Ind. Eng. Chem. Res. 1996, 35, 4257-4263
4257
Gravity-Enhanced Magnetic (HGMS) Coal Cleaning S. Zhou,* E. S. Garbett, and R. F. Boucher Department of Mechanical and Process Engineering, University of Sheffield, Sheffield, U.K.
For economic and environmental reasons, precombustion magnetic coal cleaning must be performed on dry pulverized coal. Gravity-enhanced high-gradient magnetic separation has been successfully applied to the removal of mineral impurities from coal with a 4-T superconducting solenoid magnet. Under optimum separation conditions obtained in this study, this technique effectively cleans up to 72 wt % of the pyritic sulfur and 44 wt % of the ash content from a typical pulverized coal in a vertically upward airstream rig, and the heating value recovery is 95 wt %. Increasing applied magnetic fields and separator matrix lengths and decreasing superficial gas velocity each individually improve the cleaning performance. The results have indicated that the matrix made of 0.360-mm circular AISI 430 stainless steel wires is the best one for all the matrix grades used. These tests have paved the way for pilot plant trial runs at the power station site to be conducted. 1. Introduction The use of high-gradient magnetic separation (HGMS) for the cleaning of pulverized coal depends on the difference in the magnetic moment associated with mineral particles and that of coal. Since the difference in magnetic susceptibility of the impurities and coal is minute, the separation needs the combination of an intense magnetic field and a large field gradient. The attractiveness of the process lies in the fact that coal is diamagnetic but most of the mineral impurities, including pyritic sulfur, are weakly paramagnetic. In principle, therefore, the magnetic impurities can be removed from the nonmagnetic coal. In practice, this means that the impurities must be liberated from the coal by grinding, as is normal in the combustion of pulverized coal. Pyrites are imbedded throughout the lignite as discrete tiny particles (Zhou et al., 1995a), while ashes are distributed throughout the coal by way of discrete layers and cavity fillings. Imperfect liberation leaves coal associated with the minerals, and even if these are completely separated, some coal is lost. Failure to fully appreciate the importance of this liberation was the major factor in the disappointing results of most previous investigations (Trindade, 1973; Lin et al., 1976; Burnley and Fells, 1985). In these studies, a high degree of removal of inorganic sulfur could only be achieved at the expense of high loss of heating value because so much coal was still bound to the sulfur. The study of Burnley and Fells, although apparently unsuccessful, did indicate that adequate liberation might be the key solution to the problem. A more recent study (Lua and Boucher, 1989) has pointed out that good pyritic sulfur removal and ash reduction can be successfully achieved. Numerous factors influence the efficiency of the process including field intensity, wire mesh material, size and number of wires, particle size, and particle velocity. In particular, low pyritic velocity is desirable to improve the probability of capture, while high coal particle velocity diminishes the chance of its mechanical entrapment on the mesh wires. The major part of the coal loss was probably the result of mechanical entrapment of coal particles on the wires of the separator matrix, together with some capture of composite particles of coal and pyrite. The study also * Author to whom correspondence should be addressed at Edwards High Vacuum International Ltd., Southfield Road Trading Estate, Nailsea, Bristol BS19 1JW, U.K.
