Saharan 2007
Saharan 2007
1
Central Mining Research Institute – Regional Centre, Nagpur, India
2
Department of Mining, Metals and Materials Engineering, McGill University,
Montreal, Canada
Received January 5, 2006; accepted January 3, 2007
Published online April 22, 2007 # Springer-Verlag 2007
Summary
This paper describes development of a generic nonlinear, dynamic modelling technique to simu-
late discrete rock fractures due to blasting using the finite element method. The element elimina-
tion technique together with a brittle, Rankine failure-type material model are used as a means to
simulate the initiation and growth of fractures in the rock under the effect of blast-induced
dynamic pressure pulse. Dynamic loads representing ideal and non-ideal detonations are simu-
lated and a new method, termed as optimised pressure profile, is proposed to approximate the
pressure-time profile of the blast load to model the dynamic load. Comparison of numerical model
results with previously reported observations from the literature reveals the ability of the model as
a predictive tool and supports the validity of the developed modelling procedure.
Keywords: Finite element method, dynamic analysis, nonlinear modelling, element elimination
technique, discrete fractures, rock fracturing.
List of Symbols
[B] Strain-displacement matrix
[C] Damping matrix
[C] Consistency condition crack tensor written in the crack direction coordinate system
cd Dilatational wave speed at a particular time step in an element
Cp, Cs Longitudinal and transverse shock wave velocity, respectively
[D] Elasticity matrix
E Elastic modulus
f(t) Body forces
f int and f ext Internal resistance and external force vectors
G Unit charge length (kg=m) (TNT equivalent)
[K] Time dependant stiffness matrix
[M] Mass matrix
[N] Shape function matrix
642 M. R. Saharan and H. S. Mitri
1. Introduction
Drilling and blasting is the main and most economical procedure to extract valuable
mineral resources from the earth. Researchers still make different hypotheses in ex-
plaining the fundamental operative mechanisms responsible for rock fracturing by
explosive energy in spite of its prevalent use at a large scale. Enormous experimental
research efforts have been made over the last five decades to understand the rock
fracturing mechanisms. Table 1 summarises the different rock breaking mechanisms
as propounded by various authors. As can be seen, there is no agreement on a unified
theory. The differences may stem from observational difficulties associated with
experimental techniques due to the extremely short duration of the explosive shock
wave (in the order of micro-seconds), a very fast fracturing process covered under
explosive gaseous products and rock debris (fracture propagation speed up to one third
of the shear wave speed in the rock), as well as the heterogeneous nature of the rock.
Moreover, the prevalent experimental techniques developed so far do not have full
control on the experiments. Continuous efforts are being made to develop better ex-
perimental techniques in order to understand the fundamental operative rock fractur-
ing mechanisms. Such an improvement will be a great help in the development of
better explosive products as well as safer and more economical blast designs.
Rapid advances made with the numerical modelling tools and the availability of
powerful computational resources at affordable cost have made numerical simulation
nowadays a most promising tool to study the dynamic rock fracturing processes. The
use of numerical simulation for the dynamic rock fracturing is appealing, essential and
most suitable due to the large number of complex variables involved.
A numerical procedure is established in this study, which encapsulates the relevant
phenomena related to the rock fracturing due to explosive energy. The dynamic load
pulse used in this study considers the role of shock wave in rock fracturing, while the
pressure exerted by gas energy is indirectly considered in the process of the element
elimination technique (EET) using the finite element method. Experimental results
indicate that shock waves are the primary cause of rock fracturing and the gas pressure
Numerical Procedure for Dynamic Simulation 643
Table 1. Blasting theories and their breakage mechanisms (revised after Anon, 1987)
energy has more roles in the fragmentation process (Brinkmann, 1990; Nie and
Olsson, 2000). The role of thermal energy in fracturing is not considered in the current
study due to the fact that the expansion of explosive gaseous products takes place
adiabatically (Hustrulid, 1999). The discrete fracture network growth in the developed
procedure is simulated without introducing geometric or material imperfections. Char-
acteristic models of blast simulation as an explosion in an unbounded thick plate are
presented in this paper, while considering two broad explosive categories of pressure
pulse, namely, ideal- and non-ideal detonation.
In the above, is the material density and [N] is the shape function matrix.
644 M. R. Saharan and H. S. Mitri
The damping matrix, [C], for the Rayleigh damping is in the form
½C ¼ ½M þ ½K ð3Þ
and are pre-defined constants.
The time-dependant stiffness matrix is defined as
ð
½K ¼ BT ½D½BdV ð4Þ
v
[D] and [B] represent the constitutive and strain-displacement matrices, respectively.
