Bender Thesis

Ghent University
Faculty of Engineering
Department of Civil Engineering (TW15)
Laboratory of Soil Mechanics
Dynamic Soil Properties out of SCPT

and Bender Element Tests with
Emphasis on Material Damping
Lutz Karl
A thesis submitted in accordance with the requirements of

the degree of Doctor of Philosophy in the Faculty of
Engineering, Department of Civil Engineering
Academic year 2004–2005

Supervisor: Prof. dr. ir. W. Haegeman
Laboratory of Soil Mechanics

Department of Civil Engineering (TW15)
Faculty of Engineering
Ghent University
Technologiepark 905
9052 Ghent (Zwijnaarde)
Belgium
Tel. +36 9 264 57 23

+36 9 264 57 17
Fax +36 9 264 57 49
http://terzaghi.ugent.be
Preface
I wish to thank all persons who have contributed to this thesis with their
work, their ideas and helpful comments. First of all, I would like to thank my
supervisor Prof. Wim Haegeman for providing the testing equipment and for
the extensive assistance throughout this research.
I am also very grateful to Prof. Geert Degrande from the K.U. Leuven for
his numerous valuable suggestions and support.
Furthermore I wish to thank the co-workers of the STWW-project ”Traffic
induced vibrations in buildings” from the K.U. Leuven, especially Lincy Pyl,
Dr. Janusz Kogut, Serge Jacobs and Kathleen Geraedts.
The help of Prof. Mia Loccufier, Peter Buffel, Wouter Ost and the staff of
the Laboratory of Soil Mechanics in Ghent is very much acknowledged. Special
thanks deserves also Lou Areias for shearing his knowledge on the SCPT, for
performing of field tests and for the visual wave velocity interpretation at the
sites in Retie, Waremme and Lincent.
The research presented in this thesis is embedded in the STWW-project
IWT000152 ”Traffic induced vibrations in buildings”. The financial support of
the Ministry of the Flemish Community for this project is likewise gratefully
acknowledged.
Lutz Karl
January 2005
v
Contents
Preface v
Frequently used symbols and units xi
I Introduction and background 1

1 Introduction 3
1.1 Purpose and scope . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Dynamic soil properties 7

2.1 Dynamic shear modulus . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Attenuation parameters of soils . . . . . . . . . . . . . . . . . . 11
3 Methods to determine G and D 19

3.1 Laboratory tests . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.1.1 Piezoelectric bender and compression element tests . . . 21
3.1.2 Cyclic triaxial tests . . . . . . . . . . . . . . . . . . . . . 22
3.1.3 Cyclic simple shear tests . . . . . . . . . . . . . . . . . . 23
3.1.4 Cyclic torsional shear tests . . . . . . . . . . . . . . . . 23
3.1.5 Resonant column test . . . . . . . . . . . . . . . . . . . 23
3.1.6 Free torsion pendulum test . . . . . . . . . . . . . . . . 24
3.2 In situ tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1 Seismic reflection test . . . . . . . . . . . . . . . . . . . 28
3.2.2 Seismic refraction test . . . . . . . . . . . . . . . . . . . 29
3.2.3 Spectral analysis of surface waves (SASW) . . . . . . . 30
3.2.4 Seismic cross-hole test . . . . . . . . . . . . . . . . . . . 31
3.2.5 Seismic down-hole and up-hole test . . . . . . . . . . . . 32
3.2.6 Seismic cone penetration test . . . . . . . . . . . . . . . 33
3.2.7 Geotomography . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.8 High-strain tests . . . . . . . . . . . . . . . . . . . . . . 34
vii
viii CONTENTS
II Characterization of the testing sites 35

4 Test site Retie 37
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 Borings and undisturbed sampling . . . . . . . . . . . . . . . . 38
4.3 Cone penetration test (CPT) . . . . . . . . . . . . . . . . . . . 41
4.4 Seismic cone penetration test (SCPT) . . . . . . . . . . . . . . 41
4.4.1 Test description . . . . . . . . . . . . . . . . . . . . . . . 41
4.4.2 Test results for the wave velocities . . . . . . . . . . . . 44
4.4.3 Test results for the damping ratio . . . . . . . . . . . . 45
4.5 Spectral analysis of surface waves (SASW) . . . . . . . . . . . . 47
4.5.1 Test description . . . . . . . . . . . . . . . . . . . . . . . 47
4.5.2 Test results . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.6 Overview of the test results Retie . . . . . . . . . . . . . . . . . 49
5 Test site Lincent 53

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.4.1 Remarks on the testing Setup . . . . . . . . . . . . . . . 60
5.4.2 Test results for the wave velocity . . . . . . . . . . . . . 61
5.5 Spectral analysis of surface waves (SASW) . . . . . . . . . . . . 62
5.5.1 Remarks on the testing setup and inversion assumptions 62
5.5.2 Test results . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.6 Overview of the test results Lincent . . . . . . . . . . . . . . . 64
6 Test site Waremme 67

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.3 In situ tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.3.1 Cone penetration test (CPT) . . . . . . . . . . . . . . . 71
6.3.2 Seismic cone penetration test (SCPT) . . . . . . . . . . 72
6.4 Laboratory tests to obtain Gmax and D . . . . . . . . . . . . . 77
6.4.1 Bender elements with time arrival interpretation . . . . 77
6.5 Overview of the test results Waremme . . . . . . . . . . . . . . 81
7 Test site Sint-Katelijne-Waver 85

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.3 In situ tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
7.3.1 Marchetti dilatometer test (DMT) . . . . . . . . . . . . 87
CONTENTS ix
7.3.2 Cone penetration test (CPT) . . . . . . . . . . . . . . . 88

7.3.3 Seismic cone penetration test (SCPT) . . . . . . . . . . 88
7.3.5 Seismic refraction test (SRT) . . . . . . . . . . . . . . . 90
7.4 Laboratory tests to obtain Gmax and D . . . . . . . . . . . . . 91
7.4.1 Bender elements with time arrival interpretation . . . . 91
7.5 Overview of the test results St.-Katelijne-Waver . . . . . . . . . 94
8 Test site Ghent 97

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
8.2 Borings and sampling . . . . . . . . . . . . . . . . . . . . . . . 97
8.4.1 Testing setup . . . . . . . . . . . . . . . . . . . . . . . . 99
8.4.2 Test results for the wave velocity . . . . . . . . . . . . . 100
8.5 Overview of the test results Ghent . . . . . . . . . . . . . . . . 101
III Studies on testing methods 105

9 SCPT based techniques 107
9.1 Motivation and purpose . . . . . . . . . . . . . . . . . . . . . . 107
9.2 Applied equipment and selection criteria . . . . . . . . . . . . . 108
9.2.1 Seismic source . . . . . . . . . . . . . . . . . . . . . . . 108
9.2.2 Seismic cones . . . . . . . . . . . . . . . . . . . . . . . . 111
9.2.3 Data acquisition system . . . . . . . . . . . . . . . . . . 117
9.3 Methods for the shear modulus . . . . . . . . . . . . . . . . . . 119
9.3.1 Direct time methods . . . . . . . . . . . . . . . . . . . . 121
9.3.2 Indirect time methods . . . . . . . . . . . . . . . . . . . 121
9.4 Methods to obtain the Damping Ratio . . . . . . . . . . . . . . 125
9.4.1 Attenuation coefficient method . . . . . . . . . . . . . . 125
9.4.2 Spectral ratio slope method . . . . . . . . . . . . . . . . 128
9.4.3 Hysteresis stress-strain-curves . . . . . . . . . . . . . . . 131
9.5 Artificially generated signals . . . . . . . . . . . . . . . . . . . . 142
9.5.1 Calculated velocities from the simulated signals . . . . . 151
9.5.2 Calculated damping ratio . . . . . . . . . . . . . . . . . 152
9.5.3 Conclusions for the analysis of real SCPT data . . . . . 156
9.6 Measurements performed at the test sites . . . . . . . . . . . . 160
9.6.1 Campaign I (geophones) . . . . . . . . . . . . . . . . . . 160
9.6.2 Campaign II (HAMLIN accelerometers) . . . . . . . . . 160
9.6.3 Campaign III, Ghent (Brüel & Kjær accelerometers) . . 166
9.7 Summary and remaining problems . . . . . . . . . . . . . . . . 169
x CONTENTS
10 Bender element technique 173

10.1 Motivation and purpose . . . . . . . . . . . . . . . . . . . . . . 173
10.2 Description of the equipment . . . . . . . . . . . . . . . . . . . 174
10.2.1 Bender elements . . . . . . . . . . . . . . . . . . . . . . 174
10.2.2 Signal generation and measurement apparatus . . . . . 176
10.3 Techniques to determine Gmax . . . . . . . . . . . . . . . . . . 178
10.3.1 Wave travel distance . . . . . . . . . . . . . . . . . . . . 178
10.3.2 Selection of the input-signal shape . . . . . . . . . . . . 179
10.3.3 Methods for determining the travel time . . . . . . . . . 179
10.3.4 Difficulties in the arrival time determination . . . . . . . 182
10.4 Methods to obtain the damping ratio . . . . . . . . . . . . . . . 186
10.4.1 Resonant method . . . . . . . . . . . . . . . . . . . . . . 186
10.4.2 Use of multiple shear wave arrivals . . . . . . . . . . . . 204
10.4.3 Use of different travel paths . . . . . . . . . . . . . . . . 210
10.5 The resonant method . . . . . . . . . . . . . . . . . . . . . . . . 213
10.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 213
10.5.2 Site in Waremme . . . . . . . . . . . . . . . . . . . . . . 213
10.5.3 Site in Sint-Katelijne-Waver . . . . . . . . . . . . . . . . 215
10.6 Summary and remarks . . . . . . . . . . . . . . . . . . . . . . . 216
11 Conclusions and summary 235
IV Appendix 239
A Accuracy and costs of soil tests 241
A.1 Laboratory tests . . . . . . . . . . . . . . . . . . . . . . . . . . 242
A.2 Field tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
A.3 Selection conclusions . . . . . . . . . . . . . . . . . . . . . . . . 244
Bibliography 249
Frequently used symbols
and units
The following table presents the most frequently used symbols and abbrevia-
tions. The symbols are, in most cases, also defined when they are introduced
in the text. Time histories are in general represented by small letters; their
frequency domain representation by capital letters.
Symbols and units
A signal amplitude, amplitude spectrum; m, m/s, m/s2 ;

area m2
a sample radius, distance; m;
time history m, m/s, m/s2
c cohesion; damping coefficient; Pa; kg/s;
wave velocity m/s
CA added mass coefficient −
D damping ratio −, %
d diameter, depth, distance m
Ds damping ratio of the s-wave −, %
E elasticity modulus; energy Pa; J
Emax , E0 dynamic elasticity modulus Pa
EI bending stiffness Nm2
e void ratio −
f frequency Hz
fcut cut-off frequency Hz
Fe void ratio function −
FFT Fast Fourier Transformation
as a synonym for a discrete
Fourier Transformation in general
G shear modulus Pa
Gmax , G0 dynamic shear modulus Pa
H transfer function; height −; m
h height m
IFFT inverse transformation to the FFT
xi
xii FREQUENTLY USED SYMBOLS
I area moment of inertia; impedance m4 ; kg/(m2 s)

Ip plasticity index −, %
J mass polar moment of inertia kg m2
K bulk modulus Pa
k stiffness; wavenumber; kg/s2 ; 1/m;
shear coefficient; permeability −; m/s
ki intrinsic permeability m2
l, L length m
m mass kg
m̄a added mass per unit length kg/m
OCR over consolidation ratio −
p pseudo simultaneous signal m, m/s, m/s2
pa atmospheric pressure Pa
Q quality factor −
qc cone resistance N/m2
R cross power spectrum; receiver
r cross-correlated signal; receiver signal; s · m2 ; ...;
radius, distance m
S spectrum of signal s; source
s time history m, m/s, m/s2
s∗ mean absolute deviation
T oscillation period s
t time s
u(t) displacement function m
Vp compression wave velocity m/s
Vs shear wave velocity m/s
x, y, z coordinates in space m
y mode shape (displacement) m
z derivative of the logarithmic s/m;
amplitude ratio; distance m
α attenuation coefficient 1/m

