1078 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
Figure 10. Relative value-add to operations (Chandramohan, 2021)
Table 2. Example of implementing an AI-driven controller for a SAG mill operation
Application of
the Controller Input data
Controller
Objective Pros &Cons of implementation
Potential
Value-Add
Basic • Mill load
• Mill speed
• Mill noise
• Mill density
Maximize Power
Draw
Pros
• A simple system works well for consistent feed
• A cost-effective solution
Cons
• Requires trained operators to define setpoint
and operating range
• Difficult to control for highly variable feed
(competent and size)
+5%
Advanced Basic controller data +
• Mill feed PSD
• Feed density
• Liner design
• Liner Wear Profiles
Stabilize load/
operation for ore
variability
Pros
• Works well for highly variable feed
• Requires some moderately trained operators
Cons
• Cannot integrate cost drivers
• Cannot drive dynamic constraints automation
+6–15%
AI-driven Advanced controller data +
• Consumable Wear Rate
• Operating Costs
• Ore properties
• Asset management data
• All constraints data
• ESG KPIs
Minimize
operating cost or
Increase efficiency
(reduce $/t metal)
Pros
• Works well for highly variable feed
• Requires no or limited operator involvement
• Can integrate cost analysis
• Can integrate ESG KPIs
• Include dynamic constraints automation
Cons
• Requires experienced operators to input data
• Needs lots of sensors/measurement data
+2–5%
Figure 10. Relative value-add to operations (Chandramohan, 2021)
Table 2. Example of implementing an AI-driven controller for a SAG mill operation
Application of
the Controller Input data
Controller
Objective Pros &Cons of implementation
Potential
Value-Add
Basic • Mill load
• Mill speed
• Mill noise
• Mill density
Maximize Power
Draw
Pros
• A simple system works well for consistent feed
• A cost-effective solution
Cons
• Requires trained operators to define setpoint
and operating range
• Difficult to control for highly variable feed
(competent and size)
+5%
Advanced Basic controller data +
• Mill feed PSD
• Feed density
• Liner design
• Liner Wear Profiles
Stabilize load/
operation for ore
variability
Pros
• Works well for highly variable feed
• Requires some moderately trained operators
Cons
• Cannot integrate cost drivers
• Cannot drive dynamic constraints automation
+6–15%
AI-driven Advanced controller data +
• Consumable Wear Rate
• Operating Costs
• Ore properties
• Asset management data
• All constraints data
• ESG KPIs
Minimize
operating cost or
Increase efficiency
(reduce $/t metal)
Pros
• Works well for highly variable feed
• Requires no or limited operator involvement
• Can integrate cost analysis
• Can integrate ESG KPIs
• Include dynamic constraints automation
Cons
• Requires experienced operators to input data
• Needs lots of sensors/measurement data
+2–5%