SUPERCONDUCTING MAGNETIC SEPARATORS

The fields available with conventional electromagnets are limited to 2 Tesla by the saturation of iron. In addition, the creation of large field volumes is costly in terms of copper, iron and electric power. For these reasons superconducting separator magnets were proposed as early as 1970. This paper deals with the special requirements of cryomagnets in the mineral processing industry, and the relatively slow rate of replacement of conventional magnets. Finally, the impact of the new high-temperature superconductors is examined briefly.


INTRODUCTION
In magnetic separation use is made of the enormous differences in the magnetic susceptibility of minerals ( Fig.   1). A small weakly magnetic particle of mass m and mass susceptibility will experience a magnetic force [2] F = re(X/ O )B. V B where B is the field and V B the field gradient. There are two ways of producing a gradient: i. Open-gradient magnetic separators (OGMS) use the actual windings or poles to create weak gradients and forces that deflect magnetic particles.
ii. High-gradient magnetic separators (HGMS) use a finely divided ferromagnetic matrix to create large enough gradients and forces to capture magnetic particles.
The design of superconducting open-gradient magnets is difficult, since magnetic flux must be both large and rapidly varying, external to the cold region. Opposed current windings are generally used, as in cusp-mode split pair coils or "racetrack" linear dipoles [3]. In all cases the gradient is  Table i shows some existing magnetic separators and Figure 2 indicates the regions of applicability in a susceptibilityparticle size continuum. Wet methods are invariably used for particle sizes below 75 /m [5].
MAGNET,S,,,FOR OPEN-GRADI,E.NT S,,PARATION We shall illustrate the requirements by a simple straight conductor of circular cross-section carrying a current I.
The field at an external point r is OI/2 rr and the field gradient is B/r. The force constant is thus For I = 1 MA and r = 0.1 m the field will be 2 T and the force factor 40 T=/m. A high force factor arises from: i. high current density in the conductor ii. small thermal insulation distance between the winding and the outside world.
A single winding is impractical and current practice entails dipole-type windings with a separation distance 2b.
In this case the gradient is of the order B/b. As b is increased the force is reduced but the height of the field region is increased, so that a practical compromise must be reached [6].
There are two ways to use this force: i. In a "falling curtain" separator the particles fall under gravity past the magnet (see Fig. 3), the mags being deflected inward [2,3].
ii. In a drum separator the mags are held against a rotating drum, while the non-mags are thrown off under the influence of gravity and rotation [7,8]. This is illustrated in fig. 4.
Both designs have advantages and disadvantages, but in general, it may be said that falling curtain separators are better suited to fine particles (0.1 to 1 mm), while drum separators are best for larger particles (up to 10 mm). In dense streams of small particles collisions play a dominant role and the deflection is reduced to (2so )I/,, where is the "collision mean free path".
For particles of mean size of 0.1 mm the effect is to reduce the deflection by a factor of about 3 [9].
The capacity of such a separator is where L is the curtain length, t its width, D the mineral density r the relative particle density and v the stream velocity, t is limited to so or s and typical values are 5 to 10 t/h per meter of magnet length. The presence of collisions leads to a deterioration of the magnetic product, so that further passes may be required [i0], In drum separators of drum length L and drum speed V the capacity for particles of average size d is where N is the number of turns and i the current. As a practical design we consider a solenoid with L = D = 0.5 m. and a current of 100 A. The central field will be 4 T for 22 500 turns, or a total wire length of 36 km. The inductance will be 170 H and the stored magnetic energy 850 kJ (see [ii] for details).
The chief limitation is the maximum single turn hoop stress JBD/2. where J is the current density in the wire and B the maximum field in the winding (a little greater than Be). An acceptable limit for NbTi-Cu composite is about 250 MPa [11].

Capacity of matrix separ,.a.t, ors
The capture and retention of magnetic par+/-icles is governed mainly by the matrix and the flow conditions [12]. An upper limit for the slurry velocity through the matrix is v. = dfDXBB=II8 a (7) where is the slurry viscosity. Typical values range from 1 to 100 cm/s. The lower values are encountered in kaolin processing, where the particle size is below 5 tm [5,13].
The capacity of a matrix separator is where A is the cross-sectional area of the matrix canister, w is the solids density of the feed, v the slurry velocity and F the duty factor or the fraction of time the separator is receiving feed. If t, is the feed time and td the "dead time", (10) Figure 5 shows the behaviour of Q with w and t. A small dead time can be achieved in three ways: i. by ramping the field up and down rapidly ii. by using a "reciprocator" (see Fig. 6). where t is limited to the time to interchange matrix canisters iii. by using a continuously moving matrix, e.g. carousel.

J. KOPP
The advantages of super.c.onducting magnets From (7) it appears that the magnetic velocity increases linearly with the applied field, so that doubling the field should allow one to double the slurry velocity and hence the capacity.
The B/v rule seems to apply to kaolin processing [14], but has not yet been fully accepted for other mineral separation problems [15].
The higher velocity can only be utilized fully if the duty factor is high. This implies a reciprocating or a continuously moving matrix, where the dead time is small [14].
It is possible to sweep a superconducting magnet fast (in about one minute or less) albeit at the cost of: i. high helium boil-off due to hysteresis in the superconducting filaments and eddy currents in the connecting matrix (see below) ii. large power supply to overcome the back emf of a high inductance. The power required to ramp a magnet of stored magnetic energy E in a time T is 2E/T. For E = 850 kJ and T = 60 s, this is 28 kW.

