Tunnelling and Underground Space Technology 28 (2012) 49–56

Contents lists available at SciVerse ScienceDirect

Tunnelling and Underground Space Technology

journal homepage: www.elsevier .com/ locate/ tust

Prediction of blast-induced overbreak from uncontrolled burn-cut blastingin tunnels driven through medium rock class

Kaushik Dey a,⇑, V.M.S.R. Murthy b

a Department of Mining Engineering, Indian Institute of Technology, Kharagpur 721302, West Bengal, Indiab Department of Mining Engineering, Indian School of Mines, Dhanbad, India

a r t i c l e i n f o a b s t r a c t

Article history:Received 24 February 2010Received in revised form 8 July 2011Accepted 26 September 2011Available online 1 November 2011

Keywords:BlastingThreshold levelTunnelBurn cutPPV

0886-7798/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.tust.2011.09.004

⇑ Corresponding author. Tel.: +91 3222283728, m+91 3222 282282 (Dept.), +91 3222 255303/282700 (

E-mail addresses: kausdey@yahoo.co.uk (K. Dey(V.M.S.R. Murthy).

URL: http://www.iitkgp.ac.in (V.M.S.R. Murthy).

Drilling and blasting is the predominant rock excavation technique in driving horizontal tunnels. Thisoften results in large overbreak. One of the prime reasons for overbreak is the unacceptable levels ofground vibration generated in blasting. From the literature survey and practical experience, it was foundthat threshold levels of PPV for overbreak depends on rock properties, namely, rock strengths, P-wavevelocity, specific gravity, Poisson’s ratio and rockmass parameters. Determination of threshold level ofpeak particle velocity (PPV) is crucial for controlling blast-induced overbreak and can be approximatedby extrapolating the vibration predictor established from near-field vibration monitoring. This paperreports the experiments carried out in five horizontal tunnels for monitoring near-field ground vibrationusing accelerometer-based-seismograph planted in roof/sidewalls to establish ground vibration predic-tors. Blast-induced overbreak for each blast has been measured using a telescopic profiler. The thresholdlevels of PPV for overbreak have been estimated from the extrapolated vibration predictors to the over-break zone and also using near field approximation technique. The estimated threshold levels of PPV foroverbreak ranged between 590 and 1050 mm/s in extrapolation model and 410–890 mm/s in near-fieldHP model. Apart from these, a relationship between the percentage overbreak and rock/rockmass, chargeand blast design parameters has been established through multivariate regression analysis of the datapertaining to five investigating sites. The result has been validated for four blasts within reasonableaccuracy.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Blasting is the most popular means of excavation for under-ground tunnels despite the rapid developments in the applicationof mechanical excavators, namely, tunnel boring machines, roadheaders, continuous miners, etc. Faster driving rates are possiblewith the recent developments in explosives (emulsion), initiatingsystems (Nonel, electronic detonator) and drilling (automation)systems. However, longer pulls, associated with high concentrationof explosives, often lead to overbreak due to excess ground vibra-tions. Overbreak can become expensive phenomena in terms of ex-tra grouting and concrete backfilling and may also give rise toadditional mucking time. Most of the existing controlled blastingtechniques, to reduce the blast-induced overbreak, need extra dril-ling and are costly. Blasting in horizontal tunnels aims at the fol-lowing objectives:

ll rights reserved.

obile: +91 8906536801; fax:Inst.).), vmsr_murthy@yahoo.com

– Achieving longer pulls.– Reducing overbreak and rock damage.– Optimizing drilling and blasting cost.

It is rational to assess blast-induced overbreak in productionblasting and control the same by modifying the blast design, whichis largely affected by a host of rock, blast design and explosiveparameters. Several researchers have attempted to study over-break/blast-induced rock damage either based on experimentalstudies or relating some of the above influencing parameters. Abrief discussion on the previous works is provided in Section 2.

2. Previous work

Holmberg and Persson (1978) stated that for an extendedcharge of charge length ‘H’ of linear charge concentration of l(kg/m), a first approximation of the resulting v has been obtainedby integrating the generalized equation for the total charge lengthand is given by,

v ¼ Kqal

Z H

0

dx

fR20 þ ðZ � xÞ2g

b2a

" #a

ð1Þ

50 K. Dey, V.M.S.R. Murthy / Tunnelling and Underground Space Technology 28 (2012) 49–56

where v is the PPV (mm/s); R0 the horizontal distance between blastand point of interest (m); Z the vertical distance between blast holeand point of interest (m); H the charge height in the hole (m); ql isthe linear charge concentration (kg/m); and K, a and b are constantsbased on site characteristics. Vibration values for overbreak couldbe arrived at using this near-field model. Different researchers haveproposed threshold level of damage based on the vibration mea-surement and extrapolation of the PPV predictors. A brief reviewof the same is given in Table 1.

