-
ABSTRACT
This investigation has concentrated on optimizing the
newlydeveloped Pansep linear screen and developing empirical models
topredict the screening performance. Most empirical models
developedin the past to predict the performance for industrial
screens aresuitable for the vibratory screens. Understandably,
these models aredeficient in predicting the performance of linear
screens due to thedifference in their operating principle.
Thus, the main objective of this study has been to
developempirical models suitable for linear screens to predict the
partitioncoefficient for a given size fraction as a function of the
aperture size.The inadequacy of the conventional model equation
originallydeveloped for vibratory screen has been demonstrated and
amodified equation developed to predict the partition coefficient
as afunction of the normalized mean particle size (d/d50c). In
addition,new model equations have been developed for linear screens
topredict the screening efficiency as a function of the separation
size(d50) and to predict the separation size as a function of the
screenaperture. It is believed that these model equations will be
useful forthe plant operators in selecting the mesh panels of
correct aperturesize and in predicting the product size
distribution based on theselected aperture size.
INTRODUCTION
Screening is achieved through two independent process
steps,i.e., stratification of the solid material into a bed having
the undersizeparticles closest to the screen surface followed by
the passage ofthese undersize particles through the screen
openings. Both stepsare equally important for the achievement of
high-efficiencyscreening. However, there is a distinct difference
in the mechanismsutilized to achieve these two steps in case of
vibratory screens andthe new generation linear motion screens. In
case of the former,mechanical vibration is transferred from the
screen surface to thesolid particles resulting in a continuous lift
and fall of solid particles onthe screen surface. The continuous
upward and downwardmovements of the particles result in the
formation of a stratified bedon the screen-surface with the finest
particles forming the bottommost layer that is closest to the
screen surface due to the well knownconsolidated trickling
phenomenon. Upon reaching the screen-surface, the fine particles
pass through the screen openings and
report to the underflow stream subject to their relative size.
Whereas,the passage of the particles through the screen openings is
notpossible for the oversize particles, the through-passage for
theundersize particles is considered as a probability process. The
greaterthe difference in size between the undersize particle and
the screenopening, greater is the probability of particle-passage
or in otherwords lesser is the probability of particle being
retained on the screensurface in a single trial. Multiple trials
allowed by a longer screensurface further decreases the probability
of particle retention on thescreen surface or in other words
increases the probability of particlepassage in vibratory
screens.
Vibratory screens are widely used for coarse coal and
mineralseparation; however they are rarely used for size separation
at orbelow 150 micron size due to several reasons discussed in
otherpublications (Buisman and Reyeneke, 2000). Cyclone
classification isthe most widely used size separation process in
the fine particle sizerange. The diameter of classifying cyclones
varies depending on themagnitude of the separation size. Small
diameter cyclones having theability of producing greater
centrifugal field are used to achieve finersize separations whereas
larger diameters are more suitable forcoarser size separations. In
many coal preparation plants, 38 cmdiameter and 15 cm diameter
cyclones are routinely used forachieving size separations at 150
micron and 45 micron, respectively.Although associated with high
throughput capacity, classifyingcyclones allow misplacement of a
significant amount of fine particlesto the underflow stream and
thus impair the overall efficiency of sizeseparation. Several
studies (Buisman and Reyeneke, 2000; Brown etal., 2000 and Mohanty,
2002) conducted in recent years havesuccessfully demonstrated more
efficient fine screening performanceusing a new generation linear
motion screen, known as the Pansepscreen.
The first significant application of linear screens reported is
DelkorTechniks screen in South African gold industry in the year
1985(Anon, 1986 and Wills, 1997)). Over the years, Delkor Technik
hasimproved the screen design to expand its range of
industrialapplications jointly with the Anglo American Corporation
of SouthAfrica. The Delkor screen, commonly used for
pre-screening,desanding, carbon scavenging and desliming of loaded
carbons isbeing developed to be used to reduce over-grinding in
millingoperations (Delkor Technik, 1999). Delkor has started using
theTrackmatic cloth as a carrier for fine weave screen cloth with
highopen area especially suitable for fine particle screening.
