Keywords

Sorting principles; history; sensor technologies; applications; generic flowsheet

14.1 Introduction

The term “sensor-based (ore) sorting” (SBS) is introduced as an umbrella term for all applications where particles are singularly detected by a sensor technique and ejected by an amplified mechanical, hydraulic, or pneumatic process (Wotruba and Harbeck, 2012). Analogous terms include ore sorting, electronic sorting or automated sorting. A variety of sensor types are available and see use in the minerals, recycling, and food industries. SBS can be implemented at various positions in the mineral processing flowsheet:

Separation of one or more ore types that are fed alternately as batches into the same plant or parallel into multiple plant lines for specialized treatment.

Note that SBS is almost always used in diamond processing flowsheets where the terminology differs slightly: waste rejection is referred to as “concentration” and concentration is referred to as “recovery.”

SBS refers to a concentration stage which identifies certain physical or chemical characteristics of individual rock particles and separates them from the process stream via a physical mechanism (Arvidson, 1987). Ore sorting is commonly undertaken after primary or secondary crushing, after sufficient liberation is achieved. Applications show that many mines have about 30 wt% barren waste liberated in the size range 10–100 mm, which allows material to be discarded without significant loss of value. Pre-concentration by sorting is seen as a method of improving the sustainability of mineral processing operations by reducing specific materials handling requirements, minimizing energy consumption and water in grinding and concentration, and achieving more benign tailings disposal (Cutmore and Ebehardt, 2002; Lessard et al., 2014). The ultimate goal is to minimize specific investment and processing costs while reducing the environmental footprint of an operation. Sensor-based sorting can be applied to a waste rejection stage (e.g., base and precious metals) or concentration (i.e., the production of an intermediate or final product, e.g., industrial minerals, ferrous metals, and gemstones). Sorting has also been used to upgrade previously mined/processed waste-rock material prior to re-processing (von Ketelhodt, 2009; Wotruba and Harbeck, 2012).

SBS is the automation of hand sorting, which is now extended by the use of additional sensing/detection technologies (i.e., techniques are not limited to optical sensors). Hand sorting is the original mineral concentration process, having been used by the earliest workers several thousand years ago. The practice was recorded by Agricola (1556) (Figure 14.1).

Hand sorting involves the visual assessment of individual ore particles and the rejection of those particles that do not warrant further treatment. Figure 14.2 shows hand sorting in the early days of operation at the Sullivan Mine (Cominco), British Columbia, Canada, which began operation in 1909 (Ednie, 2006).

Hand sorting has declined in importance due to the need to treat large quantities of low-grade ore which can require extremely fine grinding. Hand sorting of some kind, however, is still practiced (e.g., the removal of large pieces of timber or tramp iron from the run-of-mine ore). It is also still applied in countries with low labor pay rates.

Early workers identified ore sorting as a key variable which affected the economics of a mine. Rickard (1905) in the book “The Economics of Mining” stated “whether to (hand) sort or not, is a question of vital importance to the economics of a mine; it may mean the choice between a small yield of high-grade material or a large output of low-grade, which immediately affects all the operations carried on at the surface, as well as underground….”

SBS is reviewed here, but several other waste elimination methods exist, which are primarily gravity-based processes (Chapters 10 and 11) that could be employed underground or at the mine. The techniques are not currently employed underground due to the need for complex water handling and retreatment systems (Hughes and Cormack, 2008; Murphy et al., 2012). Complete SBS installations are relatively compact (small volume to throughput ratio), especially when operated dry. Thus, SBS is commonly employed using a semi-mobile operation. About 30% of today’s machines being installed are containerized systems. This being said, there is great potential for SBS systems integration underground (near-to-face sorting) especially where stoping and cut-and-fill methods are used (Schindler, 2003; Dammers et al., 2013; Robben, 2014).

SBS was first introduced in the late 1940s, and although its application is fairly limited, it is an important technique for the processing of certain minerals (e.g., diamonds, uranium, limestone, magnesite, gemstones) (Sassos, 1985; Salter and Wyatt, 1991; Sivamohan and Forssberg, 1991; Collins and Bonney, 1998; Arvidson, 2002). Arvidson (1987) stated that vein-type, layered, brecciated or pebbly mineralizations were typically good candidate ores for sorting. There is a general misconception that sorters cannot be used in massive ores, but full liberation is not necessarily required for successful waste elimination (Chapter 1). Complete liberation may be required in certain instances for concentration, such as when limestone is required for filler purposes (Arvidson, 1987).

