carl bergmann, mintek
TRANSCRIPT
Iron Ore Beneficiation Africa
Modelling of Physical Separation Processes of fine ores
Carl Bergmann 17 March 2014
Agenda
Objectives Using Particle Tracking Analysis – Method Using Particle Tracking Analysis – Results Flowsheet Simulation Conclusions
Objectives
FINE ORE CHARACTERISATION
How to characterize a fines sample in terms of size, density, shape, liberation, grade?
Size and shape – laser scattering, cyclosizer, screening, mineralogical
Density – particle pycnometry, heavy liquids, mineralogical
Liberation – grain free surface area, volume percent, mass percent
Objectives FLOWSHEET MODELLING
How to model the performance of fines separators in terms of size, density, shape liberation and
grade?
Milling WHIMS Cyclone
Flotation
Spirals
Screening
Ore characterization - method
A sample of fine hematite ore (-1mm) was characterized using Particle Tracking Analysis (PTA)
• XRD and SEM were used to identify 12 different minerals in the feed sample
• These minerals were analysed and assigned chemical compositions using – Microprobe (on individual mineral types) – ICP chemical assay (on sized and spiral products) – Standard mineralogical compositions (Quartz = SiO2)
• A relative density was assigned to each mineral
Ore characterization - method
Two types of hematite were identified: • Clean hematite with composition close to theoretical Fe2O3 • Hematite with ultra-fine inclusions of silicates
Mineral list
Ore characterization - method
• Approximately 124 000 particles were mapped using AutoSEM. • Data exported as Excel files including: mass, perimeter, area,
assay composition, density, shape parameters, state of liberation of each grain.
Ore characterization - method
• Data sorted into 10 size classes and 14 density classes • Choose the range of each size class • Choose each density interval to match identified minerals
Ore characterization - method What particle size should be used?
Particle Morphology Circularity and Shape factor distributions for each mineral can be measured
Ore Characterization - method Factors that influence PTA reliability
• Stereological Effect – 2D sections of 3D reality – traditionally overestimates liberation • Sufficient data to reliably represent ‘reality’ – 10% mineral vs 0.1% mineral
– 100 000 particles – 1 000 000 particles?
• Number of coarse particles • Mineral surface roughness • Porosity
Perhaps 3D tomography in the future
Ore Characterization - results
Feed sample - Mineral Content
Ore Characterization - results
Total feed – Size x Density
Hematite grains – Size x Density
Ore characterization - results
Feed sample - Size x Grade
Ore characterization - results
Feed sample - Density x Grade
Ore characterization - results
Ore characterization - results
Ore characterization - results
Ore characterization - results
Ore characterization - results
Circuit simulation
Circuit simulation
Partition Surfaces • The performance of each process unit must be modelled in
terms of both size separation and density separation • These partition surfaces are generated experimentally by taking
multiple product and tailings samples at various mass yields
Mill breakage rates per mineral type can also be determined
Circuit simulation
Circuit simulation
Circuit simulation
Circuit simulation
Cyclone performance data
Circuit simulation
Circuit simulation
Spiral 1 performance data
Circuit simulation
Circuit simulation
Spiral 2 performance data
Circuit simulation
Mill performance – no liberation
Mill performance – total liberation
Circuit simulation
Circuit simulation
Input overall circuit data
Circuit simulation
• Opportunity to optimize product grade/recovery with
changing grind size
• Full grade recovery curves can be constructed for each
process unit
• Distinction is made between middlings type – misplaced or
locked particles
• Opportunity to refine unit models over time
• Can include the effect of particle shape
Conclusions
Automated SEM with PTA can be used to characterize fine ore samples in terms of size, density, shape, liberation and composition
The performance of gravity/size separation units at varying yields can be modelled using the PTA data and partition surface models
Simple flowsheets including recycle streams can be simulated
Recommend that this technique be expanded to model other processes such as magnetic separation and flotation
Refine technique for low grade ores (particle count)