International Journal of Rock Mechanics and Mining Sciences 107 (2018) 48–54

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics and Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

Dynamic model validation using blast vibration monitoring in mine backfill T

a,⁎ b c
Muhammad Zaka Emad , Hani Mitri , Cecile Kelly
a
Department of Mining Engineering, University of Engineering & Technology, Lahore, Punjab 54890, Pakistan
b
Department of Mining & Materials Engineering, McGill University, Montreal, Quebec, Canada
c
Birchtree Mine, Vale Manitoba Operations, Thompson, Manitoba, Canada

A R T I C LE I N FO A B S T R A C T

Keywords: Sublevel stoping mining methods with delayed backfill are employed for mining steeply dipping ore deposits at
Underground mining deeper horizons. Cemented rockfill (CRF) is used as a backfilling material for open stopes formed during mining.
Blast-induced vibrations In primary-secondary mining sequence, blast vibrations are induced in CRF during the extraction of adjacent
Numerical modelling stopes. Higher magnitude of blast vibrations may destabilize CRF causing failure and dilution of precious ore.
Calibration
This paper presents a methodology for calibrating a FLAC3D numerical model, encompassing mine backfill with
FLAC3D
blast-induced vibration data recorded from field instrumentation. Two FLAC3D models are prepared, one for
simulating the blasting process and the other is prepared for backfill stability analysis. The first model resulted in
load vectors to be applied to backfill and rock interface. The second model showed that the backfill fails under
the blast vibrations induced by the adjacent production stope. The results of the back analysis of fill failure are
presented and discussed. The methodology may serve as numerical model calibration procedure for mine
backfill.

1. Introduction
Canadian hard rock mines have adopted sublevel stoping method
with delayed backfill or its variations such as vertical block mining
(VBM) for extraction of steeply dipping ore bodies.1 In VBM, the ore-
body is divided into rectangular prism blocks or stopes; which are
mined-out while following pyramidal sequence. Stope production is
carried out in three blast rounds or lifts as shown in Fig. 1. Each lift is
blasted and mucked until sufficient room is available for the anticipated
material from proceeding blast. Backfill is required to fill up the empty
stopes. Cemented rockfill or consolidated rockfill (CRF) is a type of
backfill, made by mixing cement slurry with rock aggregates from ei-
ther development waste or quarry.2,3
Generally, CRF is designed with either limit equilibrium methods4
or simple numerical models, which is rare. Static limit equilibrium
methods are used to estimate the required uniaxial compressive
strength of the backfill material and are based on gravity loading only.4
However many authors believe that blast loading may influence the
stability of CRF, in addition to the static loading.2,4–9 An effective CRF
design should accommodate static loading due to gravity and dynamic
loading from blast-induced vibrations.2,3,8–10 It is reported that blast
vibrations from stopes adjacent to mined ones are perhaps one of the
major causes of CRF failure.2,8,9,11–14 Yu and Counter2 and Yu4 pro-
posed that a safety factor of 2.5 can help CRF to withstand blast
loading. Recently,7 the effect of different simulated blast loads is ex-
amined through numerical modelling study. In that study a stable CRF
stope under static load fails when dynamic load is applied.
In order for dynamic modelling to be used as a reliable tool for the
design of CRF material, it must be first calibrated with field measure-
ments of blast-induced vibrations. This paper presents a stepwise
methodology for numerical model calibration using the results of a field
investigation into the effect of blast-induced vibrations on CRF failure
at an underground mining operation. A FLAC3D dynamic numerical
model is constructed for a typical layout of a primary-secondary stoping
system, encompassing a previously mined and backfilled primary stope.
2. Case study
The case study mine utilize VBM for extraction of a tabular steeply
dipping ore-body with a thickness of 8–30 m. Length of ore-body is
1450 m along strike (N45E) and dip is 85 West. The ore-body is divided
into different zones as per geological settings. The zone studied is lo-
cated at a depth of 839 m. The geology of the ore-body can be generally
described as a brecciated-ultramafic rock in a sulphide matrix. The
geology department of mine terms this rock as peridotite. Footwall rock
comprise of biotite, plagioclase, pyrrhotite and quartz. The hanging-
wall rock is schist in major portion of ore-zone under study.
The massive ore-body is mined out with VBM in transverse retreat

