UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Critical assessment of the canlex blast experiment to facilitate a development of an in-situ liquefaction… Pathirage, Kapila S. 2000

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2001-0094.pdf [ 9MB ]
Metadata
JSON: 831-1.0063997.json
JSON-LD: 831-1.0063997-ld.json
RDF/XML (Pretty): 831-1.0063997-rdf.xml
RDF/JSON: 831-1.0063997-rdf.json
Turtle: 831-1.0063997-turtle.txt
N-Triples: 831-1.0063997-rdf-ntriples.txt
Original Record: 831-1.0063997-source.json
Full Text
831-1.0063997-fulltext.txt
Citation
831-1.0063997.ris

Full Text

CRITICAL ASSESSMENT OF THE CANLEX BLAST EXPERIMENT TO FACILITATE A DEVELOPMENT OF AN IN-SITU LIQUEFACTION METHODOLOGY USING EXPLOSIVES B Y KAPILA S. PATHJRAGE M.Sc, The Russian People's Friendship University, Russia A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF T H E REQUIREMENTS FOR T H E DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES T H E F A C U L T Y OF APPLIED SCIENCE DEPARTMENT OF CIVIL ENGINEERING We accept this as conforming to the required standard T H E UNIVERSITY OF BRITISH COLUMBIA December 2000 © Kapila S. Pathirage 9 2000 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of £-/' \S ( t~ &*S$?L *s££yL2T*./ The University of British Columbia Vancouver, Canada DE-6 (2/88) ABSTRACT In current engineering practice, the liquefaction susceptibility of soil is assessed by in-situ testing or by laboratory testing. The in-situ testing approach is based on field performance correlation approach and it is good for sandy soils, but becomes difficult to apply in soils such as gravels, silts and clays. In laboratory conditions, the cyclic loading tests could indicate inaccurate results unless those have been performed on absolutely undisturbed samples. The research in this thesis is a part of larger investigation, which is intended to explore the potential for evaluating liquefaction susceptibility in-situ by blasting. The main advantage of such a methodology is that the liquefaction resistance of the ground is evaluated under its existing stresses and ground water conditions. The major part of this research is a critical assessment of a field trial blast test, which was carried out during the Canadian Liquefaction Experiment (CANLEX) in Alberta. This includes the evaluation of measuring instruments used in the test, sampling rate and evaluation of measured data. In addition, this research investigates the induced wave patterns due to an explosion and the near field blast effects and the far field blast effects. Also, the research evaluates the possibilities of simulating earthquake like ground motions using explosives. The thesis establishes empirical relationships for liquefaction evaluation in loose saturated sandy soils. The state of art to establish these relationships is discussed. A comparison of these relationships with those proposed by other authors is also presented. In addition, the thesis discusses observed inadequacies of the C A N L E X blast test to help in planning and conducting blast tests in the future. ii TABLE OF CONTENTS Page ABSTRACT ii LIST OF TABLES vi APPENDICES vi LIST OF FIGURES vii ACKNOWLEDGEMENTS ix 1.0 BACKGROUND AND OBJECTIVES 1 1.1 Background 1 1.2 Purpose & Objectives 2 1.3 Organization of Thesis 3 2.0 PREVIOUS EXPERIENCE WITH T H E USE OF EXPLOSIVES TO M O D E L E A R T H Q U A K E LOADING AND C R E A T E LIQUEFACTION IN SOILS 4 2.1 Liquefaction Phenomenon 4 2.2 Cyclic Loading of Soils 5 2.2.1 Earthquake Loading 5 2.2.2 Blast Loading 5 2.3 Prediction of Liquefaction Susceptibility of Saturated Soils Subjected to Cyclic Loading 8 2.4 Explosives and Explosion 10 2.4.1 Physical Characteristics 13 2.4.2 Process of Energy Transfer 13 2.4.3 Typical Response in Soils 15 2.4.4 Previous Use of Explosives and Simulation of Earthquake Characteristics. 16 iii 2.5 Prediction of Soil Response to Explosive Detonation. 17 2.5.1 Empirical Procedures 17 (i) Hopkinson's Number 18 (ii) Normalized Weight 19 (iii) Powder Factor 19 2.5.2 Analytical Approach 20 (i) Static Theory 20 (ii) Dynamic Cavity Expansion 22 3. FIELD TRIAL BLAST AT SYNCRUDE J - Pit - C A N L E X 26 3.1 Introduction 26 3.2 Site Location 26 3.3 Soil Conditions 26 3.4 Test Program and Blast Layout 33 3.5 Characteristics of Explosives 37 3.6 Instrumentation 38 3.7 Post blast Observations in the J-Pit 39 4 D A T A ASSESSMENT FROM C A N L E X EXPERIMENT 40 4.1 Introduction 40 4.2 Assessment of instrumentation housing 40 4.2.1 Effect of Instrumentation Characteristics on Data Measured 40 4.3 Assessment of the Quality of Acceleration Data 42 4.3.1 Sampling Rate 43 4.3.2 Orientation of Accelerometer 46 4.4 Assessment of the quality of Pore Pressure Data 52 5. D A T A ANALYSIS 63 5.1 Introduction 63 5.2 Characteristic Wave Traces 64 5.2.1 Earthquake Ground Motion 64 5.2.2 Blast Loading 65 iv 5.2.3 Discussion 66 5.3 Energy Approach- Arias Intensity 73 5.4 Determination of Wave Propagation Velocity 81 5.4.1 P-Wave Velocity 81 5.4.2 S-Wave Content 88 5.6 Determination of the Peak Particle Velocitys and Displacement of a Blast Pulse 93 5.7 Peak Shear Strain 94 5.8 Pore Pressure 94 5.8.1 Trends of Pore Pressure 95 5.9 Relationships between the Dynamic Responses of Blasting to the Empirical Scaling Laws 107 6. CONCLUSIONS AND RECOMMENDATIONS 114 BIBLIOGRAPHY. 116 v LIST OF TABLES Table Page Table 1: Typical detonation pressures 12 Table 2: Soil parameters at the site-CANLEX 28 Table 3: Description of blast events-CANLEX 33 Table 4: Schedule of blast events-CANLEX 33 Table 5: Observed pore water pressure parameters-Blast # 2 at PI and P2 52 Table 6: Observed pore water pressure parameters-Blast # 4 at PI and P2 53 Table 7: Observed pore water pressure parameters-Blast # 5 at PI 54 Table 8: Observed pore water pressure parameters-Blast #6 at PI 55 Table 9: Observed pore water pressure parameters-Blast # 6 at P2 56 Table 10: Description of earthquakes 63 Table 11: Computed Propagation Velocity 84 APPENDICES: APPENDIX-A PHOTOGRAPHS 122 APPENDIX-B Additional Figures 128 APPENDIX-C Determination of Orientation of Accelerometers 142 APPENDIX-D Baseline Correction of Recorded Acceleration Time Histories 149 vi LIST OF FIGURES Page Figure 1: Typical blast set-up 7 Figure 2: The schematic view of an internal process of an explosion 14 Figure 3: The schematic view of an explosion 15 Figure 4: An idealized model for the analysis of non-linear response to blast loads 22 Figure 5: A typical modeled velocity and shear strains 25 Figure 6: Plan view of the site-CANLEX 27 Figure 7: Location of CPT tests carried out 29 Figure 8: CPT logs at #26 30 Figure 9: CPT logs at #27 31 Figure 10: Results of standard penetration test 32 Figure 11: Blast location #1, #2, #3 34 Figure 12: Blast location #4, #5 35 Figure 13: Blast location #6 36 Figure 14: Typical blast hole 37 Figure 15: Typical blast acceleration time histories 48 Figure 16: Acceleration time histories-blast # 4-P1 and P2 49 Figure 17: Fourier Amplitude Spectrum-Blast # 2-P1 and P2 50 Figure 18: Fourier Amplitude Spectrum-Blast # 4 51 Figure 19: Typical Blast Acceleration and Pore pressure time histories-blast 57 Figure 20: Pore pressure time histories-blast # 4-P1 & P2 58 Figure 21: Pore pressure time histories-blast #5-Pl 59 Figure 22: Pore pressure time histories-blast # 5-P2 60 Figure 23: Pore pressure time histories-blast # 6-P1 61 Figure 24: Pore pressure time histories-blast # 6-P2 62 Figure 25: Acceleration time histories-El-Centro site 67 Figure 26: Acceleration time histories-Treasure Island site 68 Figure 27: Velocity & displacement time histories-El-Centro site 69 Figure 28: Velocity & displacement time histories-Treasure Island site 70 Figure 29: Fourier Amplitude Spectrum-El-Centro site 71 Figure 30: Fourier Amplitude Spectrum-El-Centro & Treasure Island sites 72 vii Figure 31: Arias intensity-Treasure Island Site 76 Figure 32: Arias intensity-El-Centro site 77 Figure 33: Arias intensity-Blast 2 at P l , #4 at P l , 78 Figure 34: Arias intensity-El-Centro site and # 4 P2 79 Figure 35: Arias intensity for Liquefaction Assessment 80 Figure 36: Sketch for Determination of Propagation Velocity 82 Figure 37: Propagation Velocity versus Charge Number 87 Figure 38: Schematic View of a Blast Event 88 Figure 39: Determination of shear waves 91 Figure 40: Fourier Amplitude Spectrum-shear wave segment- blast #4 92 Figure 41: Integrated velocity & displacement time histories 96 Figure 42: Integrated velocity & pore pressure response- blast #4 97 Figure 43: Velocity Time Histories & Pore Pressure- # 4P1 and P2 98 Figure 44: Determination of Pore Pressure Propagation Velocity 99 Figure 45: Peak Dynamic Pore pressure versus Hypocentral Distance 102 Figure 46: Pore Water Pressure versus Shear strain 103 Figure 47: Peak Pore Pressure versus Peak Particle Velocity 104 Figure 48: Pore Water Pressure versus Strain Cycles 105 Figure 49: Pore Pressure Ratio versus Peak Particle Velocity 106 Figure 50: Scaled Distance versus peak particle velocity 110 Figure 51: Pore Pressure Ratio Vs Scaled Distance I l l Figure 52: Peak shear strains Vs Scaled Distance 113 viii ACKNOWLEDGEMENTS I would like to extend my gratitude to my thesis advisor, Dr. John A. Howie, for providing me with the opportunity to carry out this research and lending his valuable insight. I wish to thank Dr. Peter Robertson of the University of Alberta, Principal Investigator of the C A N L E X project for permission to use the C A N L E X test blast data. I would also like to thank Dr. Blair Gohl, Pacific Geodynamics Inc. for his assistance. Additional thanks are also extended to Dr. Carlos Ventura for his assistance in signal processing concerns. Finally, I would like to thank my wife and my son for their support and understanding throughout the last several months. I wish to acknowledge the financial support of the science council of B.C. and NSERC in the form of a graduate research assistantship. ix 1. BACKGROUND AND OBJECTIVES 1.1 Background Loose sands, when subjected to static or cyclic loading show significant decrease in volume. For saturated sand, if drainage is unable to occur during the time span of the loading sequence, then the tendency for volume reduction results in an increase in pore-water pressure. As the pore water pressure builds up, the effective stress is reduced resulting in a reduction in strength and stiffness. In the extreme, this process is responsible for a variety of effects, collectively known as liquefaction. In current engineering practice, the liquefaction susceptibility of soil is assessed by in-situ testing or by laboratory testing. In-situ testing is based on a correlation between field evidence of liquefaction and an index test such as Standard Penetration Test (SPT), Cone Penetration Test (CPT) or Shear Wave Velocity (Vs). This approach has been shown to work well for sandy soils, but becomes difficult to apply in soils such as gravels, silts and clays. In laboratory conditions, results of cyclic loading tests are affected by sample disturbance. There is a need for an in-situ test, which will allow assessment of liquefaction susceptibility in soils other than clean sands. The use of blasting has been considered for this purpose. For this to be possible, a greater understanding of soil response to explosive detonation is required. The research in this thesis is part of a larger investigation, which is intended to explore the potential for evaluating liquefaction susceptibility in-situ by blasting. The ultimate goal of this larger investigation is to develop an in-situ testing methodology to evaluate liquefaction susceptibility in all types of liquefaction-prone soils by using controlled detonation of 1 explosives. The main advantage of such a methodology is that the liquefaction resistance of the ground is evaluated under its existing stresses and ground water conditions and also, such a methodology can indicate the result of the test immediately at the site. This thesis presents a critical assessment of the results of a test blast program designed to investigate the potential for creating liquefaction in the foundation of the C A N L E X (Canadian Liquefaction Experiment) test embankment. This test embankment was constructed in J-Pit, at the Syncrude Plant site, Fort McMurray, Alberta. The embankment was built rapidly on a loose sand foundation in an attempt to induce static liquefaction of the foundation soil, and to initiate large deformation in the embankment and foundation. The test was unsuccessful in its attempt to induce liquefaction. Subsequently it was decided to carry out a test blast to determine whether it would be possible to induce liquefaction by blasting. The proposed full-scaled blast was never carried out. The data obtained in the test blast provide information about the relationship between explosive detonation and soil response, including the relationship between ground response and peak and residual pore pressures. 1.2 Purpose and Objectives The purpose of the thesis was to assess whether explosive detonations could be used to evaluate the liquefaction susceptibility of soils by matching the ground motion characteristics of earthquakes by careful sequencing of explosive detonations. The research included the following objectives: (i) Present current understanding of interaction between saturated soil and explosives, including the ability to cause liquefaction. 2 (ii) Assess the test data to distinguish between reliable and unreliable data and to identify practical issues in monitoring of ground motion and pore pressures due to blasting. (iii) Assess whether the reliable data indicated soil response consistent with (i) above and analyze the data for new insights into the effect of explosive detonation on loose saturated sands. These objectives fit in with a wider study of the potential for using blasting as a method of in-situ liquefaction assessment. 1.3 Organization of Thesis The initial part of this research explores the liquefaction concept, cyclic loading phenomenon, explosives and the explosion process in saturated soils (Chapter 2). Chapter 3 presents the details of the test carried out at the Syncrude Plant Site in Northern Alberta as part of the Canadian Liquefaction Experiment. Data assessment is discussed in Chapter 4. The results obtained in the test blast are analyzed and discussed in Chapter 5. The characteristics of ground motion caused by blasting are compared to earthquake shaking and the potential for using blasting to create ground motion typical of earthquakes is examined. Differences and similarities are discussed. In addition, empirical relationships between characteristics of the blast and the resulting ground motions and pore pressures are investigated. Conclusions and recommendations for further research are presented in Chapter 6. 3 2. PREVIOUS EXPERIENCE WITH THE USE OF EXPLOSIVES TO MODEL EARTHQUAKE LOADING AND CREATE LIQUEFACTION IN SOILS. 2.1 Liquefaction Phenomenon In 1978, the committee on soil dynamics of the Geotechnical division, American Society of Civil Engineers defined liquefaction as, "the act or process of transforming any substance into a liquid. In cohesionless soils, the transformation is from a solid state to a liquefied state as a consequence of increased pore pressure and reduced effective stress" Karl Terzaghi was the first to observe the phenomenon of liquefaction in loose saturated sand when subjected to unidirectional loading. In the late 50's, this concept was investigated by other researchers who found that liquefaction might be triggered in saturated loose soils subjected to other types of cyclic loading such as blast induced ground motion, and earthquake induced ground motion. Florin and Ivanov (1961) published the first account of Russian laboratory and field experiments on blast-induced liquefaction where pore pressure measurements and settlement were used to monitor the liquefaction susceptibility of the soil. However, earthquake ground motions have received more attention by the research community since the mid 1960's than blast induced ground motions, as two major earthquakes occurred in 1964. These caused major destruction due to earthquake liquefaction. For this reason, earthquake-induced liquefaction has been extensively studied. The prediction of pore-water pressure build up is the key for evaluating the potential for liquefaction. 2.2 Cyclic Loading of Soils 2.2.1 Earthquake Loading Earthquakes produce two basic types of waves, (i) Body waves made up of a variety of wave types, such as P-waves or compression waves, where the direction of particle motion is parallel to the direction of wave propagation, and S-waves or shear waves, particle motion is perpendicular to the direction of wave propagation. S-waves can travel only in a solid, as the material must be able to transmit shear. (ii) Surface waves, which travel only at ground surface. P-waves travel the fastest, S-waves next and lastly Surface waves. In geotechnical engineering, it is commonly assumed that the main source of cyclic loading contributing to pore pressure build up and liquefaction are the S-waves (Seed, 1975; Higgins, 1978). In earthquakes, S-waves provide the greatest transverse ground shaking, as there are usually many cycles with low frequencies and cyclic deformation can be large. The characteristics of earthquake-induced ground shaking at a site of interest depend on fault dislocation, length of fault rupture, distance and travel path between the fault and the particular site and the type of overburden soils. In engineering practice, the above ground shaking is usually measured as an acceleration or velocity time history. In general, the peak acceleration is less than gravity (g) and an earthquake induced ground-shaking lasts for about 20 sec to 60 sec. The ground shaking causes pore pressure rise in loose saturated soils, and it may induce a state of liquefaction in undrained conditions. 