You are on page 1of 10

Powder Technology 139 (2004) 165 – 174

www.elsevier.com/locate/powtec

Open and closed circuit dry grinding of cement mill rejects in a pilot scale
vertical stirred mill
C.C. Pilevneli *, S. Kızgut, İ. Toroğlu, D. Cßuhadaroğlu, E. Yiğit
Mining Engineering Department, Zonguldak Karaelmas University, Maden Muhendisligi Bolumu, 67100 Zonguldak, Turkey

Received 20 July 2003; received in revised form 15 November 2003; accepted 1 December 2003

Abstract

In this study, separator rejects of a closed circuit cement tube milling were dry ground in a 10-l pilot scale vertical stirred mill. Tests were
done as open and closed circuit. The results were evaluated by size distributions, Blaine surface area of the products and the specific energy
consumed during the tests. Mill capacity and the stirring speeds were the investigated parameters for the open circuit tests, whereas closed
circuit tests were carried out at a constant stirring speed but variable discharge rate. Five speed levels used in the open circuit grinding tests
are 200, 400, 600, 800 and 1000 rpm. The capacities were determined to obtain product Blaine surface areas in the limits between 2000 and
5000 cm2/g. Closed circuit tests have resulted in 87%, 182% and 452% circulating loads.
D 2004 Elsevier B.V. All rights reserved.

Keywords: Cement; Clinker; Dry grinding; Open circuit; Closed circuit; Vertical stirred mill

1. Introduction ture, medicine, photography, energy, etc., extensively uses


stirred milling technology in their production [1 –4].
The increase of the world’s population day by day and the The main reason for using stirred mills is their energetic
development of the human relations in many aspects follow- efficiency in the production of submicron particles com-
ing the globalisation of the earth has led to the increase in pared to other conventional grinding systems such as ball
motion which in turn has expanded the world’s energy milling. So nowadays regrinding circuits tend to shift to the
consumption. This increase in energy consumption of the stirred milling system.
consumed energy has caused the investigation of new energy The coarse size of the clinker is a disadvantage to stirred
sources and the invention of novel less power-consuming mills since their applicability is valid for a maximum feed size
technologies. of a couple of millimetres. However, stirred mills can be
The building sector is an indicator of the civilisation applied to cement grinding if the feed size is made suitable.
level. This also shows the importance of cement to the For example, instead of regrinding the separator reject in the
civilisation progress. Growing population and development tube mill, it can be directed to a stirred mill. Thus a possible
both increase the need for cement production. Half of the energy saving can be obtained compared to ball mill grinding.
energy used in cement production is consumed in clinker Because, stirred mills are reported to supply energy savings
grinding. Thus new technologies and researchers tend to below 100 Am product sizes [5] such as for finish cement
focus on this huge energy necessity to diminish it. ( 30 Am). So, this paper was aimed to clarify this reasoning.
Stirred milling is becoming popular in fine and super fine
grinding. Today, many industries including ceramics, min-
eral, coal, metallurgy, electronics, chemical, food, agricul- 2. Stirred mills

Fig. 1 shows the schematic representation of a horizontal


stirred mill. The mill body is a cylindrical tube with a stirring
* Corresponding author. Tel.: +90-533-7457170; fax: +90-372-
mechanism situated at the centre of the longitudinal axis.
2574023. Mainly a rotor activates the stirring parts such as the pins or
E-mail address: ccpilevneli@mynet.com (C.C. Pilevneli). discs mounted on it. The remaining empty volume is filled by

0032-5910/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.powtec.2003.12.002
166 C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174

Fig. 1. A horizontal stirred mill with pin stirrers.

