MINPRO-02659; No of Pages 5

International Journal of Mineral Processing xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

International Journal of Mineral Processing

journal homepage: www.elsevier.com/locate/ijminpro

An energy based comparison of vertical roller mills and tumbling mills

A. Boehm a,⁎, P. Meissner b, T. Plochberger c
a
Institute of Mineral Processing, Montanuniversitaet Leoben, Max Tendler Strasse 4, 8700 Leoben, Austria
b
Institute of Mineral Processing, Montanuniversitaet Leoben, Max Tendler Strasse 4, 8700 Leoben, Austria
c
CEMTEC Cement and Mining Technology, Ennshafenstrasse 40, 4470 Enns, Austria

a r t i c l e i n f o a b s t r a c t

Article history: The Institute of Mineral Processing at the Montanuniversitaet Leoben in cooperation with the company CEMTEC
Received 2 January 2014 has developed a pilot scale, 200 mm table diameter, vertical roller mill for energy controlled laboratory tests. The
Received in revised form 19 September 2014 mill provides the technical options to vary process parameters like air-flow, mass flow, grinding force and clas-
Accepted 24 September 2014
sifier speed in a wide range and is equipped to analyze the internal circulating load. In order to address grinding
Available online xxxx
efficiency (defined as the increase in mass specific surface vs. net specific energy input) the energy consumption
Keywords:
of two pilot scale vertical roller mills (200 mm and 670 mm table diameter) was compared to that of a sequence
Vertical roller mill of laboratory comminution equipment covering the same size range. The sequence consisting of a laboratory jaw
Tumbling mill crusher and three laboratory tumbling mills of differing grinding media was operated following the principles of
Energy efficiency energy optimized comminution according to the OCS-method. With respect to the grinding energy only all the
Optimized comminution sequence method results for marble, siderite and hematite ore show energy savings when using the vertical roller mill at optimized
(OCS) settings. The methods used are outlined including the special features of the equipment.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction below 10 kW power supply was measured mechanically. The specific

energy consumption related to the generated mass of fines and the
Comminution is one of the most energy intensive mechanical treat- characterization of the dispersity by the Kozeny surface form the basis
ments in common mineral processing plants. In consideration of of comparison.
increasing energy prices and the demand for sustainable industrial
processes the reduction of energy expense attracts more and more
2. Methods, equipment and materials
attention. Although the potential of alternative methods such as high
voltage treatment or material's pretreatment by ultrasonic or micro-
2.1. The OCS method
waves is intensively investigated, tumbling mills and vertical roller
mills still represent the state of the art of comminution in a size range
The concept (Steiner, 1991) of energy optimized comminution
between 102 and 10−2 mm. The vertical roller mill equipped with an
(OCS) is based on the idea to establish a comminution process – at
internal classifier is often said to be more energy efficient than tumbling
least at the laboratory scale – that minimizes the energy expense (Δe)
mills working on the same material and the same size step (e.g. Gerold
for a given size reduction step i (refer Fig. 1). According to Steiner the
et al., 2012). It is long established in the cement industry and gains more
general principles of energy optimized comminution (which does not
and more importance in the industrial mineral and iron ore sectors. In
necessarily correspond to cost optimized comminution) are defined as
order to address the energy consumption of the grinding tools a labora-
follows:
tory as well as a pilot scale vertical roller mill were benchmarked by the
Subdivide a given total comminution step into a sequence of size
OCS-method.
reduction steps of low size reduction ratio using comminution tools
The lab scale comminution tools used in this (energy) optimized
optimized to the feed size. For an optimum mechanical energy transfer
comminution sequence (OCS) are arranged and operated in a way to
from the comminution tool to the material, keep the amount of fines as
reveal the minimum energy expense for a defined size reduction step.
small and the retention time of fines after their generation in the
The energy input into the grinding chamber of all the lab systems
comminution chamber as short as possible. The fines cause the absorp-
tion of energy due to compaction. A closed circuit design with
prescreening and high circulating load result from these requirements.
⁎ Corresponding author at: Institute of Mineral Processing, Montanuniversitaet Leoben,
Max Tendler Strasse 4, Leoben 8700, Austria. Tel.: +43 3842 402 1804, +43 664 80898
The general setup of an OCS is shown in Fig. 1.
1804 (mobile). Close to Bond's method the continuous operation is simulated in
E-mail address: Andreas.Boehm@unileoben.ac.at (A. Boehm). each step by batch wise comminution with pre- and intermediate

