Coal Preparation » Gravity Separation
Although coal-washing dense medium cyclones can be extremely efficient, they continue to lose yield due to inadequate attention to simple issues such as spigot size and control of medium density, particularly in plants with parallel separators.
Dense medium cyclones (DMC) circuits in Australian preparation plants lose around three percent yield through surging, vortex finder overload or poor control of medium condition. If half that additional yield was recovered for half of Australia's DMC circuits, the increase in mine gate annual value of coal produced would exceed $30 million.
The objectives of this research were to conduct accurate performance assessments of single DMCs and multiple DMCs operating in parallel at the Dartbrook, Macquarie and Norwich Park plants. The research examined what factors influenced DMC performance. It also determined covering feed and operating ranges (typical and atypical situations) over the very short term (one minute) and the medium term (days).
The results were used to produce a database to:
- Confirm or identify key factors which reduce yield in DMC plants
- Quantify their effects under many combinations of raw coal characteristics, product specifications and plant configurations
- Provide a simple guide to help producers effectively direct yield maximisation effort to the critical factors.
The test sites were selected due to the diversity of their DMC circuits.
Dartbrook: With a single 1000mm Multotec ceramic lined DMC operating in a region of low-moderate near-gravity material and steady control of medium density and inlet pressure, this plant was selected as a benchmark.
Macquarie: This plant is configured in two modules using 900mm cast steel Minco DMCs of Dutch State Mines design. Each module incorporates a primary circuit of two DMCs fed by two pumps from a single wing tank. With a feed medium density of 1.26RD units, the proportion of near-gravity material is extremely high. Primary sinks from each module are rewashed in a single secondary DMC.
Norwich Park: This single-stage process has six parallel circuits, each fitted with two 710mm cast units based on the McNally "cycloid" design.
In terms of partitioning performance, no significant flaws were identified in the control and operation of the Dartbrook DMC. Partition curves were near-symmetrical and reproducible in the short and medium-term. Over the normal range of operating conditions, for multiple tests with 16mm density tracers and including experimental errors, offset estimates encompassed a range of only 0.015RD units and Ep estimates encompassed a range of only 0.003RD units.
Additional tests were conducted under conditions outside the normal range of operation at Dartbrook. Results were in broad agreement with predictions from Wood's dense medium cyclone performance model.
Despite plant procedures which maintain reasonable operating conditions and careful calibration of density gauges, a consistent difference in primary DMC cutpoints was observed. Simulations indicated that, due to the extreme level of near-gravity material, this discrepancy was misdirecting around 1.9 percent of DMC feed to thermal product rather than the higher-value metallurgical product. The main cause was a sampling point, used to calibrate the Module 1 primary DMC medium density gauge, which generated a biased medium sample. This was to be confirmed and corrected by Macquarie personnel. Minor factors related to maintenance of DMC spigot dimensions and of correct inlet pressures.
Similar problems, including biased medium samples and dimensional differences, were identified in the secondary DMC circuits. The effect on yield was small due to the small proportion of near-gravity material. Inadequate supply of medium to the DMC feed pumps caused very large, short-term fluctuations of inlet pressures. Despite this, partitioning efficiency was excellent in the relatively small number of tests conducted on those circuits. Plant personnel are taking steps to correct the medium supply problem.
At Norwich Park, the control of cutpoints was poor. Under nominally identical operating conditions, the six parallel circuits exhibited cutpoints that varied from each other by up to 0.13RD units. However, the loss of readily recoverable yield was only one percent due to the small quantities of gravity material involved and the absence of surging or overloading.
A major reason for poor control of cutpoints was the presence of varying amounts of froth in the circulating medium. This made it impractical to accurately calibrate the nucleonic density gauges and for them to remain calibrated. Another significant problem was poor control of medium density.
Although it was difficult to compare loop performance, the influence of differing spigot diameters on cutpoints was evident. The research also found that some of the cycloids were in poor condition and components were mismatched, such as large steps at joints and low-wearing ceramic components fitted to high-wearing steel components in a single separator. Norwich Park personnel have started a program of plant and control system modifications to address these problems.