Beruflich Dokumente
Kultur Dokumente
)
2015 Australian Centre for Geomechanics, Perth, ISBN 978-0-9924810-3-2
https://papers.acg.uwa.edu.au/p/1511_05_Moulding/
CMouldingGoldFieldsAustraliaPtyLtd,Australia
PAndrewsGoldFieldsAustraliaPtyLtd,Australia
Abstract
This paper outlines the protocols introduced for the management of potential seismic risk, including
monitoring,numericalmodelling,extractionsequencingandthenecessarysupportupgradestomitigatethe
risksatDarlotGoldMine.
TheWaltersLodeislocated400mbelowsurfaceandusesalongholeminingmethodretreatingbacktoa
regional pillar. The lode is hosted within a strong dolerite host rock (uniaxial compressive strength
(UCS)=220MPa+),withaUCSthatoftenexceeds200MPa,withthemainmineralisedlodefollowingalong
a structure called the first slip fault. This structure pinches and swells at varying depths and is usually
identified by quartz intrusions and a 0.51.0m sheared zoned along the plane. This fault dips at
approximately 50 to the west. In addition to the first slip fault, there are multiple intersecting faults
throughoutthearea.
Originally the ground support for the development drives was not designed for seismic conditions and
predominantlyutilisedresinboltsinstalledoverfibrecrete.Astheminingfrontbegantoretreattowardsthe
pillar, mining induced stresses began acting upon the rock mass and the structures in the area. The rock
mass response to these changes included an increase in seismic activity and deformation. This increased
activityledtoareviewoftheseismicmanagementriskproceduresatDarlot.
Initialchangesincludedupgradingthegroundsupportregimetoayieldingsystemcapableofretainingthe
rock mass in response to predicted levels of seismicity. Expected levels of seismicity were obtained by
undertakingareviewofhistoricalseismicdatainthearea.
Nonlinearelasticmodellingwasalsoundertakentodeterminethepotentialseismicresponseandpredicted
damage to the rock mass based on the proposed extraction sequence. Based on these results, a
geotechnicallyoptimisedstopingsequencewasproposed.
Internalnumericalmodellinghasbeenconductedtoaccountfornewlyfoundstopeshapesrelativetothe
originalextractionsequence.Theresultsthatshowpotentialdepthsoffailureareusedtostrategicallyplan
andaccountforadditionaloverbreakorbrowdamageintheminingplan.
Ongoingseismicmonitoringandanalysisisusedtodetermineseismicexclusionzonesandreentrytimes.
Alloftheabovechangesareusedtocontinuallycalibratetheexistingmodels,andongoingdatacollection
hasresultedinanoptimalextractionsequence,thereforeallowingasafeandproductiveextractionofthe
WaltersLode.
1 Introduction
Theselected extractionmethodandminingsequencehaveasignificantinfluenceon the performanceof
deepundergroundstopes.Theperformanceisprimarilyassociatedwithstopegeometry,fieldandinduced
stresses, geological structures, intact rock strength and rock mass characteristics. Darlot Gold Mine has
miningareasthatcapturealloftheseperformancefactors,predominantlywithintheWaltersLodewhich
hashighinducedstressesandcomplexstructuralconditionswithinaremnantminingzone.Thesefactors
have historically led to increased stope over break and an increased seismic response, that both affect
production(i.e.timedelay,increasedcost,safetyconcerns).Asetofmanagementstrategiesandpractices
wereimplementedtomanagethesefactors,includingbothelasticandnonlinearnumericalmodellingto
help geotechnical engineers optimise the stope geometry and extraction sequence. Additional
administrativecontrolshavealsobeenimplemented,suchasexclusionzonesandreentrytimestoensure
seismic risk to personnel is mitigated. This paper presents an overview of the mine design, geological
conditions,stressinfluencesandmanagementprocessestosafelyextracttheorebody.
2.1 Location
DarlotGoldMineislocated58kmeastofLeinster,WesternAustralia.Since1October2013,theminehas
beenownedbyDarlotMiningCompanyPtyLtd,anindirectsubsidiaryofGoldFieldsLimited.Darlotbegan
asanopenpitoperationsin1989,andundergroundoperationscommencedin1995undertheHomeStake
Mining Company when the Darlot orebody was discovered. Underground mining is now located in the
Centenaryorebody.
