XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 3907
competence variability based on the novel concept of ‘tn-
family’ per particle rather than per set of particles. A proof
of concept of this strategy was reported for a single size
fraction of four different rock types based on breakage
percentile curves, and then showed how this variability
had implications for the simulation of SAG mill perfor-
mance (Faramarzi et al., 2018). In a comminution case
study, the ExDWT was applied to samples from three key
ore domains of a copper mine located in South America,
and then competence variability was modelled based on
a component-based description technique (Faramarzi et
al., 2019). In their studies the JKMRC SAG mill model
(Morrell &Morrison, 1989) was used to estimate ore-
induced throughput variation through simulations, assum-
ing that single percentile fractions constitute the entire
mill feed. That is, in practice the expected variations for
mixed distributions should be much smaller, hence a multi-
component simulation should be used to estimate the likely
impact of variations in the competence distribution.
In this paper, the ExDWT results from character-
izing four samples supplied by the Barrick Cortez gold
mine, USA, were used to describe the extent of breakage
variability within each ore sample through comminution
percentile curves based on the JKMRC breakage model.
Furthermore, a novel methodology is proposed to estimate
proportion of different components from the ExDWT
data. Simulation is an established approach for design,
optimisation and evaluation that assists with exploring fea-
sibility and impact of ‘change(s)’ on process performance
through analysis of “what if” scenarios. There are numerous
examples in the literature, some of which are quantifying
the effect of mine-to-mill (Burger et al., 2006 Diaz et al.,
2015 Grundstrom et al., 2001 Hart et al., 2001 Jankovic
et al., 2004 Kanchibotla et al., 2015 Kanchibotla et al.,
1999), ore competence variability (Faramarzi et al., 2018
Faramarzi et al., 2019) or using Grade Engineering ® tech-
niques to modify mill feed particle size distribution with
upgraded undersize fractions (Carrasco et al., 2017). In this
paper, several scenarios were simulated to estimate how dif-
ferent components of competence within the ore may affect
the Barrick Cortez SAG mill performance relying on the
ExDWT-based multi-component breakage modelling and
description approach.
The methodology underpinned in this paper assists in
evaluating risks arising from ore competence variability and
identifying likely circuit constraints when the mill is fed
from different ore domains. It should also support mining
companies in establishing a more realistic risk profile by
integrating orebody knowledge into the strategic and tacti-
cal plans of the mine. This integration would enhance their
confidence in project valuations, thereby facilitating more
informed decision-making regarding the expansion or con-
traction of projects.
THE EXTENDED DROP WEIGHT
TESTING (EXDWT)
Breakage Characterisation
The concept of ‘tn-family per particle’ adopted by the
ExDWT involves measuring three particle properties of
particle mass, progeny (particle) size distribution (tn) and
residual height of the drop weight after breakage for each
particle within the sample compared to the average mea-
sured in the standard JKDWT. Accounting for these vari-
ables assist with capturing the breakage variability between
the particles within a sample of ore. On the contrary, the
standard JKDWT uses the conventional concept of ‘tn-
family per set of particles’, where it averages the above vari-
ables, hence it does not allow to separate the inter-particles
competence heterogeneity. Details on the ExDWT are
described elsewhere (Faramarzi et al., 2020).
Multi-Component Breakage Modelling and
Description
Stage 1—Generating Comminution Percentile Curves
The JKMRC breakage model is used to describe the break-
age behaviour of each sample over a range of input energies
(Napier-Munn et al., 1996):
e t A61
10
)
cs =--b(E @(1)
where, t10 is the percentage of progenies that are smaller
than 1/10th of original size, Ecs is the specific comminution
energy in kWh/t, A is limiting value of t10 and b parameter
defines the steepness of t10-Ecs curve. The model param-
eters, A and b values estimated by non-linear least squares
regression.
