Artificial intelligence in the post-deep learning era
2. Jonathan Nortier and Ali Sirkeci, AMC - Evaluate Your Options with Minemax
1. Evaluate your options with Minemax &
Studio OP
Jonathon Nortier – jnortier@amcconsultants.com
& Ali Sirkeci – asirkeci@amcconsultants.com
AMC Consultants
Datamine Symposium, Perth, March 2021
4. AMC and Datamine / Minemax
4
• 30 year history with Datamine
• Datamine and Minemax’s fundamental structure and
flexibility suits project work where we are constantly faced
with new and unusual challenges.
• Auditable process
• Allows us investigate ideas to add value
5. 5
Why Minemax
• Flexibility
• Ability to ensure
a practical
mining
schedule
• Produce
numerous
scenarios
quickly to allow
study team to
evaluate
options.
Minemax approach
6. When to use Minemax
6
Scoping and Pre-feasibility studies,
Minemax is especially as useful at looking
at options
- Plant size
- SMU and equipment specifications
- Fleet size
- Expansion capital and timing
- Mill blend configurations
- Multi-pit blending and product
specification limitations
- Waste dump optimisation
7. 7
Case Studies
• Capacity Study
• TSF constrained project & Autopit
• Pit sequencing and plant feed determination
• Waste dump optimisation
8. 8
Capacity Study
Capacity study completed at the start of PFS to define what options to take
forward.
1. Setting: Large gold mine
2. Goal: Determine the optimal plant size
3. Other objectives:
• Guidance for future studies within PFS – SMU, mining equipment class,
bench height, ultimate pit shell selection
9. Pit optimisation inputs
9
SMU Options
X 3
Resource
Model
Mining
Equipment
X 3
Plant Capacity
Options
X 6
Pit Optimisations
54 combinations, 16 selected
Evaluate results
Generate MM
Inventories
Stage Selection
Diluted
Models
Unit Costs
Unit Costs
16. Scenario Options
16
Throughout the capacity study 64 options were evaluated, testing:
• Impact of ultimate shell selection
• Impact of Inferred mineral resources on plant capacity selection
• Class of mining fleet
• Bench height and SMU size
17. Output
17
• Results allowed the client to
clearly see the impact of
increasing the plant capacity
has on project NPV.
• Knowing the outputs had
practical schedules to back
them up improved confidence
in data presented.
NPV / Initial
Capital
NPV
Increasing plant size
18. Output
18
• Results allowed the
client to clearly see the
impact of increasing
the plant capacity has
on project NPV.
• Knowing the outputs
had practical
schedules to back
them up improved
confidence in data
presented.
NPV / Initial
Capital
NPV
Increasing plant size
19. Output Schedules
19
Increasing plant throughput
Total movement
Mining by stage
Processing tonnes
Feed grade
Stockpile inventory
Cash flow
Gold production
20. Outcomes of study
20
• Provided guidance on plant capacity to allow PFS to progress
• Demonstrated that selected plant capacity was robust in terms of
final pit selection, conversion (or not) Inferred material and mining
equipment selection.
• Allowed study team to set the basis of design.
Minemax allowed us to the complete the study at level of detail and
confidence that the client expected for the magnitude of the
investment decision.
21. 21
TSF Constrained
1. Setting: Existing operation with limited TSF expansion capacity.
2. Goal: Determine the optimal pit design for TSF capacity
3. Challenges:
• Select a pit too large and high grade at depth is deferred and processing plant
underutilized or low grade feed displaces future high grade
• Select a pit too small and less high grade feed is available later in the schedule.
• Fit in with existing pit development
• Limited mining width to split final stage to improve ore presentation to the plant
22. TSF Constrained - Approach
22
1. Completed pit optimization
2. Assess 4 different shells as ultimate pit limits
3. Use Auto-pit to generate pit designs that tie in with existing stages
4. Test options in Minemax
23. Studio OP Auto design pit demonstration
23
What is Auto design
Why AMC uses Auto design?