http://dx.doi.org/10.1016/j.minpro.2014.09.014
0301-7516/© 2014 Elsevier B.V. All rights reserved.

Please cite this article as: Boehm, A., et al., An energy based comparison of vertical roller mills and tumbling mills, Int. J. Miner. Process. (2014),
http://dx.doi.org/10.1016/j.minpro.2014.09.014
2 A. Boehm et al. / International Journal of Mineral Processing xxx (2014) xxx–xxx

2.2. Tumbling mills and jaw crushers

The sequence to evaluate the vertical roller mills consisted of a labo-

ratory jaw crusher, a rod mill and two ball mill stages with differing
i-1  ei-1 grinding media. The dimensions and settings of the laboratory mills
S Z are given in Table 1 and are illustrated in details in (Boehm and
Flachberger, 2006). The jaw crusher used is a standard jaw crusher
ei-1 (RETSCH BB 100). The gap width is usually set to obtain a product size
of 100% −6.3 mm at a circulating load of 50%. The size reduction ratio
should not exceed 1:4. The rod mill is typically operated within the
i
S Z  ei size reduction step 100% α6.3 to 100% −1 mm at a circulating load of
100%. The mass specific energy input ΔeRod [J/kg] is derived from
Eq. (2). Details are given in Steiner (1996). Calibration by means of
ei torque measurement revealed a dimensionless power conversion factor
for the mechanical power transferred into the material cp of 1.1. The
mass of grinding media (MK, kg) and the inner diameter (Di, m) are
i+1  ei+1
S Z listed in Table 1. The number of revolutions (U, 1) is adjusted to the de-
sired circulating load, which is defined as the mass of recycled material
ei+1 to the mass of produced fines (MF, kg). g (m/s2) corresponds to the
gravitational constant.

U
ΔeRod ¼ cp  g  MK  Di  ð2Þ
Fig. 1. The optimized comminution sequence (OCS) (Steiner, 1991). Flow sheet of the cir- MF
cuit design of the laboratory comminution (Z) and classification (S) apparatus.

