BLAST VIBRATION MODELING USING IMPROVED SIGNATURE HOLE

TECHNIQUE FOR BENCH BLAST

DISSERTATION


A dissertation submitted in partial fulfillment of the
requirements for the degree of Doctor of Philosophy in the
College of Engineering at the
University of Kentucky

By

Jhon Jairo Silva-Castro

Lexington, Kentucky

Director: Dr. Braden T. Lusk, Associate Professor of Mining Engineering



Lexington, Kentucky

2012

Copyright Jhon Jairo Silva-Castro 2012















ABSTRACT OF DISSERTATION




BLAST VIBRATION MODELING USING IMPROVED SIGNATURE HOLE
TECHNIQUE FOR BENCH BLAST



Mining blast vibration prediction is a complex task due to the complexity of the
variables involved in the problem. The lack of consistency in the blast parameters such
as the geometry of the blastholes, the composition of the explosives, and the geology in
mining operations make each blast a unique event. Despite the technological advances in
the application of blasting, the design and the prediction of the results is based on
empirical equations, or in best cases statistical information, with a limited or no
theoretical support.
The objective of this research was to improve signature hole technique with a new
methodology. The scatter in the initiation system, the geology, the consistency of the
explosives, the changes in the vibration path between the source and the monitoring point
and the geometry of blastholes are considered in the methodology. Parameters including
the initiation timing, the traveling time of the vibration waves, and the vibration
waveform generated by each hole are assigned a random behavior. To randomize the
vibration waveform for each hole, one equation was developed base on Fourier series.
An equation called the Silva-Lusk equation captures the main properties of the vibration
waveform for the location where blast vibrations are under study. Every time a hole is
blasted, the methodology generates a complete random vibration waveform for such hole
using random normal distributions for the amplitude of the signal, the frequency content
and the attenuation of the signal. In the proposed methodology, to superpose the random
signals of each blasthole and assess the complete vibration waveform, a Monte Carlo
scheme is used. Using this technique, a series of likely waveforms are generated. When
all the likely outputs are plotted, an envelope waveform is generated containing the actual
vibration for the blast. Along with the envelope, a peak particle velocity histogram is
generated, providing an opportunity to assess the vibration levels measuring the
percentage of confidence in the final result. The validation of the proposed methodology
was achieved through several field blasting tests performed in a surface coal mine in
West Virginia. Recommendations and future work are provided to improve the
methodology.
KEYWORDS: Electronic detonators, Waveform, Signature hole,
Fourier Series, Blast vibration.




































___________________________
Students Signature

____________________________
Date














BLAST VIBRATION MODELING USING IMPROVED SIGNATURE HOLE
TECHNIQUE FOR BENCH BLAST



By

Jhon Jairo Silva-Castro












______________________________
Director of Dissertation



______________________________
Director of Graduate Studies



_____________________________






0B















"Quasi nanos gigantum humeris insidentes Bernard of Chartres
This dissertation is dedicated to the people who were, are and always will be part
of my life. Thanks to my parents, Trinidad and Pedro, for my being, for all your love,
and all the life lessons that you are still giving me. To my brothers, Henry and Daniel,
who always believed in my capabilities and for whom I have been trying to be a role
model. To Trinidad Melo, thanks for all her support and encouragement. To Eduardo
Torres, my mentor, who guided me into the mining industry, thanks for his advice and
friendship. To Edmundo, who started the spark that made all this possible. Without his
encouragement and the support of his lovely family I would be in another world, nowI
am glad to be where I am. To Isabel, who I have the privilege to call my friend, thanks
for her support, and the technical and philosophical discussions through all this time. To
my friends and mentors Braden and Shannon for all their support and encouragement
during my studies. I would like to thank my fellow graduate students: Kumar, Todor,
J osh and Gosh.
Finally, thanks to Sandra my lovely wife who has always been there for me.
Thanks for her support and encouragement, thanks for your patience and understanding
along the way to achieve this goal; without her and the baby we are expecting, this
accomplishment would not have any meaning. Sandra, you are giving a beautiful
meaning to all this effort, I love you, a thousand thanks.











iii

ACKNOWLEDGEMENTS

Thanks to the entire faculty and staff of the Mining Engineering Department here
at The University of Kentucky. Special thanks to those who served on my committee;
Dr. Bryson, Dr. Honaker, Dr. Wala, Dr. Perry and Dr. Kalinski. The comments you
provided on my dissertation where very valuable and improved a lot this work, thank
you.
Thank you to Rick Townsend of Nelson Brothers. You helped me setup many of
the seismographs and collect the information. Another especial thanks goes to Ken
Eltschlager of OSM you provide valuable comments during this research.
Finally thanks Dr. Lusk, you have been the perfect mentor for me, thanks for your
support, encouragement and guidance.










iv

TABLE OF CONTENTS
ACKNOWLEDGEMENTS ................................................................................... iii
TABLE OF CONTENTS ....................................................................................... iv
LIST OF TABLES ............................................................................................... viii
LIST OF FIGURES ............................................................................................... ix
Chapter 1 ................................................................................................................. 1
INTRODUCTION ...................................................................................................1
1.1 Background ............................................................................................ 1
1.2 Fundamentals of blast vibration ............................................................ 2
1.2.1 Characteristics of the explosive ....................................................... 4
1.2.2 Initiation system used ...................................................................... 4
1.2.3 Shot Geometry and Timing ............................................................. 5
1.2.4 Distance source monitoring point ................................................... 5
1.2.5 Geology ........................................................................................... 5
Chapter 2 ................................................................................................................. 7
BLAST VIBRATION ..............................................................................................7
2.1 Blast vibration characteristics ................................................................ 7
2.2 Blast vibration measurement ................................................................. 9
2.3 Blast vibration prediction .................................................................... 11
2.3.1 Scaled distance estimation ............................................................ 11
2.3.2 Analytical and numerical approaches ........................................... 14
2.3.3 Signature hole technique ............................................................... 18
Chapter 3 ............................................................................................................... 26
DETAILED DISCUSSION OF CURRENT SIGNATURE HOLE TECHNIQUE26
3.1 Introduction ......................................................................................... 26
3.2 Blast vibration energy .......................................................................... 26
3.3 Seed waveform .................................................................................... 27
3.4 Explosion sequence ............................................................................. 31
3.5 Linear superposition ............................................................................ 34
Chapter 4 ............................................................................................................... 37
v

MODEL DEVELOPMENT ...................................................................................37
4.1 Introduction ......................................................................................... 37
4.2 Monte Carlo scheme ............................................................................ 37
4.2.1 Discrete Random Variables ........................................................... 37
4.2.2 General Scheme of the Monte Carlo Method ................................ 41
4.2.3 Pseudorandom numbers generation .............................................. 42
4.3 Seed waveform variability ................................................................... 42
4.3.1 Modeling of changes in signature waveform ................................ 45
4.3.1.1 Signature waveform comparison ............................................ 48
4.3.1.2 Fourier series and Signature Waveform ................................. 51
4.3.1.3 Signature Waveform approach based on Fourier series ......... 55
4.3.1.4 Numerical example random signature waveform generation 65
4.4 Wave arrival time distribution and time sequence .............................. 67
4.4.1 Wave arrival time .......................................................................... 67
4.4.2 Blasting time sequence .................................................................. 70
4.4.2.1 Experimental setup ................................................................. 70
4.4.2.2 Non-electric detonators results ............................................... 76
4.4.2.1 Electronic detonators results .................................................. 80
4.5 Linear superposition and discrete convolution .................................... 84
4.5.1 Practical example .......................................................................... 89
Chapter 5 ............................................................................................................... 95
IMPROVED SIGNATURE HOLE TECHNIQUE VALIDATION......................95
5.1 Introduction ......................................................................................... 95
5.2 Field Experiments ................................................................................ 95
5.2.1 Instrumentation and data collection .............................................. 95
5.2.1.1 Site description ....................................................................... 95
5.2.1.2 Instrumentation....................................................................... 96
5.3 Tests description ............................................................................... 102
5.4 Analysis and results of the models using improved signature hole
technique 107
5.4.1 Model 1. Airblast modeling (test No.5) ...................................... 107
vi

5.4.1.1 Signature test No.5 ............................................................... 108
5.4.1.2 Synthetic signature signals ................................................... 109
5.4.1.3 Timing sequence .................................................................. 110
5.4.1.4 Results Model 1 .................................................................... 111
5.4.2 Model 2. Particle Velocity (test No.10) ...................................... 116
5.4.2.1 Signature test No.10 ............................................................. 117
5.4.2.2 Synthetic signature signals ................................................... 118
5.4.2.3 Timing sequence .................................................................. 119
5.4.2.4 Results Model 2 .................................................................... 120
5.4.3 Model 3. Particle Velocity (test No.12) ...................................... 123
5.4.3.1 Signature test No.12 ............................................................. 125
5.4.3.2 Synthetic signature signals ................................................... 126
5.4.3.3 Timing sequence .................................................................. 127
5.4.3.4 Results Model 3 (test No.12)................................................ 128
5.4.4 Model 4 Pyrotechnic initiation (test No.14) ................................ 131
5.4.4.1 Signature test No.14 ............................................................. 133
5.4.4.2 Synthetic signature signals ................................................... 134
5.4.4.3 Timing sequence .................................................................. 135
5.4.4.4 Results Model 4 (test No.14)................................................ 136
5.4.5 Model 5 electronic initiation system (test No.15) ....................... 139
5.4.5.1 Signature test No.15 ............................................................. 140
5.4.5.2 Synthetic signature signals ................................................... 140
5.4.5.3 Timing sequence .................................................................. 140
5.4.5.4 Results Model 5 (test No.15)................................................ 141
Chapter 6 ............................................................................................................. 143
DISCUSSION OF RESULTS AND FUTURE WORK RECOMMENDATIONS143
6.1 Discussion about the improved methodology and the validation results
143
6.1.1 Number of frequencies to model the signature signal ................. 143
6.1.2 Decay factor ................................................................................ 151
6.1.3 Initiation system, timing and blast vibration ............................... 151
vii

Chapter 7 ............................................................................................................. 159
CONCLUSIONS AND FUTURE WORK ..........................................................159
7.1.1 Conclusions ................................................................................. 159
7.1.2 Novel contributions ..................................................................... 161
7.1.3 Recommendations for Future Work ............................................ 161
REFERENCES ....................................................................................................163
VITA ....................................................................................................................170
APPENDIX A ......................................................................................................171
Blasting log reports from the mine ......................................................................171
APPENDIX B ......................................................................................................199
Plan Layout of tests performed in 2011 ...............................................................199
APPENDIX C ......................................................................................................206
Vibration records for events recorded in 2011 ....................................................206

viii

LIST OF TABLES


Table 2-1 Range of typical blast parameters (after Dowding, 1985) ...................... 7
Table 4-1 Statistical parameters to modeling random signals .............................. 66
Table 4-2 Typical rock velocities (from Bourbi, Coussy, and Zinszner,
Acoustics of Porous Media) .................................................................................. 69
Table 4-3 Detonator Matrix .................................................................................. 74
Table 4-4 Summary of Results for Non-Electric Validation Example ................. 76
Table 4-5 Summary non-electric detonator results ............................................... 78
Table 4-6 Summary of Results for Electronic Validation Example ..................... 80
Table 4-7 Summary statistics electronic system ................................................... 81
Table 4-8 Blast test parameters ............................................................................. 89
Table 5-1 Seismograph location and their characteristics .................................... 98
Table 5-2 Seismographs triggering parameters ................................................. 100
Table 5-3 Triggering levels used in the second round of test ............................. 102
Table 5-4 Tests including signature hole ............................................................ 103
Table 5-5 Blast test parameters test No.5 ........................................................... 107
Table 5-6 Blast test parameters test No.10 ......................................................... 116
Table 5-7 Blast test parameters test No.12 ......................................................... 123
Table 5-8 Blast test parameters test No.14 ......................................................... 131
Table 5-9 Blast test parameters test No.15 ......................................................... 139
Table 6-1 Comparison results according to the number of frequencies used ..... 145
Table 6-2 Decay factor signature waveforms at 3599 seismograph location ..... 151
Table 6-3 Results test 14 and 15 particle velocity peak values (actual readings)153
Table 6-4 Delay timing and initiation system simulation far seismographs (beyond
1440ft) ................................................................................................................. 155
Table 6-5 Modeling results for seismograph 4906 (Pyrotechnic vs electronic)
Radial component ............................................................................................... 156


ix

LIST OF FIGURES
Figure 1.1 Waveform idealization in a blast event (Adapted from Dowding,
1985). ...................................................................................................................... 3
Figure 1.2 Types of Detonators (After Miller, D., Drew, M. 2007) ....................... 4
Figure 1.3 Factors affecting ground vibration (after Khandelwal, M and Singh, T.
2006) ....................................................................................................................... 6
Figure 2.1 Typical earthquake and blast waveforms .............................................. 8
Figure 2.2 Blast vibration recommended regulation Z curve. Safe level blasting
criteria from USBM RI 8507 (After Siskind, 2000) ............................................... 9
Figure 2.3 Scaled distance law, analysis for different components (Lusk, Silva,
Eltschlager, Hoffman, 2010) ................................................................................. 14
Figure 2.4 Main components of rock fracturing modeling process (Adapted from
Saharan, Mitri and Jethwa, 2006) ......................................................................... 15
Figure 2.5 pressure pulse shape A. Ideal detonation. B. Non-ideal detonation
(Aimone, 1992., Olsson et al., 2001) .................................................................... 16
Figure 2.6 Sketch of a system with continuous and discrete signals. ................... 18
Figure 2.7 Time Invariant systems single input - single output (Adapted from
Signal and Systems 3th edition) ........................................................................ 19
Figure 2.8 SI-SO, C, L, TI systems (Adapted from Signal and Systems 3th
edition) .................................................................................................................. 20
Figure 2.9 Systems Theory and Signature Hole Technique similarity. ............... 21
Figure 2.10 Finite impulse response from one hole blasted ................................ 22
Figure 2.11 Summary of the most common signature hole techniques currently
available (After Spathis, 2010) ............................................................................. 25
Figure 3.1 Particle velocity spectra for single charge shots of short column weight
(After Crenwelge 1988) ........................................................................................ 28
Figure 3.2 Particle velocity spectra for single charge shots of tall column weight
(After Crenwelge 1988) ........................................................................................ 29
Figure 3.3 Particle velocity spectra for single charge shots (a) short column
weight (b) tall column weight (Crenwelge 1988) ................................................. 29
Figure 3.4 Seismograms recorded for three different charge weights and two
locations (After M. Sakamoto, et al., 1989) .......................................................... 30
Figure 3.5 Seismograms recorded for three different charge weights (After
Bonner, et al., 2008) .............................................................................................. 30
Figure 3.6 Spectral analysis for three different charge weights (After Bonner, et
al., 2008) ............................................................................................................... 31
Figure 3.7 Explosion sequence function ............................................................... 32
Figure 3.8 Graphical representation of Equation 3.7 ............................................ 32
Figure 3.9 Systems homogeneity property ......................................................... 34
Figure 3.10 Numerical homogeneity concept between the quantity of explosive
and the particle velocity. ....................................................................................... 35
Figure 3.11 Additivity concept for signature hole technique .............................. 36
Figure 4.1 Histogram for the ronJn function in Matlab. .................................. 42
Figure 4.2 Signature hole reproducibility (adapted from Blair 1999) .................. 43
Figure 4.3 General view of the tested area ........................................................... 45
Figure 4.4 Test plan layout ................................................................................... 46
x

Figure 4.5 Seismograph network at Guyan mine .................................................. 47
Figure 4.6 Complete signature waveform three components ............................... 47
Figure 4.7 Complete waveform, Radial component Seismograph 01 .................. 48
Figure 4.8 Signature waveform comparison ......................................................... 49
Figure 4.9 Frequency content of the signals and four frequency zones ................ 50
Figure 4.10 Signature 1 Isolated from complete record (Radial component) ....... 52
Figure 4.11 Magnitude coefficients from Fourier series ...................................... 53
Figure 4.12 FFT and magnitude coefficients from Fourier series ........................ 54
Figure 4.13 Fourier series and signature 1 ............................................................ 54
Figure 4.14 Measured signal Vs base equation .................................................... 58
Figure 4.15 Exponential decay calculation ........................................................... 59
Figure 4.16 Measured signal Vs base equation including exponential decay factor
............................................................................................................................... 60
Figure 4.17 Measured signal Vs final approach ................................................... 61
Figure 4.18 Frequency domain comparison .......................................................... 62
Figure 4.19 Frequency parameter for Equation 4.48 ............................................ 64
Figure 4.20 Numerical example frequencies signature 1 test 01 ......................... 65
Figure 4.21 Random generated signatures using Silva-Lusk equation. ................ 66
Figure 4.22 Signature signals measured vs Generated using one and three signals
............................................................................................................................... 67
Figure 4.23 Delay times involved in the signature analysis ................................. 68
Figure 4.24 Illustration of Interaction between Break Wire and Counter ............ 71
Figure 4.25: Blasting Machine Counter Interaction ............................................. 71
Figure 4.26: Bunch Block Configuration.............................................................. 72
Figure 4.27: Graphical User Interface test setup .................................................. 73
Figure 4.28 Test Cells ........................................................................................... 74
Figure 4.29 Break Wire Placement ....................................................................... 75
Figure 4.30 Detonators Awaiting Test .................................................................. 75
Figure 4.31: Electronics Break Out Box ............................................................... 76
Figure 4.32 Frame grabs from non-electric sample showing detonation sequence
............................................................................................................................... 77
Figure 4.33 Normal distribution, density function, non-electric detonators tested.
............................................................................................................................... 78
Figure 4.34 Normal distribution, density function, non-electric detonators at 9, 25
and 100 ms nominal delays. .................................................................................. 79
Figure 4.35 Normal distribution, density function for nominal delays, non-electric
detonators at 700, 1000 and 1400 ms.................................................................... 79
Figure 4.36: Frame grabs from electronic sample showing detonation sequence 80
Figure 4.37 Normal distribution, density function, electronic detonators tested. . 81
Figure 4.38 Normal distribution, density function, electronic detonators at 10ms
nominal delays. ..................................................................................................... 82
Figure 4.39 Normal distribution, density function, electronic detonators at 1000ms
nominal delays. ..................................................................................................... 82
Figure 4.40 Normal distribution, density function, electronic detonators at 8000ms
nominal delays. ..................................................................................................... 83
xi

Figure 4.41 comparison 9ms and 10ms nominal delay (non-electric Vs electronic)
............................................................................................................................... 83
Figure 4.42 Pair impulse signature waveform ................................................... 85
Figure 4.43 Three holes numerical example ......................................................... 85
Figure 4.44 Result of three holes detonation mathematical convolution .............. 86
Figure 4.45 Implementation of convolution in VB ............................................ 87
Figure 4.46 Convolution in Matlab .................................................................... 87
Figure 4.47 Convolution using Matlab ................................................................. 87
Figure 4.48 Convolution results using matlab and a more discretized signal. ..... 88
Figure 4.49 Shifted signals to perform the sum. .................................................. 88
Figure 4.50 Vibration record for test blast 4 holes detonated at the same time ... 90
Figure 4.51 Plan view from the blast report for the test ....................................... 90
Figure 4.52. Radial signature (seed) waveform .................................................... 91
Figure 4.53 Matlab command to load signature and timing vectors ..................... 91
Figure 4.54 Convolution for radial component ..................................................... 92
Figure 4.55 Graphical procedure to calculate vibration waveform in test ............ 92
Figure 4.56 Final waveform after sum 17 signals using graphical procedure ..... 93
Figure 4.57 Final waveform simulation using commercial software Apha-Blast
............................................................................................................................... 93
Figure 4.58 Complete waveform comparison using different methodologies. ..... 94
Figure 5.1 Location of the mine where the field experiments were conducted .... 95
Figure 5.2 Drill Hole GY 9411, stratigraphic column .......................................... 96
Figure 5.3 Case for the seismograph setup ........................................................... 97
Figure 5.4 Seismographs location ......................................................................... 99
Figure 5.5 Seismographs location for second round of test. (Summer 2011) ..... 101
Figure 5.6 NOMIS 5400 System used to collect blast vibrations and airblast 102
Figure 5.7 Plan layout test 06/24/2011 ............................................................... 104
Figure 5.8 Vibration record for test No.12, seismograph 4906 (approx. 767ft from
source) ................................................................................................................. 104
Figure 5.9 Radial vibration component for test No.12 and all the seismographs in
summer 2011 ....................................................................................................... 105
Figure 5.10 Airblast records for test No.12 and all the seismographs in summer
2011..................................................................................................................... 106
Figure 5.11 Airblast record for Test No.5 (Four holes signature) ...................... 107
Figure 5.12 Plan view from the blast report for the test No.5 ............................ 108
Figure 5.13. Airblast signature waveform test No.5 ........................................... 108
Figure 5.14 Airblast signature waveform and synthetic waveform (approach). . 109
Figure 5.15 Frequency interval and scale factor for test No.5 ............................ 110
Figure 5.16 Random airblast signature waveform signals test No.5 ................. 110
Figure 5.17 Delay time series for test No.5 ........................................................ 111
Figure 5.18 Time function for test No.5. Zero and 1ms standard deviation in
detonator ............................................................................................................. 111
Figure 5.19 Results using one iteration and zero delay standard deviation ........ 112
Figure 5.20 Variables involved in the prediction test No.5 using improved
signature technique. ............................................................................................ 112
Figure 5.21 Monte Carlo result using initial parameters for test No.5 ............... 113
xii

Figure 5.22 Monte Carlo results using 3Hz frequency as main frequency in test
No.5 ..................................................................................................................... 114
Figure 5.23 Histogram absolute maximum calculated values ............................ 115
Figure 5.24 Convergence plot for test No.5 ........................................................ 115
Figure 5.25 Test No.10, plan view from survey ................................................. 116
Figure 5.26 Vibration record seismograph 3599 test No.10 ............................... 116
Figure 5.27. Ground vibration signature waveform test No.10 .......................... 117
Figure 5.28 Signature waveform and synthetic waveform (approach) test No.10.
............................................................................................................................. 118
Figure 5.29 Frequency interval and scale factor for test No.10 .......................... 118
Figure 5.30 Random ground vibration signature waveform signals test No.10 119
Figure 5.31 Time series for test No.10 ............................................................... 119
Figure 5.32 Time function for test No.10. Including both timing parameters ... 120
Figure 5.33 Results using one iteration for test No.10 ....................................... 120
Figure 5.34 Variables involved in the prediction test No.10 using improved
signature technique. ............................................................................................ 121
Figure 5.35 Monte Carlo result using fixed parameters for test No.10 .............. 122
Figure 5.36 Histogram absolute maximum calculated values test No.10 ........... 122
Figure 5.37 Convergence plot for test No.10 ...................................................... 123
Figure 5.38 Test No.12, plan view from survey ................................................. 124
Figure 5.39 Vibration record seismograph 3599 test No.12 ............................... 124
Figure 5.40. Ground vibration signature waveform test No.12 .......................... 125
Figure 5.41 Signature waveform and synthetic waveform (approach) test No.12.
............................................................................................................................. 126
Figure 5.42 Frequency interval and scale factor for test No.12 .......................... 126
Figure 5.43 Random ground vibration signature waveform signals test No.12 127
Figure 5.44 Time series for test No.12 ............................................................... 127
Figure 5.45 Time function for test No.12. Including both timing parameters ... 128
Figure 5.46 Results using one iteration for test No.12 ....................................... 128
Figure 5.47 Variables involved in the prediction test No.12 using improved
signature technique. ............................................................................................ 129
Figure 5.48 Monte Carlo result using fixed parameters for test No.12 .............. 130
Figure 5.49 Histogram absolute maximum calculated values test No.12 ........... 130
Figure 5.50 Convergence plot for test No.10 ...................................................... 131
Figure 5.51 Test No.14, plan view from survey ................................................. 132
Figure 5.52 Vibration record seismograph 3599 test No.14 ............................... 132
Figure 5.53 Signature waveform for test No.14 ................................................. 133
Figure 5.54. Ground vibration signature waveform test No.14 .......................... 133
Figure 5.55 Signature waveform and synthetic waveform (approach) test No.14.
............................................................................................................................. 134
Figure 5.56 Frequency interval and scale factor for test No.14 .......................... 135
Figure 5.57 Random ground vibration signature waveform signals test No.14 135
Figure 5.58 Time series for test No.14 ............................................................... 136
Figure 5.59 Time function for test No.14. Including both timing parameters ... 136
Figure 5.60 Results using one iteration for test No.14 ....................................... 137
xiii

Figure 5.61 Variables involved in the prediction test No.14 using improved
signature technique. ............................................................................................ 137
Figure 5.62 Monte Carlo result using fixed parameters for test No.14 .............. 138
Figure 5.63 Histogram and convergence plot of the prediction for test No.14 .. 138
Figure 5.64 Test No.15, plan view from survey ................................................. 139
Figure 5.65 Vibration record seismograph 3599 test No.15 ............................... 140
Figure 5.66 Results using one iteration for test No.15 ....................................... 141
Figure 5.67 Monte Carlo result using fixed parameters for test No.15 .............. 141
Figure 5.68 Histogram and convergence plot of the prediction for test No.14 .. 142
Figure 6.1 Signature signal test No.5 using five frequencies ............................ 143
Figure 6.2 Random airblast signature waveform signals test No.5 using 25
frequencies. ......................................................................................................... 144
Figure 6.3 Final prediction for airblast using 25 frequencies and amplification
factor of 2.2 for main frequency (3Hz). .............................................................. 145
Figure 6.4 Results comparison using four and twenty five frequencies for model
02, test No.10. ..................................................................................................... 146
Figure 6.5 Results comparison using four and twenty five frequencies for model
03, test No.12. ..................................................................................................... 147
Figure 6.6 Results comparison using four and twenty five frequencies for model
04, test No.14. ..................................................................................................... 148
Figure 6.7 Results comparison using four and twenty five frequencies for model
05, test No.15. ..................................................................................................... 149
Figure 6.8 Synthetic signal for test No.14 before apply decay factor ................. 150
Figure 6.9 Accuracy and precision electronic vs non-electric system ................ 151
Figure 6.10 Waveform comparison test No.14 vs test No.15 transversal
component (seismograph 3599) .......................................................................... 152
Figure 6.11 Frequency domain comparison test No14 vs test No.15 (seismograph
3599) ................................................................................................................... 153
Figure 6.12 Test No.15 simulating both initiation systems and including only time
delay due to initiation sequence. ......................................................................... 154
Figure 6.13 Signature used to model Tests No.14 and 15 (average) in seismograph
4906 location. ...................................................................................................... 156
Figure 6.14 Modeling results test No.14 seismograph 4906 (Pyrotechnic) ........ 158
Figure 6.15 Modeling results test No.15 seismograph 4906 (Electronic) .......... 158










1

Chapter 1
INTRODUCTION
1.1 Background
Vibrations as a result of blasting practices in mining engineering are complex
phenomena controlled by many variables. Mining blast vibration modeling and
prediction are becoming more important as a consequence of the general negative
perception of the public to mining activity. Many communities (all over the world) that
are close to mines complain about airblast and vibrations as a consequence of mine
blasting activity. It is important for the mining industry to have tools and understandable
methodologies to model and predict blast vibration.