5 2.2.2 Blast Loading When an explosive charge is detonated in soils, body waves and surface waves are generated (Dowding, 1985). The dominant wave type are the P-waves that propagate outward in a spherical manner until they intersect a boundary such as a change in stratigraphy or ground surface (Dowding). At such an intersection, shear and surface waves are generated. However, to distinguish these waves separately on a recorded time history is very difficult because in most cases, blast induced soil responses are captured relatively close (10m to 20m) to the origin of the blast. Also, most explosives are detonated as a series of smaller explosions, with short delays in between detonations. Differences in travel path and delay time results in overlapping arrivals of both wave fronts and wave types. As an illustration, Figure 1 shows an arrangement of a blast event. Transducers are located either at close distance such as A or at far distance such as B. In some blast tests, transducers are located at both locations. A typical response of a transducer at location A, would show a single spiked pulse due to the direct shock wave transmission from the origin. At the far-field position B, an explosion will produce a sinusoidal-like pulse due to a combination of direct wave transmission, reflection, and refraction. In blast induced ground motions, it is commonly assumed that the P wave is the dominant wave with high frequencies and with high accelerations close to the blast origin. In general, blast induced ground motion lasts for only a few milliseconds, i.e. a lot less than the ground motion induced by earthquakes. The common consequence for both types of ground motion is the pore pressure rise in saturated soils in undrained conditions. Figure 1: An Arrangement of a Blast Event and Anticipated Wave Pattern at Close and Relatively Far Distance (after Dowding, 1985) 7 2.3 Prediction of Liquefaction Susceptibility of Saturated Soils Subjected to Cyclic Loading Cyclic loading of sands results in a tendency to contract. For saturated sands, the tendency for volume reduction results in an increase in pore-water pressure if drainage cannot occur sufficiently quickly. Seed and Idriss (1971), and Seed (1979) pioneered a method called the "Simplified Procedure" to identify soils susceptible to pore pressure build up during earthquakes. This simplified procedure is widely used in engineering practice because it requires only a limited number of input parameters. The parameters involved in this procedure are a function of the relative density, D r and the initial effective stresses. Some of these parameters need to be determined by laboratory tests. Cyclic tests performed in the last few years have revealed that a number of other factors affect the cyclic strength besides relative density (Dr). Such factors include consolidation and time under pressure (ageing effect). The influence of all these factors on the cyclic strength of the sands certainly complicates the accuracy of the results and makes its practical use more difficult. Peck (1979) argued that, considering all these difficulties, the cyclic loading tests used to evaluate liquefaction potential would probably give inaccurate results unless the tests have been performed on absolutely undisturbed samples. However, it is very difficult to obtain undisturbed samples of sands. The cyclic strain approach is another way to approach the problem of predicting pore water pressure buildup in saturated loose sands due to earthquake like loading patterns. Silver and Seed (1971) showed experimentally that the cyclic shear strain y c = T C / G rather than cyclic shear stress T c , controls the densification of dry sands. Dobry (1979) suggested the existence 8 of a threshold shear strain, y t in saturated sands. The threshold strain is that below which no pore pressure is generated by loading. The sand used in his experiments was Monterey No. 0 sand. This sand was a commercially available washed uniform medium to fine beach sand. His experiments revealed that the threshold shear strain yt, for Monterey sand is 0.01 %. Above this y t , excess pore pressures were generated by cyclic loading. One of the main advantages of the strain approach is that the parameters involved are less affected by factors such as relative density (Dr) and time under pressure as both T c (cyclic shear strength) and G are similarly affected by the same factors (Youd, 1972; Dobry and Ladd, 1982). The liquefaction susceptibility of soil is conventionally assessed by laboratory testing of undisturbed samples or by in-situ testing. Due to the great 'difficulty in obtaining undisturbed samples of soils, many engineers have preferred to adopt the field performance correlation approach. This approach is based on correlation between observed cases of liquefaction and soil index parameters, such as standard penetration resistance, cone penetration resistance, shear wave velocity, etc. All these methods have been used with some success in sandy soils, but become difficult to apply in soils such as gravels, silts and clays as few correlations are available and large adjustments are required to allow the use of the sand database. Since these liquefaction evaluation methods are subject to considerable uncertainty, a reliable method is required for assessment of liquefaction susceptibility of gravel, silts and clays. To accomplish such purpose, this research investigates the possible use of explosives to develop an in-situ liquefaction testing methodology. In summary, • In earthquakes, relatively low frequency S waves and surface waves are the major contributors to damaging ground shaking. The cyclic strain causes soil to contract 9 resulting in a pore pressure being generated in saturated loose soils. In general, an earthquake ground motion lasts for 20-60 sec with peak acceleration less than the acceleration due to gravity, g. • Detonation of explosive charges mainly produces relatively high frequency P waves that cause ground shaking and also cause pore pressure rise in saturated sands. Blast induced ground motion lasts for only a few milliseconds, and causes high accelerations close to the detonation point. These accelerations attenuate rapidly with distance from the blast. • In conventional practice, the liquefaction susceptibility of soil is assessed by laboratory testing or by in-situ testing. In laboratory conditions, cyclic loading tests could indicate inaccurate results unless those have been performed on absolutely undisturbed samples. In-situ testing used for the field performance correlation approach is good for sandy soils, but becomes difficult to apply in soils such as gravels, silts and clays. 2.4 Explosives and Explosions 2.4.1 Physical Characteristics Explosives are compounds or mixtures that are capable of undergoing extremely rapid decomposition thereby releasing gases. The gases require a space many times the original volume of the explosive. In general, there are two classifications of the use of explosives, commercial and military. The interest of this research is on the commercially available explosives, which include Dynamite, TNT, ANFO, and Slurries (Dick et al; 1993; Konya and Walter 1990). 10 Dynamite was first developed by Alfred Nobel and consists of a mixture of Nitroglycerin and Kieselguhr (Sio2 or diatomaceous earth). The Kieselguhr is used to make the explosive safer to handle and the end product was in granular form. At present dynamites can be found in various forms such as Straight-Nitroglycerin Dynamite, Ammonia Dynamite, Gelatin Dynamite and Semigelatins (Hemphill, GB. 1981). The Straight - Nitroglycerin dynamite is very much like the original dynamite, except that dynamite of this type contains additional ingredients that take part in the chemical reaction. Straight - Nitroglycerin dynamite consists of Sodium Nitrate with carbonaceous materials such as wood pulp added to absorb Nitroglycerin and to provide fuel for the detonation (Dick, 1993, DuPont, 1997). Ammonia Dynamite, also a granular material, was developed to replace Straight-Nitroglycerin dynamite. Ammonium Nitrate was used to absorb Nitroglycerin, as the Ammonium Nitrate is less sensitive to shock and heat which offers safer use in practice. However, this combination has lower density and is less resistant to water. Gelatin dynamite has a base of water-resistant "gel" made by dissolving nitrocotton in nitroglycerin. The nitrocotton gel is insoluble in water and tends to bind together other ingredients, making them water resistant. Semi-gelatin is designed to combine the resistance to water cohesiveness of gelatins with the lower cost of Ammonia Dynamites. Semi-gelatins tend to have poor storage and temperature resistance characteristics. Trinitrotoluene, commonly known as TNT, is commercially available in a pelletized form and has very good performance in wet holes (Hachey, et al; 1993). ANFO is a mixture of Ammonium Nitrate and fuel oil and it is the least expensive explosive available (Konya and Walter, 1990). ANFO is soluble in water and therefore, unless protected in some way, packages cannot be 11 used under water. To solve the ANFO and water problem, slurries were developed. These are a mixture of an Ammonium Nitrate base in an aqueous solution with a combustible fuel. These explosives are available as either cartridges or in bulk form. As noted above, these explosive types differ from each other and they can be categorized considering some characteristics such as detonation pressure, velocity of detonation (VOD), water resistance, density, and sensitivity as shown in the following table. Tablel - Typical detonation pressures (after Konya and Walter, 1990) Explosive Detonation Pressure (kilo bars) Density (g/cm3) Water Resistance Hazard Sensitivity Performance Sensitivity VOD m/sec Granular 20-70 0.8-1.4 Poor to Moderate Excellent 4500 Dynamite Good to high Gelatin 20-140 1.0-1.6 Good to Moderate Excellent 4700 Dynamite Excellent Packaged 20-45 1.0 Very Good Low Good to very 4800 ANFO good Cartridge 20-100 1.1-1.3 Very Good Low Good to very 4600 Slurry good TNT 75-150 1.1-1.2 Excellent Low Excellent 4900 It is clear from this table that the performance characteristics of different explosives are quite variable. For comparison purposes, researchers often present their results in terms of'equivalent TNT" as the TNT type explosive has almost been accepted as the standard explosive in the blasting community. TNT is a constituent of many explosives, such as amatol, pentolite, tetrytol, torpex, tritonal, picratol, and ednatol. In a refined form, TNT is one of the most stable of high explosives and can be stored over long periods of time. A high 12 explosive is characterized by the extreme rapidity with which its decomposition occurs; this action is known as detonation. When initiated by a blow or shock, it decomposes almost instantaneously, either in a manner similar to extremely rapid combustion or with rupture and rearrangement of the molecules themselves. One of the main advantages of TNT explosive is that it is relatively insensitive to blows or friction and also does not form sensitive compounds with metals. TNT is readily acted upon by alkali to form unstable compounds that are very sensitive to heat and impact. Despite, its reaction to alkali, the TNT explosive is the most commonly used explosive in the construction industry, mainly due to its excellent characteristics such as safe handling, high detonation pressure, and excellent performance in water. 2.4.2 Process of Energy Transfer An explosion is a sudden physical or chemical change of the state of a mass, accompanied by a release of energy (Hentych, 1979). It is convenient to consider an explosive as made up of internal and external components. The internal part is concerned with the process occurring in the material that releases the energy and the external part concerns the media surrounding the charge. Figure 2 shows a schematic view of an internal process of an explosion of a spherical charge. The radius of charge is Rcharge and the time needed to detonate for whole charge is T wh 0 i e charge- The explosion begins at point O at time To = 0 sec and at time Ti the charge can be divided into two zones as reacted (Zone-I) and unreacted (Zone-II) explosive. A thin layer of chemical reaction separates these zones and the reacted explosive zone consists of gases in a state of high pressure and high temperature. Where; Rcharge:- The radius of the charge Twhoie charge: - Time needed to detonate whole charge Zone-I: - Reacted Zone Zone-II: - Unreacted Zone Figure 2: The schematic view of an internal process of a spherical explosion Within less than a millisecond, when the charge has detonated completely, the volume of the reacted explosive occupies the same volume as the solid explosive material (Figure 3). An explosive discharges energy by releasing hot gases. The gas begins to expand, as the pressure of the gas is higher than the resistance of the surrounding medium. The phenomenon of existence of a gas front after a detonation of an explosive was verified by experimental studies on TNT explosives, carried out by Kedrinskii and Soloukhin (1960). They found that the explosive gases contained a mixture of 2H2 + O2 or 2C2H2 + 5O2. Others have documented that these gas pressures could be as high as 5 G P a to 15 GPa (Ivanov, 1967; Konya and Walter, 1990). 14 Figure 3 shows a schematic view of completely exploded sphere of explosive. Note that the explosive material has been substituted completely by the gas soon after the detonation. The explosive gas front expands until its pressure comes to equilibrium with the pressure in the surrounding medium. Hentych (1979) and Charlie (1985) note that the expansion of the explosive gas front occurs in accordance with Charles's gas law. i.e., as the volume expands the pressure drops. Figure 3: The schematic view of a completely exploded spherical explosive 2.4.3 Typical Response in Soils Due to an Explosion In general, soils have a complex structure. The mechanical properties of soils are widely varying and not well enough understood, especially in the case of the large stress zone close to an explosion. According to the above discussion, the gas pressure of the explosive gases 15 would have begun to act on the soil at all points of contact with the charge simultaneously. Accordingly, the energy released by the explosion in saturated sands radiates outward in the form of a dynamic pressure wave (Hryciw, R.D., 1986). The initial stress transient is very high relative to the strength of the material, and the material will deform plastically. As the dynamic pressure wave attenuates with distance, the strains caused by the waves become very small and the wave can be considered to be an "elastic wave". The elastic wave is the compressive (P-wave) wave. When this type of wave intersects a medium of different elastic properties, part of the energy is reflected into the first medium and part is transmitted into the new layer. Other wave types are created at such intersections (Dowding, 1985; Hryciw, R.D., 1986). Also, there are some other factors that might affect blast induced wave patterns including burial depth, and geometry of charge. For instance, if the charge is buried at considerable depth, then the explosion does not lead to ejection or considerable movement of the surface layer and in this case the wave pattern has time to develop sufficiently before the leading edge of the explosion wave reaches the free surface. This justifies the idealized consideration of explosives in an unbounded medium. 2.4.4 Previous Use of Explosives and Simulation of Earthquake Characteristics In civil engineering construction, explosives are often used for fragmenting rock or other material prior to its removal during the construction process. Also, explosives have been used successfully over the last 60 years to compact loose saturated sandy and silty soils by means of explosive energy, i.e. blast densification or explosive compaction. Florin and Ivanov (1961) used explosives to assess the liquefaction susceptibility of the soil. These investigators refer to their experiment as a "standard explosion at standard depth". 16 However, this method has not gained widespread acceptance as an index test. The main criterion for liquefaction was an index of the average settlement (>5 cm) of the surface of the soil in a radius of 5 metres from the place of explosion. A major work on the use of explosives to simulate earthquake loading of structures was performed by Higgins (1978). He used borehole line arrays containing decked explosives to cause earthquake-like loading of buried and above ground structures. The benefit of this work is that it demonstrated how the use of carefully controlled detonation sequences could provide adequate earthquake ground motion simulation in close proximity (50 to 100m) to blast arrays. Also, Higgins stated that P and S waves attenuate at large distances from the source in proportion to the distance R, while surface waves attenuate in proportion to -JR , and as a result, surface waves predominate at more distant ranges. What is important from these investigations is the knowledge that the P-wave is the dominant component in explosive induced ground motions whereas S-wave component is the dominant component in earthquake induced ground motions (Higgins). Higgins concluded that the use of explosives for simulation of earthquake like ground motion had limited success. He also noted that the simulation of all characteristics of an earthquake using explosives is impossible. 2.5 Prediction of Soil Response to Explosive Detonation 2.5.1 Empirical Procedures The energy input at a soil element is a function of explosive type, size and geometry of the charge, distance from the detonation, and the soil resistance. Input energy from an explosive detonation is commonly described in three different ways (Van Court and Mitchell, 1994): Hopkinson's number; normalized weight; and powder factor. However, the response at a 17 particular location is not completely described by the above, as the attenuation of the input energy is expected to be different at different sites. (i) Hopkinson's Number Commonly used guidelines for preliminary blast designs are based on Hopkinson's number, as shown in the results of work by Ivanov (1967). Ivanov carried out his work with concentrated or pinpoint charges at shallow depth and the relationship with Hopkinson's number may be valid in cases where his condition was adopted. Hopkinson's number may provide a reasonable description of the energy attenuation with distance, but it is only a function of the charge weight and distance from the blast, and does not account for the geometry of the charge. Despite these concerns, Hopkinson's number has been correlated with a number of parameters such as peak pore pressure, residual pore pressure, peak particle velocity, and settlement recorded during various blast experiments (Kok, 1981; Van Impe, 1986). Hopkinson's number is equal to the inverse of the Scaled Distance (R/W 0 3 3 3 ) (where R is hypocentral distance, m and W is the weight of explosive, kg) and is given by H N = W ° ' 3 3 3 / R (1) where, HN=Hopkinson's number in kg°' 3 3 3 /m W= weight of explosive, kg R= distance from the centre of the charge, m 18 (ii) Normalized Weight The normalized weight method has been proposed to describe the input energy a given distance from the detonation of a columnar charge (Dembicki et al., 1992; Imiolek, 1992). Hopkinson's number assumes a spherical charge. The normalized weight is calculated as follows: NW = (W 1 / 2/R)/h 0 1 / 2 (2) where: NW = normalized weight in kg 1 / 2 /m 3 / 2 W = Total weight of explosive in bore hole, kg R = distance from centre of the charge column, m ho= height of the sublayer, m (iii) Powder Factor The powder factor is defined as: PF = (W/R2)/(7ih0) (3) where: PF = powder factor in g/m3 W = weight of explosive charge, grams R= distance from the detonation, m ho= height of the layer treated, m All these methods are just different attempts to relate the volume of soil influenced by the explosive to the amount of soil treated. All are reasonable quantities to empirically relate to observed response. 