Вы находитесь на странице: 1из 9

Int. J. Miner. Process.

74S (2004) S55 S63


www.elsevier.com/locate/ijminpro

Optimization of continuous ball mills used for finish-grinding


of cement by varying the L/D ratio, ball charge filling ratio,
ball size and residence time
R. Schnatz
Polysius AG, Beckum, Germany

Abstract
During the last decade, semi-finish-grinding plants have been used more and more for the energy efficient grinding of highquality cement. In 1999, it was found that by decreasing the ball charge filling ratio it was possible to lower the specific energy
demand for grinding significantly.
It was obvious, too, that the L/D ratio influences the specific energy demand and the mill throughput as well. Therefore, a
huge test program was carried with a semi-industrial ball mill, which was operated in closed circuit. The mass-specific surface
area of the two feed materials (intermediate product) used were quite typical for industrial semi-finish grinding plants. The
values were 2200 and 3000 cm2/g according to Blaine. The product finenesses were 3000 and 3800 cm2/g, respectively. The L/
D ratio of the ball mill was varied in four steps of 1.75, 2.1, 2.79 and 3.49, and the ball charge filling ratio was varied in three
steps of 15%, 20% and 25%. The experiments clearly indicated that the optimal L/D ratio and the optimal ball charge filling
ratio are different for each feed fineness.
The influence of the ball charge grading on the specific energy demand, characterised by the average ball diameter, was
tested by means of a discontinuous laboratory ball mill. The results showed that by using a finer ball grading the specific energy
demand could be lowered considerably.
The obtained results can be explained well by theoretical considerations regarding the ruling stress intensity and the number
of stress events. The stress intensity expressed as the power input per ball is dependent on the ball diameter to the third power
and only slightly dependent on the inner diameter of the mill. The number of stresses can be characterised by the average
retention time of the ground material inside the mill if the ball charge grading remains unchanged. The optimal retention time
depends not only on the feed material and the desired comminution result but also on the ball charge filling ratio and particularly
on the L/D ratio. On the basis of the present results and considerations, a specific optimisation of ball mills in semi-finishgrinding plants can be done.
D 2004 Elsevier B.V. All rights reserved.
Keywords: ball mills; L/D ratio; finish-grinding

E-mail address: schnatzr@kp.thyssenkrupp.com.


0301-7516/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.minpro.2004.07.024

S56

R. Schnatz / Int. J. Miner. Process. 74S (2004) S55S63

1. Introduction
During the last decade, semi-finish-grinding plants
have been used more and more for the energy efficient
grinding of high-quality cement. Recent research
work has shown that the operational performance of
the ball mill, which is situated downstream of the
high-pressure grinding roll separator cycle can be
improved significantly. The results of extensive
investigations carried out on industrial semi-finishgrinding plants indicated that the specific power
consumption of the ball mill could be decreased by
about 20% when lowering the ball charge filling ratio
from about 30% to about 20%. However, the
throughput is decreased also by about 20% (Schnatz
and Knobloch, 2000).
Furthermore, it has shown that a L/D ratio of about
3.0, which is a quite common value for cement mills,
might be to high for a ball mill in a Combi-grinding
plant. This was proven by the fact that there was no
grinding progress observed in the last third of the total
grinding path length. This finding was not effected by
different ball charge filling ratios and different mill
throughput mass flows. Quite typical are the results
shown in Fig. 1 in which the mass-specific surface
area (Blaine) and the residues on the 32-, 63- and 90-

Am screens of meter-samples are plotted against the


grinding path length.
The ball charge filling ratio and the L/D ratio are
influencing the specific power consumption and the
throughput simultaneously. Furthermore, the absolute
fineness of the ball charge and the ball charge
grading are important factors for the optimal operation of a ball mill. Systematic investigations, which
are necessary for quantifying the different influences,
cannot be executed in industrial plants. Therefore,
the test runs were carried out in the R&D facilities of
the Polysius AG.