The nodal force vector due to surface tractions, [q(t)], and body forces, [f(t)], is
given by
ð ð
fPðtÞg ¼ ½NT qðtÞS þ ½NT f ðtÞdV ð5Þ
s v
The explicit procedure integrates through time by using many small increments. The
central difference operator is conditionally stable, and stability limit for the operator
(with no damping) is given in terms of the highest eigen value in the system as,
2
t ð8Þ
!max
where ! is the circular frequency of the natural modes.
Numerical Procedure for Dynamic Simulation 645
The current dilatational wave speed in the element is calculated with the following
expression
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð^ þ 2^Þ
cd ¼ ð9Þ
where ^ and ^ are effective Lame constants and is the material’s mass density.
Details of the integration procedure are given in finite element textbooks (e.g. see,
Hughes, 2000).
Full model domains are simulated and model boundary truncation due to sym-
metry are not considered. A schematic of the numerical model geometry is shown
in Fig. 1.
mathematical expressions are given below, more detailed information of the brittle
cracking model can be obtained from ABAQUS manual (ABAQUS, 2003),
(a) Crack detection
A simple Rankine criterion is used to detect crack initiation. This states that a
crack forms when the maximum principal tensile stress exceeds the tensile strength of
the material. The Rankine crack detection surface in the plane stress space is shown in
Fig. 2.
(b) Element strain
This is defined as follows:
de ¼ deel þ deck ð10Þ
where de is the total mechanical strain, deel is the elastic strain representing the
uncracked rock, and deck is the cracking strain associated with any existing cracks.
The intact continuum between the cracks is modelled with isotropic, linear elasticity.
(c) Constitutive equation
ds ¼ Del ðde Tdeck Þ ð11Þ
el
where ds is the stress increment, D is the isotropic linear elasticity matrix and T is a
transformation matrix constructed from the direction cosines of the local coordinate
system. T is constant in the present fixed crack model.
(d) Consistency condition
A consistency condition for cracking (analogous to the yield condition in classical
plasticity) written in the crack direction coordinate system has the following tensor
form:
½C ¼ Cðt; sI; II Þ ¼ 0 ð12Þ
where [C] ¼ [Cnn Ctt Css Cnt Cns Cts]T and sI,II represents a tension softening
model (Mode I fracture) in the case of the direct components of stress and a shear
648 M. R. Saharan and H. S. Mitri
Fig. 3. Schematic representation of the functioning of the EET approach using a non-linear brittle failure
model
Numerical Procedure for Dynamic Simulation 649
Table 2. Typical granite rock properties used for the numerical modelling
are possible: actively opening crack state and closing=reopening crack state. The brit-
tle cracking model is characterized by a stress-displacement response rather than a
stress-strain response. This characterization is based on Hilleborg et al. (1976) fracture
energy proposal to avoid unreasonable mesh-sensitive numerical results. Thus, crack
opening displacement is selected as a criterion for element elimination from the
model.
The rock material chosen for the present study is granite, whose properties are well
tested and known. Although crack detection in the present development is based
purely on Mode I fracture considerations due to crack detection by the Rankine failure
criterion, ensuing cracking behaviour includes both Mode I (tension softening) and
Mode II (shear softening=retention) behaviour.
The cracked element is removed from the calculations after an amount of crack
opening displacement is attained at which it can no longer sustain stresses. To
eliminate an element, all components of stress tensors in this element are set to
zero. As a result, all forces in this element become zero as well and therefore, this
element stops to transmit load to neighbouring non-eliminated elements. In order to
avoid numerical problems related to strong local loss of equilibrium, the stress is set
equal to zero in several relaxation steps. The modulus of elasticity in the eliminated
elements is set to zero in the last relaxation step (Mishnaevsky and Schmauder,
2001). A removed element represents a macro crack or fracture in the context of
the present development. Figure 3 provide information about logical steps used
by the EET using the brittle failure material model and Table 2 gives a summary
of the rock properties used for the dynamic simulation as well as reference of
these values.
Velocity components are calculated using the explicit central difference integration
rule as follows:
1 1 tðiþ1Þ þ tðiÞ ðiÞ
u_ ðiþ2Þ ¼ u_ ði2Þ þ €u
2 ð13Þ
1
ðiþ1Þ
u_ ¼ u_ ðiÞ þ tðiþ1Þ €
uðiþ2Þ
where u_ and €u are velocity and
acceleration,
respectively.
The superscript (i) refers to
the increment number and i 12 and i þ 12 refer to mid-increment values. The
central difference integration operator
isexplicit in that the kinematic state may be
advanced using known values of u_ i 12 and € u from the previous increment. Details
of the integration procedures are introduced in the theory manual of ABAQUS (2003).
