oEPA
EPA/600/R-12/547 | June 2012 www.epa.gov/research
United States
Environmental Protection
Agency
Evaluation of the Seismic
Characterization of Select
Engineered Nanoparticles
in Saturated Glass Beads
RESEARCH AND DEVELOPMENT
-------
Evaluation of the Seismic
Characterization of Select
Engineered Nanoparticles
in Saturated Glass Beads
Mohamed Nihad Rajabdeen1
Barbara Luke1
D. Dale Werkema Jr.2
Department of Civil and Environmental Engineering
University of Nevada - Las Vegas
Las Vegas, NV 89119
2U.S. Environmental Protection Agency
Office of Research and Development
National Exposure Research Laboratory
Environmental Sciences Division
Las Vegas, NV 89119
Although this work was reviewed by EPA and approved for publication, it may not necessarily reflect official
Agency policy. Mention of trade names and commercial products does not constitute endorsement or
recommendation for use.
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
EVALUATION OF THE SEISMIC CHARACTERIZATION OF SELECT
ENGINEERED NANOPARTICLES IN SATURATED GLASS BEADS
Mohamed Nihad Rajabdeen1, Barbara Luke1, D. Dale Werkema Jr.2
1. Department of Civil and Environmental Engineering, University of Nevada - Las
Vegas.
2. U.S. EPA, Office of Research and Development, National Exposure Research
Laboratory, Environmental Sciences Division, Las Vegas, NV.
11
-------
EXECUTIVE SUMMARY
A laboratory testing apparatus was developed for the study of seismic body wave
propagation through nanoparticles dispersed in pore fluid that is essentially saturating
glass beads. First, the responses of water-saturated glass bead specimens were studied to
establish baseline signatures. Then the seismic responses in the presence of engineered
nanoparticles of various concentrations dispersed in the pore fluid of the specimen
chamber were studied to observe variances from baseline.
The testing apparatus incorporates piezoceramic bender elements to actuate and
receive seismic body waves through a cylindrical column filled with glass beads and
back-saturated at ambient pressure with liquid. The system was calibrated in air, water,
and water-saturated glass beads. System repeatability was checked after the system was
saturated and flushed once to soak and seat the beads. The water-saturated glass bead
specimens were tested for compression, shear, and spectral response, from which baseline
signatures were established. Criteria were proposed to evaluate the detectability of
nanoparticle dispersions.
Nanoparticle dispersions of zinc oxide (nZnO), titanium dioxide (nTiOz), and silver
(nAg) were tested. The testing system showed itself to be capable of registering subtle
changes in the response caused by varying consolidation states of the glass beads and
pore fluid content. The presence of nZnO was detectable at 0.03%, 0.3%, and 2.7%
concentrations for all the test methods except compression wave arrivals; nAg was
detectable at 3.7% concentration only by compression wave amplitude and spectral
in
-------
response and nTi02 showed only subtle detectability for spectral response at 4.9%
concentration.
IV
-------
TABLE OF CONTENTS
LIST OF TABLES vii
LIST OF FIGURES viii
CHAPTER 1 INTRODUCTION 1
1.1 Necessity of the Research 1
1.2 Research Objectives 3
1.3 Research Questions 3
1.4 Report contents 3
CHAPTER 2 TESTING APPARATUS 5
2.1 Piezoceramic Bender Elements 5
2.2 How Piezoceramics Work 6
2.3 Bender Element Configuration 7
2.4 Column Testing System Design 8
2.5 Testing Layout 10
2.6 Equipment 10
CHAPTER 3 SIGNAL PROCESSING 19
3.1 Source Signal 19
3.2 Signal Interpretation: Potential Sources of Error 21
3.2.1 Near-field Effects 21
3.2.2 Electrical Crosstalk 22
3.2.3 Boundary Conditions 22
3.2.4 Mechanical Impedance Traps 23
3.2.5 Coupling Effects 23
3.3 Testing System Delay 24
CHAPTER 4 TESTING METHODS: LITERATURE REVIEW 31
4.1 Time Domain Methods 32
4.1.1 First Arrival 32
4.1.2 Characteristic Points 32
4.1.3 Cross Correlation 33
4.2 Frequency Domain 34
4.2.1 Discrete Methods 34
4.2.2 Frequency Sweep Method 35
4.3 Test Methods: Summary 36
CHAPTER 5 CALIBRATION IN AIR AND WATER 37
5.1 Testing System Setup 37
5.2 Testing Methodology 39
5.3 Results for Testing in Air 40
5.4 Results for Testing in Water 42
CHAPTER 6 GLASS BEAD SPECIMEN PREPARATION 55
6.1 Methods of Specimen Preparation 55
6.2 The Dumping Method 56
6.3 The Stage Fill Method 57
6.4 Chosen Method 57
CHAPTER 7 BASELINE TESTING GLASS BEADS IN WATER 58
7.1 Test Setup and Preparation 58
-------
7.2 Testing Methodology 59
7.3 Data Processing 59
7.4 Pulse Signals to Highlight P-Waves 60
7.5 Pulse Signals to Highlight S-Waves 63
7.6 Frequency Sweep Method 65
7.6.1 Amplitude Spectrum Results 65
7.6.2 Phase Angle Results 66
7.7 Summary: Detection Criteria 67
CHAPTER 8 TESTING WITH NANOPARTICLE DISPERSIONS 88
8.1 Test Setup and Preparation 88
8.1.1 Pluming Process 89
8.2 Testing Methodology 90
8.3 Validating Water Trials 90
8.4 Testing with nZnO 91
8.4.1 0.03% Concentration 91
8.4.2 0.3% Concentration 93
8.4.3 2.7% Concentration 94
8.5 Summary: nZnO Testing 95
8.6 Testing with nTiOz at 4.9% Concentration 96
8.7 Testing with nAg at 3.7% Concentration 97
8.8 Overall Analysis: Nanoparticle Detectability by Seismic Methods 98
CHAPTER 9 CONCLUSIONS AND RECOMMENDATIONS 126
9.1 Conclusions 126
9.2 Recommendations 128
9.3 New Research Questions 130
ACKNOWLEDGEMENTS 131
Formulas used for calculations 133
APPENDIX 3 BACKGROUND SIGNAL ANALYSIS 154
REFERENCES 159
VI
-------
LIST OF TABLES
Table 2.1. Key dimensions of oedometers modified for bender element testing, compared
against dimensions of cell for this study 12
Table 5.1. Anticipated and measured P-wave travel times in air 44
Table 5.2. Test to differentiate electrical crosstalk from P-wave arrivals in water 44
Table 5.3. Cross correlation method: anticipated and measured P-wave arrivals in water
45
Table 7.1. Velocities associated with received pulse signals for water-saturated glass bead
specimens 70
Table 7.2. Amplitudes associated with received signals for water-saturated glass bead
specimens 70
Table 8.1. Velocities associated with received pulse signals from water and nanoparticle
dispersions in glass bead specimens 101
Table 8.2. Amplitudes associated with received pulse signals from tests in water and
nanoparticle dispersions in glass bead specimens 102
Table 8.3. Summary of detectability of nanoparticle dispersions in glass bead specimens
using time domain methods 103
vn
-------
LIST OF FIGURES
Figure 2.1. Bender element, waterproofing and grounding process 13
Figure 2.2. Bender element placed in a vinyl cap, at the required depth, and set in a
wooden block mold, shown in four different views (a, b, c and d). The mold
supported curing of epoxy used for anchoring bender element in the vinyl
cap 14
Figure 2.3. A bender element cased in a vinyl cap with epoxy 14
Figure 2.4. Schematic cross-section of test column showing anticipated P- and S-wave
travel paths; D: Column inner diameter; L: Tip-to-tip distance 15
Figure 2.5. Complete testing system layout, shown in two halves, left (a) to right (b);
equipment described in section 2.6. The nanoparticle tank is in the sonicator.
16
Figure 2.6. Electrical component layout of the testing system; connections between
equipment were made using BNC cables; soldered coaxial cable connections
to the bender elements were made as shown in Fig.2.1. S and R represent
source and receiver bender elements, respectively 17
Figure 2.7. Fluid component of the testing system: layout 18
Figure 3.1. System delay test: source (top) and receiver (bottom, in top cap) bender
elements touching to make the travel distance zero 27
Figure 3.2. System delay test, no filter: pulse signal at 8 kHz, applied at time 10 ms.
Three repetitions superimposed, 1000 recordings averaged per repetition... 28
Figure 3.3. System delay test, high pass filter at 1 kHz: pulse signal at 8 kHz, applied at
time 10 ms. Three repetitions superimposed, 1000 recordings averaged per
repetition 29
Figure 3.4. System delay test, band pass filter at 1 kHz and 16 kHz: pulse signal at 8 kHz,
applied at time 10 ms. Three repetitions superimposed, 1000 recordings
averaged per repetition, a: Extended view, green box shows zoom window b:
Detail view demonstrating system offset of-20 us with 20 us system delay.30
Figure 5.1. Frequency response for a O-to-30 kHz sweep showing the resonance
frequency of the test system in air; average of 1000 recordings 46
Figure 5.2. Frequency response for a O-to-30 kHz sweep showing the resonance
frequency of the test system in water; average of 1000 recordings 46
Figure 5.3. Acrylic spacers used for testing in air and water to hold top cap at the required
height 47
Figure 5.4. Direct (red) and reflected (green) travel paths 48
Figure 5.5. 8 kHz pulse signal in air with no filter applied, showing single repetition of
1000 recordings averaged per repetition 49
Figure 5.6. 8 kHz pulse signal in air with 1 kHz high-pass filter, showing single repetition
of 1000 recordings averaged per repetition 49
Figure 5.7. First arrival test in air with 1 kHz high pass filtering, shows three repetitions
of an 8 kHz sine pulse with 1000 recordings averaged per repetition, arrival
pick, source signal (offset for display purposes), anticipated arrival times, and
background noise threshold 50
Figure 5.8. Cross correlation response in air showing peak times for three repetitions in
air, 1000 recordings averaged per repetition 51
Vlll
-------
Figure 5.9. First arrival test in water with no filtering of an 8 kHz sine pulse, shows
source (blue), receiver (red) with three repetitions of 1000 recordings
averaged per repetition, a: Expanded view b: Detailed view 52
Figure 5.10. Differentiating electrical crosstalk from P-wave arrivals in water by varying
the tip-to-tip distance, 1000 recordings averaged per received signal 53
Figure 5.11. Cross correlation response in water showing peak times of three repetitions
of 1000 recordings averaged per repetition, a: No filtering applied b: 1 kHz
high-pass filter applied 54
Figure 7.1. Frequency response for a O-to-30 kHz sweep showing the resonance
frequency of the bender element in a water-saturated glass bead specimen;
result of 1000 recordings averaged, with no filtering applied 71
Figure 7.2. 8 kHz pulse signal (1000 recordings averaged) in a saturated glass bead
specimen with no filter applied 72
Figure 7.3. Representative result for consecutive 8 kHz pulses with 200 Hz high-pass
filter applied, received signals of first and second trials are shown; result of
1000 recordings averaged for each 73
Figure 7.4. Representative result of an 8 kHz sine pulse and received signals of 1000
recordings averaged for each, emphasizing reflected-path P-wave
propagation. Trials 1 and 2 are conducted sequentially under near-identical
test conditions. Note the irregularity present at the initiation of the source sine
pulse 74
Figure 7.5. Summary of 8 kHz sine pulse highlighting P-wave velocities in water-
saturated glass bead specimens 75
Figure 7.6. Representative picks of characteristic points used to compare the amplitudes
of received slow P-wave signals, 1000 averages and 200 Hz high-pass filter.
76
Figure 7.7. Summary of received signal amplitudes of characteristic points (described in
text) from 8 kHz pulse signals highlighting P-waves in water-soaked glass
bead specimens 77
Figure 7.8. Check for optimal sine pulse frequency to test for shear in saturated glass
beads; received signals are 1 repetition of 1000 recordings averaged per
repetition, under 200 Hz high-pass filter 78
Figure 7.9. Representative result of a 1 kHz sine pulse test in saturated glass beads
showing two consecutive pulses, demonstrating that disturbances due to the
first pulse do not completely decay prior to the arrival of the second pulse. A
single repetition from each trial is shown; 1000 recordings averaged per
repetition 79
Figure 7.10. Representative result of 1 kHz sine pulse (shear); shows near-field effects,
trial 2 first arrival earlier than trial 1 arrival; single repetition of 1000
recordings averaged per repetition, 200 Hz high-pass filter 80
Figure 7.11. Summary of 1 kHz sine pulse highlighting S-wave (shear) received signal
velocities in water-saturated glass bead specimens 81
Figure 7.12. Representative picks of characteristic points used to compare the amplitudes
of received S-wave signals, 1000 averages and 200 Hz high-pass filter 82
IX
-------
Figure 7.13. Summary of received-signal peak-to-peak amplitude differences from 1 kHz
pulse signals highlighting S-waves in water-saturated glass bead specimens.
83
Figure 7.14. Representative coherence for 30 kHz sweep with 200 Hz high-pass filter in a
water-saturated glass bead specimen, showing result of 1000 recordings
averaged each, for Trials 1 and 2 84
Figure 7.15. Representative amplitude spectrum of 30 kHz sweep for all water-saturated
glass bead specimens; 1000 recordings averaged per repetition 85
Figure 7.16. Residual signals equal to the difference between the spectral responses of
trial 1 and trial 2 in water-saturated glass bead specimens, used to quantify
the sensitivity of the test system. The average residual signal is the averaged
result of the three residual signals of the specimens, and it is used as the
baseline 86
Figure 7.17. Representative result of unwrapped phase angles for trials 1 and 2 in water-
saturated glass bead specimen, three repetitions of 1000 recordings averaged
per repetition, high-pass filter at 200 Hz applied. Range of high coherence is
expected from 7 to 25 kHz 87
Figure 8.1. Summary of 8 kHz results highlighting slow P-wave velocity for all
nanoparticle dispersions 104
Figure 8.2. Summary of P-wave characteristic point amplitudes from 8 kHz pulse tests in
water and in the presence of nanoparticle dispersions in saturated glass bead
specimens. The signs indicate an increase (+), decrease (-) or no change (0) in
amplitude for the P-waves in the presence of nanoparticles from the baseline
trial 2 result. The black dashed lines show the baseline against which to
compare nano test amplitudes (from Sec. 7.4) 105
Figure 8.3. Summary of 1 kHz results highlighting S-wave velocity for all nanoparticle
dispersions 106
Figure 8.4. Summary of S-wave amplitudes from 1 kHz pulse tests in water and in the
presence of nanoparticle dispersions in saturated glass bead specimens. The
black dashed lines show the AP2 baseline amplitude range (from Sec. 7.5).
107
Figure 8.5. 8 kHz sine pulse with 200 Hz high-pass filter applied in saturated glass bead
specimen for testing differences in compression with water and nZnO at
0.03% dispersion, showing three repetitions of 1000 recordings averaged per
repetition 108
Figure 8.6. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nZnO at 0.03%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition, and consistent first arrival pick 109
Figure 8.7. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nZnO at 0.03%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition 110
Figure 8.8. Residual signals from the differences of spectral response in water and in
nZnO at 0.03%, of three repetitions of 1000 recordings averaged per
repetition, compared to the average baseline residual signal 110
x
-------
Figure 8.9. 8 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing differences in compression with water and nZnO at
0.3% dispersion, showing three repetitions of 1000 recordings averaged per
repetition, a: Expanded view of entire received signal; note amplitude
variation between the three repetitions of each trial b: Detail view showing
representative picks for characteristic amplitude points, residual fast P-waves
and slow P-wave arrivals Ill
Figure 8.10. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nZnO at 0.3%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition, and consistent first arrival pick 112
Figure 8.11. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nZnO at 0.3%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition 113
Figure 8.12. Residual signals from the differences of spectral response in water and in
nZnO at 0.3%, of three repetitions of 1000 recordings averaged per
repetition, compared to the average baseline residual signal 113
Figure 8.13. 8 kHz sine pulse with 200 Hz high-pass filter applied in saturated glass bead
specimen for testing differences in compression with water and nZnO at
2.7% dispersion, showing three repetitions of 1000 recordings averaged per
repetition, a: Expanded view of entire received signal; note amplitude
variation between the three repetitions of each trial b: Detail view showing
representative picks for characteristic amplitude points, residual fast P-wave
and slow P-wave arrivals 114
Figure 8.14. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nZnO at 2.7%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition, and consistent first arrival pick 115
Figure 8.15. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nZnO at 2.7%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition 116
Figure 8.16. Residual signals from the differences of spectral response in water and in
nZnO at 2.7%, of three repetitions of 1000 recordings averaged per
repetition, compared to the average baseline residual signal 116
Figure 8.17. 8 kHz sine pulse with 200 Hz high-pass filter applied in saturated glass bead
specimen for testing differences in compression with water and nTiOz at
4.9% dispersion, showing three repetitions of 1000 recordings averaged per
repetition 117
Figure 8.18. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nTiOz at 4.9%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition, and first arrival pick for nTiOz coming in earlier than water 118
XI
-------
Figure 8.19. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nTiOz at 4.9%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition 119
Figure 8.20. Residual signals from the differences of spectral response in water and in
nTiOz at 4.9%, of three repetitions of 1000 recordings averaged per
repetition, compared to the average baseline residual signal 119
Figure 8.21. 8 kHz sine pulse with 200 Hz high-pass filter applied in saturated glass bead
specimen for testing differences in compression with water and nAg at 3.7%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition 120
Figure 8.22. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nAg at 3.7%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition, and first arrival pick for nAg coming in earlier than for water. .121
Figure 8.23. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nAg at 3.7%
dispersion, showing three repetitions of 1000 recordings averaged per
repetition 122
Figure 8.24. Residual signals from the differences of spectral response in water and in
nAg at 3.7%, of three repetitions of 1000 recordings averaged per repetition,
compared to the average baseline residual signal 122
Figure 8.25. Signals from slow compression wave comparing response in the presence of
nanoparticles to the baseline received signal 123
Figure 8.26. Signals from shear waves comparing response in the presence of
nanoparticles to the baseline received signal 124
Figure 8.27. Signals from 30 kHz sweeps comparing response in the presence of
nanoparticles to the baseline 125
xn
-------
CHAPTER 1
INTRODUCTION
This study was part of a larger project concerning the detection of nanoparticles used
in engineered nanomaterials as they disperse throughout the environment. This study
addresses the potential for seismic methods to be implemented in detecting such
nanoparticles in a natural environment. A testing system was built and calibrated in air,
water, and essentially saturated glass beads. Testing was then conducted for the presence
of various types of nanoparticles dispersed in the pore fluid of essentially saturated glass
bead specimens.
1.1 Necessity of the Research
Nanotechnology is the manipulation and control of substances on the nanoscale. The
nanoscale measures particles in nanometers, where one nanometer is one billionth of a
meter. When particles from the nanoscale are compared to particles of the same material
on the macro-scale, the physical and chemical properties often differ. This phenomenon
enables new applications, processes and technology (National Nanotechnology Initiative,
2009).
According to the National Nanotechnology Initiative (2009), three types of
nanoparticles exist: naturally occurring; incidental; and engineered. Naturally occurring
nanoparticles for example, exist in the human body, which uses them to control many
systems and processes. An example of this is hemoglobin, which is a protein nanoparticle
1
-------
that is used to transport oxygen. Incidental nanoparticles are created as by-products of
processes such as combustion and other industrial activities. When particles are
purposefully manufactured on the nanoscale, they are known as engineered nanoparticles.
Nanotechnology is a growing industry which has the potential of improving the
standard of living and benefitting society. Industries such as medicine, energy, and
information technology are all currently exploring possibilities with nanoparticles
(National Nanotechnology Initiative, 2009). As more industries start utilizing engineered
nanoparticles, they have the potential to be released into the environment by various
processes. The impacts of engineered nanoparticles on human health and the environment
are unknown due to the fact that applications are novel and limited research has been
conducted. This is where the primary environmental concern with engineered
nanoparticles comes into play. There are no current proven methods of detecting the fate
and transport of nanoparticles in the subsurface (Conlon, 2009). For this reason, new
testing practices and detection techniques have to be explored.
Williams et al. (2005) used a column containing sand to monitor the effects of
microbial activity on metal ions over a number of days. The microbial activities led to the
development of nanoparticles along the sand surfaces and in assemblages formed within
the pore spaces. Seismic and electrical techniques were applied to observe variances from
initial readings caused by the development and presence of the nanoparticles. The authors
found that subtle changes in grain size, consolidation state, and type of pore fluid
saturation of the material can alter the velocity and amplitude of the seismic response to
varying degrees. The results from the monitoring efforts by the authors led to
development of this research in which a testing system is developed and optimized to
2
-------
identify the presence of nanoparticles in essentially saturated glass beads by seismic
methods.