There are several approaches to model and assess blasting vibration levels. They
range from the scaled distance methodologies to very complex and elaborate numerical
models. The scaled distance methodologies used a power law relationship between the
peak particle velocity (PPV) and the scaled distance to the specific point under study.
These methodologies demand statistical estimations of the main parameters involved in
the problem, so they are reliable after a good database of events is available.

Numerical models range from models that use the physical properties of the rock
and elastic solutions to simulate the propagation waves in a continuous medium to more
elaborate proposals. According to the computational developments, numerical models
for continuous medium are capable of including a large number of variables in the
simulation, but are not able to reproduce exactly the geology where the blasting takes
place and how the vibration propagates. As a consequence, the results of numerical
simulations of blasting vibrations are just models of the real phenomena. More advanced
numerical approaches are related to the analysis of block systems (Mortazavi A.,
Katsabanis P.D., 2001). In such models, systems of simultaneous equations are
formulated and solved minimizing the energy of the system to bring the system into
equilibrium.

Usually, the blast phenomenon that occurs in the hole (the explosion) is simulated
using cavity expansion and propagation crack theories around the blast hole. The
numerical techniques used to solve the wave propagation depend on the assumptions
regarding the medium. If the medium is considered as a continuum where the influence
of the rock mass joint system is neglected or simulated as interfaces, the most common
methods to use are finite element models (FEM), and boundary element methods (BEM).
On the other hand, if the presence of the discontinuities is taken into account and large
displacements and rotating blocks are allowed in the discontinuities, methods as distinct
element method (DEM), discontinuous deformation analysis (DDA) or bounded particle
method (BPM) are used (A. Bobet et al., 2009).

DEM is one of the most successful techniques used to simulate the media
subjected to vibrations (rock mass) as a discontinuous medium composed by an
assemblage of discrete blocks. The internal discontinuities are treated as boundary
2

conditions between blocks; large displacements along such discontinuities and rotations
are allowed, the individual blocks are modeled as rigid or deformable solids in
accordance with the properties of the rock and rock mass to be modeled (Wang et al.,
2009). These methodologies require significant computational time, parameters or
physical properties for the media to simulate that, in some cases, are difficult to evaluate
by field or laboratory testing, and highly-trained personnel capable of modeling and
interpreting the results. Due to this, modeling processes using these types of techniques
sometimes are not practical or become very expensive.

A technique based on the use of seismic analysis, waveform interpretations and
signal and systems theories has become used more often in mining applications. Using
this methodology, it is possible to model the complete waveform of a mining production
blast for a critical site where vibration monitoring or control is needed. In this approach,
one seismic signal from one blasthole is recorded in the site under study and using signal
and systems procedures that involve mathematical convolution of signals, the waveforms
from each blasthole can be superposed to model the complete vibration waveform for the
production blast. One of the outputs of this technique is the initiation timing between
charges to reach given vibrations levels (usually the minimum vibration level for a
production blast). Due to that reason, waveform superposition technique is becoming
increasingly popular with the use of electronic detonators since their use provides better
control over the initiation timing pattern (in other words less scatter initiation timing).
Improvements in fragmentation, vibration levels and air blast have been reported in the
literature through the use of both electronic detonators and waveform superposition
(Chistopherson, and Papillon, 2008).

The areas of study of this research were:
Background behind the vibration modeling using waveform superposition;
Waveform superposition methodologies;
Analysis of the major assumptions in the waveform superposition
methodologies;
New approach to improve these methodologies in order to obtain more
reliable and accurate results when vibration levels are assessed.

1.2 Fundamentals of blast vibration
In the literature, there are different approaches to explain and model the physical
process when a buried charge of explosives is detonated. In general, if a classification
according to the strain-stress behavior of the blasted material is made, it is possible to
distinguish two different zones, i.e., the inelastic and elastic zones. In the inelastic zone,
the energy contained in the explosive is released through a chemical reaction. In the
inelastic zone, tremendous pressure and high temperatures are developed due to the
chemical reaction. As result, the solid medium is subject to inelastic phenomena such as
breaking, shearing and crushing of the rock mass. Also, large strains are developed
within this zone (Enescu et al., 1973; Bollinger, 1980; Saharan et al., 2006). A
fundamental discussion of the inelastic process within the inelastic zone in an explosion
can be found in Cook (1958) or in Langefors and Kihlstrom (1963). At some distance
3

from the explosive reaction, the behavior is more elastic. In this zone, the disturbance
due to the energy released during the explosion propagates as seismic waves. The
behavior in this zone is considered as elastic because it is commonly assumed that the
solid medium returns to its initial state after passage of the seismic disturbance.

The seismic waves propagating through the earth media can be divided in two
major categories: body waves and surface waves. Body waves propagate through the
solid medium (soil or rock) and surface waves travel along the surface. The main surface
wave is the Rayleigh wave denoted by R-wave. Body waves can be subdivided into
compressive waves (P-waves) and shear waves (S-waves).

Explosions produce mainly body waves (P and S) at small distances while R-
waves become important at larger transmission distances (Dowding, 1985). The
waveforms can be idealized for far and close distances according to the location of the
recording site. The two idealized waveforms are explained using Figure 1.1

If the strain, pressure, or particle velocity (PV) is measured at Point A (close-in
explosion), the shape of the idealized wave will be a single-spiked pulse. This is because
at Point A, only direct-transmission of the waves generated by the explosion is measured.
On the other hand, if the recording site is located at Point B (far explosion), the idealized
waveform of the strain, pressure or particle velocity will be more like a sinusoidal shape.
At Point B the sinusoidal waveform will be a combination of direct-transmission,
reflection and refraction waves (Silva et al., 2011).



Figure 1.1 Waveform idealization in a blast event (Adapted from Dowding,
1985).
In the blast vibration phenomena, there are several factors involved in the process.
These factors are explained briefly in the following sections.
Explosion
A
B
D
D
D
R
R
GZ STZ
time time
PV
or Strain
ARecording site A
D Direct transmission
R Reflection and Refraction
transmission
Idealized waveform at A Idealized waveform at B
PV
or Strain
4

1.2.1 Characteristics of the explosive
It has been documented that the type of explosive used in mine blasting influences
the blast vibration (Hossain and Sen, 2004; Hunter et al., 1993; Harries and Gribble,
1993). There are two broad categories of explosives according to the type of detonation
generated, non-ideal and ideal detonations. Non-ideal detonation occurs when the rise
time for the peak blast hole is longer and the post peak pressure drop is much slower
when compared to ideal detonation. On the other hand, in the ideal detonations, the peak
pressure rise time is very short and the post peak pressure drop is steep (Saharan, M.R.,
Mitri, H.S. 2008). In their research, Hunter et al., (1993) reported that explosives with
lower density and lower detonation velocity (non-ideal detonation) produced lower
ground vibration levels when compared to ideal detonation.
1.2.2 Initiation system used
Safety fuses were developed to improve the safety in the blasting operations.
These devices introduced a delay before the detonations occurred, which gave the blaster
enough time to move away from the blast site. After the introduction of the safety fuse,
the electric detonator was introduced at the beginning of the 20th century to increase
performance in blasting. Electric means that there is a bridge wire in the detonator that
matches with the initiation system (electricity). In the 1960s and 1970s, Dyno Nobel
introduced the non-electric (nonel) detonator. In the nonel detonator, as well in the
electric detonator, the time delay is given by the length of the pyrotechnic element;
therefore, varying the length of this element varies the time delays. Due to the chemical
nature of the delay element (pyrotechnic), the accuracy is relatively low when compared
to electronic and high compared to safety fuse. Through the 21
st
century, scatter in the
detonation time has decreased. Since 1950 to 80s, improvements in the pyrotechnic
elements have increased the accuracy to levels of 1.5% to 2.5 % of the total time delay
(Larsson et al., 1988). Despite this effort to reduce the scattering of time delays, the
pyrotechnic delay still gives low accuracy in some situations. With longer delays, the
delay inaccuracy can potentially cause overlap and holes firing out of sequence affecting
the vibrations and performance of the blast (Lusk, et al., 2012).
Shock tube or non-electric detonators (nonel), electric and electronic detonators
are the major delay systems currently used. The basic differences between the three
types of detonators are the type of delay element used to produce the time delay and the
igniter. Figure 1.2 adapted from Miller and Drew (2007) shows the major differences
between the three systems.











Figure 1.2 Types of Detonators (After Miller, D., Drew, M. 2007)
5

In a traditional pyrotechnic detonator (nonel or electric), the delay element is
composed of a pyrotechnic device (chemical delay) while, in electronic, the time delay is
controlled by an electronic circuit and a bridge wire.

1.2.3 Shot Geometry and Timing
The shot geometry take into account six variables involved in the problem:
diameter of the hole, burden, spacing, length, stemming and sub-drilling (Ash, 1973).
The relationship between these variables determines the performance of the explosive
forces and whether or not the mining blast behaves as expected. From the blast vibration
point of view, the shot geometry affects the degree of confinement of the charges which
affects the level of the seismic waves generated in the mine blasting process. It is
commonly assumed that high degree of confinement will generate higher vibration levels
than with a one less confined blast pattern. Regarding timing the use of electronic
initiation systems, the 8 ms rule to control blast vibration levels is under analysis (Reiz, et
al., 2006). Current researchers are trying to explore the advantages of short timing delay
(under 8ms) to diminish the total time of exposure under blast vibration that a structure
under control is subject when a long timing delay is used in a mining blast.

1.2.4 Distance source monitoring point
The distance from the blast pattern to the monitoring point is one of the most
important parameters in blast vibration. As vibration waves travel away from the source,
they spread out meaning longer duration and lower frequency. Additionally, some of the
energy is absorbed by the materials they travel through. As consequences of the
spreading out of the waves with the distance, the waves change their characteristics,
including their amplitude and frequency. The spreading of energy with distance is related
to the type of waves. In the case of body waves (p and s waves), the amplitude decreases
according to the relation (1R) where R is the distance source-measurement point. For
surface waves (Rayleigh waves) the decrement relation is (1R
.5
). Additionally as
consequence of the energy absorption of the materials, the ground motion amplitude
decrease exponentially with R (Kramer, S.L., 1996).

1.2.5 Geology
The geologic discontinuities and joints act as boundaries in the medium. The
waves are reflected and or refracted every time they hit a boundary. Also, the lithology
introduces changes (amplitude and frequency changes) in the waves because the different
dynamic properties of the materials the waves travel through. Geology creates new type
of waves when original P and S waves hit an inclined boundary producing, for example,
reflected and refracted P-and SV waves. The path to reach one specific monitoring
point that the vibration energy follows coming from each hole in a blasting pattern
change according to the geological features the waves hit while they are traveling
towards the monitoring point. Figure 1.3 shows the main factors affecting ground
vibrations due to mining blasting for a single shot.


6










Figure 1.3 Factors affecting ground vibration (after Khandelwal and Singh,. 2006)

7

Chapter 2
BLAST VIBRATION
2.1 Blast vibration characteristics
Ground vibrations from blasting are the result of energy release from chemical
explosives during an explosion in a blast hole. In the detonation process, the solid mass
of explosive is converted nearly instantaneously into gaseous products. The change in
the pressure of the gas occurring during the explosion generates a rapid change in the
initial stress state of the medium (rock) which crushes the rock near the hole and
displaces the rock in to a muck of pile (Saharan et al., 2006). Beyond the hole at some
distance related to the initial hole diameter, the stress deforms the rocks elastically and
part of the energy released during detonation travels as a stress wave or a seismic wave
through the medium generating vibrations (Sally and Daemen, 1983).
Seismic waves can travel considerable distances (Frantti. G.E., 1963). In a
production blast, several holes are detonated at varying times generating different pulses
and seismic waves. The interaction of the seismic waves in a constructive or destructive
manner produces a complex vibration pattern which is recorded at a specific site. The
vibration waveform recorded in a blast event usually consists of three orthogonal
components: radial, longitudinal and vertical. Generally, the radial and longitudinal
components are in a horizontal plane while vertical component is perpendicular to the
other two. This arrangement is due to the construction of the geophone.
Most vibrations from surface mine blasting have a frequency content less than
200 Hz while underground mining blasts tend to produce much higher frequencies
(Spathis, 2010). Particle velocity is commonly used to measure or establish the permitted
levels of ground vibrations. This parameter was chosen over displacement or
acceleration from the research of Langefors, et al., (1958), Edwards and Northwood
(1960) and Duvall and Fogelson (1961) among others. In these studies, the particle
velocity was established as the best criterion to assess the structural damage due to blast
vibrations. Most current regulations stipulate particle velocity to establish maximum
vibration levels from a mining blast; however, to choose which vibration parameter to
measure (displacement, velocity or acceleration) it is necessary to assess the frequency
content of the vibration (Dowding, 1985). Table 2-1 contains the range of the main
parameters of blast vibration.

Table 2-1 Range of typical blast parameters (after Dowding, 1985)








8

In general, the waveform of ground vibrations from blasting has a lower particle
velocity and higher dominant frequencies comparatively to the waveform of ground
vibrations from earthquakes (Siskind 1993). The total duration of both types of events
(earthquake and blast) in a typical production blast differ considerably. Figure 2.1 shows
a typical earthquake wave form recorded by the Kentucky Seismic Network and a blast
vibration signal from a production blast. Figure 2.1a shows the comparison in time
domain between a signal recorded form an earthquake versus a mining blast. Earthquake
record is bigger and longer than blasting record. Figure 2.1b shows how the main
frequency is higher when compared to blast vibration signal.

















a) Time series comparison (Blast Event Earthquake)
















b) Fourier Transform comparison (Blast Event Earthquake)
Figure 2.1 Typical earthquake and blast waveforms
9

After the vibration guidelines published by the United States Department of
Interiors Bureau of Mines (USBM), (Siskind et al., 1980), most of the blast vibration
regulations are based on particle velocity and frequency content. The USBM study was
based on damage due to blast vibrations of houses ranging from modern houses with
drywall interiors to older houses with plaster or wood lath interiors. The damage was
categorized as threshold, minor and major damage. In this study, around 200 blast where
included.
Graphically, USBM blast vibration recommended regulation is expressed using
the so-called Z curve presented in Figure 2.2.



















Figure 2.2 Blast vibration recommended regulation Z curve. Safe level blasting
criteria from USBM RI 8507 (After Siskind, 2000)
In Figure 2.2, dashed lines define USBM recommended safe limits (Z curve).
Symbols shown are positive damage observations in the houses under study.

2.2 Blast vibration measurement
The objective of the measurement of blast vibration is to describe the behavior at
a point in the ground due to the vibratory motion produced by a mining blast. From the
physics point of view, it is possible to measure displacement, velocity or acceleration.
However, as explained previously, the research of Duvall and Fogelson (1961) found that
particle velocity is better to use to describe and control vibrations levels.
To describe completely the dynamic behavior of a point in the ground due to the
mining blast, it is necessary to measure three orthogonal components of any of the
physical quantities given by the displacement, velocity or acceleration as a function of
time (t). The relationships of these three physical quantities are given by:
10


o =
J:
Jt
=
J
2
x
Jt
2

[2.1]
Where:
o: acceleration
:: velocity
x: displacement
t: time

In principle, the measurement of one of the physical quantities allows the
determination of the other two through integration or derivation.

The devices to measure blast vibrations are mainly composed of four components;
a transducer, a recorder, a timing system, and a storage system. There are many types of
transducers. One of the most basic transducers is composed by a magnet suspended
inside a coil. When the magnet experiences a movement, the relative movement between
the magnet and the coil generates a current. The output current is proportional to the
movement of the magnet and in turn the movement of the magnet is proportional to the
displacement, particle velocity or acceleration in the ground. Most of the devices
currently in operation use electronic transducers composed by piezoelectric materials.
Those materials when subject to transient forces generate electrical currents proportional
to accelerations, velocity or displacements. This current is calibrated to a specific range
of motion. Regardless of the type of transducer, the controlling factor in the blast
vibration measurement is the natural frequency of the instrument and the ground
vibration frequency to measure. (Bollinger, 1980).
Initial ground motion instruments record the motion in an analog form on paper or
photographic film (Kramer 1996). Due to the computational development in technology,
analog signals are transformed into digital signals. Analog signals are also called
continuous-time signals (CT). By definition, a CT signal is defined at every time instant
in a time interval of interest and its amplitude can assume any value in a continuous
range. On the other hand, a digital signal or discrete-time signal (DT) is defined only at
discrete time instants, and its amplitude can assume any value in a continuous range
(Cheng, C.T., 2004).
Many devices commonly used in mining to measure blast vibrations use a sample
rate of 1024 samples per second (although many are capable of 2048 samples per
second), for a total recording time of 12 seconds. That means that the sampling period
(T) is given by:

I =
sompling timc 1 somplc
numbcr o somplcs in sompling timc
=
1 s 1somplcs
1u24 somplcs
= u.uuu976Ss
[2.2]
Many of the records used to analyze and predict blast vibrations using waveform
superposition have a precision close to one millisecond (0.97 ms). In other words, the
11

time step between two consecutive values in the signal is one millisecond. If the time
step of one millisecond in the signal is compared against the fact that when electronic
initiation systems are used, the standard deviation in the timing of the initiation system
for short periods is sometimes below one millisecond (Lusk, B., et.al 2012), a
disadvantage arises when waveform superposition is performed using the conventional
time sampling given by the devices used commonly to monitoring mine blast vibrations.
2.3 Blast vibration prediction
There are, in the literature, many approaches to assess the vibration levels from a
mining blast. Those methods can be categorized into five approaches (Spathis, 2010)
historical data review;
charge weight scaling laws;
waveform superposition;
scaled charge weight superposition;
analytical and/or numerical methods.
In this document, only the most common methodologies are included giving a
brief discussion about each approach.
2.3.1 Scaled distance estimation
Scaled distance approaches have been used in connection to small-scale modeling
of nuclear explosions (Dowding, 1985). In this research, commonly the empirical
expressions for particle velocity due to blast deducted from ground shock data have the
general form given by (Drake and Little, 1983):
I
o
= A B
-n
_
R
w
1
3
,
_
-n

[2.3]
where:
I
o
: peak particle velocity
: coupling factor for near surface detonations
R: distance to the explosion
A, B: constants
n: attenuation coefficient
w: explosive charge mass. It is proportional to the energy release
during detonation
Those equations are derived from the Buckingham Pi theory of dimensionless
analysis (Buckingham, 1915). The Pi theorem states that any of the parameters may be
considered to be a function of another, and that the parameters may be raised to any
power. In the case of the particle velocity, the dimensionless group is given by
(Ambraseys and Hendron, 1968):

I
c
= g
2
_
tc
R
,
w
pc
2
R
3
_
[2.4]
12

Where:
I: particle velocity |II
-1
]
c: seismic velocity of rock mass |II
-1
]
t: time |I]
R: distance from explosion (range) |I]
w: energy released by explosion |FI]
p: density of the rock mass |FI
2
I
-4
]
g
2
: unknown function of the dimensionless group

Assuming that the density p and the seismic velocity of the media c are relatively
constants when compared to the distance to the explosion R and the energy released by
the explosion w, they are sometimes dropped from the dimensionless terms. Since the
parameters may also be raised to any power, I can be plotted against (w)
13
R or
R(w)
13
, those parameters produce consistent relationships according to Ambraseys
and Hendron (1968) and Dowding (1985) and others (Villano, Charlie, 1993).

It was in Bulletin 656 of the Bureau of Mines (Nicholls, et al., 1971) data analysis
from 171 blasts at 26 quarries where the form of the equation given by Equation 2.5 was
proposed:
I

= E

_

w
u
]
[
i

[2.5]
where:
I

: particle velocity in i direction

E

: particle velocity in i direction intercept

: shot to gage distance
w: charge weight
o: exponent
[

: slope or decay factor in i direction

i: denotes component, radial, vertical, or transverse
In the case of the Equation 2.5, it was determined that using a value of o =
1
2
, for
the data of such project that when a log-log plot is made between particle velocity I and
w
1
2
a significant reduction in the spread of the data was achieved.

If Equation 2.5 is written in the traditional form, using square root of the charge
instead of cube root of the energy released by explosion, the particle velocity is expressed
by:
I

= o (S)
-b

[2.6]
where:
o and b: adjustable parameters, dependent upon the local ground conditions
S: scaled distance defined by:
13

S =
R
w
2

[2.7]
where:
R: distance from explosion
w: charge weight

Although Equation 2.7 is one of the equations most used in mining engineering to
estimate and control vibrations from blasting, this equation has fundamental problems
(Blair, 2004). The main problem about Equation 2.7 is regarding to units; if we assume
that the units in Equation 2.7 are given by:
R: length [L]
w: mass [M]
Then, the units of scaled distance should be:
S: units of [LM
-1/2
]
That means that the constant o in Equation 2.6 should have dimension or units;
for example if b=2, the units of o will be:
o: units of [M
-1
L
3
T
-1
]
However if b=1 then:
o: units of [M
-1/2
L
2
T
-1
]
If we consider that b is both, site and direction specific, then o units change
continually and Equation 2.6 is not a fundamental equation for vibration prediction
(Blair, 2004)
To predict vibrations, using scaled distance laws, it is necessary to collect
information of vibration levels from a set of standard blast events that represent the
conditions of blasting of the mine in the future. The main assumption besides the
applicability of the scale distance law is that the future blasts are going to produce equal
or similar vibration levels than the standard blasts. Figure 2.3 shows an example of data
used to produce scaling law in different vibration components (Lusk, et al., 2010).
















14







































Figure 2.3 Scaled distance law, analysis for different components (Lusk, Silva,
Eltschlager, Hoffman, 2010)
2.3.2 Analytical and numerical approaches
Currently most of the efforts to model mining blasts are more focused on the
modeling of the fracture and fragmentation process than blast vibration prediction. The
majority of numerical modeling research is focused in the study of the fragmentation
process and the evaluation of the vibration levels is just some additional information. A
15

P
h
y
s
i
c
a
l

v
a
r
i
a
b
l
e
s
P
r
o
b
l
e
m
M
a
t
h
e
m
a
t
i
c
a
l

V
a
r
i
a
b
l
e
s
Rockfracturingmodeling
Material
(Rock)
Explosive
Boundary
conditions
Geology Properties Geometry Chemistry
Distance,
size
FiniteElements
Boundaryelements
Discreteelements
Infinite elements
Distinticelements
Linear(Elastic)
Viscoelastic
Perfectlyplastic
Hardeming
Softening
Others
Waveenergy
Gaspressure
energy
Thermalenergy
Ideal(non)
detonation
Detonation
coupling
Interhole
Interaction
Resonance
summary of the main assumptions, developments and considerations when blasting
numerical modeling is performed are given next.
The components in the modeling of fracturing process can be grouped in three
branches (Saharan, et al., 2006):
Materials (Rock);
Explosive;
Boundary conditions.
From each branch there are more ramifications creating more sub-systems. Each
branch is physically independent but they interact with each other to explain the
fracturing process. Figure 2.4 adapted from (Saharan, et al., 2006) shows the main
components to explain the fracturing process.


















Figure 2.4 Main components of rock fracturing modeling process (Adapted from
Saharan, Mitri and Jethwa, 2006)
The most common simplification in the fracturing model using analytical or
numerical approaches is to assume elastic properties of the surrounding rock to the
charged hole and a cylindrical geometry for the blast hole (Spathis, 2010). Other authors
(Batzle et al., 1980, Blair and Cook, 1998, Kranz, 1983) suggest the tensile failure mode
as the basic failure mechanism in rock. Under this failure mechanism, fractures
generated radially from the blast hole grow by taking the path of least resistance, i.e.
either that of least shear stress of the rock (usually the tensile strength) or the least
confining stress (Saharan and Mitri, 2008). When detonation pressure, exceeding the
tensile strength near the blast hole perimeter is overwhelming, and a crushing zone is
developed. Beyond the crushing zone, blasting results in the formation of a discrete
fracture networks. It is also common to assume the behavior of the material having a
viscoelastic law ignoring the non-linear behavior that the blasting process implies.
Regarding the dynamic load, there are two different pressure pulse shapes. The
form of the shape is related to the time the chemical compound reaches the peak pressure.
16

Those pulses explain the ideal and non-ideal detonation. In the ideal detonation, the rise
in the peak pressure is reached in a very short time while the rise time for the peak in the
non-ideal detonation is longer and the post peak pressure drop is much slower when
compared to ideal detonation. Figure 2.5 show the two most used pressure pulses in
fragmentation and blast vibration analytical and numerical modeling.












Figure 2.5 pressure pulse shape A. Ideal detonation. B. Non-ideal detonation
(Aimone, 1992., Olsson et al., 2001)
Usually in the modeling process, it is assumed that the cylindrical hole is
pressurized simultaneously over a section of its length. However, in production
blastholes, this process may take several milliseconds.
Several analytical solution assuming elasticity and isotropy of the material can be
found in the literature. In 1952 Blake derived the solution for the governing equation of
the problem of the propagation of a spherical wave due to an impulsive pressure; the
governing equation is given by:
o
2

ot
2
= C
p
2
v
2

[2.8]
Where:
C
p
: compressional wave velocity
t: time
: a potential function
v
2
: Laplacian operator
If a pressure function is defined (pressure pulse shape) as:

p(t) = p
o
c
-ut
for t u
p(t) = u for t < u
[2.9]
Using o = u, the radial displacement at large distances (r o) can be found as:

17

u

=
o
or
= -
p
o
o
3
K
pC
p
2
r
2
_-1 +2 -20 c
-u
c
:
cos _
o
-ton
-1
_
1
4K -1
_]_
+
p
o
o
3
K
pC
p
2
r
_
o
o
C
p
+2 -20 c
-u
c
:
cos _
o
-ton
-1
_
1
4K -1
_]_
+ _

o
C
p
+2 -20 c
-u
c
:
sin _
o
-ton
-1
_
1
4K -1
_]_

[2.10]
Where:
o: radius of the sphere
K:
1-0
2(1-20)
Bulk module
0: Poissons ratio
r: radial coordinate
o
o
:
C
p
2uK
radiation damping constant
: t -
-u
C
p
time lag

o
:
C
p
2uK
4K -1 natural frequency

To calculate the particle velocity it is needed to differentiate Equation 2.10.

u

=
o
2

orot
=
p
o
o
pC
p
r
|2 -20 c
-u
c
:
cos(
o
+ton
-1
|1 -20|)]
[2.11]
This approach has the limitation that the source of the pressure is spherical and
purely compressional modes of radiation are analyzed in the problem. A recent effort to
develop a numerical model that represents the mechanism of rock blasting involving the
detonation, fracturing and movement process has been conducted by an international
collaboration project (The Hybrid Stress Model Project (HSBM)). The model proposed
is named Hybrid Stress Blasting Model (Furtney et al., 2009). In this model, a numerical
code simulating the non-ideal detonation process is coupled to a numerical model that
simulates the behavior of a 3D rock mass. The model uses a combination of discrete and
continuum numerical techniques to model the rock blasting behavior. In the near field,
close to the blast hole, the rock mass is represented as a continuum grid. Such continuum
grid is coupled to another continuum grid that represents the explosives and their
behavior while detonation, (volume expansion, pressure and axial flow). The gas
products are introduced also into the network of fractures, causing further expansion of
the fractures and heave of the resulting rock fragments. The rock in the intermediate and
far field is represented based on a lattice of nodes and springs. (Cundall, 2008).
Currently even using the most advanced computational technology, the run times
for large models makes the day-to-day application of those methodologies impractical.

