Numerical simulations of rock mass damage induced by underground explosion

8
Numerical simulations of rock mass damage induced by underground explosion X.Y. Wei a,b , Z.Y. Zhao a, , J. Gu a a School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639798, Singapore b School of Civil Engineering, Chang An University, Xi’an 710061, China article info Article history: Received 7 August 2008 Received in revised form 3 February 2009 Accepted 13 February 2009 Available online 21 March 2009 Keywords: Rock mass damage Explosion Loading density Peak particle velocity Damage criterion abstract The damage prediction of rock mass under blast loads induced by accidental explosions, rock bursts or weapon attacks is crucial in rock engineering. In this paper, parametric studies are conducted to evaluate the effect of loading density, rock mass rating (RMR) and weight of charge on the rock mass damage induced by underground explosions. The numerical simulations are carried out based on the transient dynamic finite element program ANSYS-LSDYNA. The numerical model was calibrated against the data obtained from a field blast test. A fully coupled numerical analysis, incorporating the explosion process, has been performed, where the large deformation zone near the charge is solved by the Arbitrary Lagrange–Euler (ALE) method. The deformable modulus and compressive strength of rock mass of granite are estimated by the RMR system. The peak particle velocity (PPV) damage criterion and the plastic strain criterion were adopted to study the damage zone around the charge hole, and an empirical formula considering the effects of loading density, RMR and weight of charge was obtained to estimate the damage zone in granite based on the numerical results. & 2009 Elsevier Ltd. All rights reserved. 1. Introduction The safety and stability of underground rock caverns or tunnels are often affected by blast induced damages and vibrations. The source of the blast loading may come from an accidental explosion, the drill/blast excavation method or weapon attacks. A reliable estimation of the damage zone in the rock mass under blast loads is the first step in design and stability assessment of underground structures. The amount of damage in rock mass will depend on the distance from the explosive charge, the weight of the charge, the rock mass properties and the joint distribution. However, it is quite difficult to include all those factors in the analysis because the problem involves many complicated physical processes, such as the explosion of the charge, the damage initiation and propagation. Some attempts have been made to investigate the damage of rock mass under the blast loading, and generally the analysis can be characterized by two main methods, i.e. the empirical method and the numerical simulation. Some field blast tests have been carried out and the experimental data and phenomena were recorded. Empirical damage criterion, usually based on peak particle velocity (PPV), was obtained from these observations and measurements. For example, the US Army Corps of Engineers conducted some large explosion tests between 1948 and 1952 near unlined tunnels in sandstone [1]. Four zones of failure were categorized in terms of the PPV. They found that the threshold PPV corresponding to an intermittent failure is 0.46 m/s. Kendorski et al. [2] reported that cracks in the shotcrete liner of a tunnel start to develop when the PPV reaches approximately 1.22m/s in an experiment conducted at the Cimax, Colorado. It was reported by Rupert and Clark [3] that only minor damage in the form of localized thin spalls and collapse of previously fractured coal ribs resulted from blast vibration exceeding 0.05 m/s. Siskind and Fumanti [4] concluded that the blast induced cracks are a function of rock type, and that the cracking zone will extend to 4.9–20 times the radius of the charge hole while the crush zone will extend to 1.9–4.5 times the charge hole for a fully coupled TNT charge in granite. Jensen et al. [5] found that no roof failure occurs even at vibration level of 0.445 m/s, and only a few loose stones are observed at 0.127 m/s. Kidybinski [6] reported that small roof falls of a underground coal mine openings may occur when PPV lies in the range of 50–100 mm/s and large roof falls at the PPV of 100–200 mm/s. Fourie and Green [7] reported that no significant damage was found when the PPV reaches 110mm/s. Singh [8] investigated the threshold value of PPV for the safety of underground workings based on rock mass rating (RMR) of the roof rock, and three groups of damage, i.e., major damage, minor damage and no damage, were classified according to the PPV damage criterion. Most empirical formulae available in the literature usually work well for certain specific types of rock mass and do not include the ARTICLE IN PRESS Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/ijrmms International Journal of Rock Mechanics & Mining Sciences 1365-1609/$ - see front matter & 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijrmms.2009.02.007 Corresponding author. Tel.: +65 67905255. E-mail address: [email protected] (Z.Y. Zhao). International Journal of Rock Mechanics & Mining Sciences 46 (2009) 1206–1213

description

The damage prediction of rock mass under blast loads induced by accidental explosions, rock bursts orweapon attacks is crucial in rock engineering. In this paper, parametric studies are conducted toevaluate the effect of loading density, rock mass rating (RMR) and weight of charge on the rock massdamage induced by underground explosions. The numerical simulations are carried out based on thetransient dynamic finite element program ANSYS-LSDYNA. The numerical model was calibrated againstthe data obtained from a field blast test. A fully coupled numerical analysis, incorporating the explosionprocess, has been performed, where the large deformation zone near the charge is solved by theArbitrary Lagrange–Euler (ALE) method. The deformable modulus and compressive strength of rockmass of granite are estimated by the RMR system. The peak particle velocity (PPV) damage criterion andthe plastic strain criterion were adopted to study the damage zone around the charge hole, and anempirical formula considering the effects of loading density, RMR and weight of charge was obtained toestimate the damage zone in granite based on the numerical results.