ThemineislocatedatthesouthernendoftheNNWtrendingYandalGreenstoneBeltoftheYilgarnCraton.
Mining occurs along an orebody with a 1.5km strike length, and active mining is conducted between
350and740mbelowthegroundsurface.TheDarlotmineutilisesvariousextractionsequencesthroughout
theminewhichincludelongholeretreat,roomandpillar,halfechelonretreat,primaryandsecondarybulk
stoping.Themajorityoftheseextractionsequencesarecomplimentedwithcementedpastefill.Thepaste
plantwascommissionedin2006,andstopefillingallowsforcompleteextractionoftheorebodyinsteadof
leavingorepillarsbehind.
2.2 Geology
Figure1showsthemultipleorezonescurrentlyextractedatDarlot.Goldmineralisationispredominately
hostedinmagneticdolerite(85%)andmagneticquartzdolerite(8%).Mineralisationcanalsobefoundin
thenonmagneticdoleriteandfelsicvolcanosedimentaryrocktypes,typicallyinthefollowingconditions:
Narrow(0.2to30cmwide)shallowwesterlydipping(0to20)stackedquartzveins.
Massive(50to400cmwide)bandedquartzveinsthatdipat60tothenorthwest.
The zone known as the Walters Lode is defined by a banded quartz vein. The vein varies from 50cm to
severalmetresintruethickness,witha600mstrikelengthandextends600mdowndip.Minorshallower
dippingveiningalsooccursalongthefootwall(FW)andhangingwall(HW).Theorezoneishostedinvery
strong Dolerite (UCS = 220MPa+) and is dominated by a regional fault structure known as the first slip
fault.Thisfaultvariesinthicknesswithdepthandischaracterisedbyquartzintrusions,narrowalteration
halosanda0.51.0mshearedcontactzonealongtheplane.ThisleadstoapoorerrockmasswithintheHW
ofstopes.
Figure 1 Darlot Gold Mine ore zones
Figure 2 The remaining pillar as of September 2014
The paste fill strengths were designed for 57m wide undercutting and mining beside exposures, using
equationsadaptedfromMitchellandRoettger(1989).
Astheextractionsequenceprogressed,thestressesconcentratedinthecentralpillarandotheradjacent
sill/rib pillars between stopes. Options for an alternative extraction sequences were reviewed, but were
foundtobeunsatisfactoryduetothemaindeclinelocationandothernearbyminingfrontsencroachingon
theareaandbecauseitwasuneconomicaltodevelopFWdrivestoprovidedualaccesstotheorebody.
Figure 3 BN1000 HWN G Panel design cross-section
Figure 4 BN1000 HWN D failure scenario
3 Seismicity
DarlotGoldMineinstalledfourtriaxialandtwelveuniaxialsensorswhenitreplacedanexistingsystemwith
an Institute of Mine Seismology (IMS) seismic system in December 2009. This installation has provided a
more consistent performance compared with the previous system which deteriorated to the point of
failureinearly2009.Thenewsystemallowedforbetteranalysisofallseismicparametersthroughoutthe
mine.
3.1 Analysis
Aspreviouslymentioned,theWaltersLodehasonemajorandmultiplesmallerfaultsinteractingwiththe
ore zone, asillustrated in Figure5. These smaller faults vary in seismic response as stopeblasting occurs
nearby.BackanalysisandreviewoftheseismicityaroundtheWaltersLodeindicatedthatthefirstslipfault
istheprimarystructureassociatedwithseismicity.
Ongoing seismic analysis indicates extensive clustering on the first slip fault contact and in the HW rock
mass,andithasalsobeencalibratedasthesourceofthedamagingseismicevents.Asummaryofhistorical
damagingseismiceventswithinthestresswindowisdisplayedinTable2.
ThelargesteventrecordedonthesystemwasanIMSlocalmagnitude(ML)of+1.3on20February2012,
anditoccurredfortyminutesafterfiringarise,whichintersectedtheFirstSlipfault.Thisisillustratedin
Figure6.