⁎
Corresponding author.
E-mail addresses: zaka@uet.edu.pk (M.Z. Emad), Hani.mitri@mcgill.ca (H. Mitri), Cecile.kelly@vale.com (C. Kelly).

https://doi.org/10.1016/j.ijrmms.2018.04.047
Received 16 January 2017; Received in revised form 25 April 2018; Accepted 29 April 2018
1365-1609/ © 2018 Elsevier Ltd. All rights reserved.
M.Z. Emad et al. International Journal of Rock Mechanics and Mining Sciences 107 (2018) 48–54

Fig. 1. Stope extraction sequence in vertical block mining system: a) longitudinal section; b) plan view.

Table 1 0.87 kg/ton and 0.59 kg/ton. The charge per delay ranged from 31.5 kg
Rock mass properties for different geological units at the case study mine.10,11 to 106 kg for Blast 1, 81.1 to 180.81 kg for Blast 2 and 68.18 to 180.9 kg
Material σt σc E ϕ c υ
for Blast 3.
(MPa) (MPa) (GPa) (°) (°) (MPa) Fig. 4 show the results of monitoring blast vibrations in backfill. The
maximum vibration magnitudes recorded in CRF are 33 mm/s at 0.32 s,
Ore 0.45 22.36 37.8 43 12.70 0.3 86 mm/s at 0.33 s, and 425 mm/s at 1.52 s, for the blasts 1, 2 and 3
Footwall 0.3 13.6 35.8 44 6.25 0.29
Hanging-wall 1.12 47.53 56.85 45 19.38 0.25
respectively.
Cemented Rockfill 0.03 3 2.5 37 1.1 0.35

Table 2
In-situ stress magnitude at level 2750 (839 m).16
σ1 σ2
Herget and Miles (MPa) 31.8 26.7
Orientation Perpendicular to ore Parallel to ore
body body
Computed using FLAC3D 31.75 26.8
(MPa)
Boundary stresses (MPa) 13.39 11.67
primary-secondary panels from hanging wall to footwall. The ore-body
is divided into sublevel with intervals of approximately 30 m. A stope is
mined from top sill and mucked from bottom. Fig. 1 shows extraction of
a typical stope in three blasts. CRF is placed in empty stope through top
sill drive using a load-haul-dump machine. Stope adjacent to CRF is
blasted no earlier than 28 days to ensure curing of CRF. Likewise, VBM
is employed for the extraction of secondary stopes. The rock mass
properties obtained from the mine are shown in Table 1.8,9 The me-
chanical properties of CRF are taken from the CRF testing conducted
by.3 The in-situ stress values at the mining level 2750 (839 m depth) are
shown in Table 2.15
3. Monitoring blast-induced vibrations on-site
In-situ blast vibration monitoring experiment is conducted to
monitor blast vibrations in the CRF from adjacent production stope.
Fig. 2 shows the schematic of the blast vibration monitoring experi-
ment. The details of blast vibration monitoring experiment are pub-
lished by.16
During the experiment two geophones are mounted; one of them
inside a drill-hole in CRF, and the second is mounted on the surface of
the backfill using a bolt that is cement-grouted. The detonation se-
quences employed for the three blast lifts are shown in Fig. 3. Total
explosive charge for Blast 1, 2 and 3 is 1216.9 kg, 2476.86 kg and
3175.6 kg. The total powder factor for Blast-1, 2 and 3 is 1.1 kg/ton,
4. Numerical models
Two FLAC3D18 models are developed for model calibration (Fig. 5).
σ3
The first model is “backfill model” used for simulating mining sequence.
The second model is “equivalent cavity model”, used to derive the ve-
15 locities (or blast vibrations) to be applied in the backfill model. The
Vertical equivalent cavity method is employed by19–23 for simulating blast-in-
14.98
duced vibrations. The primary and adjacent secondary stopes are
modeled with finer mesh to reduce the computation time. The stope is
8.62 18 m long, 12 m wide and has a height of 34 m. CRF and rock interface
is created through ten centimeters thick zones surrounding primary
stope. A small gap of 1 m on top of the backfill is also modeled to
prevent stress load transfer from strata above the backfill.
Model boundaries are placed at a distance of 285 m from the stope.
The model boundaries are transformed to viscous boundaries during the
dynamic analysis. The geo-materials are assumed to be homogeneous-
isotropic for this analysis. After setting boundary conditions and ma-
terial properties, the model is solved for equilibrium using linear-elastic
constitutive model.
Linear-elastic constitutive model is employed to solve the model for
initial equilibrium. The numerical model is solved with elasto-plastic
Mohr-Coulomb material model. Once equilibrium is established the
stope mining and backfilling sequence is simulated with gravity
loading. At this point, the numerical model is ready to receive vibration
data as input parameter on face of CRF. The equivalent cavity method is
used for simulating the effects of blast-vibrations on CRF stope.
The radius of equivalent cavity is considered within three to nine
times the diameter of blast hole.20 The equivalent cavity model is also
generated using FLAC3D code.18 The diameter of blast-hole is 10 cm.
One quarter of the blast hole is simulated as 30 cm diameter equivalent
cavity. The model boundaries are placed 25 m away from the cavity
with the depth of the model to be 35 m. The mesh is graded and mesh
sensitivity analysis is performed. The material is modeled as linear-
elastic and material properties for ore are taken from Table 1. The
length of blast hole charged with explosive is constructed with finer
zones, as shown in Fig. 5.