18

2.3.3 Signature hole technique
This technique is based on signals and system theories. It is well known that in a
blast event, the vibration structure response is a function of the amplitude and frequency
content of the ground vibration signal reaching the structure (Siskin et al., 1980b). It was
in the 1980s when wave interference concept began to be introduced (Anderson, et al.,
1985) and (Crenwelge et al., 1986) to control blast vibrations. Past researches had shown
the benefits of the use of wave interference to reduce the ground vibration levels in a
blast event (Lusk et al., 2006), (Christopherson and Papillon 2008), (Chiappetta, et al.,
1985).
The basic concept behind the signature hole technique is similar to the principles
applied in the signals and systems theories. In that branch of knowledge, a system is
defined as an entity with a unique relationship between the excitation or input and the
response or output (cause and effect) Figure 2.6 shows this similarity.
There are many types of systems. From an Input - Output point of view, they can
be:
SISO: Single Input Single Output
MIMO: Multiple Input Multiple Output
and his combinations:

MISO: Multiple Input Single Output
SIMO: Single Input Multiple Output



















Figure 2.6 Sketch of a system with continuous and discrete signals.
The systems can also be classified according to the characteristics in causal or
non-causal, lumped or distributed, linear or nonlinear and finally as time invariant or time
varying.
Single-Input, causal, linear and time invariant systems are very useful in the real
world because many physical phenomena can be modeled using the system theory
19

applicable to that type of systems. Figure 2.7 shows the basic concepts of the SI-SO, and
time invariant systems.
























Figure 2.7 Time Invariant systems single input - single output (Adapted from
Signal and Systems 3th edition)
Causality is related to the relationship between the Input- Output and the time.
One system is causal if the current output is only related to the current input (the current
response is not related to past or future inputs). On the other hand, linearity in the
systems theory is related to the linear superposition of different actions to produce a
response. Finally if the system does not change over time, this means that the system is
time invariant, i.e., an input in current time, produces the same output that an input given
to the system in the future.
All these concepts mean that if the system is Continuous (C) Linear (L) and Time
Invariant (TI), by knowing a pair Input Output signals, it is possible to predict the
outputs for whatever Input signal. Figure 2.8 shows this concept in more detail.








20
































Figure 2.8 SI-SO, C, L, TI systems (Adapted from Signal and Systems 3th
edition)
In blasting, the signature hole technique assumes that the vibrations generated as
energy release in a blast, and transformed into elastic waves travelling within the rock,
are a physical phenomenon developed in an SI-SO, C, L, TI system.
In such case, the system is the entity that wraps the site specific geological
conditions between the event site and the point under study (joints, faults, lithology etc.,),
and the path of the vibration waves, including reflections and refractions of waves
propagating away from the event site. Figure 2.9 shows this concept.






Known
21














Figure 2.9 Systems Theory and Signature Hole Technique similarity.

Other assumptions to the signature hole technique are (after Anderson 2008):
There is a need to control the vibrations in a specific location;
All holes are detonated at the same location, so that the path traveled by the waves
is identical;
All holes have the same explosive charge type and weight. In others word, the
quantity of energy converted into elastic waves each time a hole is blasted is the
same;
The phenomenon occurs in a system ideally SI-SO, C, L, and TI, so that all holes
have the same explosive-rock interaction. That means that the source pulse
(detonation) always generates the same response in the site under study (signature
wave).
In the signature hole technique, assuming that all the assumptions are fulfilled, the
signature wave recorded in a specific site (the signal recorded when a hole is blasted) can
be expressed as a finite impulse response (FIR). This means an impulse response with
finite nonzero entries, which can be expressed as:

b|n] = u or n = u,1,2, , N -1
[2.12]
with
b|N -1] = u
[2.13]
and
b|n] = u or n N
[2.14]



22

Graphically, it is represented in Figure 2.10. In this specific case, when the input
signal is the impulse, (one blasted hole at t
o
=0), it can be expressed as:

u|n] = o|n] or
[2.15]
u|n] = A

o|n]
[2.16]











Figure 2.10 Finite impulse response from one hole blasted
Now, if we consider an arbitrary input sequence (production blast hole), it can be
expressed as:
u|k] = A

o|t -m

t
o
]
[2.17]
and
u|n] = u|k]o|n -k]

k=0

[2.18]
Assuming the system is linear and time-invariant, and using the shifting,
homogeneity and additive properties of signals, the output y[n] excited by the input u[n],
for n0, can be given by:
y|n] = b|n -k]u|k]
n
k=0

[2.19]
Or in a general form:
y|n] = b|n -k]u|k]

k=-
=b|n] u|n]
[2.20]
This algebraic equation is called a discrete convolution. This equation relates
the input and output of a system. Due to this relation, the convolution is also sometimes
23

called input-output description of the system. In this case, the description of the system
(calculation of the output given an input) is developed without using any physical
properties of the system and is based on signal-system properties as linearity, time
invariance and causality.
In the signature hole technique, predicting the vibration levels of a production
blast on the same monitoring point that the signature wave was recorded is achieved by
using the recorded signature wave directly to calculate the blast vibration waveform of
the production blast. On the other hand, if there is no signature waveforms available in
the place where it is required to assess vibrations levels, some authors combine scaled
distance estimations and transfer functions to get the signature waveform at the point of
interest (Spathis, 2010).
In their methodology, Yang and Scovira (2010), use a set of signature waveform
tests to estimate the amplitude attenuation of the vibration due to the distance from the
source to the point of interest. Using the set of signature waveforms a graphical
representation of Equation 2.6 and Equation 2.7 for the site under study is obtained.
After calculating the maximum peak particle velocity for the point of interest according
to Equations Equation 2.6 and Equation 2.7, the closest signature waveform is modified
to match the peak particle velocity at the point of interest. It is well documented that not
only is there a change in the amplitude of the vibration according to the distance, also
when an acoustic pulse propagates, its frequency changes attenuating the higher
frequencies (Kavetsky, et al., 1990). In order to model the change in frequency, the
concept of an effective wavelength varying linearly with distance is introduced through
the use of Kjartansson transfer function (Kjartansson, 1979). This transfer function may
be used to propagate an arbitrary wave shape by Fourier transforming the convolution of
the impulse response and the source wave. The Fourier transform of the propagated
wave at the point of interest is given by:

w() = S()B()
[2.21]
Where:
w(): Fourier transform of the propagated wave at point of interest
S(): Fourier transform of the source wave (closest signature waveform
to the point of interest)
B(): Kjartansson transfer function

and the Kjartansson transfer function is given by:

B() = c
_-
xo
c
c
c
_
o
o
c
_
1-y
jtun[
ny
2
+sgn(o)[_

[2.22]
Where:
x: distance from source wave to point of interest

o
: frequency of reference
o
=
1
t
c
; t
o
: arbitrary reference time
c
o
: phase velocity at the arbitrary reference frequency
o

24

c
o
=
[
M
c

1
2
,
cos[
ny
2


H
o
: bulk modulus
p: density
y: shear strain y =
1
n
ton
-1
[
1

=
1
n

: constant creep function (Rock quality factor)
sgn() = 1 > u
sgn() = -1 < u

Using this approach, bulk modulus, density and rock quality factor of the medium
are required.
Finally, in their methodology, Yang and Scovira (2010), propose the waveform
change as a function of screening effect of earlier firing holes. The screening effect
assumes that there is a change in the amplitude and in the shape of the waveform due to
the change in the medium where the blasting process is occurring. In other words, a
previous hole blast affects the surrounding rock where the next blast hole takes place in
the vibration path. The function proposed to change the seed waveform is a ratio between
the quantity of explosive already blasted to the quantity of explosive that produce the
waveform and is given by
s() = z
q

[2.23]
where
=

t

[2.24]

t
total charge weight of earlier fired blast hole in the vibration path
charge weight of the presently firing charge, where <1

The limit conditions of this equation is given by
s() = _
1, u
u,

[2.25]

Figure 2.11 summarizes the current signature hole technique methodologies.









25





















Figure 2.11 Summary of the most common signature hole techniques currently
available (After Spathis, 2010)

The next chapter is a detailed discussion about the main assumptions in the
current signature hole technique.

26

Chapter 3
DETAILED DISCUSSION OF CURRENT SIGNATURE HOLE TECHNIQUE
3.1 Introduction
This chapter describes in detail the main aspects of the current signature hole
technique. The description includes the major assumptions of the current signature hole
technique as the seed waveform repeatability and the influence of the timing sequence in
the waveform produced. Explanations about the linear superposition are given in two
forms including the traditional one where convolution is used to calculate the total
waveform and how to do the linear superposition using graphical techniques. Finally,
results are included as comparison between the current prediction methodology and
waveforms as a result of production blasts.
3.2 Blast vibration energy
The energy stored in the chemical components of explosives (ANFO, dynamite,
Emulsions etc.) is released in a combined process of deflagration and detonation.
Sanchidrin Jos et al., (2006), proposed that the major components of the released
energy in a mining blasting process are composed by four parts: fragmentation, seismic,
kinetic, and energy used in other types of work during the process. The equation that
describes this behavior is given by:

E
L
= E
P
+E
S
+E
K
+E
NM

[3.1]
Where:
E
E
: Explosive energy in the chemical
E
F
: Fragmentation energy
E
S
: Seismic energy
E
K
: Kinetic energy
E
NM
: Not measurable energy

Not measurable energy includes the energy release as sound (airblast), heat, light
and other phenomena that occurs in the explosion.
Using the energy flux concept defined as the power or rate of work per unit area,
it is possible to relate the particle velocity of the ground to the dynamic stresses generated
when the wave passes through a specific point as:

= t

:
[3.2]
Where:
: Energy flux
t

: Stress vector
u: Particle velocity vector

27

Using stress tensors (Cauchy formula) and assuming that the energy transferred to
the rock can be evaluated as the integral of the energy flow through a control surface at a
given distance from the blast, it is possible to evaluate the seismic energy using the plane
wave approximation as (Sanchidrin Jos, et al., 2006):

E
Ss
= 4nr
2
_c
L
_ v
1
2
ut +c
T
_ (v
2
2
+v
3
2
)

0
ut

0
_
[3.3]
Where:
: rock density
c
L
and c
T
: Longitudinal and Transverse rock wave velocity
respectively.
r: distance to the source
v
1
, v
2
, v
3
: Particle velocities radial, longitudinal and transverse
respectively.

The signature hole technique assumes that each hole releases the same seismic
energy under similar blasting conditions. In order to apply the technique, the seismic
energy released by each hole can be analyzed as a pulse taking place at firing times,
according to the delay sequence used in a blast. The previous statement implies that it is
possible to use the delta function to mathematically represent the firing of each hole as:

Esis

= A

o(t -m

t
o
)
[3.4]
Where:
Esis
i
: Seismic Energy release in the hole i
A
i
: Seismic Energy Efficiency in hole i
t: time
m
i
t
o
: Time delay for hole i

According to the signature hole assumptions, the seismic energy efficiency hole to
hole ,A
i
, is equal to one (1).
3.3 Seed waveform
Many studies have shown how some parameters affect the characteristics of the
seismic waves produced due to a mine blast. The main parameters involved in the blast
vibration phenomena can be summarized as follows (Aldas, 2010):
Explosive-rock interaction;
Blast-induced wave transmission property of a rock unit (i.e waves
traveling along specific layers);
Distance between blast location and measurement point;
Geology of the propagation media (i.e faults, bedding planes, etc);
Geology at the measurement point;
28

Blasting parameters (i.e diameter, explosive type, borehole depth, spacing
and burden, delays and free faces).

The first five of six elements involved in the wave generation are site specific and
related to the geology. Despite the site specific nature of the phenomena, the assumption
that each hole produces the same waveform in a specific measurement point is based on
the fact that the frequency spectra for different blast single holes are similar according to
Crenwelge (1988) findings.
In 1988, Crenwelge reported the typical spectra of seismic waves due to single
blast hole. In addition, other variables like weight of explosives, the type explosive and
the source distance measurement point were studied to determine their influence on the
spectra. Figure 3.1 shows the particle velocity spectra for single charge - shots of small
column weight.

















Figure 3.1 Particle velocity spectra for single charge shots of small column weight
(Crenwelge, 1988)
In this case, the seismic waves were recorded at the same point of interest
(geological conditions, distance to source event and wave travel path are equal). The
difference between both events is the quantity of explosive detonated (90 and 180 lbf)
and no rock breakage influence between them. Despite the difference in the weight of
explosives, in general terms, Figure 3.1 shows the similitude in the shape of the spectra of
the two shots.

Similar results are obtained when the weight of explosives detonated is further
increased in the hole. Figure 3.2 shows the seismic wave spectra for explosives with
weights between 250 and 2,000 lbf.




29


















Figure 3.2 Particle velocity spectra for single charge shots of tall column weight
(Crenwelge, 1988)

Different results are obtained in the frequency spectrum if the distance to the
source-recording site changes. With the change in the distance, the geological conditions
and travel path of the seismic waves change. Figure 3.3 shows how the shape of the
spectra changes with the distance for both weights of explosives (small and high column
weight).


















(a) (b)
Figure 3.3 Particle velocity spectra for single charge shots (a) short column
weight (b) tall column weight (Crenwelge 1988)
30

These results show, as expected, that vibrations are site specific and the shape of
the spectra is similar for the same site even if the weight of explosive is reduced or
increased. Sakamoto et al (1989) in their study about the accuracy delay in detonators
showed the signature waveform for two locations (100 and 150 meters) using three
different weights of explosives in a limestone mine. As previously noticed with the
spectrum, the signature waveforms are similar for three different loads. This is included
in Figure 3.4












Figure 3.4 Seismograms recorded for three different charge weights and two
locations (After M. Sakamoto, et al., 1989)
Similar results corresponding to waveform similitude were found by Bonner et
al., 2008, where the effect of the explosive type in the rock damage and the source of
shear wave generation were studied. Figure 3.5 shows the waveform similitude for a
given fixed vibration recording station when different shots using different explosives
where tested.

















Figure 3.5 Seismograms recorded for three different charge weights (After
Bonner, et al., 2008)
31

When a spectral analysis is done using those waveforms the vibration frequency
content is between 1 and 22 HZ as shown in Figure 3.6.
















Figure 3.6 Spectral analysis for three different charge weights (Bonner, et al.,
2008)
The original signature technique assumes the invariability in the waveform from
different holes at the same station or measurement point. However, recently proposals
recognize the waveform variability hole-to-hole in the signature hole methodology Aldas,
(2010), Yang and Scovira (2010), Blair (1999), however the inclusion of the signature
waveform variability in the methodology is not totally understood and the parameters
used to change the waveform between holes are difficult to assess.

3.4 Explosion sequence
Based on signals and systems theories and assuming that the full-blast vibration
record occurs in a casual, linear and invariant system and the phenomenon follows a
linear superposition, the whole vibration record for the full-blast y(t) can be
mathematically expressed as the convolution of the signature waveform (impulse
response-signature hole) b(t) and a delta Dirac sequence (input sequence or blast timing
sequence) u(t). The relationship is given by:

y(t) = b(t) u(t) = _ b()u(t -)J

-

[3.5]
In discrete terms Equation 3.5 can be expressed as
y|n] = b|k]u|n -k]

k=-

[3.6]
32

time
Explosion function
u(t)
T
n
u(nT)
m t
t
Delay between holes
Hole (j-1) Hole (j)
time
The explosion sequence u(t) or u|k] is a function representing the energy
released in each hole that is transformed into vibration and depends also of the time delay
between holes.












Figure 3.7 Explosion sequence function
The full-blast sequence function u(t) can be approximated using a stair-step
function given by:
u(t) = u(nI) | o
1
(t -nI)]I

n=0

[3.7]
Where:
o
1
(t -nI) pulse occurring at t = nI
|o
1
(t -nI)]I pulse height
T pulse width
u(nI) u(t) evaluated in t = nI

The graphical representation of Equation 3.7 is shown in Figure 3.8.












Figure 3.8 Graphical representation of Equation 3.7

33

If the vibration of the ]tb hole is isolated, the vibration component of the ]tb hole
can be expressed as:
y
]
(t) = b(t) u
]
(t) = _ b()u
]
(t -)J

-

[3.8]
Using Equation 3.7 and isolating from the full-blast function the explosion of the
hole j, it is possible to express the explosion function for the ]tb hole as (Blair 1999):

u
]
(t) = A
k]
o(t -kt)
m
k=1

[3.9]
Where:
u
]
(t) explosion function for hole ] (group of delta functions)
mt duration of the Dirac sequence for the explosion in the hole j
A
k]
Amplitude of Dirac delta functions
t time interval (pulse width in Equation 3.7)
In Equation 3.9, the amplitude of Dirac delta functions A
k]
is constituted by two
components; the first one is related to the relative amount of coherent energy in the
waveform of the ]tb blasthole and the second represents the total amount of random
energy.
In this sense and using Equation 3.8 and Equation 3.9, the vibration generated for
the ]tb blasthole is given by
y
]
(t) = (1 -R)b(t) +
R
R
]
A
k]
b(t -kt)
m
k=1

[3.10]
Where:
y
]
(t) vibration for the jth blasthole,
R relative amount of random energy for each blasthole,
R
]
measurement of the total energy of the random component,
evaluated using Parsevals Theorem as R
]
= A
k]
2 m
k=1
,
A
k]
a random number in the range -1 to 1.

In Equation 3.10, if R=0, all blast hole waveforms are identical and there is not a
random energy component exist in the full-blast. Notice that, if R=0, the vibration
generated by the ]tb blasthole is equal to the signature waveform. On the other hand, if
R=1, the waveform for each hole in a full-blast are totally different and there is no
correlation between the signatures of any of the blast holes and the prediction of the
complete waveform using signature technique is not possible. Field measurements
reported by Blair (1993) suggest that a model based upon R=0.8 is reasonable.

The total vibration y(t) for a sequence of delayed blast holes is given by:
34

y(t) = y
]
(t -J
]
)
N
]=1

[3.11]
Where:
N: total number of blastholes
J
]
: is the ]tb initiation delay time.
y
]
: vibration for hole j
t: time

3.5 Linear superposition
The mathematical development behind signature hole technique requires that
linear superposition to do the summation (convolution) of the signals be possible. A
system is linear if the system satisfies two properties; homogeneity and additivity. A
system is homogeneous if scaling the input the predicted output is going to be scaled by
the same quantity (as it is illustrated in Figure 3.9 ). In other words:
ou
1
(t) oy
1
(t)
[3.12]
Where:
o: real constant
u
1
(t): input
y
1
(t): output




















Figure 3.9 Systems homogeneity property
35

Because o is a real constant, homogeneity is also called the scalar rule of a linear
system. This property is not completely satisfied in a real production blast; however
there is a relationship between the quantity of explosive used and the vibrations level
generated. Using the scale law, despite all the inconsistencies previously mentioned, the
relationship between the weight of explosive and the particle velocity in a given
explosive weight interval can be assumed that the homogeneity property is satisfied.
Figure 3.10 was elaborated using site constant values of: o = 1uu, b = -1.S and
= 1uu t.
The other property previously mentioned that a system should fulfill to be linear
is additivity. The additivity concept (illustrated in Figure 3.11) is expressed for any pairs
input output {u

(t) y

(t)], for i = 1, 2 as:

u
1
(t) +u
2
(t) y
1
(t) +y
2
(t)
[3.13]
Where:
u
1
(t), u
2
(t): input 1 and 2
y
1
(t), y
2
(t): output 1 and 2
























Figure 3.10 Numerical homogeneity concept between the quantity of explosive
and the particle velocity.

36































Figure 3.11 Additivity concept for signature hole technique
In the next chapter the Monte Carlo scheme is introduced in conjunction with the
signature waveform methodology to improve the capability of blast vibration levels
prediction.
37

Chapter 4


MODEL DEVELOPMENT
4.1 Introduction
In this chapter, the Monte Carlo scheme is introduced to the signature hole
technique. The probabilistic component for each stage of the methodology where it is
possible to introduce this type of approach is discussed next.

4.2 Monte Carlo scheme
In a mining blast, the variables involved are related to geometry (e.g., depth of the
holes, diameter, surface at the face etc). Quantity of explosives and geology are not
constant hole-to-hole and they present some degree of uncertainty. This uncertainty
makes it possible to treat some of the variables involved in the mining blast vibration
phenomena as random variables within a given range.
Randomness sources in the blasting process come from different elements; those
that are in situ such as the geology, the joint rock mass system, the underground seepage,
etc., and those involving procedures of human activities like the drilling process, the
loading of the holes with explosives and other blasting components like the detonation
system.
Despite the high randomness involved in the mining blasting vibration
phenomena, there are methodologies to study the problem using reasonable assumptions
and approaches to obtain logical and meaningful results.
By solving blast vibration phenomenon as a random process, there is not a single
answer or an exact prediction result. Most of the current deterministic methodologies to
estimate the level of the particle velocity due to blasting are based on scaled distance.
Using a probabilistic approach like the Monte Carlo scheme, the result will be a
probability distribution of peak particle velocity according to the probability distribution
of the variables in the problem. To introduce the general idea about Monte Carlo scheme
using signature hole technique, it is necessary to define some basic concepts.


4.2.1 Discrete Random Variables
According to probability theory, a random variable is a variable such that we do
not know the value of this quantity in any given case, but we know what values it can
assume and we know the probabilities with which it assumes these values Sobol, (1960).
A random variable X is called discrete if it can take any of a discrete set of values
{x
1
, x
2
, x
n
]. So we can define X as given by Equation 4.1 Sobol, (1960)

X = [
x
1
x
2
x
n
p
1
p
2
p
n

[4.1]
38

Where:
x
1
, x
2
, x
n
possible values of the variable X
p
1
, p
2
, p
n
probabilities corresponding to possible values of the
variable X
Equation 4.1 is called the distribution of the random variable X

The probability p

that the random variable has the value x

is denoted by:

P(X = x

) = p

[4.2]
The values of the numbers x
1
, x
2
, x
n
are arbitrary, however the probabilities
p
1
, p
2
, p
n
must satisfy the conditions given in Equation [4.3] and [4.4]:

p

u
[4.3]
and
p
1
+p
2
++p
n
= 1
[4.4]
Last condition ([4.4) means that in every event X must assume one of the values
x
1
, x
2
, x
n
.
The number given by:
E(X) = x

n
=1

[4.5]
is called the expected value, or mathematical expectation, of the random variable
X.

Some basic properties of mathematical expectation are given by:

E(X +c) = E(X) +c

E(cX) = cE(X)
and
E(X +) = E(X) +E()
[4.6]
Where:
c: is any constant
X, : random variables

The variance defined as the mathematical expectation of the squared deviation of
the random variable X from its average value E(X) is given by:

39

Ior(X) = E((X -E(X))
2
)
or
Ior(X) = E(X
2
) -(E(X))
2

[4.7]
and is a measure of how far a set of numbers is spread out from the mean. As in
the case of the mathematical expectation variance has some basic properties:

Ior(X +c) = Ior(X)

Ior(cX) = c
2
Ior(X)
[4.8]
Two random variables are independent when watching both variables, if the
distribution of the variable X (Equation 4.1) does not change when the value which the
variable assumes is known. If two random variables are independent the basic
properties given by Equation 4.9 are satisfied:

E(X) = E(X)E()
and
Ior(X +) = Ior(X) +Ior()
[4.9]
To define the concept of probability density or density distribution of the random
variable, it is necessary to assign a function to the probabilities of the possible values of
the variable X in a given interval (a function for p
1
, p
2
, p
n
in Equation 4.1). In other
words, if a random variable X is defined in an interval of values |o, b], (the interval of
possible values x
1
, x
2
, x
n
) and a function p(x) is assigned to these interval to
represent the probabilities of those values (p
1
, p
2
, p
n
), then p(x) is called the
probability density or density distribution of the random variable X. The significance of
p(x) is as follows: let (o
i
, b
i
) be an arbitrary interval contained in |o, b]. Then the
probability that X lies in the interval (o
i
, b
i
) is equal to:

P(o
i
< X < b
i
) = _ p(x)Jx
bi
ui

[4.10]
The type of probability density to assign or to use for the random variable X
depends of the physical process to represent. It has been seen that normal random
variables are often encountered in nature.
A normal (Gaussian) random variable is a random variable Z defined on the
whole axis (-, ) and having the density function given by:

p(x) =
1
o2n
c
_-
(x-u)
2
2c
2
_

[4.11]
40

Where:
o, o: numerical parameters where o > u.

In probability theory, it is possible to show that:

E(Z) = o and Ior(Z) = o
2

[4.12]
One of the reasons or mathematical explanations why normal random variables
are often encountered in nature is related to the concept of the central limit theorem of
probability theory. Central limit theorem says that if there are N independent, identically
distributed random variables (same function p(x)), their mathematical expectations and
their variances also will coincide. Such theorem is expressed as:

E(X
1
) = E(X
2
) = = E(X
N
) = m
and
Ior(X
1
) = Ior(X
2
) = = Ior(X
N
) = :
2

[4.13]
Also, if we denote the sum of all N variables by S
N
:

S
N
= X
1
+X
2
++X
N

and
E(S
N
) = E(X
1
+X
2
++X
N
) = Nm

Ior(S
N
) = Ior(X
1
+X
2
++X
N
) = N:
2

[4.14]
Now if there is a normal random variable Z
N
with parameters

o = Nm and o
2
= N:
2

[4.15]
By the theorem of the central limit The density of the sum S
N
approaches the
density of the normal variable Z
N
in such a way that for every x,

p _
S
N
-Nm
:(N)
< x_ = p _
Z
N
-Nm
:(N)
< x_
for all large N.
[4.16]
the significance of this theorem is clear: The sum S
N
of a large number of
identical random variables is approximately normal, or:

p
S
N
(x) = p
z
N
(x)
[4.17]
41

Using all previous probability concepts, next a brief explanation of the Monte
Carlo scheme is given.