14.2 Sensor-based Sorting Principles

Many mineral properties have been used as the basis of sensor-based sorting, including reflectance and color in visible light (magnesite, limestone, base metal and gold ores, phosphates, talc, coal), ultraviolet (scheelite), natural gamma radiation (uranium ore), magnetism (iron ore), conductivity (sulfides), X-ray fluorescence (base metals), and X-ray luminescence (diamonds). Infrared, Raman, microwave attenuation, and other properties have also been tested. Table 14.1 gives examples of the properties that can be exploited using commercially-available sensor technologies along with example industrial applications. Reviews on historical developments, the various sensor types, and applications are given elsewhere (Salter and Wyatt, 1991; Wotruba, 2006; Wotruba and Harbeck, 2012).

SBS systems inspect particles to determine the value of some property using contactless and real-time measurements that obtain both location and material properties. The data are processed and the information (e.g., visible light reflectance) creates a basis for ejection (or retention) of those particles which meet some criterion (e.g., light vs. dark particles). Therefore, a distinct difference in the required physical property must exist between the valuable minerals and the gangue. Either valuables or waste may be selected for ejection. Automated sorters consist of four basic subsystems (Arvidson, 1987; Salter and Wyatt, 1991; von Ketelhodt, 2012; Robben et al., 2013):

Robben et al. (2013) include a material conditioning stage prior to particle presentation. Most important for successful operation is the presentation of a carefully selected and screened particle size range that shows both liberation and a minimum amount of fine material which is detrimental for high availability. For photometric sorting, particle surfaces sometimes must be moistened or washed, so that blurring of the signal by a covering layer does not occur. The upper size limit is technically 350 mm, but nonetheless liberation of barren waste is often experienced below 100 mm. The lower size limit is technically 0.5 mm for most detection technologies, but as the operating costs are inversely proportional to the average particle size (and weight), the economically feasible lower size limit is often in the range of 10–20 mm. Separation efficiency decreases when a wide range of particle sizes is fed to a single machine, the ratio between maximum and minimum particle size (size range coefficient) typically should not exceed three (Robben et al., 2013).

Particle presentation can be achieved via two system types: chute or belt-type systems (shown in Figure 14.3). The chute system senses the particles as they free-fall after being guided on a high-incline chute. For the belt system, the sensor is mounted above or below the conveyor belt, which feeds a monolayer of particles. The ore must be fed in a monolayer, as individual particles must be displayed to the sorting device for effective separation.

Various sensing technologies that are available and common industrial sensors are listed in Table 14.1. Electronic processing involves analysis of the data acquired by the detector. A wide range of site-specific algorithms are used, depending on the sensor type and ore characteristics. Physical separation is typically achieved using an array of about 200 high speed air valves. Mechanical ejectors are installed in low throughput units, for example, single particle XRL (X-ray luminescence) diamond or XRF (X-ray fluorescence) sorters. Water jets have been discussed and trialed, but have not found their way into today’s industrial scale sorters (Fickling, 2011; Robben et al., 2013).

14.3 Historical Development

First patents for SBS technology originated in the 1920s (Sweet, 1928). During the 1950s Kelly and Hunter (K+H) developed the Model 6 (M6) photometric sorter, which was subsequently installed at the Mary Kathleen uranium mine in Australia (Salter and Wyatt, 1991; Stewart, 1967). In 1966, Gold Fields of South Africa undertook a joint project with Rio Tinto-Zinc (RTZ) (through their subsidiary Ore Sorters) to develop sorting technology for use in South African gold-mining operations (Barton and Peverett, 1980). The project culminated with the installation of a Model 13 prototype photometric sorter at the Doornfontein Gold Mine in 1972 (Keys et al., 1974; Barton and Peverett, 1980). The sorters were the first to use laser technology and are considered the first high tonnage sorters (Salter and Wyatt, 1991). The Gunson’s Sortex MP80 machine was probably the first sorter to employ microprocessor technology (Anon., 1980). The sorter handled minerals in the size range 10–150 mm at feed rates of up to 150 t h⁻¹.

Rocks having white or gray quartz pebbles in a darker matrix were accepted, while quartzite ranging from light green through olive green to black, were rejected. Most of the gold occurred in rocks which reported to the “accept” category. Uniform distribution of the ore entering the sorter was achieved by the use of tandem vibrating feeders and the ore was washed on a second feeder to remove slimes which could affect light-reflecting qualities. The successful implementation of the Model 13 sorter led to the development of the RTZ Ore Sorters Model 16 photometric sorter, which has been used since 1976 on a wide range of ore types (e.g., magnesite, wolframite, gold) (Anon., 1981a; Barton and Peverett, 1980).