the beads up to 90% to comminute the material which it Industrial applications of the stirred mills were recorded
surrounds. The media can be made up of different materials up to 20 t/h capacity with 6 Am median product size with
such as steel, glass, zirconium, ceramics, alumina, etc. 60% less energy consumption [1]. Recently, the installation
Among different media sizes from a couple of millimetres of the biggest stirred mill has been achieved in Australia
to hundreds of micrometers, an appropriate size is selected under the Mount Isa Mines and Netzsch-Feinmahltechnik
according to the feed size of the material. partnership. This stirred mill constitutes 3000 l net volume
The application sizes of stirred mills range from a few and 240 l/min capacity at 50% solids rate [8,9]. This large-
millimetres down to 1 Am. The product size is in the sub capacity installation encourages the mining industry to
micrometer range. Stirred mills are very effective in the replace the regrinding ball mill circuits with stirred mills
production of micronised material due to their easy pro- technology in the future. Increase in the stirring disk
cessing, simple construction, high size reduction ratio, low number and mill volume results in inputted power increase
energy consumption and less wear contamination. There- and reduction in wear loss over the beads [8] and thus
fore it has been started extensively to be used in the recent enhances the savings in energy.
decade in many industries especially such as mineral, coal,
ceramic, metallurgy, electronics, paint, chemical, food,
animal nourishment, biotechnology, rubber, agriculture, 3. Materials and method
medicine, photography and energy [1– 4,6].
The forces utilised in stirred mills are different from the The separator reject sample from a nearby cement
rod and ball mills. Stirred mills utilise mostly abrasive and factory (Oyak Bolu Cement Industry) was collected for
shear forces, whereas tumbling mills predominantly com- the experimental part of this study. The sampling point for
minute by impact and compression [6]. The stirrers of the the clinker grinding circuit was depicted in Fig. 2. The
stirred mills transfer their very high kinetic energy to the capacity of the cement mill during the sampling time was
surrounding beads in the radial direction, where their 130 t/h and the specific energy consumption was 32 kW
consolidated structure enhances abrasive and shear forces h/t. The amount of separator reject circulated back to the
rather than impact and compression. High speed of the mill was 480 t/h with a circulating load of 370%.
stirrers increases both the number or frequency and the The size analysis results were carried out with a
intensity of the comminuting events. Shear and abrasive Malvern Mastersizer-S Laser Analyser. The size distribu-
forces are the main reasons for grinding in the sub tion of the feed material is given in Fig. 3. As can be seen
micrometer range [7]. from Fig. 3, the size distribution of the separator reject is

Fig. 2. The sampling point of the cement milling circuit.


C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174 167

Fig. 3. Size analysis results of the separator reject samples used in the experiments.

below 600 Am with a median and 80% passing sizes of 75 description of the method for surface area measurement
and 150 Am, respectively. This size is more convenient to with a Blaine manometer is detailed in Turkish Standards
stirred mill grinding rather than ball mill grinding [5]. No. 24 [10]. Besides size measurements, energy measure-
The separator reject sample was brought to the ‘‘Yakup ments were also taken by a kW h meter.
KESKİN’’ Mineral Processing Laboratories of the ZEDEM The experimental setup is constituted by a magnetic
Industrial Support Centre of the Mining Engineering De- resonance feeder, a stirred mill, a helical transporter, a
partment of the Zonguldak Karaelmas University. Then the mechanical classifier, a vacuum fan, a dust collecting
milling studies were carried out with a pilot scale stirred filtering mechanism and the necessary drivers for adjusting
mill. The grinding tests were carried out both as open and the revolutions of the motors attached to the system units.
closed circuit basis. The study was financed by the Fig. 4 sketches the experimental setup.
University Research Fund and the experimental set was The magnetic feeder was produced by Eriez Magnetics
designed and prepared at local sites. The product fineness and has an hopper with a 100 kg of separator reject
was assessed by Blaine surface area manometer. The capacity. The hopper has a magnetically vibrating chute.

Fig. 4. Experimental setup of the pilot scale stirred mill.


