mine-to-plant optimization in concentrators and hydro …€¦ · mine-to-plant optimization in...
TRANSCRIPT
MINE-TO-PLANT OPTIMIZATIONIN CONCENTRATORS AND HYDRO
PLANTS
J.M. Menacho, L.A. Verdugo and G.E. Vega, De Re Metallica Ingeniería SpA, DRM-TechI Congreso Internacional de Conminución de Minerales (Intermet) , Mayo 11 – 13, 2017, Lima, Perú.
Contents
• Framework
• Fragmentation Models and Blast Simulators
• Mill Models and Milling Simulators
• Mine-to-Plant at Concentrator and Hydro Sites
• Final Remarks
The Virtuous Mine-to-Plant Cycle…
Mine Plan
ConcentrationPlant
Production, Sales, Profit
Drillingand
Blasting
ComminutionPlant
MORE…Robust monitoringdevices and analytica are
needed to verify whether thereality fit the expected resultsand how can we permanently
improve them
THIS PAPER…Topics in ComminutionAnalytica for Mine-to-
Plant Applications
• Geology• Geotechnics• Geometallurgy• Hydrogeology• Mining Operation
Fragmentation Models• Kuz-Ram model
• SveDeFo´s fragmentation equation
• Kou-Rustan´s fragmentation equation
• NBC approach
• Chung and Katsabanis model
• Bergmann, Riggle and Wu model
• JKMRC models: “The Crushed Zone Model” and “The Two Component Model”
• DRM PBM-model
• Others
https://www.diva-portal.org/smash/get/diva2:995258/FULLTEXT01.pdf
• Most are empirical equations
• The Kuz-Ram equation is the most used with poorprediction of the fines content
• The JKRMC models employ a bimodal Kuz-Ramequation to improve the estimate of fines
• DRM model is phemenological in nature
The DRM PhenomenologicalFragmentation Model
i - 1_ _
i E Ei i i j j j_
j = 1i > 1
dw t= -S w E + b S w (E)
dE
_
Ej
i_-S E
i i jj = 1
w (E) = a e
k
i
i - 1
i ikk = 1i > 1
i - 1E
ik k jE Ek = jj
0
w 0 - a
1S b a
S - S
ija =, withThe integrated Reid´s solution is:
Classical PBM approach; batch process;breakage habit and specific rate ofbreakage parameters are recognized. B.C. isw(0) = f(0) for = 0.
_
E
The SiE and Bij parameters are sensitive to
blasting design and operational parametersand also to attributes of the rock mass.
DRM has a commercial software version especiallydesigned for Mine-Planning applications
Constitutiveequations
DRM Model Performance
Imageanalysis
Crushed Product Predictedfrom the Fragmentation Model
Size distribution of the ROM productand the the crushed product are well-predicted from the DRM.Blast andDRM.Crush fragmentation models
ROM GranulometryComputed In The Muck Pile
Blast Simulators• I Blast, TBC Co., Nice, France, Physics based simulator• Blo Up, Itasca, Illinois, USA, Physics based simulator• JK Sim Blast, JKRMC, Australia• Others
http://www.dna-blast.com/Blasting-Design-Simulation-Optimization/Blasting-Optimization.html
Sellers, E. et al., (2012) “Improved Understanding of Explosive – Rock Interactions Using the Hybrid Stress Blasting Model,” in SHIRMS 2012 (Proc. of the 2nd Southern Hemisphere Int. Rock Mechanics Symp., South Africa, May, 2012).
http://www.soft-blast.com/JKSimBlast/JKSimBlast.htm
• Oriented toward design of drill and blast operations: Fragmentation Image Analysis; Muck-Pile shapeprediction; Free face and Cratering visualization, PPV, Frequencies and Waveform prediction, Air Blast model
• The Kuz-Ram model is the preferred tool to estimateROM ore granulometry
Milling Models
POWER MODELSBond, Austin, Morrell, Hogg & Fuerstenau,
others
POPULATION BALANCE MODELS
(PBM)Austin, Herbst, Lynch,
Morrell, others
DEM MODELSDesign of liner, chute,
SAG mill discharge, other similar
Classes of grinding models
Construction based on A. Doll, Technical Memo to UBC Mine331 Class, Nov. 11, 2015.