1.2 Research Obj ectives
This project comprises two research objectives; first to design, build and optimize a
laboratory seismic testing system for essentially saturated glass beads, and second to use
the system to explore the seismic response of select nanoparticle dispersions in an
essentially saturated granular matrix.
1.3 Research Questions
Three research questions are addressed in this study:
1. Which type of seismic waveform and function is most suitable for testing?
2. To what degree are test results repeatable?
3. Can the presence of select nanoparticle dispersions be detected by variations
in the seismic response?
1.4 Report contents
Chapter 2 presents test column design criteria, and test system components and
layout. It also addresses the composition of piezoceramic elements which were used to
actuate and receive seismic energy, and how they were prepared for this research.
Chapter 3 addresses signal processing, complications caused by near-field effects, and the
-------
potential sources of error with signal interpretation. Chapter 4 reviews previous research
efforts that have utilized bender elements, and presents suitable test methods for this
research. Chapter 5 presents the calibration of the system in air and water. Chapter 6
presents the methods used to prepare the glass bead specimens. Chapter 7 presents the
calibration of the system using water-saturated glass beads. Chapter 8 presents the testing
of nanoparticle dispersions. Chapter 9 presents the conclusions and recommendations.
Chapter 9 also presents new research questions that arose from this study.
-------
CHAPTER 2
TESTING APPARATUS
This chapter addresses the composition of piezoceramic elements, how they actuate
and receive seismic energy, and how they were prepared for this study. Also presented
are the column design criteria, test system layout and other components of the testing
system.
2.1 Piezoceramic Bender Elements
Piezoceramic bender elements are transducers that can be used interchangeably to
either generate or receive seismic body waves. The bender elements convert electrical
energy to mechanical energy and vice versa. Bender elements were first used to measure
shear-wave velocity of clay specimens in 1978 by Shirley and Hampton (Clayton et al,
2004). From 1978 until today, piezoceramic bender elements have been the choice of
transducer for use by many researchers when mechanical properties of sediments were
required in the laboratory (Dyvik and Olsen, 1991).
Bender elements are also utilized because of their good coupling capability between
the transducer and testing media (Lee and Santamarina, 2005) to measure variances in
response as seismic energy is propagated through saturated granular media. Lee and
Santamarina (2005) carried out a thorough study that addressed bender element
installations, prevention of electromagnetic coupling, directivity of transmitted energy,
resonance condition, detection of first arrival, and near-field effects. Da Fonseca et al.
-------
(2008) list the methods available for testing with bender elements and provide advice for
choosing the most suitable.
2.2 How Piezoceramics Work
A piezoceramic material generates and receives sound and voltage by the
phenomenon known as the piezoelectric effect (e.g., Piezo Systems, 2009). Piezoceramic
crystals have an asymmetrical lattice structure that leads to polarization densities when
the crystal undergoes mechanical deformation (flexing; Birkholz, 1995). This in turn
leads to a voltage difference being created across the crystal. Similarly, if a voltage
difference were applied on opposing faces of the crystal, this would cause the crystal to
flex.
This principle applies when piezoceramic elements are placed within test specimens
in the following manner: as seismic body waves (i.e. S- and P-waves) strike the surface
of the piezoceramic, the piezoceramic element flexes and this creates a voltage difference
that can be captured electronically. When a voltage difference is applied across a
piezoceramic bender element that is embedded within a granular specimen, the element
vibrates, creating body waves that travel through the specimen (Blewett et al., 1999).
Bender elements generate both shear (S) and compression (P) waves when they actuate in
granular media, where S-waves are generated in the form of a frontal lobe and the P-
waves as side lobes with respect to the bender element (Lee and Santamarina, 2005).
-------
2.3 Bender Element Configuration
The bender elements were purchased from Piezo Systems Inc., and were 2-piezo layer
transducers, made with PSI-5A4E piezoceramic, parallel-poled, using nickel electrodes
and brass center reinforcement. The elements were 12.7 mm square and 0.5 mm thick.
Two important parameters for the source bender elements are the free deflection and the
maximum force generated for the voltage applied. An important parameter for the
receiver bender element is the voltage generated by the force applied (Leong et al., 2005).
The force generated by the source and the voltage generated by the receiver are
dependent upon the width of the bender elements. As the width increases, the force
generated at the source increases and the voltage generated at the receiver decreases.
Widths of bender elements typically range from 6 to 15 mm (Leong et al., 2005).
The free deflection and output voltage of the bender element for a given applied
voltage is dependent upon the cantilever length. Keeping the cantilever short makes the
resonance frequency of the bender element dependent on the bender element properties
and the anchoring properties, whereas a long cantilever would make the resonance
frequency dependent on the sediment properties (Lee and Santamarina, 2005). A shorter
cantilever yields a higher resonant frequency and a shorter wavelength at resonance. A
shorter cantilever length is preferred in this study so that resonant frequency will remain
relatively constant for all testing media. The cantilever length used was 4.2 mm which is
1/3 the total length of the bender element.
Bender elements are high impedance devices and can therefore short electrically
when exposed to moisture (Dyvik and Madshus, 1986). Figures 2.1-2.3 show the
-------
process by which the bender elements were cased and water proofed: they were coated
with a thin layer of polyurethane, then painted over with silver paint and carefully potted
in vinyl caps using epoxy. The silver conductive paint coating is applied to help properly
ground the bender element to minimize electrical crosstalk (Wang et al., 2007). Professor
Carlos Santamarina and his colleagues at Georgia Tech recommended the waterproof
polyurethane coating for better actuation and reception of signals. If the bender elements
produce sound and have the resistivity of an open circuit (or very high resistance, on the
order of Mega-Ohms) after the polyurethane and silver coatings are applied, they are
properly prepared (Changho Lee, personal communication, 9/15/08). The final product of
the bender elements potted with epoxy in vinyl caps was fixed within the testing system
by applying RTV silicone. The silicone was applied on the outside of the vinyl cap of the
potted bender element, which was then placed within the test system. Silicone was
chosen due to its inert and waterproof properties (Zhihai et al., 2008). The different
materials used for preparing and holding the bender element in the test system create
impedance traps that prevent waves generated at the anchor from travelling through the
structure of the test system to the receiver, and therefore causing error.
2.4 Column Testing System Design
The testing column was constructed from a clear PVC tube of 15.2 cm inside
diameter (D), mounted on a PVC base, with an acrylic top cap made to fit snugly inside
the column (Fig. 2.4). The top cap was fitted with a rubber 0-ring. The base was mounted
on four column supports, and had an inlet valve attached to it. The purpose of the inlet
-------
was for pluming fluids into the column by gravity. The top cap had an outlet for drainage
of excess liquid or air from the specimen. A handle was attached to the top cap.
The column geometry is comparable to that of an oedometer, which is an instrument
used to measure the rate and amount of consolidation of a specimen as pressure is
applied; however, the functions of an oedometer and our testing apparatus are different.
Wave travel paths can be compared between the two systems. Apart from oedometers,
other common test systems that have been used with bender element testing are triaxial
testing systems and large tanks with the bender elements placed on stands within the
tank.
When bender element testing is incorporated in an oedometer, the tip-to-tip distance
between bender elements (L) and not the full height of the specimen must be taken as the
travel path length (Fig. 2.4; Dyvik and Madshus, 1986). The dimensions of the testing
system were selected considering three main design criteria as presented below.
The first criterion addressed the ratio of the column inner diameter (D) to the tip-to-
tip distance (L) of the bender elements; the D:L ratio. Some D:L ratios used in previous
research with oedometers modified for bender element testing are presented in Table 2.1.
The range considered in this body of research was: 4.2 > D:L > 2.2.
The second criterion addressed the relationship between (L) and the wavelength (X) of
the actuated signal; this relationship addresses the potential for P-wave coupling with S-
wave arrivals (so-called "near-field effects"). According to Wang et al. (2007), this
coupling effect can be avoided by configuring the test cell so that: L:X >2.
The third criterion was to use the shortest acceptable travel path length for the waves.
This criterion was used to minimize signal attenuation between the source and the
9
-------
receiver and to keep the volume of the testing system to a minimum in order to minimize
the quantity of experimental treatment used, to control costs and minimize waste.
For the testing system, the only set parameter out of the two ratios (D:L and L:X) was
the diameter (D). The (L) value could be varied because the top cap was mobile in the
vertical direction and (X) could be adjusted by varying the actuation frequency of the
source signal. The L value chosen for testing is presented in chapter 5. Calibration of the
column in air and water was carried out approximately at the resonance frequency of the
potted bender elements, because this improves the signal to noise ratio (Wang et al.,
2007). For testing in saturated granular media, depending on whether P-waves or S-
waves were targeted, the frequency and therefore wavelength parameters were adjusted
within the criteria provided above, until the clearest signals were received.
2.5 Testing Layout
The testing system layout was comprised of the mechanical test cell, electrical
components and a fluid system. Figures 2.5 through 2.7 illustrate the components.
2.6 Equipment
Equipment ancillary to the test column included:
Function generator: Agilent 33220A
Linear amplifier: Piezo systems Inc., EPA 104
Bender elements: Piezo Systems Inc., described previously
10
-------
Filter-Amplifier: Krohn-Hite, 3364
Signal analyzer: Dataphysics SignalCalc Dynamic Signal Analyzer
Sonicator: Branson 5510
Peristaltic pump: Ismatec C.P. 78023-10
Digital caliper: Cen-tech 47257 (not shown in figures)
The process of actuating, transmitting and receiving a signal is as follows (Fig. 2.5 and
2.6):
The source signal is generated via function generator
The signal is amplified through a linear amplifier to increase signal to noise ratio
The amplified signal is transmitted to the source bender element
The source bender element converts the electrical signal to a mechanical wave
The actuated mechanical wave is transmitted through the specimen to the receiver
bender element
The receiver bender element converts the mechanical wave to an electrical signal
The electrical signal is filtered to reduce noise, and displayed and recorded on a
digital signal analyzer
The fluid system layout (Fig. 2.7) is described in chapter 8.
11
-------
Table 2.1. Key dimensions of oedometers modified for bender element testing, compared
against dimensions of cell for this study
Reference
Dyvik and Olsen
(1991)
Zeng and Ni
(1998)
Lee and Santamarina
(2005)
Lee and Santamarina
(2005)
Lee et al. (2007)
This study
Specimen diameter
(D, mm)
66.7
152.4
70
100
74
152.4
Tip-to-tip distance
(L, mm)
16
68.6
32
19.8
29
62.5
D:L
4.2
2.2
2.2
5.1
2.6
2.4
12
-------
Solder connection to (+) wire
Bender
element
Coaxial cable
Brass (inner) electrode
Negative wire
(grounding)
Piezoceramic
Center metal
shim (brass)
Solder connection to the (+) wire
Bender
element
Polyurethane
Coaxial cable
Space left so that
the silver paint
can connect to
the electrode
Bender element
coated with
polyurethane and
silver paint
Brass (inner) electrode
Solder connection
to the (-) wire
Figure 2.1. Bender element, waterproofing and grounding process
a: Mounted piezoceramic bender element with coaxial cable wiring
b: Isometric view of piezoceramic bender element showing layering
c: Bender element coated with polyurethane and high-purity silver paint for
waterproofing and grounding.
13
-------
Figure 2.2. Bender element placed in a vinyl cap, at the required depth, and set in a
wooden block mold, shown in four different views (a, b, c and d). The mold supported
curing of epoxy used for anchoring bender element in the vinyl cap.
Figure 2.3. A bender element cased in a vinyl cap with epoxy
a: front view
b: rear view
14
-------
PVC column
Receiver bender
element
>\
.* e*Jr - =-... ^i
Handle
p
PVC base
plate
Source
bender element
Acrylic top cap with
rubber O-ring
Glass beads
(0.5 mm diameter)
Figure 2.4. Schematic cross-section of test column showing anticipated P- and S-wave
travel paths; D: Column inner diameter; L: Tip-to-tip distance.
15
-------
Figure 2.5. Complete testing system layout, shown in two halves, left (a) to right (b);
equipment described in section 2.6. The nanoparticle tank is in the sonicator.
16
-------
Signal Analyzer
Test Column
Filter/Amplifier
"S
Figure 2.6. Electrical component layout of the testing system; connections between
equipment were made using BNC cables; soldered coaxial cable connections to the
bender elements were made as shown in Fig.2.1. S and R represent source and receiver
bender elements, respectively.
17
-------
Water tank
Test
column
Nanoparticle
tank
1ST
4-wayspHtter
a
a Valve
&
Disposal
Water tank
Test
column
LnJ
TV an
op
s.
Sonicator
Peristaltic
[_i i_|l Pump
c. LJ 4-way splitter
S Valve
a
Disposal
Figure 2.7. Fluid component of the testing system: layout
a: Pluming system used for testing nano-oxides
b: Pluming system used for testing nano-metals
18
-------
CHAPTER 3
SIGNAL PROCESSING
This chapter outlines how the source signal was chosen, complications posed by near-
field effects, other potential sources of error with signal interpretation, and the signal
processing required to receive clear signals.
3.1 Source Signal
Received signals are distorted versions of the input signal due to the effects of
transfer functions. The shape of the input signal is important in reducing unwanted effects
(Arroyo et al., 2003). The most commonly used source waves with bender elements are
sine and square waves (Leong et al., 2005).
Leong et al. (2005) used bender elements to determine the shear wave velocities of
sand, mudstone, and kaolin specimens. The results were examined with respect to
characteristics of the waveform type, magnitude, and frequency applied to the transmitter
bender elements. The two types of waveforms considered were square waves and sine
waves. The authors showed that when square waves are used as the source signal, the
received signals do not resemble the transmitted signal because the rise time of a square
wave is practically zero which corresponds to an infinite frequency, leading to
uncertainty in arrival time. When sine waves were used as the source, there was less
ambiguity in the arrival times of the received signals when compared with those of the
square wave. The authors state that uncertainty in the interpretation of bender element
19
-------
tests arises due to the quality of the received signal and for this reason a sinusoidal input
is the preferred choice.
Jovicic et al. (1996) also carried out a study to determine whether sine or square
waves were more appropriate to record the shear wave arrivals. The authors found that
sinusoidal waves were the simplest way to obtain a bender element output that may be
interpreted objectively, since the actuation frequency can be adjusted according to the
required travel distance and test specimen stiffness to avoid near-field effects (described
below). The authors claimed that square waves will always have near-field effects
because they are comprised of a spectrum of frequencies, which make the square waves
complex to analyze and near-field effects difficult to nullify.
Arroyo et al. (2003) also studied near-field effects with bender elements by analyzing
multiple source waves. The amplitude of the near-field effects caused by sine waves was
10% of the output peak signal (S-wave) and that of the near-field effects caused by the
square waves was 30% of the output peak, therefore 3 times larger than those caused by
sine waves. These authors concluded that square waves were the least favorable option in
terms of picking first arrivals and reducing near-field effects.
Therefore, from the literature it was evident that sine waves were more suitable
source signals than square waves. Arroyo et al. (2003) and Jovicic et al. (1996) also
considered distorted sine waves, which reduced the near-field effects even more than
regular sine waves, but distorted sine waves were not considered for our research for
practical reasons.
20
-------
3.2 Signal Interpretation: Potential Sources of Error
Bender element testing involves numerous potential sources of error and biases.
Identifying them in advance helps to reduce their effects.
3.2.1 Near-field Effects
Picking first arrivals of S-waves can be confounded by the effects of the near-field,
which affect only Vs and not VP (Brignoli et al., 1996). In a complex test system where
received signals are not limited to plane wave propagation between source and receiver,
near-field effects, which are the mixed radiations of P-waves and S-waves (Wang et al.,
2007; Arroyo et al., 2003), occur. As the name implies, this confounding effect dies out
as distance from the source increases because of the difference between P- and S-wave
velocities. As stated earlier, S-waves are generated in the form of a frontal lobe and the P-
waves as side lobes with respect to the bender element (Lee and Santamarina, 2005). As
the direct-transmission S-waves arrive at the receiver bender element, so do P-waves
reflected off the testing system walls. Wang et al. (2007) avoided the effects of P-wave
interference on picking S-wave arrivals by placing the receiver at least two wavelengths
away from the source. (This criterion was presented in section 2.4.)
A similar criterion was also found by Jovicic et al. (1996) when testing with sine
waves. The authors found that the ratio of travel distance (L) and S-wavelength (A) of the
source signal can be optimized to limit near-field effects. For low values of L:X, the near-
field effects were present at the receiver and as the L:X ratio increased, the effects of the
near-field on the ability to pick shear arrivals decreased. This meant that the travel
21
-------
distance of the shear wave, or the tip-to-tip distance between the source and receiver (L)
had to be increased in order to significantly decrease the near-field effects for a given (A).
For this study, as mentioned in section 2.4, the volume of the test system and the
wave travel path had to be minimal. These conflicting requirements resulted in a mid-
range selection of D:L (Table 2.1). Detecting the actual shear wave arrival is less a
priority for this study than establishing a repeatable response ("signature").
3.2.2 Electrical Crosstalk
Electrical crosstalk can also be a major source of error. The wiring of the bender
elements can influence how much electrical crosstalk is present. Parallel - aligned bender
elements have a shielding effect when the outer electrodes are connected to the ground
and so crosstalk can be significantly reduced (Lee and Santamarina, 2005). In our
testing, the bender elements were aligned parallel to one another and a grounding setup
presented by Wang et al. (2007) was implemented but crosstalk was still excessive;
therefore further steps had to be taken to reduce it. A voltage divider was applied to the
signal passing from the linear amplifier to the signal analyzer (Fig. 2.6.), and the source
and receiver inputs on the analyzer were spaced as far apart as possible. With these two
additional steps, the electrical crosstalk in the received signals was significantly reduced.
3.2.3 Boundary Conditions
Rigid boundary conditions also cause wave distortions due to the interference of the
direct waves with reflected waves (e.g. Arulnathan et al., 1998). For example, for the
22
-------
direct-transmission ray path, after the incoming energy first actuates the receiver bender
element, the energy reflects off the plate on which the receiver bender element is
mounted and is seen again as another arrival as it passes the receiver bender element in
the opposite direction. In our experiment, the most significant reflecting surface for
direct-transmission energy would be the top cap in which the receiver was mounted.
3.2.4 Mechanical Impedance Traps
Source bender elements generate a signal at the anchor in addition to what is
generated along the length (Lee and Santamarina, 2005). In the absence of isolation or
mechanical impedance traps, this signal could reach the receiver element by travelling
along the cell walls and therefore introduce error in the received signal by short circuiting
the test specimen altogether. The bender elements used in this research were prepared to
minimize this error. As mentioned earlier, potting the bender elements using epoxy
within a vinyl casing, and then fixing them onto the top cap and base plate with RTV
silicone creates impedance traps which limit the transfer of mechanical wave energy (Lee
and Santamarina, 2005).
3.2.5 Coupling Effects
Coupling between transducer and test medium is critical in bender element testing.
Void formation around the source bender element is another potential source of error
(Lee and Santamarina, 2005; Wang et al., 2007). Therefore care should be taken with
installing the bender elements, and to densify and compact the specimen properly in order
to minimize the production of voids between the element and the test specimen.
23
-------
3.3 Testing System Delay
When making measurements with a complex testing system, any time offset in signal
transmission caused by the testing system must be accounted for. This section describes
system offsets and system delays caused by the testing system.
Here, we define system offset as the time difference between the time of actuation of
the source signal (by the function generator) and the recorded time for the source sensor
(on the signal analyzer). If the same system offset is present in the source signal and the
received signal, the offsets cancel out. We define a system delay to result when the
system offset of the source signal and the system offset of the received signal are unequal
and so do not cancel.
To check for system offset and delay, the source and receiver bender elements were
made to touch at the tips (Fig. 3.1); a bender element from the same production batch was
substituted for the bender element from the base plate of the testing system for
maneuverability purposes. By making the bender elements touch, the travel distance
between them was zero so that the travel time for the signal to be received was also
theoretically zero. With this configuration, in the absence of system delay, the actuation
time of the source signal should be the same as the first arrival time of the received
signal.
The time interval between pulses initially used for data collection was 10 ms. The
source bender element was actuated with single sine pulses at 8 kHz and 5 V, with 10 ms
intervals between pulses. This initial choice of time interval between pulses was
increased in future tests (Sec. 7.4). The sine pulses were generated by using a burst
24
-------
function, which allows the function generator to create individual sinusoidal pulses at
predetermined intervals. The signals were amplified to 10 V. The amplification was
applied to replicate planned testing in saturated glass beads, where the signal would be
amplified to improve signal to noise ratio. Trials were carried out with no filtering, with a
high-pass filter at 1 kHz and with a band-pass filter at 1 and 16 kHz. The high-pass filter
presents a situation where the received signal is processed through the filter equipment
once. The band-pass filter presents a situation where the received signal is processed
through the filter twice.
