DOCTORAL THES I SLule University of TechnologyDepartment of Civil, Mining and Environmental EngineeringDivision of Rock Mechanics:oo::oo|issx: +o:-+|isrx: i+u-i+ -- o: oo -- sr:oo::ooBehaviour of Blast-Induced Damaged Zone Around Underground Excavations in Hard Rock MassUniversitetstryckeriet, LuleDavid SaiangDavid Saiang Behaviour of Blast-Induced Damaged Zone Around Underground Excavations in Hard Rock Mass :oo::ooBehaviourofBlastInducedDamagedZoneAroundUndergroundExcavationsinHardRockMass

byDavidSaiangSubmittedinFulfilmentoftheThesisRequirementfortheDegreeofDoctorofPhilosophyinRockMechanicsandRockEngineeringDivisionofMiningandGeotechnicalEngineeringDepartmentofCivil,MiningandEnvironmentalEngineering,LuleUniversityofTechnology,Lule,SwedenNovember2008DavidSaiang2008iiTomymotherandsister,GetrudandPricillaSaiang,whopassedonduringthecourseofmyPhDstudy.iiiivACKNOWLEDGEMENTThis thesis would not have been possible without financial support,supervision and encouragement from various individuals and organisations.Banveket (the Swedish Rail Road Administration) is sincerely acknowledgedfor being the major financial sponsor of the research work published in thisthesis.OtherfinancialsponsorsduringthecourseofthestudyincludeLKAB,RockTechCentreABandLuleUniversityofTechnology.Professor Erling Nordlund is greatly acknowledged for providing thesupervisionandforhavingtheconfidenceinmeduringtheresearchwork.Healso ensured that this thesis is of acceptable quality. His family isacknowledgedforprovidingfriendship.IamverygratefultomyassistantsupervisorDr.PingZhang.Hisconstructiveinputs and thorough reading of this thesis is sincerely acknowledged. MycolleaguesandfriendsattheDivisionofMiningandGeotechnicalEngineeringare thanked for friendship and support. Tomas Villegas (Tech. Lic.) was avaluable asset for my numerical modelling work. Lars Malmgren (PhD) wasmypartnerintheearlystagesofmyresearch.ThefolksatTestLabaregreatlyacknowledged for having the confidence to let me use the laboratoryequipmentwithminimalsupervision.SpecialthankstoSveBeFopersonnelforletting me make a summary of the SveBeFo publications on blastinduceddamage. Professor Finn Ouchterlony provoked my thoughts whenever hevisited. Ms. Christine Saiang is acknowledged for checking the englishgrammar.ThePapuaNewGuineaUniversityofTechnologyprovidedfinancialsupportformyfamilyuntiltheendof2007.Forthattheyaregreatlyacknowledged.IamalsoverygratefulformyrelativesandfriendsinPapuaNewGuineawhogavemeencouragementandmoralsupport.Finally, I am greatly indebted to my wife, Christine, and two daughters,Hannah and Ruth. They have borne with me throughout the course of thisstudy and had so much patience with my melancholy temperaments and vunhappy life! I am grateful for the closure of this chapter and am nowlookingforwardtoacalmerone.In all and above all, I thank the Lord all Mighty in whom comes all thetreasurersofknowledgeandwisdom(Col.3:2).DavidSaiangOctober2008,Lule,Sweden viPREFACEThis thesis is written in fulfilment of the thesis requirement for the degree ofDoctorPhilosophyinRockMechanicsandRockEngineering.TheDivisionofMiningandGeotechnicalEngineeringwithintheDepartmentofCivil,Miningand Environmental Engineering at Lule University of Technology providedtheenvironmentandfacilityfortheproductionofthethesis.ThethesisitselfistheresultoftheworksponsoredbytheBanverkettostudythebehaviourandinfluenceofblastinduceddamagedrockaroundanundergroundexcavation.Itispresentedintwopartsas:I. Extended summary covers; (i) introduction, background and researchdirection,(ii)literaturereview,(iii)investigation,analysisandresults,and(iv)generalconclusionsandsuggestionsforfuturework.II. Appended papers presents the peerreviewed papers, which covervariousresearchquestionswithinthescopeofthisthesis.Itisanticipatedthattheworkpresentedinthisthesiswillcontributetowardstheunderstandingofthemechanicalbehaviouroftheblastinduceddamagedzonearoundundergroundexcavationsinhardrockmasses.Blastinduced damaged rock around an underground excavation is animportantconcernformanyrocktunneloperators,includingBanverket.Itisagenerallyheldbeliefthatthepresenceoftheblastinduceddamagedzonecanaffectthestabilityandperformanceofanundergroundexcavation.However,it is still unclear how it does that, since many of the conclusions about itseffectsare based onhuman judgementsand intuitions. Even if it isdoes thenone of the many questions that need to be answered is how significant itsinfluence is to the behaviour of the excavation. Unless we know more aboutthebehaviouroftheblastinduceddamagedzonewewillstillneedguidelinesforblastdamagecontrol,eveniftheseguidelinescanbeasourceofadditionalcosts. It was felt during the course of the work presented in this thesis that,viimuch still needs to be done before any decision can be taken about when,where and how, blast damage control guidelines can be applied to optimisetheconstructionofatunnel.DavidSaiangOctober2008,Lule,SwedenviiiABSTRACTThepresenceofblastinduceddamagedzonearoundexcavationshasbeenanimportant concern in rock construction. It is generally believed that thepresence of this zone can pose problems related to stability and seepage andconsequently impair the performance and functionality of the excavation. Infact, the immediate consequences of this zone are usually conceived in termsofsafetyandcost.Hence,someorganizationshaveputinplaceguidelinesforcontrolling the amount of damage induced by blasting. Since shotcrete orsprayed concrete is a widely used surface rock support, its performancedependsprimarilyonthecompetenceofthedamagedzone.However,insomeinstancestheuseofshotcretemaybeunnecessary,butthelackofknowledgeofthecompetencyofthedamagezonemeans,itisbetternottotakechances.Itisthereforenecessarytoincreasetheknowledgeandtheunderstandingofthecompetency and behaviour of the damaged zone in order to predict theperformanceandfunctionalityofarocktunnel.To gain an understanding of the damaged or disturbed zone in general,significanteffortshavebeenmadeoverthelastfewdecadesinabroaderarea;the excavation disturbed zone. These efforts mainly focused on thecharacterizationandclassificationofthedamagedzone.Quantificationofthiszone has also been done in terms of mechanical, hydraulic and physicalparameters, particularly to delineate the extent of the damaged zone. Thecharacterization, classification and quantification of damaged zone werepurpose specific and therefore, the definitions for damage zone are differentandvarying.Inthisthesisthedamagedzoneisdefinedasthezonewheretherockhasbeensignificantlydamagedsuchthatthemechanicalpropertieshavebeen affected and that these changes are measurable by any state of the artmeasurement techniques. This definition also applies to the blastinduceddamagedzone.ixTobeabletoassessthesignificanceoftheblastinduceddamagedzoneanditsinfluence on the performance of an excavation, the mechanical behaviour ofthis zone must be understood. This thesis is therefore aimed in that directionandthustheobjective.Severalissueswereinvestigatedincluding;effectsandconsequencesofblastinduceddamagedzone,mostlikelyfailuremechanisms,mechanical parameter sensitivity and their impact on the behaviour of thedamaged zone, numerical modelling approach for damaged zone andindicators for failure from a continuum model, etc. A literature review andindustrial questionnaire gave the direction for the investigations. Field andnumerical methods were employed in the investigations. The results of theseinvestigations are published in a series of papers that make up this thesis. Inbrief,themainresultsandconclusionscanbesummarizedasfollows:- Theblastinduceddamagedzonehasbeenlargelydefinedintermsofitsextent. A definition based on its inherent competency parameters,particularlystrengthandstiffness,isstilllacking.- Theblastinduceddamagedzonethicknessvariesinmostpracticalcasesbetween 0.1 and 1.0 m, with an average ranging from 0.3 m to 0.5 mdepending on whether perimeter blasting techniques are used or not.The reduction in the Youngs modulus varies anywhere between 10 to90 % of the undamaged rock value. In a field investigation reported inthisthesistheYoungsmodulusofthedamagedrockwasfoundtovarybetween 50 and 90 % of the value for he undamaged rock mass. Thethicknessofthedamagedzonewasbetween0.5and1.0m.- From the numerical study, the presence of the blastinduced damagedzone did affect the behaviour of the stability quantities, namely;deformation and induced boundary stresses. However, it cannot beconcludediftheeffectsaresignificantenoughtocauseproblemsaroundanundergroundexcavationinthehardrockmasstypestudied.- Theinherentpropertiesofthedamagedzonethataffecteditsbehaviour,identified in order of their significance are; the deformation modulus,xfollowed by tensile strength and compressive strength. External factorssuch as the state of the insitu stresses are also seen to significantlyinfluencethebehaviourofthedamagedzone.- The numerical studies showed that the main failure mechanism withinthe damaged zone at shallow excavations is tension, while in deepexcavationsitisshear.Thetensilemechanismsaredominantatshallowdepth and mainly occur in the walls. In deep excavations shearmechanisms are dominant and occur mainly in the roof and floor. Theshiftinthedominanceofthemechanismsfromthetension(inwalls)toshear(inroofandfloor)occursatabout100mdepth.- The modelling of the damaged zone using the coupled continuumdiscontinuum method showed that, the presence of the blastinduceddamagedzoneincreasesthedepthoffailureby20to30%.- Thestudyontheshotcreterockinterfaceshowedthatthebondstrengthof the interface is important for the shear strength. Average bondstrength of 0.5 MPa was determined for interfaces with surfacesroughnesswithJRCvaluesof13,and1.4MPaforthosewithJRCvaluesof 913. These values were determined for low normal load conditions( 0 . 1 cio Eq.2.7where is the deformation modulus of the rock mass, D is a disturbancefactor, and GSI is the Geological Strength Index. A chart was provided byHoek et al. (2002) as a guide for estimating the disturbance factor, based onfacial expression of the excavation wall. The reduction in modulus is appliedglobally.Thatis,thereductioninthemodulusisnotfinitetoacertainextent.rmESaiang (2008b) pointed out that the maximum reduction in the deformationmodulusobtainedthroughEq.2.7is50%.However,fieldinvestigationsshowthat the modulus of the damaged rock can be much lower than 50% of itsundisturbed value (see Saiang, 2008b). Hoek and Diederichs (2006) presentedan improvement to Eq. 2.7, which accounts for reductions below the 50%mark,as:( ) ( ) |.|

\|++ = + 11 / 15 6012 / 102 . 0GSI Di rmeDE E Eq.2.8iftheintactrockmodulus isknownorreliablyestimated.Otherwise,ifonlythe GSI or RMR (Rock Mass Rating) data are available then the modulus canbeestimatedfrom:iE34( ) ( ) |.|