S0888-5885(95)00568-9 CCC: $12.00
indicate that the separator matrix could not be heavily loaded without reducing the separator performance. Wet and dry methods have been used in high-gradient magnetic separation for the desulfurization of pulverized coal. The dry methods may be desirable because they require the lowest initial capital investment and have the lowest maintenance costs of all currently used methods of upgrading fine coal. Some comments (Fine et al., 1976) are acceptable on the practicality of thermal processing, such as pyrosis, flash roasting at low temperatures in an inert atmosphere, followed by magnetic separation. The thermal processes are to transform a small portion of the paramagnetic pyritic particles to ferromagnetic pyrrhotite to facilitate removal of sulfur from pulverized coal by magnetic separation. The experimental results available are, however, too incomplete to warrant the drawing of any conclusions. Increasing the magnetic field intensity increases the percentage of reductions of pyritic sulfur, total sulfur, and ash contents. To generate sufficiently large background magnetic fields of, say, more than 2 T, over the volumes required for the high-gradient magnetic cleaning of dry pulverized coal, only superconducting magnets are currently practical. With the approach of hightemperature superconductors, much simpler and cheaper liquid-nitrogen-cooled magnets can be anticipated. A simple system of matrix changing has been devised and tested in which the separator matrix is moved in or out of the field in a few seconds by large forces generated by passing a pulse of microcurrent through the separator matrix in the high magnetic field (Male and Lal, 1988). Alternatively, a carousel separator matrix can be incorporated to rotate in and out of the magnetic field to provide for continuous cleaning operation. It is, therefore, quite acceptable that superconducting magnets could be used to design a industrial system having a capacity of around 50 ton/h with the separator matrix having to be cleaned on a cycle of between 1 and 3 min. The present study reported here seeks to bring about improvements of matrix loadability and reductions of coal mechanical entrapment. These two improvements would lead to longer cycle times between matrix cleaning (an important operational factor) and reduction in heating value losses (an important economic factor) while achieving major pollution with retrofittable technology which does not add to the waste-disposal burden. The geological processes that form coal can also concentrate trace elements in the coal. When coal is © 1996 American Chemical Society
4258 Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996
burned, trace elements can be further concentrated. Many trace elements in coal are associated with particular mineral matter species (Finkelman, 1980). Arsenic and mercury are commonly associated with pyrite, cadmium is associated with sphalerite, and selenium is associated with lead selenide, pyrite, and other sulfides. The high-gradient magnetic separation can potentially clean at least some of the trace elements associated with specific minerals, thereby reducing the release of these elements into the atmosphere. Indeed, if the technology was adopted on a significant scale, the magnetic “product” would warrant investigation as a source of some scarcer minerals and of sulfur itself. Further improvement in ash-reduction efficiency should also be useful in coal beneficiation for chemical feed stock and other applications previously impeded by mineral impurities, such as coal-burning diesel engines and high-quality coke. 2. Rationale High-gradient magnetic separation is a magnetic extraction process in which the fundamental element is a ferromagnetic wire, located transversely in a strong uniform background magnetic field. The magnetic separation equipment consists of numerous extraction units formed by the interaction between the magnetized wires and fine paramagnetic particles suspended in the flowing airstream in an intense applied magnetic field. The interaction is governed by the magnetic tractive force on the paramagnetic particle, particle inertia and gravity, and air viscous drag force and buoyant force. A simple single wire of radius a is assumed to be magnetized to saturation M B s and placed transversely to an intense background field H B , large enough to saturate the wire. The airstream has a viscosity µ, a permeability of free space ν, and density Ff and flows with a uniform superficial gas velocity U B 0. The magnetic tractive force F Bm, the viscous drag force F Bs, and the sun of air buoyant force and gravitational force F B g, acting on a small spherical paramagnetic particle of susceptibility B χ, radius b, density Fs, and density U B s, can be expressed as
2 Bb3∇(H F Bm ) πνχ B ‚H B) 3
(1)
1 F Bs ) πCb2Ff|U B0 - U B s|(U B0 - U B s) 2
(2)
4 F Bg ) πb3(Fs - Ff)g b 3
κ)
(
)
2Rθ 2Rθ 12 πa πa2
(5)
However, the correction will be neglected here. The number of paramagnetic particles extracted in the second mesh can be expressed by the probability of capture in the second mesh times the probability of passage through the first mesh as n2 ) κn(1 - κ). The total length of the separator matrix is given by L ) m(a + σ) if σ is the symbol of the gap between two matrix meshes. Based on the design of the separator matrix, a < 1 mm, σ e 4 mm, m g 15, and L . m(a + σ), the number of paramagnetic particles captured on the mth mesh wires is
nm ) κn(1 - κ)m-1 = nκ exp(-mκ) + O[(a + σ)2] (6) (3)
Every paramagnetic mineral particle always experiences such a force balance between the forces above and the particle inertia force F Bi as
dU Bs 4 )F Bm + F Bs + F Bg F Bi ) πb3Fs 3 dt
For a paramagnetic particle moving initially parallel to the background magnetic field H B at a radial distance r g a, the particle may still be able to be captured by the magnetized wire. When a particle is outside the “shadow” of a wire, the outcome depends on the magnetic capture force overcoming competing drag force and particle inertia (which is reduced in upward flow). It can be imagined that there exists a value of coordinate r ) R such that if the initial position r for a paramagnetic particle satisfies r e R, then the particle will be captured. Thus, the capture cross section per unit length of the magnetized wire is 2R. It is now appropriate to consider the cleaning performance of the separator matrix. The separator matrix consists of a number of magnetizable metal meshes equally layer-built in series, located in the bore of a superconducting solenoid magnet. First, consider a mesh layer located perpendicularly to the intense uniform magnetic field and the airstream flow. The length of the wire in one mesh is θ/πa2, if the filling factor of the mesh per unit area is θ, which is the volume occupied by the mesh wires per unit area. The length of the magnetized wire is perpendicular to the background field and air flow and, therefore, effective in the extraction process. The effective capture area the mesh per unit area can present to the suspension flow is therefore κ ) 2Rθ/πa2. The number of particles extracted in the first mesh layer will be n1 ) κn, where n is the total number of paramagnetic particles. This expression is only approximately correct if θ and R are large for the separator matrix made of multimesh layers due to the overlapping of capturing areas. The firstorder correction replaces κ ) 2Rθ/πa2 by
If the separator matrix is of m mesh layers total, then the total number of paramagnetic particles extracted on the separator matrix nex is given by m
nex ) (4)
As contrasted with the vertically downward airstream (Lua and Boucher, 1989), the net gravitational force is used to assist capture of the mineral particles in the vertically upward test rig. Some advantages which will issue from conducting test operations in a vertically upward airstream are larger capturing and holding forces, which should be of significance in improving capture of mineral particles and separator matrix loadability.
