j m a t e r r e s t e c h n o l .

2 0 1 4;3(1):86–89

Available online at www.sciencedirect.com

www.jmrt.com.br

Original Article

Scale up and simulation of VertimillTM pilot test operated

with copper ore

Douglas Batista Mazzinghy a,∗ , Roberto Galéry a ,

Claudio Luiz Schneider b , Vladmir Kronemberger Alves c
a Programa de Pós-Graduação em Engenharia Metalúrgica, Materiais e de Minas, Universidade Federal de Minas Gerais, Belo Horizonte,
MG, Brazil
b CETEM – Centro de Tecnologia Mineral, Rio de Janeiro, RJ, Brazil
c Vale S.A. – Centro de Desenvolvimento Mineral, Santa Luzia, MG, Brazil

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

Article history: VertimillTM has been used in regrind circuits during the past years due to its greater efficiency
Received 6 November 2013 when compared to the conventional tube ball mill. This paper presents the VertimillTM pilot
Accepted 28 November 2013 grinding tests with a sample of copper ore carried out in closed circuit with a high frequency
Available online 16 January 2014 screen. The sample was crushed 100% to <6 mm. All operating variables were measured
under controlled conditions to produce a mass balance of the test. A laboratory batch ball
Keywords: mill was used to characterize the sample in order to determine the energy specific selec-
Vertimill TM tion function and the breakage function. Previous studies have shown that the VertimillTM
Breakage parameters produces larger values of the selection function and a constant scaling factor can be used
Grinding to simulate the product particle size distribution of the VertimillTM from the batch ball mill
Simulation grinding tests. The results of the simulations showed that it is possible to estimate the
Copper ore product particle size distribution of the VertimillTM pilot scale from breakage parameters
determined from a lab-scale batch ball mill. These results confirm that the VertimillTM and
the conventional ball mill use similar mechanisms of impact and that the main difference
between them is the intensity and frequency of the impacts.
© 2013 Brazilian Metallurgical, Materials and Mining Association. Published by Elsevier
Editora Ltda. All rights reserved.

tests with small quantities of samples were carried out in lab

1. Introduction to determine breakage parameters to simulate this type of
mill. Some studies have showed that it is possible to simulate
An investigation program of VertimillTM was conducted with the VertimillTM using the methodology that considers batch
the aim of development of models to scale up and simulate ball mill tests in different interval times and breakage param-
this type of mill. Pilot tests with VertimillTM was carried out eters given by software that performs the optimization of the
under controlled conditions to produce a mass balance and particle size distributions [1–3].

∗
Corresponding author.
E-mail: douglasmazzinghy@ufmg.br (D.B. Mazzinghy).
2238-7854/$ – see front matter © 2013 Brazilian Metallurgical, Materials and Mining Association. Published by Elsevier Editora Ltda. All rights reserved.
http://dx.doi.org/10.1016/j.jmrt.2013.11.001
 j m a t e r r e s t e c h n o l . 2 0 1 4;3(1):86–89 87