\|+= + 11 / 25 7512 / 1000 100GSI DrmeDE Eq.2.9Attemptshavealsobeendonewithothermethodssuchas;MRMR(Laubscherand Taylor, 1990), Q (Barton, 2007) and RMi (Palmstrm, 1996), to take intoconsiderationthereductioninrockmassmodulusduetoexcavationdamages.The most widely used empirical method for estimating the compressivestrengthoftherockmassistheonebyHoekandBrown(1980).Thisrelationisgivenasafunctionoftheminorprincipalstressasfollows:23 3 1 ci ci ss m o o o o o + + = Eq.2.10wheres 1o is the major principal stress at failure,3o is the minor principalstress,cio istheuniaxialcompressivestrengthoftheintactrockmaterial,andmandsarerockmaterialconstants.Since the majority of the rock mass failure analyses are based on the MohrCoulombconstitutivemodel,whichrequirefrictionangle(|)andcohesion(c)as the main inputs, Bray (Hoek, 1983) derived a method for estimating theequivalentvaluesfortheseparametersfromtheHoekBrownfailureenvelope.Theserelationsaregivenas:( ) ( ) | |( )( )( ) ( ) |( )( )|a am s ama am s m a s acan m ban b n b ci+ ++ ++ ++ + +=2 16 12 11 2 11313 3oo o o Eq.2.11and ( )( )( ) ( ) (((

+ + + ++=1313 16 2 1 26sinan b ban b bm s am a am s amoo| Eq.2.1235wherecinooomax 33 = , a, mb, s are HoekBrown rock constants. The equivalentuniaxialcompressivestrengthoftherockmassisthusgivenby:||osin 1cos 2=ccm Eq.2.13This method of estimating the equivalent cohesion and friction from HoekBrown failure envelope has been implemented in RocLab (Rocscience, 2002)andRocData(Rocscience,2005).2.3.3 FailureModelforRockMassesThree failure models are commonly used for rock masses, see Figure 2.11.Hoek&Brown(1997)gavethefollowingdescriptionforrockmassesthatsuitthemodels.Elasticbrittleplastic The rock mass is of very high quality, with intactstrengthexceeding150MPa,GSIof75andmiof25orhigher.Strainsoftening The rock mass is of average quality, with average intactstrengtharound80MPa,GSIof50andmiof12.Elasticperfectly plastic The rock mass is of very poor quality, with averageintactstrengtharound20MPa,GSIof30andmiof8.In FLAC for example there exist three constitutive models that are applicableforthefailuremodelslistedabove,forhardrockmasses.Theseare;(i)HoekBrown, recently implemented in FLAC version 5.00 (Itasca, 2005), which isapplicable for the elasticbrittleplastic model, (ii) MohrCoulomb StrainSoftening,applicableforthestrainsofteningmodel,and(iii)theconventionalMohrCoulomb,whichisapplicablefortheelasticperfectlyplasticmodel.36strain strain strainstress stress stress elastic-brittle strain-softening elastic-plastic Figure2.11:Yieldbehavioursgenerallyassumedforrockmasses,(a)elasticbrittleplastic(b)elasticstrainsofteningand(c)elasticperfectlyplastic.2.3.4 FailureCriteriaforRockMassesTwo widely used failure criteria for rock masses are the MohrCoulomb andHoekBrown. The MohrCoulomb failure criterion is expressed in terms ofshear and normal stresses (t andno ), and HoekBrown in terms of the majorand minor principal stresses (1o and3o ). Although the MohrCoulombcriterionisusuallyexpressedintermsoft andno as | o t tannc + = Eq.2.14itcanalsobeexpressedintermsofprincipalstressesas(e.g.BradyandBrown,1993)( )( ) ||||o osin 1cos 2sin 1sin 13 1++=c Eq.2.15According to the two criteria the assumptions for failure mechanisms, that isshearandtensilefailures,canbesummarizedasfollows:Fortensilefailuretooccurto o s3 and 01 s o . Eq.2.16Forcompressiveinducedshearfailuretooccurto o >3 ands 1 1 o o > . Eq.2.1737whereto is the rock mass tensile strength ands 1o is the major principalstress at failure. The above assumptions are illustrated graphically in Figure2.12belowintermsoftheprincipalstresses.Eventhoughthetensilestrengthoftherockmassisaccountedforinthetwocriteriatheprimaryassumptionisthatfailureoccursprimarilyasacompressioninducedshearfailure.tShear failure Tensile failure o to Figure2.12:FailureassumptionsaccordingtoMohrCoulombandHoekBrown.382.4 MODELLINGCONSIDERATIONSFORBLASTINDUCEDDAMAGEDZONE2.4.1 PhysicalCharacteristicsIn order to understand the behaviour of the damaged zone, the physicalcharacteristics of the damagedzonemustalso be understood. Thisis becausecharacteristics such as fracture patterns can influence the behaviour of thedamagedzone.Thisunderstandingisalsonecessaryfornumericalmodellingtasks.Thissectionthereforedescribesthebasiccharacteristicsofthedamagedzone and factors that have the potential to influence its behaviour.Understanding of these factors and characteristics will assist in makingeducated judgements on the strength parameters, failure mechanisms andmechanical behaviour. This is an important aspect of the numerical analysesworkpresentedinthisthesis.Figure 2.13 shows an example of how the blastinduced cracks around atunnel boundary would look like. The cracks created by blasting are: (i)macroscopic and microscopic fractures, with different shapes and sizes, (ii)they are radial, highly anisotropic and nonpersistent cracks, (iii) fracturedistribution and its nature (including extent and intersection of the cracks)havespatial characteristics, which depend on thefactors presented in Section2.2. Figure 2.14 is the theoretical representation of cracking around a blasthole,whichillustratespoints(i)and(ii)statedabove.Obviously,thekinematicsoftheblastinducedblockswillbedifferenttothoseformed by natural geological structures. A model must therefore beconstructed in such a way that it represents blast induced fractures so that itcapturesthebehaviourcorrectly.39Undamaged rock Damaged rock Figure2.13: Rockmassconditionaroundatunnelboundaryexcavatedbydrillandblast.Thedamagedzonecomprisesofdiscontinuousfracturesofmicroscopictomacroscopicsizeswithcomplexfracturepatternsduetoradialcracks,seeembeddedfigure(adoptedfromOlssonandBergqvist,1995).Explosive chargeFragment formation zoneFracture zoneBlastholeCrushingzoneFigure2.14: Radialcrackpatternsnormallyobservedaroundablasthole(Whittakeretal.,1992).402.4.2 StrengthCharacteristicsTomakeareasonablejudgementonthevaluesforthestrengthparameters,c,| andcmo (at low confining stress), ideas have been drawn from the work offor example; Cai et al. (2004), Diederichs (1999), Hajiabdolmajid et al. (2003;2002), Martin and Chandler (1994), Martin et al. (1999), Pelli et al. (1991) andStacey (1981). Some of the above mentioned authors suggest that, at lowconfiningstresstheyieldingprocessforbrittlerocksisgovernedbyacohesionweakening friction mobilisation phenomenon (see Figures 2.15). Under lowconfining stress conditions the most likely failure mechanisms are of tensilenature leading to spalling, axial splitting and direct tension (see Figure 2.16).Stacey(1981)forexamplestatesthattheinsitustrengtharoundexcavationsisdeterminedbytheextensionstraincapacityoftherock.Undersuchconditionsfailure can occur at stress levels in which1o (major principal stress) is lowerthans 1o (major principal stress at failure), as long as3o (minor principalstress)overcomesthetensilestrengthoftherockmass.Within the damaged zone there is a vast amount of rock bridges. DiederichsandKaiser(1999)havedemonstratedthat1%oftherockbridgeswithin1m2can produce cohesive strength equal to the strength of one cable bolt, underlow confining stress conditions. Robertson (1973) showed that rock bridgeswithin the rock mass can increase its inherent strength, as rupture must firstoccur through the intact rock before failure develops. This indicates thatcohesion(c)andtensilestrength(to )arethemostimportantparametersunderlow confining stress conditions ( MPa 0 . 1max 3 < o ). ISRM (Brown, 1981)suggested the following equation to estimate the shear strength of the rockbridges, as a function of the compressive and tensile strengths of the intactrock(co andto ).t cc o ot =21 Eq.2.18where is the shear strength and also the cohesive strength of the rockbridge.tc41Figure2.15:Cohesionweakeningfrictionmobilisationfailureprocess(Hajiabdolmajidetal.,2003)Figure2.16:Failurecharacteristicsandmechanismsatdifferentconfiningstresslevels(Diederichs,1999).422.4.3 DeformationModulusThe approach by Hoek et al. (Hoek et al., 2002) assumes a global disturbanceand therefore the deformation modulus is reduced for an entire model. Thestrength is also reduced in the entire model with this approach. On the otherhand,damagearoundatunnelisusuallylimitedtoacertainextent.Hence,thechangeineitherthemodulusorthestrengthislimitedtothisextent.Variationin the deformation modulus around a tunnel can also be verified throughseismic measurements (see Figure 2.17). Hence, one approach used in thisthesis is illustrated in Figure 2.18, which follows the theoretical behaviourillustratedinFigure1.3inChapter1.Therockmassmodulusislinearlyvariedfromthetunnelboundaryuptotheexpecteddepthofthedamagedzone(seeSaiangandNordlund,2008a).Thedeformationmodulusofthedamagedzoneis calculated using either Eq. 2.4 or 2.5. It is then assumed that this valueoccursatthetunnelboundaryandlinearlyvariesasillustratedinFigure2.18.300040005000600070000 1 2 3 4Depth (m)vP (m/s)Test area 1Test area 3Test area 5Figure2.17: ChangesinPwavevelocityobservedfromminedriftdamageinvestigationstudies(Malmgrenetal.,2007).Theoreticallythevariationinthemodulusshouldfollowthesametrend.43Deformation modulus EmEDDistance from tunnel boundaryDamaged zoneUndamaged zone Figure2.18: Therockmassdeformationmodulusislinearlyvariedfromtheexcavationboundarytothedamagedundamagedrockboundary. and aredamagedandundamagedrockdeformationmodulirespectively.DEmE44CHAPTER3:INVESTIGATIONS,ANALYSESANDRESULTSThe actual investigations, analyses and results are presented in the appended paperswhichconstitutePartIIofthisthesis.Nonetheless,thischapterpresentsasummaryofthese papers which include; the objectives of each paper, research methodology, themainresultsandthemainconclusionsreached.Linksbetweeneachpaperareoutlined.TheappendedpapersinPartIIprovidetheirrespectivedetails.453.1 EXCAVATIONDAMAGEDZONEINVESTIGATION3.1.1 SummaryofPaperA(Malmgrenetal.,2007)BackgroundandinvestigationmethodIn Paper A an investigation of the excavation disturbed zone at LKABsKiirunaavara Mine in northern Sweden is presented. This investigation wasmainlyaimedatquantifyingtheextentofthedisturbedordamagedzoneandthe deformation modulus of this zone. The terms excavation disturbed anddamaged zone were used interchangeably in this paper because thedifferences between blastinduced and stressinduced damages anddisturbanceswereundefined.Studies have also been conducted earlier by Nyberg et al. (2000) and Nybergand Fjellborg (2002) in the drifts and pillars of the same mine, to identify theblastinduced damaged zone by blast vibration measurements and sawcutexamination. Damaged zone thicknesses of up to 0.4 m were determined inthese studies. However, the measurement of the deformation modulus wasnotdone.Twomaingeophysicalmethodswereusedintheinvestigation,thecrossholeseismicsandtheSpectralAnalysisofSurfaceWaves(SASW).BoreholeopticalimagingwasalsoconductedviaBIPS(BoreholeImageProcessingSystem)toprovide information about the fracture intensity. The basic configurations ofthefieldtestsareshowninFigures3.1to3.3.Thetestswereconductedinthefootwalldriftsandinthecrosscutsintheorebody.463 m4 m5 mSource Receivers Drift wallFigure3.1: Setupofthecrossholeseismicexperiment.Impulsive noise source - a hammer Receivers (Accelerometers)Wall d1 d2Recording unit PC Minimate PlusTMFigure3.2: SetupoftheSurfaceWaveSpectralAnalysis(SAWS)experiment.47Recording unit and image processing unit Borehole Mirror Battery Bar Lighting TV-camera PCFigure3.3: SetupoftheBoreholeImageProcessingSystem(BIPS).ResultsandConclusionsTable 3.1 shows the result of the measurements. The extent of the excavationdamaged zone was estimated to be approximately 0.5 m and 1.0 m for C45mmandC48mmblastholesrespectively,independentoftherocktype.TheseextentsconfirmedthosereportedbyNybergandFjellborg(2002).TheYoungsmodulusofthedamagedrockmass( )wasreducedby1024%withC45mm blastholes and by approximately 50% with C48 mm holes. These werecalculated using Eq. 2.5, given in Section 2.1.4. In essence the extent of thedamagewasdependentontheblastholediameter.EDZETheresultsfromthegeophysicaltestsweremorecomplexthanexpected.Thiswas reflected by the large variation in the Youngs modulus values for thedamaged rock, measured by the two geophysical methods. Information fromBIPS was clear and concise in most of the measurements. Besides providingphotographic images the BIPS also provided fracture density data, whichwhenplotted,couldshowthetrendofthefracturedensity,thustheextentofthedamagedzone(seeFigure3.4forexample).48Table3.1: MeasuredPwavevelocity,damagedzonedepthandYoungsmodulusofthedamagedrockasapercentageoftheundamagedrockmodulus.AveragePwavevelocity(m/s) EDZETestareaUndamagedrockDamagedrockEDZdepth(m)aspartof undamagedEContourholeC45mmNo.1:Wasterock 5590 5300 ~0.5 90%No.2:Ironore 6220 5400 ~0.5 76%No.3:Wasterock 5690 5200 ~0.5 84%ContourholeC48mmNo.4:Ironore 6020 4300 ~1.0 51%No.5:Wasterock 5800 4200 ~1.0 52%3.03.54.04.55.05.56.06.50.