Linear
1 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
Preprint 03-135
PERFORMANCE PREDICTION OF THE PANSEP LINEAR SCREEN
M. K. MohantyA. Palit
Southern Illinois Univ. at CarbondaleCarbondale, IL
-
(motion) screens of several other types (Derrick, 2002 and
Osborn,2002) are used for mostly fine particle dewatering
applications. Thelinear motion of these dewatering screens is
provided by a vibrationmechanism unlike the linear screens used
commonly for fine sizing,which utilize a head and tail pulley
system to guide the mesh panelsthrough a linear path.
The Pansep screen is one such linear screen, which wasdeveloped
jointly by the Particle Separation Systems and the AngloAmerican
Corporation of South Africa to be used for high-efficiencyfine
particle screening. The details of the constructional features
ofthe Pansep screen have been reported in other publications
(Buismanand Reyneke, 2000; Brown et al., 2000). As shown in Figure
1, coalslurry is introduced through a feed distributor evenly on
the screensurface moving in a linear direction. A relatively thin
bed of solidmaterials formed on the screen surface is subjected to
water spraysfrom both top and bottom, which allows the required
stratification ofthe solid particles. As in the case of the
vibratory screen, greater thedifference between the particle size
and the screen opening greater isthe chance of the particles being
passed through the screenopenings. The probability of the particle
passage is further increasedby increasing the water spray pressure
up to certain extent. Inaddition, increased retention time of the
particles on the screensurface caused by a slower screen speed may
allow a longer periodfor the solid materials to be acted upon by
the water sprays and forthe undersize particles to pass through the
screen openings.Furthermore, longer retention time allows the
near-size particles to besuitably oriented on the rectangular mesh
screen to enable their easypassage through the screen-openings.
However, decreasing screenspeed may also allow more solid material
to be fed per unit time perunit screen length. This may increase
the bed thickness to a criticallevel beyond which stratification
and thus the overall screeningperformance may be impaired. Thus,
there is a trade-off between thelinear velocity of the screen mesh
panels and the mass feed rate tothe Pansep linear screen.
In light of the above discussions related to the
screeningmechanisms associated with the vibratory screens and the
linearscreens, it is understandable that the screening performance
obtainedfrom both screen types may not be the same. Therefore, a
majorityof the empirical model equations suitable for vibratory
screens maynot be useful for predicting the performance from linear
screens. Thisphenomenon has been investigated in this study using
theexperimental data obtained from a Pansep linear screen for
threedifferent coal samples as well as one iron ore sample. Due to
theinadequacy of the existing empirical model (Karra, 1979), a
modifiedmodel has been developed to predict the partition
coefficient as afunction of the normalized particle size (d/d50c).
New empirical modelequations have been developed to predict the
partition coefficient for
various size fractions and the corresponding screening
imperfectionand the effective separation size (d50) as functions of
screen aperturesize. The models have been validated using coal and
iron oresamples obtained from multiple sources.
EXPERIMENTAL
Samples
Fine coal slurry samples were collected from several
processstreams of two different preparation plants treating
Illinois No. 5 andMurphysboro seam coals. The process streams
included the raw coalclassifying cyclone feed stream, the spiral
product-sieve bend feedstream and the secondary classifying cyclone
feed stream. The bulkslurry samples were collected over a period of
several hours tocompensate for the temporary fluctuations in the
plant feedcharacteristics and thus be representative of the feed
slurry reportingto the existing size separation devices used in the
plant. The slurrysample collected from the feed streams of raw coal
cyclone and sievebend were used for evaluating the Pansep screen
for its sizeseparation performance at 150 micron and 250 micron,
respectively,whereas the secondary desliming cyclone feed slurry
sample wasutilized for evaluating the size separation performance
at 45 micronparticle size. The coal slurry sample obtained from the
preparationplant treating the Illinois seam coals were used for the
modeldevelopment tests whereas, the Murphysboro seam coals were
usedfor the model validation tests. Size-by-size weight and ash
distributionfor each sample used in this investigation is provided
in Table 1.