The results without filtering are presented in Fig. 3.2; three repetitions are shown with
each repetition consisting of the average of 1000 pulses. Averaging or stacking pulses to
improve quality of received bender element signals was demonstrated by Wang et al.
(2007). The number of averages used in this research was determined experimentally.
The source signal was actuated at time 10 ms and the source initiation recorded on the
signal analyzer was at time 9.98 ms, giving a negative offset of 20 ^is. The received signal
pick occurs one sample later than the source initiation time. The sampling interval was 10
us; which was near the shortest possible sampling rate of 9.4 us with the signal analyzer
used. Therefore a 10 ^is system delay is present, with the received signal trailing the
source signal.
To test the cause of the negative system offset, a BNC cable was connected directly
from the function generator to the signal analyzer, bypassing the linear amplifier and
voltage divider (Fig. 2.6), and the same single sine pulse was applied at 10 ms intervals,
with identical results. This test demonstrated that the system offset is not caused by the
linear amplifier or voltage divider.
25
-------
Results of the system offset/delay test using a high-pass filter at 1 kHz are presented
in Fig. 3.3. Variations in amplitude and phase were present in the shape of the received
signal with respect to the received signal without filtering (compare Fig. 3.2a to 3.3a), but
they were well after the first arrivals. The same time offset and delay as without filtering
were encountered.
The results of using the band-pass filter are presented in Fig. 3.4; these signals show
the same system offset but a longer system delay. The source signal was received at 9.98
ms as before, and the received signal was at 10 ms, yielding a system delay of 20 ^is (two
samples), with the received signal trailing the source signal.
26
-------
Figure 3.1. System delay test: source (top) and receiver (bottom, in top cap) bender
elements touching to make the travel distance zero
27
-------
0.0096
0.0098
0.01 0.0102
Time (s)
0.0104
D.
E
-0.1 -
0.2-
-0.3-
-0.4
0.0099
0.0106
0.0108
0.01
0.0101
Time (s)
0.0102
0.0103
Figure 3.2. System delay test, no filter: pulse signal at 8 kHz, applied at time 10 ms.
Three repetitions superimposed, 1000 recordings averaged per repetition.
a: Extended view, green box shows zoom window
b: Detail view demonstrating system offset of-20 us with 10 us system delay
28
-------
-0.5 L
0.0096
0.15-
0.05-
E
<
0.0098
0.01
0.0102
Time (s)
0.0104
0.0106
0.0108
-0.05
-0.1 -
-0.15-
0.0099
0.01
0.0101
Time (s)
0.0102
0.0103
Figure 3.3. System delay test, high pass filter at 1 kHz: pulse signal at 8 kHz, applied at
time 10 ms. Three repetitions superimposed, 1000 recordings averaged per repetition.
a: Extended view, green box shows zoom window
b: Detail view demonstrating system offset of-20 us with 10 us system delay
29
-------
0.0096
0.2
0.15
0.1
0.05
>^
0)
I
E
<
-0.05
-0.1
-0.15
-0.2
0.0099
0.0098
0.01
0.0102
Time (s)
0.0104
0.0106
0.0108
X: 0.00998
Y: -0.00967
X: 0.01
Y: 0.001595
0.01
0.0101
Time (s)
0.0102
0.0103
Figure 3.4. System delay test, band pass filter at 1 kHz and 16 kHz: pulse signal at 8 kHz,
applied at time 10 ms. Three repetitions superimposed, 1000 recordings averaged per
repetition.
a: Extended view, green box shows zoom window
b: Detail view demonstrating system offset of-20 us with 20 us system delay.
30
-------
CHAPTER 4
TESTING METHODS: LITERATURE REVIEW
There are many methods available for testing with bender elements and researchers
are not in complete agreement about which method is best. No standard exists for the
testing procedures or for the interpretation of the results (Da Fonseca et al., 2008).
One appeal of bender elements for measuring mechanical properties of specimens is
that the concept is simple; seismic energy is actuated and received by bender elements
and the received signal is analyzed to identify the seismic signature of the system. Many
researchers use bender element testing to find the shear wave velocity of the specimens
being tested, from which other mechanical properties can be derived. It is important to
point out that we are not primarily concerned with the received signal velocity, rather we
seek a means to monitor for change in the response of the system in the presence of an
experimental treatment. This change could be related to the frequency content and shape
of the signals, in addition to velocity and amplitude as was seen in testing carried out by
Williams et al. (2005).
This chapter presents the testing methods used by previous researchers in bender
element studies to determine shear and compression wave velocities from which other
mechanical specimen properties could be derived. The test methods are divided into time
domain methods and frequency domain methods. No prior research on the analysis of
frequency content and signal amplitude from bender element testing was found, although
this topic was addressed in this study.
31
-------
4.1 Time Domain Methods
This section analyzes time domain methods that have been used with bender element
tests in the lab to determine velocities. Three methods were considered: first arrival;
characteristic points; and cross correlation.
4.1.1 First Arrival
The first arrival method has been used successfully to determine the travel time with
bender element testing by numerous researchers (Leong et al., 2005). For example, Lee et
al. (2007) used the first arrival method in a modified oedometer cell to calculate the shear
wave velocity from recorded data to estimate consolidation characteristics of a marine
clay specimen. The first arrival method utilizes the length of the travel path between the
source and receiver bender element and the travel time derived from the transmitted and
received signals to calculate the velocity of the received energy. Picking the first arrival
has been documented as a difficult task due to uncertainty associated with correctly
picking the first deflection point (Arulnathan et al., 1998). The uncertainty results from
signal attenuation, noise such as electrical crosstalk and, in some cases, near-field effects
(Sec. 3.2).
4.1.2 Characteristic Points
The characteristic points method is like the first arrival method, except travel time is
calculated from more easily identified points on the wave train than the point of first
32
-------
arrival. This is done to avoid the uncertainty issues described above. The drawback with
this method is that wave velocities are underestimated due to signal attenuation.
Clayton et al. (2004) used the characteristic points method in research concentrated
towards improving the objectivity and repeatability of shear wave velocity measurements
by bender elements. Experiments were carried out on Leighton Buzzard sand in a triaxial
testing apparatus. The source was mounted at the base and receivers were mounted along
the side wall and in the top cap. Discrete sine pulses were actuated at frequencies from 6
kHz to 30 kHz. The characteristic points considered were the first trough and subsequent
peak associated with the first deflection. Travel times were determined from differences
between timings for characteristic points of the received signals at successive receivers.
The results showed better repeatability at higher frequencies (10 to 30 kHz) when
compared with lower frequencies (6 to 10 kHz).
4.1.3 Cross Correlation
Cross correlation indicates similarities between the source and receiver waveforms.
For well-correlated data, the time associated with the peak of the cross correlation relates
directly to the transmission time of the wave, which is simply the difference in time from
initiation to the peak of the computed function (Reynolds, 2000).
Viggiani and Atkinson (1995) used cross correlation to determine shear wave velocity
of a reconstituted clay specimen in a triaxial testing apparatus from bender element
testing. Source and receiver bender elements were placed in the end caps of the triaxial
apparatus. The authors concluded that their results from the cross-correlation method
were very accurate; however, the results could vary depending on the testing apparatus.
33
-------
Lee and Santamarina (2005) found that the results can also vary when signals being
correlated are not of the same nature. Distortions in the received signal with respect to the
source signal complicate the correlation function and confound the determination of
transmission time.
4.2 Frequency Domain
The frequency domain methods considered include discrete methods and the
frequency sweep method.
4.2.1 Discrete Methods
Considering the use of bender element testing in the time domain, Greening and Nash
(2004) found that problems caused by transient effects such as reflected waves are
removed if impulsive signals are replaced with a continuous harmonic signal. Discrete
methods use continuous sinusoids at select frequencies. According to Greening and Nash
(2004) and Da Fonseca et al. (2008), discrete methods are time consuming but provide a
way to determine the travel time of the system in the frequency domain without involving
measurements of travel distance. Two types of discrete methods were considered; the
continuous harmonic signal method and the u-point identification method.
34
-------
4.2.1.1 Continuous Harmonic Signal Method
Like pulse signals, continuous signals can also be used to derive travel time between
source and receiver (Rio, 2006), and hence, velocity. This is a manual process, meaning
that to get results at different frequencies the signal has to be stepped manually. For each
frequency, a continuous harmonic signal is actuated and data are collected via the signal
analyzer. The travel distance remains fixed. At each frequency step, the phase difference
between consecutive peaks and troughs of the source and receiver is calculated (Rio,
2006). The phase differences are then plotted against their respective frequencies. The
slope of the plot is used to calculate the travel time.
4.2.1.2 7i-Point Method
The u-point method is the reverse of the continuous harmonic signal method
(Greening and Nash, 2004). Here the frequency of the sinusoid is varied until the
received signals meet preselected phase differences (i.e. n and -n radians; Rio, 2006). As
shown by Da Fonseca et al. (2008), in an ideal material (homogenous, isotropic) this
process produces a linear relationship between phase angle and frequency, from which
the slope is determined to calculate the travel time.
4.2.2 Frequency Sweep Method
The frequency sweep initially sweeps over a broad range in which the coherence
between the source and receiver signal is used to determine an intermediate range over
35
-------
which output signals produce optimal transfer of input energy. The signal analyzer
correlates the source signal and received signal, and produces a coherence plot that
ranges from 0 to 1 over the span of the frequency sweep. If the coherence is close to 1,
more energy in the output signal is caused by the input signal and the two are well
correlated (Da Fonseca et al., 2008). High coherence between the signals is necessary to
obtain low variation in the results (Da Fonseca et al., 2008). Once the intermediate range
is identified, the frequency sweep is concentrated on that range, from which plots of
unwrapped phase angle with respect to frequency are used to determine the travel time.
The frequency sweep method has the same outcome as the discrete methods, it is more
efficient, and it also allows the calculation of the coherence function, which improves its
reliability over the discrete method (Rio, 2006).
4.3 Test Methods: Summary
The first arrival, characteristic point, and cross correlation methods in the time
domain are chosen for analysis, because of the applicability to this study and documented
success. It was decided not to use either of the discrete methods because they were time
consuming and required heavy signal processing. The frequency sweep method can be
carried out quickly and with less signal processing effort (Greening and Nash, 2004).
36
-------
CHAPTER 5
CALIBRATION IN AIR AND WATER
The system was first calibrated in air and water. This chapter presents the testing
system set up, the testing methodology, data processing and the results.
5.1 Testing System Setup
As presented in section 2.4, three parameters were considered for test column design:
D:L ratio; L:X ratio; and minimizing L to minimize volume and signal attenuation. The L
value chosen depends on the reference shear wave velocities, the D value was fixed and
the X depends on the actuation frequency, which was selected experimentally for the tests
in air and water. The test system was optimized for shear wave transmission as opposed
to compression wave transmission. This is because a test system optimized for S-wave
transmission can still receive useful P-waves, which is not the case if the test system is
optimized for P-waves. P-waves are better received with the bender elements close to
each other, which would lead to significant near-field effects on the S-waves, and render
them useless.
The bender element in the base plate was always used as the source, and the bender
element in the top cap was always used as the receiver.
Lee and Santamarina (2005) reported that when bender elements with short cantilever
lengths are used, the resonance frequency in air is higher than the resonance frequency in
saturated granular media. In the current experiment, the resonance frequency was
37
-------
identified by a sudden increase in sound from the actuating bender element during a
frequency sweep and by observing the amplitude spectrum of the received signal. In air,
the source bender element started to resonate at around 6 kHz and plateau around 8 kHz,
reaching the peak displacement amplitude around 12 kHz (Fig. 5.1). In water, the
resonance frequency of the source bender element peaked at approximately 8 kHz (Fig.
5.2). For this study in which the primary goal is to address testing with glass beads under
essentially saturated conditions and ambient pressure, the frequency for testing was
chosen to be 8 kHz. This frequency was used to calculate the value of L for the placement
of the top cap, a value which was maintained for all testing in this study.
As discussed in Sec. 3.2.1, the ratio L:X is selected to minimize near-field effects. To
do so, a reference shear wave velocity was required. Patel et al. (2009) reported shear
wave velocity (Vs) for water-saturated glass beads under ideal stacking conditions,
achieved by placing the specimen on a vibration table, and at ambient pressure to be 150
m/s. Using this reference Vs, the actuation frequency (Fr) of 8 kHz and the following
formula:
the reference X was found to be 18.8 mm. This number presents the following possible
values for L:
1. L>2X^L> 37.6mm
2. 4.2 > D:L > 2.2 -» 69.3 mm > L > 36.3 mm
These values address the criteria put forth in section 2.4. The value of L selected for
this research represented a compromise between the conflicting design criteria discussed
in sections 2.4 and 3.2.1. The distance from the top of the base plate to the bottom of the
38
-------
top cap (H) was selected to be 70 mm, which yields a value for (L) of 62 mm, given 4
mm protrusion of each bender element. This (L) value gives a D:L ratio of 2.4. Note that
measurements of (H) for top cap placement were made from outside the column using a
digital caliper which is accurate to 0.1 mm, but the readings were rounded off to the
nearest integer millimeter.
Three acrylic spacers were used to hold up the top cap at the required height for the
air and water tests (Fig. 5.3). With the spacers in place, the average of five measurements
of (H) was 71 mm, which differs from the target value by 1.5%.
5.2 Testing Methodology
The first-arrival and cross-correlation methods were used for calibrating the system.
As described previously, individual sine pulses at 8 kHz at 10 ms intervals were used for
actuating the source bender element. Each test was repeated three times, where the
average of 1000 pulses was considered as one repetition.
The travel path lengths were determined as shown in Fig. 5.4. From them, the
theoretical travel times were calculated by using expected values for P-wave velocity in
air and water. As mentioned earlier, P-waves were assumed to be generated in the form
of side lobes, therefore the reflected travel path was assumed (Lee and Santamarina,
2005).
Sengpiel (2010) reports compression wave speed in air at 70°F to be 343 m/s.
Santamarina et al. (2001) report compression wave speed in water at 70°F to be 1480 m/s.
39
-------
By using these data and assumptions the anticipated travel times were calculated. Except
where noted, testing was done without filtering.
5.3 Results for Testing in Air
The time history and Fourier amplitude spectrum (to 500 Hz) of the received signal in
air are presented in Fig. 5.5. The pulse signals are riding upon a low frequency
background signal. The Fourier amplitude spectrum shows that the dominant frequency is
32 Hz with resonances at 58 Hz, 82 Hz, and 105 Hz. A high-pass filter with a cut off
frequency of 1 kHz was applied to remove the disturbance; a lower cut off frequency
might have been adequate but was not tested. The filtered time history is presented in Fig.
5.6; the low frequency background signal is removed. Figure 5.6 shows two consecutive
pulses, which demonstrate that the energy from one pulse does not completely decay
prior to the arrival of the next pulse. This affects the ability to make an accurate first
arrival pick because it increases the background noise threshold. It is recommended that
the interval between pulses be increased for any future testing in air to reduce these
effects on the arrivals. Figure 5.7 is an amplification of data shown in Figure 5.6, which
shows a first arrival and the source pulse, offset to facilitate comparison. The first data
point on the received waveform demonstrating amplitude clearly greater than the
background noise threshold occurs at approximately 0.51 ms. Two excursions of the
waveform exceeded the background noise threshold prior to this time, but they were
considered too close to the threshold to be counted as a first arrival. The first arrival time
picked is 4% later than the anticipated arrival time (Table 5.1), assuming a reflected
40
-------
travel path (Fig. 5.4) as expected for side-lobe energy actuation of P-waves from the
faces of the bender elements as demonstrated by Lee and Santamarina (2005). Note that
both the residual noise from the previous pulse which partially obscures the actual arrival
and also the algorithm for picking arrivals cause the process to err on the slow side,
leading to the chosen arrival time being later than the actual arrival. Note that a change in
frequency and shape of the received signal within the background noise prior to the
arrival pick might be interpreted as faint evidence of the direct-transmission arrival.
The source signal and the received signal were cross-correlated using the signal
analyzer. The results for testing in air are presented in Fig. 5.8. When no filtering was
used, the cross-correlation result had a dominant "V" shape. A 1 kHz high-pass filter was
applied, and the "V" was removed. The filtered and the non-filtered results gave the same
peak time, which was 30% slower than the anticipated reflected wave travel time (Table
5.1). This peak time is of course slower still than the anticipated direct wave travel time.
Lee and Santamarina (2005) assert that the cross-correlation technique must either relate
signals of the same nature or accommodate for the testing system's transfer functions.
The received signal was a heavily modified version of the source sine pulse (e.g., Fig.
5.5). Unless the transfer function can be accounted for, or the received signal is filtered
to mask effects of multiple reflections and other scattering that dominate the wave train at
later times, the results of this cross-correlation analysis are not meaningful.
The tests in air determined the arrival time for compression wave velocity within
approximately 4% of the anticipated time using the first arrival method with a 1 kHz
high-pass filter and assuming a reflected travel path. The accuracy of results can be
improved by reducing background noise.
41
-------
5.4 Results for Testing in Water
Testing was repeated in room-temperature distilled water. The results for the first
arrival method are presented in Fig. 5.9. Unlike for testing in air, there was no low-
frequency carrier energy present. However, a first arrival pick could not be made because
it appeared to coincide with electrical crosstalk. We know this was crosstalk because the
initiation time was exactly the same and the duration is approximately the same as the
source signal, 0.14 ms. This crosstalk effect was not observable in air because it was
obscured by the background noise, which was significantly higher in air than in water
(e.g., compare Figs. 5.6 and 5.9). The arrivals in air occur after the crosstalk effects die
out.
To differentiate between the P-wave arrival and the electrical crosstalk, a set of trials
was carried out where the travel distance through the water was varied, while all other
features of the testing were kept constant. This process allowed differentiating between
where the electrical crosstalk ends and the received signals begin. As the travel distance
increases the arrival time for the P-waves should increase while the electrical crosstalk
would remain constant. Four trials were carried out, where the first was at the initial
spacing used (L = 62.5 mm), and the spacing of each successive trial was increased by
12.7mm.
The results are presented in Fig. 5.10 and Table 5.2. The majority of the arrivals
occurred at different times, and in the correct order with the shortest travel path arriving
first. The first three spacings' arrivals were successively one time sample apart, and the
third and fourth arrivals had the same arrival time. Due to the cross talk it was not
42
-------
possible to pick the P-wave arrival with certainty. The anticipated direct and reflected
wave arrival times for all the spacings used were within the duration of the crosstalk or
just beyond it. The actual arrivals of both the direct and reflected P-waves could not be
timed using the first arrival method because of the residual crosstalk and also because of
the sampling rate. It should be noted that the received signal waveforms at all four
spacings tested were similar in amplitude and frequency content between 0.3 and 0.5 ms,
indicating a resonance condition in the test chamber that is independent of the parameter
(H).
The results for cross correlation testing in water are presented in Fig. 5.11. Like in air,
there was a low-frequency disturbance present, which was removed with a 1 kHz high-
pass filter. As with tests in air, the cross-correlation results gave a peak time that was
significantly later than the anticipated direct or reflected wave travel times (Table 5.3).
Again, post processing of the received signal would be required in order for the cross-
correlation computation to be meaningful.
43
-------
Table 5.1. Anticipated and measured P-wave travel times in air
Wave
travel path
Direct
Reflected
Anticipated
travel time (s)*
1 Q9T7 n/l
l.a&IL U4
4.91E-04
Method
First arrival
Cross
correlation
First arrival
Cross
correlation
Experimental
travel time (s)
5.10E-04
7.03E-04
5.10E-04
7.03E-04
Difference
62%
73%
4%
30%
"The anticipated travel time accounts for the 10 us delay with the testing system.
Table 5.2. Test to differentiate electrical crosstalk from P-wave arrivals in water
Wave travel path
Direct
Reflected
Path length
(mm)
62.5
75.2
87.9
100.6
164.7
169.9
175.9
182.6
Anticipated
travel time (s) *
5.22E-05
6.08E-05
6.94E-05
7.80E-05
1.21E-04
1.25E-04
1.29E-04
1.33E-04
Experimental
travel time (s)
8.79E-05
9.77E-05
1.07E-04
1.07E-04
8.79E-05
9.77E-05
1.07E-04
1.07E-04
Difference
41%
38%
35%
27%
-38%
-28%
-20%
-24%
*The anticipated travel time accounts for the 10 us delay with the testing system.
44
-------
Table 5.3. Cross correlation method: anticipated and measured P-wave arrivals in water
Travel path
Direct
Reflected
Anticipated
travel time (s)*
5.20E-05
1.20E-04
Experimental
travel time (s)
A 9nin n/i
Q.LVEi U4
Difference
708%
250%
Experimental
Vp (m/s)
392
"The anticipated travel time accounts for the 10 us delay with the testing system.