β porosity −
γ shear strain; unit weight −, %; N/m3
δ logarithmic decrement −
δn , δn+1 amplitudes of an oscillation m
∆t time shift s
∆x distance m
η viscosity Pa · s
λ wavelength m
ν ratio of poisson −
ξ fraction of critical damping −, %
ρ density kg/m3
σ normal stress Pa
σ00 mean effective stress Pa
FREQUENTLY USED SYMBOLS xiii
0
σ1,2,3 isotropic confining stress Pa
τ shear stress; time shift Pa; s
φ phase shift; angle of internal friction −; o
ω circular frequency 1/s
ω0 circular frequency at resonance 1/s
Subscripts
B related to the bottom sensor of the seismic cone
k related to bulk compression
kin kinematic
max maximum
min minimum
p related to p-waves
pot potential
R related to a receiver signal
r related to Rayleigh-waves
s related to s-waves; related to a sender signal
T related to the top sensor of the seismic cone
Superscripts
0
related to an effective value or an imaginary component
∗
related to an complex value
˙ 1st derivative after the time
¨ 2nd derivative after the time
Other symbols
ˆ related to an amplitude of a parameter
¯ related to a complex conjugate
xiv FREQUENTLY USED SYMBOLS
Part I
Introduction and
background
1
Chapter 1
Introduction
1.1 Purpose and scope

The evaluation of a number of civil engineering problems involve the transmis-
sion of waves through soil, such as seismic response under earthquake loading,
foundation response under dynamic loading and vibrations in buildings, in-
duced by various sources. Important sources are industrial activities (looms,
printing presses), building activities (pile driving, installation of sheet piles and
demolition of buildings), rail and road traffic.
Mechanical waves can be divided in body and surface waves. Body waves
can exist in an ideal full space or they travel in a region that is not affected
by a free surface. P- (primary, compression, longitudinal) waves and s- (secon-
dary, shear, transverse) waves are types of body waves. The particle motion
of p-waves is in propagation direction; the particle motion of s-waves is per-
pendicular to the direction of propagation. Surface waves may only exist at
the surface or the boundary, separating media of different properties. Rayleigh
(vertically polarized) and Love (horizontally polarized) waves are examples of
surface waves.
The prediction of elastic wave propagation from the source, through the
soil to the receiver can be performed by means of numerical models. Effects of
construction and renovation of traffic infrastructure or other vibration sources
can already be evaluated in the phase of planning. Unreasonable nuisances to
residents or damage to buildings can be avoided by design of a proper vibration
isolation system. Experimental validation has shown that the estimated soil
characteristics crucially affect the vibration prediction accuracy in the free field
and in buildings.
The most important characteristic parameters are the velocity of s- and p-
waves and the material damping ratios of these body waves. Material damping
in the soil represents energy dissipation caused by friction between solid partic-
les in the skeleton and by the relative motion between the soil skeleton and the
pore fluid. Material damping must be distinguished from geometrical or radial
3
4 CHAPTER 1. INTRODUCTION
damping. Since the velocity of p-waves is highly affected by the groundwater

table, the most efforts are spent to the determination of the s-wave velocity.
S-waves are transmitted in saturated soils by the soil fabric only. The dynamic
shear modulus, Gmax , can be calculated directly based on the s-wave velocity.
There are only a few techniques available to determine the damping ratio.
Only laboratory tests, as resonant column and cyclic shear tests, but no field
testing methods, can be considered as established techniques.
This work focuses on the determination of the damping ratio of shear waves
by means of field and laboratory tests. Therefore extended interpretation tech-
niques for the seismic cone penetration test (SCPT) and the bender element
test (BE) are studied and evaluated.
The SCPT is an extension of the cone penetration test (CPT). The use
of the CPT in the geotechnical engineering practice has increased sharply in
recent years. Therefore the CPT equipment is widely spread. Since most of
this equipment is also used in a SCPT the economical barriers for the transfer
of technique to practice are low. The BE methods are in the focus of interests
because bender elements can be mounted in various laboratory testing devices
and are able to generate s-waves with very low strain amplitudes similar to
those found in situ.
The research is embedded in the project ”Traffic induced vibrations in buil-
dings” initiated from the K.U. Leuven and Ghent University. Within the frame-
work of this project soil parameters at different sites are determined for use as
input parameters in a numerical model to predict the wave propagation. Five
test sites in Belgium are chosen for testing: Waremme, Lincent, Retie, Sint-
Katelijne-Waver and Ghent. The SCPT and the BE test are a substantial part
of this exploration to measure the shear wave velocity and the damping ratio.
The extensive testing campaigns offered the unique opportunity to evaluate
and improve the methods on different soil materials.
1.2 Outline of the thesis

An introduction in Chapter 2 is devoted to the fundamentals of the dynamic
soil properties: shear modulus and the attenuation parameters. This is followed
by an overview of state of the art methods to determine the shear modulus and
the damping ratio in Chapter 3. Laboratory testing techniques and in situ tests
are considered.
The Chapters 4 to 8 provide information on the testing sites in Retie, Lin-
cent, Waremme, Sint-Katelijne-Waver and Ghent. Apart from a general de-
scription of each site, the essential results of all performed tests are given,
including free torsion pendulum and resonant column tests. The findings of
the bender element resonant tests are the only exception. These are discussed
in Chapter 10.
Readers primarily interested in the BE- and SCPT testing methods, which
are closer investigated in this work, may at first skip the five chapters on the
testing sites to come back later to certain results, if necessary.
1.2. OUTLINE OF THE THESIS 5
Chapter 9 is dedicated to the SCPT. The selected test equipment and de-
veloped data acquisition tools are described. After a summary on methods to
determine the s-wave velocity, the chapter focuses on possibilities to measure
the damping ratio based on time recods gathered by a SCPT. The results of
an evaluation of the spectral ratio slope and the attenuation coefficient method
by means of a numerical simulation of the test are given. Finally the obtained
results on the five testing sites are summarized.
Chapter 10 deals with the methods based on the bender element test. Pro-
cedures to measure the arrival time of the s-wave are described at first. Then
methods to determine attenuation parameters are given. A resonant method is
applied on samples from two of the testing sites and the results are discussed.
Chapter 11 concludes the main findings of the thesis and gives suggestions
for further research.
The obtained insights in the accuracy of the testing techniques concerning
the G-modulus and the damping ratio are used to compare the methods under
consideration of their costs. The results of the comparison can be found in the
appendix and can be considered as a conclusion of this work.
6 CHAPTER 1. INTRODUCTION
Chapter 2
Dynamic soil properties
2.1 Dynamic shear modulus

A harmonic loading of a soil element as depicted in Figure 2.1 leads to a stress-
strain diagram as shown in Figure 2.2. It is a typical outcome of a cyclic
torsion or a cyclic triaxial test. A hysteresis loop accrues for each period of the
oscillation. The shear modulus is found as the slope of the line that connects
the point of origin and the inversion point between loading and unloading. This
modulus is also called secant modulus or equivalent modulus. It decreases with
an increase of load and shear amplitude. Therefore the stress-strain relationship
is principally non-linear. A schematic sketch of the shear modulus in function
of the shear strain is given in Figure 2.3.
The first loading curve, sometimes called the backbone curve, connects
the load inversion points of oscillation periods with different levels of shear
strain and has a hyperbolic shape. The slope in the origin point to this curve
corresponds to the dynamic shear modulus Gmax or G0 . It is also called tangent
modulus. Gmax is the shear modulus in the low strain range, usually assumed
at values below the linear elastic threshold strain of about γ = 5 · 10−3 %. It
is in general expected that Gmax does not change in the low-strain range.
Vibrations caused by seismic in situ tests, traffic, construction works, weak
earthquakes or even blastings usually have shear strain amplitudes below
5 · 10−3 %. This opens the opportunity to model the wave propagation with a
comparatively simple linear elastic constitutive model with Gmax as its essential
parameter.
Gmax in Pa can be calculated as:
Gmax = ρ Vs2 (2.1)
with the soil density (ρ) in kg/m3 and the shear wave velocity (Vs ) in m/s.
This essential relation is used to obtain Gmax by means of various seismic
geophysical tests providing values of Vs .
7
8 CHAPTER 2. DYNAMIC SOIL PROPERTIES
Figure 2.1: Cyclic loading: load his- Figure 2.2: Cyclic loading: hystere-
tory. sis loops.
G
Shear modulus, G
Damping ratio, D
Shear strain, γ
Figure 2.3: Shear modulus and damping ratio in function of shear strain.
2.1. DYNAMIC SHEAR MODULUS 9
Laboratory tests have shown that soil stiffness is influenced mainly by cyclic
strain amplitude, void ratio, mean effective stress, plasticity index, overconsoli-
dation ratio and number of loading cycles. The secant shear modulus decreases
with increasing shear strain amplitude. Gmax is, according to the definition,
strain independent. The parameters which can influence Gmax are summarized
in Table 2.1 taken from Dobry & Vucetic (1987). In function of the type of soil
some of these parameters might be irrelevant.
Increasing parameter Gmax

mean effective stress, σ00 increases with σ00
void ratio, e0 decreases with e0
geological age, tg increases with tg
cementation, c increases with c
overconsolidation ratio, OCR increases with OCR
plasticity index, Ip increases with Ip if OCR > 1;
stays approximately constant
if OCR = 1
Strain rate, γ̇ no effect for non-plastic soils;
increases with γ̇ for the case of
plastic materials
number of loading cycles, N decreases after N cycles of large
cyclic strain amplitudes,
but recovers with time in clays;
increases with N for sand
Table 2.1: Influence of different parameters on Gmax for normally to moderately

overconsolidated soils, Dobry & Vucetic (1987).
A large amount of laboratory test results, primarily from resonant column

tests, are available in literature from different authors. They are the basis of
empirical correlation functions developed in the past. One of the most known
formulas can be found in Hardin & Black (1969) and can be applied for clays
and sands. It considers Gmax as a function of void ratio (e0 ), mean effective
stress (σ00 = (σ10 + σ20 + σ30 )/3), overconsolidation ratio (OCR) and plasticity
index (Ip ):
Gmax = 625 Fe pa1−n σ00 n OCRk (2.2)
with the atmospheric pressure (pa ≈ 100kPa) in the same units as Gmax and
σ00 , an empirical exponent k related to the plasticity index Ip
Ip0.72
k≈ ≤ 0.5 (2.3)
50
the void ratio function Fe , varying somewhat in different studies, and a stress
exponent n. Hardin & Black (1969) introduced originally

(2.973 − e0 )2
Fe = 0.51 (2.4)
1 + e0
this has later been improved to cover a wider range of void ratios in Hardin
(1978):
1
Fe = (2.5)
0.3 + 0.7e20
Another expression for Fe is given by Jamiolkowski et al. (1991):
1
Fe = (2.6)
e1.3
0
The stress exponent is often taken as n=0.5 but can be calculated for indi-
vidual soils from the results of laboratory tests at different effective confining
pressures.
Other empirical relationships are proposed for different soil types. Some of
them can be found in Kramer (1996) together with correlation functions based
on parameters obtained by conventional in-situ tests as CPT (cone penetra-
tion test), SPT (standard penetration test), DMT (dilatometer test) and PMT
(pressuremeter test).
The shear modulus G at higher shear strain amplitudes can be obtained if
a hyperbolic non-linear constitutive model Duncan & Chang (1970) is used. It
shows the relation between G and Gmax :
Gmax
G= (2.7)
1 + γ/γr
with
τmax
γr = (2.8)
Gmax
τmax represents the maximum shear stress before failure occurs. γ is the shear
strain related to the calculated G. Details can be found for instance in Studer
& Koller (1997).
Ishibashi & Zhang (1993) describe the modulus reduction G/Gmax with γ
as a function of mean effective stress and plasticity index only:
G
= K(γ, Ip ) (σ00 ) m(γ,Ip )−m0 (2.9)
Gmax
with
"
m(γ, Ip ) − m0 = 0.272 1 . . .
( " 0.4 #)#
0.000556 1.3
. . . − tanh ln e−0.0145 Ip (2.10)
γ
2.2. ATTENUATION PARAMETERS OF SOILS 11
"
K(γ, Ip ) = 0.5 1 . . .
( " 0.492 #)#
0.000102 + n(Ip )
. . . + tanh ln (2.11)
γ


 0.0 for Ip = 0


 (sandy soils)
−6 1.404

3.37 · 10 I for 0 < Ip ≤ 15

p



(low plastic soils)

n(Ip ) = (2.12)

 7.0 · 10−7 Ip1.976 for 15 < Ip ≤ 70
(medium plastic soils)



−5 1.115

2.7 · 10 I for I p > 70



 p
(high plastic soils)

This empirical formulation is based on a large amount of data and covers a

wide range of materials from gravelly soils to moderately overconsolidated clays.
It will show its special usefulness in this research in relation with correlation
functions for the damping ratio from the same authors.
2.2 Attenuation parameters of soils

