Sampling in DMS Streams

In dense medium separation (DMS) and other mineral processing circuits, sampling and analysis have traditionally relied on manual sample collection and delayed laboratory results. This approach stunted real-time process optimisation, with process control decisions based on periodic composite samples and operator experience, rather than continuous, representative measurement of stream behaviour.

Modern analyser technologies, big data methods and increased computing power enable real-time insights into grade, density and process variability. However, these systems can only perform to their full potential when they are fed with representative, Theory of Sampling (TOS)-compliant samples. Multotec’s webinar, Sampling in DMS Streams – Seeing Through the Black Slurry, presented by Willem Slabbert, examines how combining metal accounting grade sampling with online analysers (also read at-line, in-line or by-line analysers) enhances measurement accuracy, process control and decision-making in DMS circuits.

Why is sampling in DMS streams so important?

Sampling in DMS streams is critical because all process control, metal accounting and optimisation decisions depend on the quality, representativity, and reproducibility of the data derived from those samples. If the sample is not truly representative of the production stream, even the most advanced analyser or model will generate misleading results when analysing an inaccurate or imprecise sample.

In DMS circuits, accurate sampling underpins:

• Medium control: Representative samples are required to maintain correct medium density and stability within the DMS circuit.
• Grade and recovery tracking: Reliable samples support accurate metal accounting, reconciliation and evaluation of circuit performance.
• Process optimisation: Valid comparisons of operating conditions are only possible when changes in performance are measured using unbiased, consistent samples.

The Fourth Industrial Revolution has brought advanced online analysers, big data algorithms and increased computing power, but these tools can only perform to their full potential when they are fed with Theory of Sampling (TOS)-compliant, metal accounting grade samples. Poor sampling introduces bias and variability that no analyser can correct, while good sampling provides reliable, high-confidence data for process control and decision-making.

What types of online analysers are used in mineral processing and DMS circuits?

A range of online analyser technologies is used in mineral processing and DMS circuits to measure size, grade, density and moisture in real time.

Common analyser types include:

• Photographic (visible) imaging: Used to assess particle size, shape and wear, and often applied in flotation control from central control rooms.
• Spectral imaging: Extends visible imaging into infrared and ultraviolet ranges to identify mineralogical or elemental features not visible.
• Microwave analysers: Widely used for online moisture measurement on conveyors or in product streams.
• X ray fluorescence (XRF): Extensively applied in laboratories and, in some cases, online systems to determine elemental grades.
• Electromagnetic inductance analysers: A newer technology for measuring slurry particle size distribution.
• Prompt gamma neutron activation analysis (PGNAA) and related nuclear methods: Penetrative techniques that can measure elemental composition through the particle, not only at the surface.
• Nuclear and non nuclear density meters: Used for slurry and dense medium density control, including in DMS correct medium circuits.
• Acoustic or sound based monitoring: Applied to infer equipment or process performance from sound signatures.

Each technology has specific strengths and is selected according to the property to be measured and the process requirements.

How does stream variability affect analyser accuracy and sampling quality?

Stream variability directly affects analyser accuracy and sampling quality because most analysers only measure a portion of the stream, while the material itself is highly heterogeneous in three dimensions (height, width and length).

In pipes, slurry velocity and particle distribution differ between the inner and outer sides of bends, so two surface mounted analysers installed at different positions can report different results from the same stream. On conveyor belts, coarse particles tend to concentrate near the top of the burden and towards the centre, while finer material and different mineral populations occur lower down and towards the shoulders. Similar segregation occurs at conveyor head pulleys, where coarse particles travel further and fine material falls close to the pulley, and in stockpiles, where coarse particles migrate to the outside foothold and fines to the centre.

Surface measurement technologies, which only detect the outer layer of the stream, therefore risk reporting biased values if analyser placement does not account for this inherent 3D variability. In some applications, this is addressed by using multiple detectors or sources around the stream or sample bin – as in the Scanmin COALLAB analyser or the Chrysos Photon Assay™, which measures variability across all dimensions of a composite sample. Without representative sampling that captures the full height and width of the stream or sample, both analyser readings and laboratory samples can still systematically misrepresent the true composition of the production flow.

