tbipdf
TRANSCRIPT
Development of rational models for tunnel blast
prediction based on a parametric study
A. K. CHAKRABORTY, A. K. RAINA, M. RAMULU, P. B. CHOUDHURY,A. HALDAR, P. SAHOO and C. BANDOPADHYAYCentral Mining Research Institute, Regional Centre, 3rd Floor MECL Complex,Seminary Hills, Nagpur-440 006, India. e-mail: [email protected]
(Received 25 February 2003; revised and accepted 25 June 2003)
Abstract. The empirical models available for prediction of the tunnel blast results like pull
ratio, specific charge, specific drilling and overbreak have some inherent shortcoming inabsence of any parametric study at the backdrop. Hence, the models use different constitutingparameters and provide values which differ widely. After a thorough review of literature and
field investigations in the drivages of mines and tunnels some parameters were identified.Those parameters were subjected to Multiple Linear Regression analyses to filter out the mostinfluencing ones which represent the rockmass properties, the tunnel configurations and theblast designs. A parameter called Tunnel Blasting Index (TBI) was conceptualized and was
expressed in terms of those most influencing parameters. All the blast results observed duringthe filed investigations could be well related to a single index TBI. Some adjustments onaccount of shape of the tunnel and joint orientations, which were not addressed in the avail-
able models, are suggested in the developed models.
Key words. joint orientation adjustments, most influencing parameters, predictive models,Tunnel Blasting, Tunnel Blasting Index (TBI).
1. Introduction
The tunnel blast performance is generally measured in terms of one or more than one
of the following blast parameters:
(1) Pull (face advance/depth of round), expressed in percent,
(2) Specific charge (kg of explosive/m3 or t of yield),
(3) Specific drilling (m of drilling/m3 or t of yield), or Detonator or hole factor
(number of holes/m3 or t of yield), and
(4) Blast induced rock mass damage and overbreak or underbreak.
The underbreak is usually expressed in the field in terms of negative overbreak.
The blast induced overbreak or underbreak is measured radially and expressed
in metre. Those are also occasionally estimated volumetrically in m3 of in situ
rock mass over or under-broken and expressed in percent of the designed
volume. However, in most of the projects in India, the permissible limit of over-
break has been defined in terms of width and height of tunnel. The Swiss Society
Geotechnical and Geological Engineering 22: 477–496, 2004. 477# 2004 Kluwer Academic Publishers. Printed in the Netherlands.
of Engineers and Architects defines the permissible overbreak limit as 0:07pA,
where A is the tunnel area or 0.4m whichever is less (Innaurato et al., 1998).
All the above results jointly contribute to the safety, rate of progress and economy
of the tunnel. Hence, it would be misleading to measure the tunnel blast efficiency in
terms of one or two of the above parameters. The degree of fragmentation and the
muck profile are also important indicators of blast performance and may need to be
addressed in several cases, particularly in open pit mining operations. However, as
these two results affect the mucking operation and generally do not pose severe pro-
blems to the practising engineers in their day-to-day tunnel blasting activities, these
are not dealt in details in the present paper.
The above results, are dependent on the rock mass properties which may be
termed as the non-controllable parameters (N), the tunnel configurations which
may be called as the semi-controllable parameters (S) and the blast design para-
meters which are known as the controllable parameters (C). All the parameters used
for development of available models leading to blast design or predictions can be
classified in these three categories as is evident in Table 1.
The following aspects are prominently noted in Table 1.
(1) The non-controllable parameters have been adequately considered in most of the
available predictive models, but the semi-controllable or the controllable para-
meters are not amply represented.
(2) The degrees of influence of the parameters considered in those predictive models
are not known. It is also known if those models considered the most important
parameters. Further, the interrelations among the influencing parameters are
also not revealed.
(3) No index is available to relate all the blast results. Hence, it is difficult to assess
the overall change in tunnel blast results if one of the parameters constituting
any model is varied. For example, it may be possible to assess the change in
specific drilling or specific charge if blast hole diameter is changed but its effect
on pull or overbreak can not be estimated as the predictive models for those do
not consider the blast hole diameter, though the linear charge density increases
with the increase in hole diameter.
It, therefore, appears imperative to conduct a parametric study to define the most
influencing parameters and subsequently develop the predictive models for all the
blast results with the help of those influencing parameters. This also enables to define
and weight all the blast results on the same platform. The present paper describes the
predictive models developed by the authors to fulfil the objective, from concept to
the actual development through parametric studies.