4.2.2 General Scheme of the Monte Carlo Method
Suppose we want to determine some unknown quantity m. Let us assume a
random variable X that satisfies:
E(X) = m
and
Ior(X) = :
2

[4.18]
Consider N independent random variables X
1
, X
2,
, X
N
with distributions
identical to that of X. If N is sufficiently large, then, according to the central limit
theorem, the distribution of the sum S
N
= X
1
+X
2
++X
N
will be approximately
normal with parameters o = Nm and o
2
= N:
2
.
In a normal distribution it is determined that:

_ p(x)Jx = u.997
u+3c
u-3c

[4.19]
Or in other words the probability that a random variable Z obtain a value differing
from E(Z) = o is less than So, or:

P(o -So < Z < o +So) = u.997
[4.20]
Using Equation 4.20 in the case of the sum S
N
of random variables it is obtained:

P(Nm -S:N < S
N
< Nm +S:N) = u.997
[4.21]
Rearranging terms finally it is obtained:

P __
1
N
X
]
-m
N
]=1
_ <
S:
N
_ = u.997
[4.22]
Then if it is found N values of the random variable X, the arithmetic mean of
these values will be approximately equal to m, the quantity that needs to be determined.
Also the probability that the error of such approximation does not exceed the quantity
S:
N
,
is high and tends to zero when N increases.

42

4.2.3 Pseudorandom numbers generation
In this project, Matlab was used as an engine linked to Visual Basic to
program the signature hole methodology including Monte Carlo method. In the
programming stage, pseudorandom numbers were used to generate the random variables
involved in the problem. Pseudorandom numbers differ from true random numbers in
that they are generated by an algorithm, rather than a truly random process. However, the
generated numbers are random in the sense that, on average, they pass statistical tests
regarding their distribution and correlation. In this project, the function ronJn was used
to generate pseudorandom numbers from a normal distribution, using the algorithm called
Mersenne Twister generator (Makoto Matsumoto., Takuji Nishiura., 1998).
By definition in Matlab, the function ronJn follows a normal distribution
having a zero (0) mean and a variance equal to one (1). Figure 4.1 shows ronJn
function in Matlab.

















Figure 4.1 Histogram for the ronJn function in Matlab.

Next, a description about the introduction of the probabilistic approach in each of
the stages of the signature waveform technique is given.


4.3 Seed waveform variability
As mentioned previously, all current methodologies using signature waveform
technique assume that the signature wave does not change hole-to-hole. However, there
are at least three reasons why the signature waveform hole-to-hole is not the same:

Damage in the surrounding rock by previous holes;
Difference in the distance and the path that vibration follows between the
hole and the monitoring point;
43

Hole (i) Hole (j)
S
d(i) d(j)
Station
Afected area
limit
Variation in drilled hole, loading, contamination of explosives, priming
effects, etc.

The first phenomena that affects the waveform hole-to-hole is related to the
damage in the surrounding rock by previous holes. When a loaded hole is detonated, it
changes the rock properties around the hole in a specific area. The extension of such area
is a function of the initial conditions of the rock (i.e. rock joint system before detonation)
as well as the geometry of the hole and the efficiency of the chemical energy transferred
from the explosive to the rock. Accordingly, if the separation between holes S is enough
to have no interference, the signature waveform from hole (i) will be equal to the
signature waveform from hole (]) as show in Figure 4.2.
















Figure 4.2 Signature hole reproducibility (adapted from Blair 1999)
If the affected area from hole (i) over lay or interfere with the affected area from
hole (]), (i.e. the separation between holes S is such that affected areas interfere), there is
a need to find a relationship to describe the nonlinear variation of the signature waveform
hole-to-hole in a production blast event.
As previously mentioned, Yang and Scovira, (2009) propose the waveforms
changes as a function of the ratio between the quantity of explosive already blasted to the
quantity of explosive that produce the waveform:

s() = z
q

[4.23]
where
=

t

[4.24]

t
total charge weight of earlier fired blast hole in the vibration path
charge weight of the presently firing charge, where <1
44


The limit conditions of this equation is given by
s() = _
1, u
u,

[4.25]
Field calibration of parameter and its the relation to the rock mass properties
still is not well understood, despite the straightforward nature of Equation 4.23 and the
intuitive correct meaning of the decrement in the amplitude of the vibration due to the
detrimental quality of the rock mass as a result of previously blasted holes. Equations
4.23 to 4.25 are a good proposal to introduce a logical screening factor to reduce the
amplitude of the waveform due to the damage of the rock mass. However such damage
is considered only in the direction of the monitoring point and do not take into account
that the cracks due to the explosion of previous holes should grow in all directions and
the energy release by later holes can be affected by previous detonated holes, even if they
are not in the vibration path.
The second reason why the waveform should change is due to the difference in
the distance and the path that the vibrations follow between the hole and the station or
monitoring point as shown in Figure 4.2 (i.e. the distances J(i) and J(]) are different for
the full-blast situation). Blair (1999) proposed, using weight scaling laws, a methodology
to take into account the change in the amplitude of the seed waveform, due to the
variability of the distance between the holes and the specified monitoring vibration
location point in a full-blast. According to the traditional weight scaling law, the vector
peak particle velocity :pp: is given by the relation between distance J, charge weight w
and the geology constants o and b as:

:pp: = o(S)
-b

[4.26]
Where SD is defined as:
S =
J
w

[4.27]
As explained, despite of all limitations of Equation 4.26, this equation can be used
to scale numerically the seed waveform for each blast hole in a full blast event. (Blair
1999).
To establish the proper values of the parameters o and b in Equation 4.26, it is
recommended to develop a curve S vs :pp: for signature holes. From these graphs, o
and b coefficients can be established for a particular site. These correlated parameters
will result in over estimation of the :pp: because isolated blast holes are usually fired in
virgin ground (Blair, 1999) in contrast to ground previously affected by production
blasting as in the real mining situation. For this reason, it is desirable to get the seed
waveform in the most representative working conditions (along to one of the production
shots). Finally, if there is information regarding signature holes, i.e. vibration records for
a single blast hole, it is possible to use Monte Carlo schemes adjusting the values of o
45

and b and choosing the most representative value of these parameters for particular
ground conditions.

4.3.1 Modeling of changes in signature waveform
In this research, a methodology for varying the signature waveform hole-to-hole
was developed. This approach takes into account the change in both main parameters of
the seed waveform amplitude and frequency. Notice that current methodologies only
modify amplitude in the waveform hole-to-hole. The methodology is based on the main
characteristics of any recorded signature waveform and the use of Fourier series to
approach one equation to produce different waveforms for each hole in the vibration
prediction process. Involving the change in amplitude and frequency of the waveforms,
it is expected that changes in the surrounding material to the detonated hole, changes in
the path from hole to monitoring point and variations in explosive output due to drilling
and loading procedures be involved in the prediction process.
During the development of this research, several field blast tests were performed
in order to study the variability of the signature waveform in a mining production blast
event. Several tests were performed at the Guyan surface coal mine in West Virginia
during the summer of 2011 for validation of models.
The rock mass is comprised of layers of sandstone and shales. It is possible to see
in Figure 4.3 the geometrical arrangement used to do the test.















Figure 4.3 General view of the tested area
In total, 11 holes of 7.875 inch diameter were detonated. The depth of all holes
was 42ft and the spacing and burden used was 15ft and 17ft, respectively. The main
objective was to collect a series of signature holes combined with the detonation of two
or more holes using delays with timing lower than 8ms. To have better control over the
timing, electronic detonators were used. In this test, six holes were detonated using delay
timing between holes of 5ms, followed by two signature holes. After the two signature
holes, two holes using 5ms delay were detonated and finally a signature hole was
detonated. In summary, in this test there are three signature holes and two sequences of
46

0
10 20 30ft
N
Bl ast ed
Soli d
Soli d
06/ 22/ 2011 3: 15 pm
Si gnat ur e Hol e
holes delayed by 5ms, the first sequence was composed by six holes while the second
was composed by two holes.

Figure 4.4 shows the plan layout of the test indicating the signature holes, the
sequence number and the timing used. The layout was generated using actual GPS
surveyed locations for each hole.























Figure 4.4 Test plan layout

Along with the blast test, a seismograph network was setup at the mine. Figure
4.5 shows the location of the seismographs and the area where the test was performed
(blasting area summer 2011 in the map).












47


















Figure 4.5 Seismograph network at Guyan mine
The detailed analysis of the signature waveform variability and modeling was
performed for seismograph 01. However, this analysis is valid for any waveform.
Seismograph 01 was located 210m (689 ft) from the blast area.

The complete waveform for the closest seismograph (seismograph 01) and three
components (radial, traversal and vertical) are included in Figure 4.6.




















Figure 4.6 Complete signature waveform three components
48

Figure 4.6 shows clearly three signature waveforms and two other blast series. To
develop the approach of simulating the signature waveform using Fourier series, the
radial component was chosen; however this procedure can be used for any other vibration
component (vertical or traverse).

First, a comparison between signature holes was performed. Such comparison
was done in both time domain and frequency domain. Comparing the waveforms and the
frequency content of the signatures it is possible to see how similar or dissimilar the
signatures waveforms are. This also sheds light on the validity of the assumption
regarding the invariability of the waveform used in the current signature hole vibration
modeling techniques.

4.3.1.1 Signature waveform comparison
The radial component was used to develop the approach of simulating the
signature waveform using Fourier series. Figure 4.7 shows the complete record for the
test in this direction.























Figure 4.7 Complete waveform, Radial component Seismograph 01

Figure 4.8 is obtained when signature waveforms are isolated from the complete
record and compared. They are placed on the same time domain to compare shapes and
amplitudes.
49

It is possible to see in Figure 4.8 how the signatures waveforms change while the
total blast is progressing. The maximum positive peak occurs for signature 3, after 10
holes have been detonated. In this test, the maximum positive peak is between 0.07 and
0.095 in/s (signature 2 and 3 respectively). For the positive peaks, the range is 0.025 in/s
of variation. On the other hand, the maximum negative peak is reached in the signature 2
and the range for the negative amplitude is 0.04 in/s of variation.



























Figure 4.8 Signature waveform comparison
The maximum negative peaks are -0.095 and -0.055 in/s, presented in signature 2
and 3 respectively. In this test, the intermediate values for the amplitude of the waveform
(positive and negative) are found in signature 1.

Signature 3 is characterized by the highest vibration output for many reasons. As
shown in Figure 4.4, signature hole 3 (corresponding to hole 11) is the most confined and
nearest hole to the monitoring point. Two factors (confinement and distance) result in the
phenomenon that signature 3 produces the maximum vibration output in this test. Other
factors that could have led to this outcome include different geologic paths from the
source to the monitoring point and different energy output of each hole due to hole
loading, explosive composition and/or contamination, priming etc. These factors result in
50

varying waveforms generated for each hole, making the waveform produced by each hole
a somewhat random process within a given range.

Figure 4.9 shows the frequency content of the three signals. This representation of
the signals reflects how the energy content of the signals (area under the curve) is
diminishing while the blasting is progressing, meaning that the last blast (signature 3)
applies lower energy to the ground than the other two signature holes. This behavior is
expected because the energy applied by the explosion to the ground that becomes
vibration is lower as the quality of the rock is also diminishing. Even though the three
curves are not equal, it is possible to distinguish the same four frequency zones for the
three spectrums. Such zones are defined as the zones where the spectral amplitude peaks
are reached. The frequency limits for the zones are (Figure 4.9): between 0 and 9 Hz,
from 9 to 12, from 12 to 17 and frequencies greater than 17 Hz.

The variation of the peak frequency content between the three curves inner to
each zone is low for example for zone 1 the peak frequency is between 6.68 to 7.01 Hz,
zone 2 between 10.70 to 10.74, zone 3 between 13.05 to 13.74 and in zone 4, where most
variability in the peak frequency content is presented, ranging between 22.09 to 30.20
Hz.





















Figure 4.9 Frequency content of the signals and four frequency zones
After analyzing time vs particle velocity and the spectral content of the signals
some conclusions about the data set can be drawn as follows:

a. There are some similarities between the three waveforms such as the time where
peaks are reached in the time vs particle velocity curve.
51


b. Signature 1 and signature 2 are similar but signature 3 is different from the other
two.

c. Energy content of signature 3 is the lowest but one of the highest particle velocity.

d. There are changes in the peak amplitude of the particle velocity between the three
signals and there is not a clear trend in the values of the peaks while the blast is
progressing.

e. For the three signals, it is possible to divide the curve frequency vs spectral
amplitude in four zones.

Using the methodology proposed by Yang (2010), it is possible to reduce the
amplitude of the particle velocity while the blast is progressing, but that trend is based on
the assumption of lower values of particle velocity through the blasting process. Also in
that proposal, the holes should be in the same vibration path, so in this particular case
screening equations of Yang and Scovira (2010), are not useful. Finally, varying the
amplitude without varying the frequency content between holes in a blast event may not
provide optimized model simulations.
4.3.1.2 Fourier series and Signature Waveform
Using Fourier series, it is possible to express any arbitrary periodic function as a
sum of sine and cosine terms. In other words, Fourier series can be used to express a
function in terms of the frequencies (harmonics) that it is composed of. The
representation of such function (t) is given by:

(t) = c
o
+ _o
n
sin _
2nnt
I
] +b
n
cos _
2nnt
I
]_

n=1

[4.28]
Where:
c
o
=
1
I
_ (t) Jt
1
0

[4.29]
o
n
=
2
I
_ (t) sin _
2nnt
I
] Jt
1
0

[4.30]
b
n
=
2
I
_ (t) cos _
2nnt
I
] Jt
1
0

[4.31]
In order to express (t) in terms of sin function, from:

52

o sin(2nt) +b cos(2nt) = A sin (2nt +)
[4.32]
Where:
A = o
2
+b
2

[4.33]
And
= ton
-1
[
o
b

[4.34]
Finally we can express the function (t) as:
(t) = c
o
+ _A
n
sin _
2nnt
I
+
n
]_

n=1

[4.35]
As an example, signature 1 is expressed using Fourier series. Figure 4.10 shows
signature 1 isolated from the complete record.

















Figure 4.10 Signature 1 Isolated from complete record (Radial component)
Fourier series is used for periodic functions, however if we take a period I equal
to 1 second for Figure 4.10, we will reproduce the complete waveform every second.
After one second, the complete waveform will repeat itself in a periodic manner. Using
Equation 4.28 to Equation 4.35, Figure 4.11 is the value of the amplitudes A
n
(Equation
4.33) of the Fourier series coefficients (o
n
, b
n
) given by the Equation 4.30 and Equation
4.31 respectively. In Equation 4.28 the quantity of coefficients to calculate are infinite,
however this example will use 25 coefficients to reproduce the signature 1 waveform.


53

























Figure 4.11 Magnitude coefficients from Fourier series

Figure 4.11 shows the amplitude calculated for each coefficient in the Fourier
series, as mentioned before; the first 25 coefficients of the series were included. As
expected, the shape of Figure 4.11 is similar to the Fourier Transform (FT) of the signal.

If the magnitude of the coefficients of the Fourier series is plotted in the same
graph that the Fourier Transform, Figure 4.12 is obtained. In Figure 4.12 the two curves
are not exactly the same because Equation 4.28 is an approximation to the real signal.













54





















Figure 4.12 FFT and magnitude coefficients from Fourier series

Using Equation 4.28 and considering 25 coefficients in the series, for a period of
two seconds we get the signal included in Figure 4.13.





















Figure 4.13 Fourier series and signature 1
55

Using Fourier series we have expressed the waveform of signature 1 using 25
terms in the Equation 4.28. Because this mathematical approach is for periodic signals,
after 1 second the waveform repeats itself (same shape) and so on.

The objective in this research is to find the manner to reflect the changes in
amplitude and frequency content of the signature waveform while the blast is
progressing. Using Fourier series it is possible to find a mathematical expression for any
wave form (in this case signature 1). The mathematical expression using 25 coefficients
to reproduce signature waveform 1 is given as follows:


[4.36]
Using software, it is possible to use any number coefficients in the Fourier Series
to find the mathematical expression that represent any signature waveform. In order to
simplify the mathematical development included in the next section, only four
frequencies were used to find the approach equation.

4.3.1.3 Signature Waveform approach based on Fourier series
The main characteristics observed in the signatures waveforms from the field test
are:
1. It is possible to determine for all signatures the zones where peak frequencies are
presented. The main characteristic of those frequency zones is that they are the
same for all signatures. The peaks frequencies within those zones for different
signatures are similar. (Figure 4.9)
2. There is not a clear trend regarding the maximum amplitude of the signature
waveform while the blast is progressing. However the peak values (positive and
negative) are in a narrow range. (Figure 4.8)
3. It is a fact that after some time the blast vibration attenuate, in contrast to Fourier
series where the waveform is repeated itself each period of time I.

Based on these characteristics, the general guidelines to approach a simplified
function to represent the signature waveform are presented next.


56

a. Main frequency content of the signal and number of Fourier coefficients
From a mathematical point of view, it is possible to establish the number of
coefficients in the Fourier series that are good enough to have a good simulation of a
function. In that case, if F(t) is a simulation of (t), the number of coefficients N in the
simulation is such that the error between F(t) and (t) is minimum. In other words:

F(t) = c
o
+ _o
n
sin _
2nnt
I
] +[
n
cos _
2nnt
I
]_
N
n=1

[4.37]
and the error given by:

E =
2
I
_ ((t) -F(t))
2
Jt
1
0

[4.38]
should be minimum. In the current simulation, the proposal is to take the number
of coefficients equal to the main frequencies of the signal (however it is possible to use a
large number of frequencies if the shape of the waveform is complicated).

Following the assumptions, for example in the case of signature 1, the frequencies
used to calculate their coefficients in the Fourier series are 6.68, 10.70, 13.74 and 22.72
Hz (Figure 4.12).

Calculations for the first frequency
The first frequency corresponds to 6.68 Hz. A time interval of 1 second is
assumed in order to perform the integral for the coefficients so:







c
o
=
1
1
] (t) Jt
1
0
= 0.002353




o
n
= 2 _ (t) sin(2n6.68t) Jt
1
0

=2* 0.0012836 = 0.0025672



57





b
n
= 2 _ (t) cos(2n6.68t) Jt
1
0

=2* - -0.008023 = -0.016046

Phase calculation:
= ton
-1
[
0.0025672
0.016046
= u.1S864;
0.15864
n
= u.uSu49; u.uSu49 -u.S =
-u.449S n

So the term for the frequency of 6.68 Hz is:

Icrm
6.68 Hz
= u.u162S sin (2n 6.68 t -u.449Sn)
[4.39]
Equation 4.39 is the sin, cos component expressed using amplitude, A, and
phase, , from Equation 4.35 for the frequency at 6.68 Hz.

Calculations for the others frequencies
Following the same procedure, the results for the frequencies of 10.70, 13.74 and
22.72 Hz are presented next.

Frequency of 10.70 Hz
o
n
= 2 u.uuS646 = u.uu7292
b
n
= 2 -u.uu4S86 = -u.uu8772
= ton
-1
_
u.uu7292
u.uu8772
] = u.69SS2;
u.69SS2
n
= u.22u7S; u.22u7S -u.S
= -u.27924 n
Icrm
10.70 Hz
= u.u114u sin (2n 1u.7u t -u.27924n)
[4.40]
Frequency of 13.74 Hz
o
n
= 2 -u.uuS91 = -u.u1182 -0.011631
b
n
= 2 u.uuu9114 = u.uu18228 0.0016649
= ton
-1
_
u.u1182
u.uu18228
] = 1.4177;
1.4177
n
= u.4S129; u.4S129 +u.S
= u.9S129 n

Icrm
13.74 Hz
= u.u1196 sin (2n 1S.74 t +u.9S1Sn)
[4.41]
Frequency of 22.72 Hz
o
n
= 2 -u.uuu2uS9 = -u.uuu4u78
b
n
= 2 u.uu1SuS1 = u.uuSuu62
58

= ton
-1
_
u.uuu4u78
u.uuSuu62
] = u.1S48S;
u.1S48S
n
= u.u4291; u.u4291 +u.S
= u.S429 n

Icrm
22.72 Hz
= u.uuSuSS sin (2n 22.72 t +u.S429n)
[4.42]
When Equation 4.39 to Equation 4.42 are together, the base equation to represent
the waveform of the signature 1 given by:

(t) = u.uu2SSS +u.u162S sin(2n 6.68 t -u.449Sn) +u.u114u
sin(2 1u.7u t -u.27924) +u.u1196
sin(2 1S.74 t +u.9S1S) +u.uuSuSS sin (2 22.72 t
+u.S429)
[4.43]
When this equation is plotted against the signature waveform measured in the
blast test, Figure 4.14 is obtained.

Different from a Fourier series using 25 coefficients (Equation 4.36), Equation
4.43 is a rough approximation to the measured waveform, so there is not a perfect match
between two curves.
To improve the approximation, a decay factor is included to restrain the
waveform to the duration time of the vibration (one second in this case) and one
amplitude scale factors to match the amplitude of the waveform in the maximum peak.


















Figure 4.14 Measured signal Vs base equation


59

b. Decay factor calculation
As previously mentioned, after certain time, the blast vibration should decays to
zero. However and due to the nature of the Fourier Series, using the mathematical
expression, the signal repeats itself in a period T of time (see Figure 4.13). The decay
factor is necessary to avoid that problem. To calculate the mathematical expression for
decay, an exponential trend line is used to envelope the positive peaks of the waveform.
Figure 4.15 show the signature 1 waveform, the envelope and the exponential decay trend
line.






















Figure 4.15 Exponential decay calculation

Figure 4.15 shows the decay factor, for signature 1 waveform the value of the
decay exponent estimated is -3.35. Now a modification to Equation 4.43 is made,
introducing the exponential decay, this results in Equation 4.44:

(t) = (u.uu2SSS +u.u162S sin(2n 6.68 t -u.449Sn) +u.u114u
sin(2 1u.7u t -u.27924) +u.u1196
sin(2 1S.74 t +u.9S1S) +u.uuSuSS
sin(2 22.72 t +u.S429)) e
-3.35t

[4.44]
Figure 4.16 show Equation 4.44 when it is compared against to the measured
signal.

60




















Figure 4.16 Measured signal Vs base equation including exponential decay factor

c. Amplitude factor calculation
Finally, matching the peak value of the measured signal and the amplitude of the
curve given by Equation 4.44 at the time when the peak of the measured signal is
reached, it is possible to calculate the amplitude scale factor given by:

ASF =
u.u8S
u.u18S
= 4.64
[4.45]
Figure 4.17 show the adjusted equation after match the maximum amplitude.

(t) = 4. 4 (u.uu2SSS +u.u162S sin(2n 6.68 t -u.449Sn) +u.u114u
sin(2 1u.7u t -u.27924) +u.u1196
sin(2 1S.74 t +u.9S1S) +u.uuSuSS
sin(2 22.72 t +u.S429)) e
-3.35t

[4.46]








61























Figure 4.17 Measured signal Vs final approach
When measured signal and Equation 4.46 are deducted following the procedure
previously described, a good correlation between both curves is clearly observable.
When a cross correlation is used to measure the differences between the measured
signature waveform and the approach or simulated waveform a coefficient of 0.85 is
obtained (one is perfect correlation or exact waveform between two signals when they are
compared). Next, it is necessary also to compare the signals in frequency domain.


Comparison in frequency domain
It is necessary to compare both signals, measured and approximated, in frequency
domain in order to see if the approximation keeps the energy content of the measured
signal. Figure 4.18 shows the frequency content comparison.

Figure 4.18 shows the similarity between both signals in frequency domain. The
dominant frequencies in the approximated signal keep the peak values of the measured
signal and they are between the zones previously defined (Zone 1 to Zone 4).







62

















Figure 4.18 Frequency domain comparison
Bigger values in the peaks of the approximated signal means that the
approximated signal carries more energy content than the measured signal this is evident
when the signals are compared in the time domain.

In conclusion, in this research a basic equation base on Fourier series to approach
the signature waveform is proposed. The equation (Silva-Lusk equation) has the general
form:

(t) = _c
o
+ ASF
m
{A
m
sin(2n requency
m
t +
m
)]
m
n=1
_ c
-decay acturt


[4.47]
where:
ASF
m
: amplification scale factor for frequency m.
c
o
: first term in the Fourier series
m: number of frequencies chose to approach the measured
signature waveform.
A
m
: amplitude coefficient for frequency m in the Fourier series
rcqucncy
m
: frequency value chose to approach the measured signature
waveform.
t: time

m
: phase for frequency m
Jccoy octor: factor related to the attenuation energy in that particular
monitoring point.
To introduce the variability hole-to-hole in the signature waveform, there are
three parameters where it was assumed to follow a random normal distribution behavior;
63

they are the amplification scale factor, frequency content of the signal and decay factor.
The formulation is as follows:

ASF
m
= ASF
m

+ronJn StJ(ASF
m
)

rcqucncy
m
= rcqucncy

+ronJn StJ(rcqucncy)

cc
]uct
= cc
]uct

+ronJn StJ(cc
]uct
)
[4.48]
Where:

ASF

, rcqucncy

, cc
]uct

: mean values for the

parameters in Equation 4.47

StJ(ASF), StJ(rcqucncy), StJ(cc
]uct
): standard deviation for the
parameters in Equation 4.47

ronJn: pseudorandom values drawn
from the standard normal
distribution

Using Equation 4.47 and Equation 4.48, a random signature waveform is
generated for each hole in the modeling process of the complete waveform from a
specific production blast. In vibration modeling using the proposed approach, there are
two scenarios; the first one where there is only one record for the signature waveform. In
such case, it is necessary to assume all the statistical parameters in Equation 4.48.

If there are not statistical values for Equation 4.48, it is possible to assess at least
one value for ASF, rcqucncy onJ EC
]uct
from the signature signal. To proceed to
use the current proposal, a standard deviation equals to one third of the estimated value
for ASF onJ EC
]uct
is proposed. On the other hand, for the frequency, it is proposed to
assume a standard deviation value equal to the lower boundary of the zone where the
main frequency for that zone is present. This concept for the assumption of the frequency
is explained in Figure 4.19.