168 C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174

Low carbon content steel beads were used for grinding


purposes. The media has a 2.5 mm diameter and 7 g/cm3
density. The mill body was filled up by 25 kg of media.
The pore volume between the media was used as a
measure of the amount of material (separator reject) to
be put at the beginning of the experiments. This was held
constant for both open and closed circuit grinding experi-
ments at 3 kg.
The mill body is enclosed inside a secondary chamber
and cooling water goes through the zone in between. The
heat radiated out of the main mill body is driven off by this
water thus preventing excess heat, damaging the system and
the material inside the mill. The mill used has 10 l net
volume and eight cylindrical pins for mixing purposes. The
pins are 20 and 170 mm in diameter and length, respective-
ly. The pins were coupled on a vertically rotating rotor
Fig. 5. The relationship between the flowrate and the Blaine surface area at
simultaneously at right angles. The mill rotor is open at the
the open circuit experiments.
bottom end where on which an attached raising plate
continuously stirs the media over a screen plate. Screen
The hopper discharge can be adjusted by a sliding door; plate covers the bottom end of the mill and holds the weight
thus, a maximum capacity can be arranged in the desired of the mill charge. The material fed moves downward by
way. The magnetic displacement intensity knob controls the gravity and stirring action. The residence time is controlled
capacity from zero to the maximum setting. The vibrating by the transporting velocity of the helical transporter
chute feeds the separator reject directly to the mill inlet mounted beneath the mill bottom. Speed of the transporta-
hopper. tion was controlled by a Telemechanic brand inverter.
The stirred mill is directly activated by a 7.5-kW motor. The transported mill discharge through the helical
The mill motor is attached to a frequency inverter (Siemens carrier can either be directed to the cyclone classifier or
MD1500) and controlled via a PC over a RS 232 – 485 directly taken as an open circuit product. Cyclone classifier
converter. The PC-motor combination enables the control of operates under vacuum conditions. Suction is maintained
the rotor’s revolution infinitely up to 1500 rpm and helps the by means of a fan with 1800 m3/h air blowing capacity.
currency and torque data to be gathered continuously during The dumped carrier discharge is sucked into the cyclone
the experiment. body where necessary air inlets form a vortex to separate

Fig. 6. The effect of stirring speed to the flowrate increase for a 1000 unit change in Blaine.
C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174 169

Table 1 separator reject in the mill. The relationship follows a linear


Linear relationship parameters m and c between flowrate ( q) and Blaine
pattern with reverse proportionality. As the stirring speed
surface area (S) to best fit the equation S = mq + c
increases, the slope of the relationship decreases, showing a
Stirring speed, rpm m c
less dependence on the effect of the flowrate change
200 3.2266 4565 compared to the slower speeds. For example, the reduction
400 2.9746 5465
required to obtain the same amount of surface area incre-
600 2.2252 6023
800 1.2188 6185 ment, let us say from 3000 up to 4000 cm2/g, flowrate at
1000 0.6233 5707 200 rpm should drop from 500 to 200 g/min. However, this
300 g/min decrement expands as it moves to the higher
speed lines and reaches 1700 g/min for the top-notch speed
the coarse particles from the fines. The fines are sucked 1000 rpm. At this speed the flowrate bounces from 4300
and taken into the filtering mechanism where fine product g/min down to 2700 g/min as Blaine surface area changes
(cement) is collected continuously. The coarse particles are from 3000 to 4000 cm2/g. As seen from Fig. 5, the capacity
ejected under the apex and recirculated to the stirred mill enormously amplifies at the higher speeds at the same
feeder chute. The amounts of the fine and coarse products Blaine fineness. This is undoubtedly because of the inten-
were weighed at regular intervals to determine the mass sified energy transferred to the charge inside the mill.
flowrates. The incremental change of the flowrates for different
speeds at each 1000 unit Blaine change was depicted in
Fig. 6. The abovementioned expansion in the flowrate
4. Open circuit grinding tests increments by the stirring speed effect can easily be seen.
If the change in capacity at 200 rpm corresponding to
Open circuit experiments were carried out at different 1000 cm2/g change in the Blaine surface area is taken as
flowrates. As mentioned above, the flowrate is adjusted by one unit; 400, 600, 800 and 1000 rpm stirring speeds lead
the revolution change of the helical carrier via a frequency the capacities to change by a factor 1.2, 1.5, 2.6 and 5.3,
inverter. The flowrates were determined and selected to let respectively.
the Blaine surface area measurements of the products lie The linear relationship parameters were drawn from the
within 2500– 5000 cm2/g limits. The stirring speeds used linear least squares approximation over the flowrate vs.
are 200, 400, 600, 800 and 1000 rpm. The analysis Blaine surface area values. The best fitting parameters
samples were collected after 9 kg of ground material found are listed in Table 1. The slope of the linear lines
was discharged from the stirred mill. This constitutes more is denoted by m and its negative sign shows an inverse
than threefold the hold up inside the mill during the proportionality. The constant c is the y axis intercept of the
experiments, i.e. more than three times the nominal resi- equations which is the maximum Blaine surface area
dence time for a single particle. Capacity measurements values achievable at the theoretically maximised residence
were continuously carried out until the test was completed. time inside the mill. This means that the product cannot
The relationship between the capacity and Blaine surface have surface area increment beyond this Blaine value at
area measurements were evaluated and the findings are the specified stirring speed. The c values are as low as
sketched in Fig. 5. 4500 cm2/g for 200 rpm, but rises to as much as 6200
As can be deduced from the reasoning, the Blaine surface cm2/g for the 800 rpm and then seems to decrease. The
area increases following a reduction in the flowrate or, in
other words, an increase in the residence time of the