Fig. 4. Pressure pulse shapes for two categories of detonation (Aimone, 1992; Olsson et al., 2001)
Numerical Procedure for Dynamic Simulation 651
detonation. While the JWL EOS takes into account the rock-explosive interaction,
estimating the correct parameters for non-ideal detonation or in softer rocks poses con-
siderable difficulties (Liu, 1997). The detonation process in most rocks is non-ideal,
thus, the accuracy of using the JWL EOS is questionable.
Input in the form of decay functions is also reported (Duvall, 1953; Jung et al.,
2001; Lima et al., 2002; Olatidoye et al., 1998; Robertson et al., 1994). These equa-
tions are used to replicate the exact waveform, but they require assumptions of some
parameters whose physical significance is unknown.
The use of Gaussian function and triangular load function has also been attempted
to approximate measured dynamic-pulse load. These procedures, however, are not
close to the physical characteristics of the dynamic load and hence carry no physical
meaning. The Gaussian function is mainly introduced to avoid numerical errors asso-
ciated with the application of very high pressure, which is in the order of GPa, in a
very short duration in the order of microseconds.
A new and simple method is proposed by the authors and is termed herein as
‘‘optimized pressure-time profile’’. In this method, the peak borehole pressure and the
pressure at different time scales are required to specify the optimized pressure-time
profile. The peak borehole pressure can be estimated from the following equation
(Atlas, 1987; Clark, 1987; Nie and Olsson, 2000).
Bore Hole Pressure,
VOD2 2
Pb ¼ rc ; Pa ð14Þ
8
where,
¼ explosive density, kg=m3
VOD ¼ Velocity of detonation, m=s
explosive diameter
rc ¼ coupling ratio ¼
bore hole diameter
¼ adiabatic exponent ¼ 1:5 ðPersson et al:; 1994Þ:
Values of detonation velocity (VOD), explosive density () and coupling ratio (rc, ratio
of charge diameter to borehole diameter) are required for estimating the peak borehole
pressure. Table 4 provides values of VOD and along with the source information.
652 M. R. Saharan and H. S. Mitri
According to Eq. (14), the ideal detonation and non-ideal detonation in the present
case will have 2.9 and 1.29 GPa peak borehole pressure, respectively. The pressure-
time profile is constructed with the help of peak pressure value and by applying it over
a time period in a magnitude and a manner consistent with the explosive character-
istics. The complete process and basis of its construction is briefly described below.
The explosion pressure in a borehole decays to a stand-off pressure within a couple
of milliseconds. The stand-off pressure is much smaller than the tensile strength of the
rock. The rise time of explosion pressure to its peak is very short and varies primarily
according to the explosive characteristics and secondly according to the blasthole
diameter, blasthole confinement, rock strength, etc. Generally, the rise time for the
ideal detonation in 38 mm diameter hard brittle rocks is around 25 and 100 ms for the
non-ideal detonation characteristics (e.g. see Jung et al., 2001). The subsequent decay
in the peak borehole pressure down to the stand-off pressure is sharp in the case of the
ideal detonation and slow for the non-ideal detonation. Therefore, to construct pres-
sure-time profile, the peak borehole pressure is fully applied at the respective rise time
of the particular explosive characteristics which increases from zero at time zero. The
peak pressure magnitude drops by 90, 99 and 99.9% over a time period consistent with
the two types of detonation characteristics as shown in Fig. 5. The optimized pressure-
time profile constructed in this way is in good agreement with the observed profiles by
Frantzos (1989), Fourney et al. (1993), Jung et al. (2001) and Daniel (2003). As can be
Fig. 5. Optimized pressure-time profiles of different explosive characteristics simulated in the study for a
38 mm diameter blasthole
Numerical Procedure for Dynamic Simulation 653
Fig. 6. Comparison of different methods for the approximation of blast pressure-time profile
seen from the comparison of Fig. 5, the optimized pressure-time profile better approx-
imates the real pressure pulse than the profiles obtained by Gaussian and triangular
shape functions.
represent infinitesimal dashpots oriented normal and tangential to the boundary. This
arrangement works well provided the waves impinge orthogonally on the boundary.
This means that the boundary should be far enough from the wave source. The wave
source should also behave like a point source in this case. Further, infinite elements
provided by Lysmer and Kuhlemeyer (1969) and adopted in ABAQUS (2003) absorb
almost all of the energy of P (primary) and S (secondary) waves but they are less
efficient in absorbing R (Rayleigh) wave energy (Ramshaw et al., 1998). These infinite
elements therefore represent ‘quiet boundaries’, and not ‘silent boundaries’. The im-
petus set in the current development is the simulation of an explosive column charge
which is in contrast to the point excitation charge and its effect on the near borehole
fracturing process. So it will not be practical to construct computationally unmanage-
able large model domain to represent point source for the waves reaching at the
boundary. The infinite elements, therefore, are not suitable and an alternate arrange-
ment needs to be considered.