2. State of the art-L/D ratio and ball charge filling


ratio
The specific power consumption is, as already said,
influenced by the L/D ratio and the ball charge filling
ratio of a ball mill. The investigations known from the
literature were carried out with two compartment
cement mills in the 1980s (Kuhlmann, 1985;
T7tigkeitsbericht, 1991). This mills produced ordinary
Portland cement with a Blaine surface area of about
3000 cm2/g. The mill discharge had a fineness of
about 1800 cm2/g and the material which enters the

Fig. 1. Sieve residues on 32, 45 and 63 A and Blaine-specific surface area of meter samples as a function of the grinding path length.

R. Schnatz / Int. J. Miner. Process. 74S (2004) S55S63

second compartment had only a fineness of about 800


cm2/g. Whereas the material which enters the ball mill
in a Combi-grinding plant is much finer (~2000 cm2/
g). Therefore, this results canTt be transferred directly
to the one compartment ball mills in Combi-grinding
plants. But, nevertheless, it gives a good introduction
into the problems occurring.
The specific power consumption during this tests
had a broad optimum in the range of 26% ball charge
filling ratio.
The optimal L/D ratio range is quite broad again
with the minimum in the range of 3.2. For one
compartment mills, this optimum should be less due
to the lack of the first compartment used for coarse
grinding.
The optimal ball charge filling ratio for Combi-ball
mills is supposed to be less too. Shoji et al. (1982)
investigated the influence of the ball charge filling
ratio on the specific power consumption for the
grinding of quartz sand. They found a flat minimum
at a ball charge filling ratio of about 15%.

3. Test runs with the semi-industrial grinding plant


3.1. Description of the plant
The flow-sheet of the semi-industrial plant is
shown in Fig. 2. Fresh feed and separator grits were
combined and fed to the mill. The ball mill has an
inside diameter (inside liners) of 0.72 m. Its length
can be varied by combining different tube shells with
a length of 25 or 50 cm in the range of 1.253.5 m.

S57

Fig. 3. Semi-industrial ball mill.

Fig. 3 shows the mill. The material, which is


discharged pneumatically by the mill ventilation air,
is collected in a cyclone and a subsequent bag-filter
and fed-combined with the material which is discharged mechanically-to the high-efficiency separator
SEPOLR.
The ball mill is equipped with a classifying lining
with a conical undulated shape. The absolute lifting
height is quite similar to an industrial used liner
plate. The discharge diaphragm is of the slotted type.
The ball charge grading used during most of the test
runs was an industrial standard grading with balls
between 12 and 20 mm. The average ball diameter
was 15.8 mm; the average ball weight was 15.3 g/
piece. For the experiments in which a finer ball
grading was used a bored discharge diaphragm (bore
hole diameter: 4 mm) was built in. The air velocity
in the free mill cross-section was adjusted to 0.5 m/
s. The mill speed was kept constant at 37.4 rpm,
which relates to a relative mill speed of 75%. The
function of the classifying liners were examined
several times. The grading was very good in all
cases.
The desired product fineness was adjusted by
varying the circumferential speed of the rotor cage.
The circulation factor was limited to 3.0.
3.2. Test program

Fig. 2. Flow-sheet of the semi-industrial ball mill plant.

The two feed materials which were used in the test


runs were produced with a high-pressure grinding roll
which was operated in closed circuit with a static

S58

R. Schnatz / Int. J. Miner. Process. 74S (2004) S55S63

Fig. 4. Spec. power consumption and throughput in relation to the L/D ratio for the coarse feed material. Parameter: ball charge filling ratio.

separator. The feeds had finenesses of about 2200 and


3000 cm2/g according to Blaine, respectively. This
intermediate product finenesses are quite common in
plants, which are producing Portland cements type
CEM I 32,5 R and CEM I 42,5 R. The 0.02% grinding
aid (diethanolamine) was added to the fresh feed.
The feed were ground to product finenesses of
3000 and 3800 cm2/g, respectively. The system was
operated near its performance limit during the test

runs. In addition, some test regarding the optimisation


of the ball charge grading were executed.

