Applications of geometallurgy for waste characterisation across the - - PowerPoint PPT Presentation

Applications of geometallurgy for waste characterisation across the mining value chain

Dr Anita Parbhakar-Fox

Senior Research Fellow WH Bryan Mining and Geology Research Centre, SMI, University of Queensland, Brisbane, Australia

What is ‘geometallurgy’?

GEOLOGY MINING METALLURGY

GEOMETALLURGY

Purp rpose: In Increase profit fits

Keeney (2008): Aimed to propagate measured processing attributes (i.e., hardness, grindabillity) down in the matrix to Level 2 and Level 1

Defined linkages are essential

Geometallurgy Matrix concept

Defined linkages are essential

Geometallurgy Matrix concept

For mine waste characterisation a geometallu lurgical matrix ix ap approach could be readily adopted to de- risk projects and improve long- term financial outcomes

Small-scale, simple, low-cost Full-scale, complex, high cost

Requires the embedding of geoenvir ironmental l proxy tests at the earliest LOM stages (i.e., exploration/prefeasibility) Representative sampling to capture heterogeneity is a key issue- this helps overcome it

The (enviro)geometallurgy tool kit

Handheld tools Hyperspectral mineralogy Data mining ‘Next-gen’ technologies Automated mineralogy Simple chemical tests

Hyperspectral mineralogy

• Challenges encountered when collecting ‘representative’

geoenvironmental samples at early life-of-mine stages

• Increasing ore deposit knowledge will assist with static and

kinetic testing sample selection

• Hyperspectral data measuring VNIR and SWIR active

minerals (e.g., Corescan) and TIR (e.g., HyLogger)

• Corescan: ~2,000 m can be collected per day
• Value-add opportunity by perform geoenvironmental

domaining to support waste forecasting

• Ide

Identify ify pot potentia ially ly ac acid id form

• rmin

ing, g, non non-acid id form

• rmin

ing and and ne neutralis ising dom domains to

• en

enable was aste man anagement thr through ea early ly for

• recastin

ing of

• f geo

eoenvironmental l ch characteris istic ics

Hyperspectral mineralogy

Type Silicate Structure Mineral Group Example VNIR Response SWIR Response TIR Response Silicates Inosilicates Amphibole Actinolite Non-diagnostic Good Good Pyroxene Diopside Good Moderate Good Cyclosilicates Tourmaline Dravite Non-diagnostic Good Moderate Neosilicates Garnet Grossular Moderate Non-diagnostic Good Olivine Foresterite Good Non-diagnostic Good Sorosilicates Epidote Clinozoisite Non-diagnostic Good Good Phyllosilicates Mica Muscovite Non-diagnostic Good Moderate Chlorite Chlinochlore Non-diagnostic Good Moderate Clay minerals Illite Non-diagnostic Good Moderate Kaolinite Non-diagnostic Good Moderate Tectosilicates Feldspar Orthoclase Non-diagnostic Non-diagnostic Good Albite Non-diagnostic Non-diagnostic Good Silica Quartz Non-diagnostic Non-diagnostic Good Non-silicates Carbonates Calcite Calcite Non-diagnostic Good Good Dolomite Dolomite Non-diagnostic Good Good Hydroxides Gibbsite Non-diagnostic Good Moderate Sulfates Alunite Alunite Moderate Good Moderate Gypsum Non-diagnostic Good Good Borates Borax Non-diagnostic Good Uncertain Halides Chlorides Halite Non-diagnostic Moderate Uncertain Phosphates Apatite Apatite Moderate Moderate Good Oxides Hematite Hematite Good Non-diagnostic Non-diagnostic Spinel Chromite Non-diagnostic Non-diagnostic Non-diagnostic Sulfides Pyrite Non-diagnostic Non-diagnostic Non-diagnostic

Linton et al. (2018)

Hyperspectral mineralogy

Cor Core ph photography Min ineral Cl Class s map Ch Chlor lorite wavelength pos posit ition Ch Chlor lorite match intensit ity Geo Geotechnical par parameters

Hyperspectral mineralogy

Core photography Mineral map Carbonate match

Mixed pixels are classified based on the most abundant spectra

Class map colour index

Aspectral Sericite Sericite + chlorite Quartz-carbonate Chlorite Clinochlore Quartz/silica Carbonate Low match Carbonate match High match

Co Core pho photogr graphy Min Mineral class ma map Su Sulfid fide di distrib ibutio ion Log Log Su Sulfid ide di distrib ibutio ion

Hyperspectral mineralogy

Scaled Neutralising Potential/ Acid Potential values Relative reactivity values Calculated Mineral abundance