Fig. 8. Change of flowrate by stirring speed to keep up certain Blaine


Fig. 7. Change of slope of the flowrate vs. Blaine lines by the stirring speed. values.
170 C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174

Fig. 9. The specific energy consumption against capacity for different stirring speeds.

last decrement at 1000 rpm is illusive due to scattered flowrate vs. Blaine line was achieved, which means that
points, especially at low Blaine values. If the uppermost required Blaine fineness is obtained at a higher flowrate
points were assumed in curve fitting, the c constant would than the expected. However, the condition is reversed at
be the same for the 1000 rpm stirring speed, too. It can be 1000 rpm leading to a decrease in slope increase rate. This
concluded from these results that the maximum attainable may happen due to ineffective mixing conditions caused
Blaine surface area is 6200 cm2/g for the experimental set by fluidised beads to fly around the empty places of the
of stirred mill used in our experiments at the open circuit upper portion of the mill body. This condition was also
tests at high speeds. This limit is lowered for the slower proved by eye inspection through an opening on the top of
speeds down to 4500 cm2/g. This may be caused due to the mill. Thus the bead-material impacts cannot reach the
inefficient transfer of the energy to effectively break down correct number due to loosened matrix, which in turn led
the increased consolidation of the powder-bead matrix at to an increase in the residence time or decrease in capacity
continued grinding times. than expected.
If the change of slopes of the linear fitting equations by In the cement industry, the product is almost at certain
stirring speed is examined, Fig. 7 delineates the type of the Blaine surface area levels such as 3000, 3500 or 4000 cm2/g.
relationship. This figure is similar to Fig. 6 because the y So, it is worth mentioning in our experiments to demonstrate
axis is reciprocated. The slope of the flowrate –Blaine line how the flowrate changes with stirring speed for certain
shows the same trend by following a linear path at the
speed range between 400 and 800 rpm, whereas it switches
from the linearity for the speeds of 200 and 1000 rpm. The
solid and dashed lines are the respective change of the
slope for the 1000 rpm due to the points regarded for
fitting a linear line as mentioned above. The dashed line is
the possible slope after discarding the points with low
Blaine values at 1000 rpm. This case better emphasises the
deviation from the linearity for the uppermost speed as
was noticed for the lowest speed, i.e. 200 rpm.
The meaning of the linear trend between the speed vs.
slope relationship for the speed range 400– 800 rpm is that
the Blaine fineness change with capacity is directly pro-
portional to the change in stirring speed. The increase of
speed proportionally reflects itself to the rate of change of
Blaine with capacity. However, deflection from this line-
arity occurs at 200 and 1000 rpm. The deflection at 200
rpm shows an increase in capacity since a less steep Fig. 10. Change of the parameter a’s value with stirring speed.
C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174 171