Another suggested approach is the insertion of a time-damping term in the wave
equations and an attenuation boundary zone around the discretization mesh. The wave
field goes to zero with time when passing through the attenuation zone. Bing and
Greenhalgh (1998) and Sochacki et al. (1987) studied five kinds of such damper
(linear, power, cubic, Gaussian and exponential) for 2D acoustic and elastic modelling
in the time domain and concluded that the artificial reflections are best reduced by a
linear damper. The Bing and Greenhalgh (1998) study indicates that a larger mesh and
a wider absorbing zone are required for effective results from this linear time damper.
Further, the use of different type of element formulations, one for the solid medium
and another for the wave-absorbing zone, may pose numerical instability problems
during the dynamic analysis due to mesh incompatibility issues. Therefore, this meth-
od is not considered for the present development.
The present development is considering a discrete simulation of fracturing process
around the blasthole. This means that the model itself should absorb the majority of
the energy as it happens in nature. So, prevalent roller boundaries should be accurate
enough to represent the far-field boundary conditions. These boundaries however
should be far enough and far greater than the established norms for static analysis
in order to avoid any effect on the fracturing zone after wave reflection while provid-
ing enough time for the fracture propagation. This can be made possible by consider-
ing full model domains and no truncation of model size due to symmetry planes. As
per the rock material properties (see Table 2) and Eq. (9), the estimated maximum
primary compressional wave speed is 5166 m=s. Also, as per the following relations
by Mosinets and Garbacheva (1972), the likely maximum radius of the fracturing zone
is 1.0 m in the present numerical analyses.
sffiffiffiffiffiffi
Cs p ffiffiffi
Crushed zone radius; rcr ¼ 3q ð15Þ
Cp
rffiffiffiffiffiffi
Cp p ffiffiffi
Fracture zone radius; rf ¼ 3q ð16Þ
Cs
pffiffiffiffiffiffi
Cp p ffiffiffi
Seismic zone radius; rs ¼ 3q ð17Þ
10
Numerical Procedure for Dynamic Simulation 655
where,
q ¼ explosive weight in TNT equivalent, kg
Cp ¼ longitudinal shock wave velocity, m=s
Cs ¼ transverse shock wave velocity, m=s.
Now considering a point 1.0 m away from the borehole centre and far-field boundaries
placed at 10 m away from the borehole boundary, the fastest wave will take approxi-
mately 3.5 ms with a wave speed of 5166 m=s to reach the point under consideration
after reflection from the far-field boundaries. This time is long enough to study the
fracturing process by the primary waves as the waves have peak pressures at 25 and
100 ms, respectively for the two detonation characteristics. So, far-field roller bound-
aries can be placed at 10 m away from the central borehole of 38 mm diameter and this
distance would be equivalent to a more than 250 times the opening size. Furthermore,
a numerical solution time of about 2 ms will provide enough time to capture the
fracturing process by the primary waves and hence it will be used.
(b) Damping of wave energy
There are two reasons for adding damping to a numerical model: 1) to limit
numerical oscillations and 2) to add physical damping to the system (ABAQUS,
2003). ABAQUS has provisions of bulk viscosity in linear and quadratic forms to
limit numerical oscillations. The suggested values for bulk viscosity may be re-
tained during the analysis. Choosing a stiff element having a single integration point
can be the other source of damping to the numerical models to limit numerical
oscillations.
The use of Rayleigh damping with elastic material models is a common practice
in simulating geomechanics applications to simulate the effect of the physical damp-
ing. The damping coefficients are selected either as per the PPV (peak particle veloc-
ity) measurements from specific site data or assign 1 to 15% values (Biggs, 1964;
Massarsch, 1992). It is also suggested that the damping coefficient should not be more
than 0.5% if a linear failure criterion like Mohr-Coulomb is used (Itasca, 2001). This
is because the linear failure criterion itself induces early failure or plasticity. Also, in
rock and soil, the natural damping is independent of frequency (Gemant and Jackson,
1937) so the use of frequency dependent Rayleigh damping may not be justifiable in
such analyses. Another problem reported with the frequency dependent damping is
misrepresentative results due to sometimes over damping of the low frequency modes
(Metzger, 2003), which usually control the solution.
