You are on page 1of 12

Powder Technology, 48 (1986)

161 - 172

161

A Study of the Effect of Operating Variables on the Efficiency of a


Vibrating Screen
N. STANDISH, A. K. BHARADWAJ and G. HARIRI-AKBARI
Department

of Metallurgy

and Materials Engineering,

University

of Wollongong,

N.S. W. 2500 (Australia)

(Received February 26, 1986; in revised form May 12, 1986)

SUMMARY

The efficiency of screening sinter particles


(0.25 - 3.5 mm) and coke particles (0.25 1.94 mm) in a vibratory screen (200 X 600
mm) is studied for a range of operating variables which include flow rate, deck angle,
RPM, oversize in feed and mesh size. Kinetic
approach is used to analyse the data and efficiency is quantified by the kinetic constants
of the individual particle sizes involved. Regression analysis shows that for both materials the effect of feed rate is positive for all
sizes, that of RPM and deck angle is negative
for near-mesh particles and positive for other
particles, whilst that of oversize is reversed,
i.e., the effect is positive for near-mesh particles and negative for smaller particles.
The significance of the results, which show
for the first time the effect of operating variables on the screening of individual size particles in the feed, in modifying partition
curves, is discussed. Comparison of the results with those reported by other workers
in terms of traditional efficiency definitions
is included.

INTRODUCTION

In a recent paper [ 11, we reported the


effect of oversize particles on the screening
efficiency of ferrous sinter particles under
constant operating conditions of a laboratory vibrating screen. The results were analysed by the kinetic equation previously
developed for batch operation [ 21. It was
concluded that the two operations are comparable in all respects if deck-length in continuous screening and time in batch sieving
are interchanged.

The kinetic approach of investigating the


separation efficiency of vibratory screens is
relatively new and no data are available of
the effect of operating variables such as feed
rate, deck angle, RPM and material properties, on the kinetic constants of the various
particle sizes being screened.
The purpose of this paper, therefore, is
to report the results of a laboratory study of
continuous screening of ferrous sinter and
coke particles and the effect of a range of
operating variables on the kinetic constants
of particle screening.

PREVIOUS WORK

The effect of the various process variables


on the efficiency of vibratory screening has
been investigated by numerous authors [3 141. All have found that screening is affected
in a complex way by the operating variables
and interaction of materials properties, screen
type and screen cloth.
The bulk of the investigations were industrially based [3 - lo] with the results expressed in the form of correlations or as
graphs or tables of correction factors to be
used in the calculations of screening efficiency. In most cases, screening efficiency
was defined by d,, of a partition curve but
percentage recovery of the undersize or similar definitions was also used.
In general, the main aim of the industrially
oriented investigations has been to define
optimum operating conditions of a particular
screening practice. However, Richter [ 51,
Karra [ 61 and Rose [ 71 have developed mathematical models which were claimed to be
able to predict performance of an industrial
vibrating screen under diverse operating
@ Elsevier Sequoia/Printed in The Netherlands

162

conditions.
Richters [ 51 model involved
partial probabilities of the materials passage
through the cloth, whilst Karra [6] based his
model on the correlations
of the various
screening and plant data available in the industry, and Rose [7] combined his equations
for the diffusion of blinding material into and
out of the cloth with pilot-plant correlations.
Unfortunately,
it is not known how good
these models have been in predicting screening efficiency in practice. However, Lynch
[lo] in his keynote address at the recent
Coal and Mineral Sizing Symposium
in
Wollongong has concluded that the gaps in
our knowledge are still large and they must
be filled if we are able to design and operate
our plants so that maximum efficiency is
achieved all the time.
By contrast with industrial investigations,
laboratory
studies of continuous
screening
have been few and far between. The earliest
work on the influence of the various operating factors on the efficiency of screening
was that reported by Porter in 1928 [12].
A W. S. Tyler Co. Hum-mer screen was
used and the variables investigated included
those of the screen, material and the cloth.
The main conclusions were that capacity
increased with frequency and decreased with
amplitude. Feed rate at first improved operation, then made it worse after a certain maximum was passed. The best angle of deck was
found to be 33 and increasing fines content
in the feed decreased product quality to a
greater extent than increasing the fraction of
coarse particles did. Unfortunately,
the screen
used by Porter [12] was of an electromagnetic type with highly uneven lateral amplitudes and other features long since superseded
by other designs. Hence, that work may be
regarded as being largely of an historical
interest rather than representative
of modern
vibratory screens.
A more relevant, and also comprehensive,
laboratory study was carried out by Lim [ 131
in 1958. The experimental
work was performed according to factorial design principles and the results subjected to the statistical
analysis of variance. Material, screen and
screen cloth variables were included and the
effectiveness of screening was defined as the
product of recovery and rejection of material
of the desired mesh size. The salient results
of that work, also reported in a paper by

