1118 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
planning for smother production with rapid response to all
types of perturbations in the composition of an ore and
the process and equipment conditions. Importantly, dras-
tic reduction of energy and waste with prevention of asset
breakdown, safety and environmental issues are achieved.
Dark data are mostly unstructured data collected by real-
time historic records in process plants. Among many new
improvements are such as increase the productivity, safety
and well-being of workers, the use of phenomenological
models in mineral processes control systems, all of which
was two complex and slow to be used in the past. The
amount of dark data collected by the current mining and
mineral processing sites is huge, requiring the need of new
tools to integrate the many silos, to validate, classify and
transform data into meaningful insights to become avail-
able to advisory systems and people.
Critical problems make integrating the silos (Mine,
Mill, Maintenance, Engineering, Environmental and
Safety) difficult, since each silo has its own version of the
fact. The use spreadsheet to get the operation data (and
dark data) without subject matter expert validation, cleans-
ing, noise filtering, classification and aggregation of the
real-time operation data is inconvenient. Due to the com-
plexity of the task people stay uninformed, working with
basic data.
DIGITAL TWIN FOR A MINERAL
PROCESSING PLANT
Today digital technologies enable us to model an object
using advanced analytics to create digital twin. A digital
twin is a digital representation of a physical object or sys-
tem, a virtual replica of physical devices that can be used to
run simulations before actual devices are built and deployed
(Shaw and Frülinger 2019). Figure 2 presents the key ingre-
dients in the implementation of an Industrial Digital Twin
for a Mineral Processing Plant. Large industrial plants have
installed process historians which collect the operating data
from their process control operations, online sensors and
laboratory data. The data infrastructure is foundation and
its shown as the Realtime series operational data in the
graphic. For the operating raw data to be productive it has
to be transformed into information by providing the right
context to be made available at the right degree of details
for process systems and process engineers and management.
The process of transforming process data into process infor-
mation is described by Bascur, 2020 Digital Transformation
in the Process Industries.
There is a huge amount of untapped company asset
in these historians which is the history of the production.
When this process history is processed by data models it
is transformed into valuable information using predic-
tive analytics and machine learning algorithms (Kelleher,
Namee, and D’Arcy, 2015, Rashka, 2015, Brunton and
Kutz, 2019, Soroush, Baldea and Edgar, 2020). The key
is to apply a business goal associated to the operational
characteristics of the plant. This is a real time performance
evaluation of the plant when close to process constraints
(Goldratt, 2014). The key is to tie the Production Planning
Targets with the final production results. This part of the
process is seldom done due to the lack of adequate informa-
tion at the adequate degree of details by the process engi-
neers. This is why we used the Net Metal Production Rate
(NMPR) key indicator has the main guidance on assessing
the performance of the plant when compared to the eco-
nomic production plan.
In many mining operations, ore is crushed and wet
milled to liberate the valuable mineral as shown in Figure 3.
This slurry is concentrated for metal by flotation and then
filtered to form a dry mineral concentrate that is shipped to
metal smelters to produce the final metallic products. The
type of filtration equipment required depends on the par-
ticle size, mineralogy, and shipping requirements. As with
all mining operations, the required equipment needs to be
extremely robust and designed to be reliable even under
the toughest operating conditions. The weather conditions
in the remote high elevation in the Andean regions are
Figure 2. Digital Twin strategy for using operational data
planning for smother production with rapid response to all
types of perturbations in the composition of an ore and
the process and equipment conditions. Importantly, dras-
tic reduction of energy and waste with prevention of asset
breakdown, safety and environmental issues are achieved.
Dark data are mostly unstructured data collected by real-
time historic records in process plants. Among many new
improvements are such as increase the productivity, safety
and well-being of workers, the use of phenomenological
models in mineral processes control systems, all of which
was two complex and slow to be used in the past. The
amount of dark data collected by the current mining and
mineral processing sites is huge, requiring the need of new
tools to integrate the many silos, to validate, classify and
transform data into meaningful insights to become avail-
able to advisory systems and people.
Critical problems make integrating the silos (Mine,
Mill, Maintenance, Engineering, Environmental and
Safety) difficult, since each silo has its own version of the
fact. The use spreadsheet to get the operation data (and
dark data) without subject matter expert validation, cleans-
ing, noise filtering, classification and aggregation of the
real-time operation data is inconvenient. Due to the com-
plexity of the task people stay uninformed, working with
basic data.
DIGITAL TWIN FOR A MINERAL
PROCESSING PLANT
Today digital technologies enable us to model an object
using advanced analytics to create digital twin. A digital
twin is a digital representation of a physical object or sys-
tem, a virtual replica of physical devices that can be used to
run simulations before actual devices are built and deployed
(Shaw and Frülinger 2019). Figure 2 presents the key ingre-
dients in the implementation of an Industrial Digital Twin
for a Mineral Processing Plant. Large industrial plants have
installed process historians which collect the operating data
from their process control operations, online sensors and
laboratory data. The data infrastructure is foundation and
its shown as the Realtime series operational data in the
graphic. For the operating raw data to be productive it has
to be transformed into information by providing the right
context to be made available at the right degree of details
for process systems and process engineers and management.
The process of transforming process data into process infor-
mation is described by Bascur, 2020 Digital Transformation
in the Process Industries.
There is a huge amount of untapped company asset
in these historians which is the history of the production.
When this process history is processed by data models it
is transformed into valuable information using predic-
tive analytics and machine learning algorithms (Kelleher,
Namee, and D’Arcy, 2015, Rashka, 2015, Brunton and
Kutz, 2019, Soroush, Baldea and Edgar, 2020). The key
is to apply a business goal associated to the operational
characteristics of the plant. This is a real time performance
evaluation of the plant when close to process constraints
(Goldratt, 2014). The key is to tie the Production Planning
Targets with the final production results. This part of the
process is seldom done due to the lack of adequate informa-
tion at the adequate degree of details by the process engi-
neers. This is why we used the Net Metal Production Rate
(NMPR) key indicator has the main guidance on assessing
the performance of the plant when compared to the eco-
nomic production plan.
In many mining operations, ore is crushed and wet
milled to liberate the valuable mineral as shown in Figure 3.
This slurry is concentrated for metal by flotation and then
filtered to form a dry mineral concentrate that is shipped to
metal smelters to produce the final metallic products. The
type of filtration equipment required depends on the par-
ticle size, mineralogy, and shipping requirements. As with
all mining operations, the required equipment needs to be
extremely robust and designed to be reliable even under
the toughest operating conditions. The weather conditions
in the remote high elevation in the Andean regions are
Figure 2. Digital Twin strategy for using operational data