Transcript of Numerical simulations of rock mass damage induced by underground explosion

  • mgapo

    Keywords:

    Rock mass damage

    rock

    in

    ing

    gro

    eme

    the data obtained from a eld blast test. A fully coupled numerical analysis, incorporating the explosion

    process, has been performed, where the large deformation zone near the charge is solved by the

    rounddamagcommeth

    and phenomena were recorded. Empirical damage criterion,

    argel. [5]el ofm/s.coal

    Fourie and Green [7] reported that no signicant damage was

    ARTICLE IN PRESS

    Contents lists availab

    .els

    al&

    International Journal of Rock Mechanics & Mining Sciences 46 (2009) 12061213damage, were classied according to the PPV damage criterion.Most empirical formulae available in the literature usually workusually based on peak particle velocity (PPV), was obtained fromthese observations and measurements. For example, the US ArmyCorps of Engineers conducted some large explosion tests between

    found when the PPV reaches 110mm/s. Singh [8] investigated thethreshold value of PPV for the safety of underground workingsbased on rock mass rating (RMR) of the roof rock, and threegroups of damage, i.e., major damage, minor damage and no

    well for certain specic types of rock mass and do not include theCorresponding author. Tel.: +6567905255.

    E-mail address: [email protected] (Z.Y. Zhao).1365-16

    doi:10.1i.e. the empirical method and the numerical simulation. Someeld blast tests have been carried out and the experimental data

    mine openings may occur when PPV lies in the range of50100mm/s and large roof falls at the PPV of 100200mm/s.analysis because the problem involves many complicated physicalprocesses, such as the explosion of the charge, the damageinitiation and propagation. Some attempts have been made toinvestigate the damage of rock mass under the blast loading, andgenerally the analysis can be characterized by two main methods,

    hole while the crush zone will extend to 1.94.5 times the chhole for a fully coupled TNT charge in granite. Jensen et afound that no roof failure occurs even at vibration lev0.445m/s, and only a few loose stones are observed at 0.127Kidybinski [6] reported that small roof falls of a undergroundA reliable estimation of the damage zone in the rock mass underblast loads is the rst step in design and stability assessment ofunderground structures. The amount of damage in rock mass willdepend on the distance from the explosive charge, the weight ofthe charge, the rock mass properties and the joint distribution.However, it is quite difcult to include all those factors in the

    Colorado. It was reported by Rupert and Clark [3] that only minordamage in the form of localized thin spalls and collapse ofpreviously fractured coal ribs resulted from blast vibrationexceeding 0.05m/s. Siskind and Fumanti [4] concluded that theblast induced cracks are a function of rock type, and that thecracking zone will extend to 4.920 times the radius of the chargePeak particle velocity

    Damage criterion

    1. Introduction

    The safety and stability of undergare often affected by blast inducedsource of the blast loading mayexplosion, the drill/blast excavation09/$ - see front matter & 2009 Elsevier Ltd. A

    016/j.ijrmms.2009.02.007mass of granite are estimated by the RMR system. The peak particle velocity (PPV) damage criterion and

    the plastic strain criterion were adopted to study the damage zone around the charge hole, and an

    empirical formula considering the effects of loading density, RMR and weight of charge was obtained to

    estimate the damage zone in granite based on the numerical results.

    & 2009 Elsevier Ltd. All rights reserved.

    rock caverns or tunnelses and vibrations. Thee from an accidentalod or weapon attacks.