PBM SAG Mill Modelling
PRODUCTFEED
LOAD
PARTICLE BREAKAGE• Rate of Breakage• Breakage Habit
POWER CONSUMPTION• Charge Dynamics
MASS TRANSPORT• Fluid Dynamics
DISCHARGE SYSTEM• Classifier
Arrangement
The best known
Modest fundamental support
The poorestdescribed
The mostprofitable
Mixture of Geomet
Units
The DRM SAG Mill Model
Grinding/TransportChamber
wi, WCO
Initial values ofW, JC, rC, PM
ci
Updated values ofW, JC, rC, PM
PP´
p´i
F´
P´i
F
fi
TrunnionDischarge
Pulp LiftersGrate
pi
L.G. Austin, J.M. Menacho and F. Pearcy, “A General Model for Semi-autogenous and AutogenousMilling”, - APCOM 87. Proc. 20th Int. Symp. Appl. of Computers and Mathematics in the MineralIndustries. Vol. 2: Metallurgy. Johannesburg, SAIMM, 1987. pp. 107 - 126.
• Particle Breakage : Austin approach, with energy-based equations
• Power Consumption : Hogg/Fuerstenau
• Mass Transport : Flow through Porous Media, Menacho
• Discharge System : Classifying/Splitting devices, open knowledge
The DRM Mass Transport Model
2P μ- = q + βρq
z k
2 23C P C i i
0.5+δC Ci
P C i ii
42.86 A μ 1-ε ( w x )A ε
P´= + Wρ 1.75 1-ε L w x
The turbulent slurry flow through theSAG mill is represented by the Ergunequation. In one dimension it is given by:
Making use of the Forchheimer-Dupuit’s approach, a closed relation between thecirculating flowrate P´and the slurry hold up W is deduced. Note that W α P´1/0.5+d.
The slurry in the SAG mill behaves close to a Bingham fluid, then the apparentviscosity μP is a function of the yield stress and the Bingham viscosity of the slurry,
-1
P S Bingham0μ = du dy τ +μ
S. Ergun and A.A. Orning, "Fluid Flow Through Packed Columns", Chem. Engng. Progr. 48 89-94 (1952).
J.M. Menacho and P.A. Chávez, “Mass Transfer in SAG Milling”, PROCEMIN 2012, Santiago, Chile (2012).
DRM Mass Transport Model
10
12
14
16
18
20
22
24
26
28
30
800 900 1000 1100 1200 1300 1400 1500
Fresh Flowrate, m3/h
Mill
Fra
ctio
nal H
old
-Up J
, %
JKMRC DRM
DRM approach is more sensitive to over filling
J.M. Menacho and P.A. Chávez, “Mass Transfer in SAG Milling”, PROCEMIN 2012, Santiago, Chile (2012).