Figure 5 Fault structures and event distribution within the Walters Lode
Figure 6 Seismic events >0.0 ML associated with first slip fault
Rise firings in the Walters Lode have been found to induce an elevated seismic response, and larger
magnitudeevents,thanstandardstopefiringsintovoidsorexistingrises.Thisincreasedlevelofseismicity
from rise firings is thought to be due to the larger energy and vibration being transferred into the
surrounding rock mass in all directions, whereas other stope firings preferentially distribute the blast
energytowardsthevoidinsteadofthesurroundingrockmass.
Thenumberofseismiceventshasdecreasedsince2009,whichmaybeduetothelowerextractionrates
from the Walters Lode since then. Even though the overall number of events has reduced, there is a
generaltrendshowingthenumberoflargermagnitudeeventsincreasedbetween2010andmid2013at
whichpointeventmagnitudesbegantoplateau.Thisappearstocorrelatewiththeextractionofthecentral
accesspillarandtheincreasedstressesactingalongthefirstslipfault.
Figure 7 Frequencymagnitude graph
Withthemaximumseismiceventandwedgegeometrydeterminedfromanalyses,theupgradestoground
supportcouldbeimplemented.Basedontheanalyses,thesupportrequiredagreaterembedmentlength
andalsoahigherdynamiccapability.Thisledtoaseismicgroundsupportregimethatcouldwithstanda
localmagnitude+1.55eventascloseas10mfromthesource.Thisdynamic supportregimeutilisesnine
2.4msplitsetsperring,two6mdynamic(debonded)15.2mmcablesataringspacingof2.5m,50mm
fibrecrete and mesh down to 2m from the floor on each wall. The 6m debonded cables provide an
additional 3m of embedment length to support the larger wedges and allow for energy absorption if
dynamic ground movements occur. The mesh transfers the load to adjacent reinforcement, minimises
unravellingaroundcablesandcontainsanypotentialfibrecreteejection.
The Factor of Safety (FS) for a +1.55ML event at 10m was designed using equations proposed by
Kaiseretal. (1996) in Canadian rockburst support handbook. A target FS of 1.6 has been targeted and
achievedusingthesecalculations:
FS (1)
/
FS (2)
where:
m = massofejectedrock(kg).
v = ejectionvelocityoftheblock(m/s).
g = accelerationduetogravity(m/s2).
d = distanceejectedrockhastravelled.
q = 1,0or1forejectionfromthebacks,wallsorfloorrespectively.
The upgraded support work very well since implementation and has greatly reduced personnel and
equipment exposure to potentially hazardous rockfalls or ejection. Since upgrades were made, large
magnitude events (>1.0) have been experienced; however, all damage to the rock mass has been
effectivelyretainedbythesupportscheme.
Figure 8 W1000 F Rise Firing 2012
Backanalysisofseismicactivityafterfiringshowsthatthemajorityofeventsoccurwithinthefirsthourin
theWaltersLode.Theactualeventsgenerallysubsidefasterthanthedecaycurve;however,spikesinthe
eventsperhourcanoccurrandomly.Thesespikesdonotnecessarilyinterceptthedecaycurve;however,it
doesimplythatthereareseismicirregularitiestoaccountforbeforeanypersonnelaregrantedaccess.
Backanalysis data shows that the level of seismicity decays to background levels within 1820hours,
therefore the Walters Lode maintains a 24hour standoff upon firing. An example of the Omori Chart is
displayedinFigure9.
Figure 9 W1000 F Rise Omori Chart
Activityfromsomeseismicsources,particularlythoserelatedtofaultsliponstructures,canshowapoor
relationshipwithmineblasting.
4 Walters modelling
Both linear elastic (Map3D) and nonlinear numerical modelling have been undertaken for the Walters
extractionsequence.Theaimwastoforecastpotentialareasofdamageandseismicpotentialwithinthe
Walters Lode. Initial elastic models were run with various extraction sequences and the proposed stope
shapes.