45
-------
0
-20
-40
-60
-80 -
-100-
I -120
-140 -
-160-
-180 -
-200
0.5
1 1.5 2
Frequency (Hz)
2.5
x 10
Figure 5.1. Frequency response for a O-to-30 kHz sweep showing the resonance
frequency of the test system in air; average of 1000 recordings
-20
-40
-60
s -80
-------
Figure 5.3. Acrylic spacers used for testing in air and water to hold top cap at the required
height
a: Acrylic spacers;
b: Side view of column with spacers in place
47
-------
01+02 = 164.6 mm
152*4 mm
Figure 5.4. Direct (red) and reflected (green) travel paths
48
-------
3.6
3.4
3.2
3
2.8
x10
1.5
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s)
x10"4
0.5
X: 32.23
Y: 0.0001086
50
100 150
200 250 300 350
Frequency (Hz)
400 450 500
Figure 5.5. 8 kHz pulse signal in air with no filter applied, showing single repetition of
1000 recordings averaged per repetition.
a: Time domain result showing low frequency carrier harmonic
b: Frequency spectrum showing dominant carrier frequency at 32 Hz and its harmonics
1.5
0.5
-0.5
-1.5
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016
Time (s)
Figure 5.6. 8 kHz pulse signal in air with 1 kHz high-pass filter, showing single repetition
of 1000 recordings averaged per repetition.
49
-------
Source signal
Anticipated direct
arrival at 0.19 ms
Anticipated reflected
arrival at 0.49 ms
First arrival pick at
0.51ms
0.5V
1
-0.5
0.5 1
Time (s)
1.5
x 10'
2.5
-3
Figure 5.7. First arrival test in air with 1 kHz high pass filtering, shows three repetitions
of an 8 kHz sine pulse with 1000 recordings averaged per repetition, arrival pick, source
signal (offset for display purposes), anticipated arrival times, and background noise
threshold.
50
-------
-0.06
-0.07 f-
Anticipated direct
arrival time at 0.19 ms
Anticipated reflected
arrival time at 0.49 ms
Time (s)
x 10"
10
12
14
Time (s) x io"4
Figure 5.8. Cross correlation response in air showing peak times for three repetitions in
air, 1000 recordings averaged per repetition.
a: No filtering applied
b: 1 kHz high-pass filter applied
51
-------
-0.2 -
-0.4
-0.6 -
-0.8 ---
I
±s
Q.
Anticipated reflected
arrival time at 0.12 ms
Anticipated direct
arrival time at 0.05 ms
Anticipated reflected
arrival time at 0.12 ms
X:-2.93e-005 I
Y: 0.001749 '
X 0.0001563
Y: 0.001483
Anticipated direcf
arrival time at 0.05 ms
0.05
-0.05
-0.1
-0.15
-0.2
x 10
Figure 5.9. First arrival test in water with no filtering of an 8 kHz sine pulse, shows
source (blue), receiver (red) with three repetitions of 1000 recordings averaged per
repetition.
a: Expanded view
b: Detailed view
52
-------
0.5
a)
1 0
-0.5
L = 62.5 mm
L = 75.2 mm
L = 87.9 mm
L = 100.6 mm
0.5
1.5
2.5
Time (s)
x 10
0.8
0.6 -
0.4
0.2
-0.2
-0.4
-0.6
-0.8
Anticipated direc^
arrival time at 0.05 ms
for L = 62.5 mm
L = 62.5 mm
L = 75.2 mm
L = 87.9 mm
L= 100.6 mm
Anticipated reflected
arrival time at 0.12 ms
for L = 62.5 mm
x 10
Figure 5.10. Differentiating electrical crosstalk from P-wave arrivals in water by varying
the tip-to-tip distance, 1000 recordings averaged per received signal
a: Expanded view of entire received signal
b: Detail view showing crosstalk, anticipated arrival times for base case
(L = 62.5 mm) and possible arrival picks
53
-------
Anticipated direct
arrival at 0.05 ms
Anticipated reflected
arrival at 0.12 ms
Peak time 0.42 ms
Anticipated direct
arrival at 0.05 ms
Anticipated reflected
arrival at 0.12 ms
-0.5
-0.5
x 10"'
Figure 5.11. Cross correlation response in water showing peak times of three repetitions
of 1000 recordings averaged per repetition.
a: No filtering applied
b: 1 kHz high-pass filter applied
54
-------
CHAPTER 6
GLASS BEAD SPECIMEN PREPARATION
Prior to calibrating the testing system with water-saturated glass beads, a consistent
method for preparing the glass bead specimens had to be established. Glass beads were
used in testing because glass is inert and therefore minimizes variability caused by
chemical interaction with the nanoparticles and the granular matrix. This chapter presents
two methods tested for achieving repeatable specimens. Repeatability was judged by
comparing the saturated unit weight of specimens prepared using the same techniques.
6.1 Methods of Specimen Preparation
The two methods used to prepare the specimens are called the dumping method and
the stage fill method.
The glass beads used were 0.5 mm in diameter and purchased from Quackenbush Co.,
Inc. Distilled water was used for backfilling the pore spaces. For both methods, the glass
beads and water to backfill were dispensed into the testing system to the desired heights.
Once the glass beads were dispensed into the column and the top cap was placed and
leveled, six measurements of specimen height H (Fig. 5.4) made with a digital caliper
were averaged to establish the height of the test specimen. The samples were not fully
saturated because there was no back-pressure applied and the water used was not de-
aired. The water was plumed into the sample from the bottom until it was approximately
25 mm above the upper surface of the top cap and this is accounted for when calculating
55
-------
unit weight; recall that the top cap had a port in it to allow water to flow out of the test
cell (Fig. 2.4).
The mass of the glass beads and water dispensed into the column were recorded for
each specimen; the average mass of three specimens was used as the reference mass. The
moisture content, void ratio and saturated unit weight of the specimens prepared were
calculated using the reference masses, the manufacturer-provided specific gravity of the
glass beads (Gs =2.5) and the standard unit weight of water (yw = 9.8 kN/m3); see
appendix 1, pages 131 and 132 for details. The saturated unit weights ranged from 18 to
19 kN/m3.
The specimens were tapped with a rod and the column walls were tapped on the side
during sample preparation to reduce voids. Testing took place at atmospheric pressure
and room temperature. No other external stresses were applied to the system.
6.2 The Dumping Method
Dry glass beads were poured into the dry column until they were near the required
height. A flat disc was used to level the top surface and check if the required height (70
mm; Sec. 5.1) was achieved. This process was repeated until the required height was
reached and then water was introduced slowly from the bottom. Three specimens were
prepared; see appendix page 132 for details. The average saturated unit weight achieved
was 18.9 kN/m3with deviation from the average ranging from 0.1% to 0.5%.
56
-------
6.3 The Stage Fill Method
The stage fill method was described by Rajabdeen et al. (2011). The first step was to
introduce water into the column to a depth of approximately 35 mm, then pour the glass
beads into the column in approximately 10 stages. After each stage, more water was
plumed into the column so that the water level was kept above the surface of the glass
beads and the specimen was rodded and the column walls were tapped to reduce voids.
When the glass beads reached the required height, the top cap was placed. Three
specimens were prepared; see appendix page 132 for details. The average saturated unit
weight achieved was 18.7 kN/m3with deviation from the average ranging from 0.0% to
0.2%.
6.4 Chosen Method
The two methods showed little variability in saturated unit weight, but higher
variability was recorded with the dumping method. Therefore the stage-fill method was
used to prepare the specimens for testing. A more robust method of sample preparation
would involve the use of a vibration table for the initial seating and preparation of the
glass beads as seen in work done by Patel et al. (2008).
57
-------
CHAPTER 7
BASELINE TESTING GLASS BEADS IN WATER
This chapter addresses testing the system with water-saturated glass beads, to
establish baseline seismic responses.
7.1 Test Setup and Preparation
Testing was conducted and baseline seismic responses were established in saturated
glass beads by replicating the method presented by Rajabdeen et al. (2011). The
specimen was prepared using the stage fill method. Once constructed, the water in the
specimen was allowed to drain by gravity and was then refilled from the bottom. The
purpose of this cycle is to soak and seat the glass beads. A first set of trials which
consisted of three test methods was conducted (described below); each test method
consisted of three repetitions, where 1000 recordings were averaged per repetition. The
specimen was then drained again, re-wetted, and a duplicate set of trials was carried out
to investigate repeatability. Three identical specimens were prepared and tested in this
fashion.
The mass of water drained and added was measured at each stage. Under soaked
conditions the specimen had an average saturated unit weight of 18.5 kN/m3 and held an
average of 30% water by weight; see appendix 1, page 133 for details. Variation between
the saturated unit weights of consecutive tests was negligible.
58
-------
7.2 Testing Methodology
The first step with the testing was to find the resonance frequency of the bender
elements in the saturated glass beads. The resonance frequency was found by running a
frequency sweep over a broad range from 0 to 30 kHz, and viewing the amplitude
spectrum (Fig. 7.1). The peak amplitude occurred at approximately 8 kHz. The actuated
signals were tailored to highlight either P-waves or S-waves. Operating at higher
frequencies aids in analyzing P-waves (Deniz, 2008), and operating at lower frequencies
aids in analyzing S-waves. For testing glass bead specimens with this system, the high
and low frequency ranges were determined experimentally as presented in the results
sections to follow. The source signal was amplified by a linear signal amplifier from 10 V
to 30 V, to increase the signal-to-noise ratio.
7.3 Data Processing
The testing system was optimized to mitigate external noise. Preliminary tests were
carried out without filtering (Fig. 7.2). The results show a low frequency background
signal upon which the high frequency pulse signal is riding. An FFT of the received
signal showed the background frequency was 38 Hz, which is close to the background
frequency recorded when testing in air with no filtering (32 Hz). To remove this effect a
high pass filter with a cut off frequency at 200 Hz was applied (Fig. 7.3); this value was
found by trial and error. This high pass filter was applied for the rest of the tests
59
-------
presented in this report. All the test result plots not presented in the main body of this
report are in appendix 2.
7.4 Pulse Signals to Highlight P-Waves
Pulse testing was started at 8 kHz to identify a baseline seismic response for P-wave
propagation; 8 kHz was chosen to approximately match the resonant frequency of the
embedded bender elements. To maximize the time between the 8 kHz sine pulses and still
capture the entire received signal, an interval of 15 ms was used. The interval was
increased by 5 ms from the tests carried out in air and water, which showed residual
effects of the first pulse obstructing the second pulse arrival. If the interval between
pulses was increased any further, the entire received signal of the second pulse could not
be recorded.
The results presented in Fig. 7.3 show overlaid plots of two trials on a specimen
prepared and tested as described in section 7.1. The pulse interval was such that energy
from each pulse had decayed to where it appeared to have minimal effect on the arrival of
the subsequent pulse. The impact of the residual energy from one pulse on the onset of
the next is demonstrated in appendix 3, by comparing the quiet time in-between pulses to
background signals in the absence of any pulse.
Figure 7.4 is a representative result of time-domain testing targeting P-waves, which
shows low-amplitude sinusoidal electrical crosstalk coinciding with the actuation of the
source pulse and preceding two possible P-wave arrival picks. It should be noted that
there is irregularity present at the initiation of the source sine pulse; the clarity of the
60
-------
source signal can be improved by using a quicker sampling rate. Recall that the crosstalk
was also observable in tests in water and showed energy decaying after approximately
0.14 ms (Fig. 5.9). The same is evident in Fig. 7.4, where the amplitude decreases at 0.14
ms until the signal leads into an increase in amplitude at 0.2 ms, which corresponds to a
velocity of approximately 840 m/s, assuming a reflected travel path (length 16.5 cm; Fig.
5.4). Another significant increase in amplitude is seen corresponding to approximately
560 m/s (assuming a reflected travel path). This amplitude is clearly greater than the
arrival corresponding to 840 m/s, and is judged as a separate second arrival.
The observed responses relate to the fast and slow P-waves first described by Biot
(1956). Fast P-waves represent energy travelling through the pore fluid and slow P-waves
represent energy travelling through the skeletal structure of the saturated granular media
(Nakagawa et al., 1997). Slow P-waves have been difficult to detect in geomaterials, but
have been well documented in artificial porous media such as glass beads (Nakagawa et
al., 1997). Slow P-wave transmission through saturated glass beads has also been well
documented by Fiona (1980) and Fiona et al. (1990), among others.
A fast P-wave travels in water at approximately 1480 m/s (e.g. Santamarina et al.,
2001). Such an arrival in the current study, be it by direct or reflected path, would be
masked by the crosstalk (Fig. 7.4). We conclude that the apparent arrival at the time
corresponding to a velocity of 840 m/s is an aftereffect of the fast P-wave following a
reflected path. The slow P-wave arrivals are not obscured by the crosstalk.
The velocities corresponding to the slow P-wave arrivals average 560 m/s (Table 7.1;
Fig. 7.5). This value agrees reasonably well with the test results of Nakagawa et al.
(1996) who measured slow P-wave velocity in a saturated sand sample at 200 - 500 m/s.
61
-------
Consistency of testing configuration (e.g., placing glass beads, installing top cap) can be
evaluated by considering differences in computed slow P-wave velocity between
specimen preparations (Table 7.1). Deviations of slow P-wave velocities of the three
specimens from the mean were no larger than 0.2 percent. The largest deviation
represented just two time samples in the time history.
Referring to Fig. 7.4, test results from the first and second trials appear similar.
Amplitudes are generally larger for trial 1. Phase shifts are noticeable but small and they
do not become significant until well after the first arrivals. The time histories of the two
trials are consistent from 0 to 0.7 ms. From 0.7 to 0.9 ms the shapes differ, but
differences disappear for the largest energy excursion which peaks for both trials at about
1 ms. This large amplitude peak observed at 1 ms indicates the presence of a standing
wave, and resonance effects. The standing wave appears at a different time from the tests
in water, which occurred between 0.3 and 0.5 ms (section 5.4). This is because the
different testing media in the two cases have different effects on the resonating wave.
To quantify the difference in amplitude between the signals of trials 1 and 2, the
amplitude of the peak immediately following the slow P-wave arrival for each signal was
noted (Fig. 7.6). The differences between the amplitudes of the peak points between trials
(PI and P2) are presented in Table 7.2 and Fig. 7.7. Amplitudes from trial 1 (PI) are 26%
larger on average than from trial 2. Further, amplitudes of trial 1, ranging from 0.17 to
0.44 V, are more variable than trial 2 (P2), which range from 0.19 to 0.25 V. The range of
amplitudes from trial 2 will be used as the baseline against which variances in the
presence of nanoparticles will be compared. Specifically, peak slow P-wave amplitudes
62
-------
outside the range 0.19 to 0.25 V will be taken as indication that the nanoparticles are
affecting the measurement.
7.5 Pulse Signals to Highlight S-Waves
An optimum frequency for actuating pulse signals to highlight S-wave energy was
found through experimentation. The results are presented in Fig. 7.8, which shows source
actuation frequency increased from 1 kHz to 4 kHz in 1 kHz increments. A starting
frequency of 1 kHz was used because this is the lowest possible actuation frequency
when using a burst function with the function generator. The S-wave train became
progressively more contaminated with high frequency energy with increasing actuation
energy. The source frequency chosen for testing was 1 kHz.
Figure 7.9 shows a received signal for 1 kHz pulses actuated at 15 ms intervals.
Decaying energy from the preceding pulse appears to still be present as the new pulse is
received. The effects of the residual energy of the preceding pulse on the background
noise are presented in appendix 3.
Figure 7.10 shows the anticipated shear wave arrival, the presence of near-field
effects on received signals, and the first arrival picks that were made. The onset of the
direct-transmission S-wave is obvious from its shape, although its arrival is preceded by
low amplitude near-field effects. In this report, all the first arrival picks of the shear wave
(direct travel path) were made at the first data point with positive amplitude in the onset
shear wave energy. An arrival corresponding to the anticipated shear wave velocity (150
m/s) occurs at the beginning of what we interpret to be near-field effects. The reference
63
-------
velocity was taken from the work of Patel et al. (2009) which does not acknowledge near-
field effects. According to Arroyo et al. (2003), if near-field effects are not accounted for,
typically shear wave velocity is overestimated. In this study, the direct travel path length
(tip-to-tip distance (L)) was used to calculate S-wave velocities (Table 7.1 and Fig. 7.11).
By accounting for near-field effects and given the picking algorithm, the shear wave
velocities determined were between 25 and 35 m/s, which were much slower than
anticipated values.
For each specimen, the S-wave velocities among repetitions differ by an average of
1.5%. Velocities are 5 to 13% higher for the second trial than the first trial. This
consistent difference implies that the process of repetitive wetting and draining of the
glass beads continually improves their seating. Such an effect would logically be visible
with the S-waves and slow P-waves, which are both dependent upon the skeletal
structure, but not with fast P-waves which depend only on the pore fluid. The fact that we
observed this effect with S-waves but not slow P-waves needs further examination. The
difference in velocities measured between the two trials implies that a change in S-wave
velocity caused by the introduction of an experimental treatment would have to be larger
than 13% or smaller than 5% to be detected with the system as it was configured in this
initial test. Increasing the number of wetting and draining cycles prior to taking any
measurements might decrease this sensitivity threshold.
Considering the measured shear wave velocity and the source actuation frequency of
1 kHz, the S-wavelengths ranged between 24 and 30 mm. The L values used for the
testing were 61.2 mm on average and always greater than 60 mm; see appendix 1 page
64
-------
135 for details. Thus the test design satisfied the criterion presented in Section 2.4 that L
must be greater than two wavelengths.
To quantify the difference in amplitude between the received signals of trials 1 and 2,
the peak-to-peak amplitudes of the S-wave pulse (AP) (Fig. 7.12) were observed (Table
7.2 and Fig. 7.13). Contrary to observations of the slow P waves, differences in amplitude
between the two trials were not consistent. However, consistent with observations of the
slow P waves, amplitudes of trial 1 (API), ranging from 1.20 to 2.41 V, are more variable
than trial 2 (AP2), which range from 2.07 to 2.36 V. The range of amplitudes from trial 2
will be used as the baseline against which variances in the presence of nanoparticles will
be compared. Specifically, peak-to-peak S-wave signal amplitude outside the range 2.07
to 2.36 V will be taken as indication that the nanoparticles are affecting the measurement.
7.6 Frequency Sweep Method
Frequency sweeps were run from 0 to 30 kHz. Figure 7.14 shows an example of the
coherence plot, where for the most part, the coherence remains above 0.9 from 7 to 25
kHz for both trials.
7.6.1 Amplitude Spectrum Results
The Fourier amplitude spectra from first and second trials, computed using the
dynamic signal analyzer, were compared by overlaying the two plots. The representative
results are presented in Fig. 7.15. For all specimens, the spectra peak at 8 kHz (the
resonant frequency of the potted bender elements) after which amplitude decreases
65
-------
gradually. Two sharp dips in amplitude appear at approximately 11 and 13 kHz. Over the
range where coherence was high, differences in spectral amplitude between the first and
second trials were small. To quantify the sensitivity between trials, the difference
between the amplitudes of the trials was found; these are referred to as the residual
signals (Fig. 7.16). The residual signals of the three specimens were averaged to get a
baseline residual signal between 7 and 25 kHz, to represent the difference between the
two trials using only distilled water. For the presence of nanoparticles to be detected with
this method, they would have to cause perturbations large enough to deviate significantly
from this baseline signal.
7.6.2 Phase Angle Results
The phase component of the frequency domain data was observed by Da Fonseca et
al. (2008) with bender-element testing on granitic residual soil and Toyoura sand using a
triaxial testing apparatus. The authors reviewed common methods used for testing with
and interpreting bender element data, and proposed an outline for testing to obtain
reliable travel times. The method considers the slope of a best-fit straight line of the
unwrapped phase angle against frequency over a selected frequency range demonstrating
high coherence. This method of analysis is applied only to determine the travel time and
velocity of the signal and not other aspects of the signature, although visual or
computational comparisons of phase might be useful to document responses to
experimental treatments.
For this study, the phase angles from the frequency sweeps were unwrapped using a
function available on the signal analyzer. A representative result is presented in Fig. 7.17.
66
-------
The plots do not give a single, simple, linear slope as was obtained by Da Fonseca et al.
(2008). The four measurement results are offset with respect to one another but more or
less parallel in the range of high coherence. Dips and rises in the phase angles disrupt the
linearity, and different gradients of the slopes are observed. By visual inspection, the
different gradients represented credible shear wave velocities, ranging from 70 to 200
m/s. However, the different gradients and velocities lend uncertainty. These differences
from the more straightforward results reported by Da Fonseca et al. (2008) might be
attributed to the fact that their testing was conducted on homogenous specimens in a
triaxial apparatus under elevated effective stresses. This topic was not pursued any
further in this study, but merits further investigation.