19 2.5.2 Analytical Approach As an alternative to the above empirical approach, attempts have been made to model the detonation of a charge as the expansion of a spherical or cylindrical cavity. This theory has found application in a number of areas in geotechnical engineering, including Piles, Cone Penetration, Pressuremeter and Tunnelling. (i) Static Theory Cavity expansion theory was first developed for application to metal indentation problems (Sharpe, 1942; Bishop, 1945; Hill, 1950). The application of cavity expansion theory to geotechnical problems came later (Ladanyi, 1961; Gibson and Anderson, 1961;Vesic 1965: Palmer, 1972; Houlsby and Withers, 1988; Yu and Houlsby, 1991). Vesic was one of the first researchers to adapt this theory to a blast loading case. Although, an explosion in soil is a dynamic problem, Vesic proposed to consider the problem as a problem of static expansion. He considered the problem of an explosive charge deeply buried inside a cavity in a medium of known properties. He showed that for a cavity exposed to steadily increasing pressure P, there would be an ultimate pressure P u (limit pressure). The limit pressure, P u, is that at which the cavity would expand indefinitely. Detonation pressures of all explosives greatly exceed this ultimate pressure. Expansion will take place as long as this internal pressure exceeds or is equal to the ultimate cavity pressure P u. When the cavity reaches its ultimate radius R u, the internal pressure will be P u and equilibrium will exist, at least for a while. However, the above process occurs very rapidly and is complicated by dynamic effects. The stress and strain fields around the cavity are transmitted by wave propagation, the propagation velocity depending on the non-linearity of the soil. Additional waves will be generated at each change 20 in stiffness within the soil stratigraphy and at the ground surface, resulting in a complex ground response at any point within the zone of influence of the blast. Amplification and damping of the response may also occur. Vesic proposed procedures to determine the ultimate cavity pressure, P„, and the corresponding cavity radius, R„, as well as the radius of the plastic zone, Rp. The plastic zone represents the zone of soil in which shear failure has been induced. According to Vesic, the general solution for cylindrical and spherical cavities in a medium possessing both cohesion, c, and an angle of friction, <\>, is given by: P u = cFc + qF q (4) where q is the overburden pressure and F c , and F q are dimensionless cavity expansion factors. These factors are functions, of the angle of internal friction and rigidity index Ir, which is given by: Ir= E/(l+v)(c+qtan(|)) (5) where, E is the Young's modulus, v is the Poisson's ratio Vesic indicated that the radius of the plastic zone, Rp around the cavity is governed by this rigidity index. Vesic also showed that the static cavity expansion theory could be adapted to a blast loading case through the state of the gaseous products of the explosion. P u = C/ (V u AV) n (6) Where V u is the ultimate cavity volume, W is the explosive weight, n is a dimensionless number analogous to the adiabatic exponent, and C is a constant for the gaseous product. For a point charge and a spherical cavity, the above equation can be transformed into: 21 R» = Ci (W U i /P u 1 / J n ) (7) Where Ci is a constant. Accordingly, the theory proposed by Vesic indicates that this procedure can be used to determine Ru and Rp caused by an explosion in a soil medium using the angle of friction, <J>; the cohesion, c; the Poisson ratio, v; the Young's modulus, E, and the charge weight, W. The static theory ignores the dynamic effects. (ii) Dynamic Cavity Expansion Sharpe (1942) presented a mathematical solution for the blast model shown below based on an idealized statement of the problem and ignores many factors that exist in actual detonations of explosives. Elastic Medium Figure 4: An idealized model for the analysis of non-linear response to blast loads Sharpe assumed a spherical cavity of radius, a, within a homogeneous, ideally elastic, infinite medium of density p and compressional wave velocity, V p . In order to determine the elastic wave motion that results from application of an arbitrary pressure p(t) to the interior surface 22 of the cavity, the actual site of the explosion, which is a water-filled cylindrical cavity, was replaced by an empty spherical cavity. The assumption of a homogeneous, infinite medium excluded the consideration of reflection and refraction at elastic discontinuities near the source. The assumption of perfect elasticity is to suppose (if the cavity is so large or the charges used are so small) the elastic limit of the material surrounding the cavity is never exceeded. Sharpe established an expression to solve the above stated problem of the elastic wave displacement, which is a solution of the elastic equation of motion. Wu (1995) developed a finite element model for dynamic analysis using this idealized model concept in the time domain. In developing this model, Wu had to make several assumptions to model the dynamic soil response. This model ignores the possible effects of different cavity shapes due to blasting, the effect of residual gas and water pressures following the blast, and the effects of cavity collapse following blasting due to soil water flow. This model assumes an empty spherical cavity with uniform pressure acting on the interior surface of the cavity. In this model, the saturated soil medium is treated as a bonded single-phase material. A non-linear shear stress-strain relationship is used to simulate the shear resistance of the material and the viscous coupling is used to model the dependence of the shear resistance of the granular material on the time rate of the shear strain. This model uses linear superposition to predict peak particle velocities, shear strains and volumetric strains due to combination of detonations and uses the relationship proposed by Martin et al. (1975) for computing pore pressures. Wu incorporated this model in his computer code called "BLASTSET". This code is calibrated using site-specific test blast data. Once calibrated in this manner the code achieves reasonable agreement with field observations. Figures 5a shows the predicted peak velocities at distance 10m away from a detonation point. Figure 5b the predicted peak shear 23 strains at distance 6m and 10m away from a detonation point. It can be seen in this figure, the peak shear strain attenuates rapidly as the distance increases. Summary of Chapter 2 Energy released by fault rupture or by blasting propagates through soil as body waves or surface waves. The cyclic loading causes a tendency towards volume contraction which results in elevated pore pressures. Pore pressure rise causes a reduction in effective stress, loss of strength, stiffness and can result in liquefaction of sandy soils. The energy released by explosives depends on the chemical content of the explosives. The energy release occurred by a very rapid change of state from solid to gas, which is initially at very high pressure. The gas expands causing a disturbance to propagate outwards from the charge location. Attempts have been made to quantify this effect using, (i) Empirical methods relating the mass of the explosive charge to the volume of soil treated. Empirical relationships have been developed between charge density and significant aspects of ground response. (ii) Theoretical methods based on the theory of cavity expansion. The theory appears to capture the essential characteristics of the blasting phenomenon. As the simplified soil model cannot produce all of the complex interaction between soil and explosive, site-specific calibration is required to account for its effects. 24 a. Modelled (Computed) Peak Velocity- B L A S T S E T Code (10m away, 1.5kg) 1 0.75 0.5 a 0.25 cn i 0 'o •5 -0.25 > -0.5 -0.75 -1 ) O.C 02 O.C 04 O.C \ 06 o\ 31 O.C 12 O.C Time (sec) 0.15 0.05 -0.05 -0.15 b. Modelled (Computed) Peak Shear S t r a i n s - B L A S T S E T Code (10m and 6 m away, 1.5kg) Shear Strain -2P1-10m Shear Strain -2P1- 6m Figure 5: A typical Modelled Velocity and Shear Strains 014 Time (sec) 25 3. FIELD TRIAL BLAST AT SYNCRUDE J -Pit- CANLEX 3.1 Introduction As part of the Canadian Liquefaction Experiment (CANLEX), an attempt was made to trigger liquefaction by rapid construction of an embankment on a foundation layer of loose sand fill. This attempt was unsuccessful. Subsequently, a test blast was undertaken to determine whether explosives could be used as an alternative means of triggering liquefaction in the sand foundation. This experimental study comprised detonation of a series of explosives in soils. Acceleration and pore pressure were recorded at several locations. 3.2 Site Location The test site was situated in J-pit (an old borrow pit) at the Syncrude Canada Site near Fort McMurray, Alberta. The sand in J - Pit was deposited artificially in order to create a loose deposit. J - Pit is approximately 500m by 300m and has a maximum depth of about 10m. The site is essentially level with the ground water about 0.5m below the ground surface and the deposit was saturated below the water table. Figure 6 shows the layout of the site including the location of the test embankment and the area where explosive tests were conducted (letters from A to K indicate the location of test blast holes). 3.3 Soil Conditions The sand in J - Pit was deposited hydraulically in order to create a loose deposit. The hydraulically placed sand deposit in J-pit was first characterized by conducting 26 Cone Penetration Tests (CPT) (49thCanadian Geotechnical Conference, C A N L E X papers, 1996) in an area 80m by 80m across the site. Figure 7 shows the locations where the CPT tests were Figure 6: Plan View of the Syncrude J-Pit Site-CANLEX carried out. CPT26 and CPT27 are of most interest here as those were the closest to the monitoring locations of the blast program. CPTs were carried out by Conetec Investigations Ltd. of Vancouver, British Columbia. The cone had a tip area of 10 cm2 and a sleeve friction area of 150 cm2. During penetration, the tip resistance, q t , Sleeve Friction, fs, and Penetration Pore Pressure, u t, were recorded at 5 cm depth intervals. Shear wave velocity measurements (SCPT) were also taken at 1.0m depth intervals in CPT26 and CPT27. The above CPT logs are shown in Figures 8 and 9. These CPT logs indicate that the deposit in J - pit consists of silty sand to sandy silt with an average corrected tip resistance, q c i of 2.35 MPa in the test area (q ci =qc*(Pa/o"'vo) where q c is field measured tip resistance, Pa is O.lMPa if q c is measured in MPa, a ' v o is effective stress at the point of interest). Reference shear wave velocity measurements, taken during seismic cone penetration tests had an average normalized value, V s i , of approximately 127 m/sec in the test area (V si=V s*(100kPa/c>'vo)0 2 5 27 -Vs is the field measured shear wave velocity). In addition, Standard Penetration Tests (SPT) were performed in three locations and a Dynamic Cone Penetration Test (DCPT) was performed in one bore hole. The locations are shown in Figure 7. SPT tests were carried out with a standard 63.5 kg safety hammer dropping 760 mm using an electric winch. Energy measurements were recorded at several depths in SPT 4. The estimated average energy transferred was approximately 55 % of the maximum theoretical potential energy. All SPT measurements have been corrected for energy and overburden stress. The corrected (Ni)6o values and normalized V s i are shown in Figure 10. The average (Ni)6o in the test area was approximately 3.4 blows per 300 mm of penetration. Table 2 shows the classification properties of the deposit in J-pit such as grain size, minimum, maximum and in-situ void ratios, relative density, and unit weight. Table 2: Soil parameters at the site-CANLEX SOIL T Y P E SDLTY SAND T O SANDY SELT Grain size (D50) 0.17mm Grain size (D10) 0.08mm . Specific gravity of soils 2.62 Minimum and Maximum void ratio 0.461 and 0.986 Insitu void ratio 0.762 Relative Density 43.0% Moist unit weight 18.5kN/mJ Saturated unit weight 19.5 kN/m J Source : Data review report Mildred Lake and J - pit site, C. Wride and P.K. Robertson (1996) 28 r-r— a v. < z ,1 / V \ / T*-^E / 1 I -~ ' " f f f r 4-^ ?..:/• P>o" ••'••1% EL £ M 5 "S £ 1 ( ]- /I § \ / t-, c \\ \ vr. r -O vr, O o vr, vn CN o vr, O c ©* i n vr. r » ' O o vr. vn c c " vr. vr, CN © O vr, o o o ©* vr, CN oo" m o c CO vr, vr, r -oc 00* vr, O VTi X. eo" vr, vn CN 00. oc vr, o o ° ° . oo* vn vr, f-oo" vr. O vn oo* vr, CN 00* vr, O o 00* vr, vn e> vn2 vq oo" vn Figure 7: Locations of CPT, SPT and DCPT Tests Carried Out at Syncrude J-Pit C A N L E X (Source:- C A N L E X Papers, 1996) 29 o o CJ 1. J A A ^ ^ L ; o o" i n i o o" *-\ i o m" i ( I U ) mdBQ Figure 8: CPT Sounding test results at # 26-GTTH (9516C-26) (Source:- C A N L E X Papers, 1996) — c " i 2 d >< a S O CJ I 30 o o CJ c n - -- J i I 5 e * "a. LD I o o* I o i n I o (IU) Li^daa Figure 9: CPT Sounding Test Results at # 27 -GTTH (9516C-27) (Source:-CANLEX Papers, 1996) 31 Figure 10: (Ni)60, V,i Versus Depth- CANLEX (Source:-CANLEX Papers, 1996) 32 3.4 Test Program and Charge Layout of CANLEX Blast The blasting program was designed to provide data from single and multiple blasts in order to study the characteristics of ground motions and pore pressure response. This test blast program included a total of 6 blast events. The blast locations are shown in Figures 11 to 13. Blasts #1, #2, #4 were single charge events. Blasts #3, #5, consisted of three charges (decks), detonated sequentially in single boreholes. Blast #6 contained four blast holes each with three charges. These charges were placed at three different depths in the each blast hole and the detonation sequence was from bottom up. All charges used were of cylindrical type. The layout details are given in Table 3. Figure 14 shows a typical blast hole layout. Table 3: Description of blast events-CANLEX Blast Number Location Plan Charge Depth (m) Charge Weight (kg) Time Delay mS 1 A 6.0 1.0 -2 D 6.0 1.5 -3 B 9.0,6.0,3.0 3.0,1.5,1.5 250mS 4 L 4.0 3 -5 K 9.0,6.0,3.0 4.5,3.0,1.5 250mS . 6 J, I, H, G 9.0,6.0,3.0 4.5,3.0,1.5 250mS-between decks 500mS-between holes Table 4: Schedule of blast events-CANLEX Blast Number Location Plan Charge Location Charge Weight (kg) 1 July 24/97 A System did not trigger 2 July 24/97 D System triggered correctly 3 July 24/97 B System triggered early 4 July 24/97 K Manual trigger 5 July 24/97 L Manual trigger 6 July 24/97 J, I, H, G Manual trigger 33 o CO m O Q . + + CM CM CJ l I 00 CD Figure 11 : Blast Locations of # 1, 2, 3 -CANLEX 34 Figure 12: Blast Locations of # 4, # 5 - C A N L E X 35 LU Q o >• cn O z t-LU co < m x LU < O CQ CD + o O) _+ CN CN O + CD m O CN O % X31NV0 Figure 13:Blast Locations of # 6-CANLEX 36 * blast hole depth= 9m below ground level * blast hole PVC casing with 100mm ID * gravel stemming between each deck * bottom up detonation sequence * time delay, 250ms between decks and 500ms between blast holes Figure 14: Typical Blast Hole-CANLEX 3.5. Characteristics of Explosives-SYNCRUDE J - Pit Test Blast. Dyno Powermite 2.5 x 16 was the explosive used in this test blast. It is a TNT type explosive packed in cartridges. The velocity of detonation (VOD), and the density of these explosives are 4900 m/sec and 1.14-1.24 g/cm3, respectively. These explosives were placed in blast holes, which were cased with 100mm PVC tubing. The procedure for preparing the holes was to fill the casing with pea gravel to the specified depth of the first deck. The explosive with blasting cap was then dropped through water to the bottom of the hole. More pea gravel was added to get to the next deck, or to the top of the hole if it was a single deck blast hole. Delays of 250ms were used for multi-deck blast holes and for the multi-hole blast. A programmable blasting box (REO system 5000) was used to provide accurate time delays of 500ms between holes in blast # 6. 37 3.6 Instrumentation The response due to each blast event was recorded at several location using different instruments. These instruments included Accelerometers, Pore Pressure Transducers, and Geophones. The accelerometers and pore pressure transducers were located at Pl and P2 locations, 6m below the ground level. These P l and P2 locations were not consistent for all blast events carried out during C A N L E X test program. The positions of P l and P2 were relocated from one blast event to another (Pl and P2 are shown in Figures 10 to 12). In addition, the locations of surface geophones are also shown in Figure 12. Accelerations were recorded by piezoresistive accelerometers mounted in a triaxial package allowing acceleration in three orthogonal directions to be recorded. These accelerometers were installed at the end of a 6m hollow rod that was pushed in to the soils. The accelerometer installed at P l location had the resolution of 0.305 g / bit, 0.325 g / bit in the horizontal direction (X and Y) and 0.375 in the vertical direction (Z) (Conetec Investigations Ltd., 1997). The accelerometer installed at P2 location had the resolution of 0.0034 g / bit, 0.0023 g / bit, in the horizontal directions, (X and Y) and 0.0034 g / bit in the vertical direction, Z. Unfortunately, the maximum ranges of these accelerometers were not documented. The set-up at P l and P2 also contained a dynamic pore pressure transducer to monitor pore water pressures. The maximum range of the pore pressure transducer was 500 psi (345m of water) at Pl and 100 psi (69m of water) at P2. Additionally, two cone penetrometers with pore pressure transducers of 500 psi capacity were installed at 6m below the ground surface at C - 22 and C - 56 locations (Figure 11-13). Three Adara type piezometers of 10, 20, 100 psi capacities respectively were also installed to monitor pore pressure response. The data 38 acquisition system for this test blast consisted of a dual PC based unit, both equipped with high-speed analog digital (A/D) cards. Each PC was able to handle up to 8 channels of data. One PC recorded accelerations in the X, Y, Z directions and pore pressures at PI and P2 locations. The other one recorded data from two Cone penetrometers, two Geophones and Adara push in piezometers. The selected sampling rates for the experiment were 31,250 Hz and 5,000 Hz for blast numbers 1,2,3 and blast numbers 4,5,6 respectively. The rationale for selection of the instrumentation was not recorded. 