You are on page 1of 6

RockMechanics the National in Interest, Elsworth, 77nucci Heasley(eds), & 2001 Swets& Zeitlinger Lisse,ISBN90 2651 827 7

Probabilistic assessmentsub-level of hangingwall stability theStillwater at


Mine, Nye, Montana
RadfordB.Langston
RockMechanics Engineer, Stillwater MiningCompany, MT Nye
Dr. Hendrik A.D.Kirsten

Steffen, Robertson, Kitsten, and Johannesburg, Africa South

ABSTRACT: The Stillwater Mine, situated the Stillwater in Complexin south centralMontana, USA has

produced palladium platinum and fromtheJ-M reefsince 1987.Miningis primarily done cut-and-fill by methods. thepast years, Over 4 however, long-hole stoping practices been have playing larger in mine a role production. amplified This utilization long-hole of mining improves productivity lowers during and risk miningactivity. Greater reliance thismining on method, however, results a potential decreased in for hangingwall
stability. therefore It becomes important quantify very to sub-level hangingwall stability. Datahasbeencom-

piled outlining parameters affecting stability. Subjective assessments probability failure of the of have been
applied each to stope. From assessment, this a skeletal graph relating probability failure rockparameters of to vs.stope geometry been has proposed. Classes sub-level of hangingwall stability been have identified and
presented matfixformthatcanbe easilycommunicated. in
1 INRODUCTION 2 OBJECTIVES

points identify objectives this the of The Stillwater Mine, thelargest primary producer Thefollowing

of palladium platinum the Western and in Hemisphere situated theStillwater is in Complex south in
central Montana, USA. The Stillwater Complex a is

paper.

42-km long layeredstratiform lopolithapproximately b.y. old.It hosts palladium plati2.7 a and numrich layerknownas the J-M Reef.Stillwater Mining Company produced has palladium platiand
num from the J-M reef since 1987. Previouspro-

Specify subjectively, likelihood failure the of fora specific sub-level hangingwall. Identify classes sub-level of hangingwall stabilityin terms can easily that be communicatedto mine operators. Describe modesof failurein sub-level hangingwalls.

duction primarily has beenaccomplished variaby


tions on cut-and-fill methods. However, over the

Develop skeleton a graph geometry a of of vs.

past years, 4 long-hole stoping practices been have playing everincreasing in mine an role production. This amplified utilization longholeminingimof proves productivity lowers of injuryto perand risk
sonnel. Greater reliance on this mining method,

rock quality assesses that -hangingwall stability in terms probability failure. of of


3 PREVIOUS WORK

however, results a potential decreased in for hangPrevious work by Potvin(1988), Potvin and Milne therefore becomes important investigate very to fail- (1992),andNickson (1992)whichbuiltuponinitial ure mechanisms sub-level for hangingwalls. deA workby Matthews a1(1981), et produced Stabilthe Methodfor cable bolt support design. This sire on the part of StillwaterMining Company's ity Graph on of managementmorefullyunderstand to hangingwall workwasbased the analysis 350 casehistories from underground minesin Canada.The stability thesub-level in open stopes resulted the collected in initiation thisproject. project ongoing of The is the- method,often referredto as the Matthews or Potvin considers primaryelements affecting open sisresearch should, thispoint,be considered analysis, and at a progress report. Thirteen casestudies havebeen stope design.Rock mass strength,structure,the around opening, andshape, orithe size and compiled outliningstructure, rockmass quality, stresses of the stress, geometry, mining stope and sequence. Failure entation the openingare usedto determine stabilityof the stope with or withoutsupport. This modes alsobrieflysummarized. are

ingwallstability the sub-level in openstopes. It

417

method alsorecommends levelsof cablebolt density when cablebolt support utilized.Figure2 shows is thegraph designed refined these and by workers
4 PROCEDURE

for Jr and Ja were assessed from direct observation.