Energy is dissipated in soils and structures by several mechanisms, including
friction, heat generation and plastic yielding. For soils and structures the domi-
nating mechanisms are not understood sufficiently to allow them to be explicitly
modeled. As a result, the effects of various loss mechanisms are usually lumped
together and represented by some convenient damping mechanism.
The most commonly used mechanism for representing energy dissipation is
viscous damping. It can be illustrated by means of a viscous damped single
degree of freedom (SDOF) system as shown in Figure 2.4. This system is
subjected to a harmonic displacement u(t) governed by
u(t) = u0 sin ωt (2.13)
whereas ω is the excitation frequency. The reaction force F(t) is:
F (t) = k u(t) + c u̇(t) = k u0 sin ωt + c u0 ω cos ωt (2.14)
The energy dissipated during the oscillation can be obtained using the re-
lation du = u̇ dt as:
Z Z
ED = c u̇ du = c u̇ u̇ dt (2.15)
u(t)
c
F(t) Q(t)
m
Figure 2.4: Damped SDOF system subjected to an external displacement u(t).
After introduction of integration limits the dissipated energy in one cycle of

oscillation ED, vis , agreeing with the area inside the hysteresis loop Aloop, vis ,
can be expressed as:
Z T
Aloop, vis =E
ˆ D, vis = c u̇ u̇ dt = πu20 ωc (2.16)
0
T = 2π/ω is the period. At maximum displacement, the velocity is zero and

the strain energy ES stored in the spring of the SDOF is given by
1 2
AAOB =E
ˆ S= ku (2.17)
2 0
whereas AAOB represents the area between the points A, O, B in Figure 2.2.
The ratio of dissipated energy and strain energy ED, vis /ES gives:
ED, vis 2π c ω
= (2.18)
ES k
p
and by using the relations√for the natural frequency ω0 = k/m and the critical
damping cc = 2mω0 = 2 k m:
ED, vis c ω
= 4π = 4π ξ β (2.19)
ES c c ω0
The fraction of critical damping is abbreviated as c/cc = ξ and the frequency

ratio as ω/ω0 = β. This leads to a formulation for the ratio of critical damping:
c ED, vis Aloop, vis

ξ= = = (2.20)
cc 4π β ES 4π β AAOB
ξ is a constant for the viscous damped system. But the dissipated energy per
loading cycle ED, vis is proportional to the loading frequency ω as can be seen in
Equation (2.16). Since damping in soil is in general assumed to be frequency
independent, that means ED, vis is not a function of frequency, the viscous
damped system has to be adapted to meet this requirement. The adapted
system is called a system with hysteretic or rate independent damping.
The aim can be achieved by changing the frequency independent damping

coefficient c to a coefficient ceq , called equivalent damping coefficient, which is
inversely proportional to the loading frequency. ceq is defined as:
ηk
ceq = (2.21)
ω
η is the loss factor and independent of frequency. The fraction of critical
damping ξeq is then
√
ceq ηk η k η
ξeq = = √ = √ = (2.22)
cc 2ω mk 2ω m 2β
ξeq is a function of β and therefore also function of the loading frequency ω.
The loss factor η can be expressed as energy ratio. The dissipated energy
of the hysteretic system can be given as
ED, hys = πu20 ω ceq (2.23)
If Equation (2.17) is used one can write in analogy to Equation (2.19)

ED, hys ceq ω
= 4π (2.24)
ES c c ω0
and after introduction of Equation (2.21)
ED, hys
= 2π η (2.25)
ES
This leads to a loss factor of
ED, hys
η= (2.26)
2π ES
The afterwards used material damping ratio D can be derived from the loss
factor by
η
D= (2.27)
2
Therefore Equation (2.26) can be expressed as
ED, hys Aloop, hys
D= = (2.28)
4π ES 4π AAOB
Experiments show that some energy is dissipated even at very low strain
levels, so the damping ratio is never zero. Above the threshold strain, the
width of the hysteresis loop increases with increasing cyclic strain amplitude,
indicating an increasing damping ratio with increasing strain amplitude.
The concept of the equivalent damping coefficient ceq works well in the
frequency domain, but if the equation of motion should be used in the time
domain it cannot be applied.
In this case hysteretic damping can be approximated by an equivalent hys-

teretic model. This is a purely viscous system with a constant c matched to
a ceq at a certain frequency. Preferably the natural frequency of the system
is used (β = 1). In this way the amplification function of the viscous damped
system shows a reasonable agreement with the hysteretic system, at least in
the important range of the natural frequency.
Isotropic visco-hysteretic elastic material model The further treatment

of the wave attenuation needs a constitutive equation for a continuum. Basis is
an isotropic visco-hysteretic elastic material model as described in Molenkamp
& Smith (1980).
The relation between stress, strain and strain rate is given by
σij = 2Geij + 3Km δij + 2G0 ėij + 3K 0 m δij (2.29)
in which G is the shear modulus, K the bulk modulus, G0 the viscous shear
modulus, K0 the viscous bulk modulus, m the isotropic strain component, eij
the deviatoric strain tensor component and δij the Kronecker delta.
The equation can be divided into the isotropic stress component σm and
the deviatoric component sij
σij = sij + σm δij (2.30)
with
σm = 3Km + 3K 0 ˙m (2.31)
and
sij = 2Geij + 2G0 ėij (2.32)
For cyclic deformations the Equations (2.31) and (2.32) can be written as
σm = 3K ∗ ˆm ei(ωt+φm ) (2.33)
and
sij = 2G∗ êij ei(ωt+ξij ) (2.34)
in which φm is the phase of the isotropic stress components, ξij the tensor of
the phase of the deviatoric stress component and ˆ indicates the amplitude of
a parameter. G∗ and K∗ incorporate moduli and viscous moduli in complex
parameters
G∗ = G + iωG0 (2.35)
and
K ∗ = K + iωK 0 (2.36)
A material is considered as hysteretic when the dissipated energy per cycle

is independent of the frequency of loading, i.e. the damping ratio and the loss
factor are constant. Molenkamp & Smith (1980) express the dissipated energy
per period and the average elastic energy for cyclic deviatoric and volume-
tric deformations in the notation of the visco-elastic material model. If these
energies are introduced in an equation for the damping ratio D, analogical to
Equation (2.28), the following formulations are obtained
ωG0
Ds = (2.37)
2G
for the shear deformation and
ωK 0
Dk = (2.38)
2K
for dilatation.
Since D should be frequency independent, G’ and K’ have to be inversely
proportional to ω. This concept is already used in connection with the in-
troduction of the ceq in the SDOF system. G’ and K’ can be expressed by
converting of the Equations (2.37) and (2.38) as
2Ds G 2Dk K
G0 = K0 =
ω ω
with leads introduced in the Equations (2.35) and (2.36) to the complex moduli
G∗ = G + 2iDs G (2.39)
and
K ∗ = K + 2iDk K (2.40)
It will be useful in the methods for determination the damping ratio of

shear waves by means of the SCPT to express Ds also in terms of a complex
wavenumber k∗ . According to the elastic-viscoelastic correspondence principle
the solution of a harmonic boundary value problem in linear viscoelasticity
can be obtained from the solution of the corresponding elastic boundary value
problem. Using this principle and Equation (2.1) a complex value of the shear
wave velocity Vs∗ can be calculated by
s
G∗
Vs∗ = (2.41)
ρ
this is linked to the complex wave number by

ω
k∗ = (2.42)
Vs∗
The Equations (2.35), (2.37), (2.41) and (2.42) conclude in an expression

for k∗ = k + k0
k k0
Ds = (2.43)
k2 − k 02
which can be simplified for a small attenuation (k0 << k) and writing the at-
tenuation coefficient as α = k0 to
k0 α
Ds = = (2.44)
k k
If α is isolated the in Section 9.4.1 needed relation
ω Ds 2π f Ds
α = k Ds = = (2.45)
Vs Vs
is obtained.
Other measures of energy dissipation are the quality factor Q and the spe-
cific damping capacity ψ related to the damping ratio D and the loss factor η
by:
1
Q= , ψ = 2π D, η = 2D (2.46)
2D
In infinite, isotropic visco-elastic materials the relationship between the
damping ratio for p-waves Dp , s-waves Ds and for bulk compression Dk is
given by Fratta & Santamarina (1996) and Winkler & Nur (1979) based on an
analytical solution of the wave equation:
Dk (1 + ν) = 3 Dp (1 − ν) − 2 Ds (1 − 2ν) (2.47)
In most problems, it is assumed that there is no dissipation of energy in pure

compressive or dilational processes and therefore Dk = 0. With this hypothesis
Udı́as (1999) obtains the following relation
Dp Qs 2 1 − 2ν
= = (2.48)
Ds Qp 3 1−ν
with ν the ratio of Poisson.

Under consideration of the dependence of p-wave velocity Vp and s-wave
velocity Vs
r
Vp 1−ν
= 2· (2.49)
Vs 1 − 2ν
Equation (2.48) can also be written in terms of the wave velocities
2
Dp Qs 4 Vs
= = (2.50)
Ds Qp 3 Vp
The increase of the damping ratio with the shear strain can be estimated
under the same assumptions used for Equation (2.8):
D γ/γr
= (2.51)
Dmax 1 + γ/γr
This relationship is given schematically in Figure 2.3.
The influence factors on the damping ratio for normally consolidated and
moderately overconsolidated soils are summarized in Table 2.2. Especially
the dependence on the plasticity characteristics should be emphasized. The
damping ratios of highly plastic soils are lower than those of low plasticity soils
at the same cyclic strain amplitude.
Increasing parameter Damping ratio, D

mean effective stress, σ00 decreases with σ00 ;
effect decreases with increasing Ip
void ratio, e0 decreases with e0
geological age, tg decreases with tg
cementation, c may decrease with c
overconsolidation ratio, OCR not affected
plasticity index, Ip decreases with Ip
cyclic strain, γ increases with γ
Strain rate, γ̇ stays constant or
may increase with γ̇
number of loading cycles, N not significant for
moderate γ and N
Table 2.2: Influence of different parameters on D for normally to moderately

overconsolidated soils, Dobry & Vucetic (1987).
Ishibashi & Zhang (1993) developed a closed expression for the damping
ratio based on the modulus reduction G/Gmax and the plasticity index. It is
valid for non-plastic to highly plastic soils if the degree of overconsolidation
remains moderate.
1.3
(
0.333 1 + e−0.0145Ip
G
2
D = 0.586 ...
2 Gmax

G
. . . − 1.547 +1 (2.52)
Gmax
If Equation (2.9) is used to calculate the modulus reduction, the damping ratio
can be described as a function of mean effective stress, shear strain amplitude
and plasticity index.
Chapter 3
Methods to determine G
and D
This chapter gives an overview of the standard soil tests for the determination
of the dynamic parameters with emphasis on the resonant column test and the
free torsion pendulum test since both tests are performed in this research to
obtain reference values for Gmax and D on two of the testing sites (Waremme
and Sint-Katelijne-Waver).
It is useful to divide the testing procedures for dynamic soil parameters in
tests working under low strain conditions so the deformations can be assumed
as elastic and tests under high strain conditions with non negligible plastic
deformations. Some of the high strain tests are able to observe the dynamic
soil behavior to the range of failure.
An overview of the relevant shear strain amplitudes in different engineering
applications and test methods is given in Figure 3.1.
3.1 Laboratory tests

A limited number of laboratory tests are performed in the range of elastic
deformations. They include resonant column test, piezoelectric bender element
test, piezoelectric compression element test and ultrasonic test. The free torsion
pendulum test is also able to reach the range of elastic soil behavior. It can be
seen as a special kind of resonant column test.
The cyclic direct or simple shear test, cyclic torsional shear test and cyclic
triaxial test belong to the group of high strain tests. They are mainly developed
to study the liquefaction behavior under earthquake loading.
The analysis of the test results of the three cyclic methods is based on
the interpretation of the measured stress-strain hysteresis loops as described
in Section 2.2. Whereas direct or simple shear and torsional shear tests give,
due to the applied shear loading, a G-modulus and damping ratio, the most
19
20 CHAPTER 3. METHODS TO DETERMINE G AND D
Shear Strain Amplitude γ [%]

10−4 10−3 10−2 10−1 1 10
Sort of Problem
seismic surv ey
v ibrations caused by traffic,

blastings and weak earthquakes
machine foundations
strong earthquakes,
farfield of explosions
nearfield of explosions
refraction and reflection seismic

In situ Tests
cross−hole, up−hole
and down−hole seismic
free oscillation tests,
enforced oscillations
dynamically loaded plates,
shear tests
ultrasonic tests
Laboratory Tests
resonant column tests