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0Depth (m)P-wave velocity (km/s)Test area 1Test area 2Test area 3(a) 01234560.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0Depth (m)Fracture density (fractures/m) Test area 1Test area 2(b) Figure3.4: (a) Pwave velocity crosshole seismics tests and (b) fracture intensity data, forthe C45 mm contour holes. Both plots show an excavation damaged zonethicknessofabout0.5m.493.2 NUMERICALANALYSESOFTHEBLASTINDUCEDDAMAGEDZONE3.2.1 ModellingRemarksThereisnospecificmethodorapproachtomodelthedamagedzone.Boththecontinuum and discontinuum methods have been used (see Section 2.3.1 inChapter 2). The blastinduced damaged zone itself is complex in many waysas described in Section 2.4 in Chapter 2. A proper understanding of thecharacteristics of the damaged zone, the most likely failure mechanisms andprocesses and the conditions they are likely to occur under is essential forchoosing an appropriate modelling approach and constitutive model.Moreover,thisunderstandingisalsoneededformakingsoundandeducatedjudgementsonthenumericalresultsandtoexplaincertainobservations.Asimplewaytobeginthestudyofsuchacomplexproblemistoapproachitinapracticalway,suchasaconsultantwoulddoifgivensuchatask.Thatisbyusingthecommonlyusedtoolsandapproaches.PaperBinthisthesiswasapproachedinthismannerandalsosetthegroundworkforfurthernumericalinvestigations.Byprogressivelyimprovingthemodels,basedonlessonslearntfrompreviousmodels,moreadvancedmodelsweredevelopedtostudythebehaviourofthedamaged zone. Paper F for example presents an advanced modellingapproachwheretwocomputercodes(FLACandPFC2D)werecoupledtostudythe damaged zone behaviour. In essence the numerical study papers (i.e.PapersBtoF)arepresentedinalogicalsequence,whereeachpaperaddressesaproblemidentifiedbyitspredecessor,andsoforth.Themodellingapproachis also improved and advanced with each paper. Papers B and C specificallydeal with shallow tunnels. Because of the difficulties encountered with theHoekBrownGSI empirical method for estimating the inputs for thecontinuummodels,analternativemethodwasdeveloped,whichispresented50inPaperD.PaperEdealswithdeepexcavations(depthsgreaterthan100m).Knowledge gained from Papers B to D was used for the modelling task inPaperE.PaperF,presentstheanalysesoftheblastinduceddamagedzonebycontinuumandcoupledcontinuumdiscontinuummethods.The numerical codes used inthe numerical studyof the damagedzone were;FLAC (Itasca, 2002; Itasca, 2005), PFC2D (Itasca, 2004; Itasca, 2008) and Phase2(Rocscience, 2007). Phase2 was used mostly in Paper F, because of thesimplicitytoimplementradialcracksinthemodel.3.2.2 SummaryofPaperB(SaiangandNordlund,2008a)BackgroundandmethodNumericalanalysiswasperformedtostudythebehaviouroftheblastinduceddamaged rock around an underground excavation at shallow depth (rockcoverlessthanorequalto20m).ByvaryingthestrengthandstiffnessoftheBlastInduced Damaged Zone (BIDZ) and other relevant parameters, theresponse of the near field rock mass was analysed in terms of the effects oninduced boundary stresses and ground deformation. Table 3.2 shows thevariousscenariossimulatedandTable3.3showsthevariabledatasetforthesescenarios. Table 3.4 shows the inputs for the base case model. The base casemodelislocated10mbelowthegroundsurface.Table3.2: Modelscenarios.Case DescriptionCase0: BasecaseCase1: Undamaged(noBIDZ)Case2: VaryingYoungsmodulusoftheBIDZCase3: VaryingcompressivestrengthoftheBIDZCase4: VaryingtensilestrengthoftheBIDZCase5: VaryingthicknessoftheBIDZCase6: VaryingoverburdenthicknessCase7: Varyinginsitustresses51Table3.3: Variableparameterdataset.Case Low Basecase HighCase0: Case1: Case2: 8.5GPa 11.8GPa 17.8GPaCase3: 8.8MPa 12.7MPa 26.8MPaCase4: 0MPa 0.2MPa 0.4MPaCase5: 0.1m 0.5m 1.0mCase6: 2m 5m 20mCase7: Lowinsitustresses HighinsitustressesTable3.4: Inputsforthebasecasemodel.ValueParameterUndamagedrockmass DamagedrockmassYoungsmodulus,E(GPa) 17.8 12.4Poisonsratio,v 0.25 0.25Cohesion,c(MPa) 2.6 1.4Frictionangle,|( ) 68 65oTensilestrength, (MPa)to 0.4 0.2A large singletrack railway tunnel geometry was selected for the analysis.Figures3.5and3.6showthetunnelandthemodelgeometries,respectively.Inthemodelfinegridswereusedwithinandnearthedamagedzoneandcoarsegridselsewhere.Ablastinduceddamagezoneisaddedtothemodel.Forthepurposeofsensitivityanalysesthethicknessofthedamagedzonewasvariedbetween0.1and1.0m.Therockmasswithanintactcompressivestrengthof250MPaandGSIof60,which is considered as good or above average quality (in Sweden), was usedas the basis for deriving the inputs for the computer models. A systematicapproach was used in estimating the inputs from the HoekBrownGSIempirical method. The deformation modulus was varied linearly in a stepwise manner from the tunnel boundary into the rock mass as illustrated inFigure3.7.52(a)9.0 m9.7 m0.32 mR4.5m (b)9.0 mR4.5m9.54 mFigure3.5: (a)CrosssectionofalargesingletrackrailwaytunnelaccordingtoBanverket(2002),(b)equivalenttunnelgeometryforthetunneldimensionshownin(a).Fine gridsBlast-Induced Damaged ZoneCoarse grids80.0 m80.0 mFigure3.6: Modelgeometry.Finegridsareusednearthedamagedzoneandcoarsegridselsewhere.53ExcavationDamaged rockUndamaged rockEDDeformation ModulusEmFigure3.7: Approachusedinestimatingtherockmassdeformationmodulusasafunctionofdistance.Thedeformationmodulusatthetunnelboundary( )isfirstestimatedandthenlinearlyincreasedstepwiseuptothedamagedundamagedzoneboundarywhereitreachesitsbackgroundvalue.DEThe insitu stresses used in the models were those reported by Stephansson(1993),whicharebasedonhydraulicfracturingmeasurements. Eq.3.1 gzV o = Eq.3.2 zH04 . 0 8 . 2 + = o Eq.3.3 zh024 . 0 2 . 2 + = o is the vertical stress, and whereVohoHo are the maximum and minimumhorizontal stress respectively, is the rock density, g is the gravity and z isthe depth. In this paper (i.e. Paper B)Ho is perpendicular to the tunnel axisand isinthedirectionofthetunnel.ho54ResultsandconclusionsNumericalmeasurementsweremadeatthreelocationsaroundtheexcavationas shown in Figure 3.8. The displacement and differential stress magnitudesweremeasuredatthesepointsduringthesimulationforthescenariosshownin Tables 3.2 and 3.3. Figures 3.9 and 3.10 show the percentage variation indeformationanddifferentialstressmagnitudesatthesepoints.Thepercentagevariationsarecalculatedasthemagnitudeofchangefromthebasecasemodel.Table 3.5 shows the sensitivity classification of these parameters. There is nospecific rule for this classification, except that it isbased on the magnitude ofchangefromthebasecasemodel.MeasurementsoftheinducedstressesatPointArevealedthat,therockmassabove the tunnel roof behaves like a beam and thereby concentrating highstresses.Thiswasobservedfordepthslessthan10m.Fordepthsgreaterthan10 m it behaves like an ordinary rock mass, where the induced stressesincreaseswithdepthasexpected,untiltherockyieldsincompression.Figure3.11 shows this phenomenon. Coincidentally the 10 m depth for base casemodelisalsothetransitionpointforthisphenomenon.CABDamaged ZoneFigure3.8: Numericalmeasurementpointsaroundthetunnelwherethedifferentialstresses(ro ou )anddisplacementmagnitudeswererecorded.550%10%20%30%40%50%60%VaryingstiffnessVarying compressive strengthVarying tensile strengthVaryingoverburdenVaryingBIDZ thicknessVaring in-situ stressesVariation in (ou-or), %Variation at Point AVariation at Point BFigure3.9: PercentagevariationinthedifferentialstressatPointsAandBforthescenariostested.0%10%20%30%40%50%60%Varying stiffnessVaryingcompressivestrengthVaryingtensilestrengthVarying overburdenVaryingBIDZ thicknessVarying in-situ stressVariation in wall displacement, %Figure3.10:PercentagevariationindisplacementmagnitudesatPointCforthescenariostested.56Table3.5: Classificationofparametersensitivity,where:>20%variation=high,1020%variation=moderate,and o 03 s oand - Type(III):compression,i.e. 03 > os 1 1 o o >Note that, the stress conditions above are compared to the calculatedcompressive and tensile strengths of the rock mass (i.e. ands 1oto ). It wasobservedinthemodelsthat,thesestressconditionswerethemainsourcesforfailure around the tunnel. Figure 3.12 shows a pictorial representation of thestress fields and their conditions of occurrence on a MohrCoulomb strengthenvelope.Two main unstableregionswere identified onthe basis of the stressanalysesdescribed above. These are shown in Figure 3.13. Region A is subjected toType(I)stressconditionwherefailurewilloccurinbiaxialextension.RegionBissubjectedtoType(II)wherefailurewilloccurintensionundercompressionandextensioninabiaxialstressfield.PotentialforcompressivefailureduetoType(III)stressscenariowaslesssignificant.59o3(o1o3)/2o1 (I)Biaxial extension. Failure will occur in tensiono1s 0o3 s oto1 > 0o3 s 0 (III)Compression.Failure will occur in shearo1 > o1s(I)(II)(III)o1 o1 o3 o3 otFailureenvelope (II)Compression and extension. Failure will occur in tensiono3 > 0Figure3.12:Potentialfailuretypesidentifiedfortheshallowexcavations.ABFigure3.13:Regionsofpotentialfailureintensionidentifiedfromelasticanalyses.RegionAis in biaxial extension ( and 01 s oto o s3 ), while region B is in tension andcompression( and ). 