Other types of damping procedures are also reported, such as mass proportional
damping (Bing and Greenhalch, 1998; Takewaki, 2000), dynamic relaxation damping
(Metzger, 2003), etc., but the validity of these methods for hysteretic damping me-
dium such as rock is not well established. Also, such damping arrangements are mass
proportional rather than stiffness proportional. A physical system like rock has natural
frequencies, which are beyond the effective range of the mass proportional damping
methods. Stiffness proportional damping is much needed either in the form of plas-
ticity or by specifying stiffness proportional Rayleigh damping parameter for such
systems. Therefore, arrangements involving mass proportional damping will not be
incorporated in the present development.
656 M. R. Saharan and H. S. Mitri
4. Analysis of Results
The dynamic model set-up above now needs to be validated. It is difficult to compare
numerical model results with an exact fracturing pattern because such a precise knowl-
edge base does not exist. Prevalent experimental techniques have observational difficul-
ties due to the reasons explained in the introduction section. Therefore, the dynamic
model results will be compared with empirical observations reported in the literature.
Figure 9 shows the finite element mesh generated for the dynamic simulation; it
consists of 4560 CST elements and 2312 nodes. A zoom-in view of the model showing
the inner zone of interest around the borehole is displayed in Fig. 10. The central
borehole is subjected to peak borehole pressures of 2.9 and 1.29 GPa to represent ideal
Fig. 10. Zoom-in view of the FE mesh showing the 38 mm diameter blasthole
Numerical Procedure for Dynamic Simulation 659
and non-ideal detonation, respectively, using the optimized pressure-time profile de-
picted in Fig. 5. The material model has the values as shown in Table 1 and the model is
solved using an explicit integration scheme with the dynamic load duration of 2 ms.
The numerical model results for the two types of dynamic load are shown in
Figs. 11 through 17. These results bring out a number of important phenomena, which
are discussed below.
Strain rate dependant rock response
As can be observed from Fig. 11, the stress required to open the first crack with a
high shock load of the ideal detonation is 600 MPa while only 159 MPa is required
to open the first crack in the case of non-ideal detonation. Several researchers have
argued in favour of strain rate dependant rock properties (e.g., Prasad et al., 2000). It is
interesting to note that in the present analysis, the material behaviour, which remains
essentially elastic throughout the calculations as well as the brittle failure law using
the element elimination technique are both rate independent. The element elimination
Fig. 11. Stress magnitudes associated with fracture initiation. (a) Ideal detonation – peak load ¼ 2.9 GPa.
(b) Non-ideal detonation – peak load ¼ 1.29 GPa
660 M. R. Saharan and H. S. Mitri
technique in conjunction with inertia effects, endows the material with attributes that
ultimately account for the ability to capture the load rate effects accurately.
Characteristics of fracturing zone
Comparison of model results between ideal and non-ideal detonations are shown
in Fig. 12 for the fracture pattern at the moment of the application of the peak load. It
has been well observed and documented that ideal detonation leads to more crushing
around the borehole, which follows a large numbers of short length radial cracks (e.g.,
Fig. 12. Fracture pattern at the moment of peak load application. (a) Ideal detonation, time ¼ 1 ms. (b) Non-
ideal detonation, time ¼ 817 ms
Numerical Procedure for Dynamic Simulation 661
McHugh and Keough, 1982). In contrast to this, the non-ideal detonation results in
a smaller crushing zone followed by a few long radial cracks. As can be seen from
Fig. 12, the model results are in good agreement with the reported fracturing behav-
iour. Similar trends are observed for the final fracturing pattern, shown in Fig. 13, at
the end of the blast pressure pulse.
Extent of fracturing zone
Figure 14 presents the extent of fracturing zones for the two detonation types. As
can be seen, the numerical model predicts fracturing zones having radii of 487 and
Fig. 13. Final fracturing pattern. (a) Ideal detonation. (b) Non-ideal detonation
662 M. R. Saharan and H. S. Mitri
341 mm for the ideal and non-ideal detonation, respectively. Mosinets and Garbacheva
(1972) and Kexin (1995) provide empirical relations to predict the extent of crushing
and fracturing zone. Kexin (1995) provides the following relation
18
G 1
fractures zone radius; r ¼ 96 ð10EÞ6 ð19Þ
10c
where, r is the fractures zone radius (mm), c is the uniaxial compressive strength
(MPa), E is the elastic modulus (MPa), G is the unit charge length (kg=m) (TNT
Fig. 14. Fracturing pattern at the end of solution time (2 ms). (a) Ideal detonation. (b) Non-ideal detonation
Numerical Procedure for Dynamic Simulation 663
equivalent). Using G ¼ 1.435 and 1.13 kg=m (for respective explosive characteristics),
c ¼ 200 MPa and E ¼ 60 103 MPa, Eq. (19) predicts 356 and 348 mm for the
respective quantities thus showing close agreement with the numerical model pre-
dictions. It is noteworthy that the present modelling procedure makes it easy to dis-
tinguish between the crushing zone and the fracturing zone. It is also important to
Fig. 15. Kinetic, internal and artificial energy plots. (a) Ideal detonation, PPV at 0.5 m from the borehole
(point coordinates ¼ 0.5, 0.0). (b) Non-ideal detonation, PPV at 0.25 m from the borehole (point coor-
dinates ¼ 0.25, 0.0)
664 M. R. Saharan and H. S. Mitri
mention that in the present modelling procedure, fractures are neither pre-placed in the
model domain nor their growth paths are pre-specified.