    1948 and 1952 near unlined tunnels in sandstone [1]. Four zonesof failure were categorized in terms of the PPV. They found thatthe threshold PPV corresponding to an intermittent failure is0.46m/s. Kendorski et al. [2] reported that cracks in the shotcreteliner of a tunnel start to develop when the PPV reachesapproximately 1.22m/s in an experiment conducted at the Cimax,Explosion

    Loading density

    Arbitrary LagrangeEuler (ALE) method. The deformable modulus and compressive strength of rockNumerical simulations of rock mass daexplosion

    X.Y. Wei a,b, Z.Y. Zhao a,, J. Gu a

    a School of Civil and Environmental Engineering, Nanyang Technological University, Sinb School of Civil Engineering, Chang An University, Xian 710061, China

    a r t i c l e i n f o

    Article history:

    Received 7 August 2008

    Received in revised form

    3 February 2009

    Accepted 13 February 2009Available online 21 March 2009

    a b s t r a c t

    The damage prediction of

    weapon attacks is crucial

    evaluate the effect of load

    damage induced by under

    transient dynamic nite el

    journal homepage: www

    InternationRock Mechanicsll rights reserved.age induced by underground

    re 639798, Singapore

    mass under blast loads induced by accidental explosions, rock bursts or

    rock engineering. In this paper, parametric studies are conducted to

    density, rock mass rating (RMR) and weight of charge on the rock mass

    und explosions. The numerical simulations are carried out based on the

    nt program ANSYS-LSDYNA. The numerical model was calibrated against

    le at ScienceDirect

    evier.com/locate/ijrmms

    Journal ofMining Sciences

  • charge is solved by the Arbitrary LagrangeEuler (ALE) method.

    2.1. Material models

    The material models used in this problem include explosive(TNT), air and rock mass.

    2.1.1. TNT

    Material Type 8 of LS-DYNA (*MAT_HIGH_EXPLOSIVE_BURN)is used for TNT and the JonesWilkensLee (JWL) equation of state(EOS) is used to calculate the pressure generated by the expansionof the detonation products of the chemical explosive. The JWL EOSdenes the pressure as

    p A 1 oR1V

    eR1V B 1 o

    R2V

    eR2V oE

    V(3)

    where A, B, R1, R2 and o are the material constants, p is thepressure, V is the relative volume of detonation product, E is thespecic energy with an initial value of E0. The material parametersfor TNT used in the present study are as follows: A 3.738102GPa, B 3.747GPa, R1 4.15, R2 0.9,o 0.35, E0 6.0GPa. Thedetonation velocity and mass density of TNT is 6.93103m/s and1.63103 kg/m3, respectively.

    ARTICLE IN PRESS

    0

    Target in rock mass

    y

    A

    6m

    2.5m

    2.5m2.5m

    2.5m

    5m

    Concrete

    Explosive

    15m 25mAir

    0.8m

    Fig. 1. Conguration of measurements.

    echThe deformable modulus and compressive strength of rock massof granite are estimated by the RMR system. The numerical modelis calibrated by comparing the PPV from the numerical results andthe experimental data. Parametric studies are performed tond out the most inuential factors on the damage of rock mass,such as the RMR of rock mass and the loading density. The PPVdamage criterion related to the RMR of rock mass was chosenaccording to the test data available in the literature. The empiricalformula for the damage depth into rock under blast loading wasderived based on the numerical results. The effect of loadingdensity, weight of charge and RMR of rock mass was considered inthe formula.

    2. Numerical model and calibration

    A series of eld blast tests at a site of Bukit Timah granite wasconducted in the previous study [11,12]. The detailed geologicalinvestigation and intensive laboratory tests on the granite werealso carried out. Fig. 1 shows the eld layout of the test. Thecharge hole has the following dimensions: total depth is 11m, andthe diameters of the charge hole are 1.5m and 0.8m for the upper6m and bottom 5m, respectively. Measurement points werelocated on three lines on the ground surface, i.e., the predominantrock joint direction, the perpendicular direction and the 451direction. Nine measurement points were on the parallel line withdistance of 2.5, 5, 10, 25 and 50m from the center of charge holeon the ground surface and 8.5m below the surface. Eight testswere carried out with equivalent TNT charge weights rangingfrom 5 to 50kg [12]. In each test, explosive was placed at thecenter of the charge hole. The recorded data were used to deriveempirical attenuation relations for PPV and peak particle accel-eration (PPA). The best tted empirical attenuation relations aregiven in the following [12]:

    PPA 1928:2R=Q1=31:4531 g (1)

    PPV 0:396R=Q1=31:1455 m=s (2)effects of loading density, the chamber geometry and theexplosive distribution.

    With the rapid development of computer technology andadvanced numerical techniques, more detail and reliable predic-tions of rock mass damage under blast loading through numericalsimulations have become available. Various numerical methodshave been proposed to estimate the rock mass damage induced bythe blast loading. Yang et al. [9] developed an isotropic cumulativedamage model and calibrated the model by comparing thenumerical results with the blasting experiments. Wu et al. [10]presented an anisotropic damage model to predict the damagezone around the charge hole. Some attempts have been made toestablish an empirical formula to estimate the damage zone ofrock mass including the loading density and chamber geometry[10], where various loading densities, chamber geometries andexplosive distribution patterns are considered in their numericalmodels, and the obtained formula is established based on the PPVcriterion and the damage index criterion.