4. Test results
The test results are plotted in Figs. 4 and 5. In both
figures, the specific power consumptionleft ordinate- and the throughput-right ordinateis plotted

Fig. 5. Spec. power consumption and throughput in relation to the L/D ratio for the fine feed material. Parameter: ball charge filling ratio.

R. Schnatz / Int. J. Miner. Process. 74S (2004) S55S63

against the L/D ratio. Parameter is the ball charge


filling ratio. Fig. 4 gives the results for the coarse feed
material (2200 cm2/g according to Blaine). Fig. 5
contains the results for the fine feed. Fig. 4 shows that
the throughput is maximal at a L/D ratio of about 2.7
2.9. In the same range, the spec. power consumption
has a minimum. For the grinding of the coarse feed
material, a ball charge filling ratio of about 25% or
may be more was found to be optimal because the
throughput was significantly higher (~20%) compared
to the tests at lower filling ratios and the spec. power
consumption was lower (~7%).
For the fine feed, there is no clear dependence of
the L/D ratio and the throughput. The throughput
remains nearly constant between L/D=2.12.7. A ball
charge filling ratio of 25% increases the throughput
only by about 4% compared to a filling ratio of 20%
and 10% at 15% filling ratio. The optimal L/D ratio
regarding the spec. power consumption is situated in
the range of 2.02.4. In contradiction to the coarse
feed, a lower ball charge filling ratio of about 18%
showed to be advantageous. For an industrial application, one has to decide what should be the
optimisation target: minimum spec. power consumption or maximum throughput.
For determining the comminution progress along
the grinding path, the mill was crash-stopped during
several test runs. Samples were taken in distances of
4050 cm along the whole grinding path length. The
particle size distribution of these samples were
analyzed by means of a laser diffraction spectrometer.

S59

The parameters of the RRSB-function (Rosin Rammler Sperling and Bennet equation) were fitted to the
PSDs.
The RRSB-slope n of the bmeterQ samples is
plotted as a function of the grinding path length in
Fig. 6. Parameter is the ball charge filling ratio. The
slope n tends towards a constant value along the
grinding path length. The high values of n for low L/
D ratios decrease and the low values of n for high L/D
ratios increase along the grinding path. Furthermore,
there seems to be no influence of the ball charge
filling ratio on the RRSB slope n.
It can be summarised that for ball mills in Combigrinding plants the optimal L/D ratio is clearly below
3.0. For coarse feed material about 2.7 seems to be
appropriate and about 2.3 for finer feed. Recommendations made by Haubold (2001) who found L/D ratios
in the range of 4.05.0 to be optimal are useful only for
ball mills in semi-finish-grinding plants, which are
operated in open circuit. The importance of open
circuit ball mills is diminishing because most of the
cement customers no longer accept coarse particles of
about 500 Am in the finished product, which cannot be
completely avoided with this type of mill.

5. Stress intensity and number of stress events


It is useful to analyse the obtained results theoretically to develop procedures for the target oriented mill
optimisation.

Fig. 6. RRSB slope n of meter samples for different L/D ratios.

S60

R. Schnatz / Int. J. Miner. Process. 74S (2004) S55S63

5.1. Theoretical considerations


The comminution process in ball mills is controlled
by
!
!

how often a particle of the material to be ground is


stressed (number of stresses) and
with which intensity the particle is stressed (stress
intensity).