49
M.Z. Emad et al.
Fig. 2. Schematic showing the blast vibration monitoring setup and observation point.17
4.1. Dynamic analysis
Dynamic loading effects due to blasting on CRF stope are integrated
in the numerical model through simple procedure including conversion
of model boundaries to viscous boundaries, specifying damping coef-
ficient and applying time varying stress load on mined-out stope.
Viscous boundaries can be represented by dashpots installed along the
model boundaries in normal and shear directions. Local damping value
is considered to be 10% for this analysis. Time varying load is applied in
the form of PPV on CRF. The load is applied only once for each blast
hole during the analysis.
4.2. Failure condition
Backfill material fails with development of relaxation zones or
tensile stress in backfill. This is due to the fact that backfill is very weak
in tension. Practically it has even lower tensile strength as the in-situ
strength of CRF is lesser than laboratory determined values due to
segregation, improper mixing and placement.2,3,10 In this work major
and minor principal stresses in CRF are used to define failure in CRF.
CRF fails with the development of tensile stress in filled stope, as CRF is
weak in tension and development of tensile stress in backfill is con-
sidered as failure condition.
4.3. Numerical model calibration with in-situ stresses
The numerical model is calibrated with the field observations in-
cluding in-situ stresses determined by.15 The model calibration is ac-
complished by applying a compressive stress load on the free model
boundaries and observing the stress magnitude inside the backfill
Fig. 3. Detonation sequence employed for blast-3.17
50
International Journal of Rock Mechanics and Mining Sciences 107 (2018) 48–54
model at a location of monitoring.15 The bottom of the model is fixed
vertically. The applied stresses are fine-tuned using trial and error
method to achieve the desired level. The iterative process continued
until the error ratio is reduced to acceptable range of measured stress
values. A normal stress equal to 8.52 MPa is applied on top of the
model. Normal stress of 13.39 MPa and 11.67 MPa are applied on sides
of the model with major stress oriented perpendicular to ore-body. This
may serve as numerical model calibration with in-situ stresses. Table 2
shows results of the calibrated numerical model using stress values.
5. Numerical model calibration with blast-induced vibrations
The dynamic model is calibrated with blast-induced vibrations re-
corded during field monitoring program at the mine. The field data is
incorporated through equivalent cavity method. The model is devel-
oped to determine the magnitude of blast load to be applied on face of
CRF. Calibration process initiates with computing borehole pressure
using Eq. (1).
ρe D 2
Pd =
4 (1)
Where Pd is the borehole pressure, ρe is the density of explosive and D is
the detonation velocity of explosive. Borehole pressures for ANFO and
low-energy ANFO used by mine are 1560 MPa and 930 MPa, respec-
tively. The blast load profile used for applying dynamic load can be
represented by Gaussian curve used by.22 The equation creates positive
skew profile which approaches maximum amplitude at 0.5 millisecond.
P (t ) = Pd t ne−φt (2)
Peak particle velocity (PPV) is computed at a distance of 2 m
M.Z. Emad et al. International Journal of Rock Mechanics and Mining Sciences 107 (2018) 48–54

