XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1509
the energy consumed during grinding. Different authors
conducted experiments by installing a torque meter on the
Hardgrove mill (Austin et al., 1981 Mucsi, 2008 Mucsi et
al., 2019 Palaniandy, 2017 Shi, 2014b, 2014a Shi &Zuo,
2014), thus showing that the energy consumed for a given
test varies with different materials. However, such studies
focused on using a Hardgrove mill to determine the break-
age function or the energy required for fine grinding by test-
ing different particle size ranges. Mucsi, 2008 and Mucsi et
al., 2019 were the only authors who used the Hardgrove
mill equipped with a torque measurement device to pre-
dict the BBMWI, who adopted two approaches. In the first
study (Mucsi, 2008), he used a standard 50 g mass from
each sample to perform a grinding cycle and calculated the
BBMWI based on the applied energy and the F80 and P80
test results, with a prediction error of 5.75%. In the second
approach (Mucsi et al., 2019), the same author used a fixed
volume of 58 cm3 and performed several grinding cycles
simulating the closed-circuit Bond test. The results showed
a more significant deviation from the standard Bond test,
with a prediction error of 36%. Despite the excellent
results obtained by different authors, different parameters
were used in each study, including torque measurement,
fixed mass or volume, different feed particle sizes, either
–1.18+0.600 mm as used at the standard Hardgrove pro-
cedure or –3.35 mm as used at the Bond procedure, and
determination of the P80 of the product or only the per-
centage passing at 75 or 106 µm screens.
A different approach to assessing a grinding test, con-
sidering not only P80 or the amount of product passing
through a given screen aperture, is to evaluate the size-spe-
cific energy (SSE), which is the ratio of the total energy
input to the generation of fines of a certain size fraction
(Ballantyne et al., 2015 Hilden &Suthers, 2010 Levin,
1992 Musa &Morrison, 2009 Pamparana et al., 2022).
The use of such an indicator to quantify the product from
the test in a Hardgrove mill, along with the torque mea-
surement, the use of a fixed volume of ore, and a feed size
closer to the original Bond test, may be an option yet to be
investigated to reduce the error in predicting the BBMWI
based on the Hardgrove test, which is the focus of the pres-
ent work. An initial study developed by Bergerman et al.
(2023) demonstrated that it was possible to predict the
BBMWI using the modified Hardgrove with less than 4%
deviation from the measured value, using 33 samples.
The present study validated this new test using 135
samples. It also considered different equations to model the
results using multilinear regressions.
MATERIALS AND METHODS
The study analyzed 135 samples from copper, gold, iron
ore, bauxite, and limestone deposits in Brazil, Canada, and
Indonesia.
All samples were staged crushed below 3.35 mm using
jaw and roller crushers while seeking to minimize the gen-
eration of fines. The product passing the 3.35 mm screen
was dried in an oven at 100oC and then split into two ali-
quot parts. The Bond Ball Mill test was performed on one
of the aliquots as per the standard defined by Bond (Bond,
1961) and described by Mcivor, 2015. A lab ball mill with
a 30.5 cm inside diameter and 30.5 cm inside length with
rounded corners was used. The grinding media consisted of
285 iron or steel balls (43 Ø36.8 mm, 67 Ø29.7 mm, 10
Ø25.4 mm, 71 Ø19.1 mm, and 94 Ø15.5 mm) weighing
20.125 g in total. The mill was operated at 70 rpm. A 700
cm3 sample was fed into the mill and ground for a deter-
mined number of revolutions and dry sieved at the closing
screen aperture. The retained fraction was combined with
fresh feed. At least five cycles were carried out until the net
grindability reached equilibrium, with a maximum differ-
ence of 3% in the last three cycles. The product of the last
cycle was dry-screened. The Bond Work Index was then
calculated using Equation 1.
**
.*.5 WI
P G
P F
10 10
1 1023 44
.BM
pr 100
0 23 0.82
80 80
=
-c m
(1)
where WIBM is the Bond Ball Mill Work Index (kWh/t),
P100 is the closing screen aperture (µm), Gpr is the net grams
of product per revolution, P80 is the product 80% passing
size (µm), and F80 is the feed 80% passing size (µm).
For the Hardgrove tests, a 30 cm3 sample of
–3.35+0.600 mm feed was placed in the mill grinding bowl
with eight 25.4 mm steel balls. The test was conducted
according to the American Standard Procedure (ASTM
D409, 2002) with weights added to the driving spindle so
that the total vertical force on the balls as a result of the
weights, shaft, top grinding ring, and gear is equal to 29 kg
A total grinding time of 3 min and a mill speed of 20 rpm
was used. The Hardgrove modified mill was fitted with a
shaft-type torque transducer model 2300 from NCTE
AG ®. The no-load torque was measured before and after
each test without rock material and then subtracted from
the measured torque during the test to calculate the average
net torque. In a previous stage of the test development, the
verification of the Hardgrove mill with standard samples
and repeatability tests was done (Guimarães Bergerman et
al., 2023).
