Maximizing recovery, grade and throughput in a single stage Reflux Flotation Cell

Highlights

• Interplay between product grade, recovery and feed throughput was examined using the RFC.

• Grade and recovery were maintained using volumetric feed fluxes up to 7 cm/s.

• Efficient technical separation demonstrated at elevated volumetric throughput.

• RFC results aligned well with the theoretical separation indicated by Coal Grain Analysis.

Abstract

The Reflux Flotation Cell (RFC) utilises the Boycott Effect to decouple the overflow water flux from the gas flux, permitting in principle high product grade and recovery at a vastly higher volumetric feed flux. This study investigated this relationship between concentrate grade, recovery, and volumetric feed throughput using a single flotation stage and feed fluxes spanning 1–9 cm/s, well beyond that used in conventional flotation. Coal flotation tailings and hydrocyclone overflow provided convenient representations of “binary” feeds for the experiments, constituting liberated hydrophobic and hydrophilic particles. The results demonstrated robust recoveries through the preservation of the gas to feed flux ratio with increasing feed flux, while minimising the gas flux strengthened the capacity to maintain high product grade using inverted fluidization water as the wash water. Remarkably, a high product grade (low product ash%) was maintained over the extreme feed flux range by ensuring a net downwards flux of wash water delivered through the upper fluidized bed of bubbles. Coal Grain Analysis (CGA), an optical imaging technique, identified the maceral composition of the feed particles and validated, with close agreement, the RFC steady state separation performance. Indeed, under continuous operation the RFC data demonstrated an overall positive shift in performance relative to that of the standard tree flotation curve. The findings showed strong preservation of product grade and recovery using a single RFC stage, over a seven-fold increase in the feed flux relative to conventional flotation systems.

Introduction

Historically, flotation cell volumes have grown exponentially, with mechanical flotation cells now more than 600 m³ in volume (Govender et al., 2014, Lelinski et al., 2015, Shen et al., 2019, Suhonen et al., 2019, Yang et al., 2019). These changes, which provide improved economy of scale, reflect the fact that flotation processing rates on a per unit area basis have changed little in 100 years, languishing at about 1 cm/s, and the fact that feed grades have declined significantly. There is a clear trend towards larger flotation cells, which will likely continue into the future (Wills & Finch, 2016). Plainly, there is a need to develop technology advances that can lower the foot-print and simplify the design of the processing circuits.

Previous studies concerned with the performance of the Reflux Flotation Cell (RFC) have focussed on two distinct applications. The earlier studies investigated its cleaner application, with emphasis on obtaining high-grade concentrate at a modest feed flux and using inverted fluidization water (Galvin and Dickinson, 2014, Galvin et al., 2014). Later studies examined its performance as a rougher stage, exploiting its enhanced bubble–liquid segregation capacity and flotation kinetics to achieve elevated volumetric feed fluxes, with no focus on the product cleaning (Dickinson et al., 2015, Jiang et al., 2019). Naturally, the focus shifted to the performance of a two-stage, rougher-cleaner circuit, with a feed flux of order 10 cm/s across the circuit (Jiang et al., 2016), equivalent to an effective feed flux of 5 cm/s per stage. Here, the processing rate was of order 5 to 10 fold higher than used by conventional flotation cells.

There are major processing and flow-sheet advantages that emerge by simplifying the two-stage concept into a single stage, provided the recovery and a clean, high-grade product is maintained. Recently, Cole et al. (2020) investigated the single stage flotation performance of the RFC using a “binary” feed of liberated hydrophobic and hydrophilic particles derived from a fine coal tailings stream. That study, which was limited to a feed flux of nominally 1 cm/s, achieved levels of product cleaning well beyond that achieved using the standard tree flotation result, matching the predictions of an optical imaging technique known as Coal Grain Analysis (CGA). This paper extends that work significantly, by investigating changes in the recovery and product grade as the volumetric feed flux increases by nearly an order of magnitude. This extreme, single-flotation-stage, process intensification has never been previously achieved.