Comminution Simulators
• JKSimMet software, Nappier Munn et al., JKRMC
• ModSim simulator, R.P. King, MTI
• MolyCop Tools, J.E. Sepúlveda, MolyCop
• Usim Pac, Caspeo
• MinProSim, J. Delgadillo, MinProSim Consultants
• MetSmart, Minerality, Toronto, Canada
• MetSim, J. Bartlett, Proware (specific applications)
• Others
http://www.dna-blast.com/Blasting-Design-Simulation-Optimization/Blasting-Optimization.html
Mine-to-Plant Challenges at Concentrator Sites
• Process-oriented blasting strategies taking into consideration metallurgicalattribute distribution in the orebody
• Improved measurement/estimate of bench structure (joints and fractures)
• Better measurement/estimate of the fines content in the muck-piles
• Better geometallurgical modelling of the orebody
• Effective blastability lab test (rock strength and crack formation)
• More representative crushability lab test
• Improved monitoring devices for on-line control of the inner SAG millgeometry and rheology and its relationship to power draw and capacity
• Reliable models of mass transport through SAG mills, useful for more realisticoptimization and control issues
Case Study N°1:Mine-to-Plant at a Concentrator Site
BlastingPrimaryCrusher
PebbleCrusher
Ball Mills
SAG Mill
HydrocycloneBatteries
Stock Pile
Open Pit Mining
Primary Crusher• Capacity, t/h• Power, kW
• 4,000• 600
Pebble Crusher• Capacity, t/h• Power, kW
• 1,000• 325
Throughput (Base Case), t/h 1,630
ROM P80 (Base Case), mm 108
Powder Factor, g/t 300
Burden/Spacing, m/m 8/9
Drill Diameter, inch 10.625
SAG Mill Diameter, m 9.75
SAG Mill Lenght, m 4.57
SAG Mill Power, kW 13,263
SAG Mill Speed, rpm 9.48
J Balls, % Volume 22
Apparent Density, t/m3 4.00
BM Diameter, m 6.71
BM Lenght, m 10.98
BM Power, kW 27,748
BM Speed, rpm 12.25
J Balls, % Volume 32
Apparent Density, t/m3 4.76
Case Study N°1:Mine-to-Plant at a Concentrator Site• ROM granulometry profiles for different powder factors in the blasting step
Case Study N°1:Mine-to-Plant at a Concentrator Site• SAG mill reduced selection function and throughput/power draw relationship
2,050 - 1,630=25.8%
1,630
1,630 t/h 2,050 t/h
SAG
Mill
Mo
tor
Pow
er D
raw
, kW
Case Study N°1:Mine-to-Plant at a Concentrator Site• Mill power draw/powder factor and pebble size and tonnage vs. powder factor:
SAG
Mill
Mo
tor
Pow
er D
raw
, kW
Mine-to-Plant Challenges atHydro Process Sites
• Improved technical and economical Mine-to-Plant strategies aimed tomaximize plant capacity/recovery at minimum cost, constrained by theeconomical heap permeability
• Process-oriented blasting strategies taking into consideration metallurgicalattribute distribution in the orebody
• Improved measurement/estimate of bench structure (joints and fractures)
• Better measurement/estimate of fines in the muck-piles
• Better geometallurgical modelling of the orebody
• Effective blastability lab test (rock strength and crack formation)
• More representative crushability lab test
• Reliable dynamic heap leaching models aimed to evaluate the impact of the overall comminution step into hydrodynamic and metallurgical responses of the heap
The DRM Hydrodynamic andMetallurgical Heap Leach Model
Variable-saturation liquid infiltration through porous beds is described by theRichards´ equation. In one dimension it is,
θ h= k h + 1
t z z
Where θ is the saturation, k is the hydraulic conductivity, h is the matric potential,z is the height coordinate and t is the time. Boundary conditions are as follows:
0 0
z=0
hθ z,0 = θ and k h + 1 = q 0,t
z
The dynamic liquid inventory inside the heap and the effluent flowrate are themain responses of the model.
L.A. Richards (1931), "Capillary Conduction of Liquids through Porous Mediums“, Physics, 1(5): 318- 333.
J.M. Menacho and F.J. Troncoso (2013), “Scale Up of Heap and Dump Leaching Results”, COPPER 2013.
The DRM Hydrodynamic andMetallurgical Heap Leach Model
An advection-dispersion model is used to describe the reactive transport ofCopper
Sθc c c= θD - qc + ρ
t z z t
Where c is the mean copper concentration in the internal leach solution, D is thedispersion coefficient, q is the application rate, ρ is the apparent density of theheap and cS is the Copper grade in the ore. Leach kinetics is given by:
S
0 S SM
c q= -(λ + λ )(c - c )
t ρz
Where λ0 and λ are specific rate parameters and cSM is the lowest Copper grade inthe leached ore (residual grade).