2. Development concept
In line with the Blast Damage Index ðDibÞ developed by Yu and Vongpaisal (1996),
the tunnel blast can be defined as a function of blast induced stress and tunnel rock
478 A. K. CHAKRABORTY ET AL.
Table 1 Parameters considered for prediction of tunnel blast results
Sl. no. Blast result
Parameters considered in
predictive models Model developed by
1 Specific charge
and specific drilling
Tunnel area (S)
Drilling error (S)
Langefors & Kihlstrom (1973)
Tunnel area (S) Olofsson (1988)
Tunnel area (S) Pokrovsky (1980)
Protodyakonov Index (N)
Rock structure (N)
Relative weight strength
of explosive (C)
Explosive (charge) diameter (C)
Tunnel area (S) Hagan (1992) and
Du Pont (1977)
Blast hole diameter (C)
Density of rock (N) Ghose (1988)
Protodyakonov Index (N)
Joint spacing (N)
Joint orientation (N)
Rock Mass Description (N) Lilly (1986)
Joint spacing (N)
Joint orientation (N)
Specific gravity of rock (N)
Hardness (N)
Rock Mass Quality (Q)(N) Chakraborty (1996, 1998)
Strength Rating (N)
Number of contact surfaces (N)
Hole length (C)
2 Pull Sonic velocity in rock (N) Bergh-Christensen and
Selmer-Olsen (1970)
Specific gravity of rock (N)
Number of joints in a round (N)
Uniaxial tensile strength (N)
Joint orientation (N) Johansen (1998)
3 Rock mass
damage and
overbreak
Peak particle velocity (N+C) Langefers & Kihlstrom (1973)
and Holmberg & Persson
(1978, 80)
Vector sum of peak particle
velocities in three orthogonal
(N+C) directions
Yu & Vongsaipal (1996)
Rock density (N)
P-wave velocity (N)
Dynamic tensile strength of rock (N)
Site Quality Constant (S)
Young’s Modulus (N) Mckenzie (1994)
Uniaxial tensile strength (N)
P-wave velocity (N)
Joint orientation (N) Johansen (1998)
Rock Mass Strength (N) Innaurato et al. (1998)
(N-Non-controllable parameter, S-Semi-controllable parameter, and C-Controllable parameter).
DEVELOPMENT OF RATIONAL MODELS 479
mass resistance to fragmentation, which jointly indicate the tunnel blast environ-
ment. Hence, tunnel blast results can be expressed as a function of rock mass resis-
tance to fragmentation and blast induced stress as shown in Equation 1.
Tunnel blast results
¼ f (tunnel rock mass resistance to fragmentation, blast induced stress)
ð1Þ
As tunnel blasting is done under confined conditions, the tunnel rock mass resistance
is controlled by not only the rock mass properties but also by confinement. It was
seen that DuPont (1977), Pokrovsky (1980), Langefors and Kihlstrom (1973) and
Olofsson (1988) considered the inverse of the tunnel area to account for the tunnel
confinement while predicting the specific charge and the specific drilling (Table 1).
Thus, Equation 1 can be modified as:
Tunnel blast results
¼ f f(rock mass resistance to fragmentation, tunnel confinement),
blast induced stressg ð2Þ
An index called Tunnel Blast Index (TBI) is conceived to represent the tunnel blast
environment comprising blast induced stress, rock mass resistance to fragmentation
and tunnel confinement. Hence, the tunnel blast results can be expressed as function
of TBI as shown in Equation 3.
Tunnel blast results ¼ f (TBI) ð3Þ
If the blast induced stress, the rock mass resistance and the tunnel confinement can
be expressed in terms of factors defined by the respective influencing parameters, TBI
can be defined by Rock Mass Factor (RF) which is a function of the rock mass prop-
erties providing resistance against fragmentation, Tunnel Configuration Factor (TF)
which is a function of tunnel configuration parameters contributing to the confine-
ment and Blast Design Factor (BF) which is a function of blast design parameters
responsible for blast induced stress.
As discussed earlier, it would be essential to identify the most influencing para-
meters to define RF, TF and BF. This needs collection of detailed information during
tunnel blasting under different environments. Field investigations were carried out in
tunnels of different configurations under various rock mass conditions. This was
followed by an analytical process to filter the parameters which have the maximum
influence on the observed blast results. TBI has been defined in terms of RF, TF and
BF which includes those most influencing parameters.
3. Field investigations
In accordance with the above scheme, field investigations were conducted by the
authors in inclined drifts of a coal mine, development galleries of two metal mines
and a tunnel of a hydro-electric project. The sites are listed in Table 2 and the loca-
tions of the sites are shown in Figure 1.