m
∑ nm ) nm)1 ∑ κ exp(-mκ) = n∫0 m)1
mk -x
e
dx )
n(1 - e-mk) (7) The capturing efficiency by number of the separator matrix is found to be
η)
(
)
nex 2mθR ) 1 - exp n πa2
(8)
Equation 8 suggests that increasing the length of the separator matrix, i.e., increasing the number of mesh layers m, the filling factor of the mesh per unit area θ,
Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 4259
Figure 1. Distribution of axial magnetic field over axis in the 264-mm bore of superconducting magnet.
i.e., more wires for every mesh, and fine mesh wires improve the separator performance. Increasing the applied magnetic field intensity can enlarge the capturing diameter R of the magnetized wire and, hence, improves the capturing efficiency, although the capturing diameter is closely related to the matrix grade. 3. Experimental Superconducting solenoid magnets can provide very high magnetic intensities with the persistent mode. The superconducting magnet system (Cryogenic Consultants Limited, London) used in the investigation comprises of a 4-T superconducting solenoid housed in a low loss aluminum/glass fiber cryostat and provides a 264-mmdiameter room-temperature access bore. The magnet is of a homogeneous field zone, in which the distribution of axial magnetic field over axis in the 264-mm magnet bore is presented in Figure 1. The field homogeneity is within (2.0% at 200-mm diameter × 400-mm long volume. The terminal velocity in air of a 100-µm spherical particle is 1 m/s for pyrite and 0.35 m/s for coal particle. In a vertically upward airstream of air velocity, 1.50 m/s, the pyrite particle would travel at a velocity of 0.50 m/s, while the coal particle would be traveling at almost twice its terminal velocity, over 1 m/s. Some suggestive benefits can thus be anticipated to accrue from conducting tests in a vertically upward airstream experimental rig, producing the resulting enhanced probability of entrapment of slower moving pyrite. The schematic diagram of the recently designed rig is given in Figure 2. A fluidized-bed aerosol generator consisting of a vertically vibrating system for feeding pulverized coal into the magnetic separator was used to generate a steady-state aerosol (Zhou et al., 1995b). Polydispersed coal particles from the aerosol generator were discharged into the bell-mouth inlet of a vertical copper pipe of 101.6-mm inner diameter where they were sucked in with ambient air. Prior to magnetic cleaning, these aerosolized coal particles mixed with upward air flow were drawn in by a centrifugal fan through the superconducting magnet bore and separator matrix. This consists of a number of magnetizable metal meshes equally layer-built in series, located in the 264-mm bore of a superconducting solenoid magnet. Pyrites and other paramagnetic minerals in coal were captured on the magnetized wires of the separator matrix, forming
Figure 2. Schematic diagram of gravity-enhanced magnetic separation test rig for pulverized coal cleaning.