McKown (1984) and Singh (1992) used half cast factor as ameasure of blast-induced overbreak. Half cast factor is the ratioof total visible drill mark length in the wall and roof after blastto the total drilling length.

HCF ¼Pn

i¼1LiPnr¼1Lr

ð2Þ

where HCF is the half cast factor; Li the post-blast drill mark lengthvisible (m); and Lr is the pre-blast drilled length (m).

Grady and Kipp (1987) used a scalar, D, to describe the rockdamage. The value D lies between 0 (intact rock) and 1 (completefailure). This can also be used to estimate the rock modulus Ed, ofthe damaged rock, so that

Ed ¼ Eð1� DÞ ð3Þ

where E, Ed is the modulus of the intact rock and damaged rockrespectively.

A method proposed by JKMRC (1990), included the frequency,surface condition and density of discontinuities as a descriptor ofdamage.

Forsyth and Moss (1990) devised a method of quantifying blast-induced damage. Their proposed Drift Condition Rating (DCR) com-prised two components: firstly, the drift back condition (related tothe rockmass integrity and the percentage of half cast visible); andsecondly, the amount of overbreak. This empirical rating variedfrom 0 to 9.

Paventi et al. (1995) reviewed the development of a field proce-dure for damage monitoring through an empirical blast induceddamage index, DM given by,

DM ¼ I � II � III � IV � ðVA þ VBÞ ð4Þ

where I: considers the reduction in intact rock strength due to micro-fracturing; II: evaluates the extent of the exposed excavation surfacearea remaining in place using the post scaling half cast factor; III:determines the drift condition by assessing the drumminess of the

Table 1PPV based damage estimation models proposed by different researchers.

Bauer and Calder (1970) observed that no fracturing occurred for a PPV < 254 mm/s, Pcaused strong tensile/radial cracking and break up of rockmass occurred at PPV > 2

Langefors and Kihlström (1973) proposed that PPV in the range of 305 mm/s to 610 mtunnels

Holmberg and Persson (1979) approximated near-field PPV and found that the damaOriard (1982) proposed that most rockmass damaged at a PPV above 635 mm/sRustan et al. (1985) estimated PPV range of 1000–3000 mm/s for rock damageMeyer and Dunn (1995) computed threshold PPV level for rock damage to be 600 mm/sBogdanhoff (1996) measured near-field blast acceleration in an access tunnel at distan

for rock damage was found to lie between 2000 and 2500 mm/sMurthy and Dey (2003) reported a ground vibration predictor including the effect of

through a basaltic formation was >2050 mm/s

v ¼ k� R

W13þSF

� ��a

where v is the PPV (mm/s), R the distance (m), W the maximum charge/delay (kg), SF

constantsMcKenzie and Holley (2004), found that the threshold PPV level exceeds 700 mm/s for

and 300 mm/s for fine cracking in wall blasting

back with a scaling bar; IV: accounts for the amount of scaling arisingfrom damage; and VA and VB: considers the direction of structure withrespect to drift direction to account for the anisotropy potentiallycaused by structural features at meso- and macro-scale.

Yu and Vongpaisal (1996) proposed a new blast damage criteriabased on dynamic tensile strength, compressional wave velocity(P-wave), density of rockmass and peak particle velocity of theblast. The proposed damage criteria is as follows:

Dib ¼V � qr � cp

Kr � rtdð5Þ

where Dib is the blast damage index; V the vector sum of peak par-ticle velocity (m/s); qr the density of rock (g/cc); cp the rock P-wavevelocity (km/s); Kr the site quality constant (0–1.0) which is(RMR � ground support adjustment)/100; and rtd is the dynamictensile strength (MPa).

Based on the blast damage index the rock may be categorized asgiven in Table 2.