(cylindrical specimen)
resonant column
tests (hollow−cylinder)
cyclic torsional shear
tests (hollow−cylinder)
cyclic triaxial test
shaking table
10−4 10−3 10−2 10−1 1 10
Figure 3.1: Overview of possible shear strain amplitudes, Studer & Koller
(1997).
3.1. LABORATORY TESTS 21
common test, the cyclic triaxial test, at first provides values for the dynamic
elasticity modulus and only indirectly the G-modulus.
Besides the direct testing of soil specimens, also small-scale physical models
can be subjected to a cyclic loading. These tests are performed on shaking
tables or, for models whose stress dependency has to match that of the full-
scale problem, more commonly in centrifuges.
Figure 3.2: Cyclic simple shear test Figure 3.3: Cyclic triaxial test de-
device from Airey & Wood (1987). vice from Kramer (1996).
3.1.1 Piezoelectric bender and compression element tests

Transmitter and receiver elements can be placed at each end of a specimen as
can be seen in Figure 3.4. The elements are made of piezoelectric materials
exhibiting changes in dimensions when subjected to a voltage across their faces
and producing a voltage across their faces when distorted. An electrical pulse
applied to the transmitter causes it to deform rapidly and produce a stress
wave that travels through the specimen toward the receiver. When the stress
wave reaches the receiver, it generates a voltage pulse that is measured. The
wave speed is calculated from the arrival time and the known distance between
transmitter and receiver.
Dependent on the internal structure of the piezoelectric materials p- or
s-waves can be generated and registered. Elements generating s-waves are
called bender elements because of their shape of movement and penetrate a
few millimeters into the sample. A schematic view of such an element is given
in Figure 3.5. Elements to generate p-waves are compression elements or if
they are driven on very high frequencies, ultrasonic elements. Compression
elements usually do not penetrate into the specimen.
Instead of bender elements shear plates are also used, which transfer there
moving energy by friction without any penetration of the soil sample. Shear
plates are more effective at high confining stresses and with very coarse soils
(Brignoli et al. (1996)). However, they are bigger, require much larger driving
voltages and at low confining stresses are less efficient than benders.
Lings & Greening (2001) explain how a modification to a standard element
design can result in a single hybrid element, termed a bender/extender, capable
Top cap
(receiv er)
Bender
elements Sample
Bottom cap
(sender)
Figure 3.4: Bender elements in- Figure 3.5: Bender element, posi-
stalled in a triaxial cell. tive voltage causes the element to
bend one way, negative voltage cau-
ses it to bend the other, Kramer
(1996).
of transmitting and receiving both s- and p-waves. Such elements are already
commercially available.
The dynamic shear modulus Gmax and the dynamic elasticity modulus Emax
can be calculated out of the s- and p-wave velocity using Equation (2.1) and
(1 + ν)(1 − 2ν)
Emax = Vp2 ρ (3.1)
1−ν
with the p-wave velocity (Vp ) and Poisson’s ratio (ν).

Because the specimens are not disturbed during the tests the piezoelectric
elements are incorporated in various soil testing devices, such as conventional
triaxial devices, oedometers and direct or simple shear devices.
A Chapter 10 is devoted to the bender element technique and more details
can be found there.
3.1.2 Cyclic triaxial tests

The test device consists of the standard triaxial testing equipment extended
with a cyclic axial loading unit. In some cases, the cell pressure is also app-
lied cyclically. Isotropic or anisotropic initial stress conditions are possible. A
sketch of the device in given in Figure 3.3. Bedding errors and system com-
pliance effects generally limit the measurements to shear strains greater than
10−2 %, although local strain devices can produce accurate measurements at
strain levels as small as 10−4 %.
3.1.3 Cyclic simple shear tests

The cyclic simple shear test device, as shown in Figure 3.2 is most commonly
used for liquefaction testing. A short cylindrical specimen is restrained against
lateral expansion. By applying cyclic horizontal shear stresses to the top or
bottom of the specimen, the test specimen is deformed in much the same way
as an element of soil subjected to vertically propagating s-waves. Simple shear
devices that control the vertical and horizontal stresses independently are able
to impose stresses other than those corresponding to K0 conditions.
3.1.4 Cyclic torsional shear tests

The cyclic torsional shear test works with a torsional loading of a cylindrical
soil specimen. The equipment looks like a conventional triaxial device except
for the added cyclic loading apparatus. Isotropic and anisotropic initial stresses
are possible. The test is most commonly used to measure stiffness and damping
characteristics over a wide range of strain levels. Torsional testing of soil speci-
mens produce shear strains that range from zero along the axis of the specimen
to a maximum value at the outer edge. To increase the radial uniformity of
shear strains, testing devices for hollow cylinder specimen are used.
3.1.5 Resonant column test

The resonant column test is a well-known technique to determine the dynamic
shear modulus, dynamic elasticity modulus and damping ratio. In a triaxial cell
a soil sample is installed and excited torsionally or axially at its top end. The
excitation is most commonly harmonic, in a range between 30 and 300 Hz, but
also random white noise (Cascante & Santamarina (1997)) or pulses (Nakagawa
et al. (1996)) have been used. There are devices for cylindrical samples and
for hollow-cylindrical samples available, the latter minimize the variation of
shear strain amplitudes across the sample in the case of torsional excitation.
With a built in accelerometer the acceleration at the top of the sample can be
measured.
After the resonant column sample has been prepared and consolidated,
cyclic loading is begun. The loading frequency starts from a low value, gra-
dually increases until the response is locally maximized and the phase shift
between driving signal and measured acceleration signal is equal to π. The
lowest frequency with a local maximum in the response function is assigned to
the fundamental frequency of the sample. This frequency is a function of the
soil stiffness, the sample geometry and characteristics of the apparatus. Based
on the assumed system with a single degree of freedom as shown in Figure 3.6
the relevant formulas for torsional and axial loading, taking into account the
additional mass of the top cap and the moving parts of the driving unit, are
given in Equation (3.2) and Equation (3.3). They can be found for instance in
Studer & Koller (1997).
Top cap and

mov ing parts J 0, m 0
of the oscillator
J, m h
Sample
Figure 3.6: SDOF system assumed for the behavior of sample and apparatus.
J ωn h ωn h
= tan (3.2)
J0 Vs Vs
m ωn h ωn h
= tan (3.3)
m0 Vl Vl
J, J0 are the mass polar moments of inertia of respectively the sample and
the top cap; m, m0 are the mass of respectively the sample and the top cap; h
is the sample height and ωn the circular frequency of the system at resonance.
The G-modulus can be calculated out of Vs by means of Equation (2.1). The
constrained elasticity modulus follows from the longitudinal wave velocity Vl :
E = ρ Vl2 (3.4)
Alternatively to a harmonic sinusoidal excitation, the fundamental natural fre-

quency can also be obtained by a single pulse excitation. The frequency of the
developing free oscillation corresponds to the natural frequency of the system.
The damping ratio can be determined by the logarithmic decrement method
as shown in the description of the free torsion pendulum technique later.
The measured response curve, in the case of harmonic excitation, can be
analyzed using the half-power bandwidth method or the circle-fit method. Both
system identification techniques will be presented later on to obtain the dam-
ping ratio by means of bender elements. Finally, if the applied dynamic force
is quantitatively known by a careful calibration of the apparatus, the amplifi-
cation factor at resonance can be used to obtain the damping ratio as well.
The devices used in this research are of the Drnevich type using a Hardin
oscillator to apply a torque to the top of the sample. A dynamic axial loading
is not possible in such a device. Both isotropic and anisotropic stress stages
can be imposed.
3.1.6 Free torsion pendulum test

The free torsion pendulum test, sometimes also called Zeevaert test after its in-
ventor, is performed on a sample from the site in Waremme and Sint-Katelijne-
Waver. A schematic sketch of the device at the soil mechanics laboratory at
Ghent University is given in Figure 3.9. For earlier publications on the device
can be referred to Storrer et al. (1986) and Van Impe (1977).
Hanger
Confinement Dial gauge
pressure
Air
Hardin−
oscillator
Oil
Cell
Sample
Water
Figure 3.7: Resonant column test Figure 3.8: Resonant column test
device with Hardin oscillator of the device of the Stokoe type, Kramer
Drnevich type. (1996).
Air pressure chamber

Horizontal arm
Counter weights
Contactless
Sealing bus Torsion shaft
Confinement displacement sensor
pressure
Air
Cell
Sample
Water
Figure 3.9: Test set-up of the free torsion pendulum test at Ghent University.
A soil sample with a diameter of 100 mm and a height of approximately

200 mm, covered by a rubber membrane is installed under drained conditions
in a slightly adapted conventional triaxial cell. In this cell the soil sample,
caught by an upper and a lower cap with lamellas, is subjected to a confining
water pressure. The sample is allowed to consolidate freely under the app-
lied hydrostatic pressure. Due to the lamellas, no slip between the caps and
the sample can occur when afterwards trough the upper cap a torsional mo-
ment is applied on the sample. After consolidation, a heavy weight horizontal
beam is installed symmetrically on the axis of the sample. Due to the special
construction of the torsion shaft and also due to the special conceived upper
cap, it remains possible, after a free consolidation, to connect the beam to the
soil sample without any slipping and to eliminate a preliminary distortion of
the sample by the torsion shaft. The whole weight of the horizontal beam is
balanced through a thin steel wire, in order not to apply any supplementary
vertical load to the sample. For testing, the horizontal very stiff beam is given
a small impulse by a lightweight hammer, allowing the system to vibrate freely.
The damped oscillating vibration of the system is measured by a contactless
proximity transducer at one end of the beam. Typical amplitudes are less than
2 mm. As an example a time record is given in Figure 3.10.
0.4
Beam displacement [mm]
0.2 δn
δn+1
−0.2
−0.4
0 1 2 3 4 5
Time [s]
Figure 3.10: Displacement record free torsion pendulum test.
In order to avoid an important damping caused by the apparatus itself,

sealings on the basis of an air-pressure cushion are used and disturbances by
the measurement device are prevented by the contactless displacement sensor.
Therefore it becomes possible to neglect the influences of the apparatus itself
on the measured damped oscillations of the soil sample.
Assuming elastic deformations of the soil sample during such damped free
oscillating vibration, the dynamic shear modulus G can be derived by means
of the following expression:
2h · JS
G = ωS 2 · (3.5)
πa4
In this equation ωS is the undamped circular frequency of the equipment inclu-

ding the soil sample, h the height of the sample, a the radius of the soil sample
and JS the polar moment of inertia of all oscillating parts of the apparatus
(Ja ) and of the soil sample (Jp ). In the framework of this research Ja has been
carefully recalculated using the mass and the dimensions of all oscillating parts,
as shown in Figure 3.11. A value of Ja = 3.26158 kgm2 has been found for the
device in Ghent. The undamped natural circular frequency ωS can be calcu-
lated using the damped frequency ωD , obtained out of the measured vibration
period, and the fraction of critical damping ξ:
ωD
ωS = p (3.6)
1 − ξ2
ξ is calculated with the help of the logarithmic decrement method:
1 δn
ξ= log (3.7)
2π δn+1
with δn and δn+1 successive amplitudes of the oscillation.
ξ of the viscous system has to be related to the damping ratio D of the
hysteretic system. This can be done by comparing Equation (2.20) with Equa-
tion (2.28) assuming ED, vis = ED, hys . It is obvious that for the resonance case,
β = 1, the damping ratio D and the fraction of critical damping ξ are equal,
that means:
η
D = = ξ for β = 1 (3.8)
2
The level of shear strain can be estimated by:
2 δn,n+1 · a
γ= (3.9)
3 r·h
δn,n+1 is the mean oscillation amplitude between the two amplitudes δn and
δn+1 used to calculate the damping ratio, r is the distance between the mea-
suring point on the beam and the center of rotation. The factor 2/3 indicates
that strain is calculated for a cylindrical face on 2/3 of the sample radius.
The initial shear strain applied in this test is, in dependence of the force
of the hammer blow, in general higher than 10−3 %. Typical values for small
hammer blows are 10−2 %. However, the amplitudes decrease after a couple
of free oscillations below the linear elastic limit. A MATLAB algoritmn was
developed to pick all usable peaks of the recorded damped beam oscillation,
instead of using only the first peaks. In this way the change of damping ratio
and G-modulus could be computed during the whole decay process. Because
the strain amplitude and frequency change from cycle to cycle, G-modulus
and damping ratio can be plotted versus shear strain. The resulting curves
of several hammer blows, usually 20, are averaged to improve the accuracy of
the obtained values. The G-modulus and the damping ratio could finally be
obtained in a range between 10−4 % and 10−1 % of strain. Because of the
2·r
Horizontal arm
JS=J p+J a Ja
Jp
Sample
h
h ≈ 20 cm
2a ≈ 10 cm
2r = 99.7 cm
d=2a
Figure 3.11: Oscillating parts of the apparatus included in the calculation of

the polar moment of inertia JS .
increasing importance of noise in the measured signals at low amplitudes, the

reliability of the test decreases in the neighbourhood of the lower border of the
shear strain. The natural frequency of the sample-apparatus system is between
3 and 6 Hz for the tested soils from the sites in Waremme and Sint-Katelijne-
Waver.
3.2 In situ tests