Table 2 Table 3
Specific energies required at certain Blaine levels Average flowrates around the separator and circulating loads
Stirring speed, rpm Specific energy, kW h/t Experiment no. Flowrates, g/min Circulating load, %
2 2 2 2
3000 cm /g 3500 cm /g 4000 cm /g 4500 cm /g Feed Coarse Fine
200 26.9 38.0 61.6 370.9 1 842 387 455 87
400 35.3 46.5 67.1 101.7 2 1688 1080 608 182
600 40.5 47.5 58.8 79.2 3 2761 2198 563 452
800 35.2 43.7 52.2 68.1
1000 30.8 42.4 56.9 71.9
(having a slope 1) to the capacity, because the tonnage is
divided to the kW h consumed at the same duration. The
Blaine fineness. Fig. 8 sketches the effect of stirring speed inverse proportionated linear lines (m = 1) on a log – log
over the flowrate to obtain 3000 –4500 cm2/g Blaine range scale can be fitted to a power relation of type y = ax 1
at 500 cm2/g increments. The shape of the Blaine lines can where a is the shifting constant along the x axis. The a
be approximated by a polynomial line. The flowrates would parameters are given in Fig. 9, too. The change of the
be overestimated if the 1000 rpm polynomials were used. parameter a with the stirring speed is shown in Fig. 10. As
This was previously noticed as higher slope in the stirring written within the figure, change of the parameter a can be
speed –slope relationship (see Fig. 7) for the 1000 rpm. The represented by a second order polynomial.
four curves were dispersed apart at regular intervals for the In order to obtain certain Blaine values, the required
range 200– 600 rpm, beyond which their scattering increas- specific energy values are given in Table 2 and plotted in
ingly expands. For example, 200 g/min flowrate change is Fig. 11 below. This table shows that certain Blaine values
required to obtain a 500 cm2/g change in the 200 –600 rpm such as 3000, 3500, 4000 and 4500 cm2/g can be obtained at
range. But, it develops up to 400– 500 g/min at 800 rpm 30, 40, 55 and 70 kW h/t specific energy levels, respective-
and 500 – 1200 g/min for 1000 rpm (lower flowrate for 4500 ly. 600 rpm gives comparatively higher specific energies to
cm2/g and higher flowrate for 3000 cm2/g). If the change in reach the Blaine values of 3000 and 3500 cm2/g. At low
amounts of the flowrates by the stirring speeds at limiting stirring speeds specific energies excessively gets higher to
Blaine values (i.e. 3000 and 4500 cm2/g) were assessed: achieve Blaine levels above 4500 cm2/g. This shows that,
The capacities change from the 500 to 4500 g/min and 32 to speeds greater than 600 rpm are required to obtain further
1930 g/min for the 3000 and 4500 cm2/g Blaine lines, grinding above Blaine values of 4500 cm2/g.
respectively.

4.1. Specific energy consumption at open circuit grinding 5. Closed circuit grinding tests
experiments
Closed circuit experiments were carried out at a constant
The open circuit experiments were energetically assessed mill stirring speed of 600 rpm. The product from the mill
by the electricity input divided by the tonnage for 1 h. The was conveyed to an air separator via a spiral conveyor. As
electricity consumption was measured via a kW h-meter. an adjustable parameter the spiral’s conveying capacity was
Different speeds gave different power input; so, the capac- used at three levels, namely 800, 1700 and 2800 g/min. The
ity-specific energy relationship plotted on a log – log axis oversize from the separator was carried back to the stirred
has separate lines for each speed shifting along the capacity
(x) axis as shown in Fig. 9. In this figure, the relationship
lines show that the specific energy is inversely proportional

Fig. 12. Change of amounts of separator products and the circulating load
Fig. 11. Specific energy vs. stirring speed for certain Blaine levels. rates with the separator’s feeding rate.
172 C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174