The second factor usedin calculating modified the stabilitynumberN' is the calledthe rock stress factor A, which relates the maximum induced stressto

The procedure utilizedfor this studyinvolvedcollectionof 13 casehistories sublevel of stopes. Initial data involved identificationof rock type and assessment rock strength all three walls. Block of in size and shapewere also categorized. Additionally, support typeswere assessed, have not beenrigbut orously considered thetimeof thiswriting. at Ratherthan assessing stabilityof eachcaseas in the Potvinmethod, eachcasewas assigned subjeca tive probability of failure. A descriptionof how these probabilities were assigned discussed is in
Section 8.

the UCS of the rock. Sincethe walls of opensublevel stopes thought be in tension, factorA are to the is takento be 1, Pakalnis Vongpaisal and (1993).
Next, the rock structureorientation factor, or B
accounts for the manner in which discontinuities in-

tersect wallsof thestope. the Discontinuities are that oblique parallelto the stope or wall havemoreof an adverse impacton stabilitythan thosethat impinge at higherangles, Potvin(1988). The gravity adjustment factorC, accounts the for orientation the stopewall. Surfaces of like a stope
back are less stable than a vertical wall due to the in-

Data collectionwas conducted using a scanwindow or sectionrather than using scan lines. Joint orientations the hangingwall in were determined in termsof strikeanddip usingthe right handrule and characterized to spacingand persistence. as Joint roughness joint alteration (Jr), (Jr),joint wall alteration, joint aperturewidth and the presence abor senceof water (Jw) were also assessed. From the initial data sheets,mine data and personnel interviews, casehistories eacharea were compiled. for Stopegeometry, stateof stress, failurehistory,excavation history, subjective probability of failure, modesof failure,support historyand corrective actions were included in the case history. Pertinent
data have been abstracted from these case studies

flueneeof gravity.Dip of the wall is relatedto the factorC by equation Potvin(1988). 3, C=8-6*cosine (dipof thewall) (3)

Thusthe equation themodified for stability number is depicted equation Potvin(1988). by 4,


N'=Q' x AxB xC (4) The stabilitynumber typicallyplottedagainst is the hydraulic radius(S) of the exposed wall andis simply the areaof thewall dividedby theperimeter. Givenby equation Hutchinson Diederichs 5, and (1996).

S= A x B/2A + 2B

(5)

but not all of the collected datahasbeenpresented in thispaper. Potvin (1988) developed modifiedstability the numberN' which has been usedin this study.To calculate N', several factors must be derived. The first of theseis Q'. The original Q or rock mass quality number,Barton et al. (1974) relatesto the block size(RQD/Jn)multipliedby the shear strength of the discontinuities demarcating blocks(Jr/Ja) the multipliedby the activestress (Jw/SRF).The entire equation givenby equation is 1. Q=RQD/Jnx Jr/Jax Jw/SRF (1)

Figure1. Dimensions andB in equation A 5.


5 GEOLOGY

The SRF term,Bartonet al. (1974) represents the amount loosening therockmassasa resultof of in the excavation. useis prevalent tunneling Its in but is setas 1 in asis the factorJw whencalculating Q,' Potvin(1988) whichis represented equation by 2.

Q'=RQD/Jnx Jr/Ja

(2)

To obtaintheparameters calculate valuefor to a Q', RQD wasderived fromthemeanfracture spacing andthejoint setnumber (Jn)wasobtained from thenumber joint sets of measured. parameters The
418

Stability of open stope walls relatesclosely to a combination mineralogy of and structural geology. The mafic mineralogy and structural orientation that havemademiningof platinum andpalladium possible havealsogivenriseto stability problems subin level and long-hole open stopes.These problems mostlyresultfromthe presence weak layers of parallel to the ore zone and out in the hangingwall that intersect joint sets cutting rockmass. the Lying on the northwest flank of the Beartooth uplift,the Stillwater Complex a differentiated is layered marie to ultramarie1opolithic intrusionof late

MODIFIED
1ooo

STABILITY

GRAPH

vine rich character of the rocks in or near the ore

zonealsocanhavea largeinfluenceon dueto alteration the rockmass of causing deposition serof

pentine chlorite and richgouge claysonthejoint


surfaces.