To quantify breakage variability within a sample, a data
splitting technique has been adopted in this work that sepa-
rates the scattered data from individual particles according
to the best model fitting definition. Figure 2 represents a
schematic of the fitting technique adopted for the 25%,
50% and 75% fitted curves. Firstly, Equation 1 is fitted to
all the data points which estimates the model parameters
for the 50% curve. These parameters are then used to divide
the data points into an upper subset (data shown as triangle
points) and a lower subset (data shown as circle points)
which include the data points above and below the fitted
line, respectively. Minitab software (Minitab, 2014) was
used to divide the data points. In the next step, as shown,
the breakage model is fitted to these subsets separately to
competence variability based on the novel concept of ‘tn-
family’ per particle rather than per set of particles. A proof
of concept of this strategy was reported for a single size
fraction of four different rock types based on breakage
percentile curves, and then showed how this variability
had implications for the simulation of SAG mill perfor-
mance (Faramarzi et al., 2018). In a comminution case
study, the ExDWT was applied to samples from three key
ore domains of a copper mine located in South America,
and then competence variability was modelled based on
a component-based description technique (Faramarzi et
al., 2019). In their studies the JKMRC SAG mill model
(Morrell &Morrison, 1989) was used to estimate ore-
induced throughput variation through simulations, assum-
ing that single percentile fractions constitute the entire
mill feed. That is, in practice the expected variations for
mixed distributions should be much smaller, hence a multi-
component simulation should be used to estimate the likely
impact of variations in the competence distribution.
In this paper, the ExDWT results from character-
izing four samples supplied by the Barrick Cortez gold
mine, USA, were used to describe the extent of breakage
variability within each ore sample through comminution
percentile curves based on the JKMRC breakage model.
Furthermore, a novel methodology is proposed to estimate
proportion of different components from the ExDWT
data. Simulation is an established approach for design,
optimisation and evaluation that assists with exploring fea-
sibility and impact of ‘change(s)’ on process performance
through analysis of “what if” scenarios. There are numerous
examples in the literature, some of which are quantifying
the effect of mine-to-mill (Burger et al., 2006 Diaz et al.,
2015 Grundstrom et al., 2001 Hart et al., 2001 Jankovic
et al., 2004 Kanchibotla et al., 2015 Kanchibotla et al.,
1999), ore competence variability (Faramarzi et al., 2018
Faramarzi et al., 2019) or using Grade Engineering ® tech-
niques to modify mill feed particle size distribution with
upgraded undersize fractions (Carrasco et al., 2017). In this
paper, several scenarios were simulated to estimate how dif-
ferent components of competence within the ore may affect
the Barrick Cortez SAG mill performance relying on the
ExDWT-based multi-component breakage modelling and
description approach.
The methodology underpinned in this paper assists in
evaluating risks arising from ore competence variability and
identifying likely circuit constraints when the mill is fed
from different ore domains. It should also support mining
companies in establishing a more realistic risk profile by
integrating orebody knowledge into the strategic and tacti-
cal plans of the mine. This integration would enhance their
confidence in project valuations, thereby facilitating more
informed decision-making regarding the expansion or con-
traction of projects.
THE EXTENDED DROP WEIGHT
TESTING (EXDWT)
Breakage Characterisation
The concept of ‘tn-family per particle’ adopted by the
ExDWT involves measuring three particle properties of
particle mass, progeny (particle) size distribution (tn) and
residual height of the drop weight after breakage for each
particle within the sample compared to the average mea-
sured in the standard JKDWT. Accounting for these vari-
ables assist with capturing the breakage variability between
the particles within a sample of ore. On the contrary, the
standard JKDWT uses the conventional concept of ‘tn-
family per set of particles’, where it averages the above vari-
ables, hence it does not allow to separate the inter-particles
competence heterogeneity. Details on the ExDWT are
described elsewhere (Faramarzi et al., 2020).
Multi-Component Breakage Modelling and
Description
Stage 1—Generating Comminution Percentile Curves
The JKMRC breakage model is used to describe the break-
age behaviour of each sample over a range of input energies
(Napier-Munn et al., 1996):
e t A61
10
)
cs =--b(E @(1)
where, t10 is the percentage of progenies that are smaller
than 1/10th of original size, Ecs is the specific comminution
energy in kWh/t, A is limiting value of t10 and b parameter
defines the steepness of t10-Ecs curve. The model param-
eters, A and b values estimated by non-linear least squares
regression.
To quantify breakage variability within a sample, a data
splitting technique has been adopted in this work that sepa-
rates the scattered data from individual particles according
to the best model fitting definition. Figure 2 represents a
schematic of the fitting technique adopted for the 25%,
50% and 75% fitted curves. Firstly, Equation 1 is fitted to
all the data points which estimates the model parameters
for the 50% curve. These parameters are then used to divide
the data points into an upper subset (data shown as triangle
points) and a lower subset (data shown as circle points)
which include the data points above and below the fitted
line, respectively. Minitab software (Minitab, 2014) was
used to divide the data points. In the next step, as shown,
the breakage model is fitted to these subsets separately to