Table 4 Table 5
Blaine surface area values of the finish products at different circulating loads The specific energies of the finish products for closed circuit stirred milling
Experiment no. Circulating load, % Blaine, cm2/g Experiment no. Capacity, g/min Specific energy, kW h/t
1 87 4850 1 485 113
2 182 4060 2 537 102
3 452 4300 3 475 115
Average 500 110

mill, whereas the undersize was collected in a dust box as an


end product or finish cement. Blaine surface areas is supposed to be caused by the
The classifier products were determined by direct manipulation of the fine product collection system during
weighing at 2-min intervals. Since the oversize is to be the experimental procedure. However, the results enable us
returned to the mill’s feeding chute, only the accumulated to have insight about the closed circuit stirred mill grinding
amount of product for half a minute at the beginning of of the tube mill rejects.
each 2-min intervals was used to calculate the coarse We can conclude that the Blaine will range in a couple of
product’s capacity. In order to balance the ingoing and hundreds vicinity of 4500 cm2/g. The 90% passing size for
outgoing streams of the stirred mill, the flowrate of the the fine product is 30 Am, which is convenient to the cement
closed circuit’s fresh feed was equalised to the finish industry. The fine product fineness can be easily adjusted by
product discharge and continuously controlled during the changing the separator settings so reducing or increasing the
experiments. Blaine values to improve the concrete’s physical properties
The duration of the experiments were held constant at 20 to the standard values. But in the scope of this study the
min. The analyses of the samples were carried out by the physical properties of the concrete obtained was not
laser sizer and surface area by the Blaine manometer. The assessed.
average values of the flowrates around the separator and the Specific energy consumptions were calculated based on
circulating load ratios were given in Table 3. Increase of the the finish cement discharge rate. Their values are given in
spiral conveyor’s capacity does not have much effect on the Table 5. Closed circuit grinding tests resulted in an average
fine product’s capacity but rather on the coarse product specific energy consumption of 110 kW h/t at 4500 cm2/g
amount and thus on the circulating load. This can be better average Blaine surface area.
visualised in Fig. 12, where coarse product flowrate and the
circulating load linearly changes with the separator’s feed
rate. A small deflection from the linearity for the circulating 6. Discussion
load has of no importance since this is the result of a small
change in the fine product amount which accounts for a high Open and closed circuit experiments have resulted in
ratio after division as denominator by the nominator coarse different specific energy consumption figures. If open and
product rate. closed circuit experiments are compared in terms of specific
The Blaine surface areas of the fine products (finish energies of the products of equivalent Blaine surface area,
cement) were measured at the end of 20th minute for the closed circuit experiments uses more energy than the open
closed circuit experiments. These results were tabulated in ones. Closed circuit experiments resulted in a Blaine area of
Table 4. The Blaine values ranged between 4000 and 4900
cm2/g. The size distributions of the three experiments do not
show much difference (Fig. 13). The difference between the

Fig. 14. Size analysis results of the finish cement, ball mill and stirred mill
Fig. 13. Size distribution of the fine products from the closed circuit tests. products.
C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174 173