3.7 Post blast Observations in the J-Pit Although the C A N L E X experiment setup was designed to record data from six different blast events, Table 4 shows that the measuring system was able to gather data from only blasts events # 2, # 4, # 5, and # 6. During blast # 1 the data acquisition system was not triggered. The influence on blast # 4 from # 2 and # 3 in terms of pore pressure was minimal as these two blasts were detonated one day prior to the # 4 detonation. However, the soil response induced by blast # 5 could have been under the influence of pore pressure induced by # 4, as the time difference between these two blasts was about one hour. The induced pore pressures by blast # 5 are considered to have had little effect on the soil response induced by blast # 6, as blast # 6 took place three days after the detonation of blast # 5. Photographs taken during one of above blast events are included in Appendix-A. They clearly show the appearance of sand boils after the blast event. Sands boils did not begin to form until about 20 minutes after the blast. Sand boil formation continued for about 60 minutes after they first began forming. 39 4. DATA ASSESSMENT FROM C A N L E X E X P E R I M E N T 4.1 Introduction In this chapter, the quality of the data is assesses in order to determine which data may be used for detailed analysis. This included examination of the performance of the instrumentation including an assessment of its adequacy for recording the true ground response. Techniques of signal analysis were used to gain insight into instrumentation response. 4.2. Assessment of Instrumentation 4.2.1 Effect of Instrumentation Characteristics on Data Measured The instrumentation was installed to measure soil response to the detonation of charges. The accelerometers were mounted in a housing close to the tip of a 6m long hollow steel rod. The instruments were inserted into the ground by pushing. The rod remained in place during the test. Consequently, the soil around the probe was altered from its in situ state by disturbance. In addition, the response of the soil was affected by the presence of the measuring system. The measured response may be influenced by the vibration characteristics of the rod and surrounding soil. Such a response will not be the true response of the soil in the far field. The rod and disturbed soil are considerably stiffer than soil in the free field. To obtain a reasonably accurate response of the soil to the blast, the vibration characteristics of the rod need to be removed from the recorded trace. Some idea of the effect of rod vibration could have been eliminated easily, if the frequency characteristics of the rod were determined by carrying out a field test in actual conditions, prior to the actual experiment. Unfortunately, this type of test had not been carried out during the C A N L E X experiment. Such a process 40 was followed by (Bogdanoff, I; 1996) who reported the usefulness of such a test and the procedures of carrying it out to obtain the resonant frequency of a measuring system. The result of that field trial test was used as a cut off frequency to filter recorded data during the actual test. He revealed that there was a significant difference between filtered and unfiltered records. A band pass filter is one of the tools in signal processing technology is used to filter unwanted frequency components of a signal for a given frequency range. The range of the band pass filter should be selected in such a way that it preserves the desired frequencies of the signal. Note that the determination of the correct range of the band past filter in this blast event is difficult due to the reasons discussed above. However, a mathematical equation in mechanics can determine the natural frequency characteristics of a hollow rod. Note that these mathematical procedures cannot determine the accurate frequency characteristics of a rod subjected to the effects of surrounding media. In this thesis, the mathematically derived vibration characteristics of the rod are used as a cut-off frequency to filter the measured data. This is only for the demonstration purposes to show the significance of awareness of frequency characteristics of the instrument housing prior to the main test. The resonant frequency of the instrumentation housing described above was estimated as 1.5-1.8kHz. The measured data were filtered using this frequency as a cut-off frequency. The results of this analysis are attached in the Appendix-B. These figures indicate some difference between the unfiltered and filtered version of time histories. This result suggests that the resonant frequency of the instrumentation housing can affect the measured data. However, as the actual frequency characteristics of the instrumentation housing were not 41 known prior to the test blast, unfiltered data were analyzed. In addition, published data of the previous researchers do not address this issue directly and it is assumed that these conclusions were based on the raw experimental data. 4.3 Assessment of the Quality of Acceleration Data- CANLEX Test Blast The blast induced ground motion during the C A N L E X trial blast was recorded by the accelerometers at P l & P2 locations. Sampling rates used during this experiment were 31500SPS(samples per second) and 5000SPS. According to sampling theory, 31500 SPS can capture a response accurately of signals with maximum frequencies 15800Hz or less and the sampling rate 5000 SPS would be accurate for signals with maximum frequencies 2500Hz or less. Figure 15a shows a recorded acceleration response at P l location due to blast # 2 of a 1.5kg explosive that was detonated at 6m below the ground level and 10m away from the measuring location. The general appearance of the trace shows that there are two significant wave packets, although only one detonation took place. The reason for this behaviour is unclear. But, Biot (1941,1956, 1962) demonstrated the existence of two compressional waves and one distortional wave in blast induced ground motions. Biot's theory has been confirmed in many experiments, such as those conducted by Plona (1980), Berryman (1980). Biot indicated that the velocity of the waves of the first kind is around 1600-1900 m/sec for sands and the velocity of the waves of the second kind is around 600-900 m/sec. An analysis of this trace (Figure 15a) also revealed that the first peak of the first wave packet travels with a velocity of about 1580 m/sec. The peak of the second wave packet travels at 800 m/sec, Although, the C A N L E X data appears to confirm Biot's theory, it is difficult to explain the 42 cause for this kind of behavior by analyzing only a single response at a particular site. Data from other blast events performed at different sites could help in finding a reasonable explanation for this behavior. Figure 15b shows the acceleration time histories measured at P2 location due to blast # 2. The sampling rate was also 31500 SPS. Unfortunately, as can be seen in this figure the recorded acceleration time history induced by blast # 2 was not measured completely at P2 location. The large peaks of the record were cut off. It is obvious that the range of the instrument used at this location was not large enough to capture the induced ground motion, although the sampling rate used appeared to be adequate. These acceleration responses with clipped data cannot be used for integration to obtain velocity and displacement, but can be used for determination of wave arrival times. Figure 16 illustrates the acceleration time histories recorded at PI and P2 location due to blast # 4. The sampling rate of this blast was 5000 SPS, much lower than the sampling rate of blast # 2. Although, the recorded acceleration time histories show an apparently reasonable response, it is necessary to understand if the selected sampling rate was adequate. 4.3.1 Sampling Rate In order to assess the effects of these different sampling rates on the measurements a data set gathered at each sampling rate was analyzed using the Fast Fourier Transform (FFT) (Brault J.W and White O.R.; 1971). This technique converts a signal in the time domain into the frequency domain. This allows comparison of frequency characteristics between these signals. Figures 17 and 18 compare the measured horizontal accelerations of blast events # 2 and # 4 using this technique. Figure 17a illustrates the frequency characteristics of the signal 43 induced by blast # 2 at Pl location. Note that the main frequencies are in the range 0-4000Hz, with a second peak in the range 5500-8000Hz. In addition, this figure reconfirms that the sampling rate used during blast # 2 was quite sufficient to record all type of frequencies of a blast induced signal as no response was seen beyond 12000Hz. Figure 17b illustrates the frequency characteristics of the signal induced by blast # 2 at P2 location. P2 was located 28.3m away from the blast detonation point. The main frequencies are in the range 0-3000Hz, and very little response was seen beyond this point. It is clearly seen that the frequency content decreases with distance. Figures 18a shows the frequency characteristics of the signal induced by blast # 4 at Pl location. This blast also contained a single charge and the distance to measuring location from the blast was almost similar to that of blast # 2. The large frequency peaks occur during frequencies 500- 1000Hz and between 1400-1700Hz. Also, it is clearly seen that there was some response beyond the point of 2500Hz, which had not been captured due to the inadequacy of the sampling rate used during blast. These results indicates that the near field contains high frequencies and the sampling rate used at blast # 4 was not adequate for near field data measurements. However, frequency analysis of blast # 2 (Figure 17a) indicates that the maximum response of the signal occurs during frequencies 0-2500Hz in the near field. According to this, the sampling rate used during blast # 4 was able to capture the maximum response of the signal in the near field. As this research mainly interests in peak ground motions induced by a blast loading, the data recorded using this sampling rate were suitable for integration to obtain velocity and displacement. Figure 18b shows the frequency characteristics of the signal induced by blast # 4 at P2 location, which was 47.7m away from 44 blast detonation point. The main response occurs during frequencies 0-800Hz. These results indicate that the far field appeared to contain relatively low frequencies and the sampling rate used at blast # 4 was adequate for far field data measurements. Blast # 5 was a single sequential blast event that contained multiple decks (4.5 kg, 3 kg, 1.5 kg) in a blast hole. These decks (charges) were detonated bottom to top in sequence with a time delay of 250ms. The sampling rate for recording of blast #5 was also 5000 SPS and accelerometer was located 10m away from the detonation point. Figure B l (Appendix-B) shows recorded acceleration time histories at PI location for each deck of 4.5kg, 3.0kg, and 1.5kg respectively. According to the above discussion, these traces also can be integrated to obtain velocities and displacements. Figure B2 shows the acceleration time histories recorded at the P2 location during blast # 5. Note that these time histories shown in Figure B2 indicate that the instrument was malfunctioning. Blast event # 6 was a sequential blast array that contained 4 blast holes, spaced 10m apart (see layout shown in Figure 11). Each blast hole contained three explosive decks (4.5kg, 3 kg, 1.5kg at 9m, 6m and 3m depth, respectively) and each hole was detonated from bottom up with 500ms delay between holes and the delay between decks was 250 ms. The acceleration response induced by blast #6 was recorded at PI and P2 locations at 5000 SPS. Figure B3 illustrates the recorded acceleration time histories during blast event # 6 at PI. Figures B4 to B6 show a zoomed version of this acceleration response recorded at the PI. These acceleration time histories can be used for integration to obtain velocities and displacements. However, the recorded acceleration response at P2 location (Figures B7 to B9) also shows large peaks of the record were cut off. These acceleration responses with 45 clipped data cannot be used for integration to obtain velocity and displacement, but can be used for determination of wave arrival times. 4.3.2 Orientation of Accelerometers Blast induced ground motion during this blast event was recorded as acceleration using 3 orthogonal accelerometers. However, the orientation of these orthogonal axes of accelerometers was not documented. Since the orientation of accelerometer represents some important aspects of whole analysis, it was necessary to employ a reliable method to determine an orientation of a recorded signal. The following method (Ventura 2000, Personal communication) can be used to determine the orientation of an instrument (X & Y direction), if the following conditions are met. • Availability of both signals in the X and Y direction at a particular point • Location of instrument relative to the blast is known The methods finds the angle at which the maximum power spectrum of the signal (Code is attached in Appendix C). For instance, if the X direction of the instrument lies on the direction of approach of the signal (hypocentral line- line between the source and the accelerometer), the maximum power spectrum of the signal should be seen on "0" angle. Accordingly, the Y-axis should be perpendicular to the X-axis, which is a line between the source and the accelerometer. If, the direction of the X-axis is not on this line, then the above-mentioned angle indicates how far the X-axis is rotated from hypocentral line. For instance, the blast # 6 contained 4 blast holes and the direction of X and Y accelerometers could be established from each blast hole. Figure C3 (Appendix-C) shows the results of this 46 analysis. The orientation established by this method for locations PI & P2 resulted in consistent response to each blast event of blast #6. In summary, • Instrument housing prevented free movement in vertical direction and other directions likely affected by vibration of steel rod. Orientation of X and Y directions of accelerometer can be estimated by maximizing power spectrum. • The sampling rate (31500SPS) used during the blast # 2 was sufficient to capture all significant characteristics of motion in the near field and the far field as well. The data recorded at P2 location, were incomplete due to the selected range of the measuring instrument being too small. But due to the rapid attenuation of high frequency components, 5000SPS appears adequate at > 45m from blast. 47 a. Acceleration Time History in X Direction- Blast 2 at P1 ; CANLEX (10m. 1.5kg) 3000 -3000 Time, sec tu o o < 300 200 100 0.D1 -100 -200 -300 b. Acceleration Time History in X Direction- Blast 2 at P2 ; CANLEX (28.8m, 1.5kg) 0.02 .03 Time, sec Figure 15: Typical Blast acceleration Time History 48 a> a> o o < Acceleration Time History in X Direction-Blast 4 at P1; CANLEX (10.7m, 3kg) 3000 2000 o 1000 o a> in 2.k7 -1000 -2000 -3000 2.485 2.ft9 Time, sec b. Acceleration Time History in X Direction-Blast 4 at P2; CANLEX (47.7m, 3kg) 300 -j 1 1 1 200 u "G 100 CD g -100 o < -200 -300 Time, sec Figure 16: Acceleration Time Histories- Blast #4 at P1 and P2 49 Fourier Spectrum Amplitude- Blast # 2 at P1-X-Direction (1.5kg, R=10m, Burial Depth, 6m) 150000 5000 10000 Frequency (Hz) 15000 25000 20000 | 15000 a> •a 3 E 10000 < 5000 b. Fourier Spectrum Amplitude- Blast # 2 at P2-X-Direction (1.5kg, R=10m, Burial Depth, 6m) I ill Frequency (Hz) Figure 17: Fourier Spectrum Amplitude -Blast # 2 P1 & P2 50 50000 -40000 -E30000 -a> T3 3 a J20000 -< 10000 -0 -( a. Fourier Spectrum Amplitude- Blast # 4 at P1 -X-Direction (3kg, R=10.7m, Burial Depth, 3m) ) 500 1000 1500 20 Frequency (Hz) 00 2500 b. Fourier Spectrum Amplitude-Blast # 4 at P2-in X Direction (3kg, R=47.7m, Burial Depth, 3m) 250 , 1 1 ; , 200 0 500 1000 1500 2000 2500 Frequency (Hz) Figure 18: Fourier Spectrum Amplitude -Blast # 4 P1 & P2 51 4.4 Assessment of the quality of Pore Pressure Data - CANLEX Test Blast During the C A N L E X trial blasts, the induced pore pressure response from each blast event was recorded at several locations (Pl, P2, C-22 and C-56 locations), 6m below the ground level. Instruments at Pl and P2 also recorded induced ground response as acceleration. Blast # 2-detonation of a single 1.5kg charge at 6m below ground level, was the first successful event of this program. Figure 19a shows the recorded acceleration time histories at the P l location of blast # 2 and the corresponding pore pressure time histories to this blast are shown in Figure 19b. Comparison of these two plots indicates that the peak pore pressure occurs after the peak acceleration. Furthermore, the initial peak pore pressure, (or maximum dynamic pressure) decays rapidly with time. The maximum dynamic pressure was 128m of water (1.28MPa) during this blast event. Some residual pore pressures appear at the end of the pulse (Figure 19c). The residual pore pressure is usually estimated by subtracting induced pore pressures at the end of the record from the initial pore pressures prior to the blast event. Figure 19c shows the recorded pore water response at P2 (28m away from the blast) indicating that the instrument used at P2 also was able to capture the complete response. Table 5 shows the summary of these pore pressure pulses. Table 5: Observed pore water pressure parameters-Blast # 2 at P l Charge Weight Distance (m) Peak Peak Pore Residual Pore (kg)/ Burial Acceleration Pressure Pressure Depth, m (g) (m) (m) 1.5kg/6m P l - 10 259 128 2.8 1.5kg/6m P2- 28 Clipped data 11.5 0.8 52 The second blast event of C A N L E X test was blast # 4. The pore pressures were also recorded at PI and P2 and those locations are shown in Figure 11. Location PI was 10.7m away and P2 was 47.7m away from the blast and the buried depth of the charge was only 3m below the ground level. Figure 20 shows the pore pressure response at both locations and the measured response at both locations shows that the range of the transducers was acceptable. Table 6 shows the summary of these pore pressure pulses. Table 6: Observed pore water pressure parameters-Blast # 4 at PI and P2 Charge Weight Distance Peak Peak Pore Residual Pore (kg)/ Burial (m) Acceleration Pressure Pressure Depth, m (g) (m) (m) 3kg/ 1.5m PI- 10.7 220 66 1.8 3kg/ 1.5m P2- 47.7 6.5 4 0.8 The third blast event of C A N L E X test was blast # 5 which included 3 decks of 4.5, 3.0, 1.5 kg. The pore water response due to this blast was recorded at PI (10m away) and at P2 (42m away). The pore water pressure response at PI location is shown in Figure 21. Note that this plot indicates three large peaks. Those were induced by detonation of three explosive decks used in this blast. The Figure 21b (zoomed version of 21a) shows that the general appearance of the trace was reasonable although some data points were clipped during recording. It can be noted that these clipped data would not affect on any calculations of excess peak pore pressures or residual pressures. Table 7 shows the summary of the pore pressure pulse. 