RUNNING COSTS OF SUPERCONDUCTING MAGNETS
A large kaolin separation electromagnet may consume as much as 500 kW electric power, all of which is converted into heat. The main advantage of superconductivity is the lack of power dissipation in the magnet at fixed field. This must be paid for by the cooling of the magnet with liquid helium.
For this reason it is of paramount importance to minimize the heat leaks into the cryostat.
These heat leaks arise from three main causes: i. Radiation losses into the helium space, which are proportional to the fourth power of the shield temperature. Liquid nitrogen-cooled shields are generally used, but in addition there may be helium vapour-cooled shields at about 30 K or mechanical shield coolers. By careful design it is possible to reduce the heat inflow to less than I0 mW/m [16].
it. Conduction losses from the neck of the cryostat and the magnet supports can be minimized by careful design and the use of nylon supports.
A I00 kg magnet capable of withstanding axial and radial forces of the order of I0 kN has conduction losses of the order of 50 mW.
iii. Current leads are a major contribution. Properly optimised vapour-cooled leads contribute about 2.1 mW/A. The zero-current value is about 40 per cent of this value and the overall boil-off depends on the duty factor of the magnet [11].
A superconducting switch allows the magnet to operate in "persistent mode". The current leads may then be detached so that the boil-off is greatly reduced. This dissipation in the switch can be made a small fraction of the stored magnetic energy in the magnet [11].
If a magnet has to be ramped frequently from zero to a field B and back again, the main dissipation contributions are: i. Hysteresis in the filaments, which can be reduced below 10 kJlm by reducing the filament size below 10 ,m [11].
ii. Eddy currents in the connecting matrix, which is reduced by twisting the filaments, increasing the matrix resistance by alloying and reducing the ramp rate. For ramp times of the order of one minute this contribution is normally small [11].
For the solenoids described above the boil-off is between 20 and 50 cm" per cycle. For a small system or a system that is used only occasionally simple bath cooling offers all advantages of reliability and low cost. The cost of delivered liquid helium varies from US$ 3 to $ 15 per liter, depending on quantity and locality, and liquid nitrogen costs are an order of magnitude lower.
For larger systems a closed cycle refrigerator or separate liquefier is generally required [16]. Table II shows the reliability requirement pertaining in the mineral processing industry.   The KHD DESCOS drum separator magnet was designed by KKA and built by Siemens [8]. The field was not high enough and the design of the windings ensured a low gradient, so that the machine is only really suitable for large particles with fairly strong magnetic properties.
This and the high price has denied it a market for a long time, but it was finally installed in a Turkish chromite mine to remove iron impurities. it.
The CCL CRYOFOS is a linear dipole racetrack magnet designed for the dry separation of diamagnetic apatite from paramagnetic phlogopite in the size range 0.I to 0.4 mm. It suffered from repeated quenches, probably resulting from electro-mechanical stress [18]. The magnet was notable for its extremely small thermal insulation distance (I0 mm between winding and room temperature}. iii. The Eriez 84" iron-clad solenoid is to a traditional design, but with very fast 60 s ramp time to 2 T. This enables one to obtain a reasonable duty factor without resorting to mechanically complicated reciprocating matrix solutions [19]. The user seems very satisfied with the system and Fig.  7 shows a cost comparison with a conventional separator.  [22]).
J. KOPP iv. The Czech reciprocating matrix kaolin separator has been in operation for over two years and seems to work well [20]. Its economic viability is questionable at this stage. Fig. 6 shows the reciprocator principle where active matrix canisters are moved between "dummy" canisters to reduce the out-of-balance forces. v.
The smaller Oxford Instruments magnet is used for pyrite and ash removal from pulverized coal [21]. The matrix canister is expelled in full field in less than 3 s using a secondary winding through which a current pulse is passed. The acceptability in the very conservative mining industry will be greatly improved.
To estimate potential cost reductions using ceramic superconductors is extremely difficult at this stage.
It is not known whether commercial production of wires or tapes is feasible and what current-carrying capacity of the new superconductors will be. Figure 8 shows  as a function of relative current density and unit material cost.
The relevant parameter is the stored magnetic energy . = VB=/21o (9) where V is the field volume and B the average field. Figure   9 shows a schematic diagram of the capital cost versus stored magnetic energy It can be seen that for small systems permanent magnets are best, while for energies in excess of 1 M3 superconducting magnets are best. Considering the low running costs of permanent and superconducting magnets it appears that the range of applicability of conventional electromagnets is being severely squeezed The major factor is the acceptability of cryomagnetsin the industry, which improves as more systems are introduced and their reliability is demonstrated. Table  5 summarizes some of the current applications of magnetic separation.