Fowler and Lim [ 141, were that only two


main factors of screen aperture and feed rate
were important,
and significant first-order
interactions
involved combinations
of feed
rate, RPM and angle of deck.
The other laboratory study of continuous
screening, of which we have only recently
become aware, is that of Ferrara and Preti
[15]. These authors reported the effect of
feed rate and the effect of deck angle, both
at otherwise constant conditions, and analysed their results by a kinetic equation similar to that independently
developed by
Standish [ 21 several years later.
It should be noted here that one important
feature which distinguishes the kinetic approach to screening from other approaches
is that the results can be used to simultaneously express screening efficiency as product quality (e.g., dsa of the partition curve)
and the time required to achieve this value.
As noted earlier, the most widely used
method of characterising screening efficiency
in practice is the partition curve. However,
it was recently observed [16] that using the
normal method of its construction
in practice, partition curve does not give any obvious
information
about screening rates - so important for any screening efficiency definition
and analysis. After all, if for the same feed
rate, two identical screens give the same
partition curve but the one does it in half
the time, then this screen would be judged by any definition - to be more efficient.
The other feature of kinetic approach to
screening and sieving is that the starting differential equation is based on probability
concepts [2, 15,171, so the kinetic constants can, in principle, be obtained from
probability
theory. In fact, the kinetic constants (Ft*) of Standish and Meta [1] follow closely [ 171 the probability relationship
for normal incidence given by Gaudin [ 181,
as do also those of Ferrara and Preti [15]
under certain test conditions. Under other
test conditions,
the relationship
differed
markedly and especially so at small deck
angles. These authors made no attempt to
correlate the effects of the two screening
parameters varied (feed rate and deck angle)
on the kinetic constants of individual size
fractions. Clearly, since the kinetic constants
of all particle sizes are involved in the efficiency of screening, the knowledge of any

163

variation, and particularly the magnitude of


the variations, would be useful. Our work on
batch sieving [19] suggests that the kinetic
constants of all particle sizes being sieved,
and not just the near-mesh particles, may
be affected by sieving variables, and by inference this would be extendable to continuous screening also.
EXPERIMENTAL

Both the vibrating screen used and the


experimental procedure employed were exactly the same as previously described [ 11.
Materials used (Table 1) consisted of
quaternary ferrous sinter (p = 4.11 g/cm3)
and quaternary metallurgical coke (p = 1.12
g/cm), all in equal proportions. Three main
series of runs were conducted as follows:
-sinter (3.5 - 1.4 mm) on 3.5 mm woven
wire cloth
-sinter (1.94 - 0.25 mm) on 1.94 mm woven
wire cloth
-coke (1.94 - 0.25 mm) on 1.94 mm woven
wire cloth.
Oversize material, when used, consisted
of -4 +3.5 mm and -2.5 +1.94 mm particles
respectively.
Operating variables investigated were :
-flow
rate (g/s): 110, 210 (1.94 mm coke,
84 runs)
140, 390, 720, 875, 940
(1.94 mm sinter, 168 runs)
720, 875 (3.5 mm sinter,
60 runs)

-deck angle (9: 11.5,12.5,14.15


-RPM: 1160,1400,1700
-oversize: 10% of feed.
A number of runs in each series (26 total)
were duplicated. These results show that
errors were within 10% of the original result.
Additionally, some exploratory runs were
performed with a number of different proportions of each size fraction in the feed to
gauge any effect of feed composition in
screening.