480 A. K. CHAKRABORTY ET AL.
3.1. METHODOLOGIES FOLLOWED IN FIELD INVESTIGATIONS
The following broad methodologies were adopted by the authors during the field
investigations:
1. All the investigated tunnel lengths were thoroughly inspected.
2. Initial tunnel lengths, which mostly consisted of weathered rock masses, were
excluded.
Table 2 Names of the investigated sites
Sl. no. Type of mine/project Name of mine/project Nature of excavation
A. Mining sector
1 Coal Tandsi Inclined drifts in rock
2 Metal
(Manganese)
Chikla Development roadways, raises
and winzes in host rock and
drivages in manganese ore body
Gumgaon
B. Civil sector
3 Hydro-electric
project
Koyna Hydro-
electric
Project (KHEP),
Stage IV
Link tunnel in rock
Figure 1 Locations of the sites selected for field investigations
DEVELOPMENT OF RATIONAL MODELS 481
The blasting operation was not standardised in the initial portion of the tunnels.
The locations where the rock masses were found abnormally different from those
in the rest of the tunnel were not considered for further analysis. The locations,
which were ignored in this manner at different sites, vary from 5 to 20 percent
of the total population.
3. Rock Mass Quality (Q) was determined for all types of formations available in
the investigated tunnel lengths using the following reaction:
Q ¼ ðRQD=JnÞ � ðJr=JnÞ � ðJw=SRFÞ ð4Þ
where,
RQD ¼ Rock Quality Designation,
Jn ¼ joint set number,
Jr ¼ joint roughness number,
Jn ¼ joint alteration number,
Jw ¼ joint water reduction factor, and
SRF ¼ Stress Reduction Factor.
RQD values used for obtaining Q were determined from volumetric joint count
using the following relation provided by Palmstrom (1975):
RQD ¼ 115� 3:3 Jv ð5Þ
where,
Jv ¼ volumetric joint count.
The Jv values were determined by adding the number of visible joints per metre
length of the exposed surfaces in all three directions. The walls and the roof of
an excavation were scanned for this purpose.
4. The investigated tunnel length was categorised into various zones based on the Q
value. The joint set which appeared most frequently and consistently was consid-
ered as the major joint set in that zone. The orientation of that joint was deter-
mined using a Brunton compass.
The procedures followed for joint orientation and spacing measurements during
field investigations are in conformation with the guidelines provided by Interna-
tional Society of Rock Mechanics (ISRM) (Brown, 1981).
5. Rock samples representing all zones were collected by the authors.
The samples were tested in the laboratories of Central Mining Research Institute,
Regional Centre, Nagpur and Visvesvaraya National Institute of Technology,
Nagpur for evaluation of various physico-mechanical properties which were
found to be influencing the blast results as detailed in Table 1.
6. Detailed information on on-going blasting practice and four blast results like the
pull, the specific charge, the specific drilling and the overbreak in various rounds
were collected by the authors.
482 A. K. CHAKRABORTY ET AL.
Face advance in a round was measured at the face centre and the two sides of the
face. The average of these values was considered as the average advance per
round. The roof and the side overbreak after each round of blasting were mea-
sured at 5 to 7 locations both side-wise and height-wise. These values were aver-
aged to obtain the average overbreak. The exacavated in situ volume was
calculated by multiplying the post-blast cross-section and average face advance.
This was verified from the number of trips of loaded muck or the stock pile or
crusher data. The specific charge or the specific drilling were estimated from the
ratio of total explosive quantity or total drilling length in a round and the exaca-
vated in situ volume of rock.
7. The blast results of different rounds in a particular zone were averaged to deter-
mine the average blast results in that zone.
8. Trial blasts were conducted in these sites with modified blast design and the
results were monitored by the authors.
The various geo-mining variations covered during field investigations are listed in
Table 3 to provide an overall picture at a glance.
During the field investigations many parameters listed in Table 1 were found to
influence the blast results significantly. However, some other parameters were also
found to affect the results as described below.
The directional blast results like the pull and the overbreak or the underbreak at
the roof or the walls are substantially influenced by joint plane orientation. The pull
was affected most adversely if the joints were steep with strike parallel to the tunnel
axis. Further, such joint sets striking across to the tunnel axis proved to be the most
favourable condition to improve the pull. An opposite trend was found in case of
roof overbreak. The roof overbreak was less when joints were steep but the strike
was parallel to the tunnel axis and was more when such joints had strike across
the tunnel axis.
