XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1403
an intermediate ore stockpile. The focus is on modeling
the geologic attributes of the ore as it moves through the
processing system as this is a prerequisite for any larger pro-
duction models that track ore movement throughout the
broader primary crushing to flotation circuit. Of particu-
lar concern to the broader operation is the acid consump-
tion rate and clay content of the ore, as the presence of
acid-consuming minerals in the concentrate will increase
the acid consumable portion of the overall processing cost
considerably, while excess clay can and does contribute to
production downtimes. From the perspective of the crush
and convey division, however, both hard ore and high clay
content ore can cause operational problems resulting in
unplanned downtimes.
METHODS
Data Sources
Mining and production data is hosted in a variety of data-
bases. These include ore control data, dispatch data, pro-
cess control data, and maintenance data, the latter of which
includes information that tracks planned and unplanned
downtime events. The ore control, dispatch data, and main-
tenance data are available via Snowflake, while the process
control data is hosted in a system called PINET, which can
be accessed through an Excel add-in. For all data sources,
the data used in this study includes all records from the year
2022, as that was the most recently available full calendar
year. A full year was chosen in part to avoid seasonal effects,
as well as to ensure a robust sample size. Details on the data
sources are shown in Table 1.
The mine maintains a database view which joins the
ore control model with data from the dispatch database.
Therefore, data linking the original location and in situ
geologic characteristics of the mined ore to the actual dump
location is available at the granularity of the individual shovel
bucket. Buckets are aggregated into individual truckloads,
and the average copper grade, clay content, hardness, etc.,
of the ore is calculated at the level of the truckload. From
there, the dump destination and the dumping time stamp
is also attached to each truckload. Linking the datasets in
this way presumes that there is negligible blast movement,
and the geologic data available is limited to that which is
included in the ore control model. However, coupled with
the dispatch data, this means that it is possible to track the
geologic characteristics of the ore through the mining pro-
cess, and downstream process models can tag each compo-
nent of ore as it moves through the system with detailed
geologic descriptors, or at least include identifying data that
can be used to trace the discrete ore component back to its
original location. Thus, data accessed from the ore control/
dispatch view is used as an input to the DES model, which
covers the mineral processing component starting with the
dumping of each truckload into the primary crusher and
ending with the point at which the ore enters an intermedi-
ate ore stockpile.
Process performance data is available via PINET and
is accessed via an Excel add-in. It includes a plethora of
detailed attributes for each component of the process sys-
tems, not all of which are likely to be a function of the geo-
metallurgical attributes of the ore. Attributes of the crush
and convey division that may be impacted by the geometal-
lurgical quality of the ore include fields related to equip-
ment speed, run status, throughput, plugged chute alarms,
and power usage. Data accessed from PINET was used to
model the probability distributions for aspects such as con-
veyor speed and crusher crushing time.
Lastly, maintenance data is available from a third data-
base view on Snowflake. This view includes data on planned
and unplanned downtimes, including categorical fields
which label each down with a cause, as well as whether the
downtime event was planned or unplanned. Examples of
planned downtimes include regularly scheduled preven-
tative maintenance, while unplanned downs can include
mechanical breakdowns, downs caused by issues upstream
or downstream in the operation, or breakdowns related to
the quality of the ore. For geology-related breakdowns, there
are six categories, which are ‘Ore Hardness,’ ‘Boulders,’
‘Bridge,’ ‘Plugged,’ ‘Material Buildup,’ and ‘Material
Buildup/Spillage.’ Among unplanned downs recorded as an
‘Operating Problem’ in 2022, over 94% of primary crusher
Table 1. Overview of data sources used in the study
Ore Control/Dispatch Data PINET Process Control Data Maintenance Data
Snowflake View Accessed via Excel add-in Snowflake view
Joins ore control block model and
dispatch records of ore haulage,
including all mineralogical fields
in ore control block model and the
exact timing each truckload was
loaded and dumped
Process control database with
various measures of operational
performance
Maintenance data on planned and
unplanned downtimes, durations,
and causes
an intermediate ore stockpile. The focus is on modeling
the geologic attributes of the ore as it moves through the
processing system as this is a prerequisite for any larger pro-
duction models that track ore movement throughout the
broader primary crushing to flotation circuit. Of particu-
lar concern to the broader operation is the acid consump-
tion rate and clay content of the ore, as the presence of
acid-consuming minerals in the concentrate will increase
the acid consumable portion of the overall processing cost
considerably, while excess clay can and does contribute to
production downtimes. From the perspective of the crush
and convey division, however, both hard ore and high clay
content ore can cause operational problems resulting in
unplanned downtimes.