the energy consumed during grinding. Different authors
conducted experiments by installing a torque meter on the
Hardgrove mill (Austin et al., 1981 Mucsi, 2008 Mucsi et
al., 2019 Palaniandy, 2017 Shi, 2014b, 2014a Shi &Zuo,
2014), thus showing that the energy consumed for a given
test varies with different materials. However, such studies
focused on using a Hardgrove mill to determine the break-
age function or the energy required for fine grinding by test-
ing different particle size ranges. Mucsi, 2008 and Mucsi et
al., 2019 were the only authors who used the Hardgrove
mill equipped with a torque measurement device to pre-
dict the BBMWI, who adopted two approaches. In the first
study (Mucsi, 2008), he used a standard 50 g mass from
each sample to perform a grinding cycle and calculated the
BBMWI based on the applied energy and the F80 and P80
test results, with a prediction error of 5.75%. In the second
approach (Mucsi et al., 2019), the same author used a fixed
volume of 58 cm3 and performed several grinding cycles
simulating the closed-circuit Bond test. The results showed
a more significant deviation from the standard Bond test,
with a prediction error of 36%. Despite the excellent
results obtained by different authors, different parameters
were used in each study, including torque measurement,
fixed mass or volume, different feed particle sizes, either
–1.18+0.600 mm as used at the standard Hardgrove pro-
cedure or –3.35 mm as used at the Bond procedure, and
determination of the P80 of the product or only the per-
centage passing at 75 or 106 µm screens.
A different approach to assessing a grinding test, con-
sidering not only P80 or the amount of product passing
through a given screen aperture, is to evaluate the size-spe-
cific energy (SSE), which is the ratio of the total energy
input to the generation of fines of a certain size fraction
(Ballantyne et al., 2015 Hilden &Suthers, 2010 Levin,
1992 Musa &Morrison, 2009 Pamparana et al., 2022).
The use of such an indicator to quantify the product from
the test in a Hardgrove mill, along with the torque mea-
surement, the use of a fixed volume of ore, and a feed size
closer to the original Bond test, may be an option yet to be
investigated to reduce the error in predicting the BBMWI
based on the Hardgrove test, which is the focus of the pres-
ent work. An initial study developed by Bergerman et al.
(2023) demonstrated that it was possible to predict the
BBMWI using the modified Hardgrove with less than 4%
deviation from the measured value, using 33 samples.
The present study validated this new test using 135
samples. It also considered different equations to model the
results using multilinear regressions.
MATERIALS AND METHODS
The study analyzed 135 samples from copper, gold, iron
ore, bauxite, and limestone deposits in Brazil, Canada, and
Indonesia.
All samples were staged crushed below 3.35 mm using
jaw and roller crushers while seeking to minimize the gen-
eration of fines. The product passing the 3.35 mm screen
was dried in an oven at 100oC and then split into two ali-
quot parts. The Bond Ball Mill test was performed on one
of the aliquots as per the standard defined by Bond (Bond,
1961) and described by Mcivor, 2015. A lab ball mill with
a 30.5 cm inside diameter and 30.5 cm inside length with
rounded corners was used. The grinding media consisted of
285 iron or steel balls (43 Ø36.8 mm, 67 Ø29.7 mm, 10
Ø25.4 mm, 71 Ø19.1 mm, and 94 Ø15.5 mm) weighing
20.125 g in total. The mill was operated at 70 rpm. A 700
cm3 sample was fed into the mill and ground for a deter-
mined number of revolutions and dry sieved at the closing
screen aperture. The retained fraction was combined with
fresh feed. At least five cycles were carried out until the net
grindability reached equilibrium, with a maximum differ-
ence of 3% in the last three cycles. The product of the last
cycle was dry-screened. The Bond Work Index was then
calculated using Equation 1.
**
.*.5 WI
P G
P F
10 10
1 1023 44
.BM
pr 100
0 23 0.82
80 80
=
-c m
(1)
where WIBM is the Bond Ball Mill Work Index (kWh/t),
P100 is the closing screen aperture (µm), Gpr is the net grams
of product per revolution, P80 is the product 80% passing
size (µm), and F80 is the feed 80% passing size (µm).
For the Hardgrove tests, a 30 cm3 sample of
–3.35+0.600 mm feed was placed in the mill grinding bowl
with eight 25.4 mm steel balls. The test was conducted
according to the American Standard Procedure (ASTM
D409, 2002) with weights added to the driving spindle so
that the total vertical force on the balls as a result of the
weights, shaft, top grinding ring, and gear is equal to 29 kg
A total grinding time of 3 min and a mill speed of 20 rpm
was used. The Hardgrove modified mill was fitted with a
shaft-type torque transducer model 2300 from NCTE
AG ®. The no-load torque was measured before and after
each test without rock material and then subtracted from
the measured torque during the test to calculate the average
net torque. In a previous stage of the test development, the
verification of the Hardgrove mill with standard samples
and repeatability tests was done (Guimarães Bergerman et
al., 2023).