Further, the tunnel shape and the application of contour blasting influenced
the overbreak. Also the unevenness at the tunnel periphery was fount to be lar-
gely dependent on the spacing to burden ratio. Holmberg (1982) recommended
different spacing to burden ratio for different parts of the tunnel specially for
the contour holes to minimize the overbreak. Additionally, the specific drilling
was also considerably increased in case of contour blasting as the spacing to
burden ratio of drill holes along the contour ðmdcÞ is maintained less than
one in contrast to those maintained as one or more in the rest of the tunnel
section.
4. Identification of the most influencing parameters
Based on the field investigations and the literature review, a list of influencing para-
meters has been prepared by the authors in Table 4.
The degree of influence of the above parameters on the blast observed in the field
investigations is determined by multiple linear regression analysis.
DEVELOPMENT OF RATIONAL MODELS 483
Table3Geo-miningconditionsobservedduringfieldinvestigations
C.Site
Sl.no.
Parameter
Link
Chikla
Tandsi
Gumgaon
A.Rockmassproperties
1Typeofrock
Basaltieflowof
compact&
amygdoloidal
basaltand
volcanic
breccin
Mn-oreandfootwall
containing
manganiferousquartz
andmuscoviteschist
Sandstone
Mn-orebodyand
footwallrock
masscontaining
quartzmuscovite
schist
2Q
5.378–64.48
0.75–32.05
0.31–18.66
0.21–1.85
3UCS,MPa
21.02–91.136
18–180
18.9–32.4
60–162
4Density,1/m3
2.37–2.93
2.5–3.9
1.9–2.35
2.82–3.97
5RQD
35.45–87.27
40.75–91.4
36–82
23–47.4
6P-wave
velocity,km/s
2.487–5.816
2.913–8.117
1.9–2.9
4.582–7.694
7Majorjointsetorientation
Dip()
60–90
60–90
0–30
30–60
Strikeangle
withrepectto
tunnelaxis()
0–30
60–90
60–90
30–60and60–90
8Mixedface
Brecciaþ
Amygdoloidal
basaltþ
Compactbasalt
Nil
Nil
Nil
484 A. K. CHAKRABORTY ET AL.
B.Tunnelconfiguration
9Size,m2
36
5.04
15(with
shotcrete
support)and
17.66withsteel
support
5.04
10
Shape
Arch
Rectangular
D-shaped
Rectangular
11
Inclination
Nearlyhorizontal
Nearlyhorizontal
Inclination1:4.66
Nearlyhorizontal
C.Operatingtool
12
Drillingmachine
Manualjack
hammer
Manualjack
hammer
Hydraulicjumbo
Manualjackhammer
D.Blastdesign
13
Typeofcut
Convergent
Convergent
Parallelin
shotcrete
supportedzone
andconvergent
insteelsupportedzone
Convergent
14
Methodof
perimeterblasting
Conventional
Conventional
Conventional
lateron
switchedto
smoothblasting
Conventional
DEVELOPMENT OF RATIONAL MODELS 485
Multiple linear regression analysis (MLR) is performed to determine the combined
effect of a group of independent variables upon a dependent variable. The method
may be used to assign relative importance to the independent and may be interrela-
ted variables by sequentially including or excluding the one having largest partial
correlation (Gupta and Kapoor, 1999 and SPSS Inc., 1993).
The parameters considered for MLR are having different units and their ranges
vary widely. It is therefore considered that all these parameters should be normalised
using the following relation:
Xn ¼X
Xmax � Xminð6Þ
where, Xn is the normalised value, X is the original value and Xmax and Xmin are the
maximum and minimum values of that particular parameter in the population.
By normalising the variables and recasting them in dimensionless units, the arbi-
trary effects of similarities between the objects are removed (Flood and Kartam,
1994; Sayed and Abdewahab, 1998; Leu et al., 1998).
The gradual improvement in correlation between the specific charge and the spe-
cific drilling with the twelve input parameters obtained through MLR are presented
in Figures 2 and 3 respectively. The indices include the independent variables repre-
senting the X-axis in Figures 2 and 3.