64




















Figure 4.19 Frequency parameter for Equation 4.48

In Figure 4.19, there are four zones and four main frequencies. In this case, the
statistical parameters for frequency parameter in Equation 4.48 are given by:

rcqucncy
1
=
1
+ronJn (
1
-Iowcr BounJ
1
)
rcqucncy
2
=
2
+ronJn (
2
-Iowcr BounJ
2
)
rcqucncy
3
=
3
+ronJn (
3
-Iowcr BounJ
3
)
rcqucncy
4
=
4
+ronJn (
4
-Iowcr BounJ
4
)
[4.49]
Regarding to the other parameters in Equation 4.48, in the case that only one
signature is available, they are given by:

ASF = ASF +ronJn (
ASF
S
)

cc
]uct
= cc
]uct
+ronJn (
cc
]uct
S
)
[4.50]
When more than one signature is available, it will be possible to evaluate the
parameters using common statistics to assess the values in Equation 4.48. Next an
example using test 01 described previously, is included to illustrate the procedure and the
results in both case scenarios.
65

4.3.1.4 Numerical example random signature waveform generation
In this example, assume that only signature one was measured for the point of
interest (scenario one). As previously explained, the values for the parameters are:

Number of Frequencies: 4
Values: 6.68Hz, 10.70Hz, 13.74Hz and 22.72Hz
Amplitude Scale Factor: 4.64
Decay Factor: 3.35

The statistical parameters for the frequencies are assessed using Figure 4.20 are:

















Figure 4.20 Numerical example frequencies signature 1 test 01

rcqucncy
1
= 6.68 +ronJn (2.67)
rcqucncy
2
= 1u.7u +ronJn (1.SS)
rcqucncy
3
= 1S.74 +ronJn (1.71)
rcqucncy
4
= 22.72 +ronJn (2.67)
ASF = 4.64 +ronJn (1.S4)
cc
]uct
= S.SS +ronJn (1.11)
[4.51]
Using the values of Equation 4.51 in to Equation 4.47, it is possible to generate
random signals for each hole in a production blast event (for example 11 holes in this
test). Figure 4.21 show a comparison between the three signature signals measured in
test 01 and eleven signatures generated using the current approach. Notice that the
parameters of only one signal was used to generate the random signals; however a set of
signatures that envelop the three measured signatures is generated



66
























Figure 4.21 Random generated signatures using Silva-Lusk equation.
Two case scenarios were mentioned before. In the second case scenario, more
than one signature is available to proceed with the modeling of signatures. Using the
three signatures measured in test 01 and evaluating the statistical parameters (despite only
three values for each parameter are available) we have in Table 4-1:

Table 4-1 Statistical parameters to modeling random signals
Signature Frequencies (Hz) ASF Decay
1 6.68 10.70 13.74 22.72 4.46 3.35
2 7.01 - 13.05 22.09 5.81 3.10
3 6.71 10.74 13.42 30.20 4.50 3.19
Mean 6.80 10.72 13.40 25.00 4.92 3.21
Stdev 0.77 0.13

Regarding the standard deviations for the frequencies, to introduce more
variability to the signals and cover a more reasonable assessment of possible outcomes, it
is necessary to keep the low boundary value.




67






















Figure 4.22 Signature signals measured vs Generated using one and three signals
In this case only three signals are used to estimate the statistical parameters for
Equation 4.47 so the difference is not so evident. The benefits of using random
signatures for each hole (random but with some boundaries) will be explained in chapter
5 and 6. An explanation about the introduction of the statistical parameters in the timing
explosion sequence is included in the following section.

4.4 Wave arrival time distribution and time sequence
4.4.1 Wave arrival time
In Figure 4.23, if hole (i) is blasted, vibration waves take to travel from hole (i)
to the monitoring point a time given by Jt
s
. It is usual in a production blast to use a
delay time between holes. In Figure 4.23 delay time between holes is given by Jt
]
and
the vibration waves generated by hole (]) take to travel from hole (]) to the monitoring
point a time given by Jt
]s
. If the detonation of the hole (i) is the reference, the vibration
from hole (]) is going to be recorded at the station at a time given by Jt
]
+Jt
]s
.
In this research, the traveling time of the waves from the source to the monitoring
point are introduced in the model through the compressional or shear wave propagation
velocity according to the vibration component that is going to be modeled. If the blast
vibration component to model is the longitudinal component, the wave propagation
velocity to use is the p-wave velocity. If any other component of the vibration is needed
(vertical or transverse), s-wave velocity should be used.


68

Hole (i) Hole (j)
dt
ji
dt
js
dt
is
time
Energy
Hole (i) Hole (j)
dt
is
dt
ji
dt
js
+
dt
j

















Figure 4.23 Delay times involved in the signature analysis

In the Figure 4.23 we have:
dt
is
vibration travel time between hole i and station S.
dt
ji
delay time between hole j and hole i.
dt
js
vibration travel time between hole j and station S.
dt
j
total delay time between holes in the sequence for the hole j

The best practice to assess the numerical value of the wave propagation velocity
is through field tests similar to those used in earthquake engineering to measure the
dynamic properties of the rock and soil. There are different methods between them:
Seismic reflection
Seismic refraction
Those methods are based on the general physics equation for velocity:

: =
x
t

[4.52]
Where:
x: distance source receiver
t: arrival time

Table 4-2 contains typical rock velocities for some of the rocks existing in
Appalachian region.



69


Table 4-2 Typical rock velocities (from Bourbi, Coussy, and Zinszner,
Acoustics of Porous Media)





























In this research, assume a normal distribution for the wave velocity and 10% of
the main value as standard deviation in order to calculate the traveling time between the
hole and the station or monitoring point. The statistical parameter for the wave velocity
is given by:

: = : +ronJn (u.1 : )
[4.53]
In order to estimate the traveling time, using Equation 4.52 and Equation 4.53, it
is obtained for the time:

Jt
ns
=
x
: +ronJn (u.1 : )

[4.54]
70

Where:
Jt
ns
: traveling time between hole n and station or measuring point s
x: distance between hole hole n and station or measuring point s
: : wave velocity, assumed or measured
ronJn: pseudorandom values drawn from the standard normal distribution

In the situation that field measurements for wave velocity are available, it is not
necessary to assume any parameters and the measured statistical parameters can be used
in Equation 4.53 and Equation 4.54.

4.4.2 Blasting time sequence
The blasting sequence depends on the initiation devices used to initiate the
explosives. Two well known initiation system devices are commonly used; electronic
and non-electric. To establish the statistical parameters to use in the current vibration
prediction methodology using signature hole techniques and Monte Carlo schemes, the
accuracy of the two initiation systems were tested. In total, 674 detonators were tested.
Each system (electronic and non-electric) was tested over the viable ranges of delays
available.

4.4.2.1 Experimental setup
To collect the information from the tests, a National Instruments PCI-6602
counter-timer card along with a custom software application developed in LabView was
used. The selected PCI-6602 counter-timer card from National Instruments was an
expansion card for use with personal computers. It included eight 32 bit counter channels
and 32 configurable digital IO lines. With the onboard clock running at 80 megahertz, it
was capable of measuring events down to 6.25nanoseconds (6.25e-006 ms) making it
well suited for this testing. Six of the channels were configured for monitoring break
wires. A seventh channel was used to monitor the control signals coming from the
electronic initiator blasting machine.
To monitor the tested detonators, the break wire principle was used. The main
idea about the break wire is that at the moment of detonation, the break wire is severed,
causing a loss of continuity through the wire. The counter-timer card triggered on this
event and reported a detonation time. It was also necessary to determine the zero
reference time for the event, or the time at which the detonator was initiated. The
difference between the measured detonation time and the zero reference time represented
the realized delay achieved by the initiator.
The channels on the counter-timer card relied on two signals for measuring time.
The first was the counter source, which was connected to a known internal timebase. It
produced an 80 megahertz signal. The hardware counted every occurrence of a rising
edge produced by the clock. The gate was the other important input used when
performing period measurements. The gate signal determines when the hardware should
report a count to the software application. This is demonstrated in Figure 4.24. The arm
start trigger shown was used to determine when the counter started counting. It counts
every edge received from the internal timebase. The gate was connected to the break
71

wire. Once continuity was lost and that signal went low, the gate was asserted and the
appropriate output was captured.










Figure 4.24 Illustration of Interaction between Break Wire and Counter

For the electronic detonator systems, the blasting machine communicated with the
detonators using a low frequency AC signal. The necessary commands to program the
detonator timing, arm them, and detonate them were sent via this signal. It was observed
that communication is ceased prior to detonation, presumably because a fire signal has
been transmitted to the detonators. This break in the signal was used to determine a zero
reference time for the application to calculate the achieved delay timing. Figure 4.25
demonstrates this as realized in the counter time hardware. The arm start trigger,
common across all counter channels, signaled the hardware to begin counting. It is
important to note that all counters began counting at the same moment due to the arm
start trigger, so they are effectively synchronized. Every falling edge on the signal
generated from the blasting machine was captured at the gate. The corresponding count
values were reported to the output for processing in the software application.









Figure 4.25: Blasting Machine Counter Interaction
For non-electric detonator systems, the zero reference time was determined in a
similar fashion to the detonation time. The detonators being tested were connected via a
bunch block. The bunch block ensured that the detonators under test shared a common
start time. A break wire positioned within the bunch block captured this time which
served as the zero reference time. This can be seen in Figure 4.26. The detonators being
tested can be seen as the shock tube leading from the bunch block to the right of the
figure (orange tubes). The detonator used to initiate them can be seen as the shock tube
leading into the bunch block from the left of the figure (yellow tube). The small wires
72

(green wires) are the break wire which were positioned with and directly adjacent to the
initiator.











Figure 4.26: Bunch Block Configuration
Once the hardware captured the detonation and reported the corresponding
counts, it would convert this value to time. Due to the synchronization, it is simply a
matter of subtracting the two values and dividing the count by the frequency of the
internal timebase. This was accomplished once the measured counts were transferred to
the LabVIEW application.
The LabVIEW application was responsible for a number of activities. It provided
a graphical user interface for the person conducting the test (Figure 4.27). Fields were
included to record information pertinent to the testing. It also provided feedback to the
user to ensure proper operation.
The application was also responsible for controlling the hardware. It configured
the counter devices for the task. It controlled the various digital lines used to establish
the levels in the break wires and manage the reset-set latches. Finally, it set the arm start
trigger to synchronize the channels on the card. Another function was calculating the
times from the measured counts and accumulating those values. When convenient for the
user, it generated a report including the test times and summary statistics.
To validate the data collected by this system, a Blaster Ranger high speed camera
was used to document several of the tests from each detonator system using an
appropriate frame rate. This footage was manually reviewed and it was concluded that
the system was accurately collecting the times of detonation.













73
























Figure 4.27: Graphical User Interface test setup
Validation of the system was performed using images captured and analyzed from
high speed video data. For some tests, high speed video data was recorded, showing the
detonation process of the initiation system. Viewing the video on a frame-by-frame basis
allowed for the visual confirmation of when each detonator initiated relative to the others.
For timing analysis using high speed video, the first detonator initiating in a test sample
was considered time zero, with timing for each following detonator based on this
reference point. This was done since the true time zero could not be obtained from the
video data.
The use of high speed imaging was a relatively straight forward process. For each
data set, the video was recorded at a specified frame rate, varying from 1,000 frames per
second (fps) to as high as 16,000 fps. With the frame rate for each data set known, the
time from one frame to the next could be calculated. For example, recording at 4,000 fps
results in a time lapse of 0.25 millisecond (ms) from frame to frame. Therefore, if the
first detonator initiating is time zero and the following detonator is shown to detonate 10
frames later, it can then be calculated that the difference between the two detonators
initiating is 2.5 ms. This process was repeated for each subsequent detonator in the test
sample.
As mentioned previously, the detonators selected for testing consisted of two
electronic systems and two non-electric systems. When possible, different lots were
procured to provide a more representative sample. In total 674 detonators were tested.
Table 4-3 outlines the testing matrix.
74


Table 4-3 Detonator Matrix


The idea about the different delays used, was to include a wide range including
short and long timing to analyze the influence of the delay time in the accuracy of the
initiation system for both non-electric and electronic.
With the high degree of automation built into the testing apparatus, the
methodology proved to be fairly simple. The detonators to be tested were first loaded
into the test cell consisting of short sections of steel pipe. The steel pipe served the
purpose of deflecting the shrapnel away from adjacent test cells and directing it away
from the break wire leads. A bar running the length of the test cells had small holes
through which the detonators were placed (Figure 4.28).















Figure 4.28 Test Cells
The break wire used for all of the testing was Belden 30 AWG solid copper hook-
up wire with polyvinyl chloride insulation. The break wire was held firmly against the
tip of the detonator and secured with a piece of vinyl tape. Care was taken to ensure the
break wire was placed running through the center of the tip. This technique is displayed
in the next figure.
Manufacture
Delay(ms) 10 1000 8000 10 1000 8000
Lots 3 3 3 3 3 4
TotalDetonators 53 43 50 51 52 47
Manufacture
Delay(ms) 9 1000 1400 25 100 700
Lots 1 1 1 2 2 2
TotalDetonators 68 60 67 59 65 59
ElectronicDetonatorMatrix
ElectronicDetonatorA ElectronicDetonatorB
NonelectricDetonatorMatrix
NonelectricDetonatorA NonelectricDetonatorB
75











Figure 4.29 Break Wire Placement
Figure 4.30 shows detonators awaiting test in the test cells.



















Figure 4.30 Detonators Awaiting Test
The following image shows the break out box, housing the interface electronics,
with the break wires and blasting machine control wires attached (Figure 4.31).








76

Order
Numberof
Frames
Calculatedtimeusing
highspeedvideo
(ms)
Systemrecorded
time(ms)
5 121 30.25 30.57
3 68 17.00 16.82
1 0 0.00 0.00
4 105 26.25 26.13
2 8 2.00 1.76
Nonelectric4000fps



















Figure 4.31: Electronics Break Out Box
The only difference to note with the methodology, as it pertains to the non-electric
testing, consisted of where to place the detonator in the bunch block used to establish the
zero reference time. During much of the testing it was placed under the test cell. After a
number of misfires occurred the setup was changed. The bunch block was placed in a
galvanized trash can filled with sand and buried. After this change in the experimental
setup, no other misfires occurred.

4.4.2.2 Non-electric detonators results
The frame grabs shown in Figure 4.32 illustrate the detonation sequences for one
test sample of non-electric detonators, in this case five detonators filmed at 4,000 fps.
For this series, a detonation event is shown to occur at Frames 00, 08, 68, 105, and 121.
The frame prior to each event is also shown for comparison purposes. Table 4-4 provides
a summary of the frame number at which a detonation event occurred, the calculated time
using the given frame rate, and the time recorded by the testing system.

Table 4-4 Summary of Results for Non-Electric Validation Example








77


































Figure 4.32 Frame grabs from non-electric sample showing detonation sequence

It is important to note two temporal considerations when reviewing the visual
analysis. One is that each frame represents a window of time created by the shutter
speed. This window for a video shot at 4,000 fps is 0.25 ms long. An event shown in a
single frame could have occurred at any point in this window. The second consideration
is that the detonation event is not an instantaneous one. There is a variable amount of
time inherent with this process and how it is represented visually too is variable. This
can be seen when comparing Frames 68 and 105 in Figure 4.32. The breakwire in
immediate contact with the detonator would be a more accurate measure of this event.


78

NominalDelay(ms) 9 1000 1400 25 100 700
Numberof
detonatorsTested 68 60 67 59 65 59
DelayAverage(ms) 11.342 1125.501 1418.766 27.751 102.730 715.710
StandardDeviation 4.594 6.550 19.054 0.765 11.250 6.195
Maximum(ms) 15.756 1146.782 1462.381 29.304 123.193 730.575
Minimum(ms) 1.534 1114.704 1367.035 26.155 79.835 697.925
PercentError 26.023% 12.550% 1.340% 11.005% 2.730% 2.244%
NonelectronicDetonatorResults
NonelectronicDetonatorsA NonelectronicDetonatorsB
The results of the testing from the non-electric detonator systems are summarized
in Table 4-5 and graphically in Figure 4.33, Figure 4.34 and Figure 4.35.

Table 4-5 Summary non-electric detonator results









Normal distribution probability density function was chose to compare the results.
All the non-electric detonators results are included in Figure 4.33. In this figure, it is
clear, how the most precise delay time is presented at 25ms nominal delay, despite that
the average value for this detonator show a difference of 2.751ms when compared to the
nominal time delay (25ms), the precision of this delay is reflected in the lower value of
the standard deviation. Figure 4.33, also shows how the least accurate delay time is the
1000ms detonator, the difference between the nominal delay time and the average tested
is around 125.501ms.













Figure 4.33 Normal distribution, density function, non-electric detonators tested.
Next figures show the graphical results for short and long delay times.
79

















Figure 4.34 Normal distribution, density function, non-electric detonators at 9, 25
and 100 ms nominal delays.


















Figure 4.35 Normal distribution, density function for nominal delays, non-electric
detonators at 700, 1000 and 1400 ms.
80

Order
Numberof
Frames
Calculatedtimeusing
highspeedvideo
(ms)
Systemrecorded
time(ms)
2 11 1.38 1.48
3 18 2.25 2.26
4 22 2.75 2.27
5 26 3.25 2.76
1 0 0.00 0.00
Electronic8000fps
4.4.2.1 Electronic detonators results
The frame grabs shown in Figure 4.36 illustrate the detonation sequence for one
test sample of electronic detonators, in this case five detonators filmed at 8,000 fps. For
this series, a detonation event is shown to occur at Frames 00, 11, 18, 22, and 26. The
frame prior to each event is also shown for comparison purposes.

























Figure 4.36: Frame grabs from electronic sample showing detonation sequence
Table 4-6 provides a summary of the frame number at which a detonation event
occurred, the calculated time using the given frame rate, and the time recorded by the
testing system.

Table 4-6 Summary of Results for Electronic Validation Example






81

NominalDelay(ms) 10 1000 8000 10 1000 8000
Numberof
detonatorsTested 53 43 50 51 52 47
DelayAverage(ms) 9.950 1000.543 8003.375 9.987 999.804 7998.589
StandardDeviation 0.092 0.321 3.751 0.030 0.107 0.851
Maximum(ms) 10.201 1001.120 8015.625 10.052 999.954 7999.400
Minimum(ms) 9.816 999.960 7995.190 9.910 999.460 7995.800
PercentError 0.501% 0.054% 0.042% 0.130% 0.020% 0.018%
ElectronicDetonatorsA ElectronicDetonatorsB
ElectronicDetonatorResults
The results of the testing from the electronic detonator systems are summarized in
Table 4-7 and graphically from Figure 4.37 to Figure 4.40.


Table 4-7 Summary statistics electronic system










Normal distribution probability density function was chose to compare the results.
All the electronic detonators results are included in Figure 4.37. In this figure, it is clear,
how detonators from maker A are less accurate and precise when compared to maker B.
Also in this result, it is evident how the precision of electronic detonators decreases when
the delay time increases. It is evident when standard normal distribution is used to
compare results Figure 4.38 to Figure 4.40 (see how the shape of the standard normal
distribution change in those figures).



















Figure 4.37 Normal distribution, density function, electronic detonators tested.
82




















Figure 4.38 Normal distribution, density function, electronic detonators at 10ms
nominal delays.




















Figure 4.39 Normal distribution, density function, electronic detonators at 1000ms
nominal delays.

83


















Figure 4.40 Normal distribution, density function, electronic detonators at 8000ms
nominal delays.

Finally, when non-electric and electronic initiation systems are compared side by
side, at nominal delay time of 9 and 10ms respectively, Figure 4.41 is obtained.





















Figure 4.41 comparison 9ms and 10ms nominal delay (non-electric Vs electronic)
84

Figure 4.41 shows clearly the big difference between both initiation systems
regarding precision and accuracy.
In this research, the statistics for the initiation systems were taken in to account
assuming a random normal distribution, and using the founded parameters of mean and
standard deviation for both initiation systems.
The general equation used in the Monte Carlo scheme to predict vibrations levels
from mining blast and regarding to delay timing between holes is given by:

Jt
]
= Jt

+ronJn (o
t
)
[4.55]
Where:
Jt
]
: time interval between detonation hole (i) and hole (]).
Jt

: average delay timing, measured or assumed

o
t
: standard deviation of the normal distribution of the average delay
timing, assumed or measured
ronJn: pseudorandom values drawn from the standard normal distribution

Finally, the time used to perform the linear superposition is given by the time of
the arrival of the vibration wave plus the time interval between detonation holes. For
example and using Figure 4.23, the time of hole (]), using hole (i) as time reference is
giving by:
t
]
= Jt
]
+Jt
]s

[4.56]
Where:
t
]
: time for hole (]) reference to hole (i)
Jt
]
: time interval between detonation hole (i) and hole (]).
Equation 4.55
Jt
]s
: traveling time between hole (]) and station or measuring point (s).
Equation 4.54
4.5 Linear superposition and discrete convolution
As stated in section 2.33, signature hole technique is based on signals and system
theories. After several assumptions regarding the system to model, the response y|n]
(output) of one system can be calculated using the discrete convolution equation given
by:
y|n] = b|n -k]u|k] =
n
k=0
u|n -k]b|k]
n
k=0

[4.57]
Where:
y|n]: current output of the system (blast vibration prediction)
k, n: integer values in one sequence
85

b|n]: impulse response (signature hole); impulse response is the
response of the system under one input equal to one impulse
sequence o|n] or Dirac delta function. In blasting o|n] is assumed
like the detonation of one hole.
u|n]: arbitrary input sequence (delay pattern used in a production blast).

Discrete convolutions are algebraic equations and can be computed by direct
substitution (Chi-Tsong Chen, 2004). To show the numerical convolution procedure,
assume that a hole was blasted and one signature signal was recorded in a monitoring
point as indicating in Figure 4.42. In this example, to simplify the numerical example, the
signature signal was discretized using only the values indicated in red colors.











Figure 4.42 Pair impulse signature waveform
Using the mathematical notation, we have in the case of Figure 4.42 that:
u|n] = o|n] = {1]
b|n] = {u,2.62,1.S4, -2.uu, -u.9u,u.67, -u.62,u.27,u.uu]
[4.58]
Now assume that we will blast three holes using one unit time of delay, the
graphical representation is included in Figure 4.43









Figure 4.43 Three holes numerical example
The problem is to predict the vibration levels produced by three holes detonated
as shown in the previous figure. In such case the new input is given by;
u|n] = {1,1,1]. Using Equation 4.57 and computing manually each value of
y|u], y|1], y|2], etc. we have:

86

u|n] = {1,1,1]
b|n] = {u,2.62,1.S4, -2.uu, -u.9u,u.67, -u.62,u.27,u.uu]

n =
y|u] = b|u]u|u] = u 1 = u
n = 1
y|1] = b|1]u|u] +b|u]u|1] = 2.62 1 +u 1 = 2.62
n = 2
y|2] = b|2]u|u] +b|1]u|1] +b|u]u|2] = 1.S4 1 +2.62 1 +u 1 = S.96
n = 3
y|S] = b|S]u|u] +b|2]u|1] +b|1]u|2] +b|u]u|S]
= -2 1 +1.S4 1 +2.62 1 +u u = 1.96
n = 4
y|4] = b|4]u|u] +b|S]u|1] +b|2]u|2] +b|1]u|S] +b|u]u|4]
= -u.9u 1 -2 1 +1.S4 1 +2.62 u +u u = -1.S6
Following this procedure we have:
y|S] = -2.2S ; y|6] = -u.8S ; y|7] = u.S2 ; y|8] = -u.SS ;y|9] = u.27
y|1u] = u

So the result of three holes detonated at one delay time unit is:
y|n] = {u,2.62,S.96,1.96, -1.S6, -2.2S, -u.8S,u.S2, -u.SS,u.27,u]


The graphical result is included in Figure 4.44.















Figure 4.44 Result of three holes detonation mathematical convolution
The implementation of the convolution algorithm is quite simple. The basic code
in MS Visual Basic VB is following:



87










Figure 4.45 Implementation of convolution in VB

Also, Matlab use a built in function called con: function, the syntax is quite
simple as follows:








Figure 4.46 Convolution in Matlab
If a more fine discretization is made, for example half of the time interval unit is
used, using matlab, the graphical results are included in Figure 4.47 and Figure 4.48.











Figure 4.47 Convolution using Matlab


In this case, notice vector u increases the number of elements to keep holes
blasted at one time unit, (in this example the holes are blasted using a delay of one time
unit in Figure 4.43).


for i=0 to n-1
for k=0 to i
a= u(i-k)*h(k)
aa=aa+a
next k
y(i)=aa
next i
u=[1,1,1]
h=[0,2.62,1.34,-2.00,-0.90,0.67,-0.62,0.27,0.00]
y=conv(u,h)
y =[0.00, 2.62, 3.96, 1.96, -1.56, -2.23, -0.85, 0.32, -0.35,
0.27, 0
u=[1,0,1,0,1]
h=[0.0000,1.6062,2.6096,2.5160,1.3422,-0.6509,-2.0284,-
1.9028,-0.9096,0.3524,0.6721,-0.1698,-0.6188,-
0.1544,0.2745,0.3161,0.0000]
y=conv(u,h)
y =[0.00,1.61,2.61,4.12,3.95,3.47,1.92,-0.04,-1.60,-2.20,-
2.27,-1.72,-0.86,0.03,0.33,-0.01,-0.34,0.16,0.27,0.32,0.00]
88













Figure 4.48 Convolution results using matlab and a more discretized signal.

Mathematically, the discrete convolution is an operation between two finite
sequences u|n] and b|n]. If there is no changes in the seed waveform, using the previous
algorithms or functions in Matlab it is possible to calculate the predicted vibration
signal for a production blast. However, if the seed waveform changes hole to hole, it is
necessary to use another approach to sum the different seeds waveform and them
calculate the predicted vibration signal.
Due to the increase in the storage and computation capacity of modern computers,
a simple graphical sum term by term is proposed to calculate the predicted signal. Using
the previous example, the seed signals should be shifted in time according to the blasting
pattern as indicated in Figure 4.49.

















Figure 4.49 Shifted signals to perform the sum.
After shifting the signals, and performing the sum for a specific instant of time it
is possible to calculate the vibration level for that instant of time. So the output for the n
term is given by:
89

y|n] = b
k
|n]
N
k=1

[4.59]
Where:
N: total number of signals to sum (equal to the number of blast holes)
b
k
: seed waveform finite sequence for hole k

Following the numerical example, in Figure 4.49 for 2.5 s, the vibration level is
calculated as:
y|2.S] = -u.6S +2.S4 +1.S8 = S.47
[4.60]
This is the same numerical value calculated using convolution. In this research,
the graphical procedure was used to calculate the predicted signal.

4.5.1 Practical example
Next, using information collected in a blasting test, the results to calculate the
vibration prediction using the convolution and the graphical methodology is compared.
During the development of the current research, several tests were performed at Guyan
mine in West Virginia. The mine is a typical surface coal mine using contour mining
methods. One of the tests was performed using electronic detonators and measuring a
signature waveform followed by the complete blasting. Table 4-8 shows the main
parameters of the blast test.

Table 4-8 Blast test parameters
Parameter Value
Total holes 69
Face height (ft) 44
Depth (ft) 44
Burden = spacing (ft) 20
Diameter (in) 9
Total explosive (lbs) 55,050.03

The timing delay used was 4 ms between holes detonated at the same time. A
seismograph was setup at 900ft from the blast site. The records of three vibration
components were recorded and they are included in Figure 4.50. The particular
characteristic of this test is the use of four holes detonated at the same time (to record the
signature or seed waveform and during the complete production blast).





90




















Figure 4.50 Vibration record for test blast 4 holes detonated at the same time
A plan view from the blast report is included in Figure 4.51.


