53 Table 7: Observed pore water pressure parameters-Blast #5 at P l Blast # Charge Weight (kg)/ Burial Depth, m Distance (m) Peak Acceleration (g) Peak Dynamic Pore Pressure (m) Residual Pore Pressure (m) #5-1 4.5kg/9m 10.4 210 199 1.92 #5-2 3kg/6m 10.0 110 134 1.95 #5-3 1.5kg/3m 10.4 50 101 2.31 However, the recorded pore pressure response at the P2 location shows an unusual trend, as pore pressure goes strongly negative and remains so until the second detonation (Figure 22). This could be due to complex interaction between the rod housing of the instrumentation and the soil. For that reason, this record may not suitable for further analysis. Blast # 6 was the last blast event of this test program. It contained 4 blast holes with 3 decks at different levels in each blast hole. The pore pressure response induced by event #6 was monitored at P l & P2 locations 6m below the ground level. It is shown in Figures 23 and 24. The J, I, H, G symbols indicate the blast hole locations (see Figure 12) of blast # 6. For instance Ji indicates the response from the first (deepest) deck at this J-location, Ii indicates the response from the first deck (deepest) at this I-location and h indicates the response from the second deck at this location, etc. For instance, symbol of I1+I2 (Figure 23) indicates the detonation of two explosives occurred together. This is due to time delays used in this blast event, 250 ms between decks and 500 ms between blast holes. Due to this reason, the bottom deck (each hole contained three decks and there were four blast holes) would have been detonated together with the top deck of the adjacent hole. The recorded pore pressure at this 54 blast also shows that the general appearance of the trace was reasonable and also can be seen some data points were clipped during recording. These clipped data also would not affect on any calculations of excess peak pore pressures or residual pressures. Table 8 shows the summary of the pore pressure pulse. Table 8: Observed pore water pressure parameters-Blast # 6 at PI Blast # Charge Weight (kg) / Burial Depth, (m) Distance (m) Peak Acceleration (g) Peak Dynamic Pore Pressure (m) Ji 4.5kg/9m 22.5 67 39 I1+I2 4.5kg/9m; 3kg/6m 14.4 114 104 I3 1.5kg/3m 14.4 66 56 h 3kg/6m 22.3 42 14 J3+H, 1.5kg/3m: 4.5kg/9m 10.4 167 249 H 2 3kg/6m 10 105 64 H3+G1 1.5kg/3m: 4.5kg/9m 14.4- 192 118 G 2 3kg/6m 14.2 65 110 G 3 1.5kg/3m 14.4 93 57 The recorded pore water response at P2 of this blast was of good quality as it is shown in Figure 24a. Figure 24b shows the zoomed version of recorded pore pressures during this blast event and symbols of J, I, H, G indicate responses due to blasts at particular locations. Table 9 shows the summary of the pore pressure pulse discussed above. Note that the trend of residual pore pressures indicates that the elevated pore pressures had been maintained for a 55 considerably longer period of time than a residual pore pressure induced during a single blast event. Table 9: Observed pore water pressure parameters-Blast # 6 at P2 Blast # Charge Weight (kg) / Burial Depth, m Distance (m) Peak Acceleration (g) Peak Dynamic Pore Pressure (m) Residual Pore Pressure (m) Ji 4.5kg/9m 36.1 Clipped Data 10.41 1.73 I1+I2 4.5kg/9m; 3kg/6m 31.7 Clipped Data 15.94 1.8 I3 1.5kg/3m 36 Clipped Data 11.51 1.9 h 3kg/6m 36.1 Clipped Data 9.67 1.6 J3+Hi 1.5kg/3m: 4.5kg/9m 20.22 Clipped Data 21.15 2.4 H 2 3kg/6m 31.7 Clipped Data 13.29 2.96 H 3 + G 1 1.5kg/3m: 4.5kg/9m 30 Clipped Data 15.12 2.44 G 2 3kg/6m 31.7 Clipped Data 21.93 2.4 G 3 1.5kg/3m 30.1 Clipped Data 10.51 2.2 56 a. Acceleration Time History in X Direction- Blast 2 at P1 ; C A N L E X (10m, 1.5kg, burial depth=6m) Time, s e c 1 5 0 •g-100 "oT 3 tn g 5 0 o °- 0 - 5 0 Pore Pressure Time History-Blast 2 at P1-(10m, 1.5kg, burial depth=6m) 11 =0.002272 0 . 0 0 5 0.01 Initial Pore Pressure Level 0 . 0 1 5 0 .02 15 f 10 2. 5 0.01 c. Pore Pressure Time History-Blast 2 at P2-(28.3m, 1.5kg, burial depth=6m) t2=0.0 14816 0 . 0 1 5 0 .02 Time.sec 0 . 0 2 5 0 . 0 3 Figure 19: Typical Blast acceleration Time History and Pore Pressure Time Histories 57 3000 2000 "a o | 1000 1 = 0 o e 2. % -1000 (J u < -2000 -3000 a. Acceleration Time History in X Direction-Blast 4 at P1; C A N L E X (10.7m, 3kg, burial depth=3m) w • 2. 18 2. 19 2 5 2. 51 2. 52 2. 53 2. 150 100 50 -50 2.47 b. Pore Pressure Time History-Blast 4 at P1 (10.7m, 3kg, burial depth=3m) 2.48 2.49 2.61 Initial Pore Pressure 2.52 2.53 Time, sec 2.54 10 E. 3 Q. o Q. 2.49 c. Pore Pressure Time History-Blast 4 at P-(47.7m, 3 kg, burial depth=3m) Initial Pore Pi essure 2.5 2.51 2.52 2.53 2.54 2.55 Time, sec 2.56 Figure 20: Acceleration and Pore pressure Time Histories- Blast #4 58 a. Pore Pressure Time Histories-Blast 5 at P1, 10 m Away (4.5kg,3kg, 1.5kg- 9m, 6m, 3m in depth respectively) 250 200 E ~o 150 3 to 10 a> 100 o Q. 50 4.5kg 3 kg 1.5 kg 1 tial P o r e P r e s s u r l / v ^ r v * - . : . . . - . . . . . , : e 1.8 2.2 Time, sec 2.4 2.6 Zoomed Version of Pore Pressure Time Histories-Blast 5 at P1, 10 m Away (4.5kg,3kg,1.5kg- 9m, 6m, 3m in Depth Respectively) 15 E 1 0 3 <0 to a 0. 0> o ft- 5 1.8 2.2 Time, sec 2.4 2.6 Figure 21: Pore Pressure Time Histories-Blast #5 at P1 59 Pore Pressure Time Histories-Blast 5 at P2, 42.2m away (4.5kg,3kg, 1.5kg- 9m, 6m, 3m in Depth Respectively) 15 , . , . , , Figure 22: Pore Pressure Time Histories-Blast #5 at P2 60 a. Pore Pressure Time Histories- Blast 6 at P1 (10 m away from the blast array) J3TI-1. , L+U H 3+( '1 G 2 . l 3 H 2 G 3 1 J 2 . 1 , V- > y • fr" 3 3.5 4 4.5 5 Time(sec) b. Zoomed Version of Pore Pressure Time Histories- Blast 6 at P1 (10 m away from the blast array) - 4krJ^- i / <m r r n I i In — - -itial P o re P r e s s u r e 3 3.5 4 4.5 5 Time(sec) Figure 23: Pore Pressure Time Histories- Blast #6 at P1 61 a. Pore Pressure Time Histories- Blast 6 at P2 (30 m away from the blast array) Time(sec) b. Zoomed Version of Pore Pressure Time Histories-Blast 6 at P2 (30 m away from the blast array) 15 =. 10 E "2» 3 in in £ 0. a> i_ o a. o -\ Initial Pore Pressure 3.5 4.5 Time(sec) Figure 24: Pore Pressure Time Histories- Blast #6 at P2 62 5. DATA ANALYSIS 5.1 Introduction The initial objective of this study was to examine blast records to determine whether it would be possible to create over earthquake-like ground motion by sequential detonation of charges. Interference between wave trains at a desired location could be used to simulate earthquake ground motion. It was first necessary to examine the characteristics of earthquake motions and blast-induced ground motions. Earthquake time histories recorded on liquefaction prone sites were selected for consideration. The selected earthquake records were from the Imperial Valley (1940) and Loma Prieta (1989) earthquakes: time histories recorded at El-Centro site and Treasure Island site during those earthquakes, respectively. The following table provides general information about both time histories. Table 10: Description of earthquakes Site Instrument Location Distance from Epicentre (km) Focal Depth (km) Magnitude Peak Ground Acceleration, (g) Peak Ground Velocity, (m/sec) El-Centro Alluvium (>300m) 8.2 9 6.9 0.34 0.28 Treasure Island Alluvial deposits (100m<) 97.6 17.6 7 0.16 0.15 63 5.2. Characteristic Wave Traces 5.2.1 Earthquake Ground Motions Figure 25 shows the Acceleration time histories recorded at ground surface during El -Centro site during the Imperial Valley earthquake. As the table above shows, the recording station was located at ground level on more than 300m deep alluvium soil and 8.2 km from the epicentre (note that the R on the Figure 25, is the hypocentral distance). Figure 26 shows the Acceleration time histories recorded at Treasure Island site during the Loma Prieta earthquake. The recording instrument was located at ground level on more than 100m deep alluvial deposits and at about 97.6 km away from the epicentre. Figures 27 and 28 show the velocity and displacement time histories that were obtained by integration of the acceleration time histories. The time histories recorded at Treasure Island site indicate a long period, low frequency wave type, while time histories at El-Centro station shows a short period, high frequency content wave response. The frequency content of the signal can be quantified using signal-processing techniques. The Fast Fourier Transform (FFT) was used for this purpose. Figure 29a shows a typical result of this type of analysis carried out for the earthquake record at El-Centro station. Note that this figure shows that all large peaks are concentrated as a group and it is difficult to distinguish the actual frequency characteristics of the signal. In order to see salient features of such results, the acceleration data were smoothed using a 300-point moving average scheme. Figure 29b shows the smoothed data. This smoothing can decrease the amplitude of the FFT drastically, but it indicates the frequency characteristics of the signal reasonably well. Figure 30 shows Fourier spectrum of the smoothed data for the 64 earthquake record at El-Centro site and Treasure Island sites. The Amplitude of the Fourier Spectrum for each record shows that the main response of an earthquake lies in the low frequency range. The Treasure Island record shows its predominant response between 0-2Hz while the El-Centro record shows its response in between 0-5Hz. The Treasure Island record shows very little or no response beyond 5Hz. The response at El-Centro station continues to spread beyond 10 Hz, and even shows some response in the range close to 25Hz. This suggests that higher frequency waves attenuate as distance from the epicentre increase. 5.2.2 Blast Loading Figures 17a and 17b show the frequency characteristics of the signal recorded at P l and P2 locations due to blast #2 comprising a 1.5kg of charge weight. The distance to the detonation point from these location were 10m and 28.3m respectively. Comparison of the spectra indicates a very rapid attenuation with distance of the amplitude and reduction in the frequency content of the motion caused by the blast. The frequency content of the peak response reduces very rapidly with distance, but is still much higher than for the earthquake records (Figures 29, 30). Figure 18 shows the frequency characteristics of the signal recorded at P l and P2 locations due to blast # 4. The distance to the detonation point was 10.7m and 47.7m respectively. This blast contained 3kg of charge weight and the burial depth was 3m below the ground level. Comparison of these two spectra also indicates a very rapid attenuation with distance of the amplitude and the frequency content. These results indicate that the ground motions measured close to the blast location (near fields) contain high frequency wave content (0-4kHz) and high amplitudes and far field ground motion contains relatively low frequency wave content. 65 5.2.3 Discussion Ground motion characteristics induced by an earthquake and by a blast are completely different. A blast induces high peak ground accelerations in the near field whereas an earthquake produces very low peak ground accelerations. During earthquakes, low frequency large wavelength waves are predominant in the far field. In blasting, even in the "far field", the ground motions are still higher frequency and lower amplitude. • To obtain representative ground motion would need a lot of charges. If were closer, too high frequency. To model the same characteristics of an earthquake ground motion is a difficult task. 66 Acceleration Time History- Imperial Valley Earthquake, (1940),EI-Centro Site (R=12.7 km) .3 J _ J 1 1 1 Time (sec) Figure 25: Acceleration Time History- El-Centro Sites 67 Acceleration Time History- Loma Prieta Earthquake, (1989), Treasure Island (R=99.17km) o 0) -2 u CU 1 m E 2 j> 0 CD u u < ill 0 2 5 3 0 3 5 4 0 4 I Time (sec) Figure 26: Acceleration Time History- Treasure Island Sites 68 a. Velocity Time History- Imperial Valley Earthquake, (1940), El-Centro Site (R=12.7 km) 0.3 1 1 1 1 1 1 1 1 0.2 TIME (sec) b. Displacement Time History - Imperial Valley Earthquake, (1940), El-Centro Site (R=12.7 km)) 0.05 T 1 1 1 1 r i 1 1 Figure 27: Velocity & Displacement Time Histories- El-Centro Sites 69 a. Velocity Time History- Loma Prieta Earthquake, (1989), Treasure Island (R=99.17 km) 0.3 f\ _« f\ N .A A rtA I A r t A A A r< 0 2 5 3 0 3 5 4 0.2 -o 0.1 -<u w E_ *>. 0 -"5 o <D > -0.1 --0.2 -0.3 -Time (sec) b. Displacement Time History - Loma Prieta Earthquake, (1989), Treasure Island (R=99.17 km) 0.05 0.025 c 0) i o o re n. W ° -0.025 -0.05 2P 25 30 36 4b Time (sec) Figure 28: Velocity & Displacement Time Histories- Treasure Island Sites 70 a. FOURIER AMPLITUDE SPECTRUM - Imperial Valley Earthquake, (1940), El-Centro Site (R=12.7 km) 250 9nn 0 5 10 15 20 25 Frequency (Hz) b. Smoothed FOURIER AMPLITUDE SPECTRUM - Imperial Valley Earthquake, (1940), El-Centro Site (R=12.7 km) I I n i b 0 5 10 15 Frequency (Hz) Figure 29: Fourier Amplitude Spectrum- El-Centro Site 71 a. Smoothed FOURIER AMPLITUDE SPECTRUM - Imperial Valley Earthquake, (1940), El-Centro Site (R=12.7 km) 15 12 3 0 i I 5 10 15 Frequency (Hz) b. FOURIER AMPLITUDE SPECTRUM -Loma Prieta Earthquake, (1989), Treasure Island (R=99.17 km) 250 , , 1 200 I 150 0) T J 3 5 10 15 Frequency (Hz) Figure 30: Fourier Amplitude Spectrum- El-Centro & Treasure Island Sites 72 5.3. Energy Approach -The Arias Intensity The Arias intensity is a quantitative measure of earthquake-shaking intensity. It has been suggested as a possible means of assessing liquefaction susceptibility (Kayen and Mitchell, 1997). The Arias intensity is the total energy per unit weight absorbed by an idealized set of oscillators during earthquake motion (Arias 1970). The Arias intensity measure (also termed accelerogram energy) is the sum of the energy absorbed by a family of simple oscillators evenly spaced in space. For a single component of motion in a given direction, Arias demonstrated that the cumulative energy-per-unit weight absorbed by a set of single degree of freedom oscillators at a site could be expressed as Where, I x x (v) =viscous damping-dependent intensity measured in x-direction, v = damping ratio of oscillator, g = acceleration due to gravity, to= duration of earthquake-shaking, a x (t)= ground acceleration. The damping factor, arccos (v)/g*-N/~(l-v2) is insensitive to variations in the structural-damping ratios of the oscillators. For the case where the damping ratio approaches zero, equation 8 reduces to Note that the Arias intensity technique was developed for analysis of earthquake excitations. This research investigates the possibility of use of the Arias intensity technique for ft, -o (8) (9) 73 assessment of the effects of explosive detonation. The main purpose for this analysis is to present a comparison of an absorbed energy at a particular site from an earthquake and that from a blast. Figures 31 and 32 present computed Arias intensity for excitations recorded at the Treasure Island and El-Centro sites respectively. Both Treasure Island and El-Centro are on a thick section of loose alluvial deposits. The computed Arias intensity at the end of earthquake shaking was Ia= 0.38 m/sec for Treasure Island earthquake and Ia= 1.8 m/sec for El-Centro earthquake. Note that the computed Arias intensity of these earthquakes suggests that the El-Centro site experienced 4-5 times shaking intensity of Treasure Island rather than doubling indicated by the ratio of accelerations. Figure 33a shows the computed Arias intensity response for a blast signal, which was recorded at 10m away from a 1.5kg single charge event. Figures 33b shows the Arias intensity response at 10.7m distance, from a 3kg charge event. However, the arias intensities computed at far field from a blast shows a similar response as those computed from an earthquake. Figure 34 shows such a computed Arias intensity for an earthquake and a blast. Figure 34a illustrates the computed Aria intensity from the El-Centro site. Figure 34b shows the Arias intensity, computed from the blast trace recorded at 47.7m away from detonation point. Comparison of Figures 33b and 34b clearly indicates that the Arias intensity attenuates rapidly with distance from the source of energy. In addition, note that these two figures (34a and 34b) show the comparable trends and the similar magnitudes of computed Arias intensities with longer time period for earthquake records and shorter time period for blast records. 74 Figure 35 shows a one of Arias intensity liquefaction assessment plots, proposed by Kayen and Mitchell. This figure illustrates the plot with fines content correction to equivalent "Clean Sand". This plot has been categorized into two zones as "liquefaction zone" which is above the clean sand boundary and "no liquefaction zone". Arias intensities computed through this analysis are shown in this figure. Note that the computed values of the Arias intensity for blast records are much higher than those for earthquakes. According to classification shown in the figure 35, the El-Centro site is liquefiable and the locations at blast # 2 and blast #4 are also liquefiable. But, the actual records indicate that these locations were not liquefied during a particular blast event. These results suggest that Arias intensity does not capture all aspect of earthquake shaking relevant to liquefaction susceptibility. Also these results indicate the approach of computing Arias intensity should be modified taking the shaking duration into consideration. 75 a. Acceleration Time History-Loma Prieta Earthquake, (1989), Treasure Island (R=99.17 km) Time (sec) 2.0 1.6 1.2 0.8 0.4 0.0 b. Arias Intensity - Loma Prieta Earthquake, (1989), Treasure Island (R=99.17 km) 10 Time (sec) 15 20 Figure 31: Arias Intensity of Treasure Island Site 76 a. Acceleration Time History-Imperial Valley Earthquake, (1940), El-Centro Site (R=12.7 km) 4 -r 3 -Time (sec) 2.0 1.6 -5" 1.2 o in E « 0.8 -0.4 -0.0 ( b. Arias Intensity - Imperial Valley Earthquake, (1940), El-Centro Site (R=12.7 km) ) ! 5 1 0 - 1 5 20 25 30 35 40 Time (sec) Figure 32: Arias Intensity of El-Centro Site 77 a. Arias Intensity - Blast 2 at P1 (1.5kg ,10m, Burial Depth 6m) 750 o o in 500 250 0 0.0015 0.0025 0.0035 0.0045 Time, sec 0.0055 b. Arias Intensity-Blast 4 at P1, (10.7m, 3kg, Burial Depth 3m) 750 500 o V It) 250 0 -I L 1 1 1 1 1 1 2.47 2.475 2.48 2.485 2.49 2.495 2.5 Time, sec Figure 33: Arias Intensity- Blast 2 at P1, Blast 4 at P1 78 2.0 -1.6 -~ 1-2 o a> 10 1 - " 0.8 -0.4 0.0 ( a. Arias Intensity - Imperial Valley Earthquake, (1940), El-Centro Site (R=12.7 km) ) i 5 1 0 1 5 2 Time 0 2 (sec) 5 3 0 3 5 40 b. Arias Intensity-Blast 4 at P2, (47.7m, 3kg, Burial Depth 3m) 2 , . 1.6 Time, sec Figure 34: Arias Intensity- El-Centro site and Blast # 4-P2 79 Arias Intensity for Liquefaction Assessment (Plot with Fines Content Correction to Equivalent "Clean Sand") 1000.00 100.00 10.00 1.00 biasyPi? at 1 um / \ Blast #4 at 1 0 . 7 Clean Sand Boundary LIQUEFACTION Blast #4 a t W . 