Jan Svoboda. Magnetic Techniques for the Treatment of Materials

Подождите немного. Документ загружается.

390 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Figure 5.65: The recovery of uranium from cyanidation residues, and the grade

of the concentrate, as a function of matrix loading O

(adapted

from [S61]).

that are occasionally used are expressed by eqs. (5.10) to (5.12).

Feed mass

Matrix area

[kg/m

or g/cm

] (5.10)

Feed mass

Matrix volume

[kg/m

or g/cm

] (5.11)

Mags mass

Matrix volume

[kg/m

or g/cm

] (5.12)

Figure 5.63 illustrates the e!ciency of removal of iron from kaolin, as a

function of the volume of the kaolin slurry processed. The deterioration of

the matrix performance with an increasing number of bed volumes processed

is marked. Similarly, the matrix loading curve for beneﬁciation of silica sand,

shown in Fig. 5.64, illustrates a trade-o between the quality of the ﬁnal product

and the throughput per cycle.

In the concentration of minerals, matrix loading aects not only the recovery,

but also the grade of the product. As a rule, with increasing loading, the

recovery decreases and the grade of the concentrate increases. This trend is

demonstrated in Fig. 5.65, where the recovery of uranium from cyanidation

residues, and the grade of the magnetic concentrate are shown. Similarly, in

beneﬁciation of siderite iron, illustrated in Fig. 5.66, the same behaviour is

observed.

5.3. WET MAGNETIC SEPARATION 391

Figure 5.66: The recovery of iron from siderite ore, and the grade of the mag-

netic concentrate, as a function of matrix loading O

(adapted from

[H22]).

The trends described above can be explained by realizing that, at a higher

matrix loading, the magnetic and entrained "non-magnetic" particles or unlib-

erated grains in higher layers of the material build-up are not held su!ciently

strongly and can be stripped o from the matrix by ﬂuid shear stress. With in-

creasing matrix loading, the importance of the erosion wash-o force increases,

in comparison with the magnetic attractive force. Although such a process re-

duces the recovery of magnetic particles, the grade of the magnetic fraction and

the selectivity of the process is improved.

We have seen that matrix loading is an important parameter that allows a

magnetic separator to be operated in a regime of maximum recovery or maxi-

mum grade of a concentrate. Matrix loading also determines the throughput of

a separator, the duration of a duty cycle of a cyclic separator, and the rotational

velocity of a continuous separator. Typical values of matrix loading for various

applications are summarized in Table 5.26.

Magnetic properties of matrix material

There is ample experimental evidence that magnetization of the matrix mater-

ial plays a minor role in the metallurgical performance of a magnetic separator

[S1]. This is in contrast to predictions of simple single-collector single-particle

theoretical models, described in Section 3.4.1. On the other hand, these exper-

imental ﬁndings do agree well with predictions of theory of HGMS based on

deep bed ﬁltration approach [S31, S33]. As has been discussed in Section 5.3.3,

392 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Table 5.26: Matrix loading in dierent applications.

Application Matrix Matrix loading (O

)

Hematite ore grooved plates 180 to 300

Siderite ore rods 200 to 600 (O

)

Steel mill waste water steel wool 9000

Power plant cooling water steel wool 4000

Uranium leach residues mesh or balls 300 to 3000

Coal desulphurization steel wool 50 to 250

Silica sand steel wool 350 to 700 (O

)

Kaolin puriﬁcation steel wool 1to20bedvolumes

the modelling of the magnetic ﬁeld pattern around a matrix collector has shown

that the dierence between magnetic induction generated by matrix materials

with polarization equal to 1 T or 2 T is, for practical purposes, negligible.