Energy utilisation in fracturing
Numerically, the kinetic energy, which is provided by the explosive energy, in the
case of the ideal detonation drops from the peak value of 14.994 to 14.00 kJ (Fig. 15a)
while it drops from 5.97 to 5.80 kJ (Fig. 15b) in the case of the non-ideal detonation.
The consumption of kinetic energy amounts to 6.63% in the case of ideal detonation
and 2.85% in the case of non-ideal detonation. The fracturing process is the only
source of energy absorption in the numerical models as no artificial damping is used.
These observations are in line with the observations reported by Langefors and
Kihlstorm (1978) and Nicholls and Hooker (1962). Furthermore, the numerical mod-
elling results validate the well acknowledged fact that a very small explosive energy is
used for fracturing and the majority of the explosive energy is lost in wave propaga-
tion, heat, sound and air-overpressure (Lownds and Du Plessis, 1984).
ABAQUS (2003) reports that the ratio of kinetic energy to internal energy (which
includes recoverable strain energy and energy consumed in plastic work) should not
be more than 10% to ensure accuracy of the dynamic numerical procedure. Also,
artificial strain energy, which is an indicator for hourglass effect, should be negligible
Fig. 16. Peak particle velocity plots (PPV) in the X-axis direction
Numerical Procedure for Dynamic Simulation 665
to assure accuracy of the dynamic numerical modelling. These conditions are also well
complied with as can be seen from Fig. 15.
Fracturing distance and PPV
Typical single point PPV observations are plotted in Fig. 16 for a point 0.5 m away
from the borehole in the case of ideal detonation and 0.25 m away in the case of non-
ideal detonation. PPV plots over a 10 m distance away from the borehole are also
plotted for the two types of detonation in Fig. 17. The plots shown in Fig. 16 are in
good resemblance with the routine field PPV measurements where PPV measurements
show reduction in magnitude and frequency with the increase of distance from the
source. This reduction is due to damping of the wave energy provided by the earth. It
is noteworthy here that no artificial damping is used in the present procedure yet the
explosive energy rapidly damps out, which is similar to field observations.
Prediction of rock fracturing based on PPV observations is a common practice in
mining. Several predictor equations are based on this context and most notably by
Holmberg and Persson (1979). They predict that fractures are most likely to take place
where PPV is more than 1000 mm=s. The PPV trends presented in Fig. 17 predict
similar trends as reported by Holmberg and Persson (1979). Also the extent of frac-
tured zones is limited to the distance with PPV values greater than 1000 mm=s.
Fig. 17. PPV trends for the two types of detonation characteristics
666 M. R. Saharan and H. S. Mitri
The results presented in Figs. 11–17 also validate the assumptions made during
the development of the new numerical procedure. The following illustrates these
validations.
(i) Modeled domain size
PPV plots presented in Fig. 16 indicate that the fracture network obtained is due to
the primary shock waves only. The PPV plots do not indicate the presence of second-
ary loading due to the reflected waves. Therefore, the choice of model domain size
was appropriate for such analyses.
(ii) Damping of the wave energy
PPV plots presented in Fig. 16 as well as PPV trends obtained by Fig. 17 illustrate
that the selected material model represents damping characteristics associated with the
natural material; no artificial damping was considered in the developed procedure.
Also, the material model shows enough potential to accurately simulate two distinct
fracturing characteristics, crushing and cracking, with a single parameter for element
elimination.
(iii) Duration of the analysis
Energy plots presented in Fig. 15 as well as stable fracture network illustrated in
Figs. 13 and 14 indicate that the results obtained are stable in nature. These results also
indicate that enough time is provided for fracture propagation. Therefore, the simulation
time adopted in the analysis is appropriate to obtain a stable fracture network.