The in situ tests can also be divided in tests belonging to the small-strain
range and others belonging to the high strain range. The various seismic tests,
using an artificial vibration source and vibration sensors are classified into the
first category. To the second category belong the conventional static tests like
cone and standard penetration test, pressuremeter and dilatometer tests. They
provide an indirect way to obtain dynamic moduli by means of correlation func-
tions. Dynamically loaded steel plates open the possibility to gather stiffness
information of the first decimeters close to the surface in the low- and in the
high-strain range.
The seismic tests focus in general on the determination of the velocity of
p- and s-waves. If the generated wave is measured at several sensors also a
calculation of the damping ratio is possible. Especially for the cross-hole test
attempts are known from different authors.
3.2.1 Seismic reflection test

The seismic reflection test measures wave propagation velocities and thickness
of superficial layers. The method follows the principle of echo-sounding and
radar. The test is performed by producing an impulsive disturbance at the
source, S, and measuring the arrival time at the receiver, R, located at a certain
distance from the source as shown in Figure 3.12. Dependent on the used source
3.2. IN SITU TESTS 29
the test can analyze p- and s-waves. However, because the generation of high
energy s-waves is difficult, the separation of the s-wave from the first arriving
p-waves might fail.
S R
direct wave
I H
2 ic
v1
v2
II
Figure 3.12: Ray paths in the seismic reflection test.
Some of the wave energy will follow the direct path from source to receiver.
Another part will travel downwards until it is reflected at the boundary of an
underlaying layer. The wave velocity of the superficial layer can be calculated
from the arrival time of the direct wave.
Because the angle of incidence at the layer interface has to be equal to the
reflexion angle and using the wave velocity already known, the thickness of the
superficial layer can be calculated.
If two or more receivers are used, a possible inclination of the layer interface
can be theoretically estimated. However, the practical realization fails in the
most cases. The properties of deeper layers may be evaluated using reflections
from deeper interfaces.
The method is limited to situations where the arrival times of the direct
and the reflected wave are sufficiently different. This means for instance that
the method is especially confident for deep layers and less for shallow layers. If
used, also arrival times from waves reflected at several layer interfaces have to
be distinguishable.
Additional information on the reflection test can be found in Kramer (1996).
3.2.2 Seismic refraction test

The seismic refraction test involves the measurement of travel times of p- or
s-waves from an impulse source to a linear array of receiver points along the
ground surface at different distances from the source. The distances between
the receivers are remarkable larger than chosen in the refection test. The
refraction tests uses only the arrival time of the first wave component regardless
of its travel path. Therefore problems to distinguish between wave components

cannot appear. Figure 3.13 shows the used ray paths in the seismic refraction
method.
If it is assumed that the site consists of a two-layered elastic half-space one
part of the wave energy travels directly from the source to the receiver array,
which can be used to calculate the wave speed of the superficial layer. Other
parts travel downward toward the boundary between layer 1 and layer 2. At the
boundary, these rays are reflected and refracted. The direction of the refracted
ray is determined by Snell’s law given in Equation (3.10).
sin i1 sin i2
= (3.10)
v1 v2
At the critical angle ic , that means short before a total reflection appears,
the refracted ray travels in layer two horizontally, parallel to the boundary.
This ray will send continuously parts of its energy back to layer one. At the
passage of the layer boundary refraction reoccurs and a head wave traveling
towards the surface develops. The refraction angle in layer one is the same
as the critical angle of incidence ic . Because the wave velocity in layer two is
higher than in layer one the head wave arrives from a certain distance from the
source xc on the surface before the component taking the shorter direct path
through layer one with the lower wave speed.
The wave velocity in layer two, v2 , can be calculated based on the velocity
in the superficial layer, v1 , and the arrival time of the head wave at at least
two distances from the source or graphically from the slope in the time arrival
diagram as seen in Figure 3.14. Because the length and the shape of the travel
path of the refracted wave is known if v1 and v2 are obtained, the thickness H
of the superficial layer might be calculated.
The method is also applicable for inclined layer interfaces and multi-layered
stratifications, if the wave propagation velocity increases with the layer depth.
Closer details for the practical application can be found in Kramer (1996)
or in the geophysical literature for instance in Udı́as (1999).
3.2.3 Spectral analysis of surface waves (SASW)

The SASW-technique uses the characteristics of Rayleigh waves to obtain the
stratification of a site. Rayleigh waves travel, as surface waves, in the region
close to the soil-air interface. Due to the fact that the penetration depth
of the Rayleigh waves into the ground is approximately one wavelength, the
thickness of the layer package influencing the speed of the wave changes with
the wavelength. This leads to a wave velocity that depends on the wavelength
respectively the frequency. Such behavior is also called dispersion.
In most cases a drop weight or a hammer is used to generate Rayleigh waves.
If a harmonic source is applied the technique is called continuous surface wave
method (CSW). The wave arrival in at least two points at some distance from
the source is recorded with geophones or accelerometers as shown in Figure 3.15.
x
t
S R
head wave
direct wave
v1 v1 1/v2
I head wave
H
direct wave
ic ic (refracted wav e)
1/v1
v2>v 1 x
xc
II
Figure 3.13: Ray paths for the seismic re- Figure 3.14: Time arrival
fraction test. diagram as result of a seis-
mic refraction test.
A cross power spectrum between the two signals is calculated. The unwrapped
phase of this spectrum is used to calculate an experimental dispersion curve of
the Rayleigh wave velocity.
S R1 R2
d1 d2
Figure 3.15: Typical configuration of source and receivers in a SASW test.
Identification of the thickness and shear wave velocity of subsurface layers

involves the iterative matching of a theoretical dispersion curve to the expe-
rimental dispersion curve. A solution for a series of uniform elastic layers of
infinite horizontal extent is used to predict the theoretical dispersion curve.
Initial estimates of the thickness and the shear wave velocity of each layer are
then adjusted in an inversion procedure until the values that produce the best
fit to the experimental dispersion curve are found.
More details can be obtained from Nazarian & Desai (1993) and Yuan & Na-
zarian (1993). Lai (1998) and Lai et al. (2002) describe an advanced inversion
technique to determine simultaneously s-wave velocity and attenuation.
3.2.4 Seismic cross-hole test

The seismic cross-hole test measures the p- and s-wave velocities between
boreholes. At least two boreholes are necessary. The first for installing a
seismic source. This might be a mechanical or an explosive source. If the focus
is on the survey of the s-wave velocity the preference is on a mechanical source
able to produce s-wave impulses with reversible polarity.
In the second hole a receiver is installed at the same depth as the source in
the first borehole. The measurement is triggered at the source and the arrival
time of s- respectively p-wave is obtained by visual interpretation of the signal
from the receiver borehole. By testing at various depths, a velocity profile can
be drawn. A sketch of the test set-up is given in Figure 3.16.
Because the trigger time measurement is potentially inaccurate it is de-
sirable to use more than two boreholes, that means more than one receiver
point. The wave velocity is then calculated from the difference in the arrival
times at the receiver holes. This has the additional advantage that instead of
a visual interpretation, cross correlation can be used. Typical distances bet-
ween the boreholes are 5 to 12 m for layered soils and up to 30 m for nearly
homogeneous sites.
Mok et al. (1988) describe the application of the attenuation coefficient
method for the determination of the damping ratio based on a cross-hole test.
3.2.5 Seismic down-hole and up-hole test

The source in the down-hole test is located at the surface close to a borehole
with an installed receiver at a certain depth. The generated waves travels nearly
vertically from source to receiver. In the up-hole test the source is situated in
the borehole and the measurement is done at the surface. The set-up is shown
in Figure 3.16.
The down-hole test is more commonly used than the up-hole test because
it is more convenient to place and adjust a seismic source at the surface than
in a borehole.
The analysis of the arrival times is done as in the cross-hole test by visual
interpretation or if more than one receiver at different distances from the source
are used also cross correlation can be applied.
R S
S R1 R2 S R
a) b) c)
Figure 3.16: a) Seismic cross-hole test, b) seismic up-hole test, c) seismic down-
hole test.
3.2.6 Seismic cone penetration test

The seismic cone penetration test (SCPT) can be seen as a special version of
a down-hole test with the receivers (geophones or accelerometers) installed in
the tip of a cone pushed into the ground by a conventional cone penetration
equipment (CPT truck, Figures 3.17 and 3.18). Since no borehole is necessary
the test is much less expensive than a down-hole test.
Figure 3.17: SCPT in Ghent. Figure 3.18: SCPT in Retie with

automotive remote-controlled track
vehicle.
P- and s-wave sources are placed at the surface beside the penetration point
of the cone. The sources consist of steel beams or plates which are hit by a
hammer horizontally (s-wave source) or vertically (p-wave source). The cone
with the receivers is pushed stepwise into the ground. Usual intervals are 0.5
or 1.0 m. At each step the source generates a seismic pulse recorded by the
cone receivers. The determination of the p- or s-wave arrival can be performed
visually. If two receivers in a certain distance are installed in a cone (dual cone)
the travel time between these receivers can be calculated by cross correlation.
The travel time leads directly to the wave velocity using the direct wave travel
path from the source to the cone.
More details are discussed later when a method to obtain damping ratio
out of SCPT data is presented in Chapter 9.
3.2.7 Geotomography
Tomography is a method to obtain a two-dimensional image of a site (Johnson
et al. (1978), Lytle (1978)). Using multiple receivers and sources, a large matrix
of source-receiver travel times can be measured and compared with predicti-
ons of a ray-tracing model. The number, position and inclination of material
boundaries are adjusted until the computed travel-time matrix matches the
observed matrix. The distribution of the elastic parameters can be obtained
even for sites with difficult stratification.
3.2.8 High-strain tests

Various in-situ tests working in the high-strain range are in use. They provide
soil stiffness parameters in this strain range either directly like for instance
dilatometer test (DMT) and pressuremeter test (PMT) or by means of corre-
lation like cone penetration test (CPT) and standard penetration test (SPT).
Furthermore also correlations to the parameters in the low strain range are
developed.
The elasticity modulus of superficial layers can also be obtained by measu-
ring the settlement of plates loaded statically or dynamically. Dependent on
the type of dynamic loading device the high strain range can be covered as well
as the low strain range. The falling weight device is an example in the low
strain range. The ”water cannon” developed by the ETH in Zurich (Studer &
Koller (1997)) works for instance in the high-strain range.
In the case of cohesive soils and rocks dynamic stiffness parameters are
obtained by means of a free oscillation test of a laterally free part after releasing
an applied lateral force. Dependent on the initial deformation the values are
valid for low- or high-strain conditions.
Part II
Characterization of the
testing sites
35
Chapter 4
Test site Retie
4.1 Introduction
The site is situated on a field next to the property of the architect P. Mer-
tens, Molsebaan 43 in Retie and around the house of the architect itself. The
following tests were performed or previous data were available on the site in
Retie:
Date Available In Situ Tests Abbr. Depth

[m]
01/04/1936 Boring, creamery St. Martin, 68.0
Flemish Subsoil Database
01/08/1977 Boring, Lageweg 19, 186.0
Flemish Subsoil Database
28/01/2000 3 CPT performed by CPT1 7.4
Geologica NV, Bertem CPT2 3.2
CPT3 9.0
08/11/2000 Boring B1 18.0
3 SASW set-ups (two from KUL SASW KUL
and one from UGent, concluding SASW UGent
in an inversion calculation from
KUL and one from UGent)
13/12/2000 2 SCPT with a single geophone SCPT1 10.8
cone SCPT2 12.3
cone SCPT4 12.5
12/05/2003 1 SCPT with a dual accelerometer SCPT5 12.3
cone
Table 4.1: Overview of the available in situ tests.
37
38 CHAPTER 4. TEST SITE RETIE
The position of each boring or penetration test can be found in Figure 4.1.
A detailed description of the testing procedure of the SASW tests and the
SCPT’s in combination with the visual interpretation of the recorded signals
can be found in the report Areias & Haegeman (2001). This report gives all
numerical values of wave velocities from SCPT1 to SCPT4 and the results of
inversion approaches with different numbers of layers on the SASW test of K.U.
Leuven.
4.2 Borings and undisturbed sampling

A drilling was performed until a depth of 18 m. The soil consists of sand and fine
sand over the whole drilling. The top tertiary layer is sand from the formation
of Mol. Data of deep drillings until 68 m in close proximity to the testing site
and even deeper borings from the region of Retie are available from the Flemish
Subsoil Data Base. They confirm that the sand reaches from the surface to a
depth of at least 186 m. The stratification found by the performed boring B1
is given in Table 4.2. The soil description is not detailed enough to distinguish
between the shallow layer of quaternary deposits and the deeper tertiary layers.
The borderline is estimated at a depth of 5 m, considering the results of the
CPT, the geological map and other drillings in the neighborhood. The profile
of a drilling in Retie, Lageweg 19 is described in Table 4.3. It includes the
geological stratification of the tertiary deposits to a depth of 190 m. It can be
used to estimate the soil structure at Molsebaan 43 for great depths.
From To Color Main Component Admixtures

[m] [m]
0.00 0.50 dark brown topsoil -
0.50 1.50 light brown sand silt
1.50 2.50 beige sand -
2.50 3.50 beige fine sand silt
3.50 4.00 beige fine sand -
4.00 5.00 beige fine sand gravel
5.00 7.00 beige, green sand -
7.00 9.00 beige, green fine sand silt
9.00 14.00 beige, green sand silt
14.00 15.00 green sand -
15.00 16.00 green sand silt
16.00 17.00 green silt sand
17.00 18.00 green sand -
Lithographic Stratification: 0 to 5 m quaternary deposits, 5 to 18 m for-
mation of Mol.
Table 4.2: Results of boring B1 at the Retie site.