    In this paper, a numerical method has been used to predict thedamage depth into rock mass induced by underground explosion.The numerical simulations are carried out based on the transientdynamic nite element program ANSYS-LSDYNA. A fully couplednumerical analysis, which includes the explosion process, hasbeen performed, where the large deformation zone near the

    X.Y. Wei et al. / International Journal of Rock Mwhere R is distance in meter measured from the charge center; Qis equivalent TNT charge weight in kilogram.Plane View

    Target on rock surface

    Z

    1.5mSensor

    45Charge hole

    Predominantrock joint set

    90x

    25.0m

    15.0m

    5.0m2.5m2.5manics & Mining Sciences 46 (2009) 12061213 12072.1.2. Air

    Material Type 9 of LS-DYNA(*MAT_NULL) is used for the air.This material model must be used with an EOS. The polynomial

  • The compressive failure is governed by J2 ow. The detailed

    numerical model. An ALE zone is dened for the area around

    3. PPV damage criterion

    As stated in a previous study [10], it is very difcult to obtain auniversally accepted damage criterion since it involves manyfactors. For practical applications, the PPV damage criteria arewidely used because it can be easily recorded at a site. Adhikariet al. [16] reported the range of PPV and the correspondingdamage in two operating mines as shown in Table 1. Persson [17]reported the PPV damage criterion for Swedish hard rock asshown in Table 2. Li and Huang [18] reported the PPV damage

    ARTICLE IN PRESS

    0.01

    peak

    scaled distance (m/kg1/3)

    Test results [12]

    1 10

    Fig. 2. Peak particle velocity versus scaled distance.

    100

    1000

    peak

    par

    ticle

    acc

    eler

    atio

    n (g

    )

    scaled distance (m/kg1/3)

    Numerical resultsTest results [12]

    1 10

    Fig. 3. Peak particle acceleration versus scaled distance.

    Table 1Threshold value of PPV (mm/s) for underground structures [16].

    Type of damage Fair rock (RMR 60) Poor rock (RMR 38)

    No damage o153 o52Opening of joints 153217 52195

    Falling of loose pieces 217367 195297

    Induced cracking 367604 297557

    Excessive damage 4604 4557

    echthe charge. The smallest size elements are used in the region nearthe charge, while larger elements are used for the other region.There are a total of 424752 elements in the mesh, with thesmallest element being 50mm50mm50mm. The cylindricalcharge with equivalent TNT weight 40 kg is used in the model. Thediameter and the height of the cylindrical charge are 0.358 and0.245m, respectively.

    The PPV and the PPA obtained from the numerical simulationsare compared with experimental results, as shown in Figs. 2 and 3.It can be found that the PPV from the numerical results agree wellwith the eld test results when the scaled distance is less thandescription of this model is given in [13].Laboratory tests were conducted on the Bukit Timah granite at

    the eld site. It was found that the granite at the site is of goodquality. The average density of the rock is 2650kg/m3, Poissonsratio is 0.16 [12], Youngs modulus of intact granite is 73.9GPa, theuniaxial compressive strength is 186MPa, and the tensile strengthis 16.1MPa [14]. The previous study also found that the initialdamage in three directions in the rock mass is 0.162, 0.124, 0.222,respectively [15]. The mean initial damage 0.169 is adopted in thisstudy. Thus, the equivalent elastic modulus, compressive strengthand tensile strength for the granite can be determined as61.41GPa, 154.57MPa and 13.38MPa, respectively. There is noavailable test data for the fracture toughness gc and viscosity Z forrock mass, so a parameter study is carried out to adjust the propervalues for granite by comparing the numerical results with thetest data.

    2.2. Numerical results

    Only a quarter of the eld is modeled owing to the symmetryabout the yz and xz planes as shown in Fig. 1. The non-reectionboundaries are applied to minimize the stress wave reection atthese computational boundaries. The computational modelshould be large enough to minimize the effect of the boundaryreection. The model with size of 60m60m35m are usedhere. In order to minimize the effect of mesh size on thenumerical results, three different meshes are tried for theEOS is used in the present study. It has the form

    P C0 C1m C2m2 C3m3 C4 C5m C6m2e (4)where e is the internal energy per volume. The compression of thematerial is dened by the parameter m (r/r0)1 with r and r0being the current and initial density of the material, respectively.

    The air is often modeled as the ideal gas by settingC0 C1 C2 C3 C6 0 and C4 C5 g1 with g is the ratioof specic heat. In the present study, the standard materialconstants of air g 1.4, r 1.225kg/m3 are used.