Experimental investigation [S63] of the eect of the saturation polarization of

the matrix material on the metallurgical performance revealed that, for matrices

with magnetic polarizations greater than 0.5 T, the curves of recovery and grade

versus magnetic ﬁeld were essentially identical. For a weakly magnetic steel

matrix with a polarization of 0.1 T at the working magnetic ﬁeld, the recovery

was strongly dependent on the external magnetic ﬁeld. At the background

magnetic induction of 1 T, the recovery approached that achieved with strongly

magnetic matrices. The results are shown in Fig. 5.67.

Matrices with very feebly magnetic properties (such as austenitic stainless

steel or aluminium) can be used, for example, for magnetic scalping of ferro-

magnetic impurities from a pulp. The aluminium matrix was used by Riley and

Watson [R21] for ﬁltration of ferromagnetic particles and the beneﬁcial eect of

such matrices on the recovery of uranium was demonstrated by Watson et al.

[W23] and by Svoboda [S61].

Clogging of the matrix

Probably the most important problem encountered in pilot or production-scale

HGMS separators is a permanent retention of particles in the matrix. Indus-

trial slurries always contain various extraneous materials (wood, plastics, tramp

iron), oversize grains and ferromagnetic admixtures. Their presence and fre-

quent shut-downs of the separator for the replacement of a blocked matrix caused

by such particles, severely impair the economic and metallurgical performance

of the unit.

The basic measure to eliminate matrix clogging is to avoid straining using

su!ciently coarse matrices compatible with size distribution of the solids in the

feed. Preliminary screening, low-intensity magnetic scalping or desliming to

remove ﬁne ferromagnetic particles, are steps that help to minimize retention of

particles in the matrix.

However, none of these measures completely eliminates the clogging of the

5.3. WET MAGNETIC SEPARATION 393

Figure 5.67: The recovery and grade of the magnetic fraction of siderite ore

versus magnetic induction, for strongly (M

= 2 T) and weakly (M

= 0.1 T) magnetic matrices (adapted from [S63]).

Figure 5.68: The principle of a continuous ball cleaning system (after Corrans

and Levin [C20]).

394 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Figure 5.69: Removal of hematite particles from a ball matrix as a function of pH

of the ﬂush water, for various volumes of the ﬂush water (adapted

from [S64]).

matrix and several e!cient long-term solutions to the problem have been in-

troduced over the years. Continuous removal from a separator and external

cleaning of steel balls, proposed by Karmazin [K21] and developed by Corrans

and Levin [C19, C8], was successfully used in several production-scale applica-

tions. The principle of the system is shown in Fig. 5.68.

A back-ﬂushing system [A30] that was introduced into the cyclic and con-

tinuous Sala separators was later improved by introduction of a vertical rotor

with matrix reversal and back-ﬂush [C7]. Introduction of a ﬂexible self-cleaning

matrix with back-ﬂush [C18, S14] and of slurry pulsation in vertical rotor sep-

arators [Y2, X1], further reduced susceptibility of matrix to blockage.

Removal of particles from the matrix It was observed by Svoboda and

Corrans [S64] that total wash-o of particles from the matrix could not be

achieved without aecting the surface interactions between the adsorbed parti-

cles and the matrix. It was shown that the removal of hematite particles smaller

than 44 m from the matrix was dependent on the pH value of the ﬂush wa-

ter, as depicted in Fig. 5.69. It can be seen that the removal is not complete,

even under the most favourable conditions. Fractions adhering to the matrix,

after ten bed volumes of ﬂush water had been used, amounted to 8% of the

feed mass. A substantial increase in volume of ﬂush water resulted in only a

marginal improvement.

Flush water is usually pH 7 in practice, and adjustment to pH between

10 and 11 can result in removal improvement by about 7%. Alternatively, a

5.3. WET MAGNETIC SEPARATION 395

Figure 5.70: Removal of hematite particles (- 44 m) from a ball matrix, as a

function of velocity of ﬂush water (pH 7), for dierent volumes of

ﬂush water (adapted from [S64]).

reduction in the volume of ﬂush water, for a constant removal e!ciency, can be

achieved by pH adjustment.