Why the VertimillTM is more efficient than the tube ball mill? where (di /d1 ) is the dimensionless particle size (normalized at
Apparently, the answer is the spectrum of energies. VertimillTM 1 mm), SE∗
1 ,  1 ,  2 are characteristic parameters of the material
has higher frequency of lower energy impacts and smaller and the conditions of grinding.
frequency of higher energy impacts when compared to con- This model was implemented in the ModsimTM plant-wide
ventional ball mills [4]. simulator [9]. For models references of ModsimTM is strongly
The objective of this paper was to simulate a pilot recommendable to read King [10].
VertimillTM test from batch mill lab tests using small quantities
of samples.
2. Experimental
1.1. Modeling approach
2.1. VertimillTM pilot test
The population balance model was formulated for chemi-
cal engineering purposes [5]. The size-mass balance model A campaign of VertimillTM pilot tests was carried out at Metso’s
that describes the batch grinding process through successive York pilot plant facility. The VertimillTM was operated in direct
events of particle breakage is given in Eq. (1) [6]. circuit configuration closed with a high frequency screen (Der-
rick Vibrating Screen) and with screw speed of 87 rpm. The mill
motor gross power registered during the VertimillTM pilot test
dmi (t)  i−1
= −Si mi (t) + bij Sj mj (t) (1) was adjusted with Prony brake curve data to provide net power
dt draw. A mass balance of the test was produced through the
j=1
samples collected from all streams for solids concentration,
where, mi (t) is the mass fraction of particles contained in size density and particle size distribution determination. Fig. 1
class i after grinding time t; bij represents the size distribution shows the VertimillTM Pilot Circuit Flowsheet.
produced by a breakage event; Si represents the specific rate The samples were crushed 100% to <6.3 mm following
of breakage of particles in size class i. Metso recommendation. Fig. 2 shows the feed size distribution
of the sample.
1.1.1. Breakage function Table 1 shows the media size distribution used in the
The breakage function model is given in Eq. (2) [6]. Bij is the VertimillTM pilot test.
cumulative breakage function and the parameters ϕ, , ˇ are
dependent on the ore.

   ˇ
xi−1 xi−1
Bi,j =  + (1 − ) · (2)
xj xj

1.1.2. Specific selection function

The selection function for each size class, Si , presents a pro-
portionality relationship with the power consumed by the
grinding action according to Eq. (3) [7].

P
Si = SEi (3)
H
Fig. 1 – VertimillTM pilot circuit flowsheet.
Si (h−1 ) is the selection function for each size class, SEi (t/kWh)
is the energy specific selection function, H(t) is mill holdup and
P(kW) is net grinding power. 100
These equations have been used to determine the energy
consumption required in grinding and they can also be used
80
in the scale-up of industrial milling circuits from parameters
obtained using a simple laboratory batch mill.
% Passing

60

1.2. Scale up
40
The specific selection function SEi is independent of the dimen-
sions of the mill and may be modeled using Eq. (4) [8]. The
20
parameter SE1 was multiplied by a scaling factor of 1.35 to rep-
resent the highest efficiency of the VertimillTM , becoming SE∗ 1 .
0
The scaling factor of 1.35 is applicable for different ores [1]. 0.01 0.10 1.00 10.00

 d    d  2

Size (mm)

i i
SEi = SE∗
1 exp 1 ln + 2 ln (4) Fig. 2 – Particle size distribution of the VertimillTM pilot
d1 d1
circuit feed.
88 j m a t e r r e s t e c h n o l . 2 0 1 4;3(1):86–89

100
Table 1 – Media size distribution.
Size (mm) % Ret.
80
35 38.4
0
30 27.5
25 14.5 2 min

% Passing
60
19 17.7 23 min
12 1.9
39 min
Total 100 40
48 min

34 min
20
Table 2 – Batch mill test.
Mill diameter (m) 0.254
Mill length (m) 0.254 0
J – ball charge level 0.40 10 100 1000 10000
U – voids 1.00 Size (µm)
Critical speed 0.70
Fig. 3 – Particle size distribution of the batch ball mill test
and the model fitting by BatchMillTM software.
2.2. Lab batch test

The batch tests were carried out in a conventional batch tube

mill. Samples were crushed at 100% to <3.6 mm in order to
maintain proper proportion between the larger particles and
10.00
the diameter of the ball mill used in the tests.
Table 2 shows the batch mill test variables.
The tests were carried out considering the same media size
distribution as that used in the VertimillTM pilot test. The sam-
i (t/kWh)
Specific selection

1.00
ples were ground for four different time intervals in dry basis
and one test in wet basis was carried out considering an inter-
function SE*

val time 30% less than the last grinding interval in dry basis.
The wet test was carried out to consider the selection function
0.10
of VertimillTM pilot test that was carried out in wet basis.

3. Results and discussion

0.01
10 100 1000 10000
3.1. Breakage parameters Size (µm)

BatchMillTM , an optimization software developed by Mineral Fig. 4 – Specific selection function.