The tumbling mill used with different types of grinding media is

screening and classification respectively. Common laboratory commi-
nution equipment (jaw crushers, rod mills, ball mills) is arranged in
the sequence. The circulating load is adjusted to 100% for rod mill oper-
ation, to 250% and 300% for mill operations using balls and cylpebs
respectively. The energy input into the grinding chamber is controlled
by torque measurement. All intermediate classification down to
200 μm is done by manual screening, which can be kept closest to
ideal classification at a given separation size. For the −40 μm circuit a
laboratory air classifier (Hosokawa, 100MZR) is used. To all experi-
mental evidence at the Institute of Mineral Processing at the
Montanuniversitaet Leoben up to now, the method reveals the product
size distribution of the lowest variety of particle size classes at a given
maximum particle size.
For homogenous materials the size distribution often linearizes in
the Gates, Gaudin, and Schuhmann (GGS) plot indicating a potential
relation between the particle size and the percentage passing. As no
other machinery was found up to now to produce a distribution of
steeper inclination it is considered to be material inherent and is called
the “natural breakage characteristic” (NBC) according to Steiner.
Plotting the mass specific surfaces of the feed and the products
determined by permeametry versus the assigned mass specific energy
consumption, the course of measurement results can be fitted best
by a straight line. As the experimental findings (down to kmax
of −40 μm) indicate linear dependence between new generated surface
(Δa, cm2/g) and energy input (Δe, J/g) in the limiting case of energy op-
timized comminution, the line is called the “Rittinger line” (Eq. (1)). The
inclination as the measure of the material's grindability is called the
Rittinger coefficient (R, cm2/J).
equipped with a torque rod for direct energy measurement. The size
reduction steps are set to 100% − 1 mm to 100% − 0.2 mm for ball
milling and 100% − 0.2 to 100% − 0.04 mm using the cylpebs. The
mean torque (Md, Nm) with time derived from the recorded data to-
gether with the number of mill revolutions (U, 1) serves to calculate
the specific energy consumption (ΔeBM, J/kg) related to the produced
fines (MF, kg) (Eq. (3)).
2  π  Md  U
ΔeBM ¼ ð3Þ
MF
2.3. The vertical roller mills VRM200 and VRM600
The VRM200 is a laboratory scale vertical roller mill with two trun-
cated conical grinding rollers (LOESCHE system) of 133 mm mean
diameter, a cross flow rotary cage classifier with a wheel diameter of
160 mm, a bag filter and full sensor configuration. The mill table has a
diameter of 200 mm allowing a maximum throughput rate of about
240 kg limestone per hour at a classifier setting of d98 = 200 μm. The
mill is described in more detail in Meissner et al. (2012). Fig. 2 gives
insight into the grinding chamber. A special feature of this mill is the
external classifier drawback that makes the coarse product of the classi-
fier accessible. The experiments however were carried out in common
mode with the drawback disabled.
The mill power draw is measured by a torque rod situated in line
with the main drive shaft. The specific power consumption
ΔeVRM200 in J/kg could be calculated by the control system according
to Eq. (3). Details of the VRM600 in the CEMTEC laboratory are pub-
lished in Gleissenberg and Hofmann (2008). The mill table diameter is
around 670 mm forced by three conical grinding rollers of 430 mm

Δa ¼ R  Δe ð1Þ Table 1
Tumbling mills — parameters.

Size or parameter Rod mill Ball mill Cylpeb mill

Apart from the evaluation of grindability in tumbling mills and
Inner shell diameter, [m] 0.150 0.200 0.200
benchmarking grinding systems, the method provides reproducible Length, [m] 0.3 0.2 0.2
size distributions, characteristic for one type of material, which is need- Mass of grinding media, [kg] 70.47.4 9 12
ed e.g. for integrated intergrowth investigations. The grindability index Percentage of critical speed, [%] 70 70 70
R varies between 20 cm2/J for low grindability (thermally compacted Degree of filling, [vol-%] 40 40 40
Circulating load, [%] 100 250 300
slags) and 150 cm2/J (unconsolidated limestone).