Rinse and ﬂush water

The metallurgical performance of a matrix magnetic separator can be signiﬁ-

cantly controlled by varying the volumes of rinse and ﬂush water.

Rinse water is added at the end of the magnetic head of a continuous sep-

arator or at the end of the feed cycle in a cyclic separator. The rinse water is

used to wash out mechanically entrained non-magnetic particles and unliberated

middling particles.

Flush water is used to remove the magnetic fraction from the matrix. Matrix

ﬂushing is performed outside the magnet station in continuous separators, or

when the magnetic ﬁeld is switched o in cyclic machines.

An increase in the volume of rinse water normally results in a decrease in the

mass of the magnetic fraction, while the grade of the magnetic fraction increases.

These trends are often quite dramatic for small volumes of rinse water, although

beyond a certain limit this eect becomes negligible. The optimum volume

and ﬂow velocity of the rinse water must be determined experimentally since

a wide array of factors aect the inﬂuence of rinse water on the metallurgical

performance of a magnetic separator.

The function of ﬂush water is to remove captured magnetic material from the

matrix. As a result of stray magnetic ﬁelds, remanent magnetism of the matrix

collectors, and surface interactions between the particles and the matrix, the

ﬂow velocity of ﬂush water should be quite high to keep the matrix clean for

396 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Table 5.27: Water consumption (per tonne of feed) in commercial continuous

WHIMS machines.

Separator Rinse water [m

/t] Flush water [m

/t]

Jones 0.5 - 1.0 1.0 - 1.7 [L13]

VMS 0.75 0.80 [C7]

VMS for kaolin 6.0 (total water) [Z3]

Reading 0.54 1.0 - 1.4 [F22], [M29]

6-ERM-35/135 2.5 (total water) [U2]

SALA HGMS 2-3 1 - 2.5 [F22]

Ferrous Wheel 1 (total water) [A36]

Table 5.28: Consumption of wash water in cyclic HGMS.

Consumption of wash water expressed in terms of

Application Bed volumes % of processed slurry volume

Power plant cooling water 11 0.01 - 0.02

Kaolin beneﬁciation 10 125 - 250

asu!ciently long time. Some typical values of rinse and ﬂush water volumes,

employed in commercial continuous matrix separators are summarized in Table

5.27, while data for a cyclic HGMS are shown in Table 5.28.

Laboratory-scale investigations [S64] have shown that, while the ﬂow velocity

of ﬂush water has, up to a point, a dramatic eect on particle desorption from

the matrix, the volume of ﬂush water appears to be of secondary importance.

The results are summarized in Fig. 5.70.

5.3.5 Flow velocity of the slurry

The metallurgical performance of a magnetic separator varies with the ﬂow

velocity of a slurry to such an extent that this variable is one of the primary

factors aecting the e!ciency of magnetic separation. Moreover, ﬂow velocity

determines the throughput of a separator and thus aects the economics of the

process.

The velocity of the slurry ﬂowing through the matrix bed determines the

shear stress acting on the particles deposited on the surface of the matrix col-

lectors, and thus controls the recovery and the grade of the products. Simulta-

neously, ﬂow velocity controls the retention time of the slurry in the magnetic

ﬁeld and thus the e!ciency of particle capture onto the matrix.

The superﬁcial velocity y

, given by eq. (5.13), is deﬁned as the average

linear velocity the slurry would have in the separation chamber if no matrix was

present.

(5.13)

5.3. WET MAGNETIC SEPARATION 397

where T

is the volumetric ﬂow rate (in [m

/s]) and V is the cross-sectional area

of the matrix (in [m

]).

The mean ﬂow velocity of the slurry within the matrix is the interstitial

velocity y, which is related to the superﬁcial velocity by

y =



(5.14)

where  is the porosity of the matrix bed.

