Coal Preparation » Process Control
The objective of ACARP project C19046 was to provide the Australian coal industry with reliable supplier-independent, size-by-size data describing the performance of modern spirals under a range of operating conditions. Data was to be generated for individual spiral starts as well as for the combined bank, over a range of
conditions. Unfortunately, the project has only partially succeeded in meeting this objective.
The performance of individual Mineral Technologies and Multotec spirals in plant operation were investigated using fine colour-coded density tracers. Four sizes of tracer were used: -2+1, -1+0.5, -0.5+0.25, -0.25 mm.
Test work was carried out at Warkworth (Mineral Technologies spirals) and Red Mountain (Multotec spirals). At Warkworth, feed rate and feed pulp density were changed over the course of the sampling period, while at Red Mountain coal type was varied between the two days of sampling. At both sites two individual spirals within a column of three starts as well as the entire spiral bank were sampled. The samples were returned to the JKMRC for sizing and tracer counting.
The two larger sizes proved straightforward, if tedious, to count using a magnifying lens. Counting time was typically 45 - 60 minutes per sample.
The minus 0.5+0.25 mm could also be counted the same way, although it was very much harder to identify the tracer colours and to spread them into a monolayer for counting. The counting took very much longer (more than two hours being typical). The accuracy was also very much poorer than for the coarser sizes, as tracers were often partially hidden, and the colours were very much harder to distinguish.
It proved impossible to count the finest size fraction (-0.25 mm) successfully, either manually under a lens or a microscope, or automatically using a USB microscope and simple image analysis software.
The raw tracer counts were mass balanced and the partition curves calculated using software written in Matlab. The software is freely available to the industry on request from the author. The software also estimates error bars for each point on the partition curve, an additional advantage of using tracers compared with conventional float-sink analysis.
The project has shown that both spiral types are capable of good performance with Eps in the range 0.05 to 0.06 for -2+1 mm and 0.06 to 0.09 for -1+0.5 mm.
Partition curves for the minus 0.5+0.25 mm could not be reliably determined in most cases.
The project has shown that it is quite feasible to use the two coarser sizes of tracer to assess the performance of either individual spirals or an entire bank in the plant, far more rapidly and more cheaply than by conventional sampling and fine floatsink analysis. Sampling takes of the order of 30 minutes from start to finish; the samples can be decanted and filtered the same day. After overnight drying, subsampling by riffling and screening into size fractions takes some three to four hours per sample. Finally counting the tracers takes about one hour per sample per size fraction. These timings imply that a partition curve for one size fraction can be available within three days of sampling. The technique is not feasible at present for the two finest sizes.