Please cite this article as: Boehm, A., et al., An energy based comparison of vertical roller mills and tumbling mills, Int. J. Miner. Process. (2014),
http://dx.doi.org/10.1016/j.minpro.2014.09.014
 A. Boehm et al. / International Journal of Mineral Processing xxx (2014) xxx–xxx
Fig. 2. VRM200 laboratory vertical roller mill. Institute of Mineral Processing,
Montanuniversitaet Leoben. View on the grinding table, feed tube and the two conically
shaped rollers, adjustable in distance to the table.
mean diameter. The power draw is derived from the system data of the
main frequency converter.
2.4. Test settings of the vertical roller mills
The main process factors to vary are the grinding pressure, the table
speed, the classifier speed, the air flow and pressure drop and feed rate
respectively. The feed of the VRM600 (100% −14 mm) was scalped at
8 mm to obtain the feed for the VRM200. The size distributions were
not changed during the various settings.
Over all experiments stable process conditions must be guaranteed.
A constant pressure drop (at fixed feed rate), constant air flow, a stable
mass balance (feed mass equals product mass) and stable product
fineness over a 1.5 hour period are taken as the constraints for valid
experimental setting. The mill parameter settings for both mills are
shown in Table 2.
To guarantee an energy optimized grinding process, at the parame-
ter settings given in Table 2, the mass throughput was pushed to the
limit until one stability criterion was violated. Three respective four
mill products of varying fineness per mill and material were used for
benchmarking. The fineness was defined by the classifier speed.
2.5. Specific surface measurement
The mass specific surfaces of all the comminution products were
measured by permeametry. An average value was calculated out of
two obtained by Blaine's apparatus and two by the Permaran of
Outokumpu respectively. For evaluating the entire surface of coarse
comminution products (−500 cm2/g) the method combines the mea-
surement of the Kozeny surface in the two smallest size classes and
the calculation of the specific surface in the remaining upper size
classes.
The measurement of the specific surface in the sieve size classes (e.g.
−100 + 40 μm and −40 μm) is followed by the derivation of a material
related shape factor (f, 1) from the measured surface (aV, m−1) of the
Table 2
VRM parameter settings.
VRM200 VRM600
Table speed, [rev/min] 92 65
Normative grinding pressure, [kN/m2] 600 670
Classifier speed, [rev/min] 600–4000 600–2600
Air flow, [m3/h] 840 5000
Distance roller to table, [mm] 0.8 5
Please cite this article as: Boehm, A., et al., An energy based comparison of vertical roller mills and tumbling mills, Int. J. Miner. Process. (2014),
http://dx.doi.org/10.1016/j.minpro.2014.09.014
3
size class with defined size limits by means of the surface equivalent
particle size (Eq. (4)). The surface equivalent particle size (kequ, m) is
deduced from the M2,3 moment of the density function fitting the
best the measured size distribution. The shape factor, which is consid-
ered to be constant for all the size classes, allows the calculation of the
mass specific surfaces of those size classes beyond the measuring
range by (Eq. (4)). The method is described in detail in Boeh, (2010).
f ¼ kequ  av ð4Þ
2.6. Tested materials
In order to cover a range of grindabilities but to avoid differential
grinding behavior, marble of constant composition (provenience:
Austria, density 2.71 g/cm3), a spateisenstein concentrate from the
Steirische Erzberg (provenience: Austria, average density 3.51 g/cm3)
mainly consisting of the siderite variety spateisenstein and ankerite as
well as hematitic sinter fines from the Ukraine (average density
4.65 g/cm3), were used as feed materials. The comminution resistances
characterized by the Rittinger index (refer to Fig. 3) were found to be
68 cm2/J for the marble, 51 cm2/J for the siderite ore and 40 cm2/J for
the hematite ore.
The materials were all crushed and screened to a maximum feed size
of 8 mm in case of the VRM200. For the VRM600 the materials were
crushed in open circuit to a maximum feed size of about 14 mm.
3. Results and discussion
Fig. 3 summarizes all the results. The drawbacks of permeametry in
measuring absolute surface values are widely discussed. Applying the
method for relative comparison the obtained results are significant
within the variation limits of ± 5%. Regarding the reliability of the
torque measurement, the torque rods of the VRM200 and the ball mill
are calibrated and controlled in static mode. The energy consumption
of the VRM600 includes losses of the motor and the drive, while the
torque rods measure the energy input into the grinding chamber.
Referring to Fig. 3 the points of measurement for the roller mills
are not interpolated by a curve, as all the products of the roller
mills result from the same maximum feed size (100% − 14 mm re-
spectively 100% − 8 mm). By connecting every single point with
the origin, one will find the grindability assigned to every vertical
mill test. The data of the OCS however were collected in series,
with decreasing feed size for subsequent comminution steps.
The linear relation between mass specific surface and the cumulative
energy consumption within the OCS was found experimentally for all
the tested materials up to a surface of 4000 Blaine and energy inputs
up to 80 kWh/t. Results achieved by arbitrary comminution tools situat-
ed right from the Rittinger line indicate less energy efficient systems.
At the Institute of Mineral Processing at Montanuniversitaet Leoben
the material characteristics – the natural breakage characteristic (NBC),
the Rittinger index (R) and the shape factor (f) – have been recorded in
this way for more than 30 years to characterize the comminution
behavior in tumbling mills.
All the results achieved by the VRM200 and the VRM600 are situated
to left of the Rittinger line and thus clearly indicate lower energy
consumption to obtain the same increase of mass specific surface.
In order to compare the results of the two roller mills and to put
them in relation to data given in literature the term “energy savings”
is defined as the difference of the energy consumptions of the vertical
mill and OCS at a given surface related to the energy consumption