ALevkovitchReuschmodelwascreatedtoprovideabasisfornonlinearmodelling(BeckEngineeringPty
Ltd2012).Thismodelshowedthat,regardlessofinitialsequence,thestopesclosesttothecentralaccess
wouldhavethelargestseismicresponse.ItwasestimatedthatthevastmajorityoftheWaltersLodeore
willberecoverable.However,modelledrockmassdamageassociatedwithfiringthecentralaccessstopes
could lead to increased rehabilitation requirements due to constant seismic activity as the central pillar
deformsunderload.Thismodelwasusedasthebasecasefortheoverallextractionsequence.
Sincethenonlinearmodelwasconducted,inhouseelasticnumericalmodellingwasconductedduetothe
additionofnewstopesoutsidethecurrentextractionsequences(Figure10).Thismodellingcomplemented
the existing modelling to determine if there would be any new significant effects on the rock mass or
stressesactingonthefaultsfromfurtherstopeextraction.Backanalysisofhistoricalfailuresandseismic
eventsidentifiedthreecriteriaonwhichforwardplanningwasmade.Table3detailsthefailurecriterions
thatwereassessedalongrockmassparametersfromTable4.
Figure 10 New proposed stopes alongside original stopes in July 2014 (perspective view looking east)
Table 3 Failure criteria assessed within the elastic numerical model (based on back-analysis)
As the extraction proceeded, modelling indicated that the majority of stopes located behind previously
extracted and paste filled stopes were in a stress shadow and were not expected to have increased
seismicity. Stress concentrations remained within the central pillar access areas. Also, stopes which
encroachedonsmallersillpillarsshowedanincreasedfieldstress(70MPa+)betweenthepasteaboveand
thestopes.Thishigherstresscriterionindicatedthat4moutofa6mpillarcouldbeexpectedtofailupon
firingorinducedseismicresponses.Therefore,stopesthatreducedthesillpillarwereconsideredunsafeto
mine(Figure11).
Figure 11 W1035 S D&E proposed shape in the sill pillar
The modelling indicated high 1 values approximately 3.2m into the backs of the drive (Figure12). This
coincides with the anticipated wedge height potential discussed in Section 3.1. The debonded cable
supportwillstillhave2.8mofembedmentwithincompetentrockandshouldstillbeeffectivethroughthe
entirestopingsequence.Inaddition,allbrowsintheWaltersLodewillhaveextracablesupport.
Figure 12 Rock mass damage potential in drives
ItwasfoundthatlossofconfinementintheHWcouldbeaprimarycauseofoverbreakwithinthemajority
of inclined shapes, especially with stopes which were overcut by extracted stopes above. The loss in
confinementisbecausetherockmassintheHWisonlyconfinedontwoabutments(Figure13).
Figure 13 Loss of confinement upon extracting stopes central to stress window
The modelling also allowed for shear stress changes to be recognised at certain steps of the mining
sequence.Itwasfoundthatcertainstopescouldinitiatearegionalstresschangealongthefaultstructures.
Theminingstepsthatshowedasuddenchangewereflaggedashighriskstopesandwereresequenced,
where possible, to late in the Walters Lode sequence. An example of this is the W980 HW A (Figure14)
which induces a regional shear stress change on the first slip fault, and has potential to cause seismic
induceddamageandfallout.
Figure 14 Regional shear stress change pre and post blasting W980 HW A Stope
Based on results from modelling, the following recommendations were made to management, and
documentedinaWaltersLodesequencereport:
Groundsupportneededtobeupgradedin1020and1040Levels(HW/FWdevelopmentaccess
whenitoccurs)toaccountforthemaximumdepthofdamageof3.2mintothebacks.
Evenafterallthegroundsupportupgradeworksarecompleted,occasionalrehabilitationworks
willberequired.
Someorereservesmaynotbemined,andarecoveryof80%isestimated.
Thenumericalmodelsuggeststhatallstopingactivitieswillcorrespondwithhighseismicevent
potential.
Centralaccesspillarstopessuchason1000and1020Levelsaretobemassblastedwhere
possibleandreentrywillbedeterminedbytheseismicitylevels.
Anystopesthatareleftaspillarsshouldbedestressedbymeansofpreconditionblasting,so
thattheydonotattractseismicity.