It can be seen in the Figures 2 and 3 that, the index of correlation ðR2Þ improves
with the addition of independent variables. But the improvement is not all significant
Table 4 List of influencing parameters
Properties
Sl. no. Rock mass properties (N)
1 Density of rock mass
2 Rock strength–represented by Strength Rating (SR)
3 P-wave velocity
4 Barton’s Rock Mass Quality (Q)
5 Number of contact surfaces in multiple geological mixed rock face condition
6 Orientation of major joint set with respect to tunnel face
Tunnel configuration (S)
7 Area
8 Shape factor – the ratio between tunnel width to diameter of the curvature at roof
9 Tunnel inclination with respect to upward vertical direction
Blast design (C)
10 Type of cut
11 The deviation factor defined in terms of the ratio between drill hole length and diameter
12 Number of additional or baby cut holes blasted before the main cut holes
13 Charge per hole
14 Coupling ratio of explosive to blast hole diameter
15 Conventional or contour blast design expressed in terms of spacing to burden
ratio and coupling ratio in the contour holes
486 A. K. CHAKRABORTY ET AL.
beyond the first six parameters, when the improvement occurs up to 1 percent. In
view of this fact the following seven parameters, which include all these first six para-
meters, are finally considered as most influencing. Those seven parameters are clas-
sified in three groups like, (i) rock mass parameters, (ii) tunnel configuration
parameters and (iii) blast design parameters as shown below:
4.1. (A) ROCK MASS PARAMETERS
1) P-wave velocity (cp, expressed in km/s)
2) Number of contact surfaces in multiple geological mixed face condition (n)
3) RQD (RQD)
4.2. (B) TUNNEL CONFIGURATION PARAMETERS
4) Area (A, m2)
5) Inclination ðbiÞ with respect to vertical upward direction (expressed in radian, r)
Figure 2 Results of multiple linear regression analysis for specific charge
Figure 3 Results of multiple linear regression analysis for specific drilling
DEVELOPMENT OF RATIONAL MODELS 487
4.3. (C) BLAST DESIGN PARAMETERS
6) Cut hole angle i.e., the angle made by cut holes with the face, expressed in cotan-
gent ðCaÞ
7) Coupling ratio between explosive and blast hole diameter ðRcÞ
The Rock Mass Factor (RF), the Tunnel Configuration Factor (TF) and the Blast
Design Factor (BF) are worked out from the above selected parameters, as per the
concept of the Tunnel Blasting Index discussed earlier.
Accordingly, the Tunnel Blasting Index (TBI) is defined as:
TBI ¼Rock Mass factor (RF)
Tunnel Configuration Factor (TF)� Blast Design Factor (BF)ð7Þ
where,
RF ¼ cp þ nþ ðRQD=10Þ; ð7AÞ
TF ¼ A� r; and ð7BÞ
BF ¼ Ca þ Rc: ð7CÞ
The P-wave velocity (cp) varied between 1000 to 8000m/s. Its range is exceptionally
wide in comparison to other parameters in TBI. Therefore, it is converted to km/s
unit to keep it in harmony with the other six parameters and to restrict the values
of TBI within reasonable limit. Among others, the range of RQD is also quite broad.
One tenth of RQD values are therefore considered in RF for the similar reason.
5. Development of models
In accordance of Equation 3, different blast results observed by the authors during
field investigations have been correlated below with TBI determined for the respec-
tive zones of the tunnels. Consequently new models for blast results prediction are
developed.
The relations between the observed specific charge (q, kg/m3) and the specific dril-
ling (bs, m/m3) with TBI are shown in Figures 4 and 5 respectively. Both the relations
have index of determination (R2) more than 0.9.
The specific charge predictive model (Equation 8) is developed on the basis of the
relation shown in Figure 4.
q ¼ 1.1+0.24 TBI, kg=m3 ð8Þ
As discussed earlier, a huge amount of additional drilling is essential in contour
blasting where the required explosive quantity is distributed in the closely spaced
holes along the perimeter. The spacing to burden ratio in the contour holes (mdc)
is included in specific drilling prediction to account for the additional drilling
488 A. K. CHAKRABORTY ET AL.
required in contour blasting, if any. Further, the shape factor (sh) has also been
considered in view of the fact that additional drilling required at the perimeter in
an arch shaped roof than in a flat roof. Shape factor is defined as the ratio of the
tunnel width to the diameter of roof curvature. The ratio is 1 for a perfect D-shaped
Figure 4 Specific charge vs. TBI
Figure 5 Adjusted specific drilling vs. TBI
DEVELOPMENT OF RATIONAL MODELS 489
tunnel and 0 in a rectangular tunnel. The ratio should lie between 0 and 1 in a tunnel
where the roof is arch shaped.