• Helps to visualize and evaluate the access concepts & options
• Ability to ensure mine-ability of interim pit versus pit optimisation shells
• Allows quick generation of multiple options to be evaluated
26. 26
TSF Constrained - Results
1 2 3 4
• Pit 1 – Contained just
enough ore to fill TSF
• Pit 2 – Some excess ore
• Pit 3 – Contained enough
High and Medium grade to
fill TSF
• Pit 4 – Large pit test upper
limit
27. 27
Pit sequencing based on plant blending
constraints
1. Setting: Laterally extensive (~10 km long) potash deposit, >200 year life
2. Goal: Determine initial 20 year pit sequence, that satisfies multiple
processing and mining constraints.
3. Challenges:
• Deposit geology variable
• Multiple ore types requiring blending
• Pit optimisation does not define suitable initial pits
28. Approach – Round 1
28
1. Prepare mining model
2. Pit optimization, defined long life pit limit (>200 year mine life)
3. Use Minemax to define areas of initial interest (e.g. 20 year pit):
a) Run Minemax to identify mining location “hotspots”, using initial process plant
constraints and inputs.
b) Feedback loop with process design team (multiple iterations)
• complete sensitivity analysis of mining location to changes in process basis of
design
• test various tolerance ranges for process ratios and assess subsequent affect on
mining practicalities (e.g. mining face advancement erratic or manageable?).
• trade-off between mining and processing to establish and refine blend ratios.
29. Mining locations
29
LG Shell RF 0.5
LG Shell RF 1
Initial mining areas
identified for
Minemax
(first 20 years)
1. LG pit shells – final pit limits
reasonably insensitive to
inputs (i.e. large step
change at low RF, small
change between RF 0.5
to 1)
2. Minemax - starting areas
sensitive to changes in
process blend constraints
30. Approach – Round 2
30
1. Develop mining panel “stage” designs based on Round 1 outputs
2. Complete final schedule based on:
• Panel designs,
• Vertical rate of advance and inter-panel dependencies
• Final agreed processing constraints
• Plant feed rate and stockpile limits,
• Fleet equipment hours (two different fleets)
3. Prepare output for financial model.
32. Outcomes
32
1. Multiple mining sequences developed to articulate potential
blending scenarios to the process design team to allow the
process basis of design to be set.
2. Defined optimal pit and development sequence to meet selected
process plant constraints.
3. Minemax outputs contributed to the financial model which
ultimately allowed the client to obtain significant initial project
funding.
Minemax allowed us to successfully identify and quantify the problem
and was the tool that allowed us to develop a workable solution
33. 33
Waste Dump Optimisation
1. Setting
• Gold mine with multiple pits
2. Goal
• Determine best waste dumping strategy with and without inpit dumps
3. Challenges
• Multiple access to waste dumps and inpit dumps
34. Waste Dump Optimisation
Waste dump designs
• Need to be oversized
• Design to be split into
dump cells
• Haulage network
designed to link all pit
and dump combinations
Pits Waste
dump
Waste
dump
35. Waste Dump Optimisation
Haulage Model Haulage model
• Needs to be split into pit
and dump components
that are linked by common
nodes
• Truck related variables
linked to pit and dump
models (e.g. Travel time,
fuel burn, haul distance)
Waste dump decision tree
36. Waste Dump Optimisation
Benefits
• Waste dump strategy
(short vs long dumping
and truck number
smoothing)
• Limit schedule to fleet
capacity
• Acid generating waste
dumping sequencing
• Helps cost modelling,
smoothing truck
numbers, less iterations
between estimator and
scheduler.
Truck hour
dashboard
Financial
Outputs
37. Waste Dump Optimisation
37
Outcomes
• Defined optimal final waste dump
landform and sequence for each
schedule option
• ~15 M$ NPV improvement with inpit
dumps
• Truck hours calculated for each
scenario and optimized based
capital expenditure