Experimental Layout and Procedure
Prior to beginning the experiments, several barrels of the
slurrysample are mixed in a feed sump having a capacity of nearly
4000liters shown in the experimental layout of Figure 2. The spray
wateris turned on and the Pansep mesh panels are set to motion at
adesired speed before introducing the feed coal slurry to the
Pansepscreen. A magnetic slurry flow meter is used to monitor the
volumetricslurry feed rate to the screen. Overflow and underflow
samples arecollected after allowing the screen to run for a few
minutes to ensurea steady state condition. Since there is an
external source of water forPansep screen in the form of water
sprays, the overflow andunderflow slurry are collected together in
a settling tank, where theslurry is left for 24 to 48 hours to
allow a complete settling of all solidparticles. Upon achieving a
complete settling, a calculated amount ofsupernatant water is
siphoned out of the settling tank to adjust thesolid content to the
desired level. The second pump in the circuit,which is used as a
spray water pump while running the experiments,
2 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
(a) (b)
Figure 1: (a) A schematic of the Pansep screen and (b) the
experimental layout utilized in this investigation.
-
is utilized to pump the coal slurry from the settling tank to
the feedtank to begin a new test series.
RESULTS AND DISCUSSION
The size separation performance of the Pansep screen
wasoptimized using factorially designed experimental programs.
Differentset of mesh panels having rectangular apertures of 180 x
400 micron,100 x 400 micron and 50 x 200 micron were used for
achievingdesired d50c sizes of 250 micron, 150 micron and 45
micronrespectively. Feed solids content appeared to be the most
significantoperating parameter influencing the screening
performance whileachieving the size separation at 45 microns using
mesh panels havingthe finest apertures. On the other hand, for
coarser size separation,other process parameters including feed
volumetric flow rate, screenlinear speed and spray angle were also
found to have significanteffect on the size separation performance.
The optimizedperformance from the Pansep screen was found to be
significantlybetter than that of the competing conventional
technologies includingsieve bend and classifying cyclone. The
excellent size separationperformances obtained using mesh panels of
all three differentaperture sizes are illustrated by the partition
curves shown in Figure2. A simple analysis of the partition curves
indicates very lowscreening imperfection values in the range of
0.20 and below at eachaperture size. The details of the factorial
experimental programconducted to optimize the Pansep screen have
been discussed inanother publication (Mohanty, 2002).
Development of Empirical Models
Numerous studies have been conducted in the past to
developtheoretical models to predict the partition functions and
thus theproduct size distribution. Ferrara and Preti developed a
screeningmodel (Ferrara and Preti, 1975) for vibrating screen and
subsequentlyvalidated the model using laboratory and pilot scale
screening data
(Ferrara et al., 1988). This screening model developed for
dryscreening has been subsequently reviewed by several
otherinvestigators (De Pretis et al., 1977; Schena, 1982; Kelly
andSpottishwood, 1982; Hess, 1983; Herbst and Obald, 1984) and
foundto be a very accurate model. Karra (1979) developed an
empiricalmodel for a double deck vibratory screen, which is also
useful forconducting simulations with wet screening operations over
a very
3 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
Table I: Size-by-size distribution of weight and ash percentage
of the coal samples used in this investigation
Figure 2: Partition curves corresponding to the best
screeningperformance achieved from the Pansep linear screen using
meshpanels of three different aperture sizes.
Par
titi
on
Co
effi
cien
t (%
)
-
wide particle size range. From one of the recent books (King,
2001),it appears that Rosin Rammlers empirical model is similar to
Karrasequation. In the current investigation, it was intended to
developsimilar model equations for predicting the performance from
a linearscreen, which are lately being commercialzed for fine
particlescreening. Prior to developing new empirical model
equations for thelinear screens based on the Pansep linear screen
data, it was desiredto investigate the suitability of Karras
equation for linear screenapplications. Mathematically, Karras
equation is described as follows:
Ci = [1-exp {-0.693(di/d50)5.846}] [1]
where Ci, the oversize partition coefficient in fraction for ith
size
fraction and di, mean particle size in microns. This equation
wascompared to the partition trend generated by nearly 400 data
pointswhich include nearly 330 data points generated from fine
coalscreening using three different aperture sizes as well as 70
iron orescreening data points supplied by the then equipment
vendor(Reyeneke, 2000). As revealed from the scatter diagram and
theresidual plot in Figures 3 (a) and (b), respectively, the Karra
modelequation does not satisfactorily represent the linear screen
datapoints. It must be realized that some amount of data scatter
isexpected due to the unavoidable effects of particle shape
andrectangular mesh. However, the deviation of the data points from
thecurve representing the Karra model is excessively high
beingrevealed by the large residual values ranging from 0.4 to
+0.5indicated in Figure 3 (b). It may also be observed that the
deviation issignificantly greater in the finer particle size range.