    2.1.3. Rock mass

    There are several material models for concrete and rockimplemented in LS-DYNA, designed for special purposes to takeinto account the erosion, effect of strain rate, cracking, etc. In thepresent study, the Mat_brittle_damage in LS-DYNA library is used.This anisotropic damage model is suitable for brittle failure ofrock and concrete and contains a minimal set of materialconstants, which can be determined from the standard tests. Itadmits progressive degradation of tensile and shear strengthacross smeared cracks that are initiated under tensile loadings.

    X.Y. Wei et al. / International Journal of Rock M12084.0m/kg1/3, while a slightly larger difference can be observedwhen the scaled distance is larger than 4.0. The similar trend forthe PPA can be found, as shown in Fig. 3.0.1

    1

    parti

    cle

    velo

    city

    (m/s

    )

    Numerical results

    anics & Mining Sciences 46 (2009) 12061213criterion for hard rock and soft rock as shown in Table 3 with thefollowing denition: slight damageinitial damage; mediumdamagepartial collapse; serious damagelarge area tunnel

  • collapse. The threshold values of PPV at different RMR of roof rockof seven coal mines are reported by Singh [8], as shown in Table 4.The damage is classied into three groups, i.e. major damage,minor damage and no damage. The vibration levels in no damagezone are taken as safe level of vibration, and the threshold value ofvibration for the safety of underground workings is recom-mended. A comprehensive study of tunnel damage for sandstoneis carried by the US Armys Underground Explosion Tests (UET), asreported by Hendron [1]. In this report, the damage is classiedinto four groups: intermittent failure, local failure, general failureand tight closure. Table 5 shows that tunnel PPV damage criterionfor different damage zones. The intermittent damage is observedwhen random spalling of loose rocks occurs.

    It can be observed from the literature that the threshold PPVvalue under no damage condition is quite different because thequality of intact rock and rock mass is different from site to site.The PPV damage criteria for various rock masses were predictedusing articial neural network by Zhao et al. [19]. The tensilestrength, compressive strength of the intact rock and the RMR areused as the input data to predict the threshold value of PPV. Based

    to investigate the effect of loading density on the pressure aroundthe wall of the charge hole and the corresponding damage zone.

    obtained by varying the weight of TNT as 311.9, 559.9, 808.0,1056.0 and 5216kg while keeping the volume of charge hole3.2m3 unchanged.

    The damage zone is measured from the wall of the charge holeaccording to the PPV damage criterion as stated in Section 3. Aplastic strain damage criterion is also used and the damage zone ismeasure from the wall of the charge hole to the location withplastic strain. The results are compared with that of PPV damagecriterion.

    Fig. 4 shows an element pressure around the charge hole forvarious loading densities. It can be noted that the peak pressure of

    ARTICLE IN PRESS

    gth

    Table 4Threshold value of PPV for the safety of underground workings [8].

    RMR of roof rock Threshold value of PPV (mm/s)

    2030 50

    3040 5070

    4050 70100

    5060 100120

    6080 120

    Table 5UET tests, sandstone [1].

    Damage zone Damage PPV (m/s)

    1 Tight closure NA

    2 General failure 12

    3 Local failure 4

    4 Intermittent failure 0.91.8

    0.0000-500

    0

    500

    1000

    1500

    2000

    2500

    pres

    sure

    (MP

    a)

    Time (s)

    loading density=1630 (kg/m3)

    loading density=407.5 (kg/m3)

    loading density=83.2 (kg/m3)

    loading density=20 (kg/m3)

    0.0005 0.0010 0.0015 0.0020

    X.Y. Wei et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 12061213 1209The loading densities 20.0, 83.2 and 407.5 kg/m3 are obtained bydening the length of the charge hole as 5.00, 1.20 and 0.245mwhile keeping the weight of TNT as 64 kg. The loading densities of97.5, 175.0, 252.5, 330.0 and 1630kg/m3 (fully coupled) are

    Table 2Damage criterion for Scandinavian bed rocks [17].

    Typical effect PPV (m/s)

    Incipient swelling 0.7

    Incipient damage 1.0

    Fragmentation 2.5

    Good fragmentation 5.0

    Crushing 15.0

    Table 3Tunnel damage criterion for unlined rock tunnels [18].

    Rock type Rock parameter

    Unit weight

    (g/cm3)

    Compressive

    strength (MPa)

    Tensile stren

    (MPa)

    Hard rock I 2.62.7 75110 2.13.4

    Hard rock II 2.72.9 110180 3.45.1on the results of the articial neural network, the threshold valueof PPV corresponding to incipient damage is chosen as: 0.5m/swhen RMR is smaller than 60, 0.7m/s when RMR lies between 60and 80, 0.9m/s when RMR is larger than 80.