A subsequent development to the RTZ Ore Sorters Model 16 is the Ultrasort UFS120 photometric sorter, which is used in the processing of magnesite, feldspar, limestone, and talc. Ore passes from a vibrating feeder to high pressure water sprays and counterweight feeder where water is removed and the rocks are accelerated to form a monolayer. They drop onto a short conveyor moving at 2 m s⁻¹ where they pass via a high speed 5 m s⁻¹ “slinger” conveyor into free fall, now well separated. The rock layer, 0.8–1.2 m wide, is scanned by a laser beam at up to 4,000 times per second, and the reflection analyzed in less than 0.25 µs by photomultiplier tubes and high speed parallel processors. One or more of 120 air ejectors are fired to divert the value or waste past a cutter and into the accept/reject bins. As the position of the rock is accurately identified, and the ejector firing duration is less than 1 ms, the sorter can operate very selectively.

SBS has been employed in diamond recovery since the 1960s, initially using simple optical sorters and more recently machines based on the fact that diamonds luminesce when irradiated by X-rays (Anon, 1971; Rylatt and Popplewell, 1999a,b; Damjanović and Goode, 2000). X-ray luminescence sorters are used in almost all diamond operations for the final stages of recovery after the ore has been concentrated by DMS (Chapter 11). They replace grease separation (Taggart, 1945), which exploits the natural hydrophobicity (oleophilicity) of diamonds and is now used only in rare cases where the diamonds luminesce weakly or to audit the X-ray sorter tailings. Luminescence is a more consistent diamond property than oleophilicity, and sorters are more secure than grease belts or tables.

Figure 14.4 shows an early dry X-ray sorter, in which the DMS concentrates are exposed to a beam of X-rays in free fall from a conveyor belt, the luminescence detected by photomultiplier tubes and the diamonds ejected by air ejectors. Both dry and wet X-ray machines are now available, and the process is usually multistage to ensure efficient rejection of waste with very high diamond recoveries.

The “Lapointe picker” was probably the earliest radiometric sorter, developed in Canada and used in the 1940s (Lapointe and Wilmot, 1952; Bettens and Lapointe, 1955; Salter and Wyatt, 1991). Radiometric sorting has since been used to pre-concentrate uranium ore in South Africa (Anon., 1981b), Namibia, Australia (Bibby, 1982), and Canada (IAEA, 1967, 1980, 1993). A sorter installed at the Rössing Mine in Namibia (Gordon and Heuer, 2000) detected gamma radiation from the higher grade ore pieces using scintillation counters comprising NaI crystals and photomultiplier tubes mounted under the belt. Lead shielding was used to achieve improved resolution of detection. A laser-based optical system similar to that used in photometric sorters was used to determine rock position and size for ejection, and could be adapted to determine additional optical characteristics of the rocks. Some uranium operations undertake selective mining practices, which could be considered ore sorting (IAEA, 1980, 1993). Radiometric readings are taken on truck-loads of ore. Based on the readings, ore is placed in high-grade, low-grade, or waste stockpiles. The ore can then be appropriately blended or processed separately based on grade.

Several other physical properties of ores and minerals have been exploited in a range of sorting machines. Neutron absorption (or activation) separation has been used for the sorting of boron minerals (Mokrousov et al., 1975). The ore is delivered by a conveyor belt between a slow neutron source and a scintillation neutron detector. The neutron flux attenuation by the ore particles is detected and used as the means of sorting. The method is most applicable in the size range 25–150 mm. Boron minerals are easy to sort by neutron absorption since the neutron capture cross section of the boron atom is very large compared with those of common associated elements and thus the neutron absorption is almost proportional to the boron content of the particles. The technique has also been tested on gold ores (Uken et al., 1966, 1968).

Near infra-red (NIR), which has seen wide use in recycling applications, has recently been investigated for use in the mining industry. It has been successfully implemented for waste elimination from boron minerals (colemanite, ulexite) (Dehler et al., 2012). Development work has shown promise in separating talc from carbonate and quartz (von Ketelhodt and Bartram, 2009), processing porphyry copper samples (Dalm et al., 2014), and for waste rock removal in the diamond industry (von Ketelhodt and Bartram, 2014).

Photoneutron separation (gamma activation) is recommended for the sorting of beryllium ores. When a beryllium isotope in the mineral is exposed to gamma radiation of a certain energy, a photoneutron is released and this may be detected by scintillation or by a gas counter.

The RTZ Ore Sorters Model 19 sorter measured conductivity and magnetic properties and had application to a wide variety of ores including sulfides, oxides, and native metals (Anon., 1981a). The machine treated 25–150 mm rocks at up to 120 t h⁻¹. Such systems employ a tuned coil under the belt which is influenced by the conductivity and/or magnetic susceptibility of the rocks in its proximity. Phase shift and amplitude are used to decide on acceptance or rejection.