4.2. BORINGS AND UNDISTURBED SAMPLING 39
Figure 4.1: Site location plan, Retie.

From To Color Main Admixtures

[m] [m] Component
0.00 6.00 yellow fine sand quartz
6.00 10.00 yellow fine sand partly lignite
10.00 34.00 white fine sand quartz, partly lignite
34.00 38.00 green fine sand clay, partly much glauconite
38.00 46.00 gray, white sand much clay, glauconite, partly
quartz
46.00 58.00 gray, white unknown glauconite, partly clay
58.00 174.00 gray, green fine sand glauconite
174.00 182.00 gray, green fine sand glauconite, lime
182.00 186.00 black fine sand much glauconite
Lithographic Stratification: 0 to 6 m quaternary deposits, 6 to 26 m for-
mation of Mol, 26 to 46 m formation of Kasterlee, 46 to 102 m formation
of Diest, 102 to 158 m member of Dessel (formation of Diest), 158 to 190 m
formation of Berchem.
Table 4.3: Results of boring Lageweg 19.
However due to the non plastic behavior of the material, undisturbed samp-
ling was impossible. As a consequence laboratory tests on undisturbed material
could not be performed and the density of the material could not be obtained.
The physical characteristics of Mol sand have been widely studied through
former static and dynamic tests at Ghent University and the Flemish Geotech-
nical Institute. Some results are given in the following section.
Sand of Mol
The sand is geologically referred to a Tertiary-Pliocene deposit at Mol in the
north-east of Belgium. Mol sands are nearly pure quartz sands. A typical
composition is 96 % of quartz mineral and 4 % of mica and traces of other
minerals. The physical characteristics and the curve of grain size distribution
are given in Table 4.4, Table 4.5 and Figure 4.2. The data were collected by
Yoon (1991) and can be used as reference for Mol sand.
The majority of the grains of Mol sand falls in the fraction of fine sand.
The given uniformity coefficient and the degree of curvature classify the sand
as poorly graded (SP) following ASTM D-2487. The maximal and minimal
possible void ratio found by laboratory compaction experiments allows to cal-
culate a range of the dry and saturated unit weight. The dry unit weight is
between 13.55 and 16.39 kN/m3 . The saturated value reaches from 18.25 to
20.02 kN/m3 and can be taken as an assumption of the unit weight in situ
below the groundwater table. The density above the groundwater can not be
given because of the missing value of the saturation degree. But it has to be
situated between the range for the dry and the saturated density.
4.3. CONE PENETRATION TEST (CPT) 41
Fraction V − Gravel IV − coarse to medium Sand III − fine Sand II − Silt I − Clay
0
10
20
Sieve residue [%]
30
40
50
60
70
80
90
100
10 2 1 0.2 0.1 0.06 0.01 0.002 0.001
Grain size [mm]
Figure 4.2: Grain size distribution of Mol sand.
Fraction Grain diameter Classification Mass

[mm] [%]
I+II <0.06 Clay to Silt 0.6
III 0.06-0.2 fine Sand 63.4
IV 0.2-2.0 medium to coarse Sand 36.0
V >2.0 Gravel and Stones 0.0
Table 4.4: Grain distribution of Mol Sand.
4.3 Cone penetration test (CPT)
Three CPT’s have been performed by Geologica N.V. around the building
no. 43. A 50 kN CPT mobile apparatus has been used. The device is not
automotive. It is maneuvered by hand and needs to be anchored in the soil in
order to archive the required reaction force for a CPT sounding.
The resulting profiles of the cone resistance qc for each single CPT and an
averaged profile are given in Figure 4.3. The cone resistance of the tests clearly
indicates a weaker layer at the depth of 4 to 5 m with a thickness of 0.5 and
1 m. Such a soft layer is not seen in the profile of the drilling. Figure 4.4
shows the undrained angle of internal friction φ calculated from the averaged
cone resistance. The value is approximately 30◦ at depths below the weak
intermediate layer and 34◦ above.
The groundwater table measured in the holes of the CPT’s is 1.15 m below
the surface.
4.4 Seismic cone penetration test (SCPT)
4.4.1 Test description

SCPT1 to SCPT4 were performed using a 200 kN CPT truck fitted with ad-
ditional tracks. The tracks can be lowered to support the conventional wheel
drive. Areas of difficult access can be reached by this means. For SCPT5 a
200 kN automotive remote-controlled track vehicle was available that is trans-
Parameter Unit Value

Median grain size, d50 mm 0.195
Uniformity coefficient, dd60
10
- 1.6
d30 2
Degree of curvature, Cc = d10 ·d60 - 1.02
Specific gravity, Gs - 2.65
Void ratio: maximum emax - 0.918
minimum emin - 0.585
Dry unit weight: minimum γdmin kN/m3 13.553
maximum γdmax kN/m3 16.387
Saturated unit weight: minimum γrmin kN/m3 18.25
maximum γrmax kN/m3 20.02
Table 4.5: Properties of Mol sand.
CPT 1 CPT 2 CPT 3 CPT 1−3

0 0 0 0
Mean q
c
Range of
standard
1 1 1 1 deviation
2 2 2 2
3 3 3 3
4 4 4 4
Depth [m]
5 5 5 5
6 6 6 6
7 7 7 7
8 8 8 8
9 9 9 9
0 10 20 0 10 20 0 10 20 0 10 20
q [MN/m²] q [MN/m²] q [MN/m²] q [MN/m²]
c c c c
Figure 4.3: Cone resistance qc single and mean profiles, Retie.
ported on the whole on top of a truck in the public road traffic. The CPT
truck and the track vehicle provide a sufficient dead load therefore anchoring
in the ground is not necessary.
Test depths went to 12.5 m and were limited by the mechanical strength of
the cone and the extension rods or by the resolution of the vibration sensors.
The maximum safe load on the cone at penetration refusal is 100 kN.
A cone equipped with a triaxial geophone and a cross-section of 10 cm2 has
been applied in SCPT1 to SCPT4. A cone with two triaxial accelerometers
4.4. SEISMIC CONE PENETRATION TEST (SCPT) 43
CPT 1-3
0
1 Mean φ
Range of
standard deviation
2
4
Depth [m]
9
10 15 20 25 30 35 40
φ [°]
Figure 4.4: Undrained angle of internal friction φ calculated from qc , Retie.
mounted in a distance of about 1 m and a cross-section of 15 cm2 was used in

SCPT5. The wave arrival time in test 1 to 4 was obtained by visual interpre-
tation; in SCPT5 cross correlation in the time domain has been applied.
The seismic source for S- and P-waves consists of a loaded steel beam. A
system of rollers between the beam and the coupling load was used in SCPT1,
SCPT3 and SCPT5. It lowers the horizontal friction between beam and load,
increasing the wave energy transfered to the ground. In SCPT2 a beam with
soil blades on its bottom side was in use to interlock the beam better with the
ground. Two kinds of hammers were applied to generate the wave pulse, a
conventional sledgehammer and a mechanical swing hammer with adjustable
falling height.
The design and working principles of the source beam and the mechanical
swing hammer can be found in detail in Areias et al. (1999) and Areias et al.
(2004).
4.4.2 Test results for the wave velocities
The results are presented in graphical form for shear wave velocity Vs and com-
pression wave velocity Vp in the Figures 4.5 and 4.6. Vp values are available
for SCPT1 and SCPT2 only. Both Vs and Vp generally show a trend of in-
creasing velocity with depth. Because of less arbitrary of the cross correlation
and the smaller scatter in the profile it can be assumed that SCPT5 gives the
most reliable values. At larger depths the compression wave velocities come
close to the velocity of compression waves in the present groundwater, about
1450 m/s.
SCPT 1−4 SCPT 1−2

0 0
No. 1
No. 2
2 2
4 4
6 No. 1 6
No. 2
Depth [m]
Depth [m]
No. 3
No. 4
8 8
10 10
12 12
14 14
0 200 400 600 0 500 1000 1500 2000
S−wave velocity [m/s] P−wave velocity [m/s]
Figure 4.5: S-wave and P-wave velocity, SCPT1-SCPT4, visual interpretation,

Retie.
SCPT1 to SCPT4 belong to the first SCPT’s performed by Ghent Univer-

sity. The adjustment of the seismic source was not always optimal. Because
with SCPT5 a reliable profile is available a re-analysis of SCPT1 to SCPT4 with
the cross correlation technique is not considered and also not very promising.
SCPT 5
0
4
Depth [m]
Mechanical Hammer
Sledgehammer
10
12
0 50 100 150 200 250 300 350 400 450 500
S−wave velocity [m/s]
Figure 4.6: SCPT5, S-wave velocity, cross correlation, Retie.
4.4.3 Test results for the damping ratio

The gathered time records of SCPT5 have been used for an analysis of the
damping ratio D of the shear wave component by means of the spectral ratio
slope method. The technique is described in Section 9.4.2. The results of the
damping ratio given in Figure 4.7 show major variation. At a few depths even
negative values are found. It is not reasonable to fit a line to obtain a closed
profile. The shallow layer until about 6 m depth shows a mean damping ratio
of 4 %, below this layer the value exceeds 10 % following the increasing trend of
the shear wave velocity. Under the assumption that the reached shear strain is
below 10−3 %, the values in comparison with values found in literature appear
to be quite high. Figure 4.8 gives example results from laboratory tests on
sands collected by Seed & Idriss (1970).
To provide alternative and perhaps more reliable values of the damping
ratio the correlation function of Ishibashi & Zhang (1993) is applied (Equation
(2.52)). The correlation function has been developed from a statistical analysis
of results of a large number of laboratory tests, mostly resonant column tests,
published in literature. It allows to estimate the damping ratio of a large range
of soils, from sands to moderately over-consolidated clays. The mean effective
stress σ̄0 , the plasticity index Ip and the shear strain γ are necessary input
values. A profile of the damping ratio for the site of Retie is calculated with
the assumption of the soil parameters in Table 4.6 and γ = 10−3 %.
0 0
Mechanical Hammer
Sledgehammer
2 2
4 4
Depth [m]
Depth [m]
6 6
8 8
10 10
12 12
0 10 20 30 40 0 1 2 3
Damping ratio [%] Damping ratio [%]
Figure 4.7: Values of damping ratio found from SCPT5 (left) and profile ob-
tained by Ishibashi’s equation (right).
Figure 4.8: Damping ratio for sand, mean values and scattering range from
Seed & Idriss (1970).
4.5. SPECTRAL ANALYSIS OF SURFACE WAVES (SASW) 47
Unit Value
Unit weight: above groundwater level γ kN/m3 18.0
below groundwater level γ 0 kN/m3 10.0
Angle of internal
friction φ0 : 0 to 4 m depth ◦
34.0
◦
deeper than 4 m 30.0
K0 : 0 to 4 m depth - 0.441
deeper than 4 m - 0.500
Plasticity index Ip : all depths - 0.0
Table 4.6: Input parameters for Ishibashi’s equation.
The by the buoyancy corrected unit weight of the saturated sand below the
groundwater table γ 0 is arbitrarily chosen at the upper end of the possible rage
given in Table 4.5. The not fully saturated unit weight above the groundwater
is chosen between the rages of dry and saturated unit weight. The internal
friction angle φ0 taken from the CPT’s is used to estimate a K0 value (ratio
horizontal to vertical stress). The latter is needed together with the vertical
stress for the calculation of the horizontal stress. Both stresses lead to the
mean effective stress. It should be remarked that the parameters as used in
the Ishibashi equation should not be seen as a conclusion of the physical soil
parameters. They are only one possible configuration out of the ranges given
earlier.
D decreases from 2.1 % close to the surface to 1.2 % at 20 m depth. The
complete profile can be found on the right side of Figure 4.7.
4.5 Spectral analysis of surface waves (SASW)
4.5.1 Test description