the fraction called mags. The tails were recovered in a Whatman glass microfiber, GF/D, filter with 2.7-µm retention, and the clean air was then vented to the atmosphere after bag collectors. A tee branch is inserted in the duct downstream of the membrane filter holder to accommodate a mechanical damper to provide flow control. A butterfly valve is fitted in the copper duct after separator matrix to provide additional flow control and avoid the tails backtracking during collection of the mags on the matrix. Two sampling ports, each with an access hole of 19mm diameter, are provided on the vertical test section upstream and downstream of the separator matrix for the insertion and support of the pitot tube or the sampling tube necessary for gas velocity measurements and coal particle analyses, respectively. These four ports provide two perpendicular traverses of the sampling tube across the cross section of the test duct. Two sets of four equally spaced wall-static tappings are also positioned upstream and downstream of the separator matrix to measure the pressure drop across separator matrix. The flow rate and superficial gas velocity in the test pipe can be measured by a calibrated ISA 1932 nozzle downstream of the membrane filter. Coal particle sizes, before and after separator matrix, can be accessed in situ through two sets of four equally spaced glass windows by an optical array particle spectrometer (OAP-230LS, Particle Measuring System) upstream and downstream of the separator matrix. This laser-driven spectrometer measures particle sizes in the 10-300-µm range with 10-µm size resolution. The present work aims at treatment of precombustion pulverized coal in a vertically upward airstream rig, the size of the coal being as close as possible to that produced by the pulverizing mills in electricity-generating stations (75 wt % < 75 µm and 100 wt % < 100 µm).
4260 Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 Table 1. Particle Size Distribution of a Feed Coal (Coalfield Farm North)a
Table 3. Analysis of Feed Coal (Coalfield Farm North) on a Dry Basis
particle size, µm
wt under, wt %
size band, µm
wt in, wt %
5.8 7.2 9.1 11.4 14.5 18.5 23.7 30.3 39.0 50.2 64.6 84.3 112.8 160.4 261.6 564.0
0.0 0.1 0.3 1.1 2.5 5.1 11.0 21.5 33.5 47.6 63.7 77.9 90.1 97.9 99.9 100.0
5.8-7.2 7.2-9.1 9.1-11.4 11.4-14.5 14.5-18.5 18.5-23.7 23.7-30.3 30.3-39.0 39.0-50.2 50.2-64.6 64.6-84.3 843.3-112.8 112.8-160.4 160.4-261.6 261.6-564.0
0.0 0.2 0.8 1.4 2.6 5.9 10.4 12.0 14.2 16.0 14.2 12.2 7.8 2.1 0.1
a
Mean particle size (by mass) ) 52.32 µm. Median particle size (by number) ) 62.37 µm. Particle size (10 wt %) ) 22.74 µm. Particle size (90 wt %) ) 112.54 µm. Table 2. Particle Size Distribution of a Tail Cleaneda particle size, µm
wt under, wt %
size band, µm
wt in, wt %
5.8 7.2 9.1 11.4 14.5 18.5 23.7 30.3 39.0 50.2 64.6 84.3 112.8 160.4 261.6 564.0
0.0 0.0 0.0 0.2 0.7 2.4 7.1 16.6 29.3 43.8 60.2 77.0 90.0 97.0 99.9 100.0
5.8-7.2 7.2-9.1 9.1-11.4 11.4-14.5 14.5-18.5 18.5-23.7 23.7-30.3 30.3-39.0 39.0-50.2 50.2-64.6 64.6-84.3 843.3-112.8 112.8-160.4 160.4-261.6 261.6-564.0
0.0 0.0 0.1 0.5 1.7 4.8 9.5 12.7 14.5 16.4 16.8 12.9 7.1 2.8 0.2
a Mean particle size (by mass) ) 55.62 µm. Median particle size (by number) ) 65.72 µm. Particle size (10 wt %) ) 25.68 µm. Particle size (90 wt %) ) 113.01 µm.
Most previous investigations on magnetic separation have been with large particle sizes. Failure to fully appreciate the importance of the liberation of the impurities from the coal was the major factor in the disappointing results of the most early investigations. Lin et al. (1976) concluded that, with fine coal in an air stream, agglomerations can impede desulfurization. Burnley and Fells (1985) also reported that coals below 125 µm could not be separated magnetically as the particles tended to coalesce. Using a larger (125-212 µm) particle size range, they found incomplete liberation of pyrite and mechanical trapping of coal problematic and removed a maximum of 37 wt % pyritic sulfur. The feed coal was supplied from an open-cast mine known as Coalfield Farm North, Burton-on-Trent, England. The rationale for initially using Coalfield Farm North coal is that its analysis is well documented and it permits direct comparison of new results with previous data. The as-mined coal was jaw crushed, milled to