01 > o 03 s o60Intheplasticanalysesitwasobservedthat,theresultsfromusingtheoriginalinputs (see Table 3.6) gave large regional yielding, which was not consistentwith the predictions from the elastic models. An examination of theincrementalvolumetricstrainshowedvolumetricstrainconcentrationslimitedto Region A. The thickness of the yielded zone in Region A was consistentwiththecasetunnel.Themodifiedinputs(seeTable3.6)gaveplasticityresultsthatwereconsistentwiththepredictionsfromtheelasticmodelaswellasthecasetunnelpresentedinthepaper(i.e.PaperC).Itwasobviousthatthetensilestrengthwasanimportantstrengthcomponent.Unless its value is correctly estimated the behaviour of the nearfield rockmass, including the damaged zone, cannot be correctly simulated. It appearsthat, the tensile strength was significantly underestimated by the HoekBrownGSI empirical method for the rock mass type simulated in this thesis.Others, for example Diederichs et al. (2007) and Carter et al. (2007) have alsodrawnsimilarconclusionsformassivehardrockmasses.3.2.4 SummaryofPaperD(Saiang,2008c)BackgroundandmethodPaperCshowedthattheinputsestimateddirectlyusingtheHoekBrownGSIempirical method yielded results that were not consistent with the expectedbehaviour. The tensile strength and cohesion were significantlyunderestimated while the friction angle was overestimated. This resulted inyield mechanisms not correctly captured. One reason was that the rock wasyielding in a brittle manner, in which case yielding and failure will dependmainly on the tensile and cohesive strengths of the rock (e.g. Martin et al.,1999).Hence,aquestionwasraisedwhetherPFC2D(Itasca,2004)couldbeusedto estimate the inputs for the continuum model. This led to the developmentof PaperD. The goal was to determine a simplefailure envelope for the rockmass using PFC2D. This failure envelope can then be used to determine theMohrCoulombparametersforanalyseswithFLAC.61Numerical laboratory tests, involving biaxial and Brazilian tests onsyntheticrockmassspecimenswereperformedusingPFC2D.Thespecimenswere prepared to represent both the damaged and undamaged rock masses.Thesespecimenswereidealisedasequivalentcontinuum(intact)withreducedmechanicalproperties,whosecompressivestrengthanddeformationmoduluscanbederivedusingempiricalmethodssuchasHoekBrownGSIsystem.Theprocedure used in the calibration and the eventual biaxial and Brazilian testsareshowninFigure3.14.Toestimatedtheinitialmicromechanicalparameterscharts and methodologies presented by Diederichs (1999) and Huang (1999)wereused.ThetunnelandthemodelgeometriesusedintheFLACanalysesareshowninFigure 3.15. A doubletrack railway tunnel geometry was selected for theanalysis. The insitu stresses used in the FLAC models were those given byStephansson(1993),whicharebasedonovercoringmeasurements,as: Eq.3.4 gzV o = Eq.3.5 zH044 . 0 7 . 6 + = o Eq.3.6 zh034 . 0 8 . 0 + = o is the vertical stress, and whereVohoHo are the maximum and minimumhorizontal stress, respectively, is the rock density, g is the gravity and z isthedepth.Inthepaper isperpendiculartothetunnelaxisandhoHo isinthedirectionofthetunnel.Notethat,inthesimulationsfromPaperBtoCtheinsitustressesusedwerethose based on hydraulic fracturing measurements. However, it is nowbecoming increasingly clear that the insitu stresses from the overcoringmeasurements are more reliable than hydraulic fracturing measurements,especiallywithrespecttoshallowdepthenvironments.Hence,insitustressesfromtheovercoringmeasurementswereusedfromthispaperandothersthatfollowed.62Rock mass parameters: ci, GSI, miTarget values for Em and cm(by HB-GSI empirical method)Initial micro-parameters (by charts and empirical relations)Uniaxial compression test (by PFC2D)Have the target values for Em and cm been achieved?Final micromechanical properties achieved. Sample is ready for biaxial compression and Brazilian testsCut out Brazilian test sampleFine tune micromechanical parametersNOYESPerform biaxial test Perform Brazilian testDetermine shear strength parameters (from Mohr-Coulomb envelope)Determine tensile strengthFigure3.14:MicroparametercalibrationprocedureandeventualbiaxialandBraziliantests.(a)Fine gridsCoarse gridsBlast-induced damaged zone80 m80 m (b)12.5 m3.85 m6.54 m Figure3.15:ModelgeometryusedinFLACsimulation,(b)geometryforthedoubletracktunnel.Theexcavationislocated10mbelowthegroundsurface.63ResultsandconclusionsFigures 3.16 and 3.17 are the MohrCoulomb peak and residual strengthenvelopes for undamaged and damaged rock mass specimens. The MohrCoulomb strength parameters obtained from these envelopes are shown inTable 3.7. These parameters were used in the FLAC models. It was alsopossibletoevaluatetheaxialandvolumetricstrainsfrombiaxialcompressiontests. It was observed that the volumetric strains give a better indication ofyield and failure than the axial strains. Hence, the volumetric strains weredetermined, which are shown in Table 3.8. These strains were used with theMohrCoulomb StrainSoftening models. The traditional MohrCoulombmodel did not require the volumetric strain values. The dilation angles weredeterminedfromtheslopesofthevolumetricstraincurvesfortwospecimens.(a) (b)Figure3.16: MohrCoulomb peak strength envelopes for (a) Undamaged Rock MassSpecimen(URMS)and(b)DamagedRockMassSpecimen(DRMS).(a) (b)Figure3.17: MohrCoulombresidualstrengthenvelopesfor(a)URMSand(b)DRMS.64Table3.7: Strengthparameters.Peakvalues Residualvalues ParameterURMS DRMS URMS DRMSCohesion,c(MPa) 5.8 3.1 0.6 0.4Frictionangle,|( ) 32 31 35 38oTensionstrength, (MPa)to 2.5 1.3 0 0Dilationangle,( ) 9 7 oTable3.8: Averagevolumetricstrainvaluesatyield,peakandpostpeakresidual.Quantity URMS DRMSYieldvolumetricstrain(%) 0.05 0.04Peakvolumetricstrain(%) 0.06 0.05Residualvolumetricstrain(%) 0.46 0.35Figures 3.18 and 3.19 show the FLAC results based on the MohrCoulomb(MC) and the MohrCoulomb StrainSoftening (MCSS) constitutive models.The volumetric strain contours and plastic indicators are plotted together.PFC2Dmodelling showed that, failure is expected when the volumetric strainexceeds0.05to0.06%(seeTable3.8).Ground surfaceGround surface (a) (b)Figure3.18: Plasticity and volumetric straining resulting from the MC model; (a) withoutdamagedzoneand(b)withdamagedzone.65Ground surface Ground surface (a) (b)Figure 3.19: Plasticity and volumetric straining resulting from the MCSS model; (a)withoutdamagedzone(b)withdamagedzone.Theyieldingaroundthetunnelisconsistentwithpracticalobservationsfromacasetunnel(seePaperD)andalsothepredictionsfromthemodelsinPaperC.Both constitutive models (i.e. MohrCoulomb and MohrCoulomb StrainSoftening) showed yield at volumetric strains of 0.06%, which was expected.However, the MohrCoulomb StrainSoftening model was sensitive to thepresence of the damaged zone, due to the strainsoftening characteristics ofthiszone.Hence,theMohrCoulombStrainSofteningisconsideredimportantfor future modelling of damaged zone in a hard rock mass system. Inconclusion,theresultsdiddemonstratethatPFC2Dcanbeusedtoestimatetheinputparametersfortherockmass.Incidentally the tensile strength and cohesion determined from the PFC2Dsimulations are similar to those obtained by varying mi in Paper C (compareTable3.7tomodifiedinputsinTable3.6).663.2.5 SummaryofPaperE(SaiangandNordlund,2008b)BackgroundandmethodInPaperEtheeffectsoftheblastinduceddamagedzonewereinvestigatedfordeep excavations (depths greater than 100 m). Hence, a mining scenario wasselectedfortheanalysis.Therockmassparametersusedinthisanalysiswere:intactuniaxialcompressivestrength,180MPa,GSI,60andmi,33.Theinsitustresses were those shown in Eqs. 3.4 to 3.6 (see Section 3.2.4). Figure 3.20shows the excavation and model geometries used in this study (i.e. Paper E).The excavation dimensions conform to the underground mining driftstypicallyencounteredinSwedishundergroundmines.Following on from the work in Paper D a chart was developed from PFC2Dsimulations to estimate the input parameters for any given HoekBrown rockmass uniaxial compressive strength. This chart, shown in Figure 3.21, wasnecessary since it was not practical to perform PFC2D simulations to obtaininputs for every continuum simulation. The inputs for the simulations inPaper E were obtained from this chart. The MohrCoulomb StrainSofteningmodel was used in this paper and the volumetric strain values were thosederived in Paper D. Table 3.9 shows the inputs for the base case model. TheapplicationofYoungsmodulusinthemodelsfollowedthesameprocedureasinPaperB.(a)7.0 m3.85 m5.0 m (b)Fine gridsCoarse gridsDamaged rock zone60.0 m50.0 mFigure3.20: (a)Driftgeometryand(b)modelgeometry.671086420Cohesion and Tensile Strength (MPa)20 15 10 5Hoek-Brown rock mass uniaxial compressive strength (MPa)363432302826242220Friction angle (degrees)Strength components based on PFC simulationsTensile strengthCohesionFriction angleFigure3.21: Chart used to obtain the values of the strength components for various HoekBrownrockmasscompressivestrengths.Table3.9:Inputsusedinthebasecasemodel.Parameter Undamagedrockmass DamagedrockmassYoungsmodulus,E(GPa) 18 12Poissonsratio,v 0.25 0.25Peakcohesion, (MPa)pc 5.8 3.1Peakfrictionangle, (p| o)32 31Tensilestrength, (MPa)to 2.5 1.3Dilationangle, ( )9 7oPeakvolumetricstrain, (%)pc 0.06 0.05Residualcohesion, (MPa)rc 0.6 0.4Residualfrictionangle, ( )r| 35 38o68PaperEisessentiallyanapplicationoftheknowledgegainedfromPapersBtoD to deep excavations. Improved inputs and a more advanced constitutivemodel than the ordinary MohrCoulomb were used. Parameter analysis as inPaperB(seeSection3.2.2)wasalsoperformedforthedeepexcavationmodels.Thebasecasemodelwasconstructedat1000mdepth.ResultsandconclusionsAs in Paper B, the numerical measurements were taken at locations in thetunnel roof and the wall to study the behaviour of the damaged zone (seeFigure3.22).Theassessmentofthebehaviourofthedamagedzonewasdonebyassessinghowthevariationsinthepropertiesofthedamagedzoneaffectedthetangentialstressanddisplacementmagnitudes.The effects of the damaged zone were obvious, for example, in the base casemodelwhenthedamagedzonewasremoved.TheseareillustratedinFigures3.23and3.24.InFigure3.23(a)and(b)thetangentialstressincreasedby50%at Point A in the roof and 45% at Point C in the wall respectively, with theremovalofthedamagedzone.InFigure3.24(a)and(b)thedisplacementsinthewallandtheroofarereducedby10%.Inessence,withthepresenceofthedamagedzonethetangentialstressatthetunnelboundaryis0.2 intheroofatPointAand0.3Hovo inthewallatPointC. Without the damaged zone these values approximately double. With thepresenceofthedamagedzonethedisplacementis10%morethanthatofwiththe damaged zone. This was case for both the wall and the roof (see Figure3.24).ABC DFigure3.22:Pointsatwhichmeasurementsweredonetostudythebehaviourofthedamagedzone.690 5 10 15 20010203040500.3ov0.55ov No damaged zone With damaged zoneTangential stress, ou(MPa)Distance from tunnel wall (m)ov0 5 10 15 200204060801000.4oH0.2oHoH No damaged zone With damaged zoneTangential stress, ou (MPa)Distance from tunnel roof (m) (a) (b)Figure3.23: Behaviourofthetangentialstresses,withandwithoutdamagedzone,alongastraightlinein(a)thetunnelroofand(b)thetunnelwall.0 2 4 6 80102030401.1oundamaged With damaged zoneNo damaged zoneRoof displacement (mm)Distance from tunnel roof (m)oundamaged0 4 8010203040121.1oundamaged No damaged zone With damaged zoneWall displacement (mm)Distance from tunnel wall (m)oundamaged(a) (b)Figure3.24:Displacementswithandwithoutdamagedzonealongastraightlinein(a)thetunnelroofand(b)thetunnelwall.Figure3.25showstheimpactsoftheparametervariationsontangentialstressat the measurement points, which is a direct reflection of the response of thedamagedzoneduetotheparametervariations.Figure3.26showstheimpactsof the parameter variations on the displacement. Note that, the effects areassessedaspercentagevariationsfromthebasecasemodelresults.70Variations in the modulus and the compressive strength affected thetangentialstressmagnitudequitesignificantly,whichisexpectedduetohighinduced stress magnitudes observed in the deep excavations models. Theimpact of the compressive strength of the rock in the shallow excavationmodels in Paper B was not significant, because of the low induced boundarystresses. The variation in the thickness of the damaged zone has a significantinfluence on the peak tangential stress at the damagedundamaged rockboundary(i.e.PointsBandD).Thisindicatestheeffectivenessofthedamagedzone (with low strength and stiffness) to force high stresses away from thetunnelboundary.