for comparing the stirred milling and ball milling results to


the finish cement from a tube mill of a neighbouring cement
factory. All the products subjected to the comparison have
the same surface area of 3500 cm2/g. Here, we will demon-
strate these results.
Fig. 14 shows three size analysis results, namely finish
cement, stirred mill and laboratory ball mill products’ size
analysis based on amounts in the indicated sizes. The stirred
mill and the cement products are much closer to each other.
They have nearly equal peaks at 20– 25 Am and possess on
the similar size range which is mostly between 1 and 100
Am. However, ball mill product peaks at 70 Am and its
coarse range extends to 200 Am.
In order to see the difference of the stirred and ball mill
products from the real cement’s size frequency, Fig. 15 was
Fig. 15. Frequency difference between cement and ball and stirred mill
products.
prepared. This figure was drawn according to the difference
of the amounts between the cement and the mill product at
4000 cm2/g at a specific energy consumption of 100 kW h/t. the same sizes. The curves above the zero line denote that
However, the same surface area can be achieved at 55 kW the cement amount at these sizes is higher than the referred
h/t for open circuit stirred mill grinding. These results mill products and vice versa. This figure better visualises the
cannot be used without strength measurements of the product comparison. Table 6 shows both the areas below
concrete products since the standard requirement is based (for positive difference) or above (for negative difference)
on the product’s strength rather than Blaine surface area. the frequency difference curves between each of the two
However, 100 kW h/t specific energy consumption is rather adjacent intersecting points on the x axis and the peak of the
high compared to the real cement manufacturing process difference distributions.
(i.e. 30 –50 kW h/t). The minus signs show that mill products are in larger
In order to utilise the stirred milling in the cement amount than the cement, whereas the plus sign shows that
grinding technology, these mills should produce similar the mill products are deficient in amount compared to the
product fineness in a cheaper way. This can be expected cement. Cement and stirred milling products mostly differ at
since literature dedicates 50% energy saving compared to 0.25 – 6.1, 6.1 – 21 and 38 – 879 Am size ranges. 0.25 – 6.1
that of tube milling. This will lead to an estimation of a Am size range has 7.5% excess for the cement, i.e. the
possible consumption of 15 kW h/t. But, pilot scale tests of stirred mill weighs less at this range. However, at 6.1 – 21
stirred milling of separator rejects in this paper have resulted and 38 – 879 Am ranges stirred mill product has 2.7% and
in a minimum specific energy consumption of 38 kW h/t at 5.6% extra mass, respectively. If ball mill product is
a Blaine surface area of 3500 cm2/g. compared to the cement, the main difference occurs at
Besides, the effect of rerouting the separator rejects to the 0.25 – 38 and 38– 879 Am size ranges. The former is the
stirred milling over the tube mill performance should be amount for a ball mill product to reach the cement’s
evaluated, too. Because, separator rejects is quite a fine fineness. In other words, 30% of the 38 –879 Am size range
material as a mill feed compared to fresh clinker, which has of the ball mill product has to be comminuted down to 38
to travel a long size reduction progress until the fine sized Am size range to obtain a cement’s size distribution.
product is obtained. The positive and negative only summation of the
During the evaluation of the results [11], investigation of amounts are equal and was given at the bottom row of the
the mass distribution into size groups has proved beneficial table and expresses the total difference at all sizes between

Table 6
The total amounts hold within two adjacent sizes at which Cement-StirredMill and Cement-BallMill frequency difference curves intersect the size axis
Cement-stirred mill Cement-ball mill
Areas of intersecting points Peaking points Areas of intersecting points Peaking points
Intersecting Amount in Size, Amount, Intersecting Amount in Size, Am Amount, %
sizes, Am between, % Am % sizes, Am between, %
0.05 – 0.25 0.1 0.2 0.03 0.05 – 0.25 0.5 0.17 0.09
0.25 – 6.14 7.5 3.1 0.66 0.25 – 38 30.6 16.6 3.18
6.14 – 21 2.7 12.2 0.57 38 – 879 30.1 76.3 4.67
21 – 38 0.9 26.2 0.42 Sum of equal signs 30.6
38 – 879 5.6 76.3 0.6
Sum of equal signs 8.4
174 C.C. Pilevneli et al. / Powder Technology 139 (2004) 165–174

