Ore sorting has been “the next big thing” in mining for about fifteen years. For most of that time, it was a technology that worked brilliantly in controlled tests and struggled in production environments. Dust, throughput limitations, variable feed characteristics, and maintenance demands kept it on the fringes.

That’s changed. Over the past two to three years, sensor-based ore sorting has moved into sustained production at enough operations to say with confidence: it works. Not everywhere, and not for every ore type. But where the conditions are right, the results are compelling.

What Ore Sorting Actually Does

For anyone who hasn’t followed the technology closely, here’s the short version. Sensor-based ore sorting uses various detection methods—X-ray transmission, near-infrared spectroscopy, laser-induced breakdown spectroscopy (LIBS), or combinations of these—to classify individual rocks on a conveyor belt. Rocks that meet grade criteria pass through. Rocks that don’t get ejected by air jets or mechanical diverters.

The goal is simple: reject waste rock before it enters the processing plant. This concentrates the feed, reduces the volume of material that needs to be crushed, ground, and processed, and cuts energy and reagent consumption.

Where It’s Working in Australia

Gold operations in Western Australia. Several gold mines in the Eastern Goldfields have deployed ore sorting to reject barren or sub-economic material from ROM feed. Gold-bearing quartz veins hosted in mafic or ultramafic wallrock create good contrast for dual-energy X-ray transmission (DE-XRT) sorting. Operations are reporting waste rejection rates of 15-25% while recovering over 95% of contained gold.

Base metals in Queensland. A number of copper-gold operations in the Mount Isa and Cloncurry districts are using ore sorting to pre-concentrate sulphide mineralisation. The density contrast between sulphide-bearing and barren material works well with XRT sensors. One operation reported a 20% reduction in mill feed tonnage while maintaining metal recovery above 97%.

Rare earth projects. Several of Australia’s emerging rare earth operations are evaluating ore sorting as a front-end concentration step. The high processing costs for rare earth ores make even modest waste rejection economically significant.

What Changed to Make It Work

Three things converged to push ore sorting past the tipping point:

Better sensors. The detection systems available in 2026 are dramatically more capable than what was on the market five years ago. LIBS sensors can now provide real-time elemental analysis at belt speeds of 3-4 metres per second. Multi-sensor systems that combine XRT with optical and NIR detection cover a wider range of ore characteristics.

Improved air ejection systems. The mechanical side of ore sorting—actually separating the rocks after classification—was always a bottleneck. Modern high-frequency valve arrays can handle smaller particle sizes and higher throughputs than earlier systems. This has expanded the range of applications from coarse (+50mm) material down to 10-15mm, which captures a much larger proportion of total ROM feed.

Machine learning classification models. Early ore sorters used fixed threshold rules—if the density measurement exceeds X, keep the rock. Modern systems train on site-specific geological data and continuously refine their classification models during operation. They adapt to changes in ore characteristics that would have confused older systems.

The Economics

The business case for ore sorting depends on three main factors: the proportion of waste that can be rejected, the cost of processing that waste through the plant, and the capital and operating cost of the sorting system itself.

For a typical gold operation processing 3 million tonnes per year with a 20% waste rejection rate, the numbers look roughly like this:

• Processing cost saving: $6-8 per tonne of rejected waste, totalling $3.6-4.8 million annually
• Reduced energy consumption: 15-20% lower grinding energy, worth $1-2 million per year
• Extended mine life: by processing higher-grade feed, marginal ore that was previously uneconomic can enter the plan
• Sorting system capital cost: $8-15 million for a 500-800 tph installation
• Sorting system operating cost: $0.50-1.00 per tonne sorted

Payback periods of two to three years are common for well-suited applications. Indirect benefits—reduced tailings generation, lower water consumption, smaller plant footprint—add further value that’s increasingly relevant under ESG reporting.

Where It Doesn’t Work (Yet)

Ore sorting struggles with fine-grained disseminated mineralisation where there’s minimal contrast between ore and waste at the particle level. Many porphyry copper deposits fall into this category. At particle sizes below 10mm, current sensor and ejection technology loses effectiveness.

It also requires relatively consistent feed sizing. ROM material with a wide size distribution needs screening before sorting, adding cost and complexity.

What to Watch

The next frontier is integrating ore sorting data back into mine planning. Every rock that passes through a sorter is essentially being assayed in real time. That data stream, if captured and fed into geological models, creates a continuous grade control feedback loop that’s far more detailed than traditional sampling programs.

Mines that figure out how to close that loop—using sorting data to update block models and adjust extraction sequences—will extract more value from ore sorting than those treating it purely as a processing step.

The technology has arrived. The integration opportunity is what makes it genuinely interesting.