In flotation, there is an inherent hydrodynamic limit to the volumetric feed rate per unit of vessel cross-sectional area (feed flux). In conventional flotation systems, such as mechanical and column cells, this limit is qualitatively associated with the onset of the so-called flooding condition (Fuerstenau et al., 2007, Yianatos, 2007). The flooding condition describes the loss of the pronounced interface between the froth and pulp regions of the cell, where the bubble volume fractions converge to a common value. Failure to preserve a clear discernment between these zones results in ineffective hydrophobic-hydrophilic separation, to the point where the overflow stream resembles the underflow. Thus, to preserve these two distinct zones and avoid flooding, the conventional process is constrained to feed fluxes and gas fluxes of a relatively narrow span. In industry, conventional flotation systems typically employ feed fluxes of order 1 cm/s (Fuerstenau et al., 2007) and gas fluxes in the range of 0.5–2.5 cm/s (Dahlke et al., 2005, Yianatos, 2007, Yianatos and Henríquez, 2007). The system hydrodynamics is scalable by bubble diameters that typically range from 0.5 mm to 2.0 mm (Diaz-Penafiel and Dobby, 1994, Yianatos, 2007, Yianatos and Henríquez, 2007).

The fundamental arrangement of the RFC comprises a main vertical chamber positioned above an inclined section. Internally, parallel inclined channels formed in the lower section provide a mechanism for enhanced gas–liquid segregation capacity through an application of the Boycott effect (Boycott, 1920). This arrangement enables operation at an elevated volumetric feed flux, over an order of magnitude greater than that typical of conventional flotation cells (Jiang et al., 2014, Jiang et al., 2016, Jiang et al., 2019, Dickinson et al., 2015).

Another unique feature of the RFC is the fluidization chamber that formally encloses the top of the cell, delivering a downward flow of wash water through an inverted fluidized bed of concentrated bubbles. An important hydrodynamic characteristic of operating the RFC beyond the flooded state is the distinct absence of a froth phase within the cell, instead forming a concentrated bubbly mixture of approximately 0.5 bubble volume fraction (Dickinson and Galvin, 2014). The decoupling of the gas flux and water recovery in turn permits independent control of the water recovery while also permitting the application of fluidization water to this permeable bubbly zone, providing strong uniform flow of liquid downwards through the top section of the cell. Termed a positive liquid bias flux (Yianatos et al., 1987, Finch and Dobby, 1990), this downward flow affords significant desliming of the product overflow that escapes via a port surrounding a centralised downcomer (Dickinson & Galvin, 2014). Thus, strong counter-current washing of the product ensues.

In this study, two “binary” feeds composed of naturally hydrophobic fine metallurgical coal and hydrophilic gangue particles were utilised as the flotation feed. The feeds were sampled from flotation tailings of conventional cells. Results obtained from a third feed, a thermal coal of high mineral matter content, are briefly presented. Volumetric feed fluxes ranging 1–9 cm/s were investigated to determine the influence of the feed throughput on the recovery and product grade, inferred by the combustible recovery and product ash % respectively. A net downwards flux of clean fluidization water was imposed on the system to reject the slimes from the product overflow.

Evaluation of the separation performance in this study was achieved through comparison with the recovery-grade curves produced by two methods. The first involved the tree flotation method, a recognised industry standard for describing the separation achievable by flotation for a given coal sample. The analysis relies on a series of batch flotation tests using a bench scale mechanical cell to produce fractionated product and reject samples. The procedure is not without criticism of its validity in producing an “ideal” recovery-ash curve however, given the systematic biases of experimental design, methodology and operator dependence, as well as the difficulty in eliminating entrainment of ultrafine clays from the froth product.

The second approach involved the application of Coal Grain Analysis (CGA). CGA provides an optical analysis of each >1 μm particle in a given sample, describing the maceral and mineral composition of each particle analysed by examining the distinct reflectance, texture, and grain associations (Ofori et al., 2006, O’Brien et al., 2007). The CGA method utilises the measured area of each identified maceral or mineral phase to estimate its volume, enabling the determination of their mass simply by utilising the component density (Ofori et al., 2006, O’Brien et al., 2007). By applying a relationship describing the ash content of the maceral or mineral phases identified, a recovery-ash curve can ultimately be generated. Hence, the result is analogous in utility to a mineralogical SEM analysis of a mineral sample (Ofori et al., 2014, Atkinson and Swanson, 2018), and characterises the sample from the particles present without the potential limitations of any physical separation technique. A comprehensive explanation of the CGA technique and its applications is outlined in O’Brien et al., 2003, O’Brien et al., 2007, Ofori et al. (2006), and Atkinson and Swanson (2018).

Cole et al. (2020) introduced the approach of Coal Grain Analysis (CGA) to their assessment of the RFC. The same approach is used here in assessing the potential to achieve the strong cleaning quantified by Cole et al. (2020), however, at greatly elevated volumetric feed fluxes.