(Jambor et al., 2007; Parbhakar-Fox and Lottermoser, 2014) (Sverdrup, 1990)

Chlorite: 60 % Carbonate: 30 % Example Quartz: 10 %

* 0.02 * 1 * 0.006 * 1 = 30 = 0.00012 * 0.004 * 0 = 0 Pixel GDI = ~30

Geoenvironmental Domaining Index (GDI)

Core images Mineral maps

Hyperspectral data

* *

Jackson et al. (2018)

Hyperspectral mineralogy

Fir irst pass GDI (V (V2) valu lue risk risk as assessment with ith su sulf lfid ides ide identif ifie ied defi fines 5 ris risk grade clas lassif ific icatio ion fie field lds

GD GDI valu alue GD GDI risk risk grade De Description of

• f geoenvir

ironmental ch characteris istics

• 35,0

,000 to

• -900

900 Extreme risk risk Do Dominance of

• f acid

cid formin ing min inerals

• ls. Sulfi

lfides id identif tified as fir first t min ineral l > > 75 %. . No

• pri

rimary ry neu eutrali lisers (A (AP >> >>NP).

• 900

900 to

Hig High risk risk Sulfi lfides com

• mmon. Sulfi

lfides es id iden enti tifie ied as 2nd

nd and 3rd rd min

ineral < < 75 %. . No pri rimary neu eutralis isers (A (AP >N >NP). 0 to

• 10,0

,000 Pot

• ten

ential l risk risk Do Dominated ed by silic ilica/quartz, seri ericit ite, ch chlo lorit ite. Few sulfid lfides es present, min inor r pri rimary ry neu eutrali lisers (AP≠NP). Som

• me gy

gypsum pres esen ent. 10,0 ,000 to

• 40,0

,000 Low risk risk Ca Carbonate abundance < < 50 % (A (AP<NP). 40,0 ,000 to

• 100,0

,000 Very ery lo low risk risk Ca Carbonate dom

• minates as fir

first Co Cores escan min ineral > > 50 %. . Lon Long ter erm acid cid neu eutrali lising capacity lik likely ely (A (AP<<NP).

Jackson et al. (2018)

Hyperspectral mineralogy

Sample A: : Skarn GD GDI V2: 2: 34 34,3 ,370 Lo Low risk risk

Cor

• re ph

photography Clas lassi sifi fied min ineral map ap Sul Sulfi fide rec ecognition Car arbonate ide identifi fication

St Static testin ing= NA NAF (H (Hig igh ANC)

Jackson et al. (2018)

Hyperspectral mineralogy

Hyperspectral mineralogy

GD GDI V2: 2: 1910 1910 Potential l risk risk St Static testin ing= NA NAF (3% (3% su sulf lfid ide-sulf lfur; ; 23 23% cal alcit ite)

Chlorite dominated

Sample B: : Skarn

Cor

• re ph

photography Clas lassi sifi fied min ineral map ap Sul Sulfi fide rec ecognition Car arbonate ide identifi fication

Jackson et al. (2018)

Hyperspectral mineralogy

GD GDI V2: 2:

• 14

140= Hig igh risk risk St Static testin ing= PAF/AF

Sericite-dominated

Sample C: : Porphyry ry Au-Cu (Potassic Alt lteration Zon

• ne)

Cor

• re ph

photography Clas lassi sifi fied min ineral map ap Sul Sulfi fide rec ecognition Car arbonate ide identifi fication

Jackson et al. (2018)

Hyperspectral mineralogy

Additional applications when scanning column feed materials prior to kinetic testing – results to be published later in 2019

Handheld tools and chemical tests

En Environmental Log Loggin ging Ch Chemical St Stain ainin ing Har ardness me measurements pXR pXRF Fi Field eld che hemic ical tes ests

Integration of results provides the best quality information to feed into the geometallurgical matrix Not all are new, but not routinely applied for geoenvironmental characterisation

Handheld tools and chemical tests

Acid cid Rock

• ck Dr

Drain inage In Index (A (ARDI)

Parbhakar-Fox et al. (2011; 2018); Opitz et al. (2016); Cornelius et al. (2017)

EQUOtip

Min ineral hardness to

• deter

ermine rate

• f
• f wea

eath thering g and predic ict elu eluti tion of

• f acid

cid/ neu eutrali lisation

Handheld tools and chemical tests

Parbhakar-Fox et al. (2015)

Handheld tools and chemical tests

Parbhakar-Fox et al. (2015)

Automated mineralogy

Min ineral l Lib Liberation Analyser Curr rrent practice: Application in in predictiv ive ARD ch characteris isation testwork an and tail ailin ings ch characteris isation

Co Commonly ly use sed techniq iques do not t allo allow for r lo low-cost hig igh volu lume analy lysis is- can XMOD be use sed?