derived from the OCS.
The energy savings of the vertical roller mill decrease for marble and
hematite with increasing surface from about 62% at 2000 Blaine and
3400 Blaine respectively to 35% at 6000 Blaine. For the siderite ore the
energy savings show different behavior increasing from 48% at 3000
4 A. Boehm et al. / International Journal of Mineral Processing xxx (2014) xxx–xxx
10000
OCS
VRM200 marble
mass specific surface, cm²/g
8000 VRM600 marble
6000
4000
2000
R = 68 cm²/J
0
0 20 40 60 80 100 120 140 160
Fig. 3. Results of energy measurement. Results of the cumulative mass specific energy vs. the generated mass specific surface obtained by the OCS method and vertical roller mills for cal-
cite, siderite ore concentrate and hematite ore concentrate.
Blaine to 60% at 6000 Blaine. Reasons for this may be found in the min-
eral structure showing extraordinary increase in the surface below
40 μm.
Compared to the VRM200 the savings for the VRM600 are signifi-
cantly lower in the range beyond 6000 Blaine depending on the type
of material (about 15% for the marble, 25% for the siderite and 1.5% for
the hematite).
The difference in the behavior of the roller mills can be explained on
the one hand by the different design of the classifiers. At the VRM200
the classifier operates most efficiently between cut sizes (d98) from
70 to 100 μm, while the classifier on the VRM600 is optimized for d98
from 10 to 25 μm. On the other hand the much higher throughput rate
and restrictions in experimental time confined the search for the opti-
mum settings for each material. In addition the energy data of the
VRM600 contain the losses of the motor and the drive.
According to the most frequent application of vertical mills most of
the data reported in literature refer to cement and slag. From data pub-
lished by Joergensen (2004), energy savings for grinding ordinary
Portland cement (OPC) at increasing mass specific surface from 3200
to 3900 Blaine from 32% to 48% can be calculated. Including the energy
expense for classifying and transport the savings drop to 24 and 39%
respectively. Slag cement of higher comminution resistance is reported
to have even better savings (50% respectively 36% at 3640 Blaine). For
the nominal same type of material different results can be derived
from (Fahrland and Zysk (2013) for OPC with 14% at 3200 Blaine up
to 26% at 4000 Blaine and from 48% at 3600 to 58% at 5000 Blaine for
slag. As these values are reported for the total grinding system, the
savings with respect to grinding energy only should be higher. The
potential of the absolute values is not easy to assess due to the incom-
plete data reported. The distinct characterization of the material by its
comminution behavior, the characterization of the machinery
concerning type, setup, operation and measurement conditions (energy
contributions included, method of measurement) would be important.
Referring to Fig. 3 the opposite behavior in the energy savings
concerning the relation between surface increase and energy consump-
tion may be explained by the differing points of reference. It may be
assumed that one arrangement of ball mill and classifier served to
obtain all the results. For tumbling mill systems operated more and
more beyond their optimum settings, the grindability (surface to energy
relation) monotonously decreases following a curve, for which the
Rittinger line serves as the tangent in the origin (Steiner, 1991).
Please cite this article as: Boehm, A., et al., An energy based comparison of vertical roller mills and tumbling mills, Int. J. Miner. Process. (2014),
http://dx.doi.org/10.1016/j.minpro.2014.09.014
OCS OCS
VRM200 siderite VRM200 hemate
VRM600 siderite VRM600 hemate
R = 51 cm²/J R = 40 cm²/J
0 20 40 60 80 100 120 0 20 40 60 80 100 120 140 160 180
cumulative mass specific
energy consumption, J/g
In order to obtain a steep particle size distribution with a low
content of fines beyond 40 μm at a maximum particle size of 1 mm com-
parable to that of the rod mill (620 Blaine), additional grinding tests
with the siderite ore using the VRM200 in overflow mode and circulat-
ing load of 100% adjusted by a screen (0.5 mm) were made. The table
speed was set too high for the buildup of a particle bed. There were no
energy savings to be found compared to the rod mill. This may serve
as an additional hint that energy savings result from the better energy
utilization in the particle bed compared to that of stochastic grinding
in tumbling mills.
More detailed experiments are subject of ongoing research work to
understand the interaction of high circulating load, classifier efficiency
and energy utilization in the particle bed.
4. Conclusions
The paper aimed to provide traceable measurement results to
compare the energy consumption for grinding only of ball mills and
vertical roller mills using lab scale equipment. At similar stress condi-
tions to industrial mills the energy input into the material could thus
be obtained by using torque rods, that are free of the losses immanent
to power measurement at the power supply. In addition the settings
of the apparatus can be adjusted to an optimum with respect to the
specific energy consumption by increasing the throughput rate to a
maximum at stable conditions for the vertical roller mills and by using
the optimized comminution sequence for the tumbling mills.
Three types of material at decreasing comminution resistance (mar-
ble, siderite ore and hematite ore) were tested in a range of product
dispersity between 2000 and 7000 Blaine.
Even at optimized conditions stochastic grinding in ball mills could
not overcome the results obtained by the vertical mills stressing the
material in the compressed bed. The energy savings seem to be material
dependent with no distinct correlation to the comminution resistance
described by the Rittinger index.
References
Boehm, A., 2010. Derivation of the specific surface area and the particle shape factor from
permeametry. In: Messe GmbH Nuernberg (Ed.), World Congress Particle Technology
6, CD Proceedings. ISBN: 978-3-00-030570-2, pp. 736–739.
Boehm, A., Flachberger, H., 2006. Ueberblick ueber Methoden der Mahlbarkeitspruefung.
BHM Berg- und Huettenmaennische Monatshefte 151, 223–232.
 A. Boehm et al. / International Journal of Mineral Processing xxx (2014) xxx–xxx 5

