August 2018

Sensor-based ore sorting to maximise profit in a gold operation

  • By B Nielsen, Dry Comminution & Sorting, Outotec Pty Ltd; J Rohleder, Process Metallurgist, Outotec; H Lehto, Technology Manager, Outotec; and C Robben, Business Development Manager, TOMRA Sorting GmbH

This is an extract of a paper originally published in the MetPlant 2017 conference proceedings. The full paper can be purchased online via


Sensor-based ore sorting is being increasingly used to reduce the amount of low-grade and waste material processed in mineral concentrators. This type of preconcentration provides bottom-line benefits to users by reducing the amount of energy, water and consumables, as well as reducing capital cost. Existing operations can increase metal production, while previously uneconomic deposits and low-grade stockpiles can also be exploited. The technology can also be used to separate ore types for selective processing.

The path to implementing sensor-based sorting may include:

  • geometallurgical evaluation
  • first inspection testing to investigate sensor response
  • bench-scale testing where sensor selection is not obvious or for difficult applications
  • performance testing in full-scale sensor-based sorting machines
  • larger scale site-based piloting with a temporary semi-mobile plant installation.

Sorting requires material to be suitably prepared and presented to the machines, and typically this consists of crushing and screening to limit top size and optimise liberation. However, where material streams are suitably sized and prepared, additional equipment may not be required such as the sorting of semi-autogenous grinding (SAG) mill pebble streams.

This paper presents a case study of economic upgrading of gold ore by preconcentrating with sensor-based ore sorting. The case study examines sorting amenability, test work and the feasibility study through to implementation, with associated flow sheet development. The development process is analysed and evaluated with a view to rationalising the process for development of future projects. In addition, limited financial modelling based on expected results is shown to illustrate the benefit to the operation.


Sensor-based particle sorting solutions for the preconcentration of ore are rapidly gaining recognition in the metalliferous mining and mineral processing industry. The technology, however, has had a long and successful history in the industrial minerals processing, recycling and food industries. Since 2014, Outotec has cooperated with TOMRA Sorting GmbH (Tomra) to bring this technology to their mineral processing customers (Outotec, 2014).

Sorting can provide many process, environmental, resource and economic benefits for mine projects.

These include:

  • reduced haulage costs per unit metal production
  • reduced water consumption per unit metal production
  • reduced energy and fuel consumption per unit metal production
  • reduced tailings footprint and costs
  • increased metal production and/or reduced operating costs in existing operations
  • economic recovery of low-grade waste dumps, with possible additional application in dumps causing acid drainage
  • less-selective mining methods can be employed and therefore reducing mining cost per tonne
  • reduced mine cut-off grades can be used, thereby increasing the resource tonnage, with a corresponding increase in production and/or mine life.

Since 2016, Outotec has been involved in evaluating sensor-based particle ore sorting with a junior gold mining company for its project in the northern hemisphere. After initial discussions, test work to evaluate the feasibility of sorting was completed. Industrial scale sensor-based particle ore sorters with X-ray transmission (XRT) and laser sensors were used to upgrade the gold ore, based on the association of gold with sulfides and quartz respectively. Different machine sensitivity and scavenger arrangements were tested to optimise the sorting response. Flow sheet and development options were formulated based on Outotec’s experience and the resulting positive impact on key economic indicators of the project were calculated.

Senior-based sorting


Particle sorting of ore has been used since the times of artisanal miners, hand selecting the best rocks for smelting to optimise product quality and volume (Agricola, 1556). In modern times, mines have found ways to automate this process using physical properties such as density (eg dense media separation). In the electronic age, sensors were initially operated with logic controllers and paddles or other mechanisms to physically separate the individual particles. These machines were throughput-limited and generally operated for special applications. Advances in computing, along with the use of precision high-pressure air jets, have allowed modern sorting machines to reach throughput of over 400 t/h.

Operating principle

The key components of sensor-based particle sorting equipment are the sensors themselves, which can be any of a number of rapid response types, a logic system or computer to process sensor signals and a means of physically separating accepted particles from rejected ones.