Fig. 4. Profile of blast load monitored on site: a) Blast-1; b) Blast-2; c) Blast-3.17

(location A) and 12 m (location B) away from blast hole using Eq. (3)24 R −s
ppν = K ⎛ 0.5 ⎞
(Fig. 5b). The locations are calculated from blast design. The equivalent ⎝W ⎠ (3)
cavity is applied with normal stress denoted by P(t). It is ensured that
the numerical model obeys charge weight scaling law presented in Eq. The values of constants K and s are extracted from work done on
(3).24 The average initial stress load Pd for the model is found to be similar geologic units by25 (K = 141, and s = 1.20). The value of radial
around 101 MPa. distance in meters, and charge per delay in kg, are extracted from blast
design for each production blast hole. Once the required magnitude of

Fig. 5. Numerical models for the calibration procedure: a) backfill model; b) equivalent cavity model.

51
M.Z. Emad et al.
Fig. 6. Peak particle velocity computed for blast lifts using numerical model: a) Blast-1; b) Blast-2; c) Blast-3.
PPV is attained, the PPV distributions are extracted for each borehole in
the form of x, y and z components of peak particle velocity. The
damping coefficient is considered as 10% for the built-in local damping
feature in FLAC3D.
Blast load vectors and distributions extracted from equivalent cavity
model are applied on walls of mined-out secondary stope of the backfill
model in terms of x, y and z-velocity. While solving the dynamic model,
PPV is computed inside the CRF at the monitoring station. The fre-
quency of the PPV is 550 Hz for blast-1, 520 Hz for blast-2 and 490 Hz
for blast-3 at the monitoring station. The local damping coefficient
imported from the equivalent cavity model is also applied to the backfill
model to simulate blast wave transmission accurately.
The trapezoidal shaped dynamic load profile is applied on the ex-
posed CRF face for duration of 7 ms. It approaches its maximum mag-
nitude between 3 and 4 ms. The blast load magnitude from equivalent
cavity model and damping coefficient are varied between ± 20% to
equate the peak particle velocities observed on site and the ones cal-
culated from the backfill model. This concludes calibration of one blast-
hole. The entire blasting sequence is simulated and calibration process
is repeated for each blast-hole. Fig. 6 present results of peak vector sum
of velocities computed from the backfill model. This process serves as
numerical model calibration with on-site blast vibration monitoring
results.
52
International Journal of Rock Mechanics and Mining Sciences 107 (2018) 48–54
6. Results and discussion
The mining and backfilling sequence used for this study is the
practice of case study mine for stope extraction. No tensile stress con-
tours appear as a result of blast-1. It can be suggested that the CRF is not
affected by blast-1, though it has the highest powder factor. Blast-1 is
the smallest blast in terms of quantity of explosive consumed, and vo-
lume of rock blasted. The reason that CRF is stable can be attributed to
the support provided to the CRF by the remaining ore block on top of
blast-1. The ore block on top absorbing the reaction of blast-induced
vibrations and confines the block.
The blast-2 is simulated in a similar fashion by applying a series of
dynamic load pulses on the face of the exposed backfill. The results
show that there is a thin band of tensile stress which appears near the
top of the exposed backfill face area of blast-2. It can be suggested that
the backfill is only slightly affected by blast-2. The development of a
thin band of tensile stress band can be considered as an indication of the
onset of a critical condition. Blast-2 is an intermediate blast with re-
spect to quantity of explosive used and volume of rock blasted. The
reason again for no failure is the fact that backfill is supported by re-
maining ore block present on top of the blasted block. The ore block on
top is confining the block and damping reaction of blast vibration.
Fig. 7(a) shows contours of minor principal stress under gravity
loading. As can be seen the CRF block contains no tensile stress. Blast-3
is simulated and results are shown in Fig. 7(b). It can be seen that the
top portion of the exposed CRF is under relaxation due to development
