XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1035
with customized phenomenological and machine learning
models.
Orica technical services developed a mine-to-concen-
trate flow sheet for a copper operation in central Chile
using a combination of industry-standard phenomenologi-
cal and Machine Learning (ML) models on the IES plat-
form. Blast, comminution, and flotation sub-circuits were
developed and integrated for short-term test planning for
the mining operation and plant modifications. The paper
describes the procedure used to develop the operationally
integrated flow sheet and explains how it was utilized to
generate bankable outcomes. This flowsheet enabled the
mine management team to conduct technical evaluations
considering flowsheet configuration modifications and
environmental constraints, including water shortages.
CONFIGURATION OF THE VALUE CHAIN
FLOW SHEET
The value chain flow sheet was developed using the meth-
odology provided by Amini and Beaton (2020). A brief
description of developing the value chain flow sheet is
detailed following.
• data collection and cleaning
– data were collected and grouped into four catego-
ries 1. ore characteristics, 2. equipment specifica-
tions, 3. process operation condition, and 4. per-
formance response.
– clean the data based on non-operational time, “0”
and “negative” value
– average out the data on a monthly basis.
• Mine to mill flow sheet development, model imple-
mentation, and validation
– develop baseline mine to mill flow sheet by inte-
grating blasting, crushing, and grinding operation
– implement prior developed JKSimMet process
models (calibrated through plant survey) into the
IES platform
– assess and fine-tune the model response with the
average operational data
– identify the process constraints and set up the flow
sheet optimiser to apply them to the flow sheet
• Flotation flow sheet development using a machine
learning model
– flotation model development to determine flota-
tion process response with the change in ore char-
acteristics and mine to mill conditions
– implement the flotation flow sheet in the IES
environment
• Value chain flow sheet development
– develop the value chain flow sheet by integrating
flotation and mine to mill flow sheet using IES
embedded flow sheet feature
An integrated base case flow sheet was developed follow-
ing the above methodology as shown in Figure 1. The inte-
grated flow sheet comprises blast and primary crushing
line, a conventional SABC) circuit, and a flotation circuit
encoded with a machine learning model.
FLOW SHEET MODEL DESCRIPTION
The flow sheet within IES consists of a combination of
phenomenological and empirical models, necessitating
calibration using survey and process data gathered during
operations. The key process models for the flow sheet are
described as follows.
Kuz Ram Fines Blast Model
avThe original Kuz Ram model, proposed by Cunningham
(1983), is widely recognised as one of the most commonly
used models for estimating fragmentation resulting from
blasting. This model is based on the Kuznetsov and Rosin-
Rammler equations. However, a notable limitation of this
model is its tendency to underestimate the quantity of fines
produced. To address this issue, the ‘Crushed Zone Model’
was developed at JKMRC (Kanchibotla et al., 1999),
known as the ‘Kuz Ram Fines Blast Model’ in IES. One
advantage of this model, compared to the original Kuz-
Ram model, is its increased sensitivity of the fine particle
size distribution (PSD) to rock mass strength and explosive
performance characteristics.
The development of the operational blast model
involved a comprehensive integration of various parameters,
including rock characteristics such as Uniaxial Compressive
Strength (UCS) and Rock Quality Designation (RQD),
blast design specifications encompassing Burden, Specimen,
Bench Height, Column Charge, and others, and criti-
cal explosive details like Relative Weight Strength (RWS),
Density, and Velocity of Detonation (VOD). These param-
eters were derived from on-site data. To enhance the mod-
el’s accuracy, a meticulous tuning process was undertaken.
This involved refining the model by minimizing discrep-
ancies between the actual fragmentation data, specifically
P80, P50, and P20, and the data predicted by the model.
This iterative tuning approach aimed to align the model
more closely with real-world blast outcomes, ensuring a
robust and reliable representation of the blasting operation.
with customized phenomenological and machine learning
models.
Orica technical services developed a mine-to-concen-
trate flow sheet for a copper operation in central Chile
using a combination of industry-standard phenomenologi-
cal and Machine Learning (ML) models on the IES plat-
form. Blast, comminution, and flotation sub-circuits were
developed and integrated for short-term test planning for
the mining operation and plant modifications. The paper
describes the procedure used to develop the operationally
integrated flow sheet and explains how it was utilized to
generate bankable outcomes. This flowsheet enabled the
mine management team to conduct technical evaluations
considering flowsheet configuration modifications and
environmental constraints, including water shortages.
CONFIGURATION OF THE VALUE CHAIN
FLOW SHEET
The value chain flow sheet was developed using the meth-
odology provided by Amini and Beaton (2020). A brief
description of developing the value chain flow sheet is
detailed following.
• data collection and cleaning
– data were collected and grouped into four catego-
ries 1. ore characteristics, 2. equipment specifica-
tions, 3. process operation condition, and 4. per-
formance response.
– clean the data based on non-operational time, “0”
and “negative” value
– average out the data on a monthly basis.
• Mine to mill flow sheet development, model imple-
mentation, and validation
– develop baseline mine to mill flow sheet by inte-
grating blasting, crushing, and grinding operation
– implement prior developed JKSimMet process
models (calibrated through plant survey) into the
IES platform
– assess and fine-tune the model response with the
average operational data
– identify the process constraints and set up the flow
sheet optimiser to apply them to the flow sheet
• Flotation flow sheet development using a machine
learning model
– flotation model development to determine flota-
tion process response with the change in ore char-
acteristics and mine to mill conditions
– implement the flotation flow sheet in the IES
environment
• Value chain flow sheet development
– develop the value chain flow sheet by integrating
flotation and mine to mill flow sheet using IES
embedded flow sheet feature
An integrated base case flow sheet was developed follow-
ing the above methodology as shown in Figure 1. The inte-
grated flow sheet comprises blast and primary crushing
line, a conventional SABC) circuit, and a flotation circuit
encoded with a machine learning model.
FLOW SHEET MODEL DESCRIPTION
The flow sheet within IES consists of a combination of
phenomenological and empirical models, necessitating
calibration using survey and process data gathered during
operations. The key process models for the flow sheet are
described as follows.
Kuz Ram Fines Blast Model
avThe original Kuz Ram model, proposed by Cunningham
(1983), is widely recognised as one of the most commonly
used models for estimating fragmentation resulting from
blasting. This model is based on the Kuznetsov and Rosin-
Rammler equations. However, a notable limitation of this
model is its tendency to underestimate the quantity of fines
produced. To address this issue, the ‘Crushed Zone Model’
was developed at JKMRC (Kanchibotla et al., 1999),
known as the ‘Kuz Ram Fines Blast Model’ in IES. One
advantage of this model, compared to the original Kuz-
Ram model, is its increased sensitivity of the fine particle
size distribution (PSD) to rock mass strength and explosive
performance characteristics.
The development of the operational blast model
involved a comprehensive integration of various parameters,
including rock characteristics such as Uniaxial Compressive
Strength (UCS) and Rock Quality Designation (RQD),
blast design specifications encompassing Burden, Specimen,
Bench Height, Column Charge, and others, and criti-
cal explosive details like Relative Weight Strength (RWS),
Density, and Velocity of Detonation (VOD). These param-
eters were derived from on-site data. To enhance the mod-
el’s accuracy, a meticulous tuning process was undertaken.
This involved refining the model by minimizing discrep-
ancies between the actual fragmentation data, specifically
P80, P50, and P20, and the data predicted by the model.
This iterative tuning approach aimed to align the model
more closely with real-world blast outcomes, ensuring a
robust and reliable representation of the blasting operation.