Singh (2000) studied the roof damage in underground due tosurface blasting. Based on underground instrumentation and far-field vibration monitoring, it was found that the Dib value of lessthan 1 referred to no damage condition and Dib value of more than2 referred to severe damage condition, whereas Dib value in be-tween 1 and 2 referred to a minor damage condition.

Ibarra et al. (1996) proposed perimeter specific charge (PCF) asthe controlling parameter for the blast induced rock damage assess-ment. Perimeter specific charge is defined as the ratio of weight ofexplosives in the perimeter blast holes and the next row, divided bythe volume of rock within this annulus, ignoring the lifters in the in-vert’. Analysis of the blast data of Aquamilpa Hydroelectric Project,Diversion Tunnel No. 2, revealed a relationship between overbreak/underbreak with log of Barton’s Q index. A linear relationship be-tween the underbreak/overbreak and PCF has been established.An increase in PCF indicates an increase in overbreak therefore a de-crease in underbreak. A composite relationship including both PCFand Q value for the prediction of overbreak/underbreak was estab-lished. Although, these relations are site-specific, it is easy to estab-lish using multiple regression analysis.

Overbreak ð%Þ ¼ �Ko1 þ Ko2 � qp � Ko3 � logðQÞUnderbreak ð%Þ ¼ Ku1 � Ku2 � qp þ Ku3 � logðQÞ ð6Þ

where qp is the perimeter specific charge; Ko1, Ko2, Ko3 the site-spe-cific characteristic constants for overbreak; and Ku1, Ku2, Ku3 is thesite-specific characteristic constants for underbreak.

PV of 254–635 mm/s resulted in minor tensile slabbing, PPV of 635–2540 mm/s540 mm/sm/s resulted in formation of new cracks and fall of rock respectively in unlined

ges occur in the PPV range 700 mm/s to 1000 mm/s

and for minor damage above 300 mm/s at Perseverance Nickel mine in Australiaces between 0.25 and 1.0 m outside tunnel perimeter holes. Threshold PPV level

free face in tunnel blasting. Threshold level of PPV for overbreak in a tunnel

the stiffness factor (burden/hole length) ratio, k and a are the site-specific

intense damage, 400 mm/s for significant damage, 350 mm/s for open cracking

Table 2Blast damage index and damage type.

Dib Type of damage

60.125 No damage to underground excavation0.25 No noticeable damage0.5 Minor and discrete scabbing effect0.75 Moderate and discontinuous scabbing damage1.0 Major and continuous scabbing failure1.5 Severe damageP2.0 Major caving

K. Dey, V.M.S.R. Murthy / Tunnelling and Underground Space Technology 28 (2012) 49–56 51

The above review clearly brings out that the damage modelssuggested relate the damage/overbreak with either a single or acouple of influencing factors. It was felt that inclusion of predom-inant factors of rock, blast design and explosive could lead to amore rational overbreak predictive model. The major contributingparameters identified are given below:

– Rock parameters: P-wave velocity and Poison’s ratio.– Blast design parameters: Confinement and advance factor.– Explosive charge parameters: Perimeter specific charge.

Thus, experimental blasts have been designed such that theinfluence of above-mentioned rock, blast design and explosivecharge parameters could be studied.

3. Design of experimental blasts

The rock/rockmass properties were determined from the fieldand laboratory investigations. Poisson’s ratio was computed fromthe measured P-wave and S-wave velocities in the laboratory.Post-blast tunnel cross sectional area was measured using tele-scopic overbreak measuring rod (Fig. 1) which had been designedand fabricated under the supervision of the authors. Overbreakswere computed using Planimeter after plotting telescopic offsetmeasurements on a graph paper. The overbreaks are expressed inpercentage of tunnel area.

Peak particle velocities and accelerations were monitored asnear to the blast face as possible using accelerometer and triaxialgeophone based seismographs (Minimate Plus and Minimate 077of Instantel Inc., Canada). The fixing arrangement of the accelerom-

Half hole dept

Fig. 1. Telescopic overbreak measuring ro

eter sensors has been shown in Fig. 2. The sensors used in the studywith their broad specifications are mentioned in Table 3. Vibrationpredictor equations were developed for each site using the squareroot scaled distances. To arrive at the overbreak threshold levels ofPPV, the established predictor equations were extrapolated up tothe overbreak distances (Murthy and Dey, 2003) and named asextrapolation model. Overbreak threshold PPV were also computedusing near-field estimation of Holmberg and Persson (1978) (near-field model HP).