TREATMENT

OF DATA

Only the experimental results of successful runs were used. Unsuccessful runs were
those for which the continuous weight
records of the four undersize and one oversize materials streams [l] were erratic, or
the undersize material in the first (feed end)
container accumulated at a slower rate than
that in the next container downstream. The
former was due to uneven lateral feed distribution and the latter indicated accumulation of the material on the screen under the
feed bin - a situation that occurred for certain combinations of operating variables,
viz. small deck angle and/or high flow rate
and low RPM. Twelve test runs exhibited
irregular behaviour and the results were
rejected.
All successful results were treated in the
manner detailed previously [ 11. Briefly, for
each run, the sieve analysis data for the four

TABLE 1
Coke and sinter characteristics
Material

Size
(mm)

Average size
(mm)

No/g

Sphericitya
(-)

(-)

Relative size

Sinter

-3.5
-3.15
-2.5
-2.0
-1.94
-1.4
-1.0
-0.63

+3.15
+2.5
+2.0
+1.4
+1.4
+1.0
+0.63
+0.25

3.32
2.82
2.25
1.70
1.67
1.20
0.81
0.44

14
19
36
88
90
215
714
2900

0.646
0.642
0.634
0.630
0.628
0.618
0.615
0.612

0.94
0.82
0.64
0.48
0.85
0.62
0.42
0.22

Coke

-1.94 +1.4
-1.4 +1.0
-1.0 +0.63
-0.63 +0.25

1.67
1.20
0.81
0.44

345
790
2780
10730

0.751
0.740
0.687
0.681

0.85
0.62
0.42
0.22

aRatio of diameters of inscribed and superscribed circles of the projected particle image.

164

undersize streams were used to calculate the


kinetic constant (Fz*) of each of the size fractions used. The calculations involved the
application of the integrated equation given
previously (viz. eqn. (9)) ref. [ 21) after the
k,/k, ratios have been evaluated as detailed
elsewhere [ 1, 21.
Because the calculated Fzj*again exhibited
a decrease in the end section of the screen
and were reasonably constant elsewhere (see
Fig. 11, ref. [l]), the values of all lz,* used
in subsequent analysis were calculated for
the same condition of 50% passing weight
of the total material. To obtain the values
of the cumulative weight per cent of each
of the size fractions involved at this 50%
passing weight, undersize stream consist
plots were made for each run and the corresponding values read off, generally by interpolation. An example of an undersize stream
consist graph is shown in Fig. 1 with the
readings involved at the 50% level marked
by x.
Plotting and calculations of the data were
assisted by computer application. The calculated values of Fzj* for all runs were then
used in the analysis of the effects of process
variables on these values. The analysis was
performed using a package programme
SPSS-X multivariant regression analysis.
The above regression analysis was chosen

because it checks each term for its tolerance


and minimum tolerance prior to the entry
in the regression equation. The tolerance of
a variable is the proportion of variance remaining after the effects of the independent
variables, already in the equation, have been
partitioned out. The tolerance is given as one
minus the squared multiple correlation of
that independent variable with the independent variables already in the equation. The
calculation of multiple correlation is performed by an iterative procedure by the
computer. The minimum tolerance is the
minimum of the computed tolerances of the
variables in the equation when the particular
independent variable is entered in the equation. The value of tolerance used in the present investigation was standard default value
of 0.01.
Although many combinations of the variables, including the projected open area
(CC cos a), were tried in the regression equation, the improvement in the correlation
coefficient obtained was not considered
sufficient (from 0.73 to 0.85) to compensate
for the much increased complexity and the
significant loss in physical perception and
practical application of the results. Therefore, the simple but still valid correlation
involving only the main variables has been
used in the analysis.

1.0
.9

CUHULRTIVE

MEIGHT

PERCENT

Fig. 1. Example of an undersize stream consist graph indicating interpolation procedure.

165

exhibit a large range of values.