(iv) Model boundary conditions
Roller boundaries were selected to represent unbounded rock medium. Stable
fractures networks, energy plots and PPV plots presented in Figs. 15–17 bring out
that the results are nowhere affected by the spurious reflection of waves from the
artificial boundaries. Hence, the selection of roller boundaries in the present modelling
procedure is justified.
5. Conclusions
This paper presents in detail a nonlinear numerical modelling procedure for the dy-
namic modelling of rock fracturing by blasting. This includes description and justifi-
cation of adopted model domain size and its boundary conditions, material model,
dynamic load as well as element type and size. A new method, named as optimized
pressure pulse, is introduced to approximate the observed pressure-time profile due to
the blast load. The results obtained from the dynamic numerical model are evaluated
and validated against established empirical knowledge base and observed dynamic
rock behaviour. The numerical results also confirm the validity of the assumptions
made during the development of the numerical procedure, particularly with regard to
model size, roller-type boundary conditions and damping. The developed procedure
thus presents a promising potential for many engineering blasting research problems.
One important limitation of the current modelling technique is that it can not be
used for investigating fracturing which ultimately results in complete fragmentation.
This is due to the fact that the model domain essentially remains continuum through-
out the solution time despite the use of the element elimination technique.
Numerical Procedure for Dynamic Simulation 667
Acknowledgements
The work presented in the paper is a part of research work done for PhD thesis of the first author.
The research work is supported by several organisations and institutes, most notably, Central
Mining Research Institute, Dhanbad, India (CMRI-India); Natural Sciences and Engineering
Research Council of Canada, Canada (NSERC-Canada) and J. W. McConnel Foundation, McGill
University, Canada. The authors are grateful for their generous support. Authors also express their
gratitude towards anonymous reviewers of the paper for improving the quality of the paper.
References
ABAQUS (2003): Abaqus 6.4-1 user’s manuals. Abaqus Inc., Pawtucket, RI, USA.
Aimone, C. T. (1992): Rock breakage: Explosives, blast design. In: Hartman, H. L. (ed.), SME
Mining Engineering Handbook. Society of Mining Engineers, Littleton, 722–746.
Anon (1987): Explosives and rock blasting. Atlas Powder Company. Washington, USA.
Atlas (1987): Explosives and rock blasting. Atlas Powder Company. Washington, USA.
Batzle, M. L., Simmons, G., Siegfried, R. W. (1980): Microcrack closure under stress: direct
observation. J. Geophys. Res. 85, 7072–7090.
Bhandari, S., Vutukuri, V. S. (1974): Rock fragmentation with longitudinal charges. In: Proc. 3rd
Int. Cong. Rock Mech. Denver, US, 1337–1342.
Biggs, J. M. (1964): Introduction to Structural Dynamics. New York, McGraw-Hill.
Bing, Z., Greenhalgh, S. (1998): A damping method for the computation of the 2.5 green’s
function for arbitrary acoustic media. Geophys. J. Int. 133(1), 111–120.
Blair, S. C., Cook, N. G. W. (1998): Analysis of compressive fracture in rock using statistical
techniques: Part I. A non-linear rule-based model. Int. J. Rock Mech. Min. Sci. 35, 837–848.
Braithwaite, W., Brown, W. B., Minchinton, A. (1996): The use of ideal detonation computer
codes in blast modelling. In: Mahanty, B. (ed.), Proc. 5th Int. Sym. Rock Fragmentation by
Blasting (FRAGBLAST 5), Montreal, Quebec, Canada. Rotterdam, Balkema, 37–44.
Brinkmann, J. R. (1987): Separating shock wave and gas expansion breakage mechanisms. In:
Proc. 2nd Int. Symp. Rock Fragmentation by Blasting.
Brinkmann, J. R. (1990): An experimental study of the effects of shock and gas penetration in
blasting. In: Proc. 3rd Int. Symp. Rock Fragmentation by Blasting, FRAGBLAST 3, Brisbane,
Australia, August 26–31, Published by AusIMM, 55–66.
Cho, S. H., Miyake, H., Kimura, T., Kaneko, K. (2003): Effect of the waveform of applied
pressure on rock fracture process in one free-face. J. Sci. Tech. Energetic Mat. 64(3),
116–125.
Clark, G. B. (1987): Principles of Rock Fragmentation. John Wiley & Sons Inc., London, p 610.
Daehnke, A., Rossmanith, H. P., Knasmillner, R. E. (1996): Blast-induced dynamic fracture
propagation. In: Mohanty, B. (ed.), Proc. 5th Int. Symp. Rock Fragmentation by Blasting,
FRAGBLAST 5, Montreal, Canada, 619–626.