7 m El-Centro Site _ Inl,1.111 o ; i i , Treasure Island bite L M f i i mi i E F A C T I O N i \ i 0.00 10.00 20.00 30.00 (N1,fc)60 F i g u r e 35: A r i a s Intensity for L i q u e f a c t i o n A s s e s s m e n t 80 5.4 Determination of Wave Propagation Velocity 5.4.1 P-Wave Velocity The frequency analysis of blast-induced excitations (see section 4.3.1) indicated that such ground motions contained high proportion of high frequency wave content and a small proportion of low frequency type wave content. It was shown that it was very difficult to distinguish shear waves in the blast records. As an attempt to identify shear waves the following propagation velocity analysis was carried out. Considering the arrival time of peak acceleration is a known parameter at both locations (PI and P2) and assuming that waves propagate only spherically at a constant velocity (Figure 36), it is possible to estimate the propagation velocity from each blast event. P-waves would be expected to arrive first followed by S-waves and surface waves. By comparing the calculated propagation velocities to the range of theoretical wave velocities for a particular soil type, it may be possible to identify the arrival of S-wave content. The P-waves are produced by a blast. Once the P wave velocity is established for a certain blast then an accurate time of detonation can be determined, as the distance from detonation point to recording location was a known parameter. If the detonation time is known, it is possible to identify the shear wave content of the records by assuming a shear wave velocity range for particular soil. The process of estimating these parameters is explained below in detail. Let us assume that the explosion is detonated at "O" location below the ground level (see Figure 36). The data recording instruments were located at PI and P2, 6m below the ground level. OPi and OP2 are the hypocentral distances (straight line between monitoring point and the blast). The propagation velocity is assumed to be constant- V p . Ti and T2 are the travel 81 times of the first wave arrival from blast location at P l and P2 monitoring points respectively. The time of TI and T2 are known parameters as the ground responses were recorded at P l and P2 on a common time base. Let the detonation occur at To sec, which is unknown. Blast Hole 9, 6, or 3m Figure 36 : Sketch for determination of propagation velocity Time for a wave to travel from O to Pi: - (Ti - T 0 ) sec Time for a wave to travel from O to P 2: - (T 2 - To) sec Here, OPi, OP 2 , Ti and T 2 are known parameters Compute the velocity- Vp O P , = V P * (Ti - To) O P 2 = V P * ( T 2 - T 0 ) Solving for Vp V P = (OP 2- OPi) / ( T 2 - Ti) m/sec. (10) (12) 82 where, OPi = J~(OA2 + AP, 2 ) and OP 2 = v r ( O A 2 + AP 2 2 ) Once the propagation velocity, V p is known, the time of detonation, To can be found from the equation #10. Propagation velocities were computed for each blast event using the equation # 3 and the resulting velocities are given in the Table 11. Note that the X, Y, Z are the orthogonal axes of the triaxial accelerometer package. Blast # 5 contained three explosive decks and pulse # 1, pulse #2, and pulse # 3 indicate the response from deck # 1, 2, and 3 respectively. Blast # 6 contained four blast holes and each hole contained three decks. The four holes were at J, I, H, G in order of detonation. The time delays used during this blast event were 250ms between decks and 500ms between blast holes and the detonation sequence was bottom up. Due to these reasons, a top deck of each hole was detonated together with the bottom deck of the adjacent hole. This table shows that the calculated propagation velocities are quite consistent with the theoretical P wave velocity in a loose saturated medium, which is around 1500 to 1700 m/sec (Dowding, 1985). Blast # 2 shows that the propagation velocity was around 1580 m/sec and blast # 4 indicates the propagation velocity was around 1606 m/sec. The blast # 5, which contained 3 charges, shows a bit different behaviour of velocity propagation pattern. These charges were detonated using a bottom up sequence. The first wave pulse from the first charge (deepest charge) propagates with a velocity of 1606 m/sec, the following pulse propagates with a velocity of 1569 m/sec and the third pulse that was close to the ground surface propagates with a velocity of 1498 m/sec. The results, computed from blast # 6 also show a similar wave propagation pattern as during blast # 5. Blast # 6 contained four blast holes. 83 Table 11: Computed Propagation Velocities X, Y - Horizontal; Z - Vertical Blast # Direction Propagation Velocity m/sec Blast #2 X - 1580 Y + 1548 Z - 1512 Blast #4 X - 1606 Y - 1567 Z + 1548 Blast #5 Pulse#l X + 1606 Y - 1580 Z + 1548 Pulse #2 X + 1569 Y - 1548 Z + 1512 Pulse #3 x + 1498 Y - 1488 Z + 1464 Blast #6 J-l Pulse #1 x + 1655 Y + 1645 Z + 1640 I-f x + 1579 > Simultaneous Y + 1579 Detonation Z + 1551 i-y X + 1529 Y + 1520 Z + 1502 84 Table 11 continued Blast # Direction Propagation Velocity m/sec 1-3 Pulse #3 X + 1394 Y + 1394 Z + 1362 J-2 Pulse # 4 X - 1558 Y + 1542 Z + 1524 J-3 Pulse # 5 X + 1367 Simultaneous Y + 1348 ^Detonation Z + 1332 H-l x + 1650 Y + 1638 Z + 1632 H-2 Pulse # 6 X - 1606 Y + 1595 Z + 1588 H-3 Pulse #7 X + 1568 ^Simultaneous Y + 1548 Detonation Z + 1548 G-l x + 1510 Y + 1492 (# 8) Z + 1488 G-2 X - 1480 Y + 1464 Z + 1464 85 Table 11 Continued Blast # Direction Propagation Velocity m/sec G-3 Pulse # 9 X + 1448 Y + 1432 Z + 1432 Figure 37 shows these computed propagation velocities versus charge numbers in a blast event of sequential detonation in a single borehole. There is a clear trend for propagation velocity to decrease with subsequent detonations. The above assumes that the velocity is uniform over the distance from Pl to P2. The reason for the decreasing velocities is not clear. In blast #6, the detonations in hole, G were the 10th, 11th, 12th detonation in the sequence. The trend towards reduced propagation velocity appears to be confined to sequential blasts in a single hole. The early detonations may cause reduced stiffness of the soil around the charges subsequently detonated in such a borehole. This could have resulted in slower transmission of disturbance through the ground. 86 Propagation Velocity Versus Charge Number in a Sequential Blast Event 1800 1600 o ai in E o o c o (0 o> RS Q . O 1400 1200 1000 1 1 > > L 1 • 2 3 Charge Number • Blast # 5 El Blast #6 -J A Blast # 6-1 X Blast # 6-H • Blast # 6-G Figure 37: Propagation Velocity Versus Charge Number in a Sequential Blast Event 87 5.4.2 S-Wave Content In an attempt to recognize the S-waves in the recorded ground motion, the acceleration data for blast #4 at PI and P2 recorded during the C A N L E X trial blast were examined. A range of shear wave velocity of 120-180m/sec was assumed. This was considered to be a broad range. The in-situ tests carried out in the test area revealed that the shear wave velocity, V s was about 130 m/sec. Figure 38 shows the schematic view of the blast event. The distance to the PI and P2 locations from the blast point was 10.7m and 47.7m respectively. The P wave velocity, V p for this blast was 1580 m/sec (Table 11) from which the detonation time, To can be determined. Instrument Location Vp Blast 1/ To Figure 38: Schematic View of a Blast Event From shear wave velocity (Vs) range, it is possible to estimate the likely arrival time of the shear waves at PI and P2. This estimation indicates that the shear waves would arrive in between 2.731 and 2.863 sec on the acceleration time histories. Figure 39a illustrates the recorded acceleration time histories of blast # 4 at P2 location. Figure 39b shows the zoomed 88 version of this acceleration time histories and an arrow shows the wave arrivals (Figure 39b) between time 2.731 and 2.863 sec. Figure 39c shows the wave arrivals between t=2.66 and 2.96sec. and it can be noted that a peak exists at 2.734sec. The velocity analysis of this peak shows that it had travelled with a velocity of, Vs = 47.7 / (2.734 - T 0 ) = 178 m /sec A wave with 178 m/sec of velocity is likely to be a shear wave, traveling in saturated soil (Dowding, 1985). This is a good indication that a detonation of an explosive can produce some shear waves, but they are small in magnitude relative to the P-waves. The peak acceleration during this phase of the ground motion is about 0.13g. In addition, a frequency analysis was carried out for this acceleration time histories recorded at P2 location and for the particular segment of acceleration time histories discussed above. Figure 39c illustrates the particular segment of acceleration time histories, where shear waves would have come from the detonation of blast # 4. Figure 40a illustrates the amplitude Fourier spectrum for response of blast # 4 recorded at P2 location. Figure 40b illustrates the amplitude spectrum for the particular segment discussed above (anticipated shear wave arrival). Note that this Figure 40b indicates that the dominant frequencies of waves arriving during this particular segment shows a dominant response at relatively low frequencies (0-400 Hz), but still very much higher than the earthquake. This is likely because of much shorter wavelength of the shear wave, i.e. V=f A,, compared to the earthquake case. These analyses suggest that the initial wave packet contains both P waves and S waves, but that the ground motion is dominated by the P-waves. The P-waves are of high frequency. The 89 S-waves are much shorter wavelength than would be produced by earthquake shaking at a large radius from the epicentre. The average P wave velocity is 1300-1600 m/sec and S wave velocity is 120-180 m/sec in saturated loose sands. The S-waves appear to be much higher frequency than S-waves produced by earthquakes. 90 a. Acceleration Time Histories-Blast #4 at P2 - X Direction (3kg, 47.7m) o "3 cn c o co i _ a CP o o < 1 16 m 56 2. 56 2. ^ 6 2. S6 2. 06 Time (sec) b. Zoomed Version of Acceleration Time Histories-Blast #4 at P2 in X Direction (3kg, 47.7m) ihear W a v e 76 2.86 2.06 Time (sec) c. Accelerations Recorded Between 2.725sec and 2.76sec-Blast #4 at P2 - X Direction (3kg, 47.7m) 15 2. 72 ^ 2 . 7 2 V V 73 \ /2.7 35 'A 2.7 45 2. ^ 5 2.7 55 2. o CD W O a in Time (sec) Figure 39: Determination of Shear waves 91 a. Fourier Spectrum Amplitude-Blast # 4 at P2-in X Direction (3kg, 47.7m) 250 , , , , 200 Frequency (Hz) b. Fourier Spectrum Amplitude for the Recorded Wave component between the time 2.725 and 2.76 sec from blast # 4 P2-X (3 kg, 47.7m) 100 n , , , , 75 0 500 1000 1500 2000 2500 Frequency (Hz) Figure 40: Fourier Spectrum Amplitude- Shear Wave Segment -Blast Signal # 4 92 5.6 Determination of the Peak Particle Velocities and Displacement of a Blast Pulse One of the main objectives of this thesis is to establish relationships between recorded ground motions and soil characteristics. Commonly used ground motion parameters are velocity and displacement of soil particles. However, only the acceleration response was recorded during this blast test. The velocities and displacements have to be derived from the measured acceleration data by integrating. In general, an integration of recorded strong motion acceleration data always shows some errors in velocities and displacements. These errors usually come from background noise, recording system noise and baseline shift of the original data set and these errors cause an offset in velocity and displacement (Hung-Chie Chiu, 1997). A "base line correction" is an essential first step in obtaining correct velocities and displacement from raw acceleration time histories. The explanation of the baseline correction employed in this study is included in Appendix-D (Gohl, 1999; Personal Communication). The velocities and displacements were calculated by integration accelerations measured in both X and Y directions. The peak particle velocity (PPV) refers to the square root sum of velocities in X and Y direction at a particular point. Figures of accelerations, velocities and displacements are shown in the thesis recorded in X-direction, although the same procedures which were used in analysis for data recorded in X direction applied to the data recorded in Y direction. Figure 41a shows a typical recorded acceleration time history and Figure 41b shows the velocities and displacement time histories calculated by integration of the acceleration records. The time histories shown belong to blast # 2 that was recorded using sampling rate of 31500PS. Figure 42a also shows a typical recorded acceleration time history 93 and Figure 42b shows the velocities and displacement time histories calculated by integration of the acceleration records. The time histories shown belong to blast # 4 that was recorded using sampling rate of 5000SPS. 5.7 Peak Shear Strain Explosive charges are typically approximated as spears or cylinders. If a perfectly spherical explosive charge were detonated, in a homogeneous, isotropic soil, the only displacement would be radial. This is the assumption of spherical cavity expansion theory from which it can be shown that the maximum shear strain can be computed by using the following equation (Hryciw, 1986). Y = 0.5 [PPV/V P + U r / r ] (13) where, PPV is the peak particle velocity, Vp is P-wave propagation velocity, U r is the radial particle displacement and r is the distance between the charge and measurement point. For short cylindrical charges, spherical cavity is a reasonable approximation for soil deformation in a zone more than a few diameters away from the charge. This implies the peak pore pressures occur at the peak shear strain. 5.8 Pore Pressure The dynamic pore water pressure is directly induced by shock waves generated by an explosion and it can be seen positive and / or negative spike (Figure 24). The magnitude of this pressure pulse is a function of charge weight, distance and soil properties such as relative density and degree of saturation. High peak pore pressures do not necessarily indicate liquefaction, as high total stress may also exist. Excess residual pore pressures are indicative 94 of a tendency towards contraction due to strain induced by blasting. It is also clear that high peak pore pressure also leads to high residual pore pressures. The residual pore pressure is the pore pressure after the shaking stops and applied stresses are removed. As the residual pore pressure increases, the effective stress decreases causing a drop in strength and stiffness. This brings the soil closer to failure. Whether liquefaction occurs depends on the current strength relative to the current applied stresses. Figure 41c illustrates a measured pore pressure trend during the blast #2. It is noted that the trend of the pore pressure response is sensitive to the velocity and not very much to the displacement. The peak velocity coincides with the peak pore pressure. Figure 42 illustrates acceleration, velocities and pore pressure response due to blast # 4 at the P1 location. Again, peak pore pressure coincides with peak velocity. The later part of the velocity curve is unreliable as the sampling rate was not fast enough. Figures 43 a and 43 b show the velocities and pore pressure response at Pl and P2 locations respectively. Note that the pore pressure trend recorded at P2 (Figure 43b) shows that some peaks being cut off during recording and however, general appearance of the trend is reasonable. The trend of these figures also indicates that the pore pressure trend is closely related to the velocity. It is also possible to use pore pressure pulse to assess the wave propagation velocity in the water. The pore pressure pulse propagates (Figure 44) with a velocity of around 1455 m/sec, which is slightly less than the propagation velocity of a P-wave in water. 5.8.1 Trends of Pore Pressure Data Observed During analysis The summary of measured pore pressure data during this blast event was tabulated in Tables #5 to #9. Figure 45 shows the induced peak pore water responses versus distance. This figure 95 shows the induced peak dynamic pore pressures by the first detonation at the each event (e.g. #2, #4, #5- first deck, #6- first deck at J location). The reason for such a selection was to be consistent with the conditions during blasts #2 and 4. Comparison of pore pressure details of blast # 2 with the pore pressure details of blast # 4 indicates that the induced dynamic pore water pressure at blast # 4 at PI was 66m whereas blast #2 induces 128m of dynamic pore water pressures at its PI. In both cases, distance to the instrument location from the blast point was almost similar. The blast # 2 was of 1.5kg explosive detonated 6m below the ground level and blast # 4 was of 3kg explosive detonated at 3m below the ground level. Also, this figure shows that the first deck of blast #5 produces the greatest pore pressure that contained 4.5kg of explosives at 9m depth. These observations likely indicate that the buried depth has an influence on the rise of pore water pressure. Figure 46 shows the peak pore pressures versus computed peak shear strain. At the PPV, the computed displacement is very small and so PPV dominates. Figure 47 shows PPV versus peak pore pressure. Laboratory testing of loose sands has shown that (Dobry, 1979) the residual pore pressure caused by cyclic loading varies with the magnitude of shear strain per cycle and the number of cycles. Figure 48 shows a typical relationship developed for samples of Ottawa sand in cyclic simple shear (Finn, 1975). For larger cycles of shear strains, fewer cycles are required to attain a given residual PPR. Figure 49 shows residual PPR plotted against PPV. There appears to be a trend towards increasing residual PPR with increasing PPV. If each detonation is considered to be one cycle of loading at the computed shear strain, the measured PPR can be computed to the values obtained for Ottawa sand. The values are shown on Figure 48. The relative density of 96 the sand at the test site was considered to be 43% (CANLEX papers, 1996). The figure suggests that blasting at J-Pit was considerably were effective at generating pore pressure than a single cycle of loading at peak shear strain. It is likely that there is additional contribution from the rest of the ground shaking caused by each detonation and each pulse is equivalent to several cycles at a lower equivalent strain magnitude. Conventional understanding of sand response to cyclic loading would indicate that a threshold strain exists below which no residual pore pressure would be generated (Dobry, 1979). 97 a. Acceleration Time Histories -Blast 2 at P1-in X Direction; CANLEX(10m. 1.5kg) 3000 _2000 1 1000 (A 1-1000 -2000 -3000 -1 0.015 0.02 b. Velocity & Displacement Time Histories-Blast 2 at P1-in X Direction; CANLEX(10m. 1.5kg) E 3 in in <i> i _ Q. s o a. c. Pore Pressure Time Histories - Blast 2 at P1-in X Direction; CANLEX(10m. 1.5kg) * 12 \ J V I O.C 05 0. )1 O.C 15 0. Figure 41: Integrated Velocity and Displacement Time Histories 98 a. A C C E L E R A T I O N TIME H ISTORY-BLAST # 4 at P1; C A N L E X (10.7m, 3kg) 3000 jjj 2000 o S> 1000 s -1000" § -2000 < -3000 18 2. 19 2 5 2. 51 2. 52 2. 