Figure 4.51 Plan view from the blast report for the test
In total to perform the convolution, the finite sequence representing the blast
pattern u|n] vector will be composed by a vector with seventeen elements (17 ones) if a
discretization of 4ms is made in time. Figure 4.52 shows the radial signature waveform
isolated from the complete record included in Figure 4.50.
91


















Figure 4.52. Radial signature (seed) waveform

Using a discretization of 1 ms, and generating two vectors of two columns (first
column time, and second column values) containing the signature waveform
(Signature.txt) and the timing sequence (timing.txt), the command to load such vectors in
Matlab is given in Figure 4.53. The first ten elements of vectors b, u and y are
included in Figure 4.53





















Figure 4.53 Matlab command to load signature and timing vectors
load Signature.txt;
t=Signature(:,1);
h=Signature(:,2);
load timing.txt;
u=timing(:,2);
Vector u Vector h Vector y
time Value time Value time Value
0 1 0 0 0 0
0.001 0 0.001 -0.02 0.001 -0.02
0.002 0 0.002 -0.02 0.002 -0.02
0.003 0 0.003 -0.02 0.003 -0.02
0.004 1 0.004 -0.02 0.004 -0.02
0.005 0 0.005 -0.02 0.005 -0.04
0.006 0 0.006 -0.02 0.006 -0.04
0.007 0 0.007 -0.02 0.007 -0.04
0.008 1 0.008 -0.02 0.008 -0.04
0.009 0 0.009 -0.02 0.009 -0.06
92

Now using the convolution built in command in Matlab given by; y =
con:(u, b), the graphical result is included in Figure 4.54.


















Figure 4.54 Convolution for radial component
The graphical procedure to calculate the predicted vibration waveform using the
signature waveform is included in Figure 4.55 for the first three signals representing the
blast of twelve initial holes and the last signal in this test.



















Figure 4.55 Graphical procedure to calculate vibration waveform in test
93

When all signals are added together, it is obtained the waveform included in
Figure 4.56.



















Figure 4.56 Final waveform after sum 17 signals using graphical procedure
Alpha-Blast software from White Industrial Seismology Inc. was used to calculate
the complete waveform of the blasting test in order to compare previous results with the
results of one commercial software already tested by the mining industry,. The simulated
waveform is included in Figure 4.57.



















Figure 4.57 Final waveform simulation using commercial software Apha-Blast
94

The measured waveform for this test and the different simulations are included in
Figure 4.58.




















Figure 4.58 Complete waveform comparison using different methodologies.
Figure 4.58 shows that there is no difference between mathematical convolution
and the graphical procedure. On the other hand, there is some small difference between
graphical procedure and Alpha-Blast software. However, since Alpha-Blast is
commercial software and there is no access to the code, it is not clear what methodology
and filtering process is performed by the software in the background that can give a
different result.
In the proposed Monte Carlo method, since for each hole a signature waveform is
generated randomly for each hole, the summation of the signals is done using the
graphical procedure explained previously. Notice that, if a standard convolution were
used, there is no chance to combine different waveforms.

The field test performed at Guyan mine are described in the next sections. The
results of the blast test are compared to simulations when the improved signature hole
technique is introduced.

95

Chapter 5
IMPROVED SIGNATURE HOLE TECHNIQUE VALIDATION
5.1 Introduction
To validate the proposed improved signature hole technique, several field tests
were performed at a surface coal mine in West Virginia. Validation of the methodology
was achieved through analysis of blast vibration signals recorded at the mine for different
experimental setups. Description of the site, instrumentation setup, designed tests and
recorded data are included next.

5.2 Field Experiments
5.2.1 Instrumentation and data collection
5.2.1.1 Site description
Guyan mine is located in southern West Virginia, in Logan County. The site was
chosen according to the matching in-kind contribution for the project offered by OSM,
Patriot Coal Corporation and the University of Kentucky. This operation is a typical
surface coal mine. The mine utilizes blasting, truck and shovel/loader machines to
perform the contour mining method at the site. The coal is sourced from the Freeport,
Kittanning, Stockton and Coalburg seams, with a 15 to 1 average overburden coal ratio.
(Source: Patriot Coal Corporation). Figure 5.1 shows the location of the site where the
information was collected.



















Figure 5.1 Location of the mine where the field experiments were conducted
96

Stratigraphic units present within the area include the Homewood Sandstone,
multiple splits of the Stockton Coal seams, Upper Coalburg Sandstone and the Coalburg
Coal seam. The overburden where the blasting activity took place mainly is the Coalburg
Sandstone which is a massive Sandstone and ranges in thickness from about 70 to 100
feet. Figure 5.2 shows a simplified stratigraphic column in the area where the blasting
activity took place during the collection of information.























Figure 5.2 Drill Hole GY 9411, stratigraphic column

5.2.1.2 Instrumentation
The objective of the instrumentation was the measurement of the environmental
effects of the production blast; specifically blast vibrations and air blast for several blast.

The instrumentation was performed in two different stages according to the frame
of time where the information was collected. The first run of collection of information
was between summer and fall 2010 and the second stage occurred in summer 2011.
Next, the most important information regarding the instrumentation during each period of
time is described.

Instrumentation for summer and fall 2010
For this period of time, in a first approach, seismograph locations were planned to
follow a radial pattern having as center the Drill hole GY9411 (Figure 5.1). However
97

after three site visits to verify the site conditions for the seismographs, it was necessary to
perform several modifications. The modifications were necessary either due to access
difficulties to some places or because the seismograph location planned were located in
areas outside of the property boundary of the mine. In order to protect the seismographs,
it was necessary to adapt tool boxes to contain the seismographs. An external battery was
used to extend the internal battery of the seismograph. Figure 5.3 shows the modified
tool box to contain and protect the seismographs.



















Figure 5.3 Case for the seismograph setup
In total, during this first run of collection of information, 12 seismographs were
installed in the area under study. Table 5-1 includes the description of the seismographs
and the coordinates in both systems NAD 83 and NAD27.
Three of the seismographs (2,3 and 4) were NOMIS 5400 while the others nine
were White Industrial MINI-SEIS series. Some of the seismographs were supplied by
WVDEP, OSM, Dep Mines Minerals and Energy and UKY. All the devices were
calibrated by the original provider of the seismographs, before the data collection
activity.
The seismographs setup was designed to keep a radial pattern reference to the
centroid of the exploitation area for the years 2010-2011. This centroid was located
coinciding with the geological drill hole GY9411. Finally the arrangement of the
seismographs was completed following three well defined lines. The lines were named
according to the lines orientation as North, East and South. Figure 5.4 show the pattern
followed to setup the seismographs. North line (seismographs 9,10,11 and 12), East line
one (seismographs 2,3,4) at the bottom of the valley, East line two (seismographs 5 and
6) at the top of the valley over reclaimed area and South line (seismographs 7 and 8).



98

SEIS DESCRIPTION Owner
E W X Y
1
MINI-SEIS II
Inst#MS II
2D2G 1/4M
S/N: 4763
37.827306 81.79722222 1769695.032 302194.887 UK
2
NS 5400 S/N:
2722
37.829722 81.7926667 1771018.239 303063.285 Saul's
3
NS 5400 S/N:
2242
37.829306 81.78955556 1771915.431 302904.128 Saul's
4
NS 5400 S/N:
2774
37.828472 81.78608333 1772915.656 302591.914 Saul's
5
MINI-SEIS II
Inst#MS II
2D2G 1/4M
S/N: 4762
37.827417 81.791 1771492.406 302219.877 UK
6
MINI-SEIS II
Inst#MS II
2D2G 1/4M
S/N: 3599
37.82575 81.78622222 1772867.112 301601.13 Ken - OSM
7
MINI-SEIS
Inst#MS 2D2G
S/N: 429
37.819056 81.79911111 1769123.556 299195.638 UK
8
MINI-SEIS
Inst#MS 2D2G
S/N: 2832
37.814444 81.79691667 1769742.925 297510.868 WVDEP
9
MINI-SEIS
Inst#MS 2D2G
S/N: 2467
37.834833 81.80225 1768266.722 304948.143 WVDEP
10
MINI-SEIS
Inst#MS 2D2G
S/N: 2468
37.8385 81.80044444 1768799.694 306278.84 WVDEP
11
MINI-SEIS
Inst#MS 2D2G
S/N: 180
37.841389 81.79933333 1769129.628 307327.998 UK
12
MINI-SEIS II
Inst#MS II
2D2G 1/4M
S/N: 1513
37.846167 81.79777778 1769593.786 309063.868
Dep Mines,
Minerals and
Energy. Big
Stone GAP VA
NAD 83 NAD 27
Table 5-1 Seismograph location and their characteristics




























Initially in the North line it was planned that all devices were at the same
elevation (seismographs 9 to 12) but when it was implemented in field, many access
problems arose. It was not practical to setup those devices at the same elevation.














99






























Figure 5.4 Seismographs location
The trigger levels of the seismographs were set after several tests to guarantee the
collection of the information in all the points under study. In the adjustment of the
trigger levels, the proximity of the seismographs to roadways to prevent false triggers,
topographic conditions of the site, the distance from the source to recording point and
others factors that affect the expected levels of airblast and vibrations were considered.
Table 5-2 includes the final arrangement for the seismographs trigger levels. The
table includes the seismograph number, the distance from the drill hole GY9411, which
was taken as the exploitation area centroid, the elevation and the trigger levels. It should
be noticed that distances seismograph-blast changed while the blasting activity was
developed through the collection of the information, so distances in Table 5-2 should be
taken as one initial reference.





100

Table 5-2 Seismographs triggering parameters
Trigger parameters
Seismograph Distance
(ft)
Elevation
(ft)
Particle
velocity
(in/s)
Airblast
(dB)
Duration
(s)
Samples/second
1 667.0 1900 0.08 122 10 1024
2 2242.0 1300 0.05 118 10 1024
3 2286.8 1350 0.05 118 10 1024
4 3859.6 1355 0.05 118 10 1024
5 2404.8 1770 0.05 118 10 1024
6 3763.8 1797 0.05 118 10 1024
7 2675.4 1800 0.08 122 10 1024
8 40405.4 1900 0.08 122 10 1024
9 3191.3 1800 0.08 122 10 1024
10 4418.9 1910 0.03 120 10 1024
11 5457.1 1400 0.03 120 10 1024
12 7208.9 1200 0.03 120 10 1024

Initially it was planned to attach some of the seismographs to rock. Due to the
geological conditions of the area, where the layer of soil is more than 3 feet thick, it was
not possible to fix the geophones to the rock. All records collected were representative of
vibration records in soils where the houses in the area are usually found. Finally, to setup
the devices, all the field practices guidelines for blasting seismographs were reviewed
and applied. (ISEE Field Practice Guidelines For Blasting Seismographs 2009 Edition).
Instrumentation parameters caused a reduction in the collected number of events
for some of the seismographs. Continuous false triggers (based on low trigger levels and
high ambient vibration) deactivated the capacity of the device and only allowed recording
of peak values instead of the vibration trace. This was especially true for seismograph 1
and seismograph 5 and 6. Considerable quantities of false triggers in seismograph 1 were
due to the proximity of this seismograph to the blasting area. Machinery activity
provided additional seismic input close to this point. Seismograph 1 was also lost for a
short period because a dozer buried the device.
Seismographs 5 and 6 were in an area where significant animal activity was
occurring, perhaps close to a trail path for deer and bears. Frequently the boxes
containing seismographs were founded lying down, and had teeth marks. Despite these
problems, the data base contained enough information to analyze regarding vibrations
when the delay system is non-electric or electronic.

Instrumentation for summer 2011
Following analysis of the information collected in 2010, it was decided to perform
some specific blast tests in summer 2011. The second seismograph instrumentation setup
took centroid the mining exploitation area for summer 2011 and was performed at the
same ridge of the mine. The new centroid was located 5000 feet to the North measured
from the point of reference for 2010 tests (drill hole GY9411). Since after the 2010 tests,
101

all the seismographs were removed (for maintenance), it was necessary to setup a new
seismograph arrangement to collect the information for the second monitoring period of
time (summer 2011). In this second round of tests, five (5) seismographs were used.
The locations of the seismographs for the second round of test are included in
Figure 5.5. As reference in Figure 5.5, the drill hole GY 9411 is included as well as the
location of some of the seismographs used during the 2010 tests (locations in grey).
























Figure 5.5 Seismographs location for second round of test. (Summer 2011)
Only two seismograph locations were kept similar during the two periods of
collection of information, seismograph 9 (2010 test) and seismograph 3 (2011 test) and
the seismograph installed in the backyard of the house that belongs to the mine
(seismograph named as House). This last seismograph was located in that place by
request of the blasting crew because some of the blasting tests used delay timing less than
8 milliseconds.
Figure 5.6 shows the NOMIS 5400 system used to collect vibration and airblast
information in fall 2010.







102














Figure 5.6 NOMIS 5400 System used to collect blast vibrations and airblast

Table 5-3 includes the characteristic of the triggering used to setup the
seismographs in the second round of test.

Table 5-3 Triggering levels used in the second round of test
Trigger parameters
Seismograph Distance
(ft)
Elevation
(ft)
Particle
velocity
(in/s)
Airblast
(dB)
Duration
(s)
Samples/second
4906 (1) 691.9 1850 0.01 148 12 1024
3857 (2) 2410.1 1200 0.01 142 12 1024
4762 (3) 1937.8 1800 0.03 148 12 1024
3599 (4) 1348.1 1500 0.01 142 12 1024
180 (House) 3278.3 1200 0.01 148 12 1024


5.3 Tests description
The complete data base of vibrations collected during this project is composed by
200 events. However, only those tests where at least one signature hole was recorded are
described next. This is because for those events, it is possible to use the proposed
methodology and compare the prediction versus the vibration waveform recorded for the
complete blast event. Table 5-4 summarizes the tests including the main characteristics
like number of holes, depth of the holes diameter etc.





103

Table 5-4 Tests including signature hole
Test Date Holes
Depth
(ft)
B
(ft)
S
(ft)
Det
Total
explosive
Main
Delay
Signature
1
09/10/201
0
25 30 18 18 Elect 10,450.21 8ms NO
2
09/11/201
0
29 95 9 9 Elect 2,125.57 1ms
Three
hole@sam
e time
(Pre-split)
3
09/11/201
0
66 30 18 18 Elect 21,160.37 80ms
4
09/15/201
0
194 90 20 20 Elect
342,763.9
3
4ms One hole
5
09/16/201
0
69 44 20 20 Elect 55,050.03 4ms
Four
Holes@sa
me time
6
09/17/201
0
41 30 18 18 Elect 11,069.31
100/42
ms
One hole
7
09/18/201
0
96 95 18 18 Elect
181,778.1
4
4ms
Two
holes@sa
me time
8
09/22/201
0
67 75 20 20 Elect 86,719.03 4ms
Two
holes@
same time
9
10/01/201
0
176 95 20 20 Elect
298,139.2
6
17ms One hole
10
06/22/201
1
11 45 18 18 Elect 5,928.63 5ms
Three
signatures
11
06/23/201
1
26 30 18 18 Elect 22,090.28
100/5
ms
Two
signatures
12
06/24/201
1
29 45 18 18 Elect 24,271.73 5ms
Two
signatures
13
06/29/201
1
32 45 18 18 Elect 26,039.31 3ms
One hole
signature
14
06/29/201
1
35 45 18 18
Non-
elect
30,106.45
42/100
ms
NO-
Signature
15
06/29/201
1
40 45 18 18 Elect 33,478.25
42/100
ms
NO-
Signature

In total, 15 field tests were used to prove the proposed methodology in this
research. Appendix A contains the blasting log report from the mine and the vibration
and airblast signals for each test.
The location of the holes in the last six tests, (performed in 2011) were controlled
using topographic survey of precision. The plan layout of those tests are included in
appendix B.
Figure 5.7 shows the plan layout for the test 06/24/2011 (test No.12). Red
indicates the hole number, in black the nominal delay used. In this test two signature
holes were recorded at 2500 and 7000ms. In total, the duration of this blast was 9.172
seconds. This test accounts for 29 holes blasted. The order of the number of holes as
104

0 10 20 30ft
N
Bl ast ed
Soli d
Soli d
06/ 24/ 2011 3: 20pm
Si gnat ur e Hol e
Bl ast ed
shown in Figure 5.8 is six (6) holes, one (1) signature, ten (10) holes, one (1) signature
and finally eleven (11) holes. This figure shows the radial, vertical and transverse
component of the vibration as it is recorded normally.


















Figure 5.7 Plan layout test 06/24/2011





















Figure 5.8 Vibration record for test No.12, seismograph 4906 (approx. 767ft from
source)
105

Figure 5.9 is obtained when the other seismographs are included in the graph
using only the radial component (just for convenience).






















Figure 5.9 Radial vibration component for test No.12 and all the seismographs in
summer 2011
In Figure 5.9 it is possible to see the attenuation of the vibration with distance. In
this figure, despite seis 3599 being closer than seis 4762, vibration levels are higher for
the seismograph further away (between 3599 and 4762). This situation may be the effect
of topographic influence on blast vibrations and the change in elevation between the
source and the monitoring point. Seismograph 4906, 4762 and the source are more or
less at the same elevation (1825ft) when compared to seismograph 3599 that is
approximately 300 ft below the source of the blast vibration (1500ft). When signature
signals are isolated from the complete record (red areas in Figure 5.9), it can be seen that
it is not possible to assess, in this case, a signature for the point located at 2375 and
3400ft away from the source point, this is because at such distances the vibrations had
been completely attenuated. In those cases, it is not possible to use signature hole
techniques, because no signature is available to calculate a prediction using this
methodology. Appendix C includes the vibration records for the other fourteen events
included in Table 5-4.

Results for airblast in test No.12 are included in Figure 5.10. This figure shows
the problems regarding the sensitivity of the seismograph. The signals are stepwise and
all signals include some level of noise (this is more evident in seismographs 4762 and
4906). This is an important factor to account when using signature hole technique to
predict airblast and ground vibrations because the quality of the prediction is directly
106

related to the quality of the signature signal. By definition, a signal carries information
that we are interested in. In the case of blast vibrations, the signal that is usually recorded
contains particle velocity information of the monitoring point under the effects of the
mining blast. Under that concept, noise is anything else in the signal. If a vibration
records contains noise, it is necessary to perform several pre-process steps before
convolution or superposition to avoid, filter or minimize the noise. If a noise signal is
used to perform the superposition, due to the numerical nature of the superposition, the
noise will propagate (like a propagating error).

When it is necessary to apply filters to the signal, some frequencies are eliminated
from the original signal; for example in low pass filters high frequencies are eliminated,
if high pass filter is applied to the signal the low frequencies are eliminated leaving the
high frequencies. The risk from the prediction point of view when filters are applied is to
eliminate information that we are interested in but it is eliminated when the signal is
filtered.


























Figure 5.10 Airblast records for test No.12 and all the seismographs in summer
2011


107

A good signature signal for airblast, in order to perform a prediction using the
signature hole technique, is included in Figure 5.11.

















Figure 5.11 Airblast record for Test No.5 (Four holes signature)
Validation for the methodology is completed in detail using tests No.5, 10 12, 14
and 15.
Those tests were chosen for several reasons; test No.5 had a good signature signal
for airblast. Tests 10 and 12 have similar geometry, depth, spacing, burden etc. In both
cases six holes were detonated at different average delay time, so using these two tests it
is possible compare the results when different delay timing is used. The two final tests
(test No.14 and 15) have similar geometry, and similar delay time 42 and 100 ms. The
difference between the final two tests was the initiation system used between non-electric
and electronic.

5.4 Analysis and results of the models using improved signature hole technique
5.4.1 Model 1. Airblast modeling (test No.5)
Test No.5 was the same test used to show the practical example in Chapter 4 to
calculate convolution using different methodologies. Next the main parameters of this
test are included again.
Table 5-5 Blast test parameters test No.5
Parameter Value
Total holes 69
Face height (ft) 44
Depth (ft) 44
Burden = spacing (ft) 20
Diameter (in) 9
Total explosive (lbs) 55,050.03
108




















Figure 5.12 Plan view from the blast report for the test No.5
The signature waveform for air blast was included in Figure 5.11 (initial part of
the signal). As mentioned before, in total 69 holes were detonated using four signatures
at the same time and using electronic delay of 4ms. Next, the assumptions for modeling
airblast using the improved signature hole technique are explained.

5.4.1.1 Signature test No.5
Figure 5.13 is obtained when the signature waveform is isolated from the
complete record in Figure 5.11. In this figure, the value of the exponential decay factor is
included. In this case, the value is 3.90, as indicated in Figure 5.13.













a) Waveform and decay factor b) Waveform in frequency domain
Figure 5.13. Airblast signature waveform test No.5
109

When the frequency content of the signal is reviewed, it is difficult to assess a
specific value to generate the synthetic signals. As was explained in Chapter 4, using
more coefficients in the Fourier series, result in more accurate simulation of the measured
signal. In this case, the first five frequencies were chosen to simulate the signature
waveform. Those frequencies are 1,2,3,4 and 5 Hz.
Using an amplification factor of 2.80, after running the software developed, the
synthetic signature waveform for the airblast is calculated. The results are included in
Figure 5.14.
In this model, in total 69 holes were detonated, however four holes were
detonated at the same time using 4 millisecond delay between sets of four holes. As it
was explained in chapter 4 and is included in Figure 5.12, it is possible to simplify
complete blast to 17 detonations (four holes each). In order to model the complete
airblast waveform, it is necessary to generate 17 different random signals using the
results from Figure 5.14.





















Figure 5.14 Airblast signature waveform and synthetic waveform (approach).

5.4.1.2 Synthetic signature signals
Using a lower and upper frequency interval of 0.50 Hz from the main frequencies,
17 signature airblast signals were generated. In the calculation, a scale factor average of
2.8, assuming a standard deviation of 0.50 was also used. Figure 5.15 shows the
numerical values used to perform the calculations as entered into the software developed.



110









Figure 5.15 Frequency interval and scale factor for test No.5

The results when 17 random signature airblast waveform signals are generated are
included in Figure 5.16. The random effects can include different quantity and quality,
contamination of explosives in each of the four holes, changes in the temperature and air
current effects between the source and the recording site that generate a change in the
waveform between detonations.
























Figure 5.16 Random airblast signature waveform signals test No.5

5.4.1.3 Timing sequence
The delay between each four hole set in this test was uniform at 4ms. In the
current validation example, there were observed no effects if timing due to distance
111

between the source and the monitoring point was included, so only timing due to
detonator accuracy is calculated. The first calculation was done using zero standard
deviation for detonators and the second one includes a standard deviation of 1ms as
corresponds with the statistical calculations previously included in Chapter 4 (a linear
variation for standard deviation with delay time was assumed to assess the standard
deviation for 3000ms detonators).






Figure 5.17 Delay time series for test No.5

The time function sequences for both scenarios are included in Figure 5.18. In
this figure, it is possible to see how the sequence is affected when the scatter in the timing
is included.
















a) Zero standard deviation b) one millisecond standard deviation
Figure 5.18 Time function for test No.5. Zero and 1ms standard deviation in
detonator
As mentioned before, no traveling time was included in the calculations.


5.4.1.4 Results Model 1
Zero standard deviation and one run
Figure 5.19 shows the result using one iteration (no-Monte Carlo analysis is
performed) and zero standard deviation for delay timing.

112



















Figure 5.19 Results using one iteration and zero delay standard deviation
In this case, all the calculations overestimate the maximum measured value for
airblast. This result is more evident in the case when only the measured signature airblast
waveform is used to perform the calculations (current signature hole technique).
One millisecond standard deviation delay and Monte Carlo analysis (improved
signature hole technique)
Using the improved methodology and the developed software, it is possible to
analyze all the variables that are involved in the problem independently. Figure 5.20
shows the screen view for the current problem.
















Figure 5.20 Variables involved in the prediction test No.5 using improved
signature technique.
113

Using the parameters previously mentioned regarding frequency content,
amplification factor, decay factor and delay time, Figure 5.21 is obtained for 50
iterations.




















Figure 5.21 Monte Carlo result using initial parameters for test No.5
As shown in Figure 5.21, the peaks are overestimated for this case. This is
because in the formulations, the same weight for the amplification factor has been
assumed for frequencies between 1 and 5 Hz. If we analyze the frequency content of the
signature airblast waveform (Figure 5.13b), only frequencies between 2 and 4 Hz
contribute in an important way to the energy of the signal. Following this idea and
assuming that the frequency of 3 Hz is the frequency that is more important for the
amplitude of the signal and running the Monte Carlo analysis, we obtain the results in
Figure 5.22.














114




























Figure 5.22 Monte Carlo results using 3Hz frequency as main frequency in test
No.5

Amplitude factor for frequencies 1,2,4 and 5 Hz were assumed equal to one and
amplitude factor for the main frequency of 3 Hz was assumed as previously mentioned of
2.80.
Using the maximum absolute values (the peaks), it is possible to create the
histogram included in Figure 5.23. In this test, an average value of 258 Pa and a standard
deviation of 18.42Pa were calculated using the improved signature hole technique. The
airblast measured in this test was -224 Pa. So the absolute value of the measured peak is
between two standard deviations from the mean calculated value.








115

















Figure 5.23 Histogram absolute maximum calculated values
In order to see the convergence of the improved signature hole methodology, the
evolution of the parameter estimated, in this case the mean of the airblast against the
number of samples or iterations, is included in Figure 5.24.

















Figure 5.24 Convergence plot for test No.5
Figure 5.24 is a graphical representation of the central limit theorem and also is a
measurement of the convergence of the mean of the peak particle velocity to certain
value. This figure also shows the minimum number of iterations that are required to start
to obtain a constant mean value for the peak particle velocity (in this particular case at
least 30 to 40 iterations).
116

0
10 20 30ft
N
Bl ast ed
Soli d
Soli d
06/ 22/ 2011 3: 15 pm
Si gnat ur e Hol e
5.4.2 Model 2. Particle Velocity (test No.10)
Test No.10 was done in summer 2011. Next the main parameters of this test are
included.
Table 5-6 Blast test parameters test No.10
Parameter Value
Total holes 11
Face height (ft) 45
Depth (ft) 45
Burden = spacing (ft) 18
Diameter (in) 7.875
Total explosive (lbs) 5,928.63













Figure 5.25 Test No.10, plan view from survey
In this test, there are three signature waveforms for the particle velocity.

















Figure 5.26 Vibration record seismograph 3599 test No.10
117

Seismograph 3599 was chosen to model the complete waveform for the six first
holes in the test. It is possible to model any component; however for simplicity only the
transverse component was used to compare the waveform prediction against the
measured.
5.4.2.1 Signature test No.10
Figure 5.27 is obtained when a signature waveform for the traverse component is
isolated from the complete record in Figure 5.26. In this test, there are three signature
waveform signals, however only signature waveform two was used for modeling. In this
figure, the value of the exponential decay factor is included. In this case, such value is
4.69, as indicated in Figure 5.27.
















b) Waveform and decay factor b) Waveform in frequency domain

Figure 5.27. Ground vibration signature waveform test No.10

In this case, it is possible to use at least four frequencies to approach the signature
waveform. Those frequencies are 4,7,13 and 23Hz.