(2) Kinetic constants decrease with relative
size.
(3) Kinetic constants for sinter particles are
higher than those for coke particles.
(4) Kinetic constants for -3.5 mm sinter
are higher than those for -1.94 mm sinter.
The maximum range of values of the
kinetic constants in each case (Figs. 2 - 4)
is exhibited by particles of largest relative

RESULTS

Plots of kinetic constants versus relative


size ratio (particle size/aperture
size) are
shown in Figs. 2 and 3 for the -1.94 mm
sinter and coke respectively,
and in Fig. 4
for the -3.5 mm sinter.
By inspection, a number of general features
of the results are apparent.
(1) Kinetic constants of each particle size

17.1SINTER

15.2 -

(-1.94 +O.Xmm)
:

13.3E
0
311.4
2
I

9.5-

'7.6 -

F
Y

5.7-

i
2

*
I

I
n

i
I

3.6 -

!
I

1.9 %0

-10

I,
.30

.20

I I I1
.40
.50
RELATIVE

I I 14
.70
.60

I
.60

. 90

1.00

SIZE. (-)

Fig. 2. Plot of kinetic constants as a function of relative size for sinter particles. Screen aperture = 1.94 mm.

6*0
3-1
7.2t

-I
COKE

6.4 -

(-1.94 +0.25mm)

7
E

5.6 ,o
4.85
3 4.02

i 3.2i=
w C.-l
3 5

1.6 -

*
.6 I

* ?I30

.10

.20

.
.

.30

I
I
.

I
i

.40

RELATIVE

.50

.60

.
I

.70

.80_

.90

1.00

SIZE, (-)

Fig. 3. Plot of kinetic constants as a function of relative size for coke particles. Screen aperture = 1.94 mm.

SINTER

16.6

a14.7
E
0

312.6

5
; 10.5

z
;

6.4-

F
;

6.3-

(-3.5 +1.4mm)

*
*

4.2

2.1

-060

t
i

.10

.20

.30

.40
RELATIVE

.50

.60

.70

.80

.90

1.00

SIZE, (--)

Fig. 4. Plot of kinetic constants as a function of relative size for sinter particles. Screen aperture = 3.5 mm.

size, i.e., the near-mesh particles. This range


is also an order of magnitude greater than
those for the other size particles. The extent
of this spread is not immediately obvious in
the graphs because the actual values involved
are very low (0.02 - 2.1) and the vertical
scale is compressed to accommodate
all the
data.
The fact that the range of values of the
kinetic constants of the near-mesh particles
is the largest of the group means that the
rate of screening of these particles is very
sensitive to changes in the operating parameters. This result is not unexpected
and it
follows all previous data that have ever been
published on sieving and screening of particles. It is also in line with Gaudins [18]
probability equation, which shows that with
increasing relative size from about 0.8 the
probability
of passage becomes extremely
low.
However, the behaviour of the smallestsize particles, which show a wider range of
kinetic constants
than the intermediate
sizes, is surprising and this result was not
expected,
since conventional
wisdom has
always been that fine particles pass through
the screen unaffected
like water. Nevertheless, the results in Figs. 2 - 4 clearly show
that the screening rate of the smaller-size
particles is affected by the operating conditions to a greater extent than that of the

intermediate-size
particles. One possible explanation for this effect may be the low kinetic energy of the small particles enhancing
dispersion in the bed rather than the directional transport to the screen surface. This
view is indirectly supported by the batch
sieving results of Hudson et al. [20] that
the transport of fine undersize and coarse
undersize to the screening surface is controlled by different transport mechanisms.
Another possible reason may be that the
somewhat lower sphericity
of the small
particles compared with that of the medium
particles has a hindering effect on their passage through the screen cloth. It is well
known that particle shape affects screening
efficiency and Riley [21], who specifically
studied the effect of particle shape in sieving,
has, in fact, shown that sieving time increases
with departure from sphericity.
The result (Figs. 2 - 4) that kinetic constants decrease with relative size is expected
from theory and confirms the fact that small
particles screen rapidly and the near-mesh
particles screen slowly. However, it is only
for certain combinations
of the operating
variables that the relationship
between kinetic constants and relative size parallels the
probability
equation for normal incidence
derived by Gaudin [ 181. For other combinations of the variables, the relationships
are
different but all still feature the expected

