Screening and Magnetic Separations for Fine Coal Dense Medium Plants

Since the early 1990s, there has been considerable development of the technology and equipment available for coal preparation which potentially can overcome the serious limitations identified in fine coal Dense Medium Cyclone (DMC) circuits. These limitations were poor size classification and magnetite recovery. Recent work on centrifuges equipped  with laser cut screens has indicated not only excellent dewatering capabilities but also significant potential for rapid size classification in the size range of interest, namely 0.100 to 0.250 mm. Other classifying units such as vibrating sieve bends or cyclones have proven to be low capacity or low efficiency units requiring significant attention and are not favoured by either operational or maintenance staff. This marked improvement in classification efficiency would also improve the performance capability of spiral/Teetered Bed/ reflux classifier circuits as well. Another development which has sparked interest in examining the option of DMC for fine coal is the use of new high intensity magnets which have virtually eliminated the need for magnetite separators in series and allow magnetite losses to be maintained at very low levels.

The objectives of this project were to assess the effectiveness of new screening units in removing sufficient ultrafines from the DMC feed and cyclone product, and assessing the effectiveness of the new high intensity magnetic separators. The aim was to improve the design of the DMC process circuit for fine coal to make it acceptable both metallurgically and with regard to maintenance. The improvement in size classification would also be relevant to other intermediate cleaning circuits. The end result would be simpler, more cost effective and higher recovery fine coal circuits.

The benefit from this project would be the identification of new technology which can be used with equipment that operate at low Ep values and has the flexibility to alter the relative density of separation to meet changes in coal characteristics and market demands.  That is, the plant has a higher recovery than existing operating circuits with separations at the same product ash.  This technology would also allow higher recovery of ultra fine magnetite and superior size classification prior to and/or after separation.  

The work was carried out by staff from the ACIRL and CSIRO Coal Preparation Groups located at ACIRL's Maitland facility and the Newcastle Energy Centre, respectively.  

Pilot scale methodology was adopted. The units to be operated were a centrifuge (LMPE), capable of feed rates of up to 4 m3/hr and creating centrifugal separation forces in excess of 300g, and a pilot scale magnetic separator (Multotec). The feed slurry assessed was a spiral circuit feed (-2+0mm) in order to determine the effectiveness of the removal of ultrafine material from feed and the impact of the removal of these ultrafines on the product and reject streams. DMC performance was assessed using well known models rather then attempting to operate a small DMC circuit.  This information was used to develop process streams based upon sizing and float/sink analysis of the coal used in the project. The work program had eight sections:

• Reconfigure ACIRL system at Maitland to handle a fine coal system operated at 4 m3/hr. Install LMPE's pilot scale fine coal scroll centrifuge and Multotec's Magnetic Separator.               
• Collect large quantities of spiral feed slurries. Process in centrifuge fitted with laser cut screen to assess removal of ultrafine from cyclone feed. Parameters to be varied are: Feed Rate Feed Solids
• CSIRO to model DMC cyclone performance to formulate cyclone O/F and U/F slurry characteristics.  
• Produce slurries of cyclone product and reject based upon CSIRO models. Slurries to include predicted coal % solids and size distribution, magnetite % and size distribution.  
• Undertake centrifuge tests on test slurries to simulate drainage using parameters such as feed rate and solids content.
• Undertake testing on a rinse section to recover magnetite by adding extra water to the centrifuge product and recentrifuging. Monitor quality of correct media from drainage tests.
• Test dilute stream from centrifuge rinse tests in magnetic separator to assess magnetite loss. Factors to be monitored are size distribution of coal and magnetite, plus screening efficiencies.
• Prepare a report outlining circuit options for DMC of fine coal, detailing the screening efficiencies of the laser cut screen in the pilot scale fine coal centrifuge unit in sizing the DMC feed and processing the DMC O/F and U/F. It will also address the issue of magnetite loss and how effective the new magnetite separators are in improving the recovery rate. The report would also comment on how the new screening technology may be applied to the more conventional fine coal processes.

The experimental work lead to an unusual set of results (unexpected behaviour of 0.030 - 0.040 mm coal), which required a major revision of the approach to analysing the partition curves in which the effects observed in hydrocyclones, screenbowl and scroll centrifuges were considered. This new analysis approach needed to be developed since the curves had odd shapes which were continually repeated from experiment to experiment and no explanation with respect to accuracy of the experimental work could be elucidated.  

New models were developed for the classification centrifuge and the magnetic separator which could be used in a LIMN simulation of dense medium cyclone performance for a fines  circuit. Models for DMC operating with fine coal hydrocyclones and sieve bends were developed from information in the literature.