7.7 Summary: Detection Criteria
From the baseline tests carried out in this chapter, the following criteria to evaluate
the detectability of nanoparticle dispersions are proposed. These criteria are tested in
Chapter 8.
1. Water trial results: If results (applied to four tests: velocities and amplitudes from
P- and S-waves) from two or more of three repetitions from the water trial are
outside the range obtained for trial 1 (Chapter 7) by 5% or more, the water trial
fails the test. Otherwise, the water trial is accepted.
67
-------
2. Nano trial results:
Slow P-waves:
If the velocity deviates by more than 1% from the mean of the trial 2
baseline velocities (Chapter 7), the nanoparticle dispersion is
detectible.
If significant phase differences between the received signals of the
consecutive water and nano trials of a specimen exist, the nanoparticle
dispersion is considered detectible. The degree of significance
assessed is strictly qualitative, from visual inspection. Further testing
would be required before quantitative criteria can be established.
If the zero-to-peak amplitude of the peak directly following the first
arrival deviates by more than 5% from the range 0.190 to 0.248 V, the
nanoparticle dispersion is detectible.
S-waves:
If the velocity is less than 5.3% quicker than the water trial velocity or
more than 13.7% quicker than the water trial velocity, the nanoparticle
dispersion is detectible. These numbers represent the extreme values of
measured difference between trials 1 and 2 (Chapter 7), incremented
by 5%.
If the peak-to-peak amplitude deviates by more than 5% from the
range 2.07 to 2.36 V, the nanoparticle dispersion is detectible. These
numbers represent the extreme values of trial 2 amplitudes (Chapter
7).
68
-------
Spectral response: If significant amplitude differences between the residual
signal (from consecutive water and nano trials of a specimen, in the high
coherence range of 7 to 25 kHz) and the baseline residual (Chapter 7) exist,
the nanoparticle dispersion is considered detectible. As with phase difference
evaluations, amplitude differences are strictly qualitative, from visual
inspection, and further testing would be required before quantitative criteria
can be established.
69
-------
Table 7.1. Velocities associated with received pulse signals for water-saturated
glass bead specimens
Specimen
1
2
3
Repetition
1
2
3
1
2
3
1
2
3
Compression (slow P-wave) :
8 kHz pulse
Trial 1
(m/s)
562.4
560.3
561.0
Trial 2
(m/s)*
562.4
560.3
561.0
Difference
0%
0%
0%
Shear (S-wave):
1 kHz pulse
Trial 1
(m/s)
28.2
28.4
28.4
23.9
24.1
24.1
29.3
29.6
29.5
Trial 2
(m/s)*
32.2
32.4
32.2
25.4
25.3
25.3
33.8
33.8
33.6
Difference**
12%
12%
12%
6%
5%
5%
13%
12%
12%
*Trial 2 represents duplicate tests following drainage and rewetting of test specimen.
**Difference is Trial 2 relative to Trial 1
Table 7.2. Amplitudes associated with received signals for water-saturated glass bead
specimens
Specimen
1
2
3
Repetition
1
2
3
1
2
3
1
2
3
Compression (slow P-wave):
8 kHz pulse
Trial 1
(P1,V)
0.174
0.230
0.267
0.379
0.327
0.350
0.233
0.416
0.441
Trial 2
(P2,V)*
0.194
0.190
0.185
0.219
0.201
0.228
0.248
0.221
0.242
Difference
12%
-18%
-31%
-42%
-39%
-35%
6%
-47%
-45%
Shear (S-wave):
1 kHz pulse
Trial 1
(AP1.V)
2.41
2.39
2.40
1.94
2.06
2.09
1.20
1.75
1.86
Trial 2
(AP2,V)*
2.07
2.23
2.19
2.19
2.18
2.17
2.34
2.34
2.36
Difference**
-14%
-7%
-9%
13%
6%
4%
95%
34%
27%
*Trial 2 represents duplicate tests following drainage and rewetting of test specimen.
**Difference is Trial 2 relative to Trial 1
70
-------
^
-50
-60
-70
-80
-100
-110
-120
-130
C
AA
I
I
I
E
Y
TJ
"~
X: 8200
Y: -49.45
rj\
[V
I
I
A, A
IWT
'Vl
I
\ n\ r
\
r\r
I
\
\ A
M?
v-\y^
V
AH
i \
0.5 1 1.5 2 2.5 C
Frequency (Hz) x 1Q4
Figure 7.1. Frequency response for a O-to-30 kHz sweep showing the resonance
frequency of the bender element in a water-saturated glass bead specimen; result of 1000
recordings averaged, with no filtering applied.
71
-------
s U-J
1
o. -U-3
< 1
_2
1 4
1.2
Amplitude (V)
o o o o
:> KJ ^ c> bo ->
)
x 10
-
-
)
I I I I I I I
I I I
I I I
>^"" ~"""~<
-^ ^^V, (\n
«f ' ' ^""^ ' ^^
/ ' ' '
/- T T r
ff~*^ i i
*WlWjWWWi/'^ ' '
vl I I
I
I I I I I
0.002 0.004 0.006 0.008 0.01 0.012 0.014 C
Time (s)
-4
I I I I I
**
V V/Vx^,~^-t_-_-__j iii
100 200 300 400 500 600 700
I |
1 ' a
1 -- A
*%::i
V" ^^v^1-'
i i
i i
i i
i i
i i
.016 0.018 0.
b
I I
800 900 10
Frequency (Hz)
Figure 7.2. 8 kHz pulse signal (1000 recordings averaged) in a saturated glass bead
specimen with no filter applied.
a: Time domain result showing low frequency carrier signal
b: Fourier amplitude spectrum to 1 kHz showing dominant carrier frequency at 38 Hz
72
-------
0.5
QJ
"On
< -0.5
" (
1
0.5
S
1 o
"On
< -0.5
.1
I I '
I
u
I
v
~^?*» "
3 0.002 0.004 0.006 0.008
,,]
. ' I
I I
-«V-1i|
l'
n |
'
i-fl
If
_
[rial 1
rrial 2
A
-ft jfr%^5>«
f
I
0.01 0.012 0.014 0.016
Time (s)
ill
rt/lt
Vf
v y
u
\ A\
y__ A
\7
vX"
0.018
^
a
0.02
b
,^
0.014 0.0145 0.015 0.0155 0.016 0.0165 0.017 0.0175 0.018 0.0185 0.019
Time (s)
Figure 7.3. Representative result for consecutive 8 kHz pulses with 200 Hz high-pass
filter applied, received signals of first and second trials are shown; result of 1000
recordings averaged for each.
a: Received signals showing two consecutive pulses and quiet time between pulses;
b: Detail view of a received pulse
73
-------
0.5V
8 kHz Source
Trial 1
Trial 2
Anticipated slow P-wave arrival range
Time (s)
x 10
Figure 7.4. Representative result of an 8 kHz sine pulse and received signals of 1000
recordings averaged for each, emphasizing reflected-path P-wave propagation. Trials 1
and 2 are conducted sequentially under near-identical test conditions. Note the
irregularity present at the initiation of the source sine pulse.
74
-------
Figure 7.5. Summary of 8 kHz sine pulse highlighting P-wave velocities in water-
saturated glass bead specimens.
75
-------
0.5V
A A
P2
- - -
V
10
15
Time (s)
x 10
Figure 7.6. Representative picks of characteristic points used to compare the amplitudes
of received slow P-wave signals, 1000 averages and 200 Hz high-pass filter.
76
-------
Baseline Amplitude Range
I PI (0.17-0.44V)
P2 (0.19-0.25V)
Figure 7.7. Summary of received signal amplitudes of characteristic points (described in
text) from 8 kHz pulse signals highlighting P-waves in water-soaked glass bead
specimens.
PI refers to the amplitude values of P-waves from trial 1, P2 refers to the amplitude
values of P-waves from trial 2, and the black dashed lines show the P2 amplitude range
that defines the baseline for nano testing (chapter 8).
77
-------
1.5
o
0
1 0
Pulse at 1 kHz
0.01 0.011 0.012 0.013 0.014 0.015
Time (s)
1.5
§ 0.5
0
1 0
±±
Q_
| -0.5
-1.5
Pulse at 2 kHz
0.01 0.011 0.012 0.013 0.014 0.015
Time (s)
1.5
1
S 0.5
1-0.5
-1.5
Pulse at 3 kHz
\J
v/
0.01 0.011 0.012 0.013 0.014 0.015
Time (s)
1.5
1
2 0.5
0
1 0
±±
Q_
| -0.5
-1
-1.5
Pulse at 4 kHz
- V-V-
/
0.01 0.011 0.012 0.013 0.014 0.015
Time (s)
Figure 7.8. Check for optimal sine pulse frequency to test for shear in saturated glass
beads; received signals are 1 repetition of 1000 recordings averaged per repetition, under
200 Hz high-pass filter.
a: Sine pulse at 1 kHz showing S-wave;
b: Sine pulse at 2 kHz showing S-wave, but not clearly;
c: Sine pulse at 3 kHz showing weak S-wave;
d: Sine pulse at 4 kHz showing dominant high frequency energy.
78
-------
0.5
>
-0.5
-1.5
0.002
0.004
0.006
0.008
0.01
Time (s)
0.012
0.014
0.016
0.018
0.02
Figure 7.9. Representative result of a 1 kHz sine pulse test in saturated glass beads
showing two consecutive pulses, demonstrating that disturbances due to the first pulse do
not completely decay prior to the arrival of the second pulse. A single repetition from
each trial is shown; 1000 recordings averaged per repetition.
79
-------
0.5V
1 1
\l
Anticipated S-wave
arrival (150 m/s)
0 0.5
1 1 1 1 1
i
i
i
i
First arrival at 32 m/s
(trial 2)
i
1
\
i
Near-field e
1
^0}
ffect
1.5
TV
/ y
/ /
p i
i
\
\^
\
\
1
\
\
\
First arrival at 28 m/s
(trial 1)
2 2.5
Time (s)
\
\ \
\
A
/ \
\
i
/
/ \ /
T^
/
/
/
/
1
\J
* \
\
Hz Source
I 1
12
3 3.5 4 4.5 £
x 10"3
Figure 7.10. Representative result of 1 kHz sine pulse (shear); shows near-field effects,
trial 2 first arrival earlier than trial 1 arrival; single repetition of 1000 recordings averaged
per repetition, 200 Hz high-pass filter.
80
-------
50.0
45.0
40.0
Specimen
I Trial 1
I Trial 2
Figure 7.11. Summary of 1 kHz sine pulse highlighting S-wave (shear) received signal
velocities in water-saturated glass bead specimens.
81
-------
0.5V
0.5
1.5
2 2.5
Time (s)
3.5
4.5
x10"
Figure 7.12. Representative picks of characteristic points used to compare the amplitudes
of received S-wave signals, 1000 averages and 200 Hz high-pass filter.
82
-------
3.00
2.50
2.00
>_
01
1.50
Q.
E
1.00
0.50
-9:06-
Repetition
I API (Water) | AP2 (Water)
Baseline Amplitude Range:
I API (1.20-2.41 V)
AP2 (2.07 - 2.36 V)
I I I I I I I I I
I I I I I I I I I
Specimen
Figure 7.13. Summary of received-signal peak-to-peak amplitude differences from 1 kHz
pulse signals highlighting S-waves in water-saturated glass bead specimens.
API refers to the peak-to-peak amplitude of S-waves from trial 1 and AP2 refers to the
peak-to-peak amplitude of S-waves from trial 2. The black dashed lines show the AP2
range, which defines the baseline for nano testing (chapter 8).
83
-------
1.5
Frequency (Hz)
x 10
Figure 7.14. Representative coherence for 30 kHz sweep with 200 Hz high-pass filter in a
water-saturated glass bead specimen, showing result of 1000 recordings averaged each,
for Trials 1 and 2.
84
-------
50V
x 10
0.8
1.2
1.4 1.6 1.8
Frequency (Hz)
2.2
2.4
x 10
Figure 7.15. Representative amplitude spectrum of 30 kHz sweep for all water-saturated
glass bead specimens; 1000 recordings averaged per repetition.
a: Shows entire frequency spectrum for the 30 kHz sweep, the box indicates the enlarged
area for (b);
b: Shows the frequency range (7 to 25 kHz) analyzed to establish a repeatable signature
between Trial 1 and Trial 2.
85
-------
20V
I
.
<
0.8
1.2
1.4 1.6 1.
Frequency (Hz)
Specimen 1
Specimen 2
2.4
X 10
Figure 7.16. Residual signals equal to the difference between the spectral responses of
trial 1 and trial 2 in water-saturated glass bead specimens, used to quantify the sensitivity
of the test system. The average residual signal is the averaged result of the three residual
signals of the specimens, and it is used as the baseline.
86
-------
2000
%
rt
JS
OH
-2000
-4000
-6000
-8000
-10000
-12000
0.5
1 1.5 2
Frequency (Hz)
2.5
3.5
x 10
Figure 7.17. Representative result of unwrapped phase angles for trials 1 and 2 in water-
saturated glass bead specimen, three repetitions of 1000 recordings averaged per
repetition, high-pass filter at 200 Hz applied. Range of high coherence is expected from 7
to 25 kHz.
87
-------
CHAPTER 8
TESTING WITH NANOPARTICLE DISPERSIONS
The nanoparticle study differed from the testing with glass beads in water described
in Chapter 7 only in that nanoparticle dispersions in water were used in place of pure
water for the second trial. Those tests with water presented in chapter 7 are referred to
here as the baseline tests. The goal of the nanoparticle tests presented in this chapter was
to check for responses that were outside of baseline; these might be attributable to the
nanoparticles.
Three nanoparticles were tested: 1. zinc oxide (nZnO), 2. titanium dioxide (nTiOz),
and 3. silver (nAg). A fourth nanoparticle was also chosen for testing; zero-valent iron
(nZVI). The nZVI considered was at 98% purity and in powdered form. After sonicating
and preparing a dispersion, it was found that the nZVI settled quickly, in no more than 3
minutes. This implies that a nZVI plume in saturated granular media would not remain
dispersed, it would rapidly settle. A dispersion with nZVI could not be created for testing
purposes and testing with nZVI was forfeited.
8.1 Test Setup and Preparation
The nanoparticles were purchased from Nanostructured & Amorphous Materials, Inc.
(www.nanamor.com). In this study, all concentrations are reported by weight. The oxides
were received pre-dispersed in distilled water at concentrations of 20% for nZnO and
40% for nTi02; they had to be diluted to the required concentrations. The metals were in
-------
powdered form, from which dispersions were created at the required concentrations. The
concentrations of the nanoparticle dispersions plumed into the column had to be higher
than the required concentration to account for water remaining in the column after
draining out the water trial (trial 1). The glass bead specimens retained an average of 240
ml of water after draining, and the average volume of additional fluid required to backfill
the specimens for the nano trial (trial 2) was 360 ml. (See appendix 1, page 133 for
details.)
8.1.1 Pluming Process
Different methods were used for introducing different nanoparticle dispersions (Fig.
2.7). The nano-oxide dispersions were transferred to a funnel flask, which was placed at a
higher elevation than the column so that the dispersion was plumed into the glass-bead-
filled column by gravity. The nano-metal dispersions could not be plumed by this process
because they clogged the valves in the column plumbing. An alternative method to keep
the dispersion homogenized and stable for pluming presented by Joyce (2011) was
adopted. The authors used a stirrer and peristaltic pump for pluming nano metal
dispersions. We use a sonicator instead of a stirrer to keep the nanoparticles dispersed.
The nano-metal dispersions were sonicated during the entire pluming process and no
clogging of valves occurred.
89
-------
8.2 Testing Methodology
The testing methods used were the same as those used for the baseline tests: 8 kHz
pulse signals to highlight compression energy in the time domain; 1 kHz pulse signals to
highlight shear energy in the time domain; and 30 kHz frequency sweep to observe
spectral response. As discussed below, all materials were tested at a concentration in the
range 3 to 5 %; in addition, the nZnO was tested at two lower concentrations.
Concentrations, fluid volumes and other details of all test specimens are provided in
appendix 1, page 134.
Following processes presented in Ch. 7, results of time-domain testing for all
nanoparticle dispersions are summarized in Tables 8.1 and 8.2. The 8 kHz pulse testing
results for velocity and amplitude are provided in Figures 8.1 and 8.2 respectively. The 1
kHz pulse testing results for velocity and amplitude are provided in Figures 8.3 and 8.4
respectively. Criteria presented in section 7.7 are applied to the results in order to
evaluate the detectability of nanoparticle dispersions.
8.3 Validating Water Trials
Recall that the acceptability criterion for the water trials is presented in section 7.7.
Compare water trial results against trial 1 results (Chapter 7) for velocities using Tables
8.1 and 7.1 and for amplitudes using Tables 8.2 and 7.2. Only one test failed the
acceptance criterion: S-wave velocity for nTiOz, which was 8% below the smallest value
measured in trial 1 for all three repetitions.
90
-------
8.4 Testing with nZnO
Testing with nZnO was conducted at three concentrations. Each set of tests is
described below, in order from low to high concentration.
8.4.1 0.03% Concentration
Recall that the nZnO was pre-dispersed with water at 20% concentration. 2 ml of the
dispersion was diluted with 498 ml water and placed on a stirrer for approximately 30
minutes to aid dispersion. This process yielded a 500 ml dispersion, of which 400 ml was
expected to be required to fill the drained specimen. The following equation presents the
dilution of the dispersed nZnO introduced into the column:
/ 0.4 ml nZnO
5OOmlH20
It was expected that 240 ml of water would remain in the column after draining the
specimen following the water trial. This retained water would dilute the concentration of
the nZnO further. The total concentration of nZnO in the column:
/ 0.08%*400ml(H20 + nZnO)
/ v ^ >
\
\ _ r\ r\ r o /
/ ~ '
V400 ml (H20 + nZnO) + 240 ml(H20)
After draining the water trial, approximately 347 ml was retained in the column. As a
result, only 185 ml of the nZnO dispersion was introduced, so that the concentration of
nZnO tested was:
' 0.08 % * 185 ml (H20 + nZnO) \ _
,185 ml (H20 + nZnO) + 347 ml (H20)J ~ 0-03°/0
91
-------
Between trials, the volumetric moisture content of the glass bead specimen decreased
by 5% and saturated unit weight decreased by 0.55 kN/m3.
For 8 kHz pulse testing to highlight P-waves, the received signals from the water and
the nano trials had minimal divergence in phase, and had identical arrival times of slow
compression waves and therefore did not satisfy the detection criteria (Figure 8.5). The
nanoparticle dispersion was however considered detectible by P-wave amplitude.
For 1 kHz pulse testing to highlight S-waves, the received signals from the water and
the nano trials had nearly identical arrival times of shear waves (Figure 8.6), which is
significantly slower than baseline and therefore satisfy the detection criterion. The
nanoparticle dispersion was not considered detectible by S-wave amplitude, although one
of the three repetitions was significantly lower than the baseline range.
For 30 kHz sweep testing to highlight spectral responses in the high coherence
frequency range, the residual signal from the nanoparticle dispersion deviated
significantly from the baseline in the ranges 8 -10, 12 - 15, and 21- 25 kHz (Figures 8.7
and 8.8). The largest deviation in voltage from zero occurred in the same frequency range
for both nano test and baseline, 12-14 kHz.
Overall, results from multiple tests showed some deviations from baseline with ZnO
at 0.03% concentration. To check if patterns develop, testing was repeated with nZnO
concentration increased by a factor of 10.
92
-------
8.4.2 0.3% Concentration
23 ml of the nZnO dispersion was diluted with 577 ml water to yield a 600 ml
dispersion of nZnO. The following equation presents the dilution of the dispersed nZnO
introduced into the column:
4.6 ml nZnO
After draining the water used for saturating the specimen to test in clean water, the water
retained in the column was approximately 277 ml. The volume of nZnO dispersion
utilized for saturating the specimen for the nano trial was approximately 213 ml. The
total concentration of nZnO in the column:
/0.8% * 213ml
7) = 0.3%
V213ml + 277mly
Between trials, the volumetric moisture content of the glass bead specimen decreased by
7% and the saturated unit weight decreased by 0.4 kN/m3.
For 8 kHz pulse testing to highlight P-waves, the received signals from the water and
the nano trials had minimal divergence in phase, and had identical arrival times of slow
compression waves and therefore did not satisfy the detection criteria (Figure 8.9). The
nanoparticle dispersion was also not considered detectible by P-wave amplitude, although
one of the three repetitions was significantly lower than the baseline range.