167

decrease of kinetic constants with relative


size. This observation, therefore, supports
the conclusion of Bandemer and Espig [22]
that either the Weibull function or a modified two-parameter power function should
be more appropriate in representing the diverse kinetic constants-relative size relationships.
The result (Figs. 2 and 3), that for otherwise identical conditions the kinetic constants for coke are lower than those for sinter,
is an expected result, which can be accounted
for by the density difference between the two
materials. Moreover, the mean difference in
the values of the kinetic constants is close
to 4 and this compares well with the density
ratio of 4.11/1.12 = 3.7. This result, therefore, agrees with that of Lim [ 131, who found
that the effectiveness of separation varied
approximately linearly with the specific
gravity of the material.
The above results are not surprising as
linear correction for density accounts for
the difference in the number of particles
between materials of differing densities. And,
since screening is a process related to the
number of particles being processed per unit
time, then it would be more rational to express kinetic constants in terms of number of
particles/(cm* s). However, as such units for
kinetic constants are unlikely to be considered generally useful the more traditional
units of g/(cm* s) have therefore been retained.
The result (Figs. 2 and 4), that increasing
aperture size and sinter particle size to keep
the relative size constants increases the values
of the kinetic constants by about 50%, may
at first sight seem strange. However, this
result parallels common experience that the
efficiency of screening fine materials is invariably lower than the screening of coarse
particles. In the literature, this phenomenon
has been explained by an increased tendency
of fine screens to become blocked. In fact,
Rose [7] has presented copious data which
show that for otherwise constant conditions,
blinding increases with decreasing aperture
size, and, after the increased blinding is accounted for in his screening model, the result
is that for a given screen type there ensues
a constant coefficient of diffusion of undersize material through screen cloth.
Similarly, Lims explanation [ 131 of his

results that screening effectiveness increases


notably with aperture size, to be due to
differences in the amount of near-mesh particles that may have existed in nominally
the same feeds employed, cannot be the
reason either. This is because the feed to
both screens in the present study was proportional identically, i.e., the feed contained
the same weight fraction of the four size
fractions used. Nevertheless, this very identity
in feeds is almost certainly responsible for
the observed difference in that the two identical feeds by weight are two different feeds
by particle number, which simple calculation
would show can account for the difference
of the same order of magnitude to that actually observed. In other words, for the same
relative size it can be shown that large aperture screen is capable of passing through a
greater number of particles per unit open
area compared with that of the small aperture screen.
It is of interest to note that for the aperture size ratio of this study, i.e., 3.5/1.94 =
1.8, the observed difference of about 50%
compares very favourably with Lims [13]
30% and 50% for the aperture size ratios of
1.6 and 2.2, respectively.

REGRESSION

ANALYSIS

The coefficients of the regression equations for individual kinetic constants, i.e., for
fixed size ratio, are given in Table 2.
Effect of density and mesh size
The effect of each operating variable on
the kinetic constants of both sinter and coke
show similar trends. The values of the coefficient, as expected, are generally higher for
coke than for sinter. However, there are
exceptions and some corresponding values
are not always proportional to density difference, suggesting that material density
interacts with each operating parameter.
Previous work [ 111 indicates that the density
of the particles being screened affects performance but the results are confused. The
results in Table 2, however, suggest that material density as such interacts in an unsystematic way with the operating parameters,
but for each operating parameter the differences are still within the systematic effect

168

of that parameter. In other words, the effect


of each parameter on the kinetic constants
is common to both materials.
Similarly, if the differences in the values
of the kinetic constants between the two
sinter feeds used are accounted for by a
constant factor, the regression coefficients
for the two different mesh size cloths used
were for all practical purposes identical. This,
therefore, means that the effect of each of
the operating variables is identical for both
screens processing the same material.
Effect of flow rate
For all practical purposes, the effect of
flow rate on the kinetic constants is to give
a linear decrease with relative size. In other
words, the effect of flow rate is least for the
near-mesh material and greatest for the fine
particles. Moreover, as Table 2 shows, flow
rate of the feed is the most significant variable throughout. In other words, the kinetic
coefficients of all particle sizes of both sinter
and coke are most significantly affected by
the feed rate.
As noted earlier, Lim [13] also found
feed rate to be one of the two important
main parameters affecting screen effectiveness. However, the effect was opposite to
that observed in this study. Since the results
of other reported studies [3, 5, 111 agree
with our result, it must be concluded that
Lims [13] result represents a special condition. Since the specific flow rates of sand
(S.G = 2.5) used by Lim were considerably
lower than those used in this study, his result
would be more relevant to particle beds than
deep beds employed in the present case.
In deep beds, the number of contacts per
unit length of the screen may be expected
to increase as the bed depth, i.e., feed rate,
increases and this explains the effect of feed
rate in increasing kinetic constants.
The observed result, that at any given feed
rate for the deep-bed operation of this study
kinetic constants increase as the particle size
decreases, can be explained by the fact that
segregation of particles towards the screen
cloth increases with particles of decreasing
size.
Effect of RPM
The effect of increasing RPM has been to
decrease the kinetic constants of the two