In spite of thedeviation, the shape of this curve and hence this
type of exponentialequation appear to provide a similar trend as
the data points.Therefore, a non-linear regression analysis was
conducted using astatistical software package to develop a similar
exponential equation(SIU Model), to fit an appropriate model to the
aforementioned 400data points. The resulting model equation can be
stated as follows:
PNi = [1-exp {-0.7838 (di/d50)2.773}] [2]
where PNi is the oversize partition coefficient corresponding to
the ith
size class. The corresponding scatter-diagram and the residual
plotsare shown in Figures 4 (a) and (b), respectively. The
adjustedstatistical coefficient of determination (R2) of this
regression fit is morethan 0.97, which means 97% of the variability
in the partition data isexplained by the exponential model equation
and only 3% is due tothe error in the model. The total sum of
square error was reducednearly 2.5 times in comparison to the Karra
model fit.
In addition, the majority of the residual data points were
inside the 0.2 partition coefficient band for the SIU model
indicating the greaterpreciseness of the model-prediction.
It must be realized that for fine particle screening, there is
asignificant difference between the magnitude of effective d50 size
andthe size of the aperture size unlike the coarse particle
screening.Therefore, both Karra and SIU Models may be useful in
predicting theoversize or undersize partition coefficients and thus
the overallproduct size distribution for only coarse particle
screening, in whichcase the effective d50 cut point of the size
separation may be fairlyclose to the aperture size. However in fine
screening, a differentmodel equation as a function of screen
aperture size instead of d50size may be more useful to predict the
partition coefficientscorresponding to individual size classes.
More than 300 partition datapoints generated from fine coal
screening tests using coal samplesoriginating from two different
coal seams and mesh panels of threedifferent aperture sizes have
been utilized to develop a best-fitequation for determining the
oversize partition coefficient. A nonlinear,step-wise regression
analysis has been conducted to generate thebest-fit model equation,
which may be mathematically described asfollows:
[3]
where PNi, oversize partition coefficient corresponding to the
ith size
fraction; A, mean aperture size in micron; di, mean particle
size for the
4 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
Figure 3: Comparison of the actual partition data with that
predicted using Karra (1979) Model.
(a) (b)
-
ith size fraction. The adjusted R2 for this model-fit is nearly
0.98,which testifies the goodness of fit of this exponential
equation. Clearly,this model equation provides a convenient
approach to predict thesize distribution of the screen oversize and
undersize products basedon the aperture size selected for the mesh
panels.
The goodness of this regression model fit is also
exhibitedgraphically by the scatter diagrams and residual plots
prepared foreach aperture size studied in this test program. As
shown in Figure 5(a), for the rectangular aperture of size 180 m x
400 m having ageometric mean size of 268 m, the model equation fits
satisfactorily
to the experimental partition coefficient data corresponding to
theindividual size fractions. The residual analysis data plotted in
Figure 5(b) indicates a maximum difference of less than 0.2 between
theexperimental and the model-predicted partition data. Nearly 90%
ofthe 30 experimental data points are within the 0.1
partitioncoefficient band of the model-predicted data. It must be
realized thatmany data points are not visible in the plot for being
overlapped withone another. For example, although it appears like
one data point,there are actually 6 data points corresponding to
the mean particlesize of 9.38 m in Figure 5 (a). Figures 6 (a) and
(b) illustrate the
5 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
Figure 4: A non-linear regression equation representing the
best-fit model (SIU Model) for predicting the partition data and
theassociated residual plot.
(a) (b)
Figure 5: Comparison of the actual experimental partition data
and the partition curve predicted using the new model
equationdeveloped for the mean aperture size of 268 m.
Par
titi
on
Co
effi
cien
t (%
)
(a) (b)
-
goodness of fit of the new model for the aperture size of 100 m
x 400m having a mean aperture size of 200 m utilizing 125
experimentaldata points. The entire data set is within +0.08 and
0.06 partitioncoefficient band of the model-predicted data. Similar
comparison ofexperimental data and the predicted partition data is
illustrated inFigure 7 (a) for an aperture size of 50 x 200 m
having a mean sizeof 100 m. The experimental data consisting of 150
partition datapoints are within + 0.06 to 0.07 partition
coefficient band of themodel-predicted data as indicated in Figure
7 (b).