For 1 kHz pulse testing used to highlight S-waves, the received signals from the water
and the nano trials had nearly identical arrival times for the shear waves (Figure 8.10),
and therefore satisfy the detection criterion. The nanoparticle dispersion was also
considered detectible by S-wave amplitude: amplitudes for all three repetitions satisfy the
detection criterion.
93
-------
For frequency response testing, the residual signal from the nanoparticle dispersion
had only slight deviations from the baseline; no significant change was identified
(Figures 8.11 and 8.12).
Detectability findings for nZnO at the mid-range concentration were consistent with
the lower concentration only for P- and S-wave velocities. Next, testing was repeated
with nZnO concentration increased again by approximately a factor of 10.
8.4.3 2.7% Concentration
150 ml of nZnO dispersion was diluted with 150 ml water to prepare a 300 ml
dispersion. After draining the water used for saturating the specimen to test in clean
water, approximately 246 ml of water remained in the column. The entire volume of 300
ml nZnO dispersion at 5.6% concentration was utilized for saturating the specimen for
the nano trial, and an additional 75 ml of clean water was required to fill the pore spaces.
The total concentration of nZnO in the column:
/ 5.6 % * 300 ml \
( q) = 2J°/0
Between trials, the volumetric moisture content of the glass bead specimen did not
change and the saturated unit weight increased by 0.06 kN/m3.
For 8 kHz pulse testing to highlight P-waves, the received signals from the water and
the nano trials had minimal divergence in phase, and had identical arrival times of slow
compression waves (Figure 8.13), and therefore did not satisfy detection criteria. The
nanoparticle dispersion was however considered detectible by P-wave amplitude, with
amplitudes for all three repetitions above baseline range.
94
-------
For 1 kHz pulse testing to highlight S-waves, the received signals from the water and
the nano trials had nearly identical arrival times and satisfy the detection criterion (Figure
8.14). The nanoparticle dispersion was not however considered detectible by S-wave
amplitude: none of the repetitions satisfy the detection criterion.
For frequency response testing, the residual signal from the nanoparticle dispersion
deviated significantly from the baseline between approximately 7 to 12 kHz (Figures 8.15
and 8.16). The residual signal remained below the baseline until 18 kHz.
8.5 Summary: nZnO Testing
Considering the three concentrations tested with nZnO, the slow P-wave arrival times
consistently lacked variation from the baseline results. Results for the slow P-wave
amplitudes were inconsistent: amplitudes increased above baseline in the presence of
nZnO at the low and high concentrations but not for the mid-range concentration. And
the increase in amplitude for nZnO at high concentration was smaller than that from the
nanoparticle dispersion at low concentration.
The S-wave arrival times increased by 5% with respect to the baseline results in the
presence of nZnO at all concentrations, demonstrating a lack of dependence on
concentration. The S-wave amplitudes showed a decrease in amplitude with respect to
baseline for the mid-range concentration but no significant deviation from baseline for
the low and high concentrations.
Spectral response analysis of the residual signal from nZnO at the mid-range
concentration showed less variation from the baseline when compared with the
95
-------
nanoparticle dispersion at low and high concentrations. The perturbations observed with
the nanoparticle dispersion at high concentration were larger in magnitude when
compared to the deviations observed at low concentration.
In the interest of efficiency, the rest of the nanoparticle dispersion tests were
conducted only in the high concentration range.
8.6 Testing with nTiOz at 4.9% Concentration
70 ml of the nTiOz dispersion at 40% concentration was diluted with 330 ml water
and placed on a stirrer for approximately 30 minutes to aid dispersion; a 400 ml
dispersion was prepared. After draining the water used for saturating the specimen to test
in clean water, the water retained in the column was approximately 156 ml. The volume
of nTiOz utilized for the nano trial was approximately 370 ml. The total concentration of
nTiOz in the column:
/ 7% * 370 ml \
= 49%
V370ml + 156ml/
Between trials the volumetric moisture content of the glass bead specimen decreased
by 1% and the saturated unit weight increased by 0.05 kN/m3.
For 8 kHz pulse testing to highlight P-waves, the received signals from the water and
the nano trials had minimal divergence in phase, and had identical arrival times of slow
compression waves, and therefore did not satisfy detection criteria (Figure 8.17). The
nanoparticle dispersion was also not considered detectible by P-wave amplitude; although
one of the three repetitions was significantly lower than the baseline range.
96
-------
For 1 kHz pulse testing to highlight S-waves, the water trial results failed to satisfy
the baseline acceptance criteria. The arrival times of the received signals from the nano
trials (Figure 8.18) did not satisfy the detection criterion. The nanoparticle dispersion was
also not considered detectible by S-wave amplitude: two of the three repetitions did not
satisfy the detection criterion.
For frequency response testing, the only significant deviation of the residual signal of
the nanoparticle dispersion from the baseline is around the resonance frequency (8 kHz)
(Figures 8.19 and 8.20).
Overall, the results indicate that nTi02 at the concentration tested is not detectible
with seismic methods, except possibly by spectral response.
8.7 Testing with nAg at 3.7% Concentration
The nAg was at 99% purity and in powdered form. The dispersion volume required
was 400 ml. The concentration of nAg after dispersing 25.0 g in 400 ml water:
25000 mg nAg mg
o o ,~ _. o - on/
(400mlH20) =62-5^-=6-3%
The 25 grams of nAg was placed in a flask, and 400 ml of distilled water was added
to it. The flask was then placed in a sonicator for approximately 300 minutes; this
duration was chosen experimentally to ensure thorough dispersion.
After draining the water used for saturating the specimen to test in clean water, the
water retained in the column was approximately 246 ml. The volume of nAg dispersion
required for filling the pore spaces of the specimen was approximately 347 ml. The total
concentration of nAg in the column:
97
-------
6.3 % * 347 ml
' = 3.7%
V347ml + 246 ml>
Between trials the volumetric moisture content of the glass bead specimen did not
change and the saturated unit weight decreased by 0.06 kN/m3.
For 8 kHz pulse testing to highlight P-waves, the received signals from the water and
the nano trials had minimal divergence in phase, and had identical arrival times of slow
compression waves (Figure 8.21), and therefore did not satisfy the detection criteria. The
nanoparticle dispersion was however considered detectible by P-wave amplitude, with
amplitudes for all three repetitions above baseline range.
For 1 kHz pulse testing to highlight S-waves, the arrival times of the received signals
from the nano trials (Figure 8.22) did not satisfy the detection criterion. The nanoparticle
dispersion was not considered detectible by S-wave amplitude: all three repetitions did
not satisfy the detection criterion.
For frequency response testing, the residual signal from the nanoparticle dispersion
deviated significantly from the baseline at the resonance frequency (8 kHz) and between
12 - 13 kHz (Figures 8.23 and 8.24).
In summary, the presence of nAg at 3.7% concentration was detectable only by P-
wave amplitude and possibly spectral response.
8.8 Overall Analysis: Nanoparticle Detectability by Seismic Methods
Three nanoparticle dispersions were tested: nZnO, nTi02, and nAg. Only nZnO
dispersions were tested at multiple concentrations levels which are referred to here as low
(0.01 to 0.05%), medium (0.1 to 0.5%) and high (1 to 5%). The nAg and nTi02
98
-------
dispersions were tested only at high concentration. A summary of the time domain tests
carried out, and the outcomes stating which nanoparticle dispersions are detectible, and
by what method, are presented in Table 8.3.
An overall comparison of the 8 kHz sine pulse received signals for all the
nanoparticle dispersions tested and one of the three baseline specimens tested is presented
in Figure 8.25. This testing addressed slow P-wave arrival times and zero-to-peak
amplitude differences. The arrival times showed no change from baseline for any
nanoparticle dispersion. The amplitudes, however, showed differences from baseline. The
nZnO was considered detectable at low and high concentration, but not at medium
concentration. The nAg was considered detectable, but the nTiOz was not.
An overall comparison of the 1 kHz sine pulse received signals for all the
nanoparticle dispersions tested and one of the three baseline specimens tested is presented
in Figure 8.26. This testing addressed S-wave arrival times and peak-to-peak amplitude
differences. The presence of nZnO at all concentration levels was detectable by S-wave
arrival times, which were outside the bounds (trailing) established for the baseline by
approximately 5%. The presence of nAg was not detectable by S-wave arrival time, and
the same was true for nTi02, however, these results are uncertain due to the water trial of
that specimen not satisfying the baseline acceptance criteria. The presence of nZnO was
detectable by S-wave amplitudes at medium concentration, but not at low or high
concentrations. This conflicting outcome is not understood and merits further study. The
presence of nTi02 was not detectable by S-wave amplitude and neither was nAg.
An overall comparison of the residual spectral responses in the presence of all the
nanoparticle dispersions tested and one of the three baseline specimens tested is presented
99
-------
in Figure 8.27. Testing for spectral response with a 30 kHz sweep addressed Fourier
amplitudes. The baseline was established by computing the residual (difference between
consecutive tests on the same specimen, separated only by draining and refilling the pore
fluid) upon water-saturated specimens (Fig. 7.16). In the presence of nZnO, spectral
responses fluctuated with respect to concentration levels, with the largest deviation from
baseline at high concentration and the smallest at mid-range. In the presence of nTiOz,
the spectral response was not distinguishable from baseline, except at around 8 kHz,
which is resonance. In the presence of nAg, some local amplitude spikes surpassed
baseline. Further tests for spectral response are needed to obtain quantifiable criteria,
from which detectability can be established.
100
-------
Table 8.1. Velocities associated with received pulse signals from water and nanoparticle
dispersions in glass bead specimens
Specimen
nZnO
(0.03%)
nZnO
(0.3%)
nZnO
(2.7%)
nTi02
(4.9%)
nAg
(3.7%)
Repetition
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
Compression (slowP-wave):
8 kHz pulse
Water
Trial
(m/s)
561.7
562.4
561
560.6
561
Nano
Trial
(m/s)
561.7
562.4
561
560.6
561
Difference
0.0%
0.0%
0.0%
0.0%
0.0%
Shear (S-wave):
1 kHz pulse
Water
Trial
(m/s)
27.1
27.3
27.2
26.4
26.7
26.9
31.0
31.0
30.8
22.0
22.0
22.0
26.3
26.5
26.2
Nano
Trial
(m/s)
27.2
27.3
27.3
26.3
26.7
26.7
31.0
31.1
31.0
24.2
24.2
24.3
28.4
28.6
28.6
Difference
0.4%
0.0%
0.4%
0.4%
0.0%
0.9%
0.0%
0.5%
0.5%
10.0%
9.7%
10.1%
8.2%
7.8%
9.1%
Red bold text indicates water trial outside range from baseline trial 1,
Black bold text in yellow box indicates detectable nano trial where the response
was lower than baseline range.
101
-------
Table 8.2. Amplitudes associated with received pulse signals from tests in water and
nanoparticle dispersions in glass bead specimens
Specimen
nZnO
(0.03%)
nZnO
(0.3%)
nZnO
(2.7%)
nTiO
(4.9%)
nAg
(3.7%)
Repetition
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
Compression (slowP-wave):
8 kHz pulse
Water
Trial
(P1,V)
0.35
0.28
0.28
0.00
0.23
0.38
0.24
0.34
0.34
0.14
0.18
0.19
0.27
0.41
0.41
Nano
Trial
(P2,V)
0.68
0.41
0.42
0.00
0.23
0.25
0.32
0.35
0.35
0.16
0.22
0.23
0.32
0.42
0.42
Difference
95%
47%
49%
-75%
-1%
-33%
35%
3%
3%
16%
23%
22%
18%
3%
4%
Shear (S-wave):
1 kHz pulse
Water
Trial
(AP1.V)
0.33
1.94
2.00
0.75
1.77
1.85
1.82
2.20
2.17
1.87
2.27
2.24
1.62
2.52
2.55
Nano
Trial
(AP2, V)
0.77
2.04
2.11
0.73
1.73
1.56
2.21
2.25
2.25
1.72
2.07
2.07
2.13
2.21
2.22
Difference
133%
5%
5%
-3%
-2%
-16%
21%
2%
4%
-8%
-9%
-8%
31%
-12%
-13%
Red bold text indicates water trial outside range from baseline trial 1,
Black bold text in green box indicates detectable nano trial where the response
was higher than baseline range,
Black bold text in yellow box indicates detectable nano trial where the response
was lower than baseline range.
102
-------
Table 8.3. Summary of detectability of nanoparticle dispersions in glass bead specimens
using time domain methods
Specimen
nZnO
(0.03%)
nZnO
(0.3%)
nZnO
(2.7%)
nTi02
(4.9%)
nAg
(3.7%)
P-wave
Velocity
No
No
No
No
No
P-wave
Amplitude
Yes
No
Yes
No
Yes
S-wave
Velocity
Yes
Yes
Yes
N/A
No
S-wave
Amplitude
No
Yes
No
No
No
Green box indicates detectible nanoparticle dispersions
Orange box indicates unusable nanoparticle dispersion
103
-------
Water Trial Nanoparticle Trial
cnn
590
£ 580
|570
% 560
01
> 550
01
|j 540
| 530
> 520
510
Bee-
Repetition
]
L :
> -
i ]
L :
> -
i ]
L 2
:
! ]
L ;
) :
! ]
L :
> -
\
Nanoparticle nZnO (0.03%) nZnO(0.3%) nZnO (2.7%) nTiO (4.9%) | nAg (3.7%)
Figure 8.1. Summary of 8 kHz results highlighting slow P-wave velocity for all
nanoparticle dispersions.
104
-------
PI (Water) | P2 (Nano)
Amplitude ranges for baseline tests
I PI (0.17-0.44V)
P2 (0.19 - 0.25 V)
Nanoparticle nZnO (0.03%) nZnO (0.3%) nZnO |(2-7%) nTiO (4.9%) nAg (3.7%)
Figure 8.2. Summary of P-wave characteristic point amplitudes from 8 kHz pulse tests in
water and in the presence of nanoparticle dispersions in saturated glass bead specimens.
The signs indicate an increase (+), decrease (-) or no change (0) in amplitude for the P-
waves in the presence of nanoparticles from the baseline trial 2 result. The black dashed
lines show the baseline against which to compare nano test amplitudes (from Sec. 7.4).
105
-------
DC
3D
< 25 -
|25
.+ -)0 .
u t-u
O
01
> 1R -
0. lb
5
s m -
to
c .
n
Repetition
Namparticle
Water trial
Nanoparticle trial
123
nZnO (0.03%)
123
nZnO (0.3%)
123
nZnO (2.7%)
123
nTiO (4.9%)
123
nAg (3.7%)
Figure 8.3. Summary of 1 kHz results highlighting S-wave velocity for all nanoparticle
dispersions.
106
-------
API (Water) AP2 (Nano)
3.00
2.50
2.00
I. 1.50
<
01
1.00
0.50
Amplitude ranges for baseline tests:
I API (1.20-2.41 V)
AP2 (2.07 - 2.36 V)
MINI
niii i
Repetition
Nanoparticle nZnO (0.03%) nZnO (0.3%) nZnO (2-7°/°) nTiO (4.9%) nAg (3.7%)
Figure 8.4. Summary of S-wave amplitudes from 1 kHz pulse tests in water and in the
presence of nanoparticle dispersions in saturated glass bead specimens. The black dashed
lines show the AP2 baseline amplitude range (from Sec. 7.5).
107
-------
1.5
0.5
>
-0.5
-1.5
0.2
0.4 0.6
0.8 1 1.2
Time (s)
1.4
1.6
1.8
x ID"'
a
"H.
-0.5
-1 -
x 10
Figure 8.5. 8 kHz sine pulse with 200 Hz high-pass filter applied in saturated glass bead
specimen for testing differences in compression with water and nZnO at 0.03%
dispersion, showing three repetitions of 1000 recordings averaged per repetition.
a: Expanded view of entire received signal; note amplitude variation between the three
repetitions of each trial
b: Detail view showing representative picks for characteristic amplitude points, residual
fast P-wave and slow P-wave arrivals
108
-------
1.5
0.5
8 0
|-0.5
-1.5
4
Time (S)
x 10
Figure 8.6. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nZnO at 0.03% dispersion,
showing three repetitions of 1000 recordings averaged per repetition, and consistent first
arrival pick.
109
-------
-40
-50
-60
-70
8
0)
| -80
"5.
E
-90
-100
-110
-120
0.8
1.2
1.4
1.6 1,
Frequency (Hz)
2.2
2.4
x 10
Figure 8.7. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nZnO at 0.03% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
0.8
1.2
1.4 1.6
Frequency (Hz)
1.8
2.2
2.4
x 10
Figure 8.8. Residual signals from the differences of spectral response in water and in
nZnO at 0.03%, of three repetitions of 1000 recordings averaged per repetition, compared
to the average baseline residual signal.
110
-------
0.5
Q.
<
-0.5
-1.5
0.2 0.4 0.6 0.8 1 1.2
Time (s)
1.4
1.6
1.8
x 10"
x 10
Figure 8.9. 8 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing differences in compression with water and nZnO at 0.3% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
a: Expanded view of entire received signal; note amplitude variation between the three
repetitions of each trial
b: Detail view showing representative picks for characteristic amplitude points, residual
fast P-waves and slow P-wave arrivals
111
-------
1.5
0.5
3 o
-0.5
-1.5
3
Time (s)
x 10
Figure 8.10. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nZnO at 0.3% dispersion,
showing three repetitions of 1000 recordings averaged per repetition, and consistent first
arrival pick.
112
-------
E
<
-40
-50
-60
-70
-80
-90
-100
-110
-120
0.8
1.2
1.4
1.6
Frequency (Hz)
1.8
2.2
2.4
x 10
Figure 8.11. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nZnO at 0.3% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
0.8
1.2
1.4 1.6
Frequency (Hz)
1.8
2.2
2.4
x 10
Figure 8.12. Residual signals from the differences of spectral response in water and in
nZnO at 0.3%, of three repetitions of 1000 recordings averaged per repetition, compared
to the average baseline residual signal.
113
-------
0.5
a.
E
-0.5
0.2
0.4
0.6
0.8
1 1.2
Time (s)
1.4
1.6
1.8
x 10
Time (s)
x 10
Figure 8.13. 8 kHz sine pulse with 200 Hz high-pass filter applied in saturated glass bead
specimen for testing differences in compression with water and nZnO at 2.7% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
a: Expanded view of entire received signal; note amplitude variation between the three
repetitions of each trial
b: Detail view showing representative picks for characteristic amplitude points, residual
fast P-wave and slow P-wave arrivals
114
-------
0.5
>,
0)
?
.
<
-0.5
-1.5
Time (s)
x10
Figure 8.14. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nZnO at 2.7% dispersion,
showing three repetitions of 1000 recordings averaged per repetition, and consistent first
arrival pick.
115
-------
-40
-50
-60
-70
-80
-90
-100
-110
-120
0.8
1.2
1.4
1.6
Frequency (Hz)
1.8
2.2
2.4
x 10
Figure 8.15. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nZnO at 2.7% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
30
20
10 -
-10
-20
-30
0.6
0.8
1.2
1.4 1.6 1.!
Frequency (Hz)
2.2
2.4
x 10
Figure 8.16. Residual signals from the differences of spectral response in water and in
nZnO at 2.7%, of three repetitions of 1000 recordings averaged per repetition, compared
to the average baseline residual signal.
116
-------
0.5
1 °
Q.
E
-0.5
-1.5
0.2 0.4
0.8 1 1.2
Time (s)
E 0
-0.5
3
Time (s)
Figure 8.17. 8 kHz sine pulse with 200 Hz high-pass filter applied in saturated glass bead
specimen for testing differences in compression with water and nTiOz at 4.9% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
a: Expanded view of entire received signal; note amplitude variation between the three
repetitions of each trial
b: Detail view showing representative picks for characteristic amplitude points, residual
fast P-wave, and slow P-wave arrivals
117
-------
0.5
Q.
E
-0.5
-1.5
Time (s)
x 10
Figure 8.18. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nTi02 at 4.9% dispersion,
showing three repetitions of 1000 recordings averaged per repetition, and first arrival
pick for nTiOz coming in earlier than water.
118
-------
40
20
E
<
-20
-40
-60
-80
-100
-120
0.8
1.2 1.4 1.6 1.i
Frequency (Hz)
2.2
2.4
10
Figure 8.19. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nTiOz at 4.9% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
1.4 1.6
Frequency (Hz)
2.2
2.4
x 10
Figure 8.20. Residual signals from the differences of spectral response in water and in
nTi02 at 4.9%, of three repetitions of 1000 recordings averaged per repetition, compared
to the average baseline residual signal.