(a)020406080100VaryingdeformationmodulusVaryingcompressivestrengthVaryingdamagedzonethicknessVaryinghorizontal toverticalstress ratioVariation in tangential stress (%)Point A Point B(b)020406080100VaryingdeformationmodulusVaryingcompressivestrengthVaryingdamagedzonethicknessVaryinghorizontal toverticalstress ratioVariation in tangentail stress (%)Point C Point DFigure3.25: Percentagevariationsinthetangentialstressatpoints(a)AandBintheroofand(b)CandDinthewall,whendifferentparametersofthedamagedzonewerevaried.71(a)04080120160200VaryingdeformationmodulusVaryingcompressivestrengthVaryingdamaged zonethicknessVaryinghorizontal tovertical stressratioVariation in roof displacement (%)Point A(b)04080120160VaryingdeformationmodulusVaryingcompressivestrengthVaryingdamaged zonethicknessVaryinghorizontal tovertical stressratioVariation in wall displacement (%)Point CFigure3.26: Percentagevariationsinthedisplacementatpoints(a)Aintheroofand(b)Cinthewall,whendifferentparametersofthedamagedzonewerevaried.MeasurementofthedisplacementsatPointsAandC,atthetunnelboundary,showed that variations in the horizontal stress to vertical insitu stress ratiosaffected the response of the damaged zone the most, and thus thedeformation. A change of over 130% from the base case model was notedwhen the insitu stress ratio was set to 3.0. The displacements were alsosensitivetothicknessofthedamagedzone.Thiswasexpectedsincethelargerthe damaged zone (with low strength and stiffness) the greater thedeformation.723.2.6 SummaryofPaperF(Saiang,2008d)BackgroundandmethodThis paper (i.e. Paper F) presents the analysis of the blastinduced damagedzone by coupled continuumdiscontinuum method. FLAC and PFC2D werecoupled to create the continuumdiscontinuum models. The idea is to studythe behaviour of the nearfield rock mass, including the blastinduceddamaged zone with PFC2D, while the farfield rock mass is simulated withFLAC. Figure 3.27 illustrates the coupling of the two codes. The mainadvantage of this method is twofold; (i) failure and fallout of the nearfieldrockmasscanbemapped,(ii)computertimeisreduceddrasticallysinceonlytheareaofinterestissimulatedusingthediscontinuummethod.ModelswerealsorunindependentlywithFLACandPhase2.Theideawasthat,sincecontinuummodelscannotsimulatefailure,theresultsfromthecoupledmodel could be used to identify indicators for failure in the continuummodels. In that way future analyses of the damaged zone can still be donewith continuum methods which appropriate indicators to analyse itsbehaviour.Figure3.27:FLACPFC2Dcoupledmodel.A9x9mhorseshoeshapedtunnelisexcavatedwithinthePFC2Dbondedparticles.Theminimummodeldimensionofthecoupledmodelis120x120m.73To model the blastinduced damaged zone explicitly, radial cracks wereindividuallytracedandimplementedinthecomputermodels(seeFigure3.28and 3.29). A sample crack pattern from the SveBeFo field studies was chosenfor the crack tracing. However, for the purpose of numerical modelling thetracing and implementation in the computer models were simplified. It wasassumed that the cracks orienting at angles up to 25o from the halfcastintersectthosewithsimilarorientationfromtheneighbouringblastholes(seeFigure 3.25). Cracks orienting at angles greater than 25odo not intersect. Forthe sake of numerical modelling the crack lengths were assumed to have thesamelength,whichinthiscasewerefixedat0.5mandareradial(seeFigure3.26).The cracks were implemented in the FLACPFC2D coupled models and inPhase2models.ItwasnotpossibletoimplementthecracksinFLAC.Hence,thedamaged zone was simply treated as an equivalent continuum in FLAC as inPapers B to E. The implementation was nevertheless simple with Phase2,because of its CAD capabilities and the options for adding various kinds ofjoints.InthecoupledFLACPFC2DmodelthecrackswereimplementedinthePFC2Dcomponent.25o 21o17oBlast-holesFigure3.28:Estimationofintersectingcracksfromadjacentblastholes.Crackswithorientationanglesupto25ofromtheblastholerowintersecteachother(photoadaptedfromOlssonandBergqvist,1995).74Figure3.29:RadialcracksgeneratedwithinthePFC2DandPhase2models.Thespacingbetweentheblastholesis50cm.TheinsitustressesusedinthemodelsofthispaperarethosegivenbyEq.3.4to 3.6. The MohrCoulomb parameters and the micromechanical propertiesused in the models were those derived in Paper D. Table 3.10 shows themicromechanical properties for the PFC2D component of the coupled model.Becauseofthelimitedinformationaboutthemechanicalpropertiesofablastinduced crack, the values shown in Table 3.11 are assumed from Saiang et al(2005a).Africtionangleof45o(i.e.frictioncoefficientof1.0)wasassumedfortheblastinducedcracks.For the sake of consistency as well as to be able make comparisons of theresults the MohrCoulomb constitutive model was used in the continuummodels (i.e. FLAC and Phase ), since Phase2 2 has no option for MohrCoulombStrainSoftening model. In the PFC2D component of the coupled model theparallel bond model was used for the particles while the smooth joint modelwasusedforthecracks.Table3.10: Micromechanicalpropertiesforsyntheticrockmass(Saiang,2008c)Parameter ValueNormalbondstiffness(GN/m) 58Shearbondstiffness(GN/m) 23Normalbondstrength(MN) 2Shearbondstrength(MN) 10Frictioncoefficient 1.07576Table3.11: Propertiesoftheblastinducedcracks(stiffnessandstrengthdataareassumedfromSaiangetal.,2005a)Parameter ValueNormalstiffness(MN/m) 1.3Shearstiffness(MN/m) 1.3Shearbondstrength(MN) 1.4Normalbondstrength(MN) 0Frictioncoefficient 1.0ResultsandConclusionTable 3.12 shows the failed zones mapped from the FLACPFC2D coupledmodels. This was done by mapping areas where the PFC2D particles haveeitherfallenoutordetachedbylossof bondcontact.Theforcesweretrackedtoidentifymechanismscausingfailure.Theblastinducedcrackswereseentoinfluencethefailurecharacteristicsfordepthsupto500m.Thesewereevidentfromthesawtoothfailuresurfaces,whichwerelessevidentwhenthemodelswere run without the cracks. In the deep excavation models this effectvanished,indicatinglessinfluencebytheblastinducedcracks.Table 3.13 shows the possible failed regions mapped from the Phase2 models.These were obtained by contouring the regions where the elements haveyielded in 100%. Plasticity indicators showed that the 100% yielded elementswere more consistent with the failed regions mapped in the coupled modelthan other indicators. At large depths however there is less consistencybetweenthecoupledandthePhase2modelresults.Table 3.14 shows the possible failed regions mapped from the FLAC models.These were done by contouring the regions where volumetric strainingoccurred. Regions where the volumetric strain exceeded 0.05% were moreconsistent with the failed regions mapped in the coupled model. As with thePhase2 model results, the FLAC model results are less consistent for deepexcavations(depthsgreaterthan100m).However,theFLACandPhase2modelresultsshowsimilarbehaviouratalltheexcavationdepthsmodelled. Table3.12:FailedzonesmappedfromtheFLACPFC2Dcoupledmodelsfordepthsof10,100,500and1000mwiththeblastinduceddamagedzone.10 m 100 m 500 m 1000 m Overburden Failed zones mapped from coupled model -Roof:-Roof:-Roof:-Roof:-102 cm -80 cm -17 cm -< 1 cm -compressive shear -compressive shear -compressive shear -compressive shear Maximum depth of failure and failure mechanisms-Wall:-Wall:-Wall:-Wall:-63 cm -45 cm -24 cm -5 cm -tensile-tensile-tensile-tensile-Floor:-Floor:-Floor:-Floor:-11 cm -4 cm -< 1 cm -25 cm -compressive shear -compressive shear -compressive shear -compressive shear 77 Table3.13:FailedzonesmappedfromthePhase2modelsfordepthsof10,100,500and1000m.Modelsarewiththeblastinduceddamagedzone.Overburden 10 m 100 m 500 m 1000 m Failed zones mapped fromPhase2 model -Roof:-Roof:-Roof:-Roof:-170 cm -100 cm -34 cm -< 1 cm -compressive shear -compressive shear -compressive shear -compressive shear Maximum depth of failure and failure mechanisms-Wall:-Wall:-Wall:-Wall:-110 cm -45 cm -43 cm -3 cm -tensile-tensile-tensile-tensile-Floor:-Floor:-Floor:-Floor:-120 cm -30 cm -< 1 cm -180 cm -compressive shear -compressive shear -compressive shear -compressive shear 78Table3.14:FailedzonesmappedfromtheFLACmodelsfordepthsof10,100,500and1000m.Modelsarewiththeblastinduceddamagedzone.79 Overburden 10 m 100 m 500 m 1000 m Failed zones mapped fromFLAC model Maximum depth of failure and failure mechanisms-Roof:-< 1 cm -compressive shear -Wall:-8 cm -tensile-Floor:-0 cm -compressive shear -Roof:-36 cm -compressive shear -Wall:-38 cm -tensile-Floor:-32 cm -compressive shear -Roof:-96 cm -compressive shear -Wall:-40 cm -tensile-Floor:-122 cm -compression shear -Roof:-180 cm -compressive shear -Wall:-86 cm -tensile-Floor:-200 cm -compression shear OntheotherhandthebehaviourobservedinthetunnelwallswithFLACandPhase models are similar to those from the FLACPFC2 2D coupled models atshallow depths (see Tables 3.12, 3.13 and 2.14). This indicates that the brittlefailure occurring in tunnel walls were captured reasonably when volumetricstrains (in FLAC) and 100% yielded elements (in Phase2) were used asindicatorsforfailure.ThedepthsoffailedzonesmappedfromthecoupledFLACPFC2Dmodelsforthevariousexcavationdepthsarewithintherangeoftenreportedinpractice.However,thecontributionsofthedamagedzonetothefailuresinpracticearenot known. Nevertheless, the coupled FLACPFC2D model results showed theincrease in the depth of failed zone with the presence of the damaged zone.ThisisdemonstratedbyFigures3.30and3.31,forthewallandtheroof.Intheshallow excavations the roof was less affected by the damaged zone (seeFigure3.31).Astheexcavationdepthincreasedtheeffectbecamesignificantasthe failure increased more in the roof than in the walls. The effects of thedamagedzoneinthewallsareconsistentwithobservationsmadeinmodelsofPapersBtoE.10 100 1000020406080Thickness offailed zone (m)Excavation depth (m) Thickness of failed zone - No blast-induced damaged zone Additional failure due to blast-induced damaged zoneWALLFigure3.30:MaximumthicknessofthefailedzoneinthewallmappedfromthecoupledFLACPFC2Dmodels.Thepresenceofthedamagedzoneincreasesthedepthoffailure.8010 100 1000020406080100120Thickness of failed zone (cm)Excavation depth (m) Thickness of failed zone - No blast-induced damaged zone Additional failure due to blast-inuced damaged zoneROOFFigure3.31:MaximumthicknessofthefailureintheroofmappedfromthecoupledFLACPFC2Dmodels.Thepresenceofthedamagedzoneincreasesthedepthoffailure.813.3 SHOTCRETEROCKINTERACTION3.3.1 BackgroundShotcrete is widely used as a surface rock support in tunnelling