The eective length of a particle trajectory is greater than the matrix length

O, because the slurry travels a tortuous path O

. The velocity y

along the

tortuous path is

= y

O

(5.15)

If Karman’s assumption O

/O =

2 is used [S65], we obtain



(5.16)

The eect of ﬂow velocity on separation e!ciency

An increase in ﬂow velocity results in a decrease in the probability of particle

collision with matrix, as illustrated by Fig. 3.35. At the same time, the hy-

drodynamic drag on particles already retained on the matrix increases. The

increase in ﬂow velocity, therefore, results in a deterioration in the e!ciency of

magnetic ﬁltration as shown in Fig. 5.71, which illustrates ﬁltration of PWR

cooling water. Removal of magnetite (- 16 m) from an industrial waste water,

as a function of ﬂow rate, is shown in Fig. 5.72.

A similar trend applies to the removal of impurities from mixtures of materi-

als. The lower the slurry ﬂow velocity, the greater is the removal of magnetizable

particles from a mixture. This eect is shown in Fig. 5.73 for coal desulphuriza-

tion by HGMS. Similarly, in kaolin puriﬁcation, the concentration of magnetic

impurities in the product increases, as is shown in Fig. 5.74, and brightness de-

creases with increasing ﬂow rate, as shown in Fig. 5.75. In contrast to magnetic

ﬁltration, the ﬂow rate must be speciﬁed with due regard for product recovery,

and a compromise must be reached between product quality, product recovery

and throughput of a separator.

The selection of the slurry ﬂow velocity in the application of matrix sepa-

ration to concentration of minerals is dictated by whether the main emphasis

is to be placed on the recovery or on the quality of the magnetic product. An

inverse relationship between product grade and recovery implies that in order

to achieve a high recovery, low ﬂow rates should be used, while a high grade can

be obtained at high ﬂow rates. Magnetic separators can, therefore, be run in a

regime of maximum recovery or in a regime of maximum grade of the concen-

trate, or any regime in between. Typical dependences of the recovery and grade

upon the ﬂow velocity are shown in Figs. 5.76 and 5.77.

The rate of decrease in recovery and increase in grade with increasing ﬂow

velocity are determined by the size distribution and dierences in the values of

398 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Figure 5.71: The ﬁltration e!ciency of PWR cooling water, as a function of ﬂow

velocity (adapted from [T7]).

Figure 5.72: The concentration of magnetite (- 16 m) in the ﬁltrate from a

chemical plant e"uent, as a function of ﬁltration rate (adapted

from [S66]).

5.3. WET MAGNETIC SEPARATION 399

Figure 5.73: The eect of ﬂow velocity of slurry on sulphur and ash removal

from coal (adapted from [W24]).

magnetic susceptibilities of the minerals in the slurry, by the strength of the

magnetic ﬁeld and by the length of the matrix.

If the dierence between the magnetic susceptibilities of a useful mineral

and the gangue is small, an increase in ﬂow velocity will result in a sharp drop

in recovery, while the improvement in grade will be small. If the dierence is

great, the drop in recovery will only marginal, as only feebly magnetic gangue

particles will be swept from the matrix, and the selectivity of separation will

improve.

Similarly, if the applied magnetic ﬁeld is su!ciently strong, an increase in

ﬂow velocity will not aect the captured magnetizable particles, while mechani-

cally entrained gangue particles will be stripped from the matrix surface, thereby

increasing the grade of the magnetic product.

When excessive amounts of slimes are present in the feed, recovery usu-

ally drops rather sharply with increasing ﬂow velocity, while the grade of the

concentrate does not improve appreciably.

Although ﬂow velocity is an important parameter which considerably aects

the performance of a magnetic separator, most industrial-scale continuous sep-

arators are not equipped with ﬂow control and their operation is often below

optimum.

5.3.6 Slurry density

Although the slurry density, or the percentage of solids, aects the throughput

of a separator and the matrix loading, its eect on metallurgical results is usually