Confinement, the ratio of drilling depth to tunnel area, has beenmeasured for every blast, because it has a significant impact on theoverbreak. Similarly, Advance factor, i.e. the ratio of advanceachieved to drilling depth has also been computed.

4. Laboratory and field investigations

Laboratory investigations are carried out on the collected rocksamples (blocks) from the each experimental site. Mostly core dril-ling was carried out to take the samples at each site. Apart fromthis, rock samples were also collected. It was observed that thereis no significant difference in the properties of the rock obtainedfrom cores and rock samples while tested in the laboratory.

Field investigations have been carried out in five horizontal tun-nels through hard metamorphic rocks representing different geo-technical conditions. The blasts investigated are the regularproduction blasts carried out with burn cut in tunnels referred toas Site-1 through Site-5. The details of the site conditions are givenbelow.

Site-1 is a horizontal tunnel of dimension 3.2 � 4.5 m. The hostrock is chlorite–serisite schist. The details of rock properties andthe rockmass classification (Q-index) ratings are provided in Tables4 and 5 respectively. Blasting was carried out with the drill jumbosof 3.2 m drill rod. The holes were charged with emulsion explosive.A schematic blast pattern for the face is given in Fig. 3a. The detailsof the experimental blasts are given in Table 6.

Site-2 is a horizontal tunnel of dimension 3 � 4 m. The host rockis chlorite schist. The details of rock properties and the rockmassclassification (Q-index) ratings are provided in Tables 4 and 5respectively. Blasting was carried out with the jack hammers with1.5 m long drill rod. The holes were charged with emulsion explo-sive. The schematic blast pattern for the face is given in Fig. 3b. Thedetails of the experimental blasts are given in Table 6.

Centre linePosition of overbreak measurement

h

Desired line of excavation

Overbreak Actual line of excavation

Blast holes

d and overbreak measuring scheme.

Fig. 2. Fixing of sensors of the accelerometer in the tunnel wall.

Table 3Major specifications of seismic sensors.

Parameters Accelerometer High frequencygeophone

Triaxialgeophone

Frequencyrange (Hz)

1–3000 1–2000 2–300 Hz

Accelerationrange

Up to 500 g(4903 m/s2)

Geophone naturalfrequency: 28 Hz

Up to254 mm/s

52 K. Dey, V.M.S.R. Murthy / Tunnelling and Underground Space Technology 28 (2012) 49–56

In both the Site-1 and Site-2, the initial openings were createdwith 5 � 3.2 m face dimension to have additional area to locateventilation ducts, etc. These blasts were carried out with drill jum-bos of 3.2 m drill rod. The blast pattern of the same is given inFig. 3c.

Site-3, Site-4 and Site-5 were the horizontal openings of a chro-mite mine driven for different purposes. However, all were hori-zontal tunnels of dimension 2.5 m � 2.5 m. In Site-3, the hostrock was serpentinite with lean chromites formation and the pur-pose of the opening was to act as the cross cut. In Site-4, the hostrock was driven through serpentinite host rock and thus the rockdensity was low. Site-5 was an ore drive driven through hard chro-mite formation. The details of rock properties and the rockmassclassification (Q-index) ratings are provided in Tables 4 and 5respectively. Blasting was carried out with the jack hammers with

Table 4Details of rock properties and overbreak threshold PPV levels of each site.

Parameters Site-1 Sit

Drilling parameters Drill pattern Burn cut BuDrill used Jumbo drill Jac

Rock parameters Density (t/m3) 2.81 2.9UCS (MPa) 63 10BTS (MPa) 6 14P-wave (m/s) 5055 50S-wave (m/s) 2855 28Poisson’s ratio 0.26 0.2Barton’s Q-index 6.15 8.4

Ground vibration predictors PPV predictor square root v ¼ 201 RffiffiffiffiWp� ��1:22

v

Extrapolation Model 591 61Threshold PPV (mm/s) Near-field HP Model 413 51

Allowable charge/hole (kg) Extrapolation Model 0.94 0.3Near field HP Model 1.59 0.4

UCS = Uniaxial compressive strength; BTS = Brazilian tensile strength; HP = Holmberg–Pecharge per delay (kg).

1.5 m long drill rod. The holes were charged with emulsion explo-sive. The schematic blast pattern for the face is given in Fig. 3d. Thedetails of the experimental blasts are given in Table 6.