STOPE DESIGN

lo

ao

HYDRAULIC RADIUS (m)


Figure2.Stability graphafterPotvin(1988) andNickson (1992).

Archcan age. Intruded into middle Archcan sedimentaryrocks at about 15 km of depth,its 6000-m of cumulus stratigraphy be subdivided three can into major series. The Basal series contains nicklecoppermineralizationwhile the Ultramafic Series
hosts 14 chromite horizons and the Banded Series

Bulk mining involves two types of stopesat the StillwaterMine. Sub-levelstopes minedbetween are two driftsand long-hole back stopes that are mined from one lower drift. Sub-levelstopesare typically designed with drill/extraction drifts on an approximate 12m vertical spacing,giving a rough on dip spacingof 20m. These stopesusually are drilled fromthetop drift andmucked from thebottomdrift. Ore widthsin sub-levelareasare typically narrow, rangingfrom 1 to 2m but may reachseveralmeters in width. Dips can range from as low as 30 up to vertical,depending the areaof the mine.Figure3 on illustrates typical stopegeometryand support a layout for a sub-leveldesign.

containsthe platinum and palladium bearing J-M


Reef. The J-M Reef, named after the Johns-Manville

Company,lies approximately 400 m abovethe top of theUltramarieSeries, Toddet al. (1982). Originally formed in a sub-horizontal position, the Complexwas tilted slightlyto the north during Cambrian timesandsubsequently moresteeply tilted northward duringthe Eocene LaramideOrogeny. Overall strikeof the J-M Reef at the mine is approximately290 and it dipsfrom 45 to 70 to the north.The dominant discontinuity orientation corresponds this orientation is sub-parallel it. At to or to leasttwo episodes intrusion doleriteor diabase of of maficdikescut the complex rocks. Thesedikestypically strikeNW anddip about35 SE. A significant numberof discontinuity are associated sets with this structural orientation, PageandZientek(1985). Stratigraphy theLower Banded of series begins with a generally monolithic noritelayercalled NoritcZoneI (NZ I) overlain a morelayeredseby quence knownasGabbro ZoneI (GZ I) thatis composed dominantly gabbronorite two pyroxenne of or gabbro, Todd et al. (1982). Troctolite- Anorthosite ZoneI (TAZ I) contains the J-M Reef. TAZ I lies atopGZ I andmarksthe reappearance olivineasa cumulus of phase. is It composed dominantly olivinerichrocksinterof digitated with plagioclase rocks.The J-M Reef rich that is the orebearingunit resides nearthebottomof TAZ I. Severalweak layerscanlie out in the hangingwallof theorezonedepending the sector on of themine,Todd et al. (1982). If present, theycan play animportant role in sub-level stability. The oli419

Figure3. Typical sublevel cablebolt layout.


7 SUBLEVEL CASE DATA

Pertinent data was extracted from 13 case histories

of sub-level openstopes. Eachstopehasa designationof 1 through Table1 presents RQD, joint 13. the set number (Jn),joint roughness number(Jr),joint alteration number (Ja), modified rock mass quality index(Q'), modifiedstability number (N'), hydraulic radius (S), probability failure(POF), anddip of of the hangingwall (Dip). The modified stability number (Eq. 4) wascalculated multiplying N' by Q' (Eq. 2) andA set equalto 1 and B which was taken to be 0.2 to 0.3 for parallelstructure C which is and the cosine of the paneltimes6 subtracted dip from 8 (Eq.3). Hydraulic radius was calculated dividing by the areaof thewall by theperimeter (Eq. 5).