The specific drilling, adjusted after the spacing to burden ratio of the periphery
holes and shape factor, has been related in Figure 5. Consequently, the specific dril-
ling predictive model is derived in Equation 9.
bs ¼ 4:79TBI0:6
m0:5dcþ sh; m=m
3 ð9Þ
5.1. ADJUSTMENTS FOR JOINT ORIENTATION
During field investigations, the joint orientation was found to influence the over-
break/underbreak and face advance which is equal to the product of the pull and
the depth of a round (Ad) divided by 100. To consider the effect of joint plane orien-
tation, the following adjustments are proposed for the major joint sets observed
during field investigations.
The gentle, moderate and steeply dipping joint planes signify the dip angles as
0–30, 30–60 and 60–90 respectively. Similarly, strikes with respect to tunnel
axis are mentioned as parallel, oblique and across to indicate that the joint strike
intersection angle with the tunnel axis as 0–30, 30–60 and 60–90 respectively.
the adjustments due to joint orientations have been suggested separately for pull
and overbreak, which are measured in particular directions.
Further, adjustments for the angle of cut (a) is also provided as a ratio of depth tocut hole length, which is also equal to sin of a ().The relation between pull, adjusted after joint orientation effect and cut angle,
with TBI is displayed in Figure 6. The model for prediction of pull, developed on
the basis of the relation shown in Figure 6, is shown in Equation 10.
Ar ¼½f1:063ðTBIÞ0:55gðsin aÞ2 þ JOAa�
Ad� 100; percent ð10Þ
In case of roof overbreak/underbreak, where the height was increased due to local
fall, specially near the junctions or where the height was abnormally low due to
underbreak caused by misfires, the observations are rejected. The number of values
rejected in this manner are 20 percent of the total population.
Table 5 Joint orientation adjustments
Joint orientation
Dip
Strike with respect to
tunnel axis
Face advance
adjustment
(JOAa), m
Roof overbreak/underbreak
adjustment (JOAr), m
Steep Parallel �0.6 0.6
Steep Across 0.45 0.03
Gentle Across 0.1 0.05
Moderate Across/oblique 0.05 0.2
490 A. K. CHAKRABORTY ET AL.
The relation between the adjusted roof overbreak/underbreak (Ior), taking into
account the adjustments due not only to joint orientation (JOAr) but also to the tun-
nel shape factor (sh) and contour blasting practice defined by the spacing to burden
ratio of the contour holes (mdc), with TBI is shown in Figure 7. Generally the cou-
pling ratio in the contour holes (Rcc) is kept different than that in production holes to
reduce the stress. In such cases the TBI along the tunnel contour (TBIc) becomes
Figure 7 Adjusted roof overbreak or underbreak vs. TBI
Figure 6 Adjusted pull vs. TBI
DEVELOPMENT OF RATIONAL MODELS 491
different that that in the rest of the tunnel section. The roof overbreak/underbreak
predictive model evolved from Figure 7 is given in Equation 11.
Ior ¼ 0:57� 0:52 lnðTBIcÞ � 0:5sh �JOArmdc
; m ð11Þ
6. Application of the developed models
A D-shaped approach tunnel was excavated by heading and benching method
through basaltic formations under aPumped Storage Scheme. The heading area is
29.74m2. The tunnel passed through different varieties of Deccan Traps like com-
pact basalts, amygdolidal basalts and volcanic breccia.
The roof overbreak in the tunnel due to blasting varied from 0.091m to 0.5m in a
length of 385m. The developed models were used to assess whether the geological
features or the drilling error was responsible for such large variations in the over-
break. The results in a the length between 300–370m chainage in the tunnel was
undertaken for this purpose (CMRI Report, 2002).
The geological conditions in the length under consideration were studied carefully.
Three numbers of joint sets were observed in most of the reviewed length of the tun-
nel. An additional joint set was observed in some locations near 370m chainage. The
joints were found tight and undulating with generally rough to very rough surfaces.
Random joints were found inconsistent. Mixed face multiple geological condition
was not observed in the tunnel.
The rock mass properties and the blast results were monitored over a length of
20m strip (covering both right and left sides of the locations) at 300m, 325m and
370m chainage of the tunnel. These three locations represented the tunnel section
between 300 and 370m. The basic properties to estimate TBI in those three locations
are listed in Table 6.
The observed blast results (Obs) in those three locations and the predicted results
(Pred) using the developed models are shown in Figure 8.
It can be seen in Figure 8 that the deviation between the observed and predicted
specific charge and overbreak are not large in between 300 and 326m despite varia-
tions in the formations. However, the predicted overbreak is much in 370m in
comparison to the observed one. This may be due the presence of an additional joint
set in this region.