Setup 1 and setup 2
A drop weight device of the K.U. Leuven has been in use for setup 1, another
from Ghent University has been used for setup 2 to generate surface waves.
In both setups accelerometers were installed at fix distances from the source
at 2, 4, 6, 8, 12, 16, 24, 32 and 48 m. Additionally two geophones have been
placed at 12 and 30 m (subsetup a), 8 and 24 m (subsetup b) and 8 and 16 m
(subsetup c). The data of the vibration signals have been acquired by K.U.
Leuven for the accelerometers and by Ghent University for the geophones. The
obtained accelerometer signals from setup 1 and setup 2 have been analyzed by
K.U. Leuven and for setup 1 the Rayleigh wave dispersion curve was calculated
and an inversion algorithm applied. The data from the geophones were not used
for the analyses of setup 1 and 2.
Detailed information on both setups can be found in Pyl et al. (2000) which
is included in the report Areias & Haegeman (2001). The found shear wave
velocity profile is referred as SASW KUL in Figure 4.9.
Setup 3
Setup 3 has been performed approximately on the line of setup 2. Two geopho-
nes were consecutively placed at spacings of 0.5, 1, 2, 4, 8 and 12 m. Elastic
stress waves were generated by the impact of a hand-held hammer for the shor-
ter distances and the drop weight of Ghent University for the greater distances.
The wave source was placed on an equal distance from the first receiver as the
corresponding receiver spacing. The data were acquired by Ghent University,
a dispersion curve was obtained and an inversion calculation was done. The
inversion led to the shear wave velocity profile shown in Figure 4.9 referred as
SASW UGent.
SASW UGent SASW KUL
0 0
1 1
2 2
Depth [m]
Depth [m]
3 3
4 4
5 5
6 6
0 100 200 300 0 100 200 300
S−wave velocity [m/s] S−wave velocity [m/s]
Figure 4.9: SASW performed by Ghent University and K.U. Leuven, S-wave
velocity, Retie.
4.5.2 Test results

A model with 3 layers on a half-space has been used in the inversion calcu-
lation of K.U. Leuven and Ghent University. The density ρ was chosen as
ρ = 2000 kg/m3 . The coefficient of Poisson ν was assumed as ν = 0.48 from
K.U. Leuven and as ν = 0.30 from Ghent University. It should be noted that
the inversion calculation is very less sensitive regarding the chosen density and
the Poisson ratio. Deviating assumptions as in this case have nearly no influ-
ence on the results. A variation of the number of layers in the initial profile of
4.6. OVERVIEW OF THE TEST RESULTS RETIE 49
the inversion was done by K.U. Leuven. However the variation in the obtained
shear wave velocities was minor.
Even though the site has a homogeneous sand profile, the shear wave pro-
files from K.U. Leuven and Ghent University show a weaker layer with lower
propagation velocity. The profile from Ghent University has the soft layer bet-
ween 1.2 and 2 m depth and the profile from K.U. Leuven at a depth between
about 0.46 and 0.77 m. The calculated velocity at depths below 3.70 m is with
250 and 253 m/s for both tests similar.
4.6 Overview of the test results Retie

Figure 4.10 summarizes the geotechnical findings for the site in Retie. It gives
a simplified profile of the performed drilling B1 and the shear wave velocity.
The latter profile is based on the results of SCPT5 for depths below 1.50 m.
The velocity of the shallow layer is taken from both SASW tests. The soft layer
indicated by the CPT’s and SASW tests is because of its uncertain depth and
thickness not reflected in the simplified shear wave velocity profile. Furthermore
it could not be confirmed by the boring and the SCPT’s. The averaged CPT
profiles for tip resistance qc and internal friction φ are repeated. Table 4.7 gives
the numerical values.
Figure 4.10: Summary of the test results at Retie site.

4.6. OVERVIEW OF THE TEST RESULTS RETIE 51
Averaged CPT-Profile Profile of SCPT5

Depth qc φ Depth Vs[MH] Vs[SH] Vs[MEAN]
[m] [MN/m2 ] [◦ ] [m] [m/s] [m/s] [m/s]
0.20 0.60 26.70 1.34 232.53 225.53 229.03
0.40 1.53 29.96 1.86 236.93 230.33 233.63
0.60 2.90 31.57 2.38 213.81 228.68 221.24
0.80 3.47 31.55 2.89 201.65 218.30 209.97
1.00 5.70 33.37 3.41 200.16 206.83 203.49
1.20 8.07 34.83 3.93 211.03 210.14 210.58
1.40 9.33 34.97 4.44 243.12 226.73 234.92
1.60 10.60 35.31 4.96 240.19 222.77 231.48
1.80 10.63 34.90 5.48 223.52 212.77 218.15
2.00 10.77 34.64 5.99 216.02 226.58 221.30
2.20 10.00 33.95 6.51 256.70 260.62 258.66
2.40 10.50 33.92 7.03 304.54 282.66 293.60
2.60 10.07 33.37 7.54 344.96 331.56 338.26
2.80 9.50 32.65 8.06 356.10 318.50 337.30
3.00 11.00 33.53 8.58 362.70 311.04 336.87
3.20 10.50 33.03 9.09 360.40 362.95 361.68
3.40 9.60 32.27 9.61 339.19 333.69 336.44
3.60 10.10 32.33 10.12 314.46 324.38 319.42
3.80 7.15 29.13 10.64 378.21 329.72 353.96
4.00 4.80 27.23 11.16 302.68 388.34 345.51
4.20 5.20 27.53 11.67 353.76 389.94 371.85
4.40 3.75 24.80
4.60 3.00 22.32
4.80 3.20 20.23 SASW UGent
5.00 9.60 30.58 Layer Thickness Vs
5.20 8.50 29.67 [m] [m/s]
5.40 7.60 28.62 1 1.2 220
5.60 9.65 30.09 2 0.3 80
5.80 9.75 29.98 3 0.5 120
6.00 8.20 28.58 4 ∞ 250
6.20 8.00 28.19
6.40 7.60 27.48
6.60 8.35 27.87 SASW KUL
7.00 9.90 29.00 [m] [m/s]
7.20 11.35 29.68 1 0.46 143
7.40 15.75 31.42 2 0.31 46
7.60 15.20 31.03 3 2.93 159
7.80 15.50 31.02 4 ∞ 253
8.00 11.00 28.80
8.20 11.20 28.77
8.40 11.00 28.48
8.60 12.00 28.95
8.80 13.50 29.63
Table 4.7: Numerical values of the averaged CPT, SCPT5 and SASW results.
Chapter 5
Test site Lincent
5.1 Introduction
The testing site is located on a field between the Rue de la Bruyère in Lincent
and the high speed railway track Brussels-Cologne. A very elaborated soil
testing campaign is performed in preparation of the construction works to the
new build high speed track in 1993 and 1994. Information on the results of in
situ and laboratory tests in close proximity to the testing site is provided by
TucRail. They delivered conventional soil parameters of the site. SASW tests
and SCPT’s are performed by the Ghent University and K.U. Leuven. Table
5.1 gives an overview of all tests performed at this site.
A site plan is given in Figure 5.1. It shows the position of each in situ
test. For the testing procedure of the SASW test and the visual interpretation
of the SCPT can be referred to the report Areias & Haegeman (2001). The
report gives the numerical values of wave velocities from SCPT1 to SCPT3.
The SASW test of K.U. Leuven is described in Pyl & Degrande (2001a).

The drilling with the closest proximity to the locations of the SCPT’s, SASW
tests and train vibration measurements is boring B1108. It reaches a depth of
15 m. A simplified description of the boring profile is given in Table 5.2.
A top layer with a thickness of 1.2 m is followed by a layer of fine sand
reaching to a depth of 3.2 m. Between 3.2 and 7.5 m a sequence of very stiff
layers of arenite embedded in clay has been found. Arenite is a sediment of the
residue of a sand stone transported from another place. The fixations between
the individual grains can reach from intact to very loose. None of the performed
CPT’s or SCPT’s could pass this layer because the maximum pushing force was
reached.
Below the arenite layers is a layer of clay from 7.5 to 8.5 m depth, followed
by fine sand from 8.5 to 10.0 m. Between 10 and 15 m thin layers of fine sand
53
54 CHAPTER 5. TEST SITE LINCENT

'
&
&

#

"

Figure 5.1: Site location plan, Lincent.


[m]
01/02/1939 Boring at the Rue de Liege in 34.0
Lincent, from the Flemish Subsoil
Data Base
24/06/1991 Boring at the corner Burgemeester 74.0
Dumont Straat to Lijsemstraat in
Waasmont, from the Flemish Sub-
soil Data Base
18/08/1993 Boring No. 1108 provided by Tuc- B1108 15.0
Rail
06/07/1993 3 CPT No. 1107, 1108, 1109 CPT1107 4.0
provided by TucRail CPT1108 5.5
CPT1109 7.2
cone SCPT2 5.5
SCPT3 7.5
01/06/2001 3 SASW set-ups SASW KUL1
SASW KUL2
SASW UGent
15/05/2003 2 SCPT with a dual accelerometer SCPT4 5.7
cone SCPT5 7.2
and clay are found in a quick succession.

The sediments in a depth from 1 to 13 m can be geologically assigned to the
tertiary formation of Hannut. The layer from 13 to 15 m refers to the tertiary
formation of Herrs.
Two other deeper borings found in the Flemish Subsoil Database are de-
scribed in Table 5.3 and Table 5.4. The location of these borings is in the
center of Maasmont and in the Rue de Liège in Lincent. Even so the layer
descriptions agree not always with the profile of boring B1108, they give an
indication about the depth of the underlaying cretaceous formation consisting
of porous chalk. The transition from the tertiary formations to the cretaceous
formation is found at a depth of about 30 m in Lincent and at 60 m depth in
Waasmont. A value in between can be assumed for the actual site close to the
railway track.
Properties of the sampled material

Undisturbed soil samples are taken from the boring B1108 and classification
tests have been performed. Table 5.5 gives an overview of the results at three
selected sampling depths. Figure 5.2 shows the grain size distribution at the
different depths. The sample of depth 1 is taken from the silt layer close to the

[m] [m] Component
0.00 1.20 brown silt gravel
1.20 2.00 gray, brown to olive-green fine sand clay, glauconite
2.00 2.50 beige to gray beige fine sand clay, glauconite
2.50 3.20 gray to dark gray fine sand silt, clay,
glauconite
3.20 7.50 Sequence of:
a) dark gray to green arenite fine sand,
glauconite
b) dark gray to green brown clay fine sand, silt,
glauconite
7.50 8.00 dark gray, dark green clay silt
8.00 8.50 gray green to dark green clay silt, glauconite
8.50 10.00 gray, dark green fine sand glauconite
10.00 10.50 brown gray to dark green clay glauconite
10.50 11.00 gray, dark green fine sand glauconite
11.00 11.50 brown gray to dark green clay glauconite
a) dark brown to dark green clay silt, fine sand,
glauconite
b) dark gray green to sand arenite
brown green
12.50 13.00 dark gray clay silt, fine sand
a) gray, dark gray clay fine sand,
glauconite
b) dark gray to dark green fine sand glauconite
Lithographic Stratification: 0 to 1 m quaternary deposits, 1 to 2.5 m mem-
ber of Halen/Lincent (formation of Hannut), 2.5 to 13 m member of Wa-
terschei (formation of Hannut), 13 to 15 m member of Gelinden/Orp (for-
mation of Heers).
Table 5.2: Results of boring B1108 at the Lincent site.
surface. Sample 2 consists of the clay material in which the hard arenite layers
are embedded. Sample 3 is finally taken from an arenite layer. The grain size
distribution curve of sample 3 shows a significant amount of gravel.
The density is obtained for the clay sample only. The wet unit weight is
15.9 kN/m3 and the dry unit weight is 11.1 kN/m3 .

[m] [m] Component
0.00 1.00 brown back filling -
1.00 13.00 tawniness clay stones
13.00 45.00 gray clay petrified clay layers
45.00 50.00 light gray silt -
50.00 58.00 gray, green sand stones
58.00 60.00 gray, green sand glauconite
60.00 74.00 white porous chalk -
Lithographic Stratification: 0 to 1 m quaternary deposits, 1 to 45 m forma-
tion of Hannut and formation of Tienen, 45 to 50 m member of Gelinden
(formation of Heers), member of Orp (formation of Heers), 60 to 74 m
cretaceous formations.
Table 5.3: Results of the boring in Waasmont.