169

larger-size particles and to increase those of


the two smaller-size particles. At constant
RPM, the effect is essentially linear with
relative size and almost identical for both
sinter and coke. In both cases, the effect of
RPM is highly significant for near-mesh particles and very significant for the next largest
size particles.
Most studies [3,4,5,13]
of the effect
of RPM have shown that screening efficiency decreases as the RPM is increased, but
Lim [ 131 has also shown that this effect
interacts with the angle of inclination
of
the deck.
The deleterious effect of increasing RPM
on the efficiency of screening is explained
by the fact that an increase in RPM lengthens
particle trajectory, so the number of particle
contacts per unit length of the screen decreases.
The result of this study, that kinetic constants of the two larger size particles only
are reduced with RPM and those of the
smaller size particles are actually increased,
is a result that has not been reported previously. A most probable explanation
for
this effect is that as the kinetic energy of the
smaller particles is lower than that of the
larger particles, their trajectory will be shorter
and therefore they will experience more contacts with the screen surface compared with
the large-size particles.
One practical result of the foregoing effect
is that a change in RPM would affect the
composition
of the undersize stream and
change the shape of the partition curve without necessarily changing the value of dSOor
Epm. Thus, at low RPM, the partition curve
is skewed at the top size end, and at high
RPM at the bottom size end. What this means
is that with usual definitions
of screening
efficiency it may be possible to obtain negative, positive, or no effect of RPM on efficiency.
Effect of deck angle
Within the range of deck angles investigated and the values of significance of this
parameter
in Table 2, the effect of deck
angle is negative for the near-mesh particles
but thereafter
it is positive and increases
with decreasing particle size. This decrease,
which is essentially linear, is relatively small
for coke and very large for the sinter parti-

cles. This result, therefore,


suggests that
particle density is an important
selective
parameter in its own right.
An increase in deck angle reduces the effective mesh area, so it is not surprising that the
passage of near-mesh particles through the
cloth should be most affected. However, the
reason why an increase in deck angle should
enhance the passage of medium and fine-size
particles is not obvious. After all, an increase
in deck angle decreases not only the effective mesh area but also the number of contacts per unit screen length because of lengthened trajectories. Nevertheless, it is of interest
to note that for the range of deck angles used
in this study Lim [ 131 has found an increase
in screen effectiveness and Ferrara and Preti
[15] an increase in the half-size kinetic constant, i.e., the kinetic constant for particles
of relative size = 0.5. However, after a maximum of about 15 was reached these investigators observed a more or less sharp decrease
in effectiveness.
One possible explanation for this effect is
that an increase in deck angle may be expected to speed the motion of particles in
proportion to their size. This would result in
a distribution
of residence times of the particles with the largest particles having shorter
time and the smallest particles longer time.
This effect, together with the vibration of
the bed, would aid stratification
of particles,
which would be expected to follow the normal gradation, i.e., fines in the bottom and
coarse particles at the top [23]. The fact that
segregation is known to be a slower process
for low-density material [23] would explain
why the effect with coke particles is much
less pronounced
than with the much denser
sinter particles.
Of course, it is not too difficult to imagine
how at some increased deck angles higher
downward velocities and reduced effective
mesh area would combine to reverse the
above effect.
Effect of oversize
Within the significance of this parameter
in Table 2, the effect of adding 10% of oversize to the feed on the kinetic constants is
to increase that of the near-mesh particles
and decrease those of the other and smaller
particles. In other words, oversize in the feed
assists the passage through the screen of the