119
-------
0.5
-0.5
0.2
0.4
0.6
0.8
1
Time (s)
1.2
1.6
x 10
1.2
1
0.8
0.6
0.2
0
-0.2
-0.4
-0.6
-0.8
0.5
1.5 2 2.5 3
Time (seconds)
3.5
4.5
x 10
Figure 8.21. 8 kHz sine pulse with 200 Hz high-pass filter applied in saturated glass bead
specimen for testing differences in compression with water and nAg at 3.7% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
a: Expanded view of entire received signal; note amplitude variation between the three
repetitions of each trial
b: Detail view showing representative picks characteristic amplitude points, residual fast
P-wave, and slow P-wave arrivals
120
-------
0.5
0
Q.
< -0.5
-1.5
Time (s)
x 10
Figure 8.22. 1 kHz sine pulse with 200 Hz high-pass filter in saturated glass bead
specimen for testing shear in the presence of water and nAg at 3.7% dispersion, showing
three repetitions of 1000 recordings averaged per repetition, and first arrival pick for nAg
coming in earlier than for water.
121
-------
40
20
-20
s
0
1 -40
-60
-80
-100
-120
0.8
1.2 1.4 1.6 1,
Frequency (Hz)
2.2
2.4
x 10
Figure 8.23. 30 kHz sweep with 200 Hz high-pass filter, testing spectral response in a
saturated glass bead specimen in water and in the presence of nAg at 3.7% dispersion,
showing three repetitions of 1000 recordings averaged per repetition.
0.8
1.2
1.4 1.6 1,
Frequency (Hz)
2.2
2.4
x 10
Figure 8.24. Residual signals from the differences of spectral response in water and in
nAg at 3.7%, of three repetitions of 1000 recordings averaged per repetition, compared to
the average baseline residual signal.
122
-------
0.5V
0.2 0.4 0.6 0.8 1.0, 1.2 14 1.6 1
.0
.0
Figure 8.25. Signals from slow compression wave comparing response in the presence of
nanoparticles to the baseline received signal.
123
-------
0.5V
0 270 3.0 CO 570 6.0 7.0
Figure 8.26. Signals from shear waves comparing response in the presence of
nanoparticles to the baseline received signal.
124
-------
10 V
Fr(kH
Fr(kll
^
Iz) 8
Wat
Nan
^7T
V"l
Iz) i
£
Iz) 8
.,
1
^
r
3 i
er:B
o: B
v^
N
i
*~~y£l*<
1
vp
1
, 1
F
1
J
2 1
rW
f
2 1
A ^
1 i
> i
4 1
4 1
eso^
%
I 1
5 1
^
5 1
fr~*f*
i 1
Baseline Average
/\
8 2
0 2
2 2
n
c
/"W
rl
8 2
Ti02 at 4.9%
oncentration
^
0 2
2 2
f 1 ^ i
_A
3 2
4
4
nAg at 3.7%
D 2
^
> 2
\
1
Fr(kllz) 8
Fr(kllz) 8
ID
II)
14 1
H 1
13 20 21 21
nZnO at 0.03%
nZnOatO.3%
1J 20 2:
I 21
tiZnOa
t 1.TA
Figure 8.27. Signals from 30 kHz sweeps comparing response in the presence of
nanoparticles to the baseline.
125
-------
CHAPTER 9
CONCLUSIONS AND RECOMMENDATIONS
This chapter presents the conclusions and recommendations of the research, and also
presents new research questions that arose due to this study.
9.1 Conclusions
A laboratory test system using bender elements was constructed for the study of
seismic body wave propagation to address the response of nanoparticles dispersed in
saturated granular media. The system was calibrated in air, water and water-saturated
glass beads. Waveforms considered for testing were square waves and sine waves. Based
on the literature, sine waves were judged to be better suited, mainly because near-field
effects are less pronounced. Sine waves at different frequencies, optimized for
transmission of shear and compression, were used in all testing described in this report.
Testing baselines for water-saturated glass bead specimens were established by
analyzing the responses between consecutive tests on the same specimen, separated only
by draining and refilling the pore fluid (trial 1 and trial 2). Baseline responses were
established with respect to compression and shear waves in the time domain, and spectral
response. Fast P-waves were not detectible with the experimental apparatus, but slow P-
waves were. Tests for slow P-waves showed negligible difference between trials in
arrival times, while characteristic-point amplitudes for trial 1 were larger than trial 2 by
about 26%. Tests for S-waves showed arrival times for trial 1 were consistently 5-13%
126
-------
slower than in trial 2. Differences are attributed to changes in consolidation state of the
test specimen caused by draining and refilling pore spaces. Only S-wave amplitudes from
trial 2 produced repeatable results, so these were used as the baseline. Spectral response
testing showed good repeatability in the range of high coherence, 7 to 25 kHz.
An acceptability criterion was proposed to compare water trials of the nanoparticle
dispersion tests against baseline; the S-wave velocity test for nTiOz failed to meet the
criterion. Criteria were also proposed to evaluate the detectability of nanoparticle
dispersions. Testing with nanoparticle dispersions showed that the system was capable of
registering subtle changes in response caused by pore fluid content. Only the nZnO was
tested at different concentrations, and detectabilities fluctuated between concentration
levels. From the quantitative criteria, testing in the presence of nZnO showed uniform
detectability for shear wave arrivals, fluctuating detectability for both shear and
compression wave amplitudes, and no detectability for compression wave arrivals. nAg
showed detectability only for compression amplitude, and nTiOz did not show
detectability. From the qualitative criteria, testing in the presence of nZnO at 0.03%,
0.3% and 2.7%; nAg at 3.7%; and nTi02 at 4.9% showed detectability for spectral
response, and no nanoparticles showed detectability for phase shift. Further tests for
spectral response and phase shift would be needed to obtain quantifiable criteria from
which detectability thresholds can be established.
Even though this report was performed in a controlled laboratory setting, the results
suggest a potential for the seismic detectability of some nanoparticles in the natural
environment. Since the seismic p-wave, s-wave, and spectral response was detectable for
the above mentioned nanoparticles the application of surface seismic methods to directly
127
-------
or indirectly detect nanoparticles in the natural environment may yield a measureable
response. If an industrial nanoparticle leak, or transportation accident releases
nanoparticles into the near subsurface resulting in detectable concentrations, then a non-
invasive surface seismic survey may assist in the characterization and mapping of such a
nanoparticle plume. This non-invasive geophysical mapping would then be utilized to
target the plume for future investigations. Of course, for this to be fully realized, future
research is required to understand the seismic response to nanoparticles within more
complicated geologic settings as well as the biogeochemical reactions which are likely to
occur from such a nanoparticle exposure. Regardless, the results from this study indicate
that it is feasible to detect the alteration of seismic properties due to the presence of some
nanoparticles within a glass bead matrix. Future research will expand upon these results
by increase the complexity of the experimentation and improving the testing apparatus.
9.2 Recommendations
It was realized that the received signals achieved with this test system could be
improved, and the following section gives recommendations on how to improve the test
system.
1. Improve repeatability of the test by refining placement practices for the glass beads
and modifying methods for soaking and seating the glass beads. The initial placement
of the glass beads might be improved in terms of seating by using an orbital shaker or
a vibration table.
128
-------
2. The number of wetting and draining cycles prior to carrying out testing should be
increased. The effects of the wetting and draining cycles can be observed with the
shear wave velocities from the baseline tests, which showed consistently higher
velocities for the second trial than the first trial, by 5 to 13%. This implies that the
process of repetitive wetting and draining of the glass beads continually improved the
seating of the glass beads.
3. The specimen height should be re-measured after each wetting and draining cycle to
check for deviation in height from initial preparation state. The change in specimen
height affects velocity (travel time), also amplitude to some extent.
4. The pulse signals were timed at 10-ms intervals; this restricted the quiet time between
them. These effects can be nullified by lengthening testing intervals.
5. The precision of results can be improved by increasing sampling rates for recording
received signals.
6. Consider testing different tip-to-tip distances (i.e. L value) depending on what types
of waves are being analyzed. When P-waves are being focused on, the L value can be
increased within limits dictated by signal attenuation so that fast P-wave arrivals are
not influenced by crosstalk. When S-waves are being focused on, the L value can be
decreased within limits dictated by near field effects to enhance the S-wave arrival
and reduce effects of side reflections.
7. To further reduce the near-field effects on the S-wave arrivals, a distorted sine-wave
can be considered for the input signal. This input was shown to reduce the near-field
effects in research done by Arroyo et al. (2003) and Jovicic et al. (1996).
129
-------
8. When testing in air, a lower cut off frequency for the high-pass filter should be
explored; in this study a 1 kHz high-pass filter was used to remove a 32 Hz
disturbance.
9. The reciprocity of the system should be tested to document accuracy by switching the
transmitter from the base plate to the top cap element.
10. The effects of nanoparticle dispersions in the absence of a granular matrix could also
be studied to further characterize them within the testing systems capabilities. A
baseline for a test such as this would be distilled water in the absence of granular
media.
11. For testing with the cross correlation method, filtering options and post processing
procedures should be explored to remove transfer functions and effects of multiple
reflections and other scattering on the received signal that dominate the wave train at
later times.
12. Processing of unwrapped phase data by shifting traces to have a common starting
point, where multiple traces would coincide with each other and lead to better
analysis should be explored.
13. The presence of nZnO at low and medium concentrations was detectable by S-wave
amplitudes, but it was not detectable at high concentration. This conflicting outcome
is not understood and merits further study. Also, future testing with nanoparticle
dispersions could be carried out at lower concentration levels to check for variations.
9.3 New Research Questions
130
-------
1. Why do nZnO particles appear to be more detectible than nTi02 or nAg by this
seismic method? Is it the substance being tested, or is it the testing method?
2. What are the effects of varying nanoparticle diameter?
3. What is the physical explanation at the nano scale for the observed results?
4. What are the effects of sample aging on detecting nanoparticle dispersions with
seismic methods?
ACKNOWLEDGEMENTS
We appreciate all the facilities provided by the UNLV Applied Geophysics Center,
and Department of Civil and Environmental Engineering and the U.S. The information in
this document has been funded partly by the United States Environmental Protection
Agency under student services contract EP09D000305 to M. Nihad Rajabdeen. It has
been subjected to the Agency's peer and administrative review and has been approved for
publication as an EPA document. Mention of trade names or commercial products does
not constitute endorsement or recommendation by EPA for use.
A special thank you to Drs. Carlos Santamarina and Changho Lee at Georgia Institute
of Technology; their advice on the design of the bender element testing system is greatly
appreciated. Also the following individuals helped with the little things that made this
research a success: Danney Glaser, John Zimmerman, Kim Rogers, Katrina Varner, Steve
Gardner and Marion Edison (U.S. EPA); Chris Cothrun, Helena Murvosh, Suchan
Lamichhane, Pinthep Kittipongdaja, Prajwol Tamrakar and Shawn Andersen (UNLV
AGC); Allen Sampson, Levia Lanier, Stacey Fisher, Kristen Young and Lily Magana
(UNLV).
131
-------
132
-------
APPENDIX 1 TABLES
Formulas used for calculations
-. T~> cc Anticipated - Experimental
1. Percentage Difference =
Anticipated
on 4. r-v-rc Trial 1-Trial 2 ^,^nnn/
L. Percentage Difference = x 100%
° Trial 1
i TT i >_ Distance travelled
6. Velocity =
Time taken
A n /r i ii/-n/\ Mw Mass of water
4. Moisture content ((£>%)=-
Mct> Mass of glass beads
5. Void ratio (e) = (
-------
Table A.I. Physical properties for glass bead specimens prepared with dumping method
Specimen
1
2
3
Average
Glass
beads
(grams)
2256
2193
2216
2221
Water
(grams)
543
555
543
547
CO (%)
24
25
24
25
e
0.60
0.63
0.61
0.62
Ysat
(kN/m3)
19.0
18.8
18.9
18.9
Difference
from
Average
0.41%
0.51%
0.09%
(co) - Moisture content of the glass beads when fully soaked
(e) - Void ratio
Ysat - Saturated unit weight
Table A.2. Physical properties for glass bead specimens prepared with stage fill method
Specimen
1
2
3
Average
Glass
beads
(grams)
2194
2200
2193
2196
Water
(grams)
566
574
562
568
CO (%)
26
26
26
26
e
0.65
0.65
0.64
0.65
Ysat
(kN/m3)
18.8
18.7
18.8
18.7
Difference
from
Average
0.02%
0.17%
0.15%
(co) - Moisture content of the glass beads when fully soaked
(e) - Void ratio
Ysat - Saturated unit weight
134
-------
Table A.3. Physical properties for tested water-saturated glass bead specimens
Specimen
1
2
3
Glass
beads
(grams)
1951
1961
1951
Physical state
Initial wetting
Drained state 1
Trial 1
Drained State 2
Trial 2
Initial wetting
Drained state 1
Trial 1
Drained State 2
Trial 2
Initial wetting
Drained state 1
Trial 1
Drained State 2
Trial 2
Volume
(ml)*
995
234
351
234
351
981
235
392
255
373
976
254
332
234
351
CO
(%)
51
12
30
12
30
50
12
32
13
32
50
13
30
12
30
e
1.28
0.30
0.75
0.30
0.75
1.25
0.30
0.80
0.33
0.80
1.25
0.33
0.75
0.30
0.75
Ysat
(kN/m3)
16.3
21.1
18.2
21.1
18.2
16.4
21.1
18.0
20.9
18.0
16.4
20.9
18.2
21.1
18.2
*Volume associated with drained state is volume retained in the column, and volume
associated with trials is volume required to saturate the specimen
Trial 2 represents duplicate tests following drainage and rewetting of test
specimen.
(co) - Moisture content of the glass beads when fully soaked
(e) - Void ratio
Ysat - Saturated unit weight
135
-------
Table A.4. Characteristics for specimens during nanoparticle dispersion testing
Specimen
(concentration)
nZnO (0.03%)
nZnO (0.3%)
nZnO (2.7%)
nTi02 (4.9%)
nAg (3.7%)
Mass
of glass
beads
(grams)
1960
1960
1971
1958
1966
Physical state
Initial wetting
Drained state 1
Water trial
Drained state 2
Nano trial
Initial wetting
Drained state 1
Water trial
Drained state 2
Nano trial
Initial wetting
Drained state 1
Water trial
Drained state 2
Nano trial
Initial wetting
Drained state 1
Water trial
Drained state 2
Nano trial
Initial wetting
Drained state 1
Water trial
Drained state 2
Nano trial
Volume
(ml)*
950
333
290
347
185
958
238
383
277
213
986
237
394
246
375
960
157
392
156
370
964
216
374
246
347
CO (%)
48
17
32
18
27
49
12
32
14
25
50
12
32
13
32
49
8
28
8
27
49
11
30
13
30
e
1.20
0.43
0.80
0.45
0.68
1.23
0.30
0.80
0.35
0.63
1.25
0.30
0.80
0.33
0.80
1.23
0.20
0.70
0.20
0.68
1.23
0.28
0.75
0.33
0.75
Ysat
(kN/m3)
16.5
20.1
18.0
20.0
18.6
16.4
21.1
18.0
20.7
18.9
16.4
21.1
18.0
20.9
18.0
16.4
22.1
18.5
22.1
18.6
16.4
21.4
18.2
20.9
18.2
*Volume associated with drained state is volume retained in the column, and volume
associated with trials is volume required to saturate the specimen
Nano Trial represents duplicate tests following drainage and rewetting of test
specimen.
(co) - Moisture content of the glass beads when fully soaked
(e) - Void ratio
- Saturated unit weight
136
-------
Table A.5. Direct (L) and reflected (D1+D2) travel path lengths
used to calculate velocities
Specimen
Water-saturated 1
Water-saturated 2
Water-saturated 3
nZnO at 0.03% concentration
nZnO at 0.3% concentration
nZnO at 2.7% concentration
nTiOz at 4.9% concentration
nAg at 3.7% concentration
L (mm)
62.3
60.7
61.0
61.6
62.3
61.1
60.9
61.1
D1+D2
(mm)
164.6
164.0
164.2
164.4
164.6
164.2
164.1
164.2
137
-------
0.5
I -0.5
APPENDIX 2 FIGURES
0.5
1.5
2 2.5
Time (s)
3.5
x 10
o
A /
A' / \ If
<- X: 0.0003027 frfl / A h
Y: 0.03791 ^ l\ \\ l\ ^
' .A A / I / \ /Av^ ^\ A //
""^ V \ / i\ // or A \y \y i I
V iu V V/v" V
i
i
i
i
A
1 /A -A\
I All -A -A
1 / \| V \/1
\y
10
15
Time (s)
x 10
Figure A.I. Water-saturated glass bead specimen 1: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 1 of trials 1 and 2;
1000 recordings averaged per repetition.
0.5
E
< -0.5
- Trial 1 |-
4 5
Time (s)
9 10
x 10"4
0.5
< -0.5
X: 0.0003223
Y: 0.1943
\ /
Time (s)
-Tfial 2 -
10
x 10
Figure A. 2. Water-saturated glass bead specimen 1: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 1 of trials 1 and
2; 1000 recordings averaged per repetition.
138
-------
1
0.5
0
-0.5
-1
1
0.5
0
-0.5
-1
^"-^ Wr/T ~
y
V
I
k
i 1
\ii
v 1
III r1 : 1
ft i Tri,, o
flifll/l/X/ AX^^'A, -^^^.ab-^
| I Vj/l/ ^ ^^ A X^T""^
If' '
1 1
1 1
T i
1 I I I I I
0 0.5
1 1.5 2 2.5 3 3.5 t
Time (s) x 1Q-3
1
1~~
X: 0.0003
Y:0.045f
0
A A II A
K--A /I fl -11 ,TY.
A/|fii A/IAA
A VAA A
U w
v
5 10 1
Time (s) x 1Q-4
Figure A.3. Water-saturated glass bead specimen 1: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 2 of trials 1 and 2;
1000 recordings averaged per repetition.
Amplitude (V) Amplitude (V)
CD O CD O
-^01001-^ -^ 01 o 01 -^
-
=-
X: 0.0003223 n
Y: 0.2303 / \
A A /
i V
i
i
i
i
i
A
v
A ~
A/
A/
/
\f
vy
01 234567891
Time (s) .jg-4
X: 0.000322
" Y: 0.1896
3
/
vy
A
K /
v
A
\ /
\ /
A
\ A
\J '
V /A
A^^
/
v
01 234567891
Time(s) ...-»
Figure A.4. Water-saturated glass bead specimen 1: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 2 of trials 1 and
2; 1000 recordings averaged per repetition.
139
-------
0.5
< -0.5
1
J1
A r
i
1 HA -All -I
on
I 1 1
I
iii p 1
|/tfV_A_A/ /V i x^^-^^^^i^-^
j/_i_y v/
-y r T
U 1 1
1 1
1 1
1 [ 1 1 [
0.5
1.5
2 2.5
Time (s)
3.5
x 10
1
0.5
1 -0.5
15
0 5 10
Time (s) ^ 1Q-4
Figure A. 5. Water-saturated glass bead specimen 1: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 3 of trials 1 and 2;
1000 recordings averaged per repetition.
0.5
< -0.5
-7-
/ \
4 5
Time (s)
9 10
x 10"4
1
0.5
| 0
~Q_
E
< -0.5
-iiiiiir"
X: 0.0003223
Y: 0.1851
r
345
Time (s)
-Trral-2
10
x 10
Figure A.6. Water-saturated glass bead specimen 1: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 3 of trials 1 and
2; 1000 recordings averaged per repetition.
140
-------
0.5
< -0.5
n
- 4[
!|
.
A j i
ilAni
A
ir ir
|T J
1
ii
i
(T r
A '
|/v^
1
1
1
1
1
^^^
^^^^i?
1
1
1
1
1
1
1
1
1
[
0.5
1.5
2 2.5
Time (s)
3.5
x 10
1
0.5
0
1 -0.5
10
15
Time (s)
x 10
Figure A.7. Water-saturated glass bead specimen 2: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 1 of trials 1 and 2;
1000 recordings averaged per repetition.
Amplitude (V) Amplitude (V)
O O CD O
-^cnocn-^ -^cnocn-^
1 1 1 1 1 1 1
X: 0.0003223 A A, '
Y: 0.3786 ' A A '
A / ' / \ A A
'/ \ /A A A \ A\ / \
/ \ /' \ / ' i /A /l\/\
^N^. /r ~\ T r " \ Trr r r i rrV/
\ / ' A /A/|V/
\y ' \ / A / A /
i i \A i \ / i \y i
i i i vj i i
i i i i i
III
01 234567
Time (s)
I I I I I I I
X 0.0003223 ] A A /V
"Y:0-219 ^""f\""7"\"""7V"">v
^"\ A / / \ /-A/A
r^^T^-A \ ~ / r -\ rn / n / A/ \
\ / i \ / / \ / A /
A/ A A '
A
i i ^ i
i i i
i
/^\ I
^A A /
\ /
\ /
^
3 9 1
x 10"4
I
/
A ;V;
i
i
i
i ii
i i i i i i i i
01 234567
Time (s)
3 9 1
x 10"4
Figure A.8. Water-saturated glass bead specimen 2: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 1 of trials 1 and
2; 1000 recordings averaged per repetition.
