Density Tracer Testing

Tracer Size(s) and Retrieval
Tracer Densities
Number of Tracers
Conduct of the Test
Interpretation of DMC Partition Curves
References

The following procedures focus on dense medium cyclone circuits but are adaptable to other units.

For a DMC circuit which has not previously been tested in this way, one must;

• select appropriate size(s) of density tracers, while ensuring that they can be retrieved.
• determine appropriate tracer densities to be employed.
• determine the number of tracers to be used at each density interval.
• assemble the required personnel and conduct the test.
• assess the data.

Tracer Size(s) and Retrieval

To ensure relevance of the data to be generated, it is strongly recommended that tests be conducted with feed on. If using magnets for retrieval, one is usually free to select tracer sizes from 2mm up. If the nominal feed size range is, for example, -25+1mm, one may select tracers of 16, 8, 4 and 2mm. To check that retrieval rates are adequate, one may position the magnets at the discharge lips of the drain-and-rinse screens, then insert tracers of the relevant sizes at the feed ends of the screens. Retrieval rates commonly approach 100%, but reliable partition curves can be generated with retrieval rates as low as 70% (Wood, 2004).

If tracers are to be retrieved manually, it is usually feasible to employ only one size of tracer, which must be large enough to be reliably seen in the load on drain-and-rinse screens. For installations with a feed topsize in the 20-70mm range, 32mm tracers are most often used. For circuits with smaller feed it is sometimes possible to use 16mm tracers. At least one person must be positioned to retrieve tracers from each drain-and-rinse screen in the circuit, for the duration of the test.

Tracer Densities

Coal

Density tracers in the coal range are available at density intervals of 0.01 g/cc (1.24, 1.25, 1.26 etc). Under common operating conditions, the separation density for a coal-washing dense medium cyclone is typically around 0.1 RD units higher than the feed medium density. In cases where the expected cutpoint is in a region of low “near-gravity” material; for example, if the feed medium density is greater than 1.40 g/cc, adequate definition of the partition curve can be obtained using only every second density increment. One may conduct a "sighting" test with small numbers of tracers at densities from say 0.1 RD units below the feed medium density to 0.3 RD units above the feed medium density. For example, if the feed medium density were 1.40, three tracers at each of the densities 1.30, 1.32, 1.34,....1.68, 1.70 may be used - a total of 63 tracers. The results of this test provide an improved estimate of separation density or definition of a density range of particle retention (refer following notes on interpretation of partition curves). If tracers larger than the feed topsize are employed, this test will also show whether they will pass freely through any unusual sieve-bend feed chute and skirt arrangements. If there is little or no retention, the full test should use tracers spanning a density range of approximately 0.3 RD units, centred on the estimated separation density. For example, if the separation density estimate from the "sighting" test is 1.52 RD units, tracers of densities 1.36, 1.38, 1.40 ...... 1.66, 1.68 may be used. After further experience with the circuit it may be possible to reduce the range to less than 0.2 RD units, but still have confidence that the entire partition curve (partition numbers from 0 to 100) will be defined. For circuits with separation densities higher than 1.40, if there is any evidence of a low-density "tail" (Figure 3 in the "Interpretation Section" which follows) to the partition curve it is desirable to include tracers of a density close to that of clean coal - say 1.30.

Diamonds & Other Minerals

Density Tracers for diamond applications are offered in 18 densities ranging from 2.5 g/cc to 3.53 g/cc. The density intervals are 0.1 g/cc, or 0.05 g/cc in the region of the commonly-targeted cutpoint of about 3.1 g/cc. Metallurgists typically select from these about 10 densities which they find to be important to their operations. Cutpoints for nickel and iron ore processing are usually similar to those for diamonds.

Number of Tracers

Coal

In coal operations it is common to use 30 tracers at each selected size and density. Thus, for a test using 3 sizes and 20 densities, with retrieval by magnets, the total number of tracers inserted would be 1,800. They would usually be added in less than 1 minute.

If tracers are retrieved manually, only one (large) size would be used, with a total of 600 tracers. To allow manual retrieval, the tracers should be mixed, to allow insertion in random density order, over a period of about 20 minutes. The test would provide no direct information about behavior of smaller particles.

In coal operations with unstable medium, it is not uncommon for large tracers of densities close to the cutpoint to be retained in a dense medium cyclone for periods of an hour or more (see below). If prior experience of a circuit suggests that more than 50% of tracers in some sizes and densities may be retained, one may consider using only about 3 tracers at each density for the suspect size. This is sufficient to define a range of retention, and little information of value would be generated by using more tracers.

If separation is efficient, partitioning of particles sized 16mm and larger is very efficient and the curves will likely be indistinguishable. If 16mm tracers are used, there is little point in testing with larger tracers unless the ore contains particles substantially larger than 16mm and small numbers of larger tracers are used to check for retention.

Diamonds

With the larger density intervals in diamond circuit tests, it is common to use 100 tracers at each selected size and density. For a test using 3 sizes and 12 densities, the total number of tracers would be 3,600.

Conduct of the Test

If magnets are used, they should be positioned in the drain-and–rinse launders, but just out of the ore streams. After conducting any preliminary observations such as determinations of the densities of feed, overflow and underflow media, drop the selected tracers into the circuit feed, typically into a deslime screen oversize launder.

Slide the magnets into their respective floats and sinks streams before arrival of the first tracers.

5 minutes after addition of the tracers, slide the magnets out of their streams. Any tracers not retrieved are considered as lost, or retained in the separator. If necessary, gently hose ore particles off the magnet, then pick off the tracers for washing, sorting into size and density fractions and counting. In cases where products from individual cyclones in a module are drained on separate screens, it can be helpful to separately record the tracers which report to those screens. A separate sheet may be used for each tracer size.

Enter the numbers in the floats and sinks columns of appropriate test sheets (Test Data Sheet)

If tracers are to be retrieved manually, they should be inserted at a rate suitable for manual retrieval, around one tracer every two-three seconds.

Interpretation of DMC Partition Curves

The figures below illustrate the common forms of density tracer partition curves for dense medium cyclones. A module of one or more well-operated and well maintained dense medium cyclones should show an efficient separation (Figure 1). By contrast with conventional float/sink techniques, density tracers provide the resolution which shows that large particles can be partitioned with an Ep of less than 0.01 RD units.

Fig. 1 Normal (efficient) partition curve

Figure 2 shows a reasonably small RD range of particle retention. Separation is still quite efficient but there is a danger that a small change in operating conditions may increase the density range of retention. The cyclones rapidly become choked with "near-density" material and frequently clear themselves by ejecting surges of slurry, including low-density coal, to underflow.

Fig. 2 Tracer Retention

The resulting partition curve is shown in Figure 3. The Ep is large; there is a low-density "tail" and a low (sometimes negative) offset between feed medium density and cutpoint. The performance shown in Figure 3 can also arise from vortex finder overload when the medium flow from the vortex finder is insufficient to carry out all the particles which should report to the low-density product. As with surging, the yield loss can be very significant.

Fig. 3 Surging or Vortex Finder Overload can cause yield loss

A curve with a plateau (Figure 4) is indicative of differing cutpoints between separators in the module. Examination of the data for individual product screens will suggest which units are separating at high, and which at low density.

Fig. 4 Two DMCs with different cutpoints

Means for the correction of these separating inefficiencies may be found in the references listed below or by contacting Partition Enterprises Pty Ltd

References