170

near-mesh particles and hinders that of the


smaller particles.
The beneficial effect of oversize in reducing
sieving and screening times has been reported
previously [ 1,2] but this is the first time
that the effect of oversize on the kinetic
constants has been considered. From previous work on the kinetics of sieving [ 21, the
effect of oversize on the near-mesh particles was not altogether unexpected and the
greater effect on the near-mesh particles of
sinter compared with that of coke is also in
line with that work. However, the reduction
of the screening rates of the smaller particles in the presence of oversize was unexpected. One possible explanation is an entanglement of the small particles by the large
oversize particles, reducing their throughput
into the undersize stream. This view is supported by the observation by Pritchard [24]
on the basis of Allis-Chalmers screening
experience that large particles tend to carry
fine particles with them or buffet the smaller
particles down to discharge over the end of
the screen. Additionally, Pritchard [ 241
also referred to instances where quite large
particles in the feed can assist in dislodging
near-size particles held in the deck apertures. This, therefore, confirms that the
beneficial effect of oversize observed in this
study has also been observed in industrial
practice.
Effect of feed composition
The results of the exploratory runs in
which feed composition only was varied have
shown that for all practical purposes kinetic
constants were not affected by this variable.
This, therefore, parallels the insensitivity of
kinetic constants to feed composition observed in our batch sieving studies [2] and
is also in agreement with the vibratory screen
results of Ferrara and Preti [ 151. These investigators correctly point out that the observed constancy of the kinetic constants
along the screen means that these values
must be independent of the size distribution
of the material that flows on the screen because this distribution varies considerably
over the length of the screen.
Moreover, as one of the conclusions of the
kinetic theory of screening [2] is that the
kinetic constants should be independent of
feed composition, the above results, there-

fore, constitute important experimental evidence in support of that theory.


Effect of relative size
For each of the three series of runs, regression of the data yielding a generalised equation for the kinetic constants has shown that
the effect of relative size is most significant
and negative, and matched only in its significance by the positive effect of the flow rate.
The highly significant effect of the relative
size is not surprising and this result simply
confirms the well-known and widely documented fact that the closer the particle size
is to that of the aperture size the more difficult is its passage through the screen. It may
be of interest to note that the high significance of the relative size was obtained even
by the regression of the combined sinter data
for which the smallest difference in the relative size between two adjacent values was
only 0.02 (0.64 - 0.62).

CONCLUDING

REMARKS

Before presenting results on any scientific


and technical subject, one must ask whether
the results contribute new and useful knowledge or merely add to the confusion that may
exist in the literature on that subject. This
question is particularly relevant to screening,
which is generally agreed to be the most confusing subject in the technical literature.
In their review article, Jansen and Glastonbury [ 111 identified a long list of factors in
three main categories, viz. those arising from
the nature of (i) screen movement, (ii) screen
surface and (iii) particulate material, that
have been established as influencing screen
performance.
That review shows that the number of
factors that may be varied in screening in
general is formidable. This, on its own, is
complex enough without adding the complexity inherent in the traditional definitions
of screen performance or efficiency. For
example, as the traditional screening efficiency definitions incorporate feed composition as a variable, the number of combinations of the effect of operating variables on
the efficiency of screening can, literally, become astronomic, However, by changing the
usual efficiency investigations to that of

171

investigating the effect of operating variables


on the kinetic constants, as has been done in
this study, the complication of feed compositions is avoided. This is useful.
Moreover, as the results of this study show,
the kinetic approach provides information
on the screening behaviour of individual size
particles. This is new knowledge.
Additionally, the results of this study have
applications to theory and to practice. Thus,
they help explain why theory (eqn. (9), ref.
[ 21) predicted lower kinetic constants for
near-mesh particles when oversize was present than when it was absent. The reason was
that the effect of oversize was at that time,
not unnaturally, assumed to be positive for
all particle sizes and not just the near-mesh
particles. However, when the negative influence of oversize on the kinetic constants
of the smaller size particles (Table 2) is taken
into account, theory then predicts higher
near-mesh kinetic constants as expected.
The application of the kinetic approach
to practical screening has already been referred to by Ferrara and Preti [15] and
Standish [16]. These authors have considered the use of kinetic constants for
(i) the comparison of different operating
conditions via the characteristic screen
number, K, [ 151;
(ii) the definition of the separation that has
taken place via the partition curve [ 15, 161;
(iii) the characterisation of the screening
process as a whole, which has not been
possible to obtain with the help of the other
parameters used up to the present time
1151;
(iv) the definition of screen acceptance tests
and standards, since in current practice different screening efficiency may be obtained
simply depending on whether we use a certain sample or whether we analyse that
sample [ 161;
(v) a more rational search for optima of
screening operations [ 15,161.
Of course, with the information of the
effects of operating variables on the kinetic
constants of individual particle sizes, greater
confidence can be placed in answers to the
foregoing questions.
This study has provided kinetic information on the effects of some of the operating
variables. Other variables that would need
to be considered to complete the picture of