Fahrland, T., Zysk, K.-H., 2013. Cements ground in the vertical roller mill fulfil the quality
requirements of the market. Cem. Int. 2 (2013), 64–70.
Gerold, C., et al., 2012. Latest installations and developments of Loesche vertical roller
mills in the ore industry. Proc. XXVI Int. Min. Proc. Congr. New Delhi, pp. 1018–1029.
Gleissenberg, J., Hofmann, F., 2008. Tailored solutions for the success of our customers.
Aufbereitungstechnik 49, 17–24.
Joergensen, S.W., 2004. Cement grinding — vertical mills versus ball mills. 13th Arab
International Cement Conference and Exhibition, Muscat, Oman. Papers, Pt.2 (16 p.).
Meissner, P., Boehm, A., Plochberger, T., 2012. Improved grinding. AT int. 53, 252–258.
Steiner, H.J., 1991. The significance of the Rittinger-equation in present day comminution
technology. Proc. XVII. Int. Min. Pro. Congr. Dresden Vol. I, pp. 177–188.
Steiner, H.J., 1996. Characterization of laboratory scale tumbling mills. In: Forssberg,
K.S.E., Schoenert, K. (Eds.), Comminution 1994. Elsevier Science B.V, Amsterdam,
pp. 373–382.

Please cite this article as: Boehm, A., et al., An energy based comparison of vertical roller mills and tumbling mills, Int. J. Miner. Process. (2014),
http://dx.doi.org/10.1016/j.minpro.2014.09.014