    4. Effect of loading density and RMR on blast loading and rockmass damage

    4.1. Effect of loading density

    The dimension of the cubic charge hole is 0.8m0.8m5m.The loading density can be obtained by two ways: (1) varying thelength of the charge hole while keeping the weight of TNTunchanged and (2) varying the weight of TNT while keeping thevolume of charge hole unchanged. These two methods are all usedHard rock III 2.72.9 180200 5.15.7

    Soft rock I 2.02.5 40100 1.13.1

    Soft rock II 2.02.5 100160 3.44.5Threshold value of PPV (m/s)

    No damage Slight

    damage

    Medium

    damage

    Serious

    damage

    0.27 0.54 0.82 1.53

    0.31 0.62 0.96 1.78

    Fig. 4. Pressure around the charge hole of various loading density.0.36 0.72 1.11 2.09

    0.29 0.58 0.90 1.67

    0.35 0.70 1.07 1.99

  • ARTICLE IN PRESS

    echX.Y. Wei et al. / International Journal of Rock M1210the fully coupled charge is about 8 times that of the loadingdensity 20kg/m3, and the peak pressure decreases very quicklyand reaches a negative value. The pressure around the charge holeis not uniform and the location where the maximum pressureoccurs is different for different loading density. The maximumpressure around the charge hole for fully coupled charge is about5.5GPa.

    Fig. 5 shows the plastic contour of different loading density at acertain time step. It indicates that the plastic zone of fully coupledcharge is much larger than that of smaller loading density.

    For a very good quality of rock mass (RMR 95), the PPVdamage criterion 0.9m/s is adopted, the damage depth atdifferent loading density is obtained as

    RPPV 0:306W0:257Q1=3 (5)

    where RPPV is the depth of the damage zone into the rock massmeasured from the wall of charge hole where the particle velocityreaches the threshold PPV,W (kg/m3) is the loading density, Q (kg)is the weight of the equivalent TNT.

    The damage zone is obtained according to the plastic straincriterion in a similar manner, and the empirical relation fordamage depth at different loading density can be obtained as

    Rplastic strain 0:359W0:172Q1=3 (6)

    Wu and Hao [10] investigated the loading density effect andalso gave an empirical equation based on their numerical results.Two damage criteria have been adopted in their study, i.e., PPV

    Fig. 5. Plastic strain contour around the charge hole of various loading densities: (a) fu83.2 kg/m3 and (d) loading density 20 kg/m3.anics & Mining Sciences 46 (2009) 12061213damage criterion (PPVX0.9m/s) and damage index criterion(DX0.71). The rock mass is granite and has the same propertiesas in this study.

    Fig. 6 compares the results from Eqs. (5) and (6) and that of Wuand Hao [10] for TNT weight of 64kg. The damage depth based onthe PPV criterion in this study is slightly larger than the results ofWu [10]. Obviously the damage depth based on the PPV damagecriterion is larger than that of the damage index criterion and theplastic strain criterion. It indicates that even for the locationwhere no plastic strain occurs, the PPV already reaches thethreshold value.

    4.2. Effects of RMR

    In this section, the inuence of rock mass properties on thepressure along the wall of charge hole and the damage zone areinvestigated. It is necessary to estimate the rock mass propertiessuch as deformation modulus and strength parameters fornumerical modeling. Although such parameters can ultimatelybe determined from the in-situ tests, at the preliminary stage,where access to the underground is limited, the practical way toobtain these parameters is to apply a rock mass classicationsystem, such as RQD, RMR, Q and Geological Strength Index (GSI)system.

    The effect of rock mass properties on the damage zone wasconducted based on the RMR system in this study. The uniaxialcompressive strength and elastic modulus are evaluated based on

    lly coupled charge 1630kg/m3, (b) loading density 407.5 kg/m3, (c) loading density

  • ARTICLE IN PRESS

    ion criterion)age Index)

    echanics & Mining Sciences 46 (2009) 12061213 12114

    6

    8

    10

    12

    14

    Dam

    age

    dept

    h in

    to ro

    ck (m

    )

    PPV criterion Plastic strain criter Results of Wu(PPV Results of Wu(Dam

    X.Y. Wei et al. / International Journal of Rock Mthe RMR system. Kalamaras and Bieniawski [20] suggested thefollowing relations

    scj=sci expRMR 100=24 (7)

    Ej=Ei expRMR 100=17:4 (8)where scj is the compressive strength of rock mass, sci is thecompressive strength of intact rock, Ej is the elastic modulus ofrock mass and Ei is the elastic modulus of intact rock.