Thesedatashowthe relationship the rock mass of parameters the stabilityand shapefactors each to of stopeas well as probabilityof failure for eachcase.

lishedjointly in 1998 by the Institutionof Civil Engineering and the Faculty and Instituteof Actuaries in theUnitedKingdomis identical concept in All of the data considered here was collected from For thepurposes thispaper,onlythe categories of sub-levelstopeswith an upper and lower access. of low throughcertainwere used.Designations on Althoughextended supportis installedin the great the stabilitygraphshownin figure4 are low (L), (M), high (H), very high (VH), andcertain majorityof sub-levelstopes, effect of installed moderate the support wasnot considered here. C). Assessments probability failurewere done of of subjectively, basedon previousexperience the on This paper focuseson assessing probabilityof sublevelfailure and severalfailuresare cataloged, mine. Acceptable thresholds a non-entrystope for manysublevel panelsare extracted the Stillwater are betweenmoderateand high due to non-exposure at Mine that experience failuresastable4 shows. no of personnel the opensub-level in panels. While the probabilities failure were subjectively of assessed, many factorsplayed into the subjective valuation. Iame : 3Iao!11II//rarame[ers umeve nan gmlwans. xor Ten years of combinedexperience conducting inWorking RQD Jn Jr Ja Q' N' S POFDip % m o stopegroundcontrolaudits at the Stillwater Mine Sub-level 1 86 9 0.5 4.0 1.8 1.8 5.6 VH 60 form the basis for these subjectiveassessments. Sub-level2 89 9 0.5 3.0 2.8 3.1 5 M 67 These audits encompass two major components. Sub-level3 88 9 1.0 4.0 2.6 2.6 4.7 M 60 One, a geomechanical aspectthat considers rock Sub-level 4 64 12 0.5 6.0 0.8 0.5 5 C 30 strength, block size, joint condition,adversestrucSub-level 5 64 9 1.0 4.0 1.5 1.2 5.6 H 50 ture, amountof stress inducedfracturingand degree Sub-level 6 86 9 0.5 6.0 1.5 1.5 6.3 H 60 of looseningof the rock mass. The other, an apSub-level 7 87 12 0.5 6.0 0.9 0.9 5 VH 60 praisal of support and excavationquality rates Sub-level8 99 12 0.5 6.0 0.7 0.6 5 VH 50 groundsupportin termsof properanchorage, plate Sub-level9 99 12 0.5 6.0 1.4 1.2 6.6 C 53 tightness, bolt angle, uniformity, adequacy, adherSub-level 10 86 12 1.0 4.0 1.6 1.9 4 L 70 enceto the support standard, securing weak and of Sub-level 11 88 9 1.0 6.0 2.2 1.6 4.3 M 45 contacts. Drilling accuracy, blasting control wall and Sub-level 12 64 9 0.5 6.0 0.7 0.7 5.6 C 60 regularityalso figure into the assessment. Although Sub-level 13 76 12 1.0 3.0 1.3 1.5 3.8 L 70 the assessments theprobability failurefor these of of sub-level openstopes were subjective, theywere not assigned arbitrarilyand are bolstered manyyears by of experience the mine. on 8 PROBABILITY CRITERION

A classification methoddevisedby Kirsten (1999) assigns quantitative a value for a qualitativeassessmentof probabilityof failure and is shownin table
2.