METHODS
Data Sources
Mining and production data is hosted in a variety of data-
bases. These include ore control data, dispatch data, pro-
cess control data, and maintenance data, the latter of which
includes information that tracks planned and unplanned
downtime events. The ore control, dispatch data, and main-
tenance data are available via Snowflake, while the process
control data is hosted in a system called PINET, which can
be accessed through an Excel add-in. For all data sources,
the data used in this study includes all records from the year
2022, as that was the most recently available full calendar
year. A full year was chosen in part to avoid seasonal effects,
as well as to ensure a robust sample size. Details on the data
sources are shown in Table 1.
The mine maintains a database view which joins the
ore control model with data from the dispatch database.
Therefore, data linking the original location and in situ
geologic characteristics of the mined ore to the actual dump
location is available at the granularity of the individual shovel
bucket. Buckets are aggregated into individual truckloads,
and the average copper grade, clay content, hardness, etc.,
of the ore is calculated at the level of the truckload. From
there, the dump destination and the dumping time stamp
is also attached to each truckload. Linking the datasets in
this way presumes that there is negligible blast movement,
and the geologic data available is limited to that which is
included in the ore control model. However, coupled with
the dispatch data, this means that it is possible to track the
geologic characteristics of the ore through the mining pro-
cess, and downstream process models can tag each compo-
nent of ore as it moves through the system with detailed
geologic descriptors, or at least include identifying data that
can be used to trace the discrete ore component back to its
original location. Thus, data accessed from the ore control/
dispatch view is used as an input to the DES model, which
covers the mineral processing component starting with the
dumping of each truckload into the primary crusher and
ending with the point at which the ore enters an intermedi-
ate ore stockpile.
Process performance data is available via PINET and
is accessed via an Excel add-in. It includes a plethora of
detailed attributes for each component of the process sys-
tems, not all of which are likely to be a function of the geo-
metallurgical attributes of the ore. Attributes of the crush
and convey division that may be impacted by the geometal-
lurgical quality of the ore include fields related to equip-
ment speed, run status, throughput, plugged chute alarms,
and power usage. Data accessed from PINET was used to
model the probability distributions for aspects such as con-
veyor speed and crusher crushing time.
Lastly, maintenance data is available from a third data-
base view on Snowflake. This view includes data on planned
and unplanned downtimes, including categorical fields
which label each down with a cause, as well as whether the
downtime event was planned or unplanned. Examples of
planned downtimes include regularly scheduled preven-
tative maintenance, while unplanned downs can include
mechanical breakdowns, downs caused by issues upstream
or downstream in the operation, or breakdowns related to
the quality of the ore. For geology-related breakdowns, there
are six categories, which are ‘Ore Hardness,’ ‘Boulders,’
‘Bridge,’ ‘Plugged,’ ‘Material Buildup,’ and ‘Material
Buildup/Spillage.’ Among unplanned downs recorded as an
‘Operating Problem’ in 2022, over 94% of primary crusher
Table 1. Overview of data sources used in the study
Ore Control/Dispatch Data PINET Process Control Data Maintenance Data
Snowflake View Accessed via Excel add-in Snowflake view
Joins ore control block model and
dispatch records of ore haulage,
including all mineralogical fields
in ore control block model and the
exact timing each truckload was
loaded and dumped
Process control database with
various measures of operational
performance
Maintenance data on planned and
unplanned downtimes, durations,
and causes