In all the experimental blasts, the same explosive was used toreduce the variation in explosive properties, however, the chargepatterns were varied.

Near-field vibrations were monitored to establish ground vibra-tion predictor equation for each site. The predictor equation hasbeen extrapolated up to the overbreak distance to estimate thethreshold level of PPV for overbreak.

Threshold levels of PPV were also computed using Holmberg–Persson near-field model. The predictor equations and estimatedvibration threshold levels for measured overbreak are given in Ta-ble 4. Using the threshold level of PPV for overbreak, allowablecharge per hole is estimated to control the overbreak for a distanceof 0.4 m from the hole (Murthy and Dey, 2003). It has been foundthat the allowable charge per hole ranges from 0.2 to 1.0 kg forextrapolation model and ranges from 0.3 to 1.6 kg for near-fieldmodel.

5. Overbreak predictive model

A composite blast-induced rock damage (BIRD) predictive mod-el has been developed for estimation of overbreak from rockparameters, blast design parameters and explosive charge param-eters (Dey, 2004). In this model (BIRD), P-wave velocity of rock(cp) and Poisson’s Ratio (t) have been considered as the rockdescriptors. As the charge descriptor, perimeter specific charge(qp) has been taken into consideration and as the blast designdescriptor advance factor (AF) and confinement (fc) have been con-sidered. The composite model developed after multivariate analy-sis carried out in SPSS software is given below:

AO ¼ 27:91þ 0:97qp � 1:59cPt� 1:89AF

fcð7Þ

where AO is the overbreak (%), qp the perimeter specific charge(kg/m3), cp the p-wave velocity in (km/s), t the Poisson’s ratio, AF

the advance factor (m/m), fc is the confinement (m/m2).For the statistical validation of the model, ‘t’ test and ‘F’ test

have been conducted. The results are given in Table 7.For the validation of the model, ‘t’ test has been conducted to

test the significance of ‘R’ (correlation coefficient). The null hypoth-esis (H0) is that ‘R’ is not significant and alternate hypothesis (H1)is that ‘R’ is significant. The calculated ‘t’ value (tcal), which is a

e-2 Site-3 Site-4 Site-5

rn cut Burn cut Burn cut Burn cutkhammer Jackhammer Jackhammer Jackhammer

2 3.24 2.98 3.354 67 81 100

13 13 1095 5500 5470 566015 3320 3190 34108 0.21 0.18 0.220 7.48 7.21 7.66

¼ 447 RffiffiffiffiWp� ��1:14

v ¼ 476 RffiffiffiffiWp� ��1:16

v ¼ 734 RffiffiffiffiWp� ��1:14

v ¼ 705 RffiffiffiffiWp� ��0:99

2 852 746 10489 683 665 886

0.45 0.16 0.363 0.52 0.34 0.48

rsson; v = peak particle velocity (PPV, mm/s); R = radial distance (m); W = maximum

Fig. 3a. Blast pattern for 4.5 m � 3.2 m jumbo tunnel face at Site-1.

Table 5Determination of ‘‘Q-index’.

Site RQD Joint set no. Rating Joint roughness Rating Joint alteration Rating Joint water condition Rating SRF Rating Q-index

Site-1 82 2 4 Silky undulating 1.5 Slightly alter 2 Minor 1 Low stress 2.5 6.15Site-2 84 2 4 Smooth undulating 2 Joint wall 2 Minor 1 Low stress 2.5 8.40Site-3 85 3 9 Rough undulating 3 Unaltered joint wall 1 Medium inflow 0.66 Low stress 2.5 7.48Site-4 82 3 9 Rough undulating 3 Unaltered joint wall 1 Medium inflow 0.66 Low stress 2.5 7.21Site-5 87 3 9 Rough undulating 3 Unaltered joint wall 1 Medium inflow 0.66 Low stress 2.5 7.66

II

II II

II

III

III

III

III IV

IV IV

IV

VIII

IX VII

3000

100

750

500

4000

100

1

1

1

1

VI

V

V

V

V

VI

VI

VI

VII

VII VII

VII VII

VII VII

VIII VIII

VIII

IX

IX IX

X X

X X

1000 150 750

150

All dimensions are in mm. Not to Scale

Fig. 3b. Blast pattern for 4 m � 3 m jack hammer tunnel face at Site-2.