9 PROPOSED

STABILITY

GRAPH

Table2. Classes Probability occurrence. for of

Qualitative
Evaluation Certain

Quantitati,'e Evaluation
EVERY TIME 1.0

Very High High


Moderate Low

1inten 1 inahundred
1 in a thousand 1 in a 10 thousand

10-' 10 -z
10-3 10-

Verylow Extremely low Practically zero

1ina 100 thousand, 1 inamillion 1inten million

10 ' 10 -6 10 '7

The points in categories low, moderateand of certain probabilitiesof failure show good linear be consistent evaluations operational in of safety in trendsevenwith a smallnumberof datapoints.Data mines in general, system integrity in engineering points that have high or very high probabilityof mechanics, economic viability of venture investfailure show more scatter but still have a defined ments,process reliability in engineering feasibility trend. studiesand of environmental protectionand rehaIn figure5, stability classes developed Potvin by bilitation in mine closures,Kirsten (1999). The 6(1988) andrefined Nickson by (1992) hasbeensupoint scale Risk Assessment Table that was pubperimposed figure Lowprobabilities failure on 4. of
sol Coal in South Africa. It has since been found to
420

The 8-point scale in the table was developed for evaluating operational safetyin the collieties Saof

A minor modification the stabilitygraphis proto posedin this paper.The modifiedstabilitynumber (N') is plottedagainst hydraulic the radius(S) as in the original case. However, rather than assessing stability, each point plotted on the graph was assigned probability failureduringdatacollection. a of Trajectoriesof equal probabilitiesof failure are drawn throughthe points. Resultsof this exercise are shownin figure4. Since many of the stopesare excavatedin sequenced patterns, many of the datapointsare clusteredaround narrowrangeof hydraulic a radii. The data doeshave somevariationin termsof stability
number.

Sublevel Hangingwall Slability Probability Failure of Trajectories


lOO

stability number along the left-hand side, a probability of failure can be read from the matrix at the
intersection of the columns and rows. The shaded

cells denoteacceptable probabilitiesof failure and the interfacebetweenthesetwo regionsin the table can be considered the design line. If the operator cannot determine thenQ' may providea reasonN', able substitute since,judging from table 1, the values for Q' and N' do not differ a great deal. It is hopedthat a more simplifiedparameter than N' can be found to perform this analysis.At the time of writing no parametricanalysishas been performed to explorethe validity of thathypothesis. Table Sub-level 3. probabilit failure of matrix.

o.1

Hydraulic Radius (S)m


4 5 6 7 8

5.0

6.0 C C

7.0 C C C VH
H M

8.0 C C C C
H H

Hydraulic Radius (m) -S


__

0.1
0.4
1.0
2.0

VH C
H
L
L

Figure4. Stabilitygraphwith probability failuretrajectories. of

agreequite well with the stablezoneon the graph. Moderateto high lie within the unsupported transitionzone.High andveryhighprobabilities failure of correspond the stablewith supportzone and the to certainprobabilityof failuresare equivalent the to lower part of the stablewith supportzone and the upper part of supported transitionzone. All of the cases certainprobability failure lie well above of of the cavedline for this rangeof S. At this point, it appears that for any given N the differencein hydraulic radius betweenthe certain probabilityof failureline and the cavedline is approximately 1.5
S.
Sublevel Hangingwall Stability Probability of Failure Trajectories
100

3.0 4.0
C

VH H

M
M

4.0

M M

8.0 L
10 MODES

OF HANGINGWALL

FAILURE

Four modes of hangingwallfailure have been put


forward at this time. Gravitational block fall out,

block topple, block sliding and wall buckling are proposedas the four modes.The first three modes
occur alone or in combination with the other two.

Indirect evidencesuggests that wall bucklingmay take place especiallyin the initial stagesof failure. However, this mode was not directly observedin any of the 13 casehistories. Table 4 illustrates mode of failureand assessed probability failure for the of sub-level hangingwalls.
Table4. Sub-level hangingwall failuremodes. Working Failure Mode Probability Failure of Sub-level 1 Topple Very High
Sub-level 2 No Failure Moderate

0.1

Sub-level 3 Sub-level 4
3 4 5 6 7 8

No Failure Block Fall Out

Moderate Certain

Hydraulic Radius - $ (m)

Figure5. Stabilitygraphwith probability failuretrajectories of and superimposed stabilityzone.