Outokumpu developed the “Precon” sorter which was installed at the Hammaslahti copper mine (now closed) (Kennedy, 1985). The Precon sorter used gamma scattering analysis to evaluate the total metal content, and had a capacity of 7 t h⁻¹ for 35 mm lumps rising to 40 t h⁻¹ for 150 mm lumps. Primary crushed ore was pre-concentrated, rejecting about 25% as waste grading 0.2% copper, compared with an average feed grade of 1.2%.

RADOS XRF sorting technology has been used (primarily in Russia) over the past two decades at 49 operational sites for over 20 commodities (Fickling, 2011; RADOS, 2014). The sensor directly identifies ore elemental composition using X-ray fluorescence technology (able to detect elements with an atomic number >20) and sorts accordingly. Figure 14.5 shows the internal arrangement of the RADOS XRF chute-type sorter and a row of units operating at a copper/zinc mine in Russia. Particles free-fall from the chute past an X-ray source and detector. Particles are separated via a mechanical ejector into concentrate and discard streams. Typically, feed grade must exceed 0.1 wt% for effective detection, although the system can also use a matrix of elements as the criteria for separation in low grade ore systems (e.g., gold, uranium or PGM) (RADOS, 2014).

Microwave attenuation has been used to sort diamond-bearing kimberlite from waste rock (Salter et al., 1989). The development was notable for the first use of high speed pulsed water ejectors. Equipment to sort asbestos ore has also been developed (Collier, 1972). The detection technique was based on the low thermal conductivity of asbestos fibers and used sequential heating and infrared scanning to detect the asbestos seams. The use of microwave heating coupled with infrared-thermography has been tested on a wide range of minerals (e.g., copper–molybdenum ore, iron ore) for sorting purposes (Sivamohan and Forssberg, 1991; van Weert et al., 2009; Ghosh et al., 2013).

A machine was installed at King Island Scheelite in Tasmania, where the scheelite was sensed by its fluorescence under ultraviolet radiation. XRT (X-ray transmission) has proven to be superior to UV fluorescence. Mittersill Mine currently operates XRT sorters for waste elimination where ca. 25% of the 130 t h⁻¹ run-of-mine stream grading 0.03% WO₃ is rejected before milling and sold as aggregate (Mosser and Gruber, 2010).

Rapid developments in sensing and computing technology have increased the capabilities of today’s SBS machines significantly. In addition, industrial food processing and recycling industries have openly adopted the technology and tens of thousands of sorters are installed for numerous separation tasks. The mining industry is now benefiting from those developments, with proven detection and data processing being available for application on mining proof machines.

14.4 Example Flowsheet and Economic Drivers

SBS installations typically consist of: crusher, screen, sorter, and compressor. Because of the often large particle sizes and low concentrations of valuables, it is advised to install mechanical sampling, size reduction and splitting equipment to enable determination of plant performance.

Figure 14.6 shows an example flowsheet for a multi-stage, multi-machine stage operation. Machines are commonly operated in parallel for high throughput operations. Cascading circuit arrangements enhance separation efficiency and allow for redundancy and the potential for multiple detection technologies to be combined.

Capital and operating costs are typically roughly half the costs of dense-media systems. Since SBS is a single particle technology, the throughput is inversely proportional to the average particle size fed to the machine. Both types of specific costs are thus inversely dependent upon the average particle size. The upper particle size is determined by the liberation characteristics of the ore and the lower particle size limit by the economically viable specific costs per ton of product.

Typically, the highest economic benefit is obtained when eliminating deleterious waste that hinders downstream processes. In such cases, the overall recovery (and potentially product quality) increases while reducing the specific costs per ton processed, and reducing the comminution and processing of waste material. Eliminating waste from the ROM stream also allows for higher dilution in the mine, which may allow for processing of marginal waste dumps or mining blocks (i.e., allowing for decreased cut-off grades and extended mine life) (Bamber et al., 2004; Pretz et al., 2011; Robben, 2013) (Figure 14.7).

In summary, SBS is a relatively low cost and versatile technology that has the potential for wide applicability in the minerals industry, if liberation of either waste or product is sufficient in coarse particle sizes. The currently available suite of detection technologies allows for discrimination of single particles based on a pre-determined separation criteria (e.g., NIR spectrum, color, conductivity, atomic density). Today’s mining machines integrate state of the art technology into mechanical designs. Increased pressure to decrease processing costs and environmental impact will undoubtedly further the use of SBS in the mining industry.