K. Dey, V.M.S.R. Murthy / Tunnelling and Underground Space Technology 28 (2012) 49–56 53

function of ‘R’, ‘n’ (no of samples) has been found to be 3.21, and islarger than ‘t’ value (ttable) at 5% significance level (i.e. 2.228 fromstudent t-table). Thus, it can be concluded that the alternatehypothesis (H1) is valid. This means ‘R’ is significant. ‘F’ test hasalso been done to test whether the variances of regression andresiduals are alike or not. The calculated ‘F’ value (Fcal) is 2.763,

which is lesser than the ‘F’ table (Ftable) value at 5% significancewith 3 and 8 degrees of freedom. In this case, the null hypothesisis valid. In other words, there is no significant difference betweenvariances of regression and residual. Thus, the proposed compositemodel is statistically and conceptually validated. The proposedcomposite overbreak model considers rock, charge and design

Z

2

7 5

I

II

II

II

II III

III III

III

IV

IV

IV

IV

V

V

V

V

VI

VI

IVIV

VI

VI

VII

VII

VII

VIII

VIII

VIII

IX IX IX

VIII

IX IXIX IX

X

VIII

VIII

X X X X X X X

3200

800

800

800

800

5000

700 720 027007 720 720 720

IX

All dimensions are in mm

Not to scale

Fig. 3c. Blast pattern for 5 m � 3.2 m jumbo tunnel face at Site-1 and Site-2.

1 1

5

3

3

5 II

II

I

I II

II II

III

III

III

III VIVI

IV IV

VII

V

V

VII

V

V

VI VI VI VI

VII VII

125 175 400 400 150

125

175

400

400

150

Face size = 2.5m×2.5m All dimensions are in mm I, II … = Long Delay 1, 2 …. = Short Delay Not to Scale

Fig. 3d. Blast pattern for 2.5 m � 2.5 m jack hammer tunnel face at Site-3, Site-4 and Site-5.

54 K. Dey, V.M.S.R. Murthy / Tunnelling and Underground Space Technology 28 (2012) 49–56

parameters, and therefore is more rational to apply for overbreakprediction. With more data sets generated the model can be finetuned further.

5.1. Parametric analysis of BIRD model

The parametric analysis attempts to investigate the reasons foroverbreak caused in tunnels using burn cut blasting. Burn cut, ingeneral, becomes successful only when the pattern of drilling isproperly implemented, supported by adequate explosive chargedesign and delay allocation. The general observations from thefield investigations pertaining to advance factor reveal that the ad-vance factor is low due to inadequate number of reamer holes,undercharging, inadequacy delay in cut holes and improper stem-ming causing loss of confinement. The proposed new overbreak

model (BIRD) is analyzed for the significance of various embeddedparameters. The proposed BIRD model is rewritten as,

AO ¼ 27:91þ 0:97qp � 1:59cPt� 1:89AF

fcð8Þ

incorporating the ranges of the parameters, the above BIRD modelbecomes,

Ao ¼ 27:91þ ð1 to 2:24Þ � ð1:87� 2:27Þ � ð4:07 to 15:81Þ ð9Þ

In the above-established relationship (BIRD Model) it is foundthat the blast design geometry (AF/fc) has the strongest influenceon the overbreak relative to other parameters. This is also evidentfrom the higher correlation coefficient obtained while relatingoverbreak with advance factor (0.60) and confinement (0.84). Inother words, the overbreak, in horizontal drifting depends signifi-

Table 6Site-wise details of each blast.

Sl no. Drill depth (m) Perimeter charge factor (kg/m3) Overbreak (%) Advance (m) Face area (m2) Advance factor (m/m) Confinement (m/m2)

Site-11 3.2 N.A. 14.24 2.6 14.4 0.81 0.222 3.2 1.34 5.83 2.8 14.4 0.88 0.223 3.2 1.22 17.01 2.0 14.4 0.63 0.224 3.2 1.34 18.8 2.0 14.4 0.63 0.225 3.2 1.19 18.43 2.0 16.0 0.63 0.206 3.2 N.A. 14.24 2.4 14.4 0.75 0.227 3.2 1.03 26.55 1.8 16.0 0.56 0.20