M.Z. Emad et al.
Fig. 7. Minor principal stress contours after blast-3: a) gravity loading; b) blast loading.
of tensile stress. This is an indication of CRF failure as described in the
failure criteria. The blast-3 is the largest in terms of volume of ore
blasted. The reason for failure is clear, there is no more ore left on top of
this blast to support CRF or to counteract the reaction from CRF. The
reaction from blast vibrations destabilizes CRF from the top of the ex-
posed face as there is very little confinement for fill mass. As can be
seen from Fig. 7, the CRF is failing from top of the exposed face and the
failure is wedge shaped. The term wedge shape failure is a generic term
applied to CRF failures but circular failure is the most common mode of
failure in CRF in three dimension. The wedge shape failure is in ac-
cordance with the field observations reported by.5–9,11
Fig. 8(a) shows the profiles of minor principal stresses computed
along the vertical center line of exposed CRF face for both gravity
loading and blast loading cases. It can be seen that the blast loading
profile shows that the top portion of CRF is under the zone of relaxation
as indicated by the tensile stress region of the curve. However the
gravity loading curve shows CRF to be stable. Notably the CRF stope is
failing from the top of the exposed face, as observed at case study and
other mines.5,6,22 Fig. 8(b) shows the profiles of major principal stress
computed along the vertical center line of exposed CRF face for both
gravity loading and blast loading cases. The result of gravity loading
shows no failure in CRF. For blast loading profile it can be seen that the
top most portion of CRF is under the relaxation zone indicated by low
compressive stress.
The profile of the peak particle velocity is also computed along the
Fig. 8. Principal stresses computed at the center line of exposed face.
53
International Journal of Rock Mechanics and Mining Sciences 107 (2018) 48–54
centerline of the exposed CRF face during the dynamic analysis and is
shown in Fig. 9. It can be seen that the higher magnitudes of peak
particle velocities are computed by FLAC3D near the top most part of
the exposed CRF face. Notably the same zones present failure for minor
principal stress shown in Fig. 8. By comparing Figs. 8(a) and 9 it can be
suggested that the peak particle velocities around the magnitude of
278 mm/s or more are initiating failure in CRF. This is consistent with
the observations of22 which is failure at ppv of greater than 250 mm/s
and by10 who suggest that a magnitude of around 300 mm/s damages
CRF.
7. Conclusions
Blast vibration monitoring in CRF and its integration into a nu-
merical model for stability analysis is of prime importance. This paper
presents dynamic numerical model setup for CRF at the case study
mine. The numerical model uses in-situ stress values measured at a
depth of 839 m at the case study mine. This is followed by dynamic
numerical model calibration with blast vibrations, which incorporates
the development of an equivalent cavity model, applying blast load on
equivalent cavity and computing vibration levels in the model with
charge-weight scaling laws. Once the model obeys charge weight
scaling law, the blast vibration distribution is extracted from the
equivalent cavity model and applied to the backfill model. The blast
vibrations in CRF are computed at the monitoring station and blast load
M.Z. Emad et al.
Fig. 9. Peak particle velocity plotted at the centerline of the exposed fill.
is varied to ± 20% to achieve blast vibration magnitude obtained from
the in-situ experiment. The numerical modelling results for Blast-3
show wedge failure of the exposed CRF face at the top of the stope. This
is in agreement with the CRF failures observed at the case study mine
and other CRF operations. Principal stresses and vibration levels are
computed at the centerline of the exposed CRF face. It is found that high
blast vibrations greater than 278 mm/s initiate tensile stresses in the
CRF. The dynamic numerical model calibration methodology presented
in this work successfully replicates the simulation of blast vibrations.
Acknowledgements
This work was jointly funded by a grant from the Natural Sciences