AvoidstopingtheW1035H/GandW980stopesatthesametime.Thiscouldcausethepillarto
crushandincreasetheshearstressalongthefirstslipfault.ThisisillustratedinFigure14.
Shapechangestothestopesthatundercutdrivesorstopesshouldbereevaluated,asthismay
helpreducestressrelatedfailuresandpotentialwedgefallouts.
6 Conclusion
TheanalysisrevealedthatminingtheWaltersLodecomeswithinherentriskstoproductionandpersonnel,
regardless of the chosen mining sequence. Increased seismicity and rock mass damage are an inevitable
consequenceofreducingtheregionalpillardimensions.Thishasbeenfactoredintotheschedule.
The final analysis showed that majority of safety issues could be mitigated with administrative control
systemsandappropriategroundcontrolstandards.
In addition to these management protocols, extraction will be closely monitored to assess whether the
seismic response and stope performance matches the modelling results. With new information available
aftereachstopefiring,thereistheabilitytofurthercalibratethemodeltoprovideanupdatedextraction
scenarioandhighlightanyneworchangingrisks.
Overall,theperformanceofthestopeextractionhasbeenasexpected,withdilutionfactorsbeingslightly
less than the modelling results. In accordance with the Australasian Joint Ore Reserves Committee
(JORC2012) requirements, indicated modelled dilution was incorporated into the ore reserve economics
(asgeotechnicalconditionshaveamaterialeffect).Theseismicityhasbeenconsistentwiththemodelling
resultstodate.However,theeventmagnitudeshavenotbeengreaterthanthe+1.3eventwhichoccurred
on 20 February 2012. There has been no fall of ground or chronic deterioration of support within the
WaltersLodesincethesupportupgradeswerecompletedin2013.Itisexpectedthat the upgradeshave
dramaticallyreducedtheamountofrehabilitationworkrequired.
Inconclusion,itwouldbeconsideredacriticalsafetyandproductionriskiftheseanalyticalmethodswere
not conducted for a regional pillar scenario. Without the forecasted effects of mining, the Walters Lode
wouldhavebeenanunknown,andwouldlikelyhaveledtounexpectedpersonnelandequipmentexposure
to rockfall hazards; increased rehabilitation after stope firings; brow failures; additional over break;
heightenedseismicresponses;andeventhepotentialforsterilisationoftheremainingorereserves.
Acknowledgement
Theauthorsthankthefollowingorganisationsandpersonsforprovidingsupportandinvolvementwiththe
Waltersregionalpillarextraction.
DarlotGoldMineforprovidingimagesandsupport.
EmmaJonesandChristopherChesterBarrickGoldofAustralia.
DavidBeckandCharlesLilleyBeckEngineeringPtyLtd.
JohanWesselooandPaulHarrisAustralianCentreforGeomechanics.
References
Barton, N, Lien, R & Lunde, J 1974, Engineering classification of rock masses for design of tunnel support, Rock Mechanics and
RockEngineering,vol.6,no.4,pp.189236.
BeckEngineeringPtyLtd2012,CalibrationandforwardanalysisofBoonNorthmiddleWaltersstresswindowarea,report,Beck
EngineeringPtyLtd,ChatswoodWest.
Gutenberg,R&Richter,CF1944,FrequencyofearthquakesinCalifornia,BulletinoftheSeismologicalSocietyofAmerica,vol.34,
pp.185188.
JORC2012,JointOreReservesCommittee,JORCCode:AustralasianCodeforReportingofExplorationResults,MineralResources
andOreReserves,JointOreReservesCommittee,SouthCarlton,pp.1819,http://www.jorc.org/docs/JORC_code_2012.pdf
Kaiser,PK,McCreath,DR&Tannant,DD1996,Canadianrockburstsupporthandbook,GeomechanicsResearchCentre,Sudbury.
Mitchell,RJ&Roettger,JJ1989,Analysisandmodellingofsillpillars,inFPHassani,MJScoble&TRYu(eds),Proceedingsofthe
Fourth International Symposium on Mining with Backfill, Innovations in Mining Backfill Technology, Balkema, Rotterdam,
pp.5362.