Daniel, R. (2003): Pressure-time history of emulsion explosives. Personal communications.
Dawding, C., Labuz, J. F., Shah, S. P. (1985): Experimental analysis of crack propagation in
granite. Int. J. Rock Mech. Min. Sci. 22, 85–99.
Diederichs, M. S. (1999): Instability of hard rockmasses: The role of tensile damage and
relaxation. (Unpublished) PhD thesis, University of Waterloo, Canada, p 610.
668 M. R. Saharan and H. S. Mitri
Donze, F. V., Bouchez, J., Magnier, S. A. (1997): Modeling fractures in rock blasting. Int. J. Rock
Mech. Min. Sci. 34(8), 1153–1163.
Duvall, W. I. (1953): Strain wave shapes in rock near explosions. Geophysics 18(2), 310–323.
Etoh, S., Hamashima, H., Murata, K., Kato, Y. (2002): Determination of JWL parameters from
underwater explosion tests. In: 12th Int. Detonation Symp. (poster session). San Diego,
California.
Fourney, W. L., Dick, R. D., Wang, X. J., Wei, Y. (1993): Fragmentation mechanism in crater
blasting. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 30(4), 413–429.
Frantzos, D. C. (1989): Finite element analysis of radial cracking mechanism around blastholes
using measured pressure-time curves for low density ammonium nitrate=fuel oil. (Unpub-
lished) PhD thesis. Dep. Mining Eng., Queen’s Univ., Canada, p 290.
Gemant, A., Jackson, W. (1937): The Measurement of Internal Friction in Some Solid Dielec-
tric Materials. The London, Edinburgh and Dublin Philosophical Magazine J Science XXII,
960–983.
He, H., Liu, Z., Nakamura, K., Abe, T., Wakabayashi, K., Okada, K., Nakayama, Y., Yoshida, M.,
Fujiwara, S. (2002): Determination of JWL equation of state parameters by hydrodynamically
analytical method and cylinder expansion test. Jl. Japan Explosives Soc. 63(4), 197–203.
Hilleborg, A., Modeer, M., Petersson, P. E. (1976): Analysis of Crack Formation and Crack
Growth in Concrete by Means of Fracture Mechanics and Finite Elements. Cement Concrete
Res. 6, 773–782.
Holmberg, R., Persson, P. A. (1979): Design of tunnel perimeter blasthole pattern to prevent rock
damage. Tunneling ‘79. Inst. Min. Metall., London, 280–283.
Hughes, T. J. R. (2000): The finite element method: linear static and dynamic finite element
analysis. Dover publications, NY, p 682.
Hustrulid, W. (1999): Blasting principles for open pit mining, Vol. I – general design concepts.
Balkema, Rotterdam, p 382.
Itasca (2001): FLAC3D 2.0 manuals. Itasca consulting Group Inc. Minneapolis, MN, USA.
Jung, W. J., Utagava, M., Ogata, Y., Seto, M., Katsuyama, K., Miyake, A., Ogava, T. (2001):
Effects of rock pressure on crack generation during tunnel blasting. Jl. Japan Explosives Soc.
62(3), 138–146.
Kexin, D. (1995): Maintenance of roadways in soft rock by roadway-rib destress blasting. China
Coal Society 20(3), 311–316 (In Chinese).
Kikuchi, N. (1983): Remarks on 4CST elements for incompressible materials. Comput. Methods
Appl. M. 37, 109–123.
Kranz, R. I. (1983): Microcracks in rocks: a review. Tectonophysics 100, 449–480.
Kutter, H. K. (1967): The interaction between stress wave and gas pressure in the fracture
process of an underground explosion in rock, with particular application to presplitting.
(Unpublished) PhD Thesis. Univ. of Minnesotta, Minneapolis, p 234.
Kutter, H. K., Fairhurst, C. (1971): On the fracture process in blasting. Int. J. Rock Mech. Min.
Sci. 8, 181–202.
Langefors, U., Kihlstorm, B. (1978): The modern techniques of rock blasting. 3rd edn., John
Wiley & Sons Inc., New York, p 438.
Lima, A. D. R., Romanel, C., Roehl, D. M., Araujo, T. D. (2002): An adaptive strategy for the
dynamic analysis of rock fracturing by blasting. In: Proc. Int. Conf. Computational Eng. &
Sci. (ICES’02). Reno, Nevada.
Numerical Procedure for Dynamic Simulation 669
Author’s address: Hani S. Mitri, Department of Mining, Metals and Materials Engineering,
McGill University, 3450 University Street, Montreal, Canada H3A 2A7; e-mail: hani.mitri@
mcgill.ca