Site-21 1.3 2.31 3.99 1.2 12.0 0.92 0.112 3.2 1.16 24.17 2.1 16.0 0.66 0.203 3.2 1.25 19.66 2.5 16.0 0.78 0.204 1.5 1.59 17.9 1.2 12.0 0.80 0.135 1.5 1.32 18.2 1.2 12.0 0.80 0.136 1.5 2.14 16.55 1.25 12.0 0.83 0.137 1.5 1.62 7.78 1.3 12.0 0.87 0.138 1.7 1.74 17.69 1.4 12.0 0.82 0.149 1.3 1.31 10.75 1.18 12.0 0.91 0.1110 1.7 1.36 14.73 1.5 12.0 0.88 0.1411 3.2 1.13 22.2 2.4 16.0 0.75 0.2

Site-31 1.6 1.41 18.44 1.2 6.25 0.75 0.262 1.6 1.34 15.42 1.4 6.25 0.88 0.263 1.6 1.40 18.38 1.3 6.25 0.81 0.264 1.6 1.24 22.36 1.25 6.25 0.78 0.265 1.6 1.44 21.48 1.25 6.25 0.78 0.26

Site-41 1.6 1.31 18.99 1.3 6.25 0.81 0.262 1.6 1.48 12.21 1.35 6.25 0.84 0.263 1.6 1.48 29.97 0.9 6.25 0.56 0.264 1.6 1.31 27.22 0.9 6.25 0.56 0.265 1.6 1.41 24.45 1.2 6.25 0.75 0.26

Site-51 1.6 1.41 21.15 1.2 6.25 0.75 0.262 1.6 1.17 17.92 1.35 6.25 0.84 0.263 1.6 1.32 22.91 1.2 6.25 0.75 0.264 1.6 1.45 22.93 1.2 6.25 0.75 0.26

Table 7Statistical analysis of the proposed overbreak model.

Dependentvariable

Independentvariables

R tcal ttable

(5%)Fcal Ftable

(5%)

Overbreak (%) Rock 0.7 3.2 2.23 2.76 4.07ChargeBlast design

Table 8Comparison of the observed and predicted overbreak.

Blast-1 Blast-2 Blast-3 Blast-4

InputPerimeter specific charge (kg/m3) 1.49 1.46 1.32 1.58P-wave velocity (m/s) 5930 5930 5930 5930Poisson’s ratio 0.27 0.27 0.27 0.27Advance per blast round (m) 1.60 1.50 2.00 1.40Face size (m2) 14.40 14.40 14.40 14.40Drill depth (m) 3.20 3.20 3.20 3.20

OutputActual overbreak (%) 24.73 23.60 19.97 24.91Predicted overbreak (%) 22.55 22.79 21.32 23.17Error (%) (±) 8.81 3.44 6.78 6.98

K. Dey, V.M.S.R. Murthy / Tunnelling and Underground Space Technology 28 (2012) 49–56 55

cantly on the advance achieved and confinement present. In fact,the rock and charge parameters become important, only when ade-quate advance is achieved.

5.2. Validation of the model

Accuracy of BIRD model was tested with the data of four extradrift rounds kept aside for testing. These data were not includedin the development of the model. The results of the validation ofBIRD model are given in Table 8. From the predicted and observedvalues, it may be concluded that the proposed model is acceptableand the error in prediction is less than 10 per cent.

6. Conclusion

Blast-induced overbreak has been investigated from the exper-imental blasts and ground vibration monitoring using the state-of-the-art seismographs. The overbreak measurements have beenutilized to establish peak particle velocity thresholds using bothnear-field and extrapolation techniques. The estimated thresholdlevels for overbreak is higher in case of extrapolation model(590–1050 mm/s) than near-field model (410–890 mm/s). How-ever, the suggested allowable charge per hole is lower in case ofextrapolation model (0.2–1.0 kg) than near-field model 0.3–1.6 kg. Thus, extrapolation model is suitable for blasting near sen-sitive structure or in extreme fragile rock and near-field model isapplicable for other cases including regular blasting.

An overbreak predictive model has been developed consideringthe rock parameters (Poisson’s ratio, P-wave velocity), blast designparameters (advance factor and confinement) and an explosiveparameter (perimeter specific charge).

The composite overbreak model (BIRD) developed is found to bestatistically significant for the cases investigated. From the testing

56 K. Dey, V.M.S.R. Murthy / Tunnelling and Underground Space Technology 28 (2012) 49–56

of BIRD model with four blast datasets, it has been found the modelcould predict overbreak within a percentage error of 10 (Table 8).Thus, this approach can be considered useful for overbreak predic-tion in tunnels so as to exercise suitable charge and blast designcontrols.