the cement and the mill products. The difference of the the production of fines at the sizes suitable for finish
stirred mill and ball mills are 8.4% and 30.6%, respectively. cement compared to ball milling. However, the stirred mill
This figure outlines that stirred mill produces a product with consumed twice the specific energy that the ball mill uses.
more fines than that of a ball mill product has. This would This is expected to happen because of motor power ratings
mean value if they also had the same specific energies. But, and too low capacities compared to commercial mills.
laboratory ball mill (20 cm in diameter) consumes 20 kW Usage of additives may enhance the grinding process,
h/t, whereas stirred mill expenses 40 kW h/t. Their differ- thereby reducing the specific energy consumption.
ence in energy consumption seems to offset the gain
obtained from the fine size range improvement by stirred
milling. It is worth mentioning here another experimental Acknowledgements
result obtained by Bond’s mill and batch stirred mill (1.75 l)
that they gave 28 kW h/t and 60 kW h/t specific energy Authors would like to appreciate very much the
consumption at the same Blaine fineness, respectively. This contributions made by the following authorities during this
shows that their motor rating and capacity has an important research: Zonguldak Karaelmas University’s Research Fund
effect on power draw. Further experiments are necessary (Project No. 97-115-001-07) and Mining Engineering
to clarify the economy of stirred milling over ball milling. Department (Turkey), A.J. Lynch (Australia), Julius
Strength measurements should support the results and form Kruttschnitt Mineral Research Centre (Australia), Universiti
basis on evaluation. Sains Malaysia School of Materials and Mineral Resources
Engineering (Malaysia), Mining Engineering Department of
Hacettepe University (Turkey), Oyak Bolu Cement Industry
7. Conclusions (Turkey).
Pilot scale open and closed circuit stirred milling tests led
to further conclusions on the grinding of separator rejects
from a cement tube mill. References

[1] M.J. Mankosa, G.T. Adel, R.H. Yoon, Effect of media size in stirred
 The maximum fineness attainable by open circuit stirred ball mill grinding of coal, Powder Technology 49 (1986) 75 – 82.
mill tests is 6200 cm2/g. [2] S. Miranda, S. Yaeger, Homing on the best size reduction method,
 During the open circuit tests, specific energies exces- Chemical Engineering 105 (1998) 102 – 110.
sively increases beyond Blaine values of 4500 cm2/g [3] H. Fadhel, C. Frances, Wet batch grinding of alumina in a stirred bead
mill, Powder Technology 119 (1991) 257 – 268.
which corresponds to a specific energy consumption of
[4] A. Kwade, J. Schwedes, Breaking characteristics of different materi-
60 kW h/t. Beyond this point, further grinding can be als and their effect on stress intensity and stress number in stirred
obtained for the speeds above 600 rpm. media mills, Powder Technology 122 (2002) 109 – 121.
 Closed circuit tests carried out at the same separator [5] Anonymous, Energy-saving ultra-fine grinding with the SALA Agi-
setting but different circulating ratios led to a constant tated Mill, Zement-Kalk-Gips 46 (9) (2002) 600 – 601.
[6] Y. Wang, E. Forssberg, Product size distribution in stirred media
product discharge rate, but different Blaine surface areas
mills, Minerals Engineering 3 (4) (2000) 459 – 465.
above 4000 cm2/g. These products have a minimum [7] B.A. Wills, Mineral Processing Technology, 3rd ed., Pergamon, Great
specific energy consumption of 110 kW h/t. Britain, 1985. 630 pp.
 Closed circuit experiments doubled the specific energy [8] M.F. Young, F. Gao, Performance of the isamills in the george
consumption. However, this is expected to be caused by fisher flowsheet, Proceedings of AusIMM Seventh Mill Operators’
Conference, Kalgoorlie, Western Australia, 12 – 14 October, 2000,
separator settings. Besides, the strength comparison of
pp. 75 – 81.
the concretes prepared from the cement products is to be [9] K.R. Weller, M. Gao, P. Bowen, Scaling-up Horizontal Stirred Mills
evaluated in addition to their Blaine surface areas and from a 4-Litre Test Mill to a 4000-Litre ‘‘ISAMILL’’, Mount Isa
specific energies. Mines Australia Internet Site, 8 pp.
 Comparison of size distribution of products form a [10] TSE, TS 24: Physical and Mechanical Test Methods of the Cements,
UDK 666.94.06, Turkish Standards Association, Ankara, 1985. 28 pp.
laboratory ball mill, open circuit pilot scale stirred mill
[11] C.C. Pilevneli, Investigation of the fine sized clinker grinding in a
and the finish cement marketed at the same Blaine surface stirred bead mill, PhD thesis, Graduate School of Natural and Applied
areas showed that stirred mill product have a pronounced Sciences, Zonguldak Karaelmas University, Turkey, 2003. 159 pp.
similarity to the finish cement. Stirred mills have doubled

You might also like