The model relationships for each unit operation was the subject of an EXCEL worksheet and these were dynamically linked using the LIMN simulation software.

For the base case in the analysis, the desliming duty was carried out by hydrocyclones and sieve bends and the drain and rinse was carried out with sieve bends. This was used for the investigation of the issues which were identified as those limiting the introduction of a cleaning circuit such as dense medium cyclones for fine coal. These were ;

• Quality of the desliming. Note the impact of ultrafine particles on final product quality and build up of non-magnetics in the medium (for example, ultrafine pyrite at Curragh)
• Contamination of the medium with non-magnetics
• Loss of magnetite
• Product yield and ash value.

Comparison of the simulation results obtained with the sieve bends and the classifying centrifuges is presented in the report.

With respect to the issues which were limiting the use of dense medium cyclones for fine coal cleaning ;

• The amount of non-magnetics in the medium have been significantly decreased. The recirculating load of non-magnetics has reduced from 18.5 to 7.6 tph.
• The level of ultrafine material in the non-magnetics has decreased from 4.4% to 3.2%. The overall impact would be a higher quality medium which could be more controllable.
• The loss of magnetite has also decreased markedly. The overall value was 0.43 tph compared to 3.75 and 1.60 for the base case with improved magnetic separation capability and the use of more sieve bend width respectively.
• The magnetite losses to the product, reject and overflow from the magnetic separators were 0.13, 0.26 0.04 tph respectively.
• The product yield appears to have increased appreciably. This is mainly an outcome of being able to use a 0.2 mm screen in the drain and rinse centrifuges (effective separation size of about 0.125mm) compared with the 0.25 mm apertures in the drain and rinse sieve bends (effective separation size greater than 0.2 mm). A reduction in the aperture for the latter device would have increased the loss of magnetite further.
• The ash value of the product also has improved noticeably. Improved medium quality and yield and decreased magnetite loss are the driving factors.   

Potential further improvements are difficult to identify ;

• The M/C ratio is already rather low for a DMC for fines system and in reality may have to be increased. For this reason, the amount of drain being directed to the DMC sump cannot be lowered from the 0.75 value used in the simulation.
• The experimental work suggests that for classifying scroll centrifuges which are operating at 350g and with apertures of 0.2 mm, the amount of water and ultrafines (magnetic and non-magnetic) reporting with the product is constant for the conditions used in this work.
• The use of further rinse steps is unlikely to be cost effective since classifying centifuges would be a relatively expensive unit operation.

The use of sprays within the rinse centrifuge may result in an improvement in removal of ultrafines, but the design of the current pilot scale unit precluded the experimental examination of the potential option.

There are a number of factors which need to be considered when considering a scroll centrifuge being used for classification in place of the conventional hydrocyclone and sieve bends;

• Generic partition curves.
• Separation size - Hydrocyclones and the scroll centrifuge can produce separations below 0.1mm. This task is somewhat problematical with a sieve bend.
• Efficiency - The sieve bend produces the sharpest partition curve. For a particular particle density the sharpness of separation produced by the centrifuge and the hydrocyclone are similar.
• Particle density - Sieve bends and scroll centrifuge are not impacted by particle density. Hydrocyclones are impacted by particle density. The overall degree of efficiency of separation of a hydrocyclone is significantly affected by particle density and is poorer than that of the centrifuge.
• Rf - For the sizes of interest in this investigation, the centrifuge produces a consistent Rf value of about 7%. Under particular operating conditions, hydrocyclones and sieve bends can produce a similar value. Under normal operating conditions the Rf values are higher. For sieve bends, the Rf value increases with increasing feed rate of solids.
• Ultrafines - These tend to have partition numbers similar to the Rf value. For the centrifuge, there is a tendency for some particles in the size range 0.02 to 0.04 mm to be captured in the oversize stream.

Clearly, the best option for a size classification duty is not immediately obvious and is strongly linked to the outcomes being sought. The differing capacities and cost of the differing devices will also affect the final solution.

The results obtained for the scroll centrifuge classifier and high intensity magnetic separator obtained in this project indicate that the employment of this new technology can significantly improve the issues which have been identified as the major problems with a dense medium cyclone for fines circuit.

The use of scroll centrifuges for size classification duties in general usage is somewhat difficult to clearly define. These units appear to have both advantages and disadvantages compared with the devices currently available.

The actual employment of this technology in the coal preparation industry will be an economic decision. The additional cost likely to be required will need to be balanced by the additional and increased value of the saleable coal which would be produced by the introduction of the technology.