53 2.5 Time, sec b. V E L O C I T Y & D I S P L A C E M E N T TIME HISTORIES- B L A S T 4 at P1; CANLEX(10.7m,3kg) 75 E 50 h 25 (A £ Q. (1) i _ O Q. 0 -252+7--50 -75 c. P O R E P R E S S U R E TIME HISTORIES- B L A S T 4 at P1; CANLEX(10.7m,3kg) 2.48 -2^4 Initial Pore Pressure -2.52- -2.53 TIME (sec) -2.54 Figure 42: Velocity Time Histories & Pore Pressure - Blast # 4 at P1 & P2 99 a. VELOCITY & PORE PRESSURE TIME HISTORIES-BLAST 4 at P1; (10.7m,3kg) E, o o > 150 100 TIME (sec) ai cn. C O t/5 LU St Q. LU K. O a. 100 150 b. VELOCITY & PORE PRESSURE TIME HISTORIES-BLAST 4 at P2; (47.7m,3kg) o 0 W I s o o > 0.06 0.04 0.02 0 2.JJ9 -0.02 -0.04 -0.06 V e l o c i t y TIME (sec) Figure 43: Velocity Time Histories & Pore Pressure - Blast # 4 at P1 & P2 100 a. Pore Pressure Time Histories-Blast # 2 at P1 (10m, 1.5kg) 150 100 E, a V— 3 (A 0) 50 a> i— o a. -50 tP1=0.002272 0.005 0.01 Initial Pore Pressure Level • ^ 0.Q15 TIME (sec) .02 b. Pore Pressure Time Histories-Blast # 2 at P2 (28.3m, 1.5kg) 15 3 10 V) 0) L_ CL 0> o CL 5 t P 2=0. D14848 Initi ilP-OjerPTeisure Level T V= 18.3/ (0.01 V= 1455 m/s< 4848-.002272) ;c 0.01 0.015 0.02 TIME (sec) 0.025 0.03 Figure 44: Determination of Pore Pressure Propagation Velocity 101 P E A K P O R E P R E S S U R E H E A D V s H Y P O C E N T R A L D I S T A N C E 250 200 E. 9> in « 150 a! £ o a. E « 100 >. Q a> a. 50 #5, 4.5/9m Depth #2, 1.5kg/6m )epth #4,3kg/3m D( #6, 4,5kg/9m De P , h © #2, 1.5kg/6m I • )epth © # 5, 4.5/9m Depth B • & — 10 20 30 Hypocentra l Distance, m 40 50 • #2, 1.5kg/6m Depth, R=10m A 0 #6, 4.5/9m Depth, R=22.4 • B #5, 4.5kg/9m Depth, R=42.3m © #4, 3kg/3m Depth, R=10.72 #2,1.5kg/6m Depth, R=28.4 #6, 4.5kg/9m Depth, R=36.2m #5,4.5kg/9m Depth, R=10.44 #4,3kg/3m Depth, R=47.8m Figure 45: Peak Dynamic Pore Pressure versus Hypocentral Distance 102 S H E A R STRAIN V E R S U S P E A K P O R E P R E S S U R E H E A D 250 200 E. | 150 CD o a. o E ra Q tu Q. 100 50 0 0 0.02 0.04 0.06 0.08 0.1 Shear Strain % • Single blast #2 0 Single blast #4 A Multiple blast #5 • Multiple blast #6 Figure 46: Shear strain versus peak pore pressure 103 P E A K PORE P R E S S U R E HEAD Vs P E A K PARTICLE VELOCITY 250 200 150 o Q. O E to c s» Q CD O a. 100 50 0 0.2 0.4 0.6 0.8 Peak Particle Velocity ( m /Sec) • Single blast #2 • Single blast#4 A Mutiple balst #5 • Mutiple blast#6 Figure 47: Peak particle velocity versus peak pore pressure 104 Figure 48: Pore water Pressure versus Strain Cycles (after Finn, 1975) 105 PORE PRESSURE RATIO (PPR- Residual) Vs PEAK VELOCITY 0.9 0.8 0.7 0.6 0£ D_ 0.5 0_ 0.4 0.3 0.2 0.1 0.2 0.4 0.6 Velocity ( m / sec) 0.8 'Mutiple blast #6 ® Single blast #2 A Single blast #4 Figure 49: Peak particle velocity versus pore pressure ratio 106 5.9 Relationships Between the Dynamic Responses of Blasting and the Empirical Scaling Laws. It is common practice in blast designs to relate PPV to scaled distance, R/W 0 ' 3 3 3 . The factor R/W 0 ' 3 3 3 is the inverse of Hopkinson's number. It is convenient to plot log PPV to log R/W 0 ' 3 3 3 . The ground response to blasting attenuates with distance, R, from the blast and increases with the size of the charge. This section examines empirical relationships between scaled distance and ground response in the C A N L E X blast trial. The use of the same-scaled distance factor in this analysis allows comparison with previous results, since many researchers have used the same-scaled distance to interpret their results. Figure 50 shows PPV versus Scaled Distance. The attenuation of PPV with increase of Scaled Distance can be viewed on this figure. Figure 51 also shows that the residual pore pressure decreases as the scaled distance increases. The best fit for this response is given by the line: PPR = - 0 . 0 3 8 5 ( R / W ° ' 3 3 3 ) + 1.26 (14) In general, the PPR equals to one, induces the liquefaction conditions. Applying this condition to the equation #14, the Scaled Distance that induces liquefaction condition could be determined. The equation #14 can be rewritten as, 1=-0.0385 ( R / W 0 3 3 3 ) +1.26 R / W 0 3 3 3 = 6.75 where R in metres and W in kg This relationship indicates that the liquefaction is induced in loose saturated tailing sands, if the scaled distance is less than 6.75. It is also possible to determine the liquefiable distance for a given charge weight or vice versa. 107 Figure 52 shows the maximum computed shear strain versus scaled distance, for the C A N L E X blast. This figure shows a trend of decrease of strain as a function of scaled distance. The best fit for this response is given by the line: (y%) = -0.11 log (R/W 0 3 3 3)+0.152 (15) This equation represents the shear strain criterion that causes liquefaction condition in particular soil. Considering the liquefaction condition from equation # 18, BJ W 0' 3 3 3 equals 6.75 in the equation #19, the required shear strain to cause liquefaction can be determined as, ( y %) = 0.06%. This shear strain criterion suggests that the induced shear strains exceeding 0.06% at a particular point by a blast is liquefiable. Discussion of the Results The relationships established above could be useful to develop the proposed in-situ testing methodology using explosives. Particularly, the relationship of strain versus scaled distance indicates that the peak strains 0.06% or above produces liquefaction of the tailings. This criterion is in good agreement with the results of experiments carried out by Youd (1972), Pyke (1973) and Dobry & Ladd (1980) as they have identified the existence of threshold shear strain of 0.01(%) percent. Below this threshold, no densification or no pore pressure build-up take place in loose sands subjected to cyclic loading. The relationship discussed above can specify standards in developing the proposed in-situ methodology. If the shear strain history is known at a particular site, the above relationship (shear strain vs scaled distance) can give an indication of the required Scaled Distance to produce such a shear 108 strain at the site. This required scaled distance ( R / W U 3 J J ) controls the in-situ test conditions at a site. In addition, these established relationships from C A N L E X blast data allow comparison with other relationships established by previous investigators. P P R = - 0 . 0 3 8 5 ( R / W ° ' 3 3 3 ) + 1.26 (14) ( y %) = -0.11 log ( R / W 0 3 3 3 ) + 0 . 1 5 2 (15) 1. Ivanov (1967) has documented that liquefaction ( P P R = 1 ) could be achieved for sands with relative density, D R = 30 to 40%, when the scaled distance ( R / W 0 3 3 3 ) is less than 8 to 6. The Eq. (14) indicates the liquefaction criterion can be established as 6.75 of scaled distance from C A N L E X test blast where the relative density of the sand was 43%. This relationship indicates that the liquefaction could occur if the scaled distance is equaled 6.75 or less. 2. In addition, the Hopkinson's number (W 0 3 3 3 / R ) is the inverse of the scaled distance(R/W ° 3 3 3 ). According to the liquefaction condition discussed above, the Hopkinson's number can be determined as 0.15 for the sand in J-pit at C A N L E X experiment. The range of Hopkinson's number also indicates a good agreement with the results of previous investigators. The comparison discussed above, indicates that the relationships established from C A N L E X blast data are in good agreement with the relationships presented by previous researchers. 109 S C A L E D DISTANCE Vs PEAK PARTICLE VELOCITY (PPV) 10 o CD (0 u o C D > o '€ co Q . JX: re (u Q . 0.1 10 100 Scaled Distance ( m / W A0.333) Figure 50: Scaled Distance versus peak particle velocity 110 PORE PRESSURE RATIO Vs S C A L E D DISTANCE-CANLEX Figure 51: Pore Pressure Ratio versus Scaled Distance 111 0.05 -I 0.045 -n HA PEAK SHEAR STRAIN Vs SCALED DISTANCE i > U.U* t n 4 • • train (°/< J C 4 CO k-n 01 § 0.02 -n 4 • • 0.01 -0.005 -0 -1 4 • 1 Scaled Distance 0 100 ( m / W A 0.333) Figure 51: Shear Strain versus Scaled Distance 112 6. CONCLUSIONS AND RECOMMENDATIONS This thesis discussed and analyzed the results of the C A N L E X test blast program. This included the evaluation of measuring instruments used in the test, sampling rate and evaluation of measured data. This evaluation helped to distinguish between reliable and unreliable data and identified practical issues in monitoring of ground motion and pore pressures due to blasting. In addition, this research investigated the induced wave patterns due to an explosion and the near field blast effects and the far field blast effects. Also, the research evaluated the possibilities of simulating earthquake like ground motions using explosives. The results of these studies lead to the following conclusions: • Blasting ground motions show much higher frequency than those in earthquake • motions in liquefiable sites. Analysis of the C A N L E X data shows that high frequency components attenuate rapidly with distance from the blast. • Peak accelerations are much higher in the near field for blast than those from earthquakes. In fact, the peak values could reach several hundred g's. However, peak accelerations from blast attenuate very fast with distance from the detonation point. • The wave propagation velocity between two points was observed to decrease when charges were detonated bottom up in sequence in a single borehole. These findings suggest that to simulate actual earthquake ground motions by blasting is not feasible, because frequency characteristics of earthquakes cannot be "created" by blasting with the type of explosives use for C A N L E X test. Even if created a very close match at one 113 point in the depth profile would not result in the soils at the site experiencing the same shear distortion as in an earthquake due to the difference in wavelength between earthquake induced and blast induced waves. However, blasting can be used effectively to create liquefaction effects such as sand boils. General conclusions are as follows: • Blasting can generate motions with significant shear wave content, but the shear waves are very small when compared with these of the P-waves. • Theory of Cavity expansion can be used to approximate and explain the effects of blasts in the field. • It is possible to obtain a relationship between peak particle velocity and peak dynamic pressures. The two coincides for a particular blast detonation. • During blasting soil experiences cyclic loading. It is therefore possible to estimate shear strains in terms of velocity and displacement at a point. Experimental data shows that maximum shear strain, y, is related to peak pore pressure. It is also shows that maximum shear strain, y, is related to residual pore pressure. This can be explained by relating shear strain at number of cycles, N, to pore pressure ratio. Multiple blasts create multiple cycles of shear strains and can produce significant pore water pressure than single blasts. Based on the above, it should be possible to use sequential detonation to investigate the potential for pore pressure generation during cyclic loading of soils and it should also be 114 possible to design in a controlled fashion a blasting sequence to create liquefaction to test the response of installations to the effects of liquefaction during an earthquake. Recommendations • Important guidelines in planning a blast test: (i) the selection of measurable bandwidth (Minimum and Maximum range) of an instrument is important. It is necessary to confirm whether the range of such instruments would be capable to capture the response due to a blast accurately. Also, it is very important to sample at a sufficiently fast rate. (ii) It is necessary to reduce impact of instrument housing or to determine the frequency characteristics of such a housing, performing an actual field test prior to the real blast event. As an alternative method, use an accelerometer attached to a code to avoid such a concern mentioned above. 115 BIBLIOGRAPHY Arias, A. (1970) "A measure of earthquake intensity" Seismic Design for Nuclear Power Plants, The M.I.T. Press, Cambridge, Massachusetts Berryman, J.G. (1980) "Confirmation of Biot's theory" Appl. Phys. 37, No. 4,382-384 Biot, M.A. (1956) "Theory of propagation of elastic waves in a fluid-saturated porous solid" J. Acoustical Society of America, v28, No.2, page 168-191 Biot, M.A. (1941) "General theory of three-dimensional consolidation" J. Appl. Phys. 12, 155-164 Bogdanoff, I. "Vibration Measurements in the damage zone in tunnel blasting" (1996), Rock Fragmentation by blasting, Mohanity (ed.) Balkenna, Rotterdam. Braulf J.W; White O.R; (1971) "The analysis and restoration of astronomical data via the Fast Fourier transform" Astronomy and Astrophysics 13 169-189. Charlie W.A; Jacobs P.J.; Doehring D.O. "Blast induced soil liquefaction of an alluvial sand deposit" (1992) Geotechnical Testing Journal A S T M 15(1) 14-23 CONETEC Investigations LTD., "Presentation of blast monitoring data" Syncrude, Fort Mc-Murray, Alberta 1997 Dembicki, E., Imioleck R. and Kisielow N. (1992). "Soil compaction with blasting method", Geomechanics and water engineering in environmental management. A . A Baklema, Rotterdam, Netherlands, 599-622 116 Dick, R.A., D, V. D' Andrea and L.R. Fletcher(1993) "The chemistry and physics of explosives" J. Explosive Engineering (10) No. 5 pp 31-41 Dobry, R.; Swiger, W.F., (1979) " Threshold strain and cyclic behaviour of cohesionless soils", Proceedings, Third A S C E / E M D E specialty conference, Austin, TX, September 17-19, pp. 521-525 Dobry, R.; Ladd, R.S.; Yokel, F.Y.; Chung, R.M.; Powell, D. (1982) "Prediction of pore water pressure buildup and liquefaction of sands during earthquakes by the cyclic strain methods" Building Science Series 138, National Bureau of Standards, US, Department of Commerce Dowding, C. H., (1985) "Blast vibration monitoring control" Prentice Hall Inc., NJ 297 pp Florin, V. A., and Ivanov, P. L., (1961) " Liquefaction of saturated sandy soils" Proc. 5 t h International Confi, Soil Mechanics and Foundation Engineering, Vol. 1, pp 107 Dupont (1977) "Blaster's hand book" 15th edition, Sales Development Section, Explosive Department, E.I. Dupont de Nemours and co., Inc., Gohl W.B.; Howie J.A; and Everard J. (1996) "Use of explosive compaction for dam foundation preparation" 49 t h Canadian Geotechnical Conference, St John's Nfld., September. Gohl W.B., Howie J.A., Hawson H.H., and Diggle D. (1994) "Field experience with blast densification in an urban setting" 5 t h US Earthquake Conference, Chicago Illinois, July pp 221-230 117 Hachey, C.J., R. Plum, J. Byrne (1993) "Blast densification of a thick loose debris flow at Mt. St. Helen's, Washington, Proc. Settlement 94, Conf, Austin, T X Hemphill, G.B. (1981) "Blasting operations" McGraw-Hill Inc. Hentych J. "The dynamics of explosion and its use" 1979 New York Higgins C.J. (1992) "Explosive simulation of earthquake-like ground motion", NSF Workshop on Experimental Needs for Geotechnical Earthquake Engineering", June. Higgins, C.J., Simmons, K.B. and Pickett S.F. (1978), "A small explosive simulation of earthquake-like ground motions" Proc. ASCE Specialty Conference on Earthquake Engineering and Soil Dynamics, Vol. 1 pp 512-529 June 19-21, 1978 Pasadena Calif. Higgins C.J.; Johnson R.L; Triandafilidis (1978), " Simulation of earthquake-like ground motions with high explosives" Final Report, University of New Mexico, Dept. of Civil Engineering, Albuquerque N M Hill, R. (1950) "The mathematical theory of plasticity" Oxford University Press Hryciw R.D., (1986) " A study of the physical and chemical aspects of blast densification of sand" Dissertation in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy, Northwestern University. Hung-Chie Chiu (1997) "Stable baseline correction of digital strong-motion data" Bulletin of the Seismological Society of America, Vol. 87, No. 4 pp 932-944 August 1997 118 Ishihara K. (1967) "Propagation of compression waves in a saturated soil" Proc. of International Symposium on Wave Propagation and Dynamic Properties of Earth Materials, Albuquerque New Mexico, U S pages 451-467. Ivanov P.L.; 1967 "Compaction of noncohesive soils by explosions" National Technical Information Service Report No TT 70-57221, U.S. Department of commerce. Imiolek R.E. (1992) "Compaction of water-saturated soils by blasts from elongated charges" Osnovaniya Fundamenty I Mekhanika Gruntorv (29) 4, 24-26 Kedrinskii V. K. " Hydrodynamics of explosives" Zhurnal Prikladnoi; Mekaniki I Tekhnicheskoi Fiziki, No- 4, pp 23-48 July - August 1987 Kok, L . (1981) "Settlement due to contained explosives in water-saturated sands" Proc. 7 th Int. Symp. Military Application of Blasting Simulations, Medicine Hat, Alberta, Canada. Konya, C.J., and E.J. Walter (1990) "Surface blast designs" Prentice Hall Inc. Englewood Cliffs, NJ, 300pp. Martin R., Finn L . A (1975) "Fundamentals of liquefaction under cyclic loading" ASCE, Vol. 101, 1975 Pacific Geodynamics Inc. (1997), "Summary report to Sato Kogyo Co. on trial blast densification, Ichikawa city test site", June 1997 Pacific Geodynamics Inc. (1998) "Molikpaq blast densification project - Annacis Island test blast report", Prepared for Sakhalin Energy Investment Co Ltd., 119 Seed, H.B. and I.M. Idriss (1971) "Simplified procedure for evaluation soil liquefaction potential" J. of Soil Mechanics and Foundation Division, ASCE (97) pp 1249-1273 Seed, H.B. and Martin P.P. (1975) "The generation and dissipation of pore water pressures during soil liquefaction" Report No. EERC 75-26, Earthquake Engineering Research Centre, University of California, 1975 Seed, H.B. and I.M. Idriss (1981) "Evaluation of liquefaction potential using field performance data" ASCE, Vol. 109, 1981 Studer, J. and Kok L. (1980) " Blast-induced excess pore water pressures and liquefaction, experience and application" Proceedings, International Symposium on Soils under Cyclic and Transient Loading. Swansea, United Kingdom, 7-11 January, pp 581-593 Sharpe, J.A. (1942) "The production of elastic waves by explosion pressures" I and II Geophysics, 7, 144-154, 311-321 Van Impe, W.F. (1989) "Soil improvement techniques and their evolution" A.A. Balkema, Rotterdam, Netherlands 125pp Ventura, C (1999) "Personal communication" Vesic, A.S. (1972) "Expansion of cavities in infinite soil mass" J. Soil Mechanics Foundation Division, A S C E 98 Wade A. Narin Van Court; James K. Mitchell (1998) "Investigation of predictive methodologies for explosive compaction" Geotechnical Earthquake Engineering and Soil Dynamics III 120 Wade A. Narin Van Court; James K. Mitchell (1994) "Densification of loose saturated cohesionless soils by blasting" Geotechnical Engineering Report No. UCB/GT/94-03 Youd, T.L. (1972) "Compaction of sands by repeated shear straining" J. of the Soil Mechanics and Foundation Divisions, ASCE, July, SM 7 pp 709-725 Yu, H.S. and G.T. Houlsby (1992) "Finite cavity expansion in dilatants soils: loading analysis" J. Geotechnique 42, No 4 Wu, G. (1995) "Dynamic response analysis of saturated granular soils to blast loads using a single phase model" A research report submitted to Natural Sciences and Engineering research Council of Canada. 121 APPENDIX- A PHOTOGRAPHS This appendix includes a collection of photos taken during blast # 5 at the C A N L E X blast experiment. Note that, this collection includes the preparation stage of the blast experiment, which shows the geometry of explosive, and stemming the blast hole with pea gravel. In addition, this collection shows the field condition right after the detonation of explosives. Note that the expulsion of water and gas from blast holes and several sand boils had been appeared on the ground surface due to the blast. 