Modern sorting machines are of either a belt- or chute-type design. In a belt-type system, material is presented to the sensor array lying in fixed position on the belt. The belt length is designed to give the particles sufficient time to settle on the belt prior to being presented to the sensor. In a chute-type system, the material slides down an angled chute and is presented to the sensors in free fall. The belt system gives better resolution because of the particles being fixed in place, whereas the chute-type solution gives the option to detect material properties on two sides of the particle during free fall. Sensor images of each particle are processed in milliseconds. The processing time is kept as short as possible to minimise the distance from sensor to separation and optimise the system accuracy.

The physical separation of the particles occurs by air jets produced by nozzles arranged in a straight line. The sensors detect the size and position of each particle in addition to the mineral properties. Air nozzles are opened in precise time frames to match the size of the particle, which can be up to 300 mm. The air jets alter the trajectory of the particles that are to be ‘ejected’ enough to be diverted over a splitting or separating plate. Particles that are ‘accepted’ follow along the original trajectory into a chute. It should be noted that the ‘eject’ stream can either be valuable or waste material, depending on the machine setup.

The successful implementation of sensor-based particle sorting also depends strongly on auxiliary processes. Material must be presented to the sensors as separate liberated particles of suitable size distribution, distributed into a monolayer over the whole system width (belt or chute). Sufficient fines removal prior to the sensor system is crucial to keeping downtime of the equipment at a minimum. The presence of fines, particularly wet fines, can greatly increase the maintenance required. The typical dust control solution for sorting consists of a combination of scalping screens prior to the sorting system and dust collection from the sorter ejection chamber through bag filters. Surface-based sensor systems require a sufficient washing process to ensure clear visibility of all surface properties.


The available sensor technologies for particle ore sorting are not able to directly detect gold occurring at the typical fine grain size of most deposits, be they refractory or free milling. Therefore, the sensors are used to detect minerals associated with gold or alternatively characteristics of gangue or waste material. For example, gold is often associated with sulfides such as pyrite or arsenopyrite, which can be detected by XRT sensor technology. This sensor type is most often used on belt-type particle sorters as shown in Figure 1. Higher atomic weight atoms have higher X-ray absorption and are therefore able to be identified by the system.

Figure 1. Diagram of a belt-type XRT sensor-based ore particle sorter.


Gold is often also associated with quartz and, in particular, concentrated in the contact zone of quartz veins. Quartz can be detected by the recently introduced laser sensors, which are most commonly used in the chute-type particle sorter. This is a multi-channel technology capable of measuring reflection, absorption, refraction, scattering and fluorescence to characterise the measured material. Multiple lasers are used on both sides of the particle stream as shown in Figure 2. In the case of quartz and other crystalline materials, the laser light enters the crystal and is reflected, refracted and scattered by the internal structure of the crystal. The crystal will then exhibit a low intensity glow over a wider area than the laser, which can be used to differentiate the crystalline minerals from opaque minerals that only reflect or absorb the laser light.

Figure 2. Diagram of a chute-type dual laser sensor-based ore particle sorter.

The detection system tracks the laser point as it scans across the particle bed and ‘sees’ the changes associated with hard (opaque) or soft (crystalline) mineral surfaces. The detection is not influenced by the surface colour of the minerals. As quartz is often harder and more brittle than the host rock, fractures tend to occur along the vein contact boundary. As a result, quartz is often over-represented on the surfaces of composite particles, further improving the selectivity of laser sorting.


Sensor-based ore particle sorting can be used to significantly upgrade run-of-mine ores prior to feeding to the concentrator. Numerous benefits can be achieved including lower plant capital costs, lower plant operating cost, lower unit mining costs and potential extension to the mine life. Benefits in the plant include a significant reduction in energy, water and reagent consumption.

A sample from a potential gold project in the northern hemisphere was tested in industrial scale sensor-based ore particle sorters with excellent results. The positive financial impact on the project was significant. The project was also more robust with the addition of particle sorting as demonstrated by the reduced NPV sensitivity to gold price downside with sensor-based ore particle sorting.

Purchase the full paper online at

Share This Article