andundergroundmining.Oneofthemainrolesofshotcreteistokeepinplaceanylooserockparticlesaswellaspreventingdilationoflooseblocksandeventualfallouts. Much of the interaction between shotcrete and rock occurs at theinterface between shotcrete and the damaged rock. The behaviour andperformance of shotcrete is influenced by among other factors, the damagedzone around the excavation boundary (e.g. Malmgren, 2005). Thus theperformanceofshotcretedependsonthepropertiesofshotcrete,thedamagedrock and interface properties. For this reason LKAB initiated a project toinvestigatetheshotcreterockinterfaceinordertounderstandthebehaviourattheinterface.The effectiveness of shotcrete for this task relies on the strength and stiffnessofthecontactbetweentherockandshotcrete.Someoftheearlystudiesonthestrength of shotcreterock interfaces were by FernandezDelgado et al (1976)andHolmgren(1979).Sincethenalargeandvariednumberoftestshavebeenperformed, including field studies and observations. However, due to thecomplexity of shotcreterock interaction the various methods could onlyprovidespecificdataforsimplegroundconditions.3.3.2 ShotcreterockinteractionTheinteractionbetweenshotcreteandrockattheinterfaceisquitecomplexasstudiesbyamongothersHolmgren(1979;1985),Stille(1992),Malmgren(2001)and Stacey (2002) have shown. These authors also show that at a genuinelycemented interface the adhesion strength is important for shotcretes82perfromance. External factors such as rock surface preparation and geometryof the rock surface on which shotcrete is applied have been found to affectadhesion strength quite significantly. Malmgren (2001) has shown that theadhesion strength of shotcreterock interface at Kiirunavaara undergroundmine (in Sweden) was significantly increased when the rock surface wasprepared by waterjet scaling. Figure 3.32 shows some of the factors thatcontributetotheinteractionbetweenshotcreteandrock.The mode of loading (tension, compression or shear, see Figure 3.33) and themagnitude of this load, will significantly impact the overall behaviour of theinterface.ThiswasalsodemonstratedinthelaboratoryinvestigationreportedinPaperG,whichispresentedintheproceedingsection.ovoHShotcreteDamaged rockUndamaged rockShotcrete-rock interaction Shotcrete properties Shotcrete-rock interface properties Damaged rock properties Rock mass properties In-situ stresses Strength Stiffness Stiffness Strength Strength Stiffness Strength Stiffness ovoHExtent Rock surface characteristics ohFigure3.32:Interactionbetweenshotcreteandrock,andsomeofthecontributingfactors.83Figure3.33:Failuremodeswhenshotcreteandrockinteract(afterCording,1974).3.3.3 SummaryofPaperG(Saiangetal.,2005a)MethodPaperGpresentsaseriesoflaboratorytestsperformedoncementedshotcreterock joints to investigate the strength and stiffness of the interfaces, whilesimulating field conditions as close as possible. Three kinds of tests wereperformed, (1) direct shear test, (2) tensile test and (3) compression test. Themain focus of the laboratory investigation was the direct shear test. Tensionand compression tests were complementary, but at the same time theyprovidedusefuldataonthenormalstrengthandstiffnessoftheinterface.Figure 3.34 shows the basic description of the samples used in theexperiments. All the samples used were cylindrical. The shotcrete and rockwere cemented, with surface roughness at the interface having JRC values of13and913.ThesetupforthedirectsheartestisillustratedinFigure3.35andFigure3.36showsthesetupforthedirecttensiontest.84Rock Fibre reinforced shotcrete Fibre reinforced shotcre-180 mm 60 - 80 mm 80 - 100 mm 80 - 100 mm 60 - 80 mm 94 mm 94 mm 80 - 100 mm (a) (b) (c)Figure3.34:Testsamples.(a)Directsheartestsample,(b)tensileandcompressiontestsampleand(c)shotcretecompressiontestsample. LVDT-2 Connection to computer LVDT-1 LVDT holder frames Shear direction Top half Bottom Figure3.35: Directsheartestsetup. CODs (3)(1) (2)Figure3.36: Directtensiontestsetup.85ResultsandConclusionsTable 3.15 shows the main data that were sought after in the experimentalwork. However, a short conclusion can also be made about the behaviour oftheshotcreterockinterface.- Thebondstrengthwasobservedtocontrolthepeakshearstrengthoftheinterface at low normal loads ( MPa 0 . 1 no )thepeakshearstrengthisinfluencedbythebondstrengthandthefriction(seeFigures3.37and3.39).- Thebondstrengthistheaccumulatedstrengthofthecementingplusthestrengthoftheshotcreteasperities,whichwasdeterminedfromthedirectsheartest.Theadhesionstrengthisobtainedthroughdirecttension.- The surface roughness affected the bond strength significantly. SurfaceswithahighJRCvalueresultedinhighbondstrength(seeTable3.15).Table3.15: Summaryofstrengthpropertiesfortheshotcreterockjointstested.ValueforJRC=13 ValueforJRC=913 ParameterJointshearbondstrength 0.50MPa 1.37Jointfrictionangles:Peak,| 40 47o opResidual,| 35 39o orJointcompressivestrength,JCS 16.0MPa Jointadhesionstrength 0.56MPa Jointcompressionstiffness:K 100MPa/mm ciK 288MPa/mm ct50K 182MPa/mm cs50Jointtensilestiffness:K 288MPa/mm tiK 251MPa/mm tt50K 261MPa/mm ts50Jointshearstiffness,K 0.94MPa/mm 1.3MPa/mm s860.0 1.0 2.0 3.0 4.0Normal stress, on, (MPa)0.01.02.03.04.0Peak shear stress, tp, (MPa)peak shear strength = bond strengthpeak shear strength = bond strength + frictionPeak shear strength controlled by bond strength under low normal stresses Figure3.37: Bondfailuredominatesunderlowconfiningstressconditions.0.0 4.0 8.0Displacement, o, (mm)0.00.20.40.60.81.0Shear stress, t, (MPa)Sample #62: Shotcrete-magnetite jointNormal stress = 0.54 MPaSecondary peak due to frictionBond failurePeak shear strength = bond strengthFigure3.38:Bondfailurecharacteristics.Peakstrengthisthestrengthofthebond.870.0 4.0 8.0 12.0Displacement,o, (mm)0.00.10.20.30.4Shear stress, t, (MPa)Peak shear strength = bond strength + frictionSample #36: Shotcrete-trachyte jointNormal stress = 0.23 MPaFigure3.39:Peakstrengthistheaccumulatedstrengthoffrictionandcohesion.88CHAPTER4:SUMMARY,CONCLUSIONSANDCONSIDERATIONSFORFURTUREWORK894.1 SUMMARYThis thesis has been an exploration of the behaviour of the blastinduceddamaged zone around underground excavations in hard rock masses. Thestudy was achieved mainly through field and numerical investigations. Thenumericalstudiesformedthemajorcomponentoftheseinvestigations.The following are the summaries from these investigations in the order theywerepresentedinthisthesis:(i) Field investigation of the excavation damaged zone was conductedat Kiirunavaara underground mine in northern Sweden. Thisinvestigation revealed the thickness of the damaged zone to bebetween 0.5 and 1.0 m, depending on the blasthole diameter. Thedetermined modulus of the damaged rock ranged between 50 and90%ofthatoftheundamagedrockmodulus.(ii) Thecontinuumandcoupledcontinuumdiscontinuummethodswereemployed in studying the behaviour of the blastinduced damagedzone. The task began with a continuum approach using techniquescommonly used methods for geomechanical analyses. However, asmore knowledge was gained about the behaviour of the damagedzone and the limitations of the numerical methods; the modellingapproachchangedand progressively improved. This led tothe finalsimulation of the damaged zone through coupled continuumdiscontinuummethod.InthismethodFLACandPFC2Dwerecoupledtostudythebehaviourofthedamagedzone.(iii) Shotcrete is a widely used surface support for undergroundexcavations. The interaction between shotcrete and rock usuallyoccurs at the interface between shotcrete and the damaged rock.Laboratorystudieswereperformedtoinvestigatethebehaviourandcharacteristics of the shotcreterock interface. The properties of the90interface investigated include; the shear strength parameters, theadhesionstrengthandtheinterfacestiffnessparameters.4.2 DISCUSSIONSANDCONCLUSIONS4.2.1 LiteratureandQuestionnaireLiterature review and a questionnaire revealed that the presence of the blastinduced damaged zone is an important concern for the stability andfunctionality of an excavation. The consequences of the blastinduceddamagedzoneareusuallyconceivedintermsofsafetyandcost.It is still unclear however, how the blastinduced damaged zone affects thestability and performance of underground excavation in a hard rock mass.Onereasonisthattheblastinduceddamagedzonehasbeenlargelydefinedintermsofitsextent.Forstabilityandperformancethestrengthanddeformationparametersofthedamagedzoneareimportantandthedamagedzonelacksadefinitionbasedontheseparameters.Thedevelopmentandextentoftheblastinduceddamagedzoneisdependenton many factors, such as; explosive parameters, blast design geometricalparameters, rock mass parameters and insitu stress conditions. However,theseareoutsidethescopeoftheworkpresentedinthisthesis.The characteristics of the damaged zone though are important. The existenceof microfractures and rock bridges within the damaged zone will affect thebehaviour of the damaged zone significantly. For example under lowconfinementconditionstherockbridgeswillplayasignificantroleinthetotalstrengthofthedamagedzone.ThisbehaviourwasseeninthemodelsofPaperF, where the blastinduced cracks significantly influenced the failure of thedamagedzoneatshallowdepth.InthedeepexcavationmodelsofPaperFthecombinedstrengthoftherockbridges,andthecohesiveandfrictionalstrengthof blastinduced cracks, provided resistance for localised failures (due totension)asatshallowdepths.914.2.2 InvestigationsandAnalysesExcavationdamagedzoneinvestigationThe field investigation of the excavation damaged zone at Kirunavaaraundergroundminerevealedthatthemodulusofthedamagedzonearoundanexcavation was reduced by as much as 10 to 50%, depending on the blastingparameters. The extent of this effect was limited to depths of 0.5 to 1.0 m.However, microcracks may extend beyond this limit. In cautious blastingoperationstheextentofthedamageisusuallylimitedtoanaverageof0.3m.It can be concluded that in practical cases the thickness of the blastinduceddamagedzonegenerallyvariesbetween0.1and1.0m,withreportedaveragesbetween0.3and0.5m.Themodulusofthedamagedzonecanbereducedbyas much as 50% of the modulus of the undamaged rock mass, depending ontheblastingparameters.NumericalstudiesNumerical studies conducted in this thesis have been presented in a logicalsequence. The first numerical paper (i. e. Paper B) presented the parameterstudy of the damaged zone. From this paper the important parameters,limitations and problems were identified which set the groundwork forfurther numerical investigations, leading to Papers C to F. By progressivelyimproving the models from lessons learnt from previous models, moreadvanced models were developed to study the behaviour of the damagedzone.Thefollowingarethehighlightsandconclusionsfromthestudies.