the kinetics of screening should include


amplitude, aperture shape, mode of vibration and a more extensive range of particle
shapes.
It is hoped that the present study has
demonstrated some of the benefits of the
kinetic approach to screening and therefore
created enough interest in this subject for
it to be studied on a wider scale and at an
increased pace.
ACKNOWLEDGEMENTS

This work was financially supported by


the University of Wollongong Research
Grants Committee and assisted by the technical staff of the Department of Metallurgy
and Materials Engineering.
REFERENCES

10

11
12
13
14
15

16
17
18
19

N. Standish and I. A. Meta, Powder Technol., 41


(1985) 165.
N. Standish, Powder Technol., 41 (1985) 57.
H. N. Wilkinson, J. Iron Steel Inst., 209 (1971)
178.
P. L. Baldwin, Trans. Instn. Chem. Engrs., 41
(1963) 255.
D. Richter, Freiberg Forschungsh., A496 (1972)
1.
V. K. Karra, CZM Bulletin, No. 4 (1979) 167.
H. E. Rose, Trans. Inst. Min. Met., 86 (1977)
ClOl.
W. Axt and P. Sauer, Proc. Symp. Coal and Mineral Sizing, Wollongong, 1984, Aust. Inst. Min.
Met., Melbourne, Australia, 1984, Paper 9.
E. Artone and D. F. Tompsitt, Proc. Symp. Coal
and Mineral Sizing, Wollongong, 1984, Aust. Inst.
Min. Met., Melbourne, Australia, 1984, Paper 10.
A. J. Lynch, Proc. Symp. Coal and Mineral
Sizing, Wollongong, 1984, Aust. Inst. Min. Met.,
Melbourne, Australia, 1984.
M. L. Jansen and J. R. Glastonbury, Powder
Technol., 1 (1967/68) 334,
J. B. Porter, Natl. Res. Council (Canada) Rept.
No. 22, 1928.
S. C. Lim, Ph.D. Thesis, Univ. Sydney (1958).
R. T. Fowler and S. C. Lim, Chem. Engng. Sci.,
10 (1959) 163.
G. Ferrara and U. Preti, Proc. 11 th Znterntl.
Mineral Processing Congress, Cagliari, 1975, p.
183.
N. Standish, Proc. Symp. Coal and Mineral Sizing,
Wollongong, 1984, Aust. Inst. Min. Met., Melbourne, Austi.dia, 1984, Paper 4.
N. Standish, kroc. Australas. Inst. Min. Metall.,
No. 290 (1985) 59.
A. M. Gaudin, Principles of Mineral Dressing,
McGraw-Hill, New York, 1939.
N. Standish and J. Eshman, Proc. 8th Znt. Con-

172
gress of Chemical Engineering, Chemical Equipment Design and Automation,
Sept. 3 - 7, 1984,
Prague, Czechoslovakia,
20 R. B. Hudson, M. L. Jansen and P. B. Linkson,
Powder Technol., 2 (1968169) 229.
21 G. S. Riley, Powder Technol., 2 (1968/69) 315.
22 H. Bandemer and D. Espig, Proc. Symp. Coal and

Mineral Sizing, Wollongong, 1984, Aust. Inst.


Min. Met., Melbourne, Australia, 1984, Paper 3.
23 J. C. Williams, %%el Sot. J., Univ. of Sheffield, 14
(1963) 29.
24 A. N. Pritchard, Proc. Symp. Coal and Mineral
Sizing, Wollongong, 1984, Auat. Inst. Min. Met.,
Melbourne, Australia, 1984, Paper 8.

You might also like