What Perth’s New Pilot Plant Can and Cannot De-Risk for Miners

A clean recovery curve from a laboratory test feels like proof. It often is not. Bench work on a few kilograms of ore can look excellent and still miss what happens when tonnes move continuously through a real circuit. That gap is the reason a new 5 tph pilot system, housed in a facility of more than 2,000 m² in Kelmscott, Perth, drew attention in the September 2026 edition of Australian Mining.

Mineral Technologies (MT) and Nagrom have built a single-site route from bench testing to pilot processing. The open question for explorers and investors is how much uncertainty that route actually removes.

The answer is useful but bounded. You will come away knowing what the facility can and cannot de-risk, and which questions to put to a developer when the next set of pilot results lands.

Why does moving from bench to pilot scale change the risk picture?

Most readers treat a laboratory result as the hard part done. The mechanics suggest otherwise, because each stage of testing proves something different.

1. Bench testing: small batch tests that show whether a mineral can be separated at all, but not how a full circuit behaves.
2. Pilot testing: continuous running on bulk samples that reveals how equipment interacts over time, but still on selected ore under close supervision.
3. Full-scale operation: the only stage that proves performance across years of variable feed, wear and real operating discipline.

The gaps open between those stages. A recirculating load, meaning material that loops back through part of a circuit for reprocessing, only builds up once a plant runs continuously. Steady state, the point where flows and loads stabilise, is rarely reached in a short batch test.

Recirculating loads are a reminder that interconnected circuit design means a change in one unit can ripple through classification, milling and separation, which batch tests rarely capture.

What single-stage tests miss

• Recirculating loads: loops in milling, classification and dense media circuits never fully develop.
• Simplified configurations: screens, desliming and recycle streams are often left out, hiding bottlenecks.
• Homogeneous samples: well-mixed bench material masks the clay, moisture and grade swings a plant faces.

MT describes Kelmscott as adding “another link” between existing bench and pilot capabilities, with a commercial focus on mineral sands, iron ore, silica sand, chromite and wolframite. Nagrom already offers sample management, characterisation, feed preparation, comminution (crushing and grinding) and physical, flotation and hydrometallurgical separation, from bench scale up to pilots above 100 t.

Commentary commonly attributed to bodies such as AusIMM, CSIRO and Austmine holds that pilots better constrain recovery, throughput, energy use and water balance. Treat that as general industry reasoning rather than verified evidence. Keeping every stage on one site shortens the handover between them, which may cut lost time and sample handling errors.

If you are weighing a developer’s flowsheet, this tells you that results from continuous, larger-scale running carry more weight than bench curves alone. The testing path matters as much as the headline recovery figure.

What does the Kelmscott rig line-up actually cover?

That principle only helps if the equipment resembles what a real plant would run. Kelmscott’s three main rigs are worth taking one at a time.

The spiral rig is the core of the new capability. Spirals are gravity separators that use a curved channel to split heavier minerals from lighter ones, and this rig can run in full open circuit across several stages, including MT’s flagship MG12 spiral.

Alongside it sits a four-pole WHIMS, or wet high intensity magnetic separator, which pulls weakly magnetic minerals out of a slurry. MT positions it as a scalable platform for mineral sands, iron ore, silica sand, chromite and wolframite.

The third rig handles dense media separation (DMS), where ore is placed in a heavy liquid so lighter particles float and denser ones sink. MT describes it as among the largest of its type, and this is where the detail starts to matter.

Why cyclone size matters MT director project development Dan Webber explained that larger DMS cyclones can accept larger feed. Smaller systems must exclude or crush coarse material, which skews results and weakens scale-up accuracy.

That point is the quiet centre of the whole facility. General technical reasoning suggests small cyclones behave differently from plant-scale units in flow pattern and wall effects, and spirals and WHIMS can also show scale effects. Feed tested finer than a plant will treat may overstate separation efficiency.

Dense medium cyclones behave according to fluid dynamics that shift with cyclone diameter and feed size, which is the underlying reason scale-up accuracy depends so heavily on how large the test unit is.