The comminution progress or resultcharacterised, e.g., by the newly produced surface area
remains constant if the number of stress events (SN)
and the stress intensity (SI) are unchanged (Kwade
and Stender, 1998). Therefore, the mass-specific
power consumption W m for a defined comminution
result is proportional to the product of the number of
stresses and the stress intensity
Wm ~SId SN

One can calculate W m from the power draw of a


mill measured at shaft P and the product throughput
m
. The throughput itself can be expressed by using
the definition of the average retention time of the
material inside the mill.
Wm

P
P
m
t t
m
m
m

The stress intensity SI here defined as the average


energy, which is transferred to the material by each
ball, should be proportional to the power input per ball
P/n k .It is quite clear that the number of stress events
is directly proportional to the average retention time
of the material inside the mill. Taking this into
consideration Eq. (2) can be written as
 
P
nk 
t
Wm
3
nk
m
n k : number of balls.
The power absorption of industrial ball mills at
Polysius has been determined for decades with an
empirical equation originally developed by Blanc
(Blanc and Eckhardt, 1928).
P

p
L
cu q 1  e D3;5
4 k k
D

Beside the mill inside diameter D and the L/D ratio


the porosity of the ball charge e, the true density of the
grinding media q k , the ball charge filling ratio u k and
an empirical factor c go into the equation. c is
dependant on the ball charge filling ratio, the used ball
grading and the mill lining.
For a constant ball diameter d, the number of balls
inside the mill can be calculated as follows
3
nk
2

D
d

3

L
u 1  e
D k

The combination of Eqs. (4) and (5) gives an


expression for the energy input per ball P/n k , which
should be proportional to the stress intensity


P
BI~
nk

p p 3
Dd cqk
6

The principal relationships can be seen from Eq.


(6): The energy per ball is linearly dependent on the
true density of the balls and the c-factor. The biggest
influencing factor is the ball diameter (to the third
power). Furthermore, interesting is the fact that more
energy is transferred per ball in large diameter ball
mills.
The second part of Eq. (3) can be written as Eq. (8)
by using a relationship for the amount of feed material
m, which stays inside the mill at a time.
m

p 3L
D
qu qe1  e
4
D k

q: true density of the feed material.



BH~


nk
6 1 t
t
p e qd 3
m

Eq. (8) shows that the number of stress events is


proportional to the retention time of the material
inside the mill and anti-proportional to the true density
of the material and the ball diameter (to the third
power).
5.2. Stress intensity and ball diameter
The ball diameter, which is able to comminuting a
particle of defined size, can be calculated according to

R. Schnatz / Int. J. Miner. Process. 74S (2004) S55S63

Bond (1958) or Perow and Brand (1954) with an


equation of the following form
p
da x

a: fitting parameter and x: particle size.


For cement clinker a can be set to a value of 24
with sufficient accuracy. A mill feed with a Blaine
surface area of 1500 cm2/g will not contain particles
N500 Am. Therefore, the largest balls in a mill should
have a diameter of about 17 mm. For 50-Am particles,
a 5-mm ball should be sufficient.
This well-known but often neglected facts were
investigated regarding the comminution of intermediate product in Combi-grinding plants. For that
purpose, tests were carried out with a discontinuous
laboratory ball mill. This mill at the Polysius R&D
centre is used for grindability testing mainly. Seventeen different ball gradings were tested. The lab-mill
(F 0.750.45 m, ball charge filling ratio 10%,
relative mill speed 0.88) were fed with clinker meal
with a Blaine surface area of 2700 cm2/g. The massspecific power consumption for the grinding to
finenesses of 3000, 4000 and 5000 cm2/g were
determined. The results are plotted in Fig. 7.
The figure contains the mass-specific power consumption in relationship to the average ball diameter
of the different ball gradings. It can be seen that the

S61

power consumption decreases with decreasing ball


diameter. The potential savings are increasing with
increasing product fineness. It can be concluded that
the finer the desired product is the finer the used ball
grading should be. Polydisperse ball gradings showed
to be slightly more efficient compared to monodisperse gradings with the same average ball diameter. A
direct transfer of this results to industrial ball mills,
which are operated continuously, however, is not
possible. This is caused by the fact that in continuous
mills the material to be ground has to be conveyed
through the ball charge. Very fine balls with diameters
b8 mm tend to swim up. Furthermore, with finer balls,
the tendency of the material to form agglomerates,
which stick to the mill liners increases. Fine balls with
a low weight do not have enough energy to clean the
liners from that agglomerates. This is a limitation for
the use of very fine balls. But it is possible to suppress
the agglomeration by an increased addition of grinding aid.
5.3. Number of stress events and retention time
The results obtained with the semi-industrial ball
mill (see Figs. 4 and 5) were analysed regarding the
number of stress events (calculated according to Eq.
(6)). For an unchanged ball grading and the same true
density of the feed material the number of stress

Fig. 7. Spec. power consumption of a laboratory ball mill as a function of the average diameter of the ball grading. Parameter: fineness of the
finished product.