Using an amplification factor of 5.25, after running the software developed, the
synthetic signature waveform is calculated. The results are included in Figure 5.28.
In this model, 11 holes were detonated, using different timing configurations.
The first six holes were detonated at 5 millisecond delay as indicated in Figure 5.25. In
order to model the complete vibration waveform for the initial part of the record, it is
necessary to generate 6 random signals.






118






















Figure 5.28 Signature waveform and synthetic waveform (approach) test No.10.

5.4.2.2 Synthetic signature signals
Using a lower and upper frequency interval as indicated in Figure 5.29, 6
signature ground vibration signals were generated. In the calculation, a scale factor
average of 5.25, assuming a standard deviation of 1.0 was also used.











Figure 5.29 Frequency interval and scale factor for test No.10

The results when 6 random signatures waveform generated are included in Figure
5.30. In this case, the random effects can include different path, cracks, different quantity
of explosives in each hole, etc.

119






















Figure 5.30 Random ground vibration signature waveform signals test No.10

5.4.2.3 Timing sequence
The delay between holes in this test was uniform at 5ms delay. In the current
validation example, the effects in timing due to distance between the source and the
monitoring point were included. This is because for this test the coordinates of each hole
and the monitoring point was obtained through a survey of the area before blasting. As
indicated in Chapter 4, the change in travel time for each hole depends on the seismic
wave velocity, in this case an average of 16,000ft/s was assumed as mean value and a
standard deviation of 4% of the mean value, in other words 640ft/s. Regarding the
standard deviation for the detonators, despite no testing was performed for this nominal
delay, 0.1ms was assumed because it is in the range of short times. Figure 5.31 shows
the values used in this validation example.







a) Detonators b) Traveling time
Figure 5.31 Time series for test No.10
120

The time function sequence including traveling time and hole delay time is
included in Figure 5.32. In this figure, it is possible to see how the sequence is affected
when the scatter in the timing is included.















Figure 5.32 Time function for test No.10. Including both timing parameters
5.4.2.4 Results Model 2
Traditional signature methodology
Figure 5.19 shows the result using one iteration (no-Monte Carlo analysis is
performed).




















Figure 5.33 Results using one iteration for test No.10
121

In this case, all the calculations underestimate the maximum measured value for
the vibration. The most close prediction value is reached using the measured signature
waveform, (current signature hole technique).

Monte Carlo analysis (improved signature hole technique) test No.10
Using the improved methodology and the developed software, it is possible to
analyze all the variables that are involved in the problem independently. Figure 5.34
shows the screen view for the current problem when all variables are included.






























Figure 5.34 Variables involved in the prediction test No.10 using improved
signature technique.
In this particular case, the complete record of six holes was available and it is
possible to perform a back analysis to establish the parameters of the signature waveform.
After a trial and error procedure, a standard deviation for the amplitude factor of 2 is used
as well as a standard deviation of 1 for the decay factor and 1000 for the wave velocity.
The results are included in Figure 5.35.
122


















Figure 5.35 Monte Carlo result using fixed parameters for test No.10
In this case, as shown in Figure 5.35, some peaks in the complete waveform are
overestimated.
Using the maximum absolute values (the peaks), the histogram included in Figure
5.36 is calculated. In this test, an average value of 0.242 in/s and a standard deviation of
0.035in/s were calculated using the improved signature hole technique. The peak
velocity in this test was of -0.32in/s. So the absolute value of the measured peak is
between 2.2 standard deviations from the mean calculated value.



















Figure 5.36 Histogram absolute maximum calculated values test No.10
123

In order to see the convergence of the improved signature hole methodology, the
evolution of the parameter estimated, in this case the particle velocity against the number
of samples or iterations is included in Figure 5.37.


















Figure 5.37 Convergence plot for test No.10

5.4.3 Model 3. Particle Velocity (test No.12)
Test No.12 also was done in summer 2011. Next the main parameters of this test
are included.

Table 5-7 Blast test parameters test No.12
Parameter Value
Total holes 29
Face height (ft) 45
Depth (ft) 45
Burden = spacing (ft) 18
Diameter (in) 7.875
Total explosive (lbs) 24,271.73


In this test in total 29 holes were detonated, this test had the same geometrical
characteristics than test No.10. In other words, the same spacing, burden, hole diameter
and explosives per hole. For modeling purposes only the first six holes are going to be
included in the calculations.


124

0 10 20 30ft
N
Bl ast ed
Soli d
Soli d
06/ 24/ 2011 3: 20pm
Si gnat ur e Hol e
Bl ast ed

















Figure 5.38 Test No.12, plan view from survey
In this test, there are two signature waveforms for the particle velocity.


















Figure 5.39 Vibration record seismograph 3599 test No.12
In order to compare two blast events where the geometry is the same and the
distance and delay time change, seismograph 3599 was chosen to model the complete
waveform for the six first holes in the test. The transverse component was used to
compare the waveform prediction against the measured.
125

5.4.3.1 Signature test No.12
Figure 5.27 is obtained when the signature waveform for traverse component is
isolated from the complete record in Figure 5.26. In this test, there are three signature
waveform signals, however only signature waveform two was used for modeling. In this
figure, the value of the exponential decay factor is included. In this case, the value is
4.69, as indicated in Figure 5.27.














a) Waveform and decay factor b) Waveform in frequency domain

Figure 5.40. Ground vibration signature waveform test No.12

It is possible to use four frequencies to simulate the signature waveform. Those
frequencies are 4.3, 6.43, 9.3 and 13.6Hz. As expected, these frequencies are similar to
those for test No.10 (4, 7, 13 and 23Hz). Also the value of the decay factor is similar,
this means that frequency content recorded at the monitoring point and decay factor are
site specific parameters.

Using an amplification factor of 4.67, after runing the software developed, the
synthetic signature waveform is calculated. The results are included in Figure 5.41.
In this model, in total 29 holes were detonated, using different timing
configurations. Here we are modeling the first six holes detonated at 5 millisecond delay
as indicated in Figure 5.38. In order to model the complete vibration waveform for the
initial part of the record, it is necessary to generate 6 random signals.










126























Figure 5.41 Signature waveform and synthetic waveform (approach) test No.12.

5.4.3.2 Synthetic signature signals
Using a lower and upper frequency interval as indicated in Figure 5.42, six (6)
signature ground vibration signals were generated. In the calculation, a scale factor
average of 4.67, assuming a standard deviation of 1.0 was also used.








Figure 5.42 Frequency interval and scale factor for test No.12

The results when 6 random signatures waveform generated are included in Figure
5.43. In this case, the random effects can include different path, cracks, different quantity
of explosives in each hole, etc.



127






















Figure 5.43 Random ground vibration signature waveform signals test No.12

5.4.3.3 Timing sequence
The delay between holes in this test was uniform at 5ms after 100ms. In the
current validation example, the effects in timing due to distance between the source and
the monitoring point were included using the coordinates for this specific test and the
same values for wave velocity used in the previous model. Regarding the standard
deviation for the detonators, a nominal delay, 0.1ms was assumed (shot time detonators).
Figure 5.44 shows the values used in this validation example.









a) Detonators b) Traveling time
Figure 5.44 Time series for test No.12

In Figure 5.45, it is possible to see how the sequence only for delay timing is
affected when the traveling time is included.
128














a) Time function detonator delay b) Time function detonator
delay + traveling time

Figure 5.45 Time function for test No.12. Including both timing parameters

5.4.3.4 Results Model 3 (test No.12)
Traditional signature methodology
Figure 5.46 shows the result using one iteration (no-Monte Carlo analysis is
performed).




















Figure 5.46 Results using one iteration for test No.12
129

In this case, the calculations using measured signature overestimate the maximum
measured value for the vibration. Using the synthetic waves, the prediction is
underestimated.


Monte Carlo analysis (improved signature hole technique) test No.12
Using the improved methodology and the developed software, it is possible to
analyze all the variables that are involved in the problem independently. Figure 5.34
shows the screen view for the current problem when all variables are included.






























Figure 5.47 Variables involved in the prediction test No.12 using improved
signature technique.
Similar to the previous case, the complete record of six holes was available and it
is possible to perform a back analysis, to establish the parameters of the signature
waveform that best reproduce the blast. After a trial and error procedure, a standard
130

deviation for the amplitude factor of 2 is used as well as a standard deviation of 1 for the
decay factor and 1000 for the wave velocity. The results are included in Figure 5.48

















Figure 5.48 Monte Carlo result using fixed parameters for test No.12
In this case, as shown in Figure 5.35, some peaks in the complete waveform are
overestimated.
Using the maximum absolute values (the peaks), the histogram included in Figure
5.36 is calculated. In this test, an average value of 0.203 in/s and a standard deviation of
0.042in/s were calculated using the improved signature hole technique. The peak
velocity in this test was of -0.145in/s. So the absolute value of the measured peak is
between 1.40 standard deviations from the mean calculated value.

















Figure 5.49 Histogram absolute maximum calculated values test No.12
131

In order to see the convergence of the improved signature hole methodology, the
evolution of the parameter estimated, in this case the particle velocity against the number
of samples or iterations is included in Figure 5.37


















Figure 5.50 Convergence plot for test No.10


5.4.4 Model 4 Pyrotechnic initiation (test No.14)
This test used pyrotechnic initiation delay system so, it was not possible to get a
signature waveform in the test as was obtained in previous tests. Next the main
parameters of test No.14 are included.

Table 5-8 Blast test parameters test No.14
Parameter Value
Total holes 34
Face height (ft) 45
Depth (ft) 45
Burden = spacing (ft) 18
Diameter (in) 7.875
Total explosive (lbs) 30,106.45


In this test, 34 holes were detonated. This test had the same geometrical
characteristics as previous tests.


132

0
10 20 30
ft
N
06/ 29/ 2011 3: 33 pm























Figure 5.51 Test No.14, plan view from survey
In this test, there are no signature waveforms for the particle velocity.



















Figure 5.52 Vibration record seismograph 3599 test No.14
133

Despite the possibility to model any component, the transverse component was
chosen.
5.4.4.1 Signature test No.14
In this case, the average between signature waveform for test No.12 and test
No.10 was used as signature in this model.



















Figure 5.53 Signature waveform for test No.14
The decay factor and the main frequencies for the signature waveform are
included in Figure 5.54.














b) Waveform and decay factor b) Waveform in frequency domain
Figure 5.54. Ground vibration signature waveform test No.14
134

In this case, it is possible to use five frequencies to simulate the signature
waveform. Those frequencies are 4.4, 6.8, 9.2, 13.19 and 18.79Hz. As expected, these
frequencies are similar to those for test No.10 and No.12. Also the value of the decay
factor is similar.

Using an amplification factor of 5.03, and after running the software developed,
the synthetic signature waveform is calculated. The results are included in Figure 5.55.
In this model, 34 holes were detonated, using different timing configurations. In
order to model the complete vibration waveform, it is necessary to generate 34 random
signals.

























Figure 5.55 Signature waveform and synthetic waveform (approach) test No.14.

5.4.4.2 Synthetic signature signals
Using a lower and upper frequency interval as indicated in Figure 5.56, 34
signature ground vibration signals were generated. In the calculation, a scale factor
average of 5.03, assuming a standard deviation of 1.0 was also used.




135









Figure 5.56 Frequency interval and scale factor for test No.14

The results when 34 random signatures waveform are generated, are included in
Figure 5.57. In this case, the random effects can include different path, cracks, different
quantity of explosives in each hole, etc.























Figure 5.57 Random ground vibration signature waveform signals test No.14

5.4.4.3 Timing sequence
The delay between holes in this test was the typical delay used in this particular
mine, 100 and 42ms. Regarding the standard deviation for the detonators, and using the
information for pyrotechnic delay systems, 6ms was used. Figure 5.58 shows the values
used in this validation example.

136











b) Detonators b) Traveling time
Figure 5.58 Time series for test No.14
The time function sequence for delay timing and travelling time is included in
Figure 5.59 In this figure, it is possible to see how the sequence only for delay timing is
affected when the traveling time is included.















b) Time function detonator delay b) Time function detonator
delay + traveling time

Figure 5.59 Time function for test No.14. Including both timing parameters
5.4.4.4 Results Model 4 (test No.14)
Traditional signature methodology
Figure 5.60 shows the result using one iteration (no-Monte Carlo analysis is
performed).





137




















Figure 5.60 Results using one iteration for test No.14
In this case, the calculations using measured signature and synthetic waves
overestimate the maximum measured value for the vibration.

Monte Carlo analysis (improved signature hole technique) test No.14
Using the improved methodology and the developed software, it is possible to
analyze all the variables that are involved in the problem independently. Figure 5.61
shows the screen view for the current problem when all variables are included.
















Figure 5.61 Variables involved in the prediction test No.14 using improved
signature technique.
138

A standard deviation for the amplitude factor of 2 is used as well as a standard
deviation of 1 for the decay factor and 1000 for the wave velocity. The results are
included in Figure 5.62



















Figure 5.62 Monte Carlo result using fixed parameters for test No.14
Peak values histogram and convergence plot is included in Figure 5.63.













a) Histogram prediction test No.14 b) Convergence plot test No.14

Figure 5.63 Histogram and convergence plot of the prediction for test No.14
In the modeling process according to the convergence plot, more than 80 runs are
needed to reach the convergence of the mean peak particle velocity. In this test, it is
possible to see that measured value is between 2.44 standard deviations of the mean
predicted value.

139

0
10 20 30
ft
N
06/ 29/ 2011 6: 45 pm
5.4.5 Model 5 electronic initiation system (test No.15)
This test had a similar geometry configuration that previous where pyrotechnic
initiation delay system was used. The main idea of this test was to compare vibrations
levels for similar shots (similar geometries) when different initiation systems are used
(nonel vs electronic). Next the main parameters of test No.15 are included.

Table 5-9 Blast test parameters test No.15
Parameter Value
Total holes 40
Face height (ft) 45
Depth (ft) 45
Burden = spacing (ft) 18
Diameter (in) 7.875
Total explosive (lbs) 33,478.25

In this test 40 holes were detonated.

























Figure 5.64 Test No.15, plan view from survey
In this test, there are no signature waveforms for the particle velocity.


140




















Figure 5.65 Vibration record seismograph 3599 test No.15
Despite the possibility to model any component, transverse component was
chosen.

5.4.5.1 Signature test No.15
In this case, same signature average used in the previous model was chosen as
signature waveform for this specific model.

5.4.5.2 Synthetic signature signals
Using the same frequency intervals for the previous test, 40 signature signals were
generated. In the calculation, a scale factor average of 5.03, assuming a standard
deviation of 2.0 was also used.

5.4.5.3 Timing sequence
The delay between holes in this test was the typical delay used in this particular
mine 100 and 42ms. The difference with the previous model is that in this case the
initiation system used was electronic. According to the information for this type of
detonator, a standard deviation of 0.10ms was used.




141

5.4.5.4 Results Model 5 (test No.15)
Traditional signature methodology

Figure 5.66 shows the result using one iteration (no-Monte Carlo analysis is
performed).















Figure 5.66 Results using one iteration for test No.15
In this case, the calculations using measured signature and synthetic waves
overestimate the maximum measured value for the vibration.

Monte Carlo analysis (improved signature hole technique) test No.15
The results in this model are included in Figure 5.67.
















Figure 5.67 Monte Carlo result using fixed parameters for test No.15
142

Peak values histogram and convergence plot is included in Figure 5.68.














a) Histogram prediction test No.15 b) Convergence plot test No.15

Figure 5.68 Histogram and convergence plot of the prediction for test No.14
In this test, it is possible to see that measured value is between 2.54 standard
deviations of the mean predicted value. Convergence plot in this case indicates that less
than 50 runs are needed to approach a constant mean peak particle velocity value. If
Figure 5.63b and Figure 5.68b are compared, when electronic initiation system is used,
fewer number of iterations are required to approach a constant mean peak particle
velocity value. This is basically due to the high scatter in timing delay in nonel initiation
system compared to electronic. The likelihood that a previous run output have a similar
value than the current run is higher when less scatter is used in the variables (in this case
the initiation timing).






143


Chapter 6
DISCUSSION OF RESULTS AND FUTURE WORK RECOMMENDATIONS

6.1 Discussion about the improved methodology and the validation results
Based on the results from the different models included in Chapter 5, a discussion
about the main parameters of the improved signature hole technique is presented.
6.1.1 Number of frequencies to model the signature signal
The main concept behind the Fourier series to model the signature signal in the
current research is to choose a finite number of frequencies (in this case the main
frequencies) to approach a synthetic signal. There are two consequences when a finite
number of frequencies are selected:
1. The shape of the synthetic waveform does not look like the original signature
waveform, and;
2. A greater value is necessary for the amplification factor when less frequencies
are used.
Those phenomenons are true for any signal (airblast or vibration signals).

For example, in the case of the signature signal for model 1 (test No. 5 airblast),
the initial part of the synthetic signal does not match to the measured signal. In that case
five frequencies were used to calculate the approach. Figure 6.1 shows the result when
five frequencies are included.


















Figure 6.1 Signature signal test No.5 using five frequencies
In this case, the initial part of the synthetic signal represents a peak value higher
than the positive peak value of the measured signal. As a consequence, the airblast
144

signature waveforms randomly generated (see Figure 5.16) will overestimate the airblast
waveform in the initial part of the signal. Finally, when the prediction is calculated, the
predicted value is going to be overestimated and it is necessary to adjust the scale factor
to reduce the predicted value as explained in Chapter 5.
In this research, it has been concluded that when a more appropriate shape of the
signature signal is needed, more frequencies are required to model the signature
waveform. Figure 6.2 shows the random signature signals when 25 frequencies are used.





















Figure 6.2 Random airblast signature waveform signals test No.5 using 25
frequencies.

In this case an amplification factor to match the maximum negative peak of the
signal of 2.22 was used instead of 2.8 (see Chapter 5). When improved signature is used,
including 25 frequencies, an average value of 227 Pa and a standard deviation of 30.20Pa
is obtained. In this case, an amplification value of 2.22 was used for 3Hz and 1 for the
other frequencies. The final waveform prediction for airblast is included in Figure 6.3.










145

















Figure 6.3 Final prediction for airblast using 25 frequencies and amplification
factor of 2.2 for main frequency (3Hz).
Next table contains the results comparison and how the parameters change when
more frequencies are used.

Table 6-1 Comparison results according to the number of frequencies used
Model
(Test
No.)
Peak
particle
velocity
absolute
value
(measured)
Model using four frequencies Model using 25 frequencies
Scale
factor
Peak
particle
velocity
mean
Standard
deviatio
n
Scale
factor
Peak
particle
velocity
mean
Standard
deviatio
n
02
(No.10)
0.32in/s 5.25 0.242in/s 0.035 2.77 0.29in/s 0.035
03
(No.12)
0.145in/s 4.67 0.203in/s 0.042 2.72 0.19in/s 0.031
04
(No.14)
0.12in/s 5.03 0.206in/s 0.035 2.75 0.22in/s 0.0325
05
(No.15)
0.125in/s 5.03 0.206in/s 0.032 2.75 0.22in/s 0.0304

Table 6-1 shows that increasing the number of frequencies used to simulate and
generate the signature waveform reduces the scale factor. However the mean values and
the standard deviation of the predictions are almost the same. To see the changes in the
shape of the vibration envelop predicted, next figures show the final result when 25
frequencies are used for all the validation examples included in Chapter 5.

146

1
4
6


Scale factor=5.25












a) Enveloped using four frequencies test No.10










Scale factor=2.77


b) Enveloped using twenty five frequencies test No.10

Figure 6.4 Results comparison using four and twenty five frequencies for model 02, test No.10.
147

1
4
7








Scale factor=4.67





a) Enveloped using four frequencies test No.12







Scale factor=2.72





b) Enveloped using twenty five frequencies test No.12

Figure 6.5 Results comparison using four and twenty five frequencies for model 03, test No.12.
148

1
4
8










Scale factor=5.03



a) Enveloped using four frequencies test No.14









Scale factor=2.75



b) Enveloped using twenty five frequencies test No.14

Figure 6.6 Results comparison using four and twenty five frequencies for model 04, test No.14.
149

1
4
9


Scale factor=5.03










a) Enveloped using four frequencies test No.15








Scale factor=2.75




b) Enveloped using twenty five frequencies test No.15

Figure 6.7 Results comparison using four and twenty five frequencies for model 05, test No.15.

150

In conclusion regarding the number of frequencies involved in the prediction,
increasing the number of frequencies involved in the prediction, shows an improvement
in the shape of the complete waveform (the shape becomes close in shape to the
measured). However, there is not a considerable improvement regarding the mean value
and standard deviation calculated for the complete blast.
When more frequencies are used in the prediction, a lower value for the scale
factor is needed. This is due to when more frequencies are used, the synthetic signal is
more like the measured signal before the decay factor is applied. The decay factor is
necessary to avoid the problem that the signal be repeated due to the character of the
Fourier series. Figure 6.8 shows the synthetic signal for test No.14 when twenty five
frequencies are used and before the decay factor is applied and before the scale factor is
calculated.



























Figure 6.8 Synthetic signal for test No.14 before apply decay factor
In this case, the synthetic signal repeats itself after 1s. Using twenty five
coefficients there is a perfect match between measured and synthetic within the first
second. This fact will lead to a lower scale factor compared to a situation where only
four frequencies are used.
151

6.1.2 Decay factor
Through the analysis of the signature waveforms for the specific chosen point, it
was observed that the decay factor is a site and directional specific parameter. This
parameter measures how the vibration energy is dissipated by the ground in the
monitoring point. Table 6-1 shows the decay factor for some tests in this research. All
values were calculated for the location of the seismograph 3599 and transverse direction.

Table 6-2 Decay factor signature waveforms at 3599 seismograph location
Test Decay
factor
10-S1 5.44
10-S2 4.70
11-S1 3.76
11-S2 4.49
12 4.63
13 4.50

The results of Table 6-2 show that having at least one signature hole, it is possible
to assess the value of the decay factor for the site under consideration.
6.1.3 Initiation system, timing and blast vibration
As mentioned in Chapter 4, there is an appreciable difference in the accuracy and
precision between electronic and non-electric initiation system. Such difference was
included in Figure 4.41 and reproduced again in Figure 6.9.



















Figure 6.9 Accuracy and precision electronic vs non-electric system
152

As shown in the previous figure, the precision of non-electric detonators is very
low when compared to electronic initiation systems.
In the case of tests No.14 and 15, they differ by four holes (34 holes test No.14
and 38 holes test No.15) and in both tests, the same nominal initiation sequence was used
(delay timing was used based on 42 and 100ms delays). Assuming that the energy
released by the four missing holes is not significant when compared to the entire blast,
due to the timing used, the electronic initiations system should lead to higher particle
velocity values. This is because through the blast, two holes are detonated at the same
nominal time 142, 242, 342ms etc. and this should be more critical for electronics than
for pyrotechnics because lower scatter in electronics increase the likelihood of two holes
detonating at the same time. So more energy is released in the case of electronic
detonators. However, it was observed that for seismograph 3599 which is in average
1440ft from the source, the complete vibration waveform, the peak particle velocity and
the main frequency of the signals are almost the same for both test, as showed in Figures
6.10 and 6.11.
























Figure 6.10 Waveform comparison test No.14 vs test No.15 transversal
component (seismograph 3599)
In this case, the difference of the peak of the particle velocity is just about
0.05in/s between both tests.



153






















Figure 6.11 Frequency domain comparison test No14 vs test No.15 (seismograph
3599)
Similar trends were observed for the other seismographs, regardless its distance
outside 1440ft. The actual readings of the peak values for all the seismographs in test
No14 and 15 are included in Table 6-2.

Table 6-3 Results test 14 and 15 particle velocity peak values (actual readings)
Seismograph
Distance
Average
(ft)
Test
14:
PPV
(in/s)
Rad.
Test
15:
PPV
(in/s)
Rad.
Test
14:
PPV
(in/s)
Vert.
Test
15:
PPV
(in/s)
Vert.
Test
14:
PPV
(in/s)
Trans.
Test
15:
PPV
(in/s)
Trans.
4906 711 0.365 0.820 0.235 0.600 0.450 0.600
3599 1440 0.115 0.120 0.095 0.095 0.120 0.125
4762 1946 0.175 0.175 0.095 0.095 0.105 0.120
3857 2627 0.020 0.025 0.025 0.025 0.025 0.030
180 3160 0.020 0.035 0.020 0.020 0.015 0.025
Note: Test 14: Pyrotechnic delay system
Test 15: Electronic delay system

In conclusion, when using traditional nominal timing delay (in this case 100
and 42ms) in this particular mine, there is no difference between electronic and non-
electric initiation system regarding the peak particle velocity for seismographs beyond
1440ft.
154

Similar results between the actual readings and the simulations were obtained
using the improved signature hole methodology in the location of the seismograph 3599
(1440ft far from the blast). When modeling, for test No.14 (nonel), an obtained
Transversal peak particle velocity of 0.206 in/s and standard deviation of 0.035,
compared to 0.206 in/s and standard deviation of 0.032 for test No.15 (electronic) (see
figures 6.6 and 6.7 (a)).

To analyze if this result is explained based on the scatter introduced by the travel
time of waves or due to the initiation timing system, a model including only the scatter of
both initiation systems was used. In other words, the traveling time due to the distance
was not included in the calculations. Results are included in Figure 6.12.















a) Pyrotechnic b) Electronic
Figure 6.12 Test No.15 simulating both initiation systems and including only time
delay due to initiation sequence.

Figure 6.12 shows once again that in this case there is not a considerable
difference for seismographs beyond 1440ft when pyrotechnic and electronic initiation
systems are used. Other timing combinations for the location of the seismograph 3599
were analyzed in order to see any possible trend between nominal timing and the
initiation system; the results are included in Table 6-3.










155

Table 6-4 Delay timing and initiation system simulation far seismographs (beyond
1440ft)
Timing Electronic Pyrotechnic
100/42ms
|: | = u.2u
in
s
,
o = u.u24
|: | = u.19
in
s
,
o = u.u24
42/17ms
|: | = u.2u
in
s
,
o = u.u4u
|: | = u.2u
in
s
,
o = u.u4u
25/9ms
|: | = u.2S
in
s
,
o = u.uS1
|: | = u.2S
in
s
,
o = u.uS4
17/9ms
|: | = u.S8
in
s
,
o = u.uS4
|: | = u.S7
in
s
,
o = u.u4u
10/5ms
|: | = u.6u
in
s
,
o = u.uS8
|: | = u.61
in
s
,
o = u.uSS

According to the results of Table 6-3, when using different nominal timing
sequence there is no difference between electronic and pyrotechnic initiation system;
however when the lower timing is used, an increment in the peak particle velocity is
observed in the location of the seismograph 3599 (1440ft from blast).