The 32 mm maximum feed size and 5 tph capacity are therefore both strengths and boundaries. Webber says recent projects have let MT move into coarse hard rock such as iron ore, manganese and cassiterite, and that the site holds ISO9001 accreditation plus safety, management and environmental certifications he links to ESG metrics commonly used in projects.

This tells you to check whether a project’s planned plant feed size and equipment class fall inside what a pilot rig could test. A mismatch weakens the value of the results.

Where do pilots still fall short, and what do the case studies show?

Matching equipment to ore is progress. It is not a guarantee, and the counter-evidence deserves equal weight.

Practitioner commentary suggests pilots often run on carefully selected samples under closer supervision than a plant can sustain. Long-term problems such as media loss, scale formation and tailings behaviour may take months or years to emerge, well beyond a pilot campaign.

• What a pilot can reduce: uncertainty over circuit design, equipment sizing, recovery under continuous running and product specification.
• What it cannot: sample representativeness across the deposit, long-term wear and media loss, workforce and logistics, funding and permitting.

Cost is the other trade-off. Money spent on a pilot can divert funds from drilling and geometallurgy, the study of how ore properties vary through a deposit and affect processing.

Pilots run on selected samples can understate ore variability, and geometallurgical models that map how recovery shifts across a deposit are the main tool for testing whether a pilot result generalises.

What the case studies suggest

Thunderbird, a heavy mineral sands operation producing leucoxene, ilmenite and zircon concentrates according to the USGS 2024 review, is reported to have used pilot work to reduce process risk while execution risk remained. Karara and Sino Iron are often cited as projects where substantial testwork still preceded throughput and reliability challenges. These accounts are unverified and should be read as illustrations.

The shared lesson in that commentary is that conventional flowsheets tied to geometallurgical models tend to fare better. No published scale-up failure rates or pilot cost benchmarks for Australian projects since 2024 were found, so the evidence base remains largely qualitative.

You should read any pilot announcement as evidence of process viability only. Look separately for sample representativeness, independent review and funding clarity.

What does this mean for Western Australian explorers and investors?

The facility sits beside a busy WA pipeline, though none of the following projects has been linked to Kelmscott. Tungsten Mining’s Mt Mulgine project spans three granted mining leases about 350 km north-north-east of Perth, according to Austrade’s February 2026 prospectus.

In iron ore, BHP approved its $1.4 billion Western Ridge Crusher in February 2024, and Mineral Resources’ $3 billion Onslow Iron shipped first ore in May 2024. The REMP 2024 report lists 35 iron ore projects yet to reach the committed stage, while REMP 2025 highlights the Mid-West Green Iron Project and the NeoSmelt Pilot Plant.

MT brings more than 80 years of experience and reports over 200 million tonnes processed through MT-designed plants in the most recent year. That is credibility context for the operator, not a promise about any client’s project.

Necessary, not sufficient General market commentary, unverified, holds that credible pilot de-risking supports feasibility studies and offtake talks, but investors still look for independent reviews, sound cost estimates and a clear financing path.

Before trusting the next pilot headline, ask:

1. Was the bulk sample representative of the whole deposit, including low-grade and high-clay zones?
2. Does the tested feed size match the planned plant feed?
3. How long did the pilot run, and did it reach steady state?
4. Has an independent party reviewed the results?
5. Is there a funded path from pilot to final investment decision?

For you as an investor or explorer, the facility lowers one specific kind of risk, process scale-up. Resource, funding, permitting and execution risk stay where they were.

Investors weighing resource risk separately from process risk can use our full explainer on completing a mineral resource estimate, which walks through the five steps that turn drilling data into quantified resources.

Reading pilot announcements with clearer eyes

Kelmscott shortens the route from laboratory to plant and lets developers test on equipment closer to commercial scale, particularly through its large DMS rig and 32 mm feed limit. That addresses one layer of project risk, and only one.

The sensible call is to weigh pilot data alongside sample quality, feed-size fit and funding clarity rather than treating a pilot milestone as the finish line. Over the coming year, watch which developers publish results from the facility, and how those results compare with plant performance once projects are built.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions. Statements about future project performance are speculative and subject to change based on market developments and company performance.