SPL Lite Target sulphide phases & characterise grain properties

Buc Buckwalt lter-Davis is (20 (2013) Six tailings samples New Caloumet mine, Canada

Ho Hours

XBSE GXMAP Characterise grain properties for mineral of interest and examine associations

Ar Aranta (20 (2010): 4 waste rock samples, Antamina Mine, Peru Par arbhakar-Fox (2012): ): 10 waste rock samples, Lode-Au mine, 9 IOCG samples, Australia Edr draki i et t al

• al. (20

(2014): Cu-Au porphyry tailings

Ho Hours

Automated mineralogy- tailings fingerprinting

30 mins

FEI Quanta 600

XM XMOD Parbhakar-Fox et al. (2017)

Data mining and machine learning

Calculate mineralogy using assay data (e.g., Berry et al., 2015; Beavis et al. 2017; Howard et al., 2019) Extract more information from existing data sets e.g., mineralogy and texture (Cracknell et al., 2018) Opportunity to enhance waste domaining e.g., using Ca and Mg from assay (Jackson et al., 2019) Matlab SQL

High-res drill core image

Data mining and machine learning

‘Next gen’ technologies

X-ray tomography + XRF

Orexplore core scanning – structural features,

• re and gangue phase morphology

(200 µm voxel resolution)

Sulphide distribution - Sunrise Dam Pyrite – Rio Blanco tourmaline breccia Cu deposit

3D A-ARDI assessments TruScanTM Minalyze CS

26

Forecast the potential for future mine wastes to fix atmospheric CO2 (using TIR data): Develop GHG consumption index Identify ‘soft’ zones based

• n classified mineralogy:

Predictive dust characterisation protocol Spent heap leach materials: identify and characterise post-leach mineralogy (e.g., alunite-group)

Spent heap leach pile, Croydon Au-mines, QLD

Additional uses of geometallurgy data and tools

Mine waste: Ore bodies of the future

New cobalt lt reso sources Zinc inc fr from sla slag Tin Tin and gold ld fr from his istori ric tail ilin ings New ind indiu ium reso sources?

Redr edraw awn fr from

• m MRT (20

(2001 01)

Parbhakar-Fox et al. (2018): https://www.mdpi.com/2075-163X/8/10/454

Mine waste: Ore bodies of the future

Planning to return and drill up to 5 drill holes @ 60 m depth perform geometallurgical and geoenvironmental testwork

Mine waste: Ore bodies of the future

Min ineralogical l & ch chem emical data analy lysis to

• predic

ict AMD ch characteristics ‘Next gen’ technologies and new ch chemical testin ting Sen ensor-based waste asses essments duri ring

• p
• perational

l stages Tailings ‘fingerprinting’ durin ring dep eposition Ch Characterisati tion of

• f

his istoric min ine e sit ites es and waste to

• deter

ermine reu euse New asses essmen ent t too

• ols

ls and processing ap approaches

“Transform how exp xplo lorers and mine iners pla lan and predic ict mini ining and envir vironmental l acti tivit itie ies, , by provi vidin ing new tools ls to guid ide th these acti tivit itie ies fr from th the init initia ial l disc iscovery ry through to end of mine life”

‘Enviro’ opportunities in geometallurgy

Thank you

Sustainable Minerals Institute University Experimental Mine 40 Isles Road, Indooroopilly, QLD 4068

T +61 7 3365 5977 M+61 400 850831 E a.parbhakarfox@uq.edu.au

What is ‘geometallurgy’?

• Through an integrated approach geometallurgy establishes 3D models which enable NPV
• ptimisation and effective orebody management, while minimising technical and
• perational risk to ultimately provide more resilient operations
• Critically, through spatial identification of variability, it allows the development of strategies

to mitigate the risks related to variability (e.g., collect additional data, revise the mine plan, adapt or change the process strategy, or engineer flexibility into the system)

• To achieve these goals, development of innovative technologies and approaches along the

entire mine value chain are being established

• Geometallurgy has been shown to intensify collaboration among operational stakeholders,

creating an environment for sharing orebody knowledge, leading to the integration of such data and knowledge into mine planning and scheduling

• Co

Companies es th that t em embrace th the e geo eometall llurgic ical l approach will ill ben enefit it fr from in incr creased net t present valu lue and sharehold lder valu lue

Dominy et al. (2018)