Model Validation
The empirical models stated in Equations 2 and 3 were
developedduring this investigation to predict the oversize and
undersize partitioncoefficients as a function of the d50 and the
aperture size,respectively. It was desired to further validate
these empirical modelsusing test data generated from new series of
experiments utilizingcoal slurry samples originating from another
coal seam, i.e.,Murphysboro seam. The SIU-model stated in Equation
2 is validatedat two different d50 sizes as shown in Figure 8 (a).
In total, 186
6 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
(a) (b)
Figure 6: Comparison of the actual experimental partition data
and the partition curve predicted using the new model
equationdeveloped for the mean aperture size of 200 m.
(a) (b)
Figure 7: Comparison of the actual experimental partition data
and the partition curve predicted using the new model
equationdeveloped for the mean aperture size of 100 m.
-
partition data points were generated from the Pansep
screeningconducted using two different coal slurry samples and two
sets ofmesh panels. Partition coefficients corresponding to
individual sizefractions were also predicted using the SIU-model of
Equation 2 to becompared with the experimental partition data as
shown in Figure 8(a). The proximity of majority of the data points
to the diagonal verifiesthe validity of the SIU-model equation.
Equation 3 states an empirical model, which is believed to
bemore useful for the plant operators from a practical point of
view. In areal-life situation, the plant operators may like to have
a reliable modelso that they may predict the size distribution of
the overflow andunderflow of a screen based on the aperture size
they use. Thus, itwas desired to check the validity and reliability
of the model equationsutilizing coals slurry samples originating
from different seamsscreening at multiple aperture sizes. In
addition, a set of iron ore dataobtained from the equipment vendor
was also utilized to check theutility of this model equation for
other minerals. In total, 222 partitiondata points were generated
from the Pansep screening conductedusing fine coal slurry samples
originating from two different coalseams and mesh panels of two
different aperture sizes, i.e., 100 mx 400 m and 50 m x 200 m.
Another set of 63 partition data pointswas obtained from iron ore
screening using mesh panels of aperturesize 200 m x 80 m. Oversize
partition coefficients corresponding toindividual size fractions
were also predicted using the empirical modelof Equation 3 to be
compared with all 285 experimental partition datapoints as shown in
Figure 8 (b). Although there is some scatter in thedata, the
diagram reveals a satisfactory comparison. It is believed thatthe
nature of different coal/mineral samples, their size distribution
andcharacteristic shapes may account for the scatter in the
data.However, it is believed that the model predictions will be
reliable withinan acceptable level of error to be used by the plant
operators.
A very useful application of this model in a real-life situation
maybe in the selection of the aperture of the correct size for
producing adesired oversize or undersize product from a feed
material of a givensize distribution. Equation 3 may simply be
written as
where, PNi, over size partition coefficient corresponding to ith
size
fraction; A, geometric mean aperture size; di, geometric mean
particlesize of the ith size fraction; fitting constants: a,
-0.0546; b, 4.2604.Therefore,
By plugging in the values a and b in the above equation, the
followingrelationship is obtained:
7 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
(a) (b)
Figure 8: Comparison of the model prediction and the partition
number calculated using the validation test data.
Pre
dict
ed P
artit
ion
Coe
ffici
ent
Act
ual P
artit
ion
Coe
ffici
ent
-
The above equation will allow the production of the type of
plotsshown in Figure 9, which will be extremely useful for the
plantoperators in the selection of suitable aperture size based on
thedesired size distribution of the product. Further manipulation
ofEquation 4 will produce new expressions for d50 cut-size and
theimperfection (I) of screening as functions of aperture size:
Understandably, Equations 5 and 6 can be used for the
predictionof screening efficiency over the fine aperture size range
investigatedduring this investigation. Some error in
model-prediction is expectedfor other minerals due to their
characteristic particle shapes and alsothe rectangular apertures
used in fine screening applications.