    Fig. 7 shows the pressure along the wall of the charge holewhen loading density is 83.2 kg/m3. It indicates that the effects ofthe RMR on the peak pressure are very small while it has obviousinuence on the gas pressure. The larger the RMR is, the larger thepressure is. It has very little effects on the peak pressure becauseof the very short arising time while the gas pressure will last for arelatively longer time.

    The effect of RMR, loading density and the weight of TNT are allincluded in the numerical simulation to investigate their effectson the damage depth into rock mass. Four types of RMR (95, 80,

    0.000

    0

    40

    80

    120

    160

    200

    pres

    sure

    (MP

    a)

    time (s)

    element 81862RMR=95RMR=80RMR=60RMR=40RMR=20RMR=0

    0.002 0.004 0.006 0.008 0.010

    Fig. 7. Effect of RMR on the blast loading along the charge hole.

    0 200 400 600 800 1000 1200 1400 1600 18000

    2

    Loading density (kg/m3)

    Fig. 6. Comparison of damage depth according to different damage criterion.

    02

    4

    6

    8

    10

    12

    14

    16

    Dam

    age

    dept

    h in

    to ro

    ck (m

    )

    Loading density (kg/m3)

    Q=64kgRMR=95RMR=80RMR=60RMR=40

    200 400 600 800 1000 1200 1400 1600 1800

    02

    4

    6

    8

    10

    12

    14

    16

    Dam

    age

    dept

    h in

    to ro

    ck (m

    )

    Loading density (kg/m3)

    Q=64kgRMR=95RMR=80RMR=60RMR=40

    200 400 600 800 1000 1200 1400 1600

    Fig. 8. Damage depth into rock versus loading density and RMR: (a) according toPPV damage criterion and (b) according to plastic strain criterion.

  • 60, 40), eight loading densities and six different weight of TNT asstated in Section 4.1 are considered in this study. Similarly, thedamage depth into rock mass under blast loading including theeffect of RMR, loading density and weight of charge can beobtained. All the obtained numerical results were used to derivethe empirical relations for damage depth. The best tted empiricalrelation is as follows:

    RPPV 3:907W0:257RMR0:578Q1=3 (9)

    Usually the scaled distance is often used to include the effect ofweight of charge and distance. If we dened the threshold scaleddistance as Z R/Q1/3, then the above equation can be expressedas

    ZPPV RPPV=Q1=3 3:907W0:257RMR0:578 (10)

    Similarly, the damage zone can be obtained according to theplastic strain damage criterion. The damage depth and thecorresponding threshold scaled distance at different loadingdensity are obtained as

    Rplastic strain 5:598W0:172RMR0:534Q1=3 (11)

    contract RPTJ0275030 for US Bureau of Mines; 1979. p. 99 .

    ARTICLE IN PRESS

    1

    2

    3

    4

    Thre

    shol

    d sc

    aled

    dis

    tanc

    e (m

    /kg1

    /3)

    RMR=95RMR=80RMR=60RMR=40

    X.Y. Wei et al. / International Journal of Rock Mech12120

    Loading density (kg/m3)

    -200

    1

    2

    3

    4

    Thre

    shol

    d sc

    aled

    dis

    tanc

    e (m

    /kg1

    /3)

    Loading density (kg/m3)

    RMR=95RMR=80RMR=60RMR=40

    200 400 600 800 1000 1200 1400 1600

    0 200 400 600 800 1000 1200 1400 1600Fig. 9. Threshold scaled distance versus loading density and RMR: (a) according toPPV damage criterion and (b) according to plastic strain criterion.[6] Kidybinski A. Design criteria for roadway supports to resist dynamic loads. IntJ Min Geol Eng 1986;4:91109.

    [7] Fourie AB, Green RW. Damage to underground coal mines caused by surfaceblasting. Int J Surf Min 1993;7:116.

    [8] Singh PK. Blast vibration damage to underground coal mines from adjacentopen-pit blasting. Int J Rock Mech Min Sci 2002;39(8):95973.

    [9] Yang R, Bawden WF, Katsabanis PD. A new constitutive model for blastdamage. Int J Rock Mech Min Sci 1996;33(3):24554.

    [10] Wu C, Hao H. Numerical prediction of rock mass damage due to accidentalexplosions in an underground ammunition storage chamber. Shock Waves2006;15(1):4354.

    [11] Wu YK, Hao H, Zhou YX. Propagation characteristics of blast-induced shockwaves in a jointed rock mass. Soil Dyn Earthquake Eng 1998;17(6):40712.

    [12] Hao H, Wu C, Zhou Y. Numerical analysis of blast-induced stress waves in arock mass with anisotropic continum damage models. Part 1: equivalentmaterial property approach. Rock Mech Rock Eng 2002;35(2):7994.