Sub-level 5 Sub-level6 Sub-level 7 Sub-level8


Sub-level 9 Sub-level 10 Sub-level 11

BlockFall Out No Failure Topple/Slide BlockFall Out


Block Fall Out No Failure No Failure

High High Very High Very High


Certain Low Moderate

Table 3 is proposed a tool for mine operators as to use as a quick assessment the probability of of sub-level hangingwall failure. By finding the hydraulicradiusalongthe top row and estimating the
421

Sub-level12
Sub-level 13

Topple
No Failure

Certain
Low

11 CONCLUSION

British Columbia.

Subjectiveassessment probabilityof failure for of sub-levelopenstopehangingwalls be usedin can conjunction with rock massquality parameters and Balkema. hangingwallgeometry predict stopestabilityA to Todd,S.G., Keith, D.W., LeRoy,L.W., Mann, E.L., andIrvine, T.N. 1982.The J-M platinum-palladium of the StillReef framework plotting for probability failuretrajecof water Complex,Montana: Stxatigraphy petrology. I. and tories on a stabilitygraph is in place and shows Economic Geology 1454-1480. 77: promise. Initial data indicatesthat this method agrees quitewell with previous work andprovides a methodthatenhances traditional stabilityanalysis. A prototype matrixis proposed give mine operators to a tool to morereadilyevaluate hangingwall stability. Additionaldatapointswill allow validationof this analyticalprocess. Modes of failure in the stopes play an importantrole in understanding how they can be managed. Furtherwork needsto be donein theseareasand is currentlyin progress. This additional data will flesh out thesegraphicaltools and testtheir validity.
12 ACKNOWLEDGMENTS

Potvin,Y. and Milne, D. 1992. Empiricalcablebolt support design.In RockSupport miningand underground in construction, proc. lnt. syrup. rock support,Sudbury, On (eds P.K. Kaiser and D.R. McCreath), 269-275. Rotterdam:

The authorsgratefully acknowledgethe management of StillwaterMining Companyfor the opportunity to undertake this studyand for permission to publishthis paper.They alsowish to express gratitude to the mine staff for valuable advice and infor-

mationin thecompilation the casestudies. of


REFERENCES

Barton,N.R., Lien, R. and Lunde,J. 1974. Engineering classification of rock masses the designof tunnel support. for RockMech. 6(4), 189-239 Hutchinson, D.J. and Diederichs, M.S. 1996. Cablebolting in underground mines. Vancouver: Bitech Institution Civil Engineering theFacultyandInstitute of and of Actuaries the UnitedKingdom.1998. Riskanalysis in and management projects. for Kirsten, H.A.D. 1999. Evaluationof risk as decisionmaking criterion. Workshop at SRKConsulting Denver,Co., held in September - 21, 1999. 20 Mathews, K.E., Hoek, E., Wyllie, D.C. and Stewart, S.B.V. 1981.Prediction stableexcavations miningat depthof for below 1000 metresin hard rock. CANMET ReportDSS Serial No. OSQ80-00081, DSS File No. 17SQ.23440-09020. Ottawa:Dept. Energy,MinesandResources. Nickson,S.D. 1992.Cablesupport guidelines underground for hardrockmineoperations. MASc. thesis. Dept. Mining and Mineral Processing, Universityof BritishColumbia. Page, N.J. and Zientek, M.L. 1985. Geologic and structural settingof the Stillwater Complex.In G.K. Czamanske & M.L. Zientek (eds), TheStillwaterComplex, Montana:geologyandguide.SpecialPublication 1-9. MontanaBu92, reauof Mines andGeology Pakalnis, and Vongpaisal, 1993. Mine designan empiriR. S. cal approach. Proceedings InternationalCongress Mine on Design,Kingston, Canada, 455-467. Rotterdam: Balkema. Potvin,Y. 1988.Empirical openstope design Canada.Ph.D. in thesis, Dept. Mining andMineral Processing, Universityof
422

You might also like