S62

R. Schnatz / Int. J. Miner. Process. 74S (2004) S55S63

Fig. 8. Spec. power consumption for the coarse feed materials a function of the retention time of the feed inside the mill. Values of Fig. 4.

events can be substituted by the average retention time


of the feed inside the mill. The calculated values for
the coarse and the fine material are plotted in Figs. 8
and 9. Both figures show that the spec. power
consumption has as significant minimum at a retention time of about 7 min. The retention is influenced
by the ball charge filling ratio and the L/D ratio. The
fact that this optimal retention time is achieved with
different L/D ratios for fine and coarse feed material
indicates that the transport behaviour of the material is
an important factor too.

6. Conclusion and outlook


On the basis of the results and considerations
presented in this paper, a target-oriented optimisation
of ball mills in Combi-grinding plants is possible. The
obtained results can be explained well by theoretical
considerations regarding the ruling stress intensity and
the number of stress events. The stress intensity
expressed as the power input per ball is dependent on
the ball diameter to the third power and only slightly
dependent on the inner diameter of the mill. The

Fig. 9. Spec. power consumption for the fine feed materials a function of the retention time of the feed inside the mill. Values of Fig. 5.

R. Schnatz / Int. J. Miner. Process. 74S (2004) S55S63

number of stresses can be characterised by the average


retention time of the ground material inside the mill if
the ball charge grading remains unchanged. The
optimal retention time depends not only on the feed
material and the desired comminution result but also
on the ball charge filling ratio and particularly on the
L/D ratio. The influence of the factors ball charge
filling ratio and L/D ratio on the retention time of the
material to be ground has to be investigated in detail.
Furthermore, the extension of the test regarding the
influence of the mill speed on the retention time is
necessary.

References
Blanc, E.C., Eckhardt, H., 1928. Technologie der Brecher, Mqhlen
und Siebvorricht ungen. Springer Verlag, Berlin.

S63

Bond, F.C., 1958. Grinding ball size selection. Mining Engineering


10 (5), S. 592 S. 595.
Haubold, S., 2001. Verbesserung des Arbeitsergebnisses von
Mahlanlagen fqr Zement nach der Technologie der Teilfertigmahlung. ZKG International 54 (10), 556 565.
Kuhlmann, K., 1985. Verbesserung der Energieausnutzung beim
Mahlen von Zement. Schriftenreihe der Zementindustrie, Heft,
vol. 44. Beton Verlag, Dqsseldorf.
Kwade, A., Stender, H.-H., 1998. Konstantes Zerkleinerungsergebnis beim Scale-Up von Rqhrwerkskugelmqhlen. AufbereiguntsTechnik 39 (8), S. 373 S. 382.
Perow, W.A., Brand, W.J., 1954. Feinmahlen der Erze. VEB Verlag
Technik, Berlin.
Schnatz, R., Knobloch, O., 2000. Einfluss des Kugelfqllungsgrades
auf den Energieverbrauch und den Durchsatz von Kugelmqhlen in Kombi-Mahlanlagen. ZKG International 53 (8),
S. 438 S. 447.
Shoji, K., Austin, L.G., Smaila, F., Brame, K., Luckie, P.T., 1982.
Further studies of ball and powder filling effects in ball milling.
Powder Technology 31, S. 121 S. 126.
Tatigkeitsbericht, 1991. Verein Deutscher Zementwerke e.V.. Beton
Verlag, Dusseldorf. 1987-90.

Вам также может понравиться