and Engineering Research Council of Canada (NSERC) (235659) and
Vale Canada (234684) under the Collaborative Research and
Development (CRD) Program. The authors are grateful for their sup-
port.
References
1. Zhang Y, Mitri HS. Elastoplastic stability analysis of mine haulage drift in the vicinity
of mined stopes. Int J Rock Mech Min Sci. 2008;45(4):574–593.
2. Yu TR, Counter DB. Backfill practice and technology at Kidd Creek mines. Can Min
Metall Bull. 1983;76(856):56–65.
3. Hassani FP, Archibald J. Mine Backfill. Montreal, Canada: Canadian Institute of
Mining, Metallurgy, and Petroleum. CD-ROM; 1998.
4. Yu TR. Mechanisms of fill failure and fill strength requirements. In: Proceedings of the
6th Canadian Rock Mechanics Symposium. Sudbury, ON; 1992:43–46.
5. Ran J, Watunga T. Application of consolidated rockfill to open stoping in under-
ground mines. In: Proceedings of the RockEng12 – Rock Engineering for Natural
Resources Conference. Edmonton, AB; 2012:197–207.
6. Chen D, Messurier ML, Mitchell B Application of cemented aggregate fill at Barrick’s
darlot gold mine. In: Minefill 2004: Proceedings of the 8th International Symposium on
Mining with Backfill. Beijing, China: 82–89.
7. Emad MZ, Mitri HS, Henning JG. Effect of blast vibrations on the stability of ce-
mented rockfill. Int J Min Rec Environ. 2012;26(3):233–243.
54
International Journal of Rock Mechanics and Mining Sciences 107 (2018) 48–54
8. Emad MZ, Mitri H, Kelly C. Modelling dynamic loading on backfilled stopes in sub-
level stoping systems. In: Proceedings of the SINOROCK2013-3rd ISRM Symposium on
Rock Mechanics, "Rock Characterisation, Modelling and Engineering Design Methods”.
Shanghai; 2013:351–356.
9. Emad MZ, Mitri H, Kelly C. Effect of blast-induced vibrations on fill failure in vertical
block mining with delayed backfill. Can Geotech J. 2014;51(9):975–983.
10. Yu TR. Some factors relating to the stability of consolidated rockfill at Kidd Creek. In:
Proceedings of the 4th International Symposium on Mining with Backfill. Montreal; 1989.
11. Aitchison GD, Kurzeme M, Willoughb DR. Geomechanics considerations in opti-
mising the use of mine fill: Part A: the investigation of the response of fill as a
structural component. In: Proceedings of the Jubilee Symposium on Mine Filling. Mount
Isa; 1973:35–48.
12. Farsangi PN. Improving Cemented Rockfill Design in Open Stoping [Ph.D. Thesis].
Montreal, Quebec: McGill University; 1996.
13. Yumlu M. Backfill practices at Çayeli mine. In: Proceedings of the 17th International
Mining Congress and Exhibition of Turkey-IMCET2001. Ankara; 2001:333–340.
14. Doerner CAC. Effect of Delayed Backfill on Open Stope Mining Methods [MASc. Thesis].
Vancouver: University of British Columbia; 2005.
15. Herget G, Miles P. Ground stress determinations at Thompson and Birchtree mines,
Thompson, Manitoba. In: C.C.F.M.a.E. Technology. Canada: Mining Research
Laboratories; 1979.
16. Emad MZ, Mitri H, Kelly C. In-situ blast vibration monitoring in cemented rockfill
stope – a case study. Int J Min Rec Environ. 2017;31(2):119–136.
17. Emad MZ. Dynamic Performance of Cemented Rockfill Under Blast-induced Vibrations
[Ph.D. Thesis]. Montreal: McGill University; 2013.
18. ITASCA. Dynamic Analysis, FLAC3D Fast Lagrangian Analysis of Continua in 3
Dimensions. Minneapolis, Minnesota: Itasca Consulting Group; 2009.
19. Sharpe JA. The production of elastic waves by explosion pressures I. Theory and
empirical field observations. Geophysics. 1942;7(2):144–154.
20. Kutter HK, Fairhurst C. On the fracturing process in blasting. Int J Rock Mech Min Sci.
1971;8:181–202.
21. Blair DP, Jiang JJ. Surface vibrations due to a vertical column of explosive. Int J Rock
Mech Min Sci Geomech Abstracts. 1995;32(2):149–154.
22. Gool BSv. Effects of Blasting on the Stability of Paste Fill Stopes at Cannington Mine
[Ph.D. Thesis]. Townsville: James Cook University; 2007.
23. Sainoki A, Mitri HS. Numerical simulation of rock mass vibrations induced by nearby
production blast. Can Geotechn J. 2014;51(11):1253–1262.
24. Atlas. Explosives and Rock Blasting. Dallas, Texas, USA: Atlas Powder Company; 1987
[ISBN 0-9616284-0-5].
25. Henning J, Mitri H. Production blast-induced vibrations in longhole open stoping. Int
J Geotechn Earthq Eng. 2010;1(2):1–11.