References

Bauer, A., Calder, P.N., 1970. Open Pit and Blasting, Seminar Mining Engg. Dept.publication, Queen’s University, Kingston, Ontario. p. 3.

Bogdanoff, I., 1996. Vibration measurements in damage zone in tunnel blasting, In:FRAGBLAST-5 (Ed.), Proceedings of Fifth International Symposium on RockFragmentation by Blasting, Mohanty, Montreal, Quebec, Canada, August 23-24,pp. 177–185.

Dey, K. 2004. Investigation of blast-induced rock damage and development ofpredictive model in horizontal drivages. Unpublished Ph. D. Thesis in IndianSchool of Mines. Dhanbad, pp. 45–103.

Forsyth, W.W., Moss, A.E., 1990. Observations of blasting and damage arounddevelopment openings. 92nd Canadian Institute Mining and Metallurgy AnnualGeneral Meeting, Ottawa, May, 1990, pp. 245–251.

Grady, D.E., Kipp, M.E., 1987. Dynamic rock fragmentation. In: Atkinson, B.K. (Ed.),Fracture Mechanics of Rock. Academic Press, London, pp. 37–76.

Holmberg, R., Persson, P.A., 1978. Swedish Approach to Contour Blasting.Proceedings of Fourth International Conference on Explosive and BlastingTechnique, Society of Explosive Engineers, New Orleans, LA, February 1–3, pp.113–127.

Holmberg, R., Persson, P.A., 1979. Swedish Approach to Contour Blasting,Proceedings of Fourth International Conference on Explosive and BlastingTechniques, pp. 113–127.

Langefors, U., Kihlström, B., 1973. The Modern Techniques of Rock Blasting. JohnWiley and Sons, New York, p. 473.

Ibarra, J.A., Maerz, N.H., Franklin, J.A., 1996. Overbreak and underbreak inunderground openings – part 2: causes and implications. Geotechnical andGeological Engineering 14 (3), 325–340.

JKMRC, 1990. Advanced blasting technology, Final research report AMIRA P93D.Brisbane, Australia.

Meyer, T., Dunn, P.G., 1995. Fragmentation and Rockmass Damage AssessmentSunburst Excavator and Drill and Blast, Proceedings of North American RockMechanics Symposium, pp. 609–616.

McKenzie, C., Holley, K., 2004. A Study of Damage Profiles behind Blasts,Proceedings of the 30th Annual Conference on Explosive and BlastingTechnique, February 1–4, 2004, New Orleans, Lousiana, pp. 203–214.

McKown, A. 1984. Some aspects of design and evaluation of perimeter controlledblasting in fractured and weathered rock. 10th Conference on explosive andblasting techniques, January 28–February 2, Orlando, Florida, pp. 120–152.

Murthy, V.M.S.R., Dey, K., 2003. Predicting overbreak from blast vibrationmonitoring in lake tap tunnel – a success story. FRAGBLAST 7 (3), 149–166.

Oriard, L.L., 1982. Blasting Effects and Their Control, SME of AIME, Littleton,Colorado, pp. 1590–1603.

Paventi, M., Lizotte, Y., Scoble, M., Mohanty, B., 1995. Measuring rockmass damagein drifting. In: Mohanty, B. (Ed.), Proceeding of Rock Fragmentation by Blasting– FRAGBLAST 5, Montreal, Quebec, Canada, 23–24 August 1995, pp. 131–138.

Rustan, A., Naarttijärvi, T., Ludvig, B., 1985. Controlled Blasting in Hard IntenseJointed Rock in Tunnels, CIM Bulletin, December, vol. 78, no. 884, pp. 63–68.

Singh, P.K. 2000. Evaluation of damages to underground coal mines caused bysurface blasting vis-à-vis establishment of blast vibration threshold. CMRI CoalS & T Project report, Project No GAP/III/BLASTING/MOC/1996, pp. 1–50.

Singh, S.P., 1992. Mining industry and blast damage. Journal of Mines, Metals &Fuels 40 (12), 465–472.

Yu, T.R., Vongpaisal, S., 1996. New blast damage criteria for underground blasting.CIM Bulletin 89 (2), 139–145.