141
-------
x 10
0 5
Y: 0.07279 A /M /
^ \7 \7
v/
A A
\ / \
1 11
It // u
y/
AA ^
7/\\ A I
/ \V \/ \ //
y/
w
w
V-S-S--A---A
M A A/
t irV/r \w \\T
Ui 1]
i/r Y/
V y
n /
10
15
Time (s)
x 10
Figure A.9. Water-saturated glass bead specimen 2: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 2 of trials 1 and 2;
1000 recordings averaged per repetition.
Amplitude (V) Amplitude (V)
CD O CD O
-^ en o en -^ -^ en o en -^
II
X: 0.0003223
Y: 0.3273
y'\ i
/
"\
~\~f~
\l
V
\ f
\ 1
I
\ 1
\ 1
\J
\
\ /
\ /
\ /
\ r
\ /
" V ""
\ A
\ / \
\ /
i
y^A /
vy \ /
\ /
\7
t
01 234567891
Time (s) ,g-4
II
=- _,
i
X: 0.0003223
" Y: 0.2006
i \ /
i
i
i
i
i
1
/ >
; f
\ /
\
^
V /
\j
\ /
\ 7
\ /
r\
\J
i
-/
\7 \J
i
01 234567891
Time(s) -4
Figure A. 10. Water-saturated glass bead specimen 2: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 2 of trials 1 and
2; 1000 recordings averaged per repetition.
142
-------
0.5
< -0.5
1-A--
0.5
1.5
2 2.5
Time (s)
3.5
x 10
1
0.5
0
< -0.5
10
15
Time (s)
x 10
Figure A.I 1. Water-saturated glass bead specimen 2: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 3 of trials 1 and 2;
1000 recordings averaged per repetition.
0.5
15.
E
< -0.5
X: 0.0003223
Y: 0.3504
-V7-
- Trial 1 |-
4 5
Time (s)
9 10
x 10"4
1
0.5
0)
E
< -0.5
X: 0.0003223
Y: 0.2282
/ \
- Trial 2 |-
10
Time (s)
x 10
Figure A. 12. Water-saturated glass bead specimen 2: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 3 of trials 1 and
2; 1000 recordings averaged per repetition.
143
-------
0.5
< -0.5
0.5
1.5
2 2.5
Time (s)
3.5
x 10
0.5
< -0.5
10
15
Time (s)
x 10
Figure A. 13. Water-saturated glass bead specimen 3: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 1 of trials 1 and 2;
1000 recordings averaged per repetition.
0 5
0)
E o
0.
E
0)
~o_
D
?
X: 0.0003223
/ A /
- \ /A / -
V/i\A
1 V
1
1
1
1
1
3 4
Time (s
X: 0.0003223
i
i
i
i
i
i
1
/"
\ /
\ /
\ I
\ 1
5
)
_r~\
\ 1
\ A
\J
3
V
\
\
\
\ ,--.
\ /
1
I
I
3
I
I
j
\ /
\ j
1 1
x 10"4
/
A /
\ /
456
Time (s)
10
x 10"
Figure A. 14. Water-saturated glass bead specimen 3: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 1 of trials 1 and
2; 1000 recordings averaged per repetition.
144
-------
0.5
< -0.5
0.5
1.5
2 2.5
Time (s)
3.5
x 10
0.5
< -0.5
10
15
Time (s)
x 10
Figure A. 15. Water-saturated glass bead specimen 3: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 2 of trials 1 and 2;
1000 recordings averaged per repetition.
1
0.5
0)
1 °
"o.
E
< -0.5
_ I L _
/K^ >\
rx^->-^r -
X: 0.0003223
Y: 0.416
r-Jv--/-rV-^--r f - r -
4 5
Time (s)
Trial 1
_ I
~ r^ V r
-V--
9 10
x 10"4
1
0.5
_
< -0.5
X: 0.0003223
Y: 0.2207
r A-/ r V-^ - - r t r
4 5
Time (s)
9 10
x 10"4
Figure A. 16. Water-saturated glass bead specimen 3: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 2 of trials 1 and
2; 1000 recordings averaged per repetition.
145
-------
0.5
< -0.5
I
w
'
i -/I
m I
J if 1
r 1
i i"
1 t\ ' _-/^
1 177 : ^\
[ u ^ W
i
i
T
/A Av
/ \
/ \
7y
v/
V^vy /71
7 -/
\ i
7
i
i
i
i
i
i
0.5
1.5
2 2.5
Time (s)
3.5
x 10
1
0.5
0
1 -0.5
. X: 0.0003027
Y: 0.1326 A
--W--
10
15
Time (s)
x 10
Figure A. 17. Water-saturated glass bead specimen 3: comparison of 8 kHz pulse signal
used to highlight slow compression wave arrivals, showing repetition 3 of trials 1 and 2;
1000 recordings averaged per repetition.
0.5
< -0.5
X: 0.0003223
Y: 0.4406
- Trial 1 |-
4 5
Time (s)
9 10
x 10"4
0.5
< -0.5
4 5
Time (s)
Trial 2
9 10
x 10"4
Figure A. 18. Water-saturated glass bead specimen 3: comparison of 8 kHz pulse signal
used to highlight slow compression wave amplitude, showing repetition 3 of trials 1 and
2; 1000 recordings averaged per repetition.
146
-------
0.5
Q.
E
-0.5
-1.5L
3
Time (s)
x 10
Figure A. 19. Water-saturated glass bead specimen 1: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 1 of trials 1 and 2; 1000 recordings averaged
per repetition.
0.5
Q.
E
-0.5
-1.5
3
Time (s)
x 10
Figure A.20. Water-saturated glass bead specimen 1: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 2 of trials 1 and 2; 1000 recordings averaged
per repetition.
147
-------
0.5
-0.5
-1.5
3 4
Time (s)
x 10
Figure A.21. Water-saturated glass bead specimen 1: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 3 of trials 1 and 2; 1000 recordings averaged
per repetition.
0.5
-0.5
-1.5
W---7P
v' r
!: 0 002402 / /
': 0005334 / /
-\ h
3 4
Time (s)
x 10'
Figure A.22. Water-saturated glass bead specimen 2: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 1 of trials 1 and 2; 1000 recordings averaged
per repetition.
148
-------
0.5
-0.5
-1.5
/ /
X: 0.002539
Y: 0.01557
\
3 4
Time (s)
Y. 10
Figure A.23. Water-saturated glass bead specimen 2: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 2 of trials 1 and 2; 1000 recordings averaged
per repetition.
0.5
Q.
E
-0.5
-1.5
X. 0.002412
Y: 0.03175 / /
l\*.
/ ] Y: 0.01097
3 4
Time (s)
x 10
Figure A.24. Water-saturated glass bead specimen 2: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 3 of trials 1 and 2; 1000 recordings averaged
per repetition.
149
-------
0.5
-0.5
-1.5
X: 0.001816
Y: 0.01884 /
3
Time (s)
x 10
Figure A.25. Water-saturated glass bead specimen 3: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 1 of trials 1 and 2; 1000 recordings averaged
per repetition.
0.5
Q.
E
-0.5
-1.5
3 4
Time (s)
x 10
Figure A.26. Water-saturated glass bead specimen 3: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 2 of trials 1 and 2; 1000 recordings averaged
per repetition.
150
-------
0.5
-0.5
-1.5
3
Time (s)
x 10
Figure A.27. Water-saturated glass bead specimen 3: comparison of 1 kHz pulse signal
used to highlight shear, showing repetition 3 of trials 1 and 2; 1000 recordings averaged
per repetition.
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.5
1.5 2 2.5
Frequency (Hz)
3.5
x 10
Figure A.28. Water-saturated glass bead specimen 2: analysis of coherence for frequency
sweep of 0 - 30 kHz used to find high coherence range, showing a repetition of trials 1
and 2; 1000 recordings averaged per repetition.
151
-------
0.5
1.5 2 2.5
Frequency (Hz)
3.5
x 10
Figure A.29. Water-saturated glass bead specimen 3: analysis of coherence for frequency
sweep of 0 - 30 kHz used to find high coherence range, showing a repetition of trials 1
and 2; 1000 recordings averaged per repetition.
2000
-2000
-4000
-6000
-8000
-10000
0
0.5
1.5 2
Frequency (Hz)
3.5
x 10
Figure A.30. Water-saturated glass bead specimen 2: analysis of unwrapped phase angles
of 0 - 30 kHz sweep, showing three repetitions each for trials 1 and 2; 1000 recordings
averaged per repetition.
152
-------
1000
0
-1000
-2000
-3000
-4000
-5000
-6000
-7000
-8000
0.5
1.5
2 2.5
Frequency (Hz)
3.5
x 10
Figure A.31. Water-saturated glass bead specimen 3: analysis of unwrapped phase angles
of 0 - 30 kHz sweep, showing three repetitions each for trials 1 and 2; 1000 recordings
averaged per repetition.
153
-------
APPENDIX 3 BACKGROUND SIGNAL ANALYSIS
The background signal in the absence of any actuated pulse signal was recorded to
quantify the extent to which the residual energy from a preceding pulse would affect the
onset of the following pulse. This was demonstrated by comparing the quiet time in-
between pulses to the background signal in the absence of any pulse.
To record the background signal, the equipment was connected in the same manner as
described in section 2.5, except the coaxial cable connected to the BNC cable from the
receiver bender element was disconnected. Therefore with no receiver bender element
connected to the dynamic signal analyzer, what was received was the background signal
(noise) within the wires and equipment of the system. It should be noted that this
background study was not conducted with the same bender elements used for the testing;
these bender elements were the same type as those used for the testing, but were newly
wired and prepared.
The isolated background signals were recorded for three repetitions, with 1000
recordings averaged per repetition. The received signals from the baseline testing with
essentially saturated glass bead trials presented in Chapter 7 were used for comparison
with the isolated background signals. Figure A3.1 shows the received 8 kHz pulse signals
(shown in Fig. 7.3) and Figure A3.2 shows the received 1 kHz pulse signals (shown in
Fig. 7.4), each overlaid on the background signal to compare the difference in amplitude,
where the impact of the residual energy from the preceding pulse on the onset of the
following can be visualized.
154
-------
To quantify the differences present between the background and the effects of the
residual pulse energy, the signal amplitudes from the data points of the three repetitions
were compared from time 10 ms to 15 ms; 10 ms is the midpoint between the two pulses
and time 15 ms is the point of actuation of the second pulse. Table A3.1 presents average
and maximum amplitude differences between the isolated background signal and the
background signal in-between actuated 8 kHz pulses and 1 kHz pulses for the above
mentioned interval. The results show the maximum difference recorded for the 8 kHz
pulse was 6.69E-1 V and the maximum difference recorded for the 1 kHz pulse was
8.81E-1 V. The average amplitude difference for both the 8 kHz and 1 kHz pulses ranged
from 7.86E-5 V to 8.86E-5 V.
The maximum amplitude difference established for the 8 kHz pulse is larger than the
amplitudes of the slow P-wave arrival picks that were made during testing. The
differentiation between the background noise and a slow P-wave arrival was made by
considering the change in frequency which led to the amplitude gain, and showed
deviation from the background signal. The maximum amplitude difference established
for the 1 kHz pulse is minuscule when compared to the S-wave arrival; the S-wave
arrival was also differentiable by its shape.
The difference in amplitude between the isolated background signal and the
background signal in-between pulses can be reduced by increasing the interval between
pulses, and further refining signal filtering and processing as mentioned in the
recommendations section of the report to minimize these effects on received signals.
155
-------
Table A3.1. Difference between isolated background signal and background signal in-
between actuated pulse signals
Signal
8kHz
1kHz
Difference
Average
Maximum
Average
Maximum
AR1 (V)
8.80E-05
6.69E-01
8.80E-05
8.75E-01
AR2 (V)
8.86E-05
7.00E-01
8.86E-05
8.81E-01
AR3(V)
7.86E-05
6.82E-01
7.86E-05
8.61E-01
AR1, 2 and 3 indicate the difference between amplitudes of the isolated background
signal and the background signal in the presence of actuated signals for three
repetitions.
156
-------
0.5
I -0.5
0.05
-0.05
0.01
8 kHz signal
Background signal
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s)
0.011
0.012
0.013
Time (s)
0.014
0.015
0.016
Figure A3.1. 8 kHz sine pulses (200 Hz high pass filter) overlaid on the isolated
background signal (no filter). Each signal is a single repetition of 1000 recordings
averaged.
a. Received signals showing two consecutive pulses and quiet time between pulses;
b. Detailed view of residual effects on the isolated background signal
157
-------
o
1 ,
tlr~~~
X
r\
l\
1 \
\J
\l
y
V
K^
I |
I |
(- ""
I |
I |
I |
I |
I |
I |
1 [
A
/A
D
V
r
\i
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s)
0.05
E
<
-0.05
X. 0.01087
Y: 0.01145
X. 0.01087
Y: 0.000619
0.01
0.011
0.012
0.013
Time (s)
0.014
0.015
0.016
Figure A3.2. 1 kHz sine pulses (200 Hz high-pass filter) overlaid on the isolated
background signal (no filter). Each signal is a single repetition of 1000 recordings
averaged.
a. Received signals showing two consecutive pulses and quiet time between pulses;
b. Detailed view of residual effects on the isolated background signal
158
-------
REFERENCES
Arroyo, M., Wood, D. M., and Greening, P. D. (2003). Source near-field effects and
pulse tests in soil samples. Geotechnique, 53(3), 337-345.
Arulnathan, R., Boulanger, R. W., and Riemer, M. F. (1998). Analysis of bender element
tests. Geotechnical Testing Journal, 21(2), 120-131.
Biot, M. A. (1956). Theory of propagation of elastic waves in fluid saturated porous
solid. Journal of the Acoustical Society of America. 28(2), 168-191.
Birkholz, M. (1995). Crystal-field induced dipoles in heteropolar crystals II: Physical
significance. The European Physical Journal B, 96(3), 333-340.
Blewett, J., Blewett, I. J., and Woodward, P. K. (1999). Measurement of shear-wave
velocity using phase-sensitive detection techniques. Canadian Geotechnical Journal,
36(5), 934-939.
Brignoli, E. G. M., Gotti, M., and Stokoe, K. H. (1996). Measurement of shear waves in
laboratory specimens by means of piezoceramic transducers. Geotechnical Testing
Journal, 19 (4), 384-397.
Clayton, C. R. I., Theron, M., and Best, A. I. (2004). The measurement of vertical shear-
wave velocity using side-mounted bender elements in the triaxial apparatus.
Geotechnique, 54(7), 495-498.
Conlon, M., (2009). EPA science in action: Nanotechnology research program, National
Exposure Research Laboratory, EPA Office of Research and Development. Retrieved
11/30/09, from
http://www.epa.gov/nanoscience/quickfinder/pdf/nanotech_nanomaterials.pdf
Da Fonseca, A. V., Ferreira, C., and Fahey, M. (2008). A framework interpreting bender
element tests, combining time-domain and frequency-domain method. Geotechnical
Testing Journal, 32(2), 91-107.
Deniz, R. 0. (2008). Bender elements and bending disks for measurement of shear and
compression wave velocities in large sand specimens. Master's Thesis, Northeastern
University, Boston, MA.
Dyvik, R., and Madshus, C. (1986). Lab measurements of Gmax using bender elements.
Publication - Norwegian Geotechnical Institute, 161 pp.
159
-------
Dyvik, R., and Olsen, T. S. (1991). Gmax measured in oedometer and DSS tests using
bender elements. Publication - Norwegian Geotechnical Institute, 181 pp.
Greening, P. D., and Nash, D. F. T. (2004). Frequency domain determination of GO using
bender elements. Geotechnical Testing Journal, 27(3), 288-294.
Jovicic, V., Coop, M. R., and Simic, M. (1996). Objective criteria for determining Gmax
from bender element tests. Geotechnique, 46(2), 357-362.
Joyce,R.A., Glaser,D.R., Werkema Jr.,D.D., Atekwana,E.A., Spectral induced
polarization response to nanoparticles in a saturated sand matrix, Journal of Applied
Geophysics, doi:10.1016/j.jappgeo.2011.11.009, 2011.
Lee, J., and Santamarina, J. C. (2005). Bender elements: Performance and signal
interpretation. Journal of Geotechnical and Geoenvironmental Engineering, 131(9), 1063-
1070.
Lee, C., Lee, J.S., Lee, W., and Cho, T. H. (2007). Experiment setup for shear wave and
electrical resistance measurements in an oedometer. Geotechnical Testing Journal, 31(2).
Leong, E. C., Yeo, S. H., and Rahardjo, H. (2005). Measuring shear wave velocity using
bender elements. Geotechnical Testing Journal, 28(5), 488-498.
Nakagawa, K., Soga, K., and Mitchell, J. K. (1996). Pulse transmission system for
measuring wave propagation in soils. Journal of Geotechnical Engineering, 122(4), 302-
308.
Nakagawa, K. Soga, K. and Mitchell, J. K. (1997). Observation of Biot compressional
wave of the second kind in granular soils. Geotechnique, 47(1), 133-147.
National Nanotechnology Initiative (2009). What is nanotechnology? Retrieved 12/04/09,
from http://www.nano.gov/html/facts/whatIsNano.html
Patel, A., Bartake, P.P., and Singh, D.N. (2009). An empirical relationship for
determining shear wave velocity in granular materials accounting for grain morphology.
Geotechnical Testing Journal, 32(1), 1-10.
Piezo Systems, I. (2009). History of piezoelectricity. Retrieved 11/20/09, from
http://www.piezo.com/tech4history.html
Fiona, T. J. (1980). Observation of a second bulk compressional wave in a porous
medium at ultrasonic frequencies. Applied Physics Letters, 36, 259-261.
160
-------
Fiona, T. J., D'Angelo, R., and Johnson, D. L. (1990). Velocity and attenuation of fast,
shear and slow waves in porous media. Ultrasonics Symposium Proceedings, Institute of
Electrical and Electronics Engineers Inc. (IEEE), Piscataway, NJ, 3, 1233-1239.
Rajabdeen, M. N., Luke, B., Werkema, D., and Glaser, D. (2011). Application of
piezoceramics in characterizing granular media. Engineering Geology and Geotechnical
Engineering Symposium, eds. Biggar, N., Luke, B., and Werle, J. Idaho State University,
Pocatello, 490-496.
Reynolds, G. (2000). The fundamentals of signal analysis. Agilent Technologies.
Application note 243.
Rio, J. F. M. E. (2006). Advances in laboratory geophysics using bender elements.
Dissertation, University College London, UK.
Santamarina, J. C., Klein, K. A., and Fam, M. A. (2001). Soils and waves - particulate
materials behavior, characterization and process monitoring, John Wiley and Sons, New
York.
Sengpiel, E. (2010). Sengpielaudio, Speed of sound in air calculation and temp - air.
Retrieved 11/04/10, from http://www.sengpielaudio.com/calculator-speedsound.htm
Viggiani, G., and Atkinson, J. H. (1995). Interpretation of bender element tests.
Geotechnique, 45(1), 149-154.
Wang, Y. H., Lo, K. F., Yan, W. M., and Dong, X. B. (2007). Measurement biases in the
bender element test. Journal of Geotechnical and Geoenvironmental Engineering, 133(5),
564-574.
Williams, K. H., Ntarlagiannis, D., Slater, L. D., Dohnalkova, A., Hubbard, S. S., and
Banfield, J. F. (2005). Geophysical imaging of stimulated microbial biomineralization.
Environmental Science and Technology, 39(19), 7592-7600.
Zeng, X., and Ni, B. (1998). Measurement of Gmax under anisotropic loading condition
using bender elements. Geotechnical Earthquake Engineering and Soil Dynamics III, ed.
P. Dakoulas and M. Yegian. Geotechnical Special Publication 75(1), 189-200.
American Society of Civil Engineers, Reston, VA.
Zhihai, X., Ping, T., Zhidong, J., Zhicheng, G., and Liming, W. (2008). Eliminate
corrosion on grounding system by a conductive RTV silicone coating. Annual Report -
Conference on Electrical Insulation and Dielectric Phenomena 49-51, Institute of
Electrical and Electronics Engineers Inc. (IEEE), Piscataway, NJ.
161
-------
&EPA
United States
Environmental Protection
Agency
Office of Research
and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
EPA/600/R-12/547
June 2012
www.epa.gov
Please make all necessary changes on the below label, detach
or copy and return to the address in the upper left hand corner.
If you do not wish to receive these reports CHECK HERE I I:
detach, or copy this cover, and return to the address in the
upper left hand corner.
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA PERMIT No. G-35
Recy cl ed /Recy cl able
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
------- |