How is DMS sampling data used to evaluate and optimise circuit performance?

DMS sampling data is used to quantify how effectively the circuit separates valuable mineral from waste and to identify where process adjustments are required. Without reliable sampling, cut density, recovery and separation efficiency cannot be evaluated with confidence.

Representative, metal accounting grade samples from streams such as feed, sinks, floats and correct medium are typically used to:

• Calculate mass balances and recoveries, showing how much valuable mineral reports to product versus waste.
• Determine cut density and separation efficiency, using tools such as Tromp curves and Ep-values.
• Monitor medium quality and stability, including density, contamination by non magnetics and medium recovery performance.

This information provides the basis for assessing circuit performance, comparing operating conditions and implementing optimisation initiatives in DMS plants.

What is a TOS-compliant/metal accounting grade sample, and why must analysers be fed with it?

A TOS compliant, metal accounting grade sample is one that has been collected and prepared according to the Theory of Sampling (TOS), so that every particle in the production stream has an equal probability of being selected, preserved and analysed. In practice, this requires correctly designed automated mechanical samplers that cut the full height and width of the stream, followed by unbiased splitting and handling. This approach reduces the stream to a one dimensional variability problem along its length, which can be controlled by taking sample increments frequently enough to meet required precision levels. In dry sampling plants, sample streams are generally conveyed on narrower belts than the main production conveyors, allowing analysers to measure the full belt width and thereby eliminate one dimension of variability in the sample stream.

Such samples are essential because they accurately represent the true composition of the production flow. By contrast, commonly used non compliant devices – such as shark fin, pressure pipe, thief, poppet and wire samplers – do not cut the full cross section of the stream and can contribute up to 70% of total sampling error, while the analysis itself typically contributes only 2 to 5% (the remainder accumulated as preparation or other errors). If analysers are fed with biased or non representative samples, even advanced technologies will produce misleading results.

When fed with TOS compliant, metal accounting grade samples, online analysers can provide reliable data for process control, calibration and metal accounting, rather than attempting to interpret a distorted view of the stream.

How does a RAMA slurry handling system feed online analysers with representative DMS samples?

A RAMA® (Real time Automated Metal Accounting) slurry handling system feeds online analysers with representative samples by integrating proven sampling and splitting equipment into an automated circuit.

Metal accounting grade primary samples are first extracted from DMS slurry streams using proven primary samplers. These samples are then representatively split with Vezin samplers and associated segregation control equipment. One portion is retained as a composite metal accounting sample, while the equally representative reject portion is pumped at a controlled feed rate to the online analyser.

The RAMA system is automatic and self cleaning, prevents cross contamination between streams and consecutive samples, and can recirculate slurry through the analyser for multiple passes within one analysis cycle. This configuration ensures that online analysers receive a continuous, representative feed that meets metal accounting standards, enabling high confidence process control and reporting.

Closing thoughts

The move towards real-time analysis, big data and advanced process control in DMS circuits often starts with a focus on analyser technology. The real step change comes when plants place sampling quality on the same level of importance as instrumentation.

Analyser performance is ultimately constrained by the quality of the sample it receives; no amount of computing power can remove bias introduced at the primary sampling stage.

For operations serious about improving control, metal accounting and optimisation, the opportunity lies in aligning three components: TOS-compliant primary sampling, correctly engineered sample handling (such as dry sampling plants and RAMA slurry systems), and appropriate online analysers. Together, these allow plants to measure what truly matters in DMS streams – not an approximation of the process, but a defensible representation of it.

The practical challenge – and invitation – for the industry is clear: review existing sampling and analyser configurations not only in terms of technology, but in terms of representativity and total sampling error. Plants that close this gap will be best positioned to extract full value from Fourth Industrial Revolution tools and to run their DMS circuits with higher confidence, tighter control and improved profitability.

For more information on sampling and online analysis in DMS streams, contact Multotec or watch the full webinar here.