Performance of the DRMHeap Leach Model
PLS Flowrate
Cu Concentration in PLS
Cu Production
Mine Plan Production P1:Tonnage x Cu Grade x Recovery
Hydro Plant Production P2:PLS Flow x Cu Concentration x SX Efficiency
HS
HT
CHS
CHT
F1 ROM
F2
F3F4
F5
F7
F6
F8F11
F9
F10
Producto Final
PF
HS
HT
CHS
CHT
F1 ROM
F2
F3F4
F5
F7
F6
F8
F11
F9
F10
Producto Final
PF
Heap Leaching SX-EW EW-CathodesAgglomeration
Drum
Blasting
Crushing Plant
Options
Case Study N°2:Mine-to-Plant in a Leach Site
Reference Case 1 Case 2
Case Study N°2:Mine-to-Plant in a Leach Site
1 Increase in Powder Factor2 Id 1 + 3ry Screen from 12.7 mm to 15 mm
3 Id 2 + 3ry Circuit Open4 Id 1 + 2ry Screen U/F to Final Product
Hard 1 Hard 2 Hard 3 Hard 4Soft 1 Soft 2 Soft 3 Soft 4
Case Study N°2:Mine-to-Plant in a Leach Site
1 Increase in Powder Factor2 Id 1 + 3ry Screen from 12.7 mm to 15 mm
3 Id 2 + 3ry Circuit Open4 Id 1 + 2ry Screen U/F to Final Product
Hard 1 Hard 2 Hard 3 Hard 4Soft 1 Soft 2 Soft 3 Soft 4
+8,000 Cu t/y
Final Remarks• Most of the existing fragmentation models are empirically-based. The Kuz-
Ram model is the most used by the blasting operators
• Even the modern JKMRC fragmentation models are improved variation of thesame Kuz-Ram model
• Conversely, the DRM fragmentation model is phenomenological in natureproviding a fundamental frame to understand the nature of the process andoptimizing parameters. Intensive testing at industrial operations validates itspredictive capacity and practical utility as a Planning Tool
• Blast simulators existing in the market are mostly oriented toward design ofdrill and blast operation rather than process optimization objectives
• Three kind of milling models can be found in the current market: (i) DEMmodels mainly oriented to design of pieces and machines; (ii) Power models,mainly oriented to equipment dimensioning and (iii) PBM models orientedtoward industrial optimization issues like the Mine-to-Plant applications
Final Remarks• There exist at least 20 different experimental tests to estimate specific energy
consumption linked to power draw models for machine dimensioning. None isfully satisfactory
• Mass transport through the SAG mill chamber is a pending task. The Austinempirical power-equation is the most used with relative success
• DRM mass transfer model is based on fluid transport through porous media.Explicit dependency on the slurry rheology, load apparent density and internalgranulometry naturally appear in this fundamental model. Initial evaluation isquite encouraging
• The bottle neck to maximize capacity at Concentrator Sites is often in thegrinding step. Early fines production at the mine may increase capacity from5% to 30% depending on compatibility with the existing facilities
Final Remarks• Mass transfer becomes “throughput controlling” when the SAG mill feed gets
finer as result of more intensive blasting practices. Reliable tools to deal withthis situation grow into a first need
• In Hydro Sites comminution end at the crushing plant. Moderate increase incapacity compared to Concentrator are here expected. An excess of fines maydrastically reduce heap permeability and hence reduce efficiency in theleaching step.
• The blasting practice in Hydro Sites may be foccused to maximize productionbelow 2” subjected to produce less than 15% below 100# in the final crushedproduct
• Relevant current technological challenges in Mine-to-Plant issues are:
Process-oriented blasting strategies taking into consideration metallurgicalattribute distribution in the orebody
Improved measurement/estimate of bench structure (joints and fractures)
Final Remarks Better measurement/estimate of the fines content in the muck-piles.
Current methods are “blind“ below 2 inches
Better geometallurgical modelling of the orebody
Effective blastability lab test (rock strength and crack formation)
More representative crushability lab test
Improved monitoring devices for on-line control of the inner SAG millgeometry and rheology and its relationship to power draw and capacity
Better models for mass transport through SAG mills, useful for morerealistic optimization and control issues
Better dynamic heap leaching models aimed to evaluate the impact of theoverall comminution step into hydrodynamic and metallurgical responsesof the heap