Technologies International [11], was used to determine the
grinding parameters. Table 3 shows the selection and break-
age function parameters determined from the laboratory tests
using a laboratory batch ball mill. The parameter SE1 was mul-
1.00
tiplied by a factor of 1.35 to correct for the higher efficiency of
the VertimillTM with respect to the standard ball mill [1].
Fig. 3 shows the particle size distributions obtained in
the batch mill tests and the model fitting obtained by the
Breakage function Bij

BatchMillTM software.
Figs. 4 and 5 show the specific selection and breakage func-
0.10
tions determined from batch ball mill test results.

Table 3 – Selection and breakage functions.

0.01
Selection function Breakage function 10 100 1000 10000
size (µm)
S1E* (t/kWh) 1 2 ϕ  ˇ
Fig. 5 – Breakage function.
0.941 0.665 −0.011 0.463 0.615 3.328
 j m a t e r r e s t e c h n o l . 2 0 1 4;3(1):86–89 89

100
Conflicts of interest

80
The authors declare no conflicts of interest.
% Passing

60
Acknowledgements

40 The authors thank Vale S.A. for providing the support for this
research.
20
references
0
0.01 0.10 1.00 10.00
Size (mm) [1] Mazzinghy DB [Doctoral thesis] Methodology for simulation
and scale-up of Tower Mills. Federal University of Minas
Fig. 6 – Simulation results obtained using ModsimTM .
Gerais; 2012.
[2] Mazzinghy DB, Alves VK, Schneider CL, Faria P, Alvarenga T,
Galéry R. Predicting the size distribution in the product and
3.2. Simulations the power requirements of a pilot scale Vertimill. In:
Proceedings of the 9th International Mineral Processing
Data from the mass balance were used to perform simulations Conference – Procemin. 2012. p. 412–20.
[3] Mazzinghy DB, Schneider CL, Alves VK, Galéry R. Applying
using the VertimillTM model implemented in the ModsimTM
the population balance model to predict the vertimill
plant-wide simulator. The feed and product size distributions
product particle size distribution. In: Proceedings of the 13th
of the mill were considered to check the accuracy of the model. European Symposium on Comminution & Classification.
Fig. 6 shows the feed and product size distributions measured 2013.
in the pilot tests as well as the predicted product from the [4] Morrison RD, Cleary PW, Sinnott MD. Using DEM to compare
scale-up model. the energy efficiency of pilot scale ball and tower mills.
The square symbols represent the measured particle size Miner Eng 2009;22:665–72.
[5] Hulburt HM, Katz S. Some problems in particle technology: a
distribution obtained in the VertimillTM pilot test product
statistical mechanical formulation. Chem Eng Sci
stream and the solid line is the corresponding model response
1964;19:555–74.
using the parameters determined and power scale-up. [6] Austin LG, Klimpel RR, Luckie PT. Process engineering of size
reduction. SME AIME 1984.
[7] Herbst JA, Fuerstenau DW. Mathematical simulation of dry
4. Conclusions ball milling using specific power information. Trans AIME
1973;254:343.
The simulations indicated that it is possible to predict the par- [8] Rajamani K, Herbst JA. Simultaneous estimation of selection
ticle size distribution and the net power of a VertimillTM using and breakage functions from batch and continuous grinding
population balance model. The energy-based model [7] used to data. Trans Inst Mining Metall 1984;93:74–85.
[9] MTI – Mineral Technologies International, Inc. ModsimTM –
simulate a ball mill, was modified to include a scaling factor of
MODular SIMulator for mineral processing plants, user
1.35 applied on one of the parameters that describe the specific
manual. 12a ed; 2009.
selection function to represent the most efficient of VertimillTM [10] King RP. Modeling and simulation of mineral processing
[1]. These results indicate that the breakages by attrition, systems. Oxford: Butterworth-Heinemann; 2002. p. 147.
abrasion or chipping are not favored in the VertimillTM . The [11] MTI – Mineral Technologies International, Inc. BatchMillTM
differences of VertimillTM and Ball Mill are in the frequency of standard batch test analyzer, user manual. 1st ed; 2003.
high and low energy impacts.