It appears that not only the joint spacing, which as been taken into account for
RQD estimation, but also the number of joint sets may need to be considered for
more precise prediction of overbreak.
However, it was concluded that the overbreak was caused mostly due to the geo-
logical features and blast design and not because of drilling error. But, the need of
controlled blasting including decoupling between the explosive and blast hole (Rc)
and proper spacing to burden ratio at the periphery holes (mdc) was stressed to
control overbreak.
492 A. K. CHAKRABORTY ET AL.
Table6BasicpropertiesforestimatingTBIintheapproachtunnel
A)Rockproperties(RF)
Locations(m)
RQD
c p(km/s)
nJointorientationwithrespecttotunnelaxis
300
90
5.7
Nil
Dipmoderateandstrikeatobliquetothe
tunnelaxis
326
60
3.1
Dipsteepandstrikeparalleltotunnelaxis
370
88.6
4.4
Dipsteepandstrikeparalleltotunnelaxis
B)Tunnelheadingconfigurations(TF)
Tunnelshape
Heading
area(m2)
Width(m)
Radiusof
curvature(m)
Vertical
wallheight(m)
Tunnel
direction
s h
D-type
29.74
73.5
1.5
Nearly
horizontal
1
C)Blastdesignconfigurations(BF)
Typeofcut
Ca,
Blast
hole
dia.(mm)
Explosive
dia(mm)
Rc
Peripheral
blasting
mdc
Production
holes
Contour
holes
Wedge
60
33
25
0.76
0.76
Contour
blasting
1
DEVELOPMENT OF RATIONAL MODELS 493
6. Conclusions and discussions
As the available models for prediction of blast results do not have a holistic
approach, an index called Tunnel Blasting Index (TBI) has been developed by the
authors to represent tunnel blast environment. TBI includes the most important
parameters influencing the blast results. The blast results like specific charge, specific
drilling, pull ratio and roof overbreak observed during various rounds in four tun-
nels could be well related with TBI. The effects due to tunnel shape and joint orien-
tations, which were not considered in the models developed by others, have been
taken into account in the newly developed relations by the authors.
The predictive models developed in the present paper by the authors are based on
limited site investigations and do not cover many of the possible combinations of the
non-controllable (specially joint orientations), the semi-controllable and the control-
lable parameters. Further, the said models are not dimensionally balanced and are
obtained from near optimised case studies having low to medium advance upto
3m. Hence, the applicability of the models are limited to medium pull only. The
detailed explosive properties and the charge concentration parameters have been
overlooked during field investigations (Section 3) due to lack of variations in the
explosive type and size, and the blast hole diameter in the fields where the investiga-
tions were conducted. The variation in delay timing has not been addressed in the
presently developed models as long delays of more than 100ms were used in most
of the monitored blasting rounds to account for tunnel confinement and wall
damage reduction. Hence, there is a scope to add the variation effects of long delay
time in the models. Though the wall overbreak is not as seriously considered by the
field engineers as the roof overbreak, an effort was made to develop a model for wall
overbreak prediction. However, no acceptable correlation could be established prob-
ably because the overbreak data on both the sides were not collected separately. The
developed models were used in a tunnel to indicate the overbreak possibly occurred
due to geological features and conventional blast design. Further, the control of the
number of joint sets on overbreak was observed.
Figure 8 Observed vs. predicted blast results at various locations in approach tunnel
494 A. K. CHAKRABORTY ET AL.
Acknowledgements
The authors are thankful to Director, Central Mining Research Institute for permis-
sion to publish the paper. Thanks are due to the authorities of the mines and tunnels
where the field investigations were conducted.
References
Barton, N., Lien, L. and Lunde, J. (1974) Analysis of rock mass quality and support practice
in tunnelling, and a guide for estimating support requirements, Internal Report of Norwe-gian Geotechnical Inst., Oslo, pp. 6–9.
