuits was still in coarse, easily recoverable size ranges.

The practical recommendations follow directly from the diagnosis. For these coarse-gold saprolite horizons, the authors suggest the hammer mill is an unnecessary bottleneck that could ideally be bypassed or replaced with a rotary scrubber, while strict density control, roughly 15 to 25 percent solids for sluices and 30 to 45 percent for centrifuges, would address the immediate failure mechanisms. More broadly, they argue that standardized on-site ore characterization before equipment purchase, accessible sampling nodes built into plant layouts, and daily mass balance audits matter more for profitability than technological sophistication. A robust sluice running 22 hours a day, they note, can yield a better net present value than a high-tech mill running only 16 hours, particularly when reactive maintenance and absent process monitoring quietly erode the margin at every stage of the value chain.

Subject of Research: Field performance comparison of sluice-based and hammer mill-centrifuge gravity circuits processing coarse-gold saprolite ore at an artisanal gold mining operation in Suriname

Article Title: Field performance evaluation of comminution and gravity processing circuits for artisanal saprolite gold ores

Article References: Druiventak, A., & Chen Poun Joe, K. (2026). Field performance evaluation of comminution and gravity processing circuits for artisanal saprolite gold ores. Discover Metals, 59(1), Article 5. https://doi.org/10.1007/s13404-026-00374-4

Image Credits: AI Generated

DOI: 10.1007/s13404-026-00374-4

Keywords: artisanal and small-scale gold mining, gravity concentration, hammer mills, saprolite ore, Suriname, gold recovery, centrifugal concentrators, sluice boxes, nugget effect, comminution, tailings losses, process control