In the bottom row timing of the Table 6-3 all holes basically are detonated 5ms
apart with some holes detonated at the same time as explained. On the other hand, when
a uniform sequence of 5ms is used, an average value of 0.32 in/s with a standard
deviation of 0.045 was obtained; this result indicates the influence of the timing initiation
sequence in the final peak particle velocity of the whole blast event.

Considerable differences in peak particle velocity between tests No.14 and 15
emerge when Table 6-2, is reviewed for seismograph 4906. The difference in peak
particle velocity generated by the two initiation systems is around 1.5 and 2.5 times for
all components. Due to this fact, the modeling for seismograph 4906 was performed and
the results are presented next.
As previously mentioned, there is not a specific signature for tests No.14 and 15,
so for modeling proposes, the average of the signatures between tests No.12 and 13 was
used as the signature for modeling. Figure 6.13 shows the two signatures at the location
of seismograph 4906 recorded and the average used for modeling.

Following the procedure for modeling explained previously, four frequencies; 2.4,
9.6, 10.8 and 16 Hz were chose to simulate the signature signal, additionally, a scale
factor of 2.48 and a decay factor of 4.85 were used. It was assumed a standard deviation
of 10ms for the pyrotechnic detonators.





156



















Figure 6.13 Signature used to model Tests No.14 and 15 (average) in seismograph
4906 location.

In the modeling of test No.15, a standard deviation of 0.1ms was assumed for
electronic initiation system. In both cases, the traveling time was introduced in the model
using a wave velocity of 16000ft/s and a standard deviation for wave velocity of 1000ft/s.
Finally, the standard deviation of the amplitude and the decay factor was assumed as 1 in
both cases. In modeling test No.15, it was necessary to adjust the amplitude factor to 4
and it was assumed a variation in amplitude of 2. This is needed because test No.15 is
closer than test No.14, also it is necessary to take into account that here for modeling we
are using signatures from test No.12 and 13 that are even further from test No.15.
The results for modeling Test No.14 and 15 are included in Table 6-5, Figure 6.14
and Figure 6.15 respectively.

Table 6-5 Modeling results for seismograph 4906 (Pyrotechnic vs electronic)
Radial component

Test No.
Peak particle
velocity absolute
value (measured)
Modeling results
Peak particle
velocity mean
Standard
deviation
No.14
(Pyrotechnic)
0.365in/s 0.41in/s 0.062
No.15
(Electronic)
0.82in/s 0.65in/s 0.010

It is remarkable how the proposed methodology, using the same signature
waveform to model pyrotechnic and electronic initiation systems show the same trend
157

than the actual readings for seismograph 4906. As previously stated, it is expected higher
peak particle velocity values when electronic initiation system is used compared to
pyrotechnic system. For this particular case the fact that two holes been detonated at the
same delay time increase the peak particle velocity value.

158

1
5
8















Figure 6.14 Modeling results test No.14 seismograph 4906 (Pyrotechnic)














Figure 6.15 Modeling results test No.15 seismograph 4906 (Electronic)
159

Chapter 7
CONCLUSIONS AND FUTURE WORK
7.1.1 Conclusions
Current methodologies to assess vibrations levels from blasting are low in
accuracy and precision. This is an inevitable consequence of the high
uncertainty of the variables involved in the problem. Specifically, in most
of the cases, the high variability in the geological conditions made each
mining blast a unique event, regardless of the consistency of the geometry,
quantity and type of explosives, and nominal initiation sequence.
The most current methodology used to assess blast vibrations from blasting
(scaled distance) presents important disadvantages and inaccuracies.

A reliable and extensive database is required
Weakness in the theoretical and physical support of the equations.
There is no other parameter regarding delay time in a blast than the 8
milliseconds rule. No clue about how different initiations timing affect the
vibration levels.

The first disadvantage is related to the requirement of a confident and extensive
database to calculate the site specific geological constants in the scaled distance equation.
This fact makes this methodology impossible to apply for some areas at the mine if no
vibration information was collected near the site where vibration levels are needed to
calculate. Another disadvantage is the weak theoretical and physical justification
regarding to the square root of the weight of explosive used in the scaled distance
calculations. This concept was developed by Blair (2000) and was included in this
document in Chapter 4. Finally, through the validation of the current improved
methodology, it was observed the high incidence of the nominal initiation timing
sequence in the vibrations levels. Scaled distance methodologies do not take into account
the initiation sequence timing when vibration levels are calculated.

Current signature hole techniques assume the invariability of the signature
waveform hole-to-hole (linear superposition). While the invariability of
the signature waveform can be true under some exceptional conditions,
like a rock mass containing few or no joint systems and a massive rock
layer, in general the geological conditions change, even between holes
affecting in some grade the signature waveform that each hole generate.
In this research, the methodology proposed to improve current signature
hole techniques is based in a probabilistic approach. The proposed
methodology allows the change of the signatures hole to hole in a random
fashion using Fourier series to generate different signatures for each hole.
Through this mathematical tool variations in geology, geometry hole to
hole, different explosives, contamination, change in the distance etc. are
considered implicitly in the model.

160

The equation (Silva-Lusk) base in Fourier Series to introduce random
behavior in the signatures hole to hole is given by:


(t) = _c
o
+ ASF
m
{A
m
sin(2n requency
m
t +
m
)]
m
n=1
_ c
-decay actur


[7.1]
where:
ASF
m
: amplification scale factor for frequency m.
c
o
: first term in the Fourier series
m: number of frequencies chose to approach the measured
signature waveform.
A
m
: amplitude coefficient for frequency m in the Fourier series
rcqucncy
m
: frequency value chose to approach the measured signature
waveform.
t: time

m
: phase for frequency m
Jccoy octor: factor related to the attenuation energy in that particular
monitoring point.

Based on the results from field tests conducted in this research, it can be
concluded that the usage of electronic detonators against nonel has more
impact in vibration levels closer to the site of the blast (this is included in
the results of test No.14 and 15). In this particular case, the low scatter in
electronic detonators increased the likelihood of two holes being detonated
at the same time as initially designed. For the topographical and
geological particular conditions where the test No.14 and 15 were
developed. For distances further than 1440ft from the blast source, there
is no difference in the vibration levels when electronic or pyrotechnic
initiation system is used.


The probabilistic methodology proposed in this research using a Monte
Carlo scheme, allows the design of the initiation timing in mining blasts.
According to the initiation sequence and timing selected for different
scenarios, using the proposed methodology it is possible to predict or
calculate vibration levels in a particular monitoring point base on the
signature of one hole with the same geometrical characteristics than the
production holes.

The usage of signature holes recorded along with production holes is a
practice that improves the quality of the results and confidence of the
signature methodologies used in the assessment of vibration levels. This
is because the signatures become more representative of the geological
161

conditions and the structural conditions of the rock mass where the
explosions take place.

It is possible to use signature techniques to assess the levels of airblast
from a mining production blast.


7.1.2 Novel contributions
In the development of the current research several novel contributions where
performed regarding blast vibration prediction:

This research presents to the academia and the industry a clear and well
supported methodology to assess blast vibration levels based on improved
signature techniques. The methodology is based on one probabilistic
approach using Monte Carlo scheme, thus, it allows to calculate vibration
levels from a mining blast using confidence intervals according to the
available information to perform the analysis.

The randomization of the waveforms hole to hole allows to perform non-
linear superposition when the complete waveform is calculated.

The simulation of signature waves based on Fourier Series as part of
signature hole techniques is a novel contribution to this methodology.

The implementation of Monte Carlo scheme to signature hole technique is
a novel contribution in the area of blast vibration prediction.

The performance of signature holes along to the production blast is a novel
contribution to signature hole techniques.


7.1.3 Recommendations for Future Work
The recommendations for future work in the area of blast vibration prediction
include:
In this research to randomize the variables involved in the problem,
normal standard deviation distributions were used. It is necessary to
perform more field test in different mines to verify or modify the
probability distributions used for each variable.

More field tests are required to verify, validate and adjust the proposed
methodology. In this research, the methodology was used to match the
recorded vibration from different production blasts. In a second stage of
this research, it is necessary to predict the vibration levels before the
production blast occurs using the appropriate information.

162

Using the proposed methodology, it is possible to assess the vibration
levels at the specific monitoring point where a previous signature had been
recorded. It is necessary to implement a methodology to assess vibration
levels where no signature information is available. This can be done
through the use of transfer functions propagating the signature from the
monitoring point to the point of interest and using the current proposed
methodology.

More research regarding the current monitoring devices used in mining
industry is required. The limitations of the devices, the internal filtering
processes of the recorded signals are not clear. How those parameters
affect the assessment of vibration levels in blasts mining is not totally
clear.


163


REFERENCES

AbuelmaAtti M.T., (2001). A New method for Fourier Analysis of discontinuous
and peak periodic waveforms. Journal Computers and Mathematics with applications, 41,
pp 1417-1423.

Aimone, C.T., (1992). Rock breakage: Explosives, blast design. Hartman, H.L.
(ed.), SME Mining Engineering Handbook. Society of Mining Engineers, Littleton, pp
722-746.

Aldas, G.G.U., Ecevitoglu, B., (2008). Waveform analysis in mitigation of blast-
induced vibrations. Journal of Applied Geophysics 66, pp 25-30.

Aldas, G.G.U., (2010). Investigation of blast design parameters from the point of
seismic signals. International Journal of Mining, Reclamation and Environment. Vol. 24,
No. 1, March 2010, 8090

Aldas, G.G.U., (2010). Non-Linear Behavior of Blasting Noticed on Seismic
Signals. Gazi University Journal of Science. 23(4):401-411 (2010)

Ambraseys, N. R. and Hendron, A. J., (1968), Dynamic Behavior of Rock
Masses, Rock Mechanics in Engineering Practice, John Wiley & Sons, Inc., London.

Anderson, D. A. (2008). Signature Hole Blast Vibration Control- Twenty Years
Hence and Beyond. The Journal od Explosives Engineers. September/Octuber., pp 6-14.

Anderson D. A., Winzer S. R., and P. Ritter A. (1982). Blast Design for
Optimizing Fragmentation while Controlling Frequency of Ground Vibration.
International Society of Explosives Engineers General Proceedings,. Pp 69-84.

Anderson, D.A., Winzer, S.R., and Reil, J.W., (1985). A method for Site-Specific
Prediction and Control of Ground Vibration from Blasting. Proc. 11
th
Anual Conf. Expl.
And Blast Tech.

Anderson D. A., (1989). The 8 Millisecond Criterion: Have We Delayed Too
Long in Questioning It?. International Society of Explosives Engineers General
Proceedings, pp 381-395.

Ash, R.L., (1963). The mechanics of rock breakage, standards for blasting design.
Pit and Quarry.

Bajpayee., T.S., Mainiero, R.J. (1990). Firing accuracy of electric detonators.
International Society of Explosives Engineers, General proccedings., pp 89-98.

164

Barkley R.C. and Daemen J.J.K. (1986). Computer Simulation Predictor of
Ground Vibration Indiced by Blasting. Report to U.S. Bureau of Mines, OFR 105(3)-81.

Batzle, M. L., Simmons, G., Sigfried, R. W. (1980). Microcrack Closure in Rocks
Under Stress: Direct Observation. J. Geophysical Research, 85(B12), 70727090.

Blair, D.P. (2004). Charge weight Scaling Laws and the Superposition of Blast
Vibrations Waves. Fragblast Vol.8. No.4, pp. 221-239.

Blair, S.C., Cook, N.G.W., (1998) Analysis of compressive fracture in rock usign
statistical techniques. Int. J. Rock Mech. Min. Sci. 35 (7).

Blake, F. C. (1982) Spherical wave propagation in solid media. J. Acoustical
80c. Am., 24(2): 211-215.

Bobet, A., Fakhimi, A., Johnson S., Morris J., Tonon F., and M. Ronald Yeung.,
(2009). Numerical models in discontinuous media: Review of advances for rock
mechanics applications. Journal of Geotechnical and Geoenvironmental Engineering,
Nov 2009., pp 1547-1561.

Bollinger, G. A. (1980). Blast vibration analysis. Carbondale: Southern Illinois
University Press, [1971].

Bourbie, T., Coussy, O., Zinszner, B., (1987), Acoustics of Porous Media, Gulf
Pub. Co.

Buckingham, E. (1915). The principle of similitude. Nature 96.

Chiappetta, R.F., S.L. Burchell, J. W. Reil, and Anderson, D.A. (1986). Effects of
accurate ms delays on productivity, energy consumption at ther primary crusher,
oversize, ground vibrations and airblast. Proceedings of the 12
th
Annual Conference on
Explosives and Blasting Techniques, Atlanta, GA, pp 213-234.

Chi-Tsong Chen (2004). Signals and Systems 3th edition. Oxford University
Press

Christopherson, K., and Papillon B., (2008). Vibration reduction through
production-signature hole blasting. The Jornal of Explosives Engineers,
September/October, pp 16-20.

Cook, M.A., (1958) The Science of High Explosives, ASC Monograph No. 139,
Reinhold.

Cundall, P. A., Detournay, C., (2008) Modeling Shock and Detonation Waves
with FLAC. Continuum and Distinct Element Numerical Modeling in Geo-Engineering
165

(Proceedings, 1st International FLAC/DEM Symposium, Minneapolis, August 2008),
Paper No.10-06. R. Hart et al., Eds. Minneapolis: Itasca Consulting Group, Inc., 2008.

Crenwelge Jr., O.E. and Peterson T. A., (1986). Overburden blasting vibrations:
Analysis, prediction, and control. International Society of Explosives Engineers, General
proccedings., pp 269-281.

Crenwelge Jr., O.E. (1988). Method for Determining Amplitude-frequency
Components of Blast Induced Ground Vibrations. Research proceedings of the Fourth
Mini-Symposium on explosives and Blasting Research. Society of Explosives Engineers.
Houston, TX.. pp73-88.

Crenwelge Jr., O.E. (1988). Use of Single Charge Vibration Data to Interpret
Explosive Excitation and Ground Transmission Characteristics. International Society of
Explosives Engineers, General Proceedings., pp151-160.

Drake, J.L., Little, C.D., (1983). Ground Shock from Penetrating Conventional
Weapons. Proceedings Symposium of the Interaction of Non-Nuclear Munitions with
Structures, U.S. Air Force Academy, Colorado Springs, CO.

Dowding, C.H. (1985). Blast Vibration Monitoring and Control. Library of
Congress. p6.

Duvall, W. I. Fogelson, D. E. (1961). Review of criteria for estimating damage to
residences from blasting vibrations, Report of Investigation 5968, Bureau of Mines.

Edwards, A. T. Northwood, T. D. (1960). Experimental studies of the effects of
blasting on structures, The Engineer Sept. 30.

Enescu D., Anca Georgescu, and Vasile Mrza (1973). Simulations of the
underground explosions generating longitudinal and transverse waves. Bulletin of the
Seismological Society of America. June 1973 63:753-760

Frantti G. E. (1963). Spectral energy density for quarry explosions. Bulletin of the
Seismological Society of America. Vol.53. No.5, pp 989-996.

Frantti, G.E. (1977). Near-Field Elastic Ground Response Spectra for Multi-Hole
Surface Explosions. General Proceedings., pp 149-164.

Furtney, J. K., Cundall, P. A., Chitombo, G. P. Developments in Numerical
Modeling of Blast Induced Rock Fragmentation: Updates from the HSBM Project. Rock
Fragmentation by Blasting (FRAGBLAST 9, Granada, Spain, September 2009), pp. 335-
342.

166

Greenhalgh S. A. (1980). Effects of delay shooting on the nature of P-Wave
seismograms. Bulletin of the Seismological Society of America. Vol.70. No.6, pp 2037-
2050.
Harries, C., Gribble, D.P. (1993). The development of low shock energy
explosive ANRUB, Proc. 4th Int. Symp. Rock Fragmentation By Blasting, Vienna. pp.
379-386.

Nicholls, H.R., Charles, F.J., Duvall, W.I. (1971). Blasting vibrations and their
effects on structures. Bureau of Mines. U.S. Bulletin 656.

Hinzen, K.G., and R. Ludeling., (1987). A New Approach to predict and reduce
blast vibration by modeling of seismograms and using a new electronic initiation system.,
Proceedings of the 13
th
Annual Conf. on Exp. And Blasting Techniques, New Orleans,
LA.

Hossaini, S.M.F., Sen, G.C. (2004). Effect of explosive type on particle velocity
criteria in ground vibration, The Journal of Explosive Engineering, July/August, pp. 34-
39.
Hunter, C., Fedak, K. and Todoeschuck, J.P. (1993). Development of low density
explosives with wall control applications, Proc.19th Annual Conf. Explosives and
Blasting Techniques, Jan, 31-Feb. 4, San Diego, California, USA, ISEE, pp. 549-555.

International Society for Rock Mechanics. (1992). Suggested Methods for Blast
Vibration Monitoring. ISRM., pp 145-156.

Kavetsky, A, Chitombo, G P F, McKenzie, CK and Yang, RL, (1990). A Model
of Acoustic Propagation and its Application to Determine Q for a Rock Mass, Int J Rock
Mech Min Sci Geomech Abstr, 27(1):33-41.

Khandelwal M., Singh T.N., (2006). Prediction of blast induced ground vibrations
and frequency in opencast mine: A neural network approach. Journal of Sound and
Vibration 289, pp 711-725.

Kjartansson, E. (1979). Constant Q wave propagation and attenuation. J.
Geophys. Res. 84: 47374748.

Kramer, S.L., (1996). Geotechnical Earthquake Engineering. Prentice Hall.

Kranz, R.L. (1983) Microcracks in rocks: a review. Tectonophysics 100: pp 449-
480.

Kreyszig, E., (2005) Advanced Engineering Mathematics. John Wile & Sons, pp
478.
Langefors, U., Kihlstrom, B., (1963). The modern technique of rock blasting.
John Wiley & Sons, p 263.

167

Langefors, U., Kihlstrom, B., Westerberg, H. (1958). Ground vibration in
blasting.Water Power, Sept-Nov.

Larsson B., Holmberg R., Westberg J., (1988) Super accurate detonators - a
rockblasters dream. International Society of Explosives Engineers General Proceedings.

Linehan, P. ; Wiss, J. F. (1983). Vibration and Air Blast Noise From Surface Coal
Mine Blasting. AIME Transactions V. 272, 1982.

Lusk, B., Worsey, P., Oakes, K., Chambers, J., Crabtree, S., Brasier, T., and
Wheeler R., (2006). Destructive Wave Interference in Underground Blasting Utilizing
Precise Timing. International Society of Explosives Engineers General Proceedings,. Vol.
1.

Lusk, B., Hoffman, J., Silva C., Wedding W., Morris, E., Calnan J. (2012)
Evaluation of Emergent Electronic Detonators and Modern Non-Electric Shocktube
Detonaros Accuracy. Blasting and fragmentation. Vol.6. No.1. pp 1-17.

Lusk, B., Silva, C., Eltschlager, K., Hoffman, J. (2010) Acoustic Response of
Structures to Blasting Analyzed Against Comfort Levels of Residents Near Surface Coal
Operations. Office of Surface Mining. Final Report OSM cooperative Agreement
S07AP12481.

Matsumoto, M., Nishimura, T. (1998) Mersenne Twister: A 623-dimensionally
equidistributed uniform pseudorandom number generator. ACM Trans. on Modeling and
Computer Simulation Vol. 8, No. 1, January pp.3-30

Miller, D., Drew, M. (2007). A review of the benefits being delivered using
electronic delay detonators in the quarry industry. The Institute of Quarry Australia 50
th

National Conference, Hobart, Tasmania, October, pp 1615-1645.

Mortazavi, A. Katsabanis, P.D. (2001). Modelling burden size and strata dip
effects on the surface blasting process. International Journal of Rock Mechanics and
Mining Sciences, Vol 38, pp 481-498.

Olsson, M., Nie, S., Bergqvist, I., Ouchterlony, F. (2001). What causes cracks in
rock blasting?. Proc. EXPLO2001. Hunter valley, NSW, Australia, pp 191-196.

Pollack, R.N (1963). Effect of delay time and number of delays on the spectra of
ripple-fire shots. Earthquake Notes, 34, pp 1-12.

Randall, M. W., (1991). An Analysis of Firing Time Scatter Effects on Vibration
Simulations from Waveforms with Low and High Frequency Components. International
Society of Explosives Engineers, Research proccedings., pp 135-142.

168

Reisz, D.W., McClure, R., and Bartley, D. (2006). Why the 8MS rule Doesn`t
Work. International Society of Explosives Engineers, Vol2. General proccedings.

Rholl S.A., Stagg M.S. (1988). A Computer Program to Predict the Probability of
Overlap or Crowding of Adjacent-Period Millisecond-Delayed Initiators. Research
proceedings of the Fourth Mini-Symposium on explosives and Blasting Research.
Society of Explosives Engineers. Houston, TX.. pp91-104.

Rossmanith, H.P. (2003). The Mechanics and Physics of Electronic Blasting.
General Proceedings Vol.1.

Sakamoto, M., M. Yamamoto, K. Aikou, E. Suzuki, H. Fukui, and K. Ichikawa
(1989). A study on high accuracy delay detonator. Procceding of the 15
th
Annual Conf.
Exp. And Blasting Techniques, New Orleans, LA., pp 185-195.

Sally,S. A., Daemen, J.J.K (1983). Ground and Air Vibrations Caused By Surface
Blasting. Volume 2 Ground Vibration Monitoring And Assessment Of Conventional
Predictors. Report to U.S. Bureau of Mines OFR 105(2)-84. p11

Saharan, M.R., Mitri, H.S. and Jethwa, J.L. (2006) Rock fracturing by explosive
energy: review of state-of-the-art. Fragblast, 10: 12,61-81.

Saharan, M.R., Mitri, H.S. (2008) Numerical procedure for dynamic simulation of
discrete fractures due to blasting. Rock Mechanics and Rock Engineering, 41 (5), pp 641-
670.

Sanchidrin J.A., P. Segarr., L.M Lpez. (2007). Energy components in rock
blasting. International Journal of Rock Mechanics & Mining Sciences V.44, pp 130-147.

Silva Castro J, Sebastian L. B., Gamber, N., Lusk, B (2011). Numerical modeling
of subsurface blasts. Pan-Am CGS Geotechnical Conference. October 2-6, Toronto,
Canada.

Siskind, D.E,. Stagg, M.S., Kopp, J.W., and Dowding, C.H. (1980b). Structure
response and damage produced by ground vibration from surface mine blasting. U.S.
Bureau of Mines RI 8507.

Siskind, D.E,. (2005). Vibrations From Blasting. International Society of
Explosives Engineers.

Sobol, I.M., (1960). The Monte Carlo Method. Popular lectures in mathematics.
U.S Department of Health Education and Welfare National Institute of Education. The
University of Chicago Press.

169

Spathis, A.T., 2010, A brief review of measurement, modeling and management
of vibrations produced by blasting. Vibrations from blasting: Proceedings and
Monographs in Engineering, Water and Earth Sciences, Taylor & Francis Group.

Stachura, V. J., D. E. Siskind, et al. (1981). Airblast instrumentation and
measurement techniques for surface mine blasting. [Washington, D.C.], U.S. Dept. of the
Interior, Bureau of Mines.

Villano, E.J., Charlie, W.A. (1993). Stress wave propagation in unsaturated sands-
Vol II Field explosive tests. Colorado State University, Final report ESL-TR-92-73.

Wang, Z.L., and Konietzky, H. (2009) Modelling of blast-induced fractures in
jointed rock masses. Engineering Fracture Mechanics 76 (2009) pp. 1945-1955.
Watson, J. T., (2002)

Willis, D.E. (1963). Comparison of seismic waves generated by different types of
source. Bulletin of the Seismological Society of America. Vol.53. No.5, pp 965-978.

Willis, D.E. (1963). A note on the effect of ripple firing on the spectra of quarry
shots. Bulletin of the Seismological Society of America. Vol.53. No.1, pp 79-85.

Worsey P. N., Tyler L. J. (1983). The Development Concept of The Integrated
Electronic Detonator. International Society of Explosives Engineers General
Proceedings,. Pp 489-496.

Yang, R., Scovira, D., 2010, A Model for Near and Far Field Blast Vibration
Based on Multiple Seed Waveform and Transfer Functions. Blasting and Fragmentation
Vol.4, No.2, 2010, pp91-116.

Young, R.P., and Hill, J.J. (1982). Statistical Analysis of Seismic Spectral
Signatures for Rock Quality Assessment. Geoexploration, 20, pp 75-91.

Yu YaLun and He Jun. (1997) A New Method for Blasting Seismic Signals
Processing. International Society of Explosives Engineers General Proceedings,. pp 349-
353.









170


VITA
Jhon Silva-Castro was born on August 9, 1972 in Zipaquir, Colombia, South
America to Pedro Silva and Trinidad Castro. He attended the Universidad Nacional de
Colombia in Bogot, Colombia, and was awarded a Bachelor of Sciences Degree in Civil
Engineering. Upon graduation, he was working for two years as field engineering in road
and tunneling construction. Later he attended to the Universidad Nacional de Colombia
in Bogot, Colombia, and was awarded a Master of Scienciae in Geotechnical
engineering. After five years working in mining related jobs, he moved to Lexington,
Kentucky to attend the University of Kentucky to study Mining Engineering. He worked
as both a research and teaching assistant to Dr. Braden Lusk. He expects to graduate in
December 2012 with a Philosophy Doctorate in Mining Engineering. While in graduate
school, he was awarded the Most Outstanding Graduate Student in the Mining
Engineering Department in April, 2011 and had been a member of both SME and ISEE
since 2008. He currently has three refereed journal publications and has several more
under review. He also has numerous local and international conference papers in which
he presented the findings of the papers. The three refereed journal publication citations
can be found below.


B.T. Lusk, J.M. Hoffman, J. Silva Castro, W.C. Wedding, E.G. Morris &
J. Calnan Evaluation of Emergent Electronic Detonators and Modern Non-
Electric Shocktube Detonators Accuracy. 2012 Blasting and Fragmentation. Vol
6. No.1, June 2012.

Lusk, B.T., J. Silva Castro, J. Hoffman, Case Study of Blast Vibration
Induced Sounds Recorded Inside a House Near a West Virginia Coal Mine.
Transactions of Mining Engineering (Accepted for Publication, May, 2010).

Lusk, B., J. Silva Castro, A Public Relations Plan Based on Structure
Generated Sounds and Public Input. The Journal of Explosives Engineering
(Accepted for Publication).






171





















APPENDIX A
Blasting log reports from the mine























172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199























APPENDIX B
Plan Layout of tests performed in 2011




















200

201

202

203

204

205

206






















APPENDIX C
Vibration records for events recorded in 2011

207

208

209

210

211

212

213

214

215

216

217

218

219

220