CONCLUSIONS
The conclusions obtained from this Pansep linear screen studymay
be summarized as follows:
The existing fine screening model (by Karra and RosinRammler),
which were originally developed for vibratory screens, hasbeen
found to be deficient in predicting the performance from a
linearscreen. A more suitable empirical model, named as SIU Model,
hasbeen developed using nearly 400 experimental partition data
andvalidated to predict oversize partition coefficient as a
function of thenormalized mean particle size (d/d50).
For fine particle screening, a new empirical model has
beenproposed to predict oversize and undersize partition
coefficients andthus the product size distribution for a given feed
size distribution asa function of the screen aperture size. A total
of 285 validation testdata obtained from fine coal and iron ore
screening comparefavorably with the model-prediction. Some of the
errors in the modelprediction may be attributed to the effect of
particle shape notconsidered while formulating the model
equation.
The new model equations developed for predicting the
screening imperfection and separation size (d50) may
significantlyhelp the plant operators in selecting the correct
aperture size toproduce a desired oversize or undersize particle
size distribution.
ACKNOWLEDGMENTS
The author sincerely acknowledges the funds provided by
theOffice of Coal Development of the Illinois Department of
Commerceand Community Affairs for this investigation under
therecommendation of the Illinois Clean Coal Institute. In
addition, theauthor greatly appreciates the technical guidance and
supportprovided during the course of this investigation by Mr. Rein
Buisman,Mr. Kobus Reyneke and Mr. Shelby Akers of the then Pansep
group.
REFERENCES
Anon., New linear screen offers wide applications,
Engineeringand Mining Journal, September, 187: 79 (1986).
Brown, J. V., Buisman, R., Imhof R. M., Cost savings Potentials
ofPansep Screening Technology, Proceedings, Major Trends
inDevelopment of Sulfide Ores Up-Grading in the 21st century,
Norilisk,April 24-28 (2000).
Buisman Rein and Kobus Reyneke, Fine Coal Screening Usingthe New
Pansep Screen, Proceedings, 17th International CoalPreparation
Conference, Lexington, Kentucky: 71-85 (2000).
Delkor Technik, www.delkor.co.za (1999).De Pretis, A., Ferrara,
G., Gaurascio, M. and Preti, U., A new
approach to screen design, Proceedings, 12th IMPC, Sao
Paulo,Brazil (1977).
Derrick, www.derrickcorp.com (2002).Ferrara, G. and Preti U., A
contribution to screening kinetics.
Proceedings, 11th IMPC, Calgiari, Italy (1975).Ferrara, G.,
Preti U. and Schena G. D., Modeling of Screening
Operations, , International Journal of Mineral Processing, 22:
193-222(1988).
Herbst, J. A. and Obald, A. E., A population balance model
forscreening. Proceedings, 9th Powder in Bulk Solids
Conference,Chicago, Illinois (1984)
Hess, F., Mathematical modeling of screen and related units
for
8 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
Cum
ulat
ive
Wei
ght
% F
iner
Figure 9: Model Simulation results to help select the
appropriate aperture size for fine particle screening.
-
plant simulation, Ph. D. thesis, University of Queensland,
Australia(1983)
Karra, V. K., Development of a model for predicting the
screeningperformance of a vibrating screen, CIM Bulletin, April:
167-171(1979).
Kelly E. G. and Spottiswood D. J., Screening and
Sieving,Introduction to Mineral Processing, John Wiley & Sons,
Inc., Canada:169-198 (1982).
King R.P., Modeling and Simulation of Mineral ProcessingSystems,
Butterworth Heinemann, Chapter 4: 89 (2001).
Mohanty, M. K., Fine coal screening performance enhancementusing
the Pansep screen, International Journal of Mineral Processing,in
press, (2002)
Osborn, www.osborn.co.za (2002).Reyneke, K., Waltech group,
Personal communication (2000).Schena, G. D., The processing of
indutrial screening data, A
modeling approach. Internal Report, 1st Miniere e Geofisica
Appl.,Universita di Trieste. (1982).
Wills, B. A., Industrial Screening, Mineral Processing
Technology,Butterworth Heinemann, Chapter 8: 177-191 (1997).
9 Copyright 2003 by SME
2003 SME Annual MeetingFeb. 24-26, Cincinnati, Ohio
HomeContents (Author/Title List)SearchSearch Results