122 123 Photo -3: Preparation of the Blast Hole (stemming with pea gravel) 124 Photo-5: Just After the Detonation (water is shooting up from a blast hole) 125 126 8.Photo-8: Several Sand Boils that Appeared und Surface Photo-9: View of Sand Boils 127 APPENDIX - B ADDITIONAL FIGURES This appendix includes figures of acceleration time histories recorded at PI and P2 locations during blast # 5 and blast #6. Note that, this collection includes good quality data and bad quality data as well. The explanation for recording good quality and bad quality data is discussed in the chapter 3 of the thesis. 128 Acceleration Time Histories in X-Direction -Blast 5 at P1,10.12m Away-Deck 1 C A N L E X (10.12m,4.5kg) 3000 •5 2000 o V) s 1000 tn £ c 0 o •S 1 S-1000 u < -2000 -3000 34 8425 1.845 85 1.655 1.86 1.865 1.87 1.875 Time, sec Acceleration Time Histories in X-Direction-Blast 5 at P1,10m Away- Deck 2 C A N L E X (10m, 3.0kg) 3000 ^ 2000 o in 8 1000 (f) 1 ^ 0 o 1 -100$' o u < -2000 -3000 2.20i 205 2.21 2.215 2.22 2. 225 2.23 2.235 2.24 Time, sec 3000 •5-2000 s 1000 in E •c 0 ,o f -1000 2 © 1-2000 -3000 Acceleration Time Histories in X-Direction-Blast 5 at P1,10.12m Away-Deck 3 C A N L E X (10.12m,1.5kg) * \ / \ll Z""-\ H_«_nYS 46 2.4 65 2. Y7 2/ 75 2. 18 2.4 85 2. 19 2/ Time, sec Figure B1: Acceleration Time Histories # 5 at P1 129 Acceleration Time Histories in X-Direction-Blast 5 at P2, 36.4m Away- Deck 1 C A N L E X (4.5kg) s> -ioo Time, sec Acceleration Time Histories in X-Direction-Blast 5 at P2, 36.3m Away-Deck 2 C A N L E X (3.0kg) 200 0 -2.; -100 ; I 2.23' -300 Time, sec Acceleration Time Histories in X-Direction -Blast 5 at P2, 36.4m Away-300 200 100 0 2. -100 < -200 -300 '85 2. (9 2.4 95 2 5 2.E 05 2. >1 2.1 15 2. 52 2.1 25 2. Time, sec Figure B2: Acceleration Time Histories # 5 at P2 130 Acceleration Time Histories- Blast # 6 at P1-X Direction (10m away from blast array) 3000 2000 1000 u o < -1000 -2000 -3000 H Time (sec) Figure B3: Acceleration Time Histories- Blast # 6 P1 131 3000 _2000 f iooo I 0 I-1008' *-2000 81000 CA 1 I ° llOOO' o u < -2000 -3000 505 Acceleration Time Histories-Blast #6 at P1 From J (4.5kg, Pulse 1,22.12m) (45 u 3. f-v^~ ~ • )5 3.C 55 3. 36 3.C 65 3. )7 3.C 75 3. TIME {sec| 3.52 3.525 3.53 3.535 Acceleration Time Histories-Blast #6 at P1 From \^+\2 (7.5kg, 14.6m) 3000 2000 3.54 TIME (sec) - BLAST I 6- 1 AX 3000 2000 Si 000 f 0 000s--2000 -3000 655 Acceleration Time Histories-Blast #6 at P1 From l 3 (1.5kg, 14.72m) 3.66 3.665 3.67 3.675 3.68 TIME (SEC) 3.685 3.69 Figure B 4 : Acceleration Time Histories-Blast # 6 at P 1 132 Acceleration Time Histories-Blast #6 at P1 From J 2 (3kg, 22m) 3000 _ 2000 u 1 1000 2 3 ! % -1000' -2000 -3000 *\ A- , £05 3. v — 51 3.1 15 3. 32 3.! 25 3. 33 3.c 35 3. TIME (sec) c 0 o 2 4. i -1000 u o < -2000 -3000 TIME (sec) Acceleration Time Histories-Blast #6 at P1 From J3+H., (6kg, 10m) 3000 2000 1000 A 75 4. C65 4. y 4.( — U v V ^ v y ^ )8 4.C ^-V — — 85 4. )9 4.( 95 4 3000 2000 1 1000 (A 1 0 2 4 | -1000' u -2000 -3000 Acceleration Time Histories-Blast #6 at P1 From H 2 (3kg, 10m) -\.--95 4 2 4.2 05 4. >1 4.2 15 4. >2 4.2 25 4. >3 4.2 TIME (SEC) - BI6I2-P1 AX Figure B5: Acceleration Time Histories-Blast # 6 at P1 133 3000 - 2000 1 1000 0 |-1000 u •*-2000 -3000 ,4.36 Acceleration Time Histories-Blast #6 at P1 From H3+G., (1.5kg, 10.12m) 4V365 4.37 4.375 4.38 4.385 4.39 4.395 3000 2000 1000 0 4. -1000 -2000 -3000 55 Acceleration Time Histories-Blast #6 at P1 From G 2 (7.5kg, 14.6m) 4.555 4.56 4.565 4.57 TIME (sec) 4.575 4.58 4.585 3000 _ 2000 U a 1 1000 CA t ° | -100f5" o * -2000 -3000 Acceleration Time Histories-Blast #6 at P1 From G 3 (1.5kg, 14.72m) A \l £05 4. 31 4.£ 15 4. 32 4.c 25 4. 33 4.£ 35 4. TIME (sec) Figure B6: Acceleration Time Histories-Blast # 6 at P1 134 300 ^200 f lOO 0 OO3' S-1 1-200 -300 Acceleration Time Histories-Blast #6 at P2 From J (4.5kg, Pulse 1,36.72m) 75 3. )5 3.C )6 3.C 65 3. )7 3.C 58 3.C TIME (sec) 300 200 1 100 in ¥ 0 4 E 3. 1-100 -200 -300 515 Acceleration Time Histories-Blast #6 at P2 From \^+\2 (7.5kg, 31.5m) 3.52 3.535 3.54 3.545 3.55 TIME(sec) -BI6I1-P2-AX 300 200 "0" •S 100 1 I 0 I 3. 1-100 u Z -200 -300 Acceleration Time Histories-Blast #6 at P2 From l 3 (1.5kg, 31.88m) \ rs. r-—v /*v /\ 57 3.e 75 3. 58 3.6 85 3. 39 3.e 95 3 TIME (SEC) Figure B7: Acceleration Time Histories - Blast # 6 at P2 135 300 _200 u CD 1100 CO 1 ¥ 0 Acceleration Time Histories-Blast #6 at P2 From H 3 (1.5kg, 30.12m) s-100 o ^oo -300 4.37 4.375 4.395 4.405 $ 100 O I 1 ° 2 4. I -100 555 -2  -300 tl 100 (A 1 1 ° 2 4. 1-100 815 -200 -300 Acceleration Time Histories-Blast #6 at P2 From G 2 (7.5kg, 36.6m) 300 200 4.56 4.575 4.585 4.59 TIME (sec) Acceleration Time Histories-Blast #6 at P2 From G3 (1.5kg, 36.72m) 300 200 4.82 4.835 4.845 4.85 TIME (sec) -BI6H3-P2-AX Figure B8: Acceleration Time Histories - Blast # 6 at P2 136 Acceleration Time Histories (Unfiltered and Filtered)- Blast # 6 at P1(J.,)- C A N L E X Band Pass Filter 0.1-1500 Hz 3000 2000 1000 u o o CO c o to k_ o a o o < -1000 -2000 -3000 i Time, sec Figure B9: Comparison of Filtered and Unfiltered Accelerations 137 Acceleration Time Histories (Unfiltered and Filtered)- Blast # 6 at P1 (l.,+l2)-CANL.EX Band Pass Filter 0.1-1500 Hz 3000 2 0 0 0 1000 o m u> o a in c o 0) Cl) o o < -1000 -2000 -3000 • Unfiltered •Filtered Time, sec Figure B10 : Comparison of Filtered and Unfiltered Accelerations 138 Acceleration Time Histories (Unfiltered and Filtered)- Blast # 6 at P1(l3)- CANLEX Band Pass Filter 0.1-1500 Hz 3000 2000 1000 o o -2 o o in c o o o o < -1000 -2000 -3000 • Unfiltered "Filtered 3.65 3.655 565 Time, sec Figure B11 : Comparison of Filtered and Unfiltered Accelerations 139 ACCELERATION TIME HISTORIES- BLAST 2 - P1-X Direction; CANLEX (Band Pass Filter 0.0001-1.5 kHz) 300,000 200,000 u 8> 1 0 0 , 0 0 0 o E o TO i-OJ O - 1 0 0 , 0 0 0 o < -200,000 -300,000 0 0 . 0 D 0 1 0.0201 TIME (sec) Unfiltered A C C 2 P 1 - A X Filtered Acc 2P1AX Figure B12: Unfiltered and Filtered Acceleration Time Histories-Blast 2-P1 140 ACCELERATION TIME HISTORIES- BLAST 2 - P1-Y Direction; CANLEX (Band Pass Filter 0.0001-1.5 kHz) 300,000 200,000 100,000 0 o <u -in "o a; in E c o i_ a> o -100,000 -200,000 -300,000 0.0001 0.0201 TIME (sec) • A C C - 2 P 1 _ A Y Filtered A c c e - A Y _ 2 p1 Figure B13: Unfiltered and Filtered Acceleration Time Histories-Blast 2-P1 141 APPENDIX-C Determination of Orientation of Accelerometers In general, the orientation of accelerometers used at a test is to be documented during the test itself. Unfortunately, the orientation of accelerometers used in the C A N L E X blast test was not documented. This appendix discusses the determination of orientation of accelerometers used at the C A N L E X blast test. 142 Determination of Orientation of Accelerometers As mentioned in section 3.5.6 the orientation of accelerometers used in the C A N L E X test blast was not documented. In contrast other parameters, such as locations of the measuring instruments and the coordinates of the origin of blasts, were well documented. As this research required the orientation of accelerometers for data analyses, the following method was used to determine accelerometer's orientation. In order to determine the orientation of an accelerometer two conditions must be met. One is that the recorded time histories in two orthogonal directions should be available. The second one is that the excitation should come from a known direction. The source code to determine sensor's orientation was written by Dr. Ventura and it is included in this appendix-C. This is a M A T H C A D worksheet, and it is implemented as follows. The recorded acceleration time histories in two orthogonal directions X and Y direction should be saved as ASCII ".dat" extension files in the same folder where the source code has been saved. The recorded number of points of acceleration time histories and the time interval should be provided as input parameters ("N=" points, time interval "A"). As these values entered, the program then will run automatically. Plots shown at the bottom of the source code, will appear as F A m a x Vs fj (frequency) and aj (angle) vs fj (frequency). The idea behind this approach is to find the angle, which gives the maximum Fourier Amplitude Spectrum of a signal. If the above angle is zero, it means that the one axis of the measuring instrument was on the same path that the signal is coming from. 1 4 3 Furthermore, if the above angle is non-zero, it means that the axis of the instrument is diverted with that angle from the path of the signal. The code was written such a way that the X-axis is determined first, once the direction this axis is determined, then Y axis is always perpendicular to it. The procedures mentioned above were used to determine the orientation of accelerometers used in blast #6- CANLEX. Initially, it was assumed that the X direction of the instrument was on the same approaching path of the excitation from location J, that faces the north and the Y direction is perpendicular to X axis, that faces the West (see the Figure CI). The initial analysis carried out for accelerations recorded from the blast at J location. The Figure C2-a shows the maximum Fourier amplitude versus frequency. This figure shows that the max Fourier amplitude occurs at a frequency of 1000Hz. Knowing this frequency, the angle could be found from Figure C2-b. In this case this angle is 18° 32' and the X-axis of the instrument should be with an angle 18° 32' from the direct signal path, (see Figure C3). As this blast event has 4 signal origin points (J, J, H, G) one can accurately determine if the above assumption of X direction is correct. Figure C3 shows the orientation of accelerometer at both locations PI and P2 determined by using this procedure. 144 Determination of Orientation of Acce le rometers U s e d @ P 1 and P 2 L o c a t i o n s - C A N L E X Trial Blast I I I I I I I I I C : > v x : * A : \ , \ i i i 30.00 -20.00 -10.00 0.00 (m) F i g u r e d : Plan View of Blast #6 145 Maximum & Minmum Amplitude Fourier Spectra of Strong Motion Data Type Of Data : Acceleration Units : m / sec A 2 Data Set: Canlex # 6-J Data Files: N:=400 points i : = 0 . . ( N - l ) T ime interval : A := 0.0002 Xj := READ("b6xplj.dat")(AcceinX) Y i : = READ( "b6yplj.dat") Nyquist Frequency: NF : 1 2-A Four ier Transform: Fx := cfft(X) Fy := cfft(Y) Power Spectrum: E X X J := Fxj-Fxj j :=0 . . -N -Frequency Interval: NF = 2.5 x 10 Af := C r o s s Spec t rum: Exyj :=Fxj-Fyj A ( N - 1) Af = 12.531 f j : = j . A f E y y j := F y j - F y j Maximum Fourier Amplitude Spectrum (MaxFAS): Number of A v e r a g e s : na := Fmax; Exxj + Eyyj ' E X X J - Eyyj v 2 ^2 1 2 Re(Exyj)2 FAmax := movavg(Re(Fmax),na Min imum Four ier Ampl i tude Spect rum (M inFAS) : Fmin ,• E X X J + Eyyj ( EXXJ - Eyyj V l 2 Re(Exyj)2 FAmin := movavg(Re(Fmin),na) Ang le of Rotat ion (degrees): aj := Re| 8 0 - ^ Direction of M F A S : Re(Exyj) := —-atan J ' 2 [_ (Exx j -Eyy j ) a := movavg(a,na) 0 1000 2000 3000 Source:Dr.Ventura ; Folder:mathcad/Canlex Test Blast/Orien_6JP1 1000 2000 3000 146 a. Fourier Amplitude Spectrum (Maximum) b. Angle versus Frequency 50 n . 1 , , . 5 0 J 1 1 Frequency, Hz Figure C2: Determination of the angle of rotation 147 Determination of Orientation of Acce le rometers U s e d @ P1 and P2 L o c a t i o n s - C A N L E X Trial Blast APPENDIX- D Baseline Correction of Recorded Acceleration Time Histories This appendix discusses the significance of adopting the base line correction to an integration process of acceleration time histories. Also, this appendix discusses the procedures on how to employ it, in order to obtain accurate velocities and displacement from raw acceleration time histories. The code used for this analysis was written on a FOTRAN worksheet by Dr. Blair Gohl (Pacific Geodynamics Inc.). 149 Baseline Correction of Recorded Acceleration Time Histories Although, the most baseline errors are not quite noticeable in acceleration waveforms, but they become larger and larger in velocity and displacement waveforms. A displacement waveform integrated directly from the raw accelerograms without applying any preprocessing may deviate greatly from a zero baseline. Actually, most of these baseline errors are much larger than the ground displacement itself and, therefore, baseline correction is an essential step in obtaining displacement from raw acceleration time histories. In general, baseline errors in recorded strong motion data come from background noise, recording system noise, and baseline drift etc (Trifiinac, 1971; Trifunac and Lee, 1974; Hudson, 1979). Recent development in digital strong-motion accelerogram has made great improvement in data quality. However, baseline errors still exist in digital data despite the improved data quality. One can note that, a baseline correction is still an important issue in the processing digital data as both low-and high frequency errors are present in digital strong-motion data. Instrument Errors Five typical sources of errors come from the instrument, imperfect instrument response, insufficient resolution, insufficient sampling rate, electronic noise and baseline drift. The deviation of the instrument response from the flat amplitude response and the linear phase are small in the low frequency but large in the high-frequency signal. Therefore, the imperfect response of a sensor has a small influence on the baseline correction of a low-frequency signal. But, the instrument correction is important for high-frequency signals. 150 Background Noise A random waveform and wide distribution in the frequency content are also features of background noise. The frequency content and the characteristics of background noise are strongly site dependent. The acceleration of background noise is much smaller than that of a seismic signal, but the effect of the noise on the displacement cannot be ignored. Background noise contributes to the nonzero initial values that cause accumulated baseline errors. As results, background noise also has a large effect on the baseline errors. Initial Values In order to record pure seismic waves, it is necessary to have zero acceleration, velocity and displacement on the instrument before the arrival of these waves. In reality, these initial values are not zero because of the presence of electronic noise and the background micro-tremor. These initial values may cause a large final offset in the displacement waveforms, as a small initial acceleration might be magnified in the integrated displacement. Data Manipulation Errors Data processing removes some type of errors that come from the recording, but at the same time, it introduces new errors. Filtering, and windowing are typical forms of data manipulation that may introduce side effects to modify the processed waveform as well as the baseline. A small change in the wave from does not significantly modify the integrated displacement, but a small baseline drift causes a large final offset in displacement. In summary, most of the low-frequency errors in digital data cause significant baseline errors, despite the fact the quality of the digital data has been greatly improved. These 151 base line errors may be grouped into two categories. The first type errors arise from the baseline drift and the initial values in both velocity and acceleration. These errors relate to the long period noise and can be accumulated when the acceleration is integrated to velocity or displacement. The second type errors are random errors. These errors are not accumulated in the integrated velocity or displacement waveforms. Procedure Used for Baseline Correction In CANLEX Trial Blast Data The source code used to integrate acceleration time histories was attached in Appendix-D. The procedure was selected based on the characteristics of the baseline errors described above. Firstly, the input file should be prepared with an initial velocity and displacement as zero (sample of input file is attached). When the code is run, it reads the acceleration time histories file, which to be defined as "accO (i)= "ax (i)" "(ax (i)= acceleration in X direction) and computes the mean value of the acceleration record and the mean value will be subtracted from each point of the record. Accordingly, it creates a new acceleration time histories and that will be integrated to obtain velocities. In general, the computed velocities show a small positive or negative shift at the end of the velocity time history. If the velocity was computed accurately, it should be zero at the end. In order to make the velocity to be zero the code should be run again with non-zero velocity. The non-zero velocity is the velocity was found at the end of first integration. Then changing the sign of that non-zero velocity, should be replaced it with the zero velocity in the input file. Also, it is required to run the code again with new input file. The velocities obtained by this procedure shows reasonable accurate pattern, which has no baseline shift. 152 C O B T A I N D I S P L A C E M E N T t i m e h i s t o r y f r o m a c c e l e r a t i o n t i m e h i s t o r y C D I M E N S I O N A C C 0 ( 6 0 0 0 ) D I M E N S I O N a x ( 6 0 0 0 ) C H A R A C T E R * 8 0 C H I , E Q T I T L C O P E N ( 5 , F I L E = 1 D i s p 6 j P I . I N ' ) O P E N ( 6 , F I L E = ' D I S P . O U T ' ) C R E A D ( 5 , 1 0 1 ) E Q T I T L R E A D ( 5 , * ) N P O I N T , D T , F A C T , V E L O 0 , D I S 0 R E A D ( 5 , 1 0 1 ) C H I C R E A D ( 5 , C H I ) ( A C C 0 ( J ) , J = l , N P O I N T ) d o i = l , n p o i n t R E A D ( 5 , * ) a x ( i ) e n d d o w r i t e ( 6 , * ) 1 E Q T I T L ' w r i t e ( 6 , * ) E Q T I T L w r i t e ( 6 , * ) ' N P O I N T , D T , F A C T , V E L O 0 , D I S 0 ' w r i t e ( 6 , * ) N P O I N T , D T , F A C T , V E L O 0 , D I S 0 w r i t e ( 6 , * ) ' a x ' d o i = l , n p o i n t w r i t e ( 6 , * ) a x ( i ) e n d d o d o i = l , n p o i n t a c c O ( i ) = a x ( i ) e n d d o 1 0 1 F O R M A T ( A 8 0 ) C A M E A N = 0 . 0 D O 2 I = 1 , N P O I N T A C C O ( I ) = A C C O ( I ) * F A C T C * * * C O M P U T E M E A N A C C E L E R A T I O N 2 A M E A N = A M E A N + A C C 0 ( I ) A M E A N = A M E A N / N P O I N T C * * * S U B T R A C T O F F M E A N A C C E L E R A T I O N F R O M A C C E L E R A T I O N T I M E H I S T O R Y D O 3 1 = 1 , N P O I N T 3 A C C O ( I ) = A C C 0 ( I ) - A M E A N C D O 1 0 I = 1 , N P O I N T T I M E = D T * I A A 1 = A C C O ( I ) A A 2 = A C C O ( 1 + 1 ) C V E L O = ( A A 1 + A A 2 ) / 2 . 0 * D T + V E L O 0 D I S = ( V E L O + V E L O 0 ) / 2 . 0 * D T + D I S 0 W R I T E ( 6 , 1 0 2 ) d i s V E L O 0 = V E L O D I S 0 = D I S 1 0 C O N T I N U E C 1 0 2 F O R M A T ( F 1 4 . 3 , 7 F 2 0 . 5 ) S T O P E N D 153 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0063997/manifest

Comment

Related Items