- Because there was no specific method for estimating the inputparameters for the damagedzone, a systematic method was developedin Paper B to estimate the inputs for the deformation and strengthparameters.ThecommonlyusedHoekBrownapproachwasusedasthebasis for this procedure. It was evident however that the direct of use92the HoekBrown approach may not be adequate to estimate the inputparametersforthedamagedzone.The important properties of the damaged zone identified in the studyreportedinPaperBinorderoftheirsignificanceare:(i)thedeformationmodulus, (ii) the tensile strength and (iii) the compressive strength.Changes in the deformation modulus affected the behaviour of thedamaged zone most, followed by the tensile strength and thecompressive strength. Of the external parameters tested the changes intheinsitustressesaffectedthebehaviourofthedamagedzonethemostcomparedtotheoverburdenandthethicknessofthedamagedzone.For the cases studied, the maximum deformation was 30% more whenthe modulus was reduced by 50%. In the base case models where themodulus of the damaged zone was reduced by 30% the deformationwas 10% more than without the damaged zone. This observation wassimilar for both shallow and deep excavation models. Note that, in theFLAC models the modulus was linearly increased from the tunnelboundary to the boundary between the damaged and the undamagedrockmasses.It can be concluded that, the presence of the blastinduced damagedzonedoesaffecttheoverallbehaviouroftherockmassintheimmediatevicinity of the excavation. Two main effects were observed with theFLACbasedcontinuummodels:i. increase in deformation due to the reduction in the modulus ofthedamagedzone.ii. reduction in tangential stress near the excavation boundarybecause of the reduced modulus and hence high stresses wereconcentratedoutsideofthedamagedzonelimit.- ItwasidentifiedinPaperBthatthefailureprocessofthedamagedzonewas an influencing factor for its behaviour. This led to Paper C where93the failure mechanism of the damaged zone was studied, althoughfocusingonshallowexcavations.Inthisstudytwoimportantsourcesoffailurewereidentified(i)biaxialextensionleadingtotensilefailureand(ii) extension and compression in a biaxial stress field, also leading totensilefailure.Thesetwoconditionsweredominantinthewalls,whichled to tensile failure of the damaged zone. Compressive shear failures,the third source, occurred mostly in the roof and floor but wereinsignificantforshallowexcavationsandthereforelessinfluentialonthedamaged zone. In deep excavations however, the compressive shearfailures of the damaged zone were significant. This observation wasmadeinPaperF,wherethebehaviourofthedamagedzonewasstudiedusingthecoupledcontinuumdiscontinuummethod.- InPaperBitwasidentifiedthattheassessmentofthevolumetricstrainwas a reliable indicator for tensile failure of the damaged zone. Thisobservation was later verified by Paper F. Other plasticity indicatorstend to overpredict the failure zones. This can be expected for acontinuummodelwhereaperfectlyplasticmodelisused.Theperfectlyplastic model tends to capture shear failure well but not the brittlefailure.- It was observed in Paper C that the continuum models could still beusedtocapturethebehaviourofthedamagedzoneifcorrectinputsareused. It was obvious thattheuse ofthe HoekBrown empiricalmethodwas limited and any inputs to be obtained were by trial and error.However, in this thesis a method was developed using PFC2D which ispresented in Paper D. By performing a series of biaxial and Brazliantests on synthetic rock mass specimens the MohrCoulomb failureenvelopes were obtained for the damaged rock mass as well as theundamaged rock mass. These tests were extended and a chart wasdevelopedtoestimatetheshearandtensilestrengthparametersforanygivenHoekBrownuniaxialcompressivestrengthoftherockmass.Thischartwasusedinestimatingtheinputsforthenumericalsimulationsofthe damaged zone in Papers E and F. The inputs obtained from the94PFC2D simulations and the chart gave results that were logically soundandalsoconsistentwithpracticalobservations.- It was observed from Papers C to E, that the brittle failure of thedamaged zone (due to tensile mechanism) can be best captured by adiscontinuum code such as PFC2D (3D can also be used). Hence PFC2Dwas coupled with FLAC to create a hybrid continuumdiscontinummmodel. This enabled the study of the damaged zone in Paper F. Blastinduced cracks were explicitly implemented as joints in the coupledmodel.ThefollowingspecificconclusionscanbemadefromtheFLACPFC2Dcoupledmodelresults. The brittle failure of the damaged zone was captured well byFLACPFC2Dcoupledmodel.Theresultswerealsoconsistentwithpredictions from Papers C to E. At shallow depths, much of thefailure of the damaged zone was tensile and occurred mostly inthewalls.Asthedepthincreasedthedominantfailuremechanismshifted from tensile in the walls to compressive induced shear inthe roof. This behaviour was captured reasonably well by thecoupledmodel.Themaximumdepthsofthefailureswereashighas5cminthewallfor10mmodeltoasmuch102cminroofforthe1000mmodel. Thecoupledmodelsalsoshowedthatatexcavationdepthsupto500 m the failure surfaces were influenced by the radial cracks.These were evidenced by sawtooth failure surfaces, wherefailures were propagating along the cracks and then failing intension. The sawtooth failure surfaces vanished in the 1000 mmodel. This indicates that the blastinduced cracks had lessinfluenceonthefailurecharacteristicsofthedamagedzone.Thatis, although the damaged zone will increase the thickness of thefailure zone, it may not dictate how the failure should occur.Failurewillbelargelydrivenbyhighcompressivestresses.95 Paper F also revealed that the tensile failures occurring in thewalls were captured well by FLAC and Phase2. However, theindicators used were the volumetric strain (in FLAC) and 100%yielded elements (in Phase2). A volumetric strain value of 0.05%(inFLACmodels)wasfoundtobethelimitwhenyieldingand/orfailurebegins.ThisvaluewasalsodeterminedearlierinPaperDthroughPFC2Dmodelling.ThecompressiveinducedshearfailureswereoverpredictedbytheFLACandPhase2.Thiscanbeexpectedsincethemodelswerebasedonaperfectlyconstitutivemodel;theMohrCoulomb.4.2.3 ShotcreteRockInteraction.Aspartoftherocksupportstudies,theinteractionbetweenshotcreteandrockwas also studied as part of the work presented in this thesis. Because theinteractionbetweenshotcreteandrockusuallyinvolvethedamagedzonetheinterface parameters are important for design and analyses. This led tolaboratory tests presented in Paper G. This study provided the data forsimulatingtheinteractionbetweenshotcreteandthedamagedrockmass(seeMalmgren,2005).Theimportantshearstrengthparameterdeterminedforlowconfinement condition was the bond strength, which was about 0.5 MPa fornormal loads less than 1.0 MPa. The adhesion strength of the cementedshotcreterockinterfaceswasalsodetermined,withthevalueof0.56MPa.Thestiffness parameters (normal and shear) were also determined which arepresentedinPaperG.4.3 SUGGESTIONS- The effects of the damaged zone on the behaviour of the rock pillardirectly above the tunnel at depths less than 10 m needs furtherinvestigation. It was observed that, since high stresses are normallyconcentrated there, the presence of the damaged zone can reduce thethicknessofthispillarandthusitsriskoffailure.96- Fieldmappingoffailuresintunnelsareneededtospecificallyidentifyifany of these failures are related to the blastinduced damaged zone.Failure mapping have seldom been done for shallow excavations,mainlybecausefailureshavebeeninsignificant.However,themappingisessentialifwewanttoknowwhetherthedamagedzoneisimportantornot.- In an attempt to improve inputs for continuum models PFC2Dsimulationswereconducted,bynumericallaboratorytestsonsyntheticrock mass specimens. A chart was then developed from thesesimulations to estimate the inputs for both the damaged andundamaged rock masses. However, more work needs to be done toimprove this chart through verification measurements and additionalPFC2D simulations. This chart can then be used as a practical tool toestimate the strength parameters for any HoekBrown rock massuniaxialcompressivestrength,foruseinacontinuummodel.- The coupling of FLACPFC2D to create a hybrid continuumdiscontinuum in this thesis sets a new approach for studying thebehaviour of the damaged zone. It is felt that the models can beimprovedsothatotherparameterssuchasthevariationinthemodulusaroundthetunnelcanalsobestudied.Furthermore,failuremappingareneededtoverifythereasonabilityofthesemodels.- The failure process of the damaged zone need to be further studiedusing the coupled FLACPFC2D model. It was observed that at shallowdepths the blastinduced cracks acts like a discrete body, where thestrength of the damaged zone is controlled by the tensile strengths ofthe individual rock bridges between the blastinduced cracks. In deepexcavations the damaged zone acts like a single continuum, where itsstrengthisprovidedbythecohesiveandfrictionalstrengthofthecracksaswellasthetensilestrengthsofrockbridges.Thesemechanismsneedtobestudiedwell.97- A test of the constitutive models needs further work. It was seen fromthe work in this thesis that, although the MohrColoumb model gavereasonable results, when appropriate inputs are used, the MohrCoulomb StrainSoftening appeared to capture the strain softeningcharacteristicsofdamagedzonewell.TheinputsforthestrainsofteningmodelwereobtainedthroughPFC2Dmodelling.Thisalsoneedsfurthertesting.- The long term strength of the damaged zone is important and needsfurther study. The strength of rock bridges within the damaged zonecan deteriorate over time, thus reducing the overall strength of thedamagedzone.98REFERENCESAlheid, H.J. and Knecht, M., 1996. Distribution of seismic wave velocities aroundunderground drifts. In: M. J.B. and M. C.D. (Editors), Proceedings of theExcavation Disturbed Zone Workshop Designing the Excavation DisturbedZoneforaNuclearWasteRepositoryinHardRock,Manitoba,Canada.Andersson, P., 1992. Excavation disturbed zone in tunneling. SveBeFo Report No. 8SwedishRockEngineeringResearch,Stockholm.Andrieux, P.P., Brummer, R.K., Liu, Q., Simser, B.P. and Mortazavi, A., 2003. Largescale panel destress blast at Brunswick mine. Journal Canadian Institute ofMining,MetallurgyandPetroleum:7887.AnlggningsAMA98, 1999. General materials and works description forconstruction work, sectionCBC: Bergschakt(SvenskByggtjanst: Stockholm),inSwedish.Autio, J., SiitariKauppi, M., Timonen, J., Hartikainen, K. and Hartikainen, J., 1998.Determination of the porosity, permeability and diffusivity of rock in theexcavationdisturbed zone around fullscale deposition holes using the14CPMMA and Hegasmethods. Journalof ContaminantHydrology,35(13):1929. Barla, G., Barla, M. and Repetto, L., 1999. Continuum and discontinuummodeling for design analysis of tunnels, 9th Int. Congress on Rock Mech,Paris,France.Banverket,2002.BVTunnel.StandardBVS585.40.Banverket.Barla, G., Barla, M. and Repetto, L., 1999. Continuum and discontinuum modelingfordesignanalysisoftunnels,9thInt.CongressonRockMech,Paris,France.Barton,N.,2007.FuturedirectionsforrockmassclassificationandcharacterizationTowards a crossdisciplinary approach. In: E. Eberhardt, D. Stead and T.Morrison (Editors), Rock Mechanics: Meeting Societys Challenges andDemands.Taylor&FrancisGroup,London,Vancouver,Canada,pp.179189.Bauer,C.,HommandEtiene,F.,Slimane,K.B.,Hinzen,K.G.andReamer,S.K.,1996.DamagezonecharacterizationinthenearfieldintheSwedishZEDEXtunnelusing in situ and laboratory measurements. In: G. Barla (Editor), Eurock 96.Balkema:Rotterdam,pp.13451352.99Bernier, F. and Davies, C., 2004. Impact of the Excavation Disturbed or DamagedZone(EDZ)onthePerformanceofRadioactiveWasteGeologicalRepositories.EUR21028EN,Luxembourg.Blake,W.,1972.DestressingtestatGalenamine,Wallace,Idaho.Trans.SMEAIME,252:294299.Boler,F.M.andSwansson,P.L.,1993.SeismicityandstresschangessubsequenttodestressblastingattheGalenamineandimplicationsforstresscontrolstrategies,ReportofInvestigations,RI/9448,BureauofMines,USDepartmentofInterior.Brady, B.H.G. and Brown, E.T., 1993. Rock Mechanics for underground mining.Chapman&Hall,571pp.Brown,E.T.