    [13] Govindjee S, Kay GJ, Simo JC. Anisotropic modeling and numerical simulationof brittle damage in concrete. Int J Numer Methods Eng 1995;38(21):361133.

    [14] Hao H, Wu YK, Ma GW, Zhou YX. Characteristics of surface ground motionsinduced by blasts in jointed rock mass. Soil Dyn Earthquake EngZplastic strain Rplastic strain=Q1=3 5:598W0:172RMR0:534 (12)Figs. 8a and b show the relationship between the damage depth

    and the loading density and RMR based on PPV criterion andplastic strain criterion, respectively, when the weight of TNT is64 kg. Obviously the loading density and RMR all have signicantinuences on the damage zone. The larger the loading density, thelarge the damage zone is. It is straightforward to estimate thedamage depth under various loading density with different RMRfrom those gures.

    The relationship between the threshold scaled distance,loading density and RMR are shown in Figs. 9a and b. The resultscan be used to estimate the safe scaled distance with variousloading density and the RMR. The damage zone can then beestimated if the weight of charge is given.

    5. Conclusions

    The damage of rock mass under blast loading was investigatedthrough numerical simulations in this study. The fully coupledmethod has been used, in which the large deformation zone nearthe charge is solved by the Arbitrary LagrangeEuler (ALE)method. A new PPV damage criterion was proposed includingthe effect of RMR of rock mass. The empirical formula for damagedepth including the effect of loading density, RMR and weight ofcharge, was derived based on the numerical results. It indicatesthat the loading density, RMR and weight of charge signicantlyinuence the damage depth. By introducing the scaled distance,the formula of safe scaled distance can be obtained. The safescaled distance for RMR 40 is about 1.7 times that of RMR 95under the fully coupled charge condition. It should be noted thatthe study is based on one type of rockgranite, so the PPVdamage criterion for other rocks should be different as they mayhave different rock mass properties but with the same RMR.

    References

    [1] Hendron AJ. Engineering of rock blasting on civil projects. In: Hall WJ, editor.Structural and geotechnical mechanics, a volume honoring NM Newmark.Englewood Cliffs, NJ: Prentice-Hall; 1977. p. 2427.

    [2] Kendorski FS, Jude CV, Duncan WM. Effect of blasting on shortcrete driftlinings. Min Eng 1973;25(12):3841.

    [3] Rupert GB, Clark GB. Criteria for the proximity of surface blasting tounderground coal mines. In: Proceedings of the 18th US symposium on rockmechanics. Colorado School of Mines; 1977. p. 3C3-110.

    [4] Siskind DE, Fumanti R. Blast-produced fractures. Lithonia granite. Report ofInvestigations 7901, US Bureau of Mines; 1974.

    [5] Jensen DE, Munson RD, Oriard LL et al. Underground vibration from surfaceblasting at Jenny mine, KY Woodward-Clyde consultants, Orange CA. Final

    anics & Mining Sciences 46 (2009) 120612132001;21(2):8598.[15] Hao H, Wu CQ, Seah CC. Numerical analysis of blast-induced stress waves in a

    rock mass with anisotropic continuum damage models. Part 2: stochasticapproach. Rock Mech Rock Eng 2002;35(2):95108.

  • [16] Adhikari GA, Rajan B, Venkatesh HS, Thresraj AI. Blast damage assessment forunderground structures. In: Proceedings of the national symposium onemerging mining and ground control technologies, Varanasi, India; 1994. p.24755.

    [17] Persson PA. The relationship between strain energy, rock damage, fragmenta-tion, and throw in rock blasting. Int J Blast Fragmentation 1997;1:99110.

    [18] Li Z, Huang H. The calculation of stability of tunnels under the effects ofseismic wave of explosions. In: Proceedings of the 26th department of

    defense explosives safety seminar. US Department of Defense ExplosivesSafety Board; 1994.

    [19] Zhao ZY, Zhang Y, Wei XY. Prediction of PPV damage criterion using a neuralnetwork approach, in preparation.

    [20] Kalamaras GS, Bieniawski ZT. A rock strength concept for coal seamsincorporating the effect of time. In: Proceedings of the 8th internationalcongress on rock mechanics, Tokyo; 1995. p. 295302.

    ARTICLE IN PRESS

    X.Y. Wei et al. / International Journal of Rock Mechanics & Mining Sciences 46 (2009) 12061213 1213

    Numerical simulations of rock mass damage induced by underground explosionIntroductionNumerical model and calibrationMaterial modelsTNTAirRock mass

    Numerical results

    PPV damage criterionEffect of loading density and RMR on blast loading and rock mass damageEffect of loading densityEffects of RMR

    ConclusionsReferences