Bergh-Christensen, J. and Selmer-Olsen, R. (1970) On the resistance to blasting in tunnelling,
International Conference on Rock Mechanics, Belgrade, pp. 59–64.Brown, E.T. (1981) (Ed.) Rock Characterization Testing & Monitoring ISRM Suggested
Methods, Published for The Commission on Testing Methods, International Society
For Rock Mechanics, Pergamon Press, U.K. pp. 6–19.Chakraborty, A.K., Jethwa, J.L. and Dhar, B.B. (1996) Predicting powder factor in mixed-
face condition: development of a correlation based on investigations in a tunnel throughbasaltic flows, Engineering Geology, Elsevier Science B.V., Netherlands, No. 47,
pp. 31–41.Chakraborty, A.K., Pal Roy, P., Jethwa, J.L. and Gupta, R.N. (1998) Blast performance in
small tunnels – a critical evaluation in underground metal mines, Tunnelling and Under-
ground Space Technology, Elsevier Science Ltd., 13(3), 331–339.CMRI Report (2002) An appraisal of geological features and their effect on strata classifica-
tion and tunnelling conditions for approach tunnel of Ghatghar Project, 6p.
Du Pont (1977) Blasters’ Handbook, E.I. du Pont de Nemours & Co. (Inc.), Wilmington,Delaware, pp. 526–541.
Flood, I. and Kartam, N. (1994) Neural networks in civil engineering I: principles and under-
standing, Journal of Computing in Civil Engineering, 8(2), 234–251.Ghose, A.K. (1988) Design of Drilling and Blasting Sub Systems—A Rock Classification
Approach, Proc. Symposium on Mine Planning and Equipment Selection, 1988, Balkema,Rotterdam, pp. 335–340.
Hagan, T.N. (1992) Safe and Cost-Efficient Drilling and Blasting for Tunnels, Caverns, Shaftsand Raises in India, Manual prepared by Golder Associates Pty. Ltd., Australia, pp.12.17, 12.25 and 12.46–12.51.
Holmberg, R. and Persson, P.A. (1978) The Swedish approach to contour blasting, In: Proc.of Annual Conference on Explosives and Blasting Research, Explosives Reference Databaseon CD-ROM, International Society of Explosives Engineers, Ohio, USA, 1997.
Holmberg, R. and Persson, P.A. (1980) Design of tunnel perimeter blasting patterns to preventrock damage, Trans. Inst. of Mining and Metall., London, 89, A37–A40.
Holmberg, R. (1982) Charge calculations for tunnelling, Underground Mining Methods
Handbook, Society of Mining Engineers, American Inst. of Mining, Metall. and Pet.Eng., New York, pp. 1580–1589.
Innaurato, N., Mancini, R. and Cardu, M. (1998) On the influence of the rock mass quality onthe quality of the blasting work in tunnel driving, Tunnelling and Underground Space Tech-
nology, Elsevier Science, Great Britain, 13(1), 81–89.Irrigation Department, Govt. of Maharashtra (1998) Specifications for Underground Excava-
tions, Notifications 67/98, 22 p.
Johansen, J. (1998) Modern trends in tunnelling and blast design, IDL Industries Ltd.,Hyderabad, India, pp. 34–41.
DEVELOPMENT OF RATIONAL MODELS 495
Langefors, U. and Kihlstrom, B. (1973) The Modern Technique of Rock Blasting, John Willey
& Sons, pp. 188–257, 299–301.Leu, S.-S., Lin, S.-F., Chen, C.-K. and Wang, S.-W. (1998) Analysis of powder factors for
tunnel blasting using neural networks, The International Journal for Blasting and Frag-
mentation, A. A. Balkema, Rotterdam, Netherlands, 2(4), 433–448.Lilly, P.A. (1986) An empirical method of assessing rock mass blastability, In: Proc. Large
Open Pit Conference, Australia, IMM, pp. 89–92.
McKenzie, C.J. (1994) Blasting for Engineers, Blastronics Pty. Ltd., Brisbane, Australia.Olofsson, S.O. (1988) Applied Explosives Technology for Construction and Mining, Applex,
Arla, Sweden, 303 pp.Palmstrom, A. (1975) Karakterising av Oppsprekningsgrad og Fjellmassers Kvalitet, Internal
Report by Ing., A. B. Berdal A/S, Oslo, In Norwegian, pp. 1–26.Pokrovsky, N.M. (1980) Driving Horizontal Workings and Tunnels, Mir Publishers, Moscow,
pp. 38–41.
Sayed, T. and Adewahab, O. (1998) Comparison of fuzzy and neural classifiers for road acci-dents analysis, Journal of Computing in Civil Engineering, 12(1), 42–47.
SPSS Inc. (1993) SPSS for WINDOWSTM, Base Systems for User’s Guide Release 6.0,
pp. 311–363.Yu, T.R. and Vongpaisal, S. (1996) New blast damage criteria for underground blasting, CIM
Bulletin, No. 998, 89, 139–145.
496 A. K. CHAKRABORTY ET AL.