(Editor),1981.ISRMCommissiononStandardizationofLaboratoryandFieldTest:Suggestedmethodforquantitativedescriptionofdiscontinuitiesinrockmass.PergamonPress.Bckblom,G.andMartin,C.D.,1999.Recentexperimentsinhardrockstostudytheexcavation response: impacts for the performance of a nuclear wastegeological repository. Tunnelling and Underground Space Technology, 14(3):377394.Cai, M., Kaiser, P.K. and Martin, C.D., 2001. Quantification of rock mass damage inunderground excavations from microseismic event monitoring. Int. J. RockMech.&Min.Sci.,38(8):11351145.Cai, M. et al., 2004. Generalized crack initiation and crack damage stress thresholdsof brittle rock masses near underground excavations. International Journal ofRockMechanicsandMiningSciences,41(5):833847.Carter,T.G.,Diederichs,M.S.andCarvalho,J.L.,2007.AunifiedprocedureforHoekBrownpredictionofstrengthandpostyieldbehaviourforrockmassesatextremeendsofthecompetencyscale.In:R.e.Sousa,OlallaandGrossman(Editors),11thCongressoftheInternationalSocietyofRockMechanics.Taylor&Francis.,Lisbon,pp.161167.Cording,E.J.,1974.Geologicconsiderationsinshotcretedesign,ProceedingsofEngineeringFoundationConference,UseofShotcreteASCEACI,SP45,pp.175199.Cosma,C.,Olsson,O.,Keskinen,J.andHeikkinen,P.,2001.Seismiccharacterizationof fracturing at the Aspo Hard Rock laboratory, Sweden, from the kilometer100scale to the meter scale. International Journal of Rock Mechanics and MiningSciences,38(6):859865.Cundall,P.A.,Potyondi,D.O.andLee,C.A.,1996.Micromechanicsbasedmodelsforfracture and breakout around the Mineby Tunnel. In: J.B. Martino and C.D.Martin (Editors), Proceedings of the Excavation Disturbed Zone Workshop Designing the Excavation Disturbed Zone for a Nuclear Waste Repository inHardRock,Winnipeg,Manitoba,Canada.,pp.113122.Davies, C. and Bernier, F., 2003. Impact of the Excavation Disturbed or DamagedZone (EDZ) on the performance of the radioactive waste geologicalrepositories, Proceedings of a European Commission CLUSTER ConferenceandWorkshop,Luxembourg.Davies, N. and Mellor, D., 1996. Review of excavation disturbance measurementsundertaken within the ZEDEX project: Implications for the Nirex RockCharacterizationFacility,Eurock96.Balkema:Rotterdam,pp.pp.13451352.Deng, J.H., Lee, C.F. and Ge, X.R., 2001. Characterization of the disturbed zone in alarge rock excavation for the Three Gorges Project, China. Can. Geotech. J. ,38:95101.Diederichs, M.S., 1999. Instability of hard rock masses: The role of tensile damageandrelaxation.Ph.D.Thesis,Ph.D.Thesis,UniversityofWaterloo,566pp.Diederichs, M.S.and Kaiser,P.K., 1999.Tensilestrength andabutmentrelaxation asfailurecontrolmechanismsinundergroundexcavations.InternationalJournalofRockMechanicsandMiningSciences,36(1):6996.Diederichs,M.S.,2003.ManuelRochaMedalRecipientRockFractureandCollapseUnderLowConfinementConditions.RockMechanicsandRockEngineering,36(5):339381.Diederichs,M.S.,Carvalho,J.L.andCarter,T.G.,2007.AmodifiedapproachfropredictionofstrengthandpostyieldbehaviourforhighGSIrockmassesinstrong,brittleground.In:Eberhardt,SteadandMorrison(Editors),RockMechanics:MeetingSocietysChallengesandDemands.Taylor&Francis.,Vancouver.Fairhurst, C. and Damjanac, B., 1996. The excavation damaged zone aninternational perspective. In: J.B. Martino and M. C.D. (Editors), Proceedingsof the Excavation Disturbed Zone Workshop Designing the ExcavationDisturbed Zone for a Nuclear Waste Repository in Hard Rock, Manitoba,Canada.101Farmer,I.W.,1968.EngineeringPropertiesofRocks.E.&F.N.SponLtd,180pp.FernandezDelgado,G.,Mahar,J.W.andParker,H.W.,1976.Structuralbehaviourofthinshotcretelayersobtainedfromlargescalestests,EngineeringFoundationConference,ShotcreteforGroundSupport,TidewaterInnEaston,Maryland,USA,pp.339442.Fishman, Y.A. and Lavrov, V.E., 1996. Designing of rock supports of tunnels on thebasisofgeneralizeddataonthicknessofdistressedzone.In:G.Barla(Editor),Eurock96.Balkema:Rotterdam,pp.915921.Fjellborg, S. and Olsson, M., 1996. Long drift rounds with large cut holes at LKAB.SveBeFoReportNo.27,SwedishRockEngineeringResearch,Stockholm.Forsyth,W.W.(Editor),1993.Adiscussionontheblastinducedoverbreakaroundundergroundexcavations.RockFragmentationbyBlasting,FRAGBLAST4,161166pp.Forsyth,W.W.andMoss,A.E.(Editors),1991.Investigationofdevelopmentblastingpractices.CANMETMRL,91143pp.Fossum, A.F., 1985. Effective elastic properties of rock mass strength. Int. J. RockMech.Sci.Geomech.Abstr.,22:467470.Goodman,R.E.,1989.IntroductiontoRockMechanics.Wiley,NewYork.Grodner, M., 1999. Fracturing around a preconditioned deep level gold mine stope.GeotechnicalandGeologicalEngineering,17(34):291304.Hajiabdolmajid,V.,Kaiser,P.andMartin,C.D.,2003.Mobilisedstrengthcomponentsinbrittlefailureofrock.Geotechnique,53(3):327336.Hajiabdolmajid, V., Kaiser, P.K. and Martin, C.D., 2002. Modeling brittle failure ofrock. International Journal of Rock Mechanics & Mining Sciences, 39(6): 731741.Huang,H.,Detournay,E.andBellier,B.,1999.Discreteelementmodelingofrockcutting.In:Amadei,Kranz,ScottandSmeallie(Editors),37thUSRockMechanicsSymposium:RockMechanicsforIndustry.Rotterdam:Balkema,Vail,Colorado.Hildyard, M.W. and Young, R.P., 2002. Modeling seismic waves aroundundergroundopeningsinfracturedrock.PureandAppliedGeopyhsics159(13):247276.102Hoek, E., 1983. Twentythird Rankine lecturer. Strength of jointed rock masses.Geotechnique,33:185222.Hoek, E. and Brown, E.T., 1980. Empirical strength criteria for rock masses. J.Geotech.Engng.Div.,ASCE.,106:10131035.Hoek, E. and Brown, E.T., 1997. Practical estimates of rock mass strength.International Journal of Rock Mechanics and Mining Sciences, 34(8): 11651186.Hoek, E., CarranzaToress, C. and Corkum, B., 2002. HoekBrown failure criterion 2002edition,Proc.5th.NorthAmericanRockMechanicsSymposiumand17thTunneling Association of Canada Conference. ATMTAC 2002. University ofToronto,UniversityofToronto,pp.267271.Holcomb, D.J., MacDonald, T. and Hardy, R.D., 2003. Using ultrasonic waves toassess the disturbed rock zone (DRZ) in an alcove corner excavated in salt atthe WIPP (Waste Isolation Pilot Plant). In: J.B. Martino (Editor), The 2002International EDZ Workshop. Atomic Energy of Canada Limited, Ontario,Canada.Holmberg,R.andPersson,P.A.,1980.Designoftunnelperimeterblastholepatternstopreventrockdamage.Transc.Inst.Min.Metall.:A37A40.Holmgren,J.,1979.PunchloadedshotcreteliningsonhardrockPhDDissertation.RoyalInstituteofTechnology.BeFoReportNo.7:2/79,SwedishRockEngineeringResearchFoundation,Stockholm.Holmgren,J.,1985.Boltachoredsteelfibrereinforcedshotcretelinings.ReportNo.BeFo73:1/85orFortFA2:85,SwedishRockEngineeringResearchFoundation(BeFo)andRoyalSwedishFortificationAdministration(FortF),Stockholm.Itasca,2002.FLAC(Version4.00).ItascaConsultingGroup,Minneapolis,Minnesota.Itasca,2004.PFC2D(Version3.10).ItascaConsultingGroup,Minneapolis.Itasca,2005.FLAC(Version5.00).ItascaConsultingGroup,Minneapolis.Itasca,2008.PFC2D(Version4.00):Prerelease.ItascaConsultingGroup,Minneapolis.http://www.soc.nii.ac.jp JapanTunnelingAssociation. (dateaccessed:Sept.2004).Koopmans, R. and Hughes, R.W., 1986. The effect of stress on the determination ofdeformation modulus, Workshop on excavation response in deep radioactive103wasterepositories;implicationforengineeringdesignandsafetyperformance(OECD/AECC),Winnipeg,Canada,pp.2628.Krauland,N.andSder,P.E.,1988.RockstabilisationbydestressblastingexperiencesfromBoldienmines,RockMechanicsMeeting,BeFo,Stockholm,Sweden.LadegaardPedersen,A.andDaly,J.W.,1975.Areviewoffactorsaffectingdamageinblasting,Mech.Engg.Dept.,Univ.ofMaryland.Laubscher,D.H.andTaylor,H.W.,1990.Ageomechnicsclassificationsystemfortheratingofrockmassinminedesign.J.S.Afr.Inst.Min.Metall.,90(10):257273.Lightfoot, N. et al., 1996. Preconditioning to reduce the incidence of face bursts inhighlystressedfaces,CSIR,DivisionofMiningTechnology,SIMRACReport,GAP030.Luke, B.A., Stokoe, K.H., II, Bay, J.A., Lee, N.J. and Nelson, P.P., 1999. Seismicmeasurements to investigate disturbed rock zones. Geotechnical SpecialPublication(90):303314.MacKown,A.F.,1986.Perimetercontrolledblastingforundergroundexcavationsinfracturedandweatheredrocks.Bull.Assoc.Engg.Geol.,XXIII(4):461478.Malmgren,L.,2001.shotcreterocksupportexposedtovaryingloadingconditions,LuleUniversityofTechnology,Lule,136pp.Malmgren,L.,2005.Interactionbetweenshotcreteandrockexperimentalandnumericalstudy,DoctoralThesis,LuleUniversityofTechnology,Lule.Malmgren,L.,Saiang,D.,Toyra,J.andBodare,A.,2007.Theexcavationdisturbedzone(EDZ)atKiirunavaaramine,Swedenbyseismicmeasurements.JournalofAppliedGeophysics,61(1):115.Martin,C.D.andChandler,N.A.,1994.ProgressivefractureofLacduBonnetgranite.InternationalJournalofRockMechanicsandMiningSciences,31(6):643659.Martin, C.D., Kaiser, P.K. and McCreath, D.R., 1999. HoekBrown parameters forpredicting the depth of brittle failure around tunnels. Canadian GeotechnicalJournal,36(1):136151.Martino,J.B.,2003a.The2002InternationalEDZWorkshop:Theexcavationdamagedzonecauseandeffects,AtomicEnergyofCanadaLimited.104Martino,J.B.,2003b.TheexcavationdamagedzoneandrecentstudiesattheURL.In:J.B. Martino (Editor), Proceedings of the 2002 International EDZ Workshop,Toronto.Martino, J.B. and Chandler, N.A., 2004. Excavationinduced damage studies at theUnderground Research Laboratory. International Journal of Rock MechanicsandMiningSciences,41(8SPECISS):14131426.Martino, J.B. and Martin, C.D. (Editors), 1996. Proceedings of the ExcavationDisturbed Zone Workshop Designing the Excavation Disturbed Zone for aNuclearWasteRespositoryinHardRock,Winnipeg,Manitoba,Canada.Mas Ivars, D., Deisman, N., Pierce, M. and Fairhurst, C., 2007. The synthetic rockmass approach a step forward in the characterisation of jointed rock mass,11thInternationalSocietyofRockMechanicsSymposium.TaylorandFrancis,Lisbon,Portugal,pp.485490.Maxwell, S.C. and Young, R.P., 1998. Propagation effects of an undergroundexcavation.Tectonophysics,289(13):1730.Meglis, I.L., Chow, T., Martin, C.D. and Young, R.P., 2005. Assessing in situmicrocrack damage using ultrasonic velocity tomography. InternationalJournalofRockMechanicsandMiningSciences,42(1):2534.Monsen, K. and Barton, N., 2001. A numerical study of cryogenic storage inunderground excavations with emphasis on the rock joint responseInternationalJournalofRockMechanics&MiningSciences,38(7):10351045Mostkov,V.M.,1979.USSRexperienceinconstructionofhydraulictunnels,ElectricPowerResearchInstitute(Report)EPRIEL.PergamonPress,NewYork,NY,Tokyo,Jpn,pp.8188.Nyberg,U.andFjellborg,S.,2002.Controlleddriftingandestimatingblastdamage.In:R.Holmberg(Editor),ProceedingsFirstWorldConferenceonExplosivesandBlastingTechnique.Ba