[m] [m] Component
0.00 4.00 brown silt glauconite, clay, sand
4.00 9.00 gray clay sand, glauconite
9.00 17.00 gray silt -
17.00 21.00 gray silt sand, mica, sporadic glauconite
21.00 31.00 gray silt much sand
31.00 34.00 white chalk -
Lithographic Stratification: 0 to 4 m quaternary deposits, 4 to 9 m member
of Lincent (formation of Hannut), 9 to 31 m member of Gelinden (formation
of Heers), 31 to 34 m chalk (cretaceous formations).
Table 5.4: Results of the boring in Lincent, Rue de Liège.
Groundwater table
The borehole of drilling B1108 was converted to a monitoring station for the
groundwater table. Form August 1993 to July 1997 the water level is measured.
It changes in a range between 6.0 m to 12.2 m below the surface, the mean
value is 10.4 m.
Data of three CPT soundings provided by TucRail in close proximity to the

testing field are taken into consideration. The profiles of the cone resistance
qc are given in Figure 5.3. The figure on the most right side shows the mean
value of all three tests. CPT1108 is performed on the same location as boring
B1108. Both other CPT’s are adjusted in height to the level of CPT1108 and
Unit Sample 1 Sample 2 Sample 3

Depth m 1.0-1.4 3.0-3.4 5.0-5.4
Grain size distribution
Frac- Grain
tion diameter
I <0.002 mm
(clay) % 16 34 18
II 0.002-0.06 mm
(silt) % 57 26 15
III 0.06-0.2 mm
(fine sand) % 24 32 15
IV 0.2-2.0 mm
(medium/
coarse sand) % 3 8 13
V 2.0-20.0 mm
(fine gravel) % 0 0 32
VI >20.0 mm
(coarse gravel/
stones) % 0 0 7
Specific gravity, GS - 2.633 2.594 2.514
Water content, w % 12.87 42.99 24.50
Void ratio, e - 1.2864
Degree of Saturation, Sr % 86.7
Unit weight, γ kN/m3 15.9
Dry unit weight, γd kN/m3 11.1
Saturated unit weight, γr kN/m3 16.7
Liquid limit, wL % 31.6 71.1 59.0
Plastic limit, wP % 21.7 39.8 38.3
Plasticity index, IP % 9.9 31.3 20.7
Classification CL OH GC
silty organic gravel-sand-
clay clay clay mixture
Table 5.5: Properties of samples taken in Lincent.

VI V − fine Gravel IV − coarse to medium Sand III − fine Sand II − Silt I − Clay
0
10
20 Depth 1 (1.0−1.4 m)
coarse Gravel to Stones
Sieve residue [%]
30 Depth 2 (3.0−3.4 m)
40
50
60
70
Depth 3 (5.0−5.4 m)
80
90
100
20 10 2 1 0.2 0.1 0.06 0.01 0.002 0.001
Grain size [mm]
Figure 5.2: Grain size distribution at the Lincent site.
CPT 1107 CPT 1108 CPT 1109 CPT 1107−1109

0 0 0 0
Mean q
c
Range of
standard deviation
1 1 1 1
2 2 2 2
3 3 3 3
Depth [m]
4 4 4 4
5 5 5 5
6 6 6 6
7 7 7 7
0 50 100 0 50 100 0 50 100 0 50 100
q [MN/m²] q [MN/m²] q [MN/m²] q [MN/m²]
c c c c
Figure 5.3: Cone resistance qc single and mean profiles, Lincent.

CPT 1107, 1108, 1109

0
Mean φ
Range of
1 standard deviation
3
Depth [m]
7
0 5 10 15 20 25 30 35 40 45
φ [°]
Figure 5.4: Undrained angle of internal friction φ calculated from qc , Lincent.
B1108 so they can be directly compared.

The penetration depth of the CPT’s is limited to a maximum of 6.5 m
because of the hard stratum of the arenite. The first appearance of the arenite
in the boring profile at 3.2 m is confirmed by CPT1108 and CPT1109. In
CPT1107, the farthest from B1108, it appears at a depth of about 4.4 m.
Figure 5.4 gives the profile of the correlated undrained angle of internal
friction φ.
5.4.1 Remarks on the testing Setup

SCPT1 to SCPT3 are performed using the 200 kN CPT truck fitted with
additional tracks. SCPT4 and SCPT5 are done by the 200 kN automotive
remote-controlled track vehicle. The tests penetrate to depths between 5.0 and
7.2 m and are limited again by the mechanical strength of the cone and the
extension rods as soon as the hard arenite layer is reached.
The seismic single cone with the triaxial geophone array is used in SCPT1
to SCPT3. The dual cone with two triaxial accelerometers in 1 m distance
is used for SCPT4 and SCPT5. The de-coupled seismic source provides the
needed shear wave excitation in most cases. Only for SCPT2 the coupled
seismic source is in use. Some closer details on cone and source can be found
in Section 4.4.1.

0 0
No. 1 No. 4
No. 2 No. 5
1 No. 3 1
2 2
3 3
Depth [m]
Depth [m]
4 4
5 5
6 6
7 7
8 8
0 100 200 300 400 500 0 100 200 300 400
Figure 5.5: S-wave velocity from SCPT1-SCPT3 using visual interpretation

(left) and SCPT4-SCPT5 using cross correlation (right), Lincent.
5.4.2 Test results for the wave velocity

Figure 5.5 presents the profiles of shear wave velocity from all performed
SCPT’s. The plot on the left side shows the results of SCPT1 to SCPT3,
analysed using the visual interpretation of the first shear wave arrival. The
right side gives the profiles of SCPT4 and SCPT5. These are calculated by
means of cross correlation between the two simultaneously recorded signals of
the dual cone in one of the horizontal axes. Each of the lines in the right figure
is the average of the two independently obtained velocity profiles using a sled-
gehammer on one side of the beam and the mechanical swing hammer on the
other side. The results of the shear wave velocity show a very good agreement
with both types of hammers.
As already seen on the site of Retie, the profiles obtained by visual inter-
pretation show a much higher scatter and variation than those obtained by
the cross correlation technique. Therefore also in this case it is assumed that
SCPT4 and SCPT5 deliver more reliable values on the soil profile than the
other SCPT’s.
The curves of SCPT4 and SCPT5 show a quite linear increase of the shear
wave velocity from about 160 m/s at 1 m depth to about 310 m/s at 6.5 m
depth. A discontinuity at the interface to the arenite layer cannot be found.

SCPT1 shows a significant increase of the velocity at about 4.5 m depth. In
SCPT2 and SCPT3 a sudden change may be seen at a depth of 3.5 m but is
doubtful.
Vp values are not calculated accurately because of poor signal definition
and are therefore not presented here.
SCPT 5
0
Mechanical Hammer
Sledgehammer
3
Depth [m]
8
0 1 2 3 4 5 6 7 8 9 10
Damping Ratio [%]
Figure 5.6: Values of damping ratio obtained from SCPT5, Lincent.
The signals recorded in the SCPT5 are analyzed to obtain the damping
ratio. The spectral ratio slope method is used for this purpose. The results
are presented in Figure 5.6. The values are in the range between 0 and 6 % at
all depths. The large variation cannot be considered as really satisfying. For
closer details it can be refered to Section 4.4.3 on the Retie site where a similar
scatter is found.
5.5 Spectral analysis of surface waves (SASW)
5.5.1 Remarks on the testing setup and inversion assump-

tions
SASW performed by K.U. Leuven - SASW KUL1/KUL2
Two setups are performed by K.U. Leuven. The first test uses the falling weight
device of K.U. Leuven; the second test uses a falling weight of TNO with a 5
times higher mass. The measurement line is chosen along the access road and
5.5. SPECTRAL ANALYSIS OF SURFACE WAVES (SASW) 63
the excitation weight is placed close to the rail track as shown in the site plan.
In the first setup the accelerometers are installed at 3, 4, 6, 8, 12, 16, 24, 32
and 48 m distance from the source; setup two uses accelerometers at distances
of 8, 12, 16, 20, 24, 32, 40, 48, 64, 80, 96, 128, 160 and 198 m.
Detailed information on the SASW tests of K.U. Leuven are given in Pyl &
Degrande (2001a). Inversion approaches with different numbers of layers are
commented in this report. The final inversion calculation for the first setup uses
a model with 4 layers on a halfspace; for setup two three layers on a halfspace
are used. The coefficient of Poisson ν = 0.33 and the density ρ = 1800 kg/m 3
are kept constant in the inversion procedure.
SASW performed by Ghent University - SASW UGent

A SASW test is performed in the direct neighbourhood to the SASW measu-
rement line of K.U. Leuven. The drop weight of the Ghent University and a
hand-held hammer is used to generate the surface waves. The testing procedure
is the same as at the Retie site.
The inversion calculation shows a profile of two layers on the halfspace.
The density and Poisson’s ratio are chosen ρ = 2000 kg/m3 and ν = 0.3, re-
spectively.
5.5.2 Test results
SASW UGent SASW KUL1/KUL2

0 0
Setup 1 (KUL1)
Setup 2 (KUL2)
1 1
2 2
3 3
Depth [m]
Depth [m]
4 4
5 5
6 6
7 7
8 8
0 100 200 300 400 0 100 200 300 400
Figure 5.7: SASW performed by Ghent University and K.U. Leuven, S-wave
velocity, Lincent.
The shear wave velocity profiles are given in Figure 5.7. The left side shows
the profile obtained by the Ghent University and the right side the two profiles
of K.U. Leuven.
The results of all inversion calculations show an increasing stiffness with
depth. At 3.0 respectively 3.5 m depth a stiffness increase is observed. This is
probably caused by the arenite layer.
5.6 Overview of the test results Lincent

Figure 5.8 summarizes the gathered geotechnical data for the site in Lincent.
It gives a simplified profile of drilling B1108. The layers are labeled based on
there main components. Names in brackets give information on the components
besides this main component. They are partly questionable. Which means a
clayey sand could also be a silty sand, a sandy clay, a silty clay.
The profile of the shear wave velocity is summarized out of the results of
SCPT4 and SCPT5 for depths below 1 m. It is assumed that both tests deliver
the most reliable values. They show a very uniform increase of the velocity
with depth. Furthermore the SASW profiles plot in the same range. The s-
wave velocity of the very shallow layer is taken as a mean value from all three
performed SASW tests.
Table 5.6 presents selected numerical values for CPT, SCPT and SASW
tests.
Simplified Stratification Averaged CPT-Profiles Simplified S-Wave
of Drilling B1108 Velocity Profile
Tip Internal
0.00 m 0 Resistance 0 Friction 0.00 m
Vs [m/s] 0.70 (Surface altitude of
Silt 1.00 0
1.20 1 1 the testing field)
140
Deposits
1 1.70
Quaternary
Fine Sand 2 2 160
(Clayey Sand)
2
3.20 3 3
3
4 4
Depth [m]
Depth [m]
Sequence of 4
Depth [m]
5 5
Arenite / Clay
(Sandy Clay 5
to Silty Clay) 6 6
6
7 7 310 7.20
7.50 0 20 40 60 0 20 40
qc [MN/m²] phi [°] 7
Clay (Silty Clay) 0 200 400
8.50 S-wave velocity [m/s]
(Tertiary Deposit)
Mean
Formation of Hannut
Fine Sand
Groundwater
10.40 m 10.00 Remark: The s-wave velocity profile starts from a depth of
0.70 m with the surface of the boring B1108 as
Sequence of reference-level.
Fine Sand / Clay
(Silty Clay)
5.6. OVERVIEW OF THE TEST RESULTS LINCENT
13.00
15.00
Site Lincent
(Tertiary Deposit)
Formation of Heers
Figure 5.8: Summary of the test results at the Lincent site.

65
Averaged CPT-Profile Profiles of SCPT’s

Depth qc φ SCPT4 SCPT5
[m] [MN/m2 ] [◦ ] Depth Vs Vs
0.20 12.20 40.00 [m] [m/s] [m/s]
0.40 9.75 40.00 0.83 421.31 173.21
0.60 9.37 37.80 1.34 174.63 166.65
0.80 6.50 34.10 1.86 166.23 190.10
1.00 9.03 34.67 2.38 176.08 203.97
1.20 10.90 34.97 2.89 216.80 201.14
1.40 8.33 34.40 3.41 223.96 221.10
1.60 4.17 29.47 3.93 231.00 251.71
1.80 1.37 20.80 4.44 231.92 274.00
2.00 2.53 23.53 4.96 253.00 279.62
2.20 13.53 30.93 5.48 - 272.43
2.40 14.37 32.37 5.99 - 280.41
2.60 1.30 17.13 6.51 - 324.83
2.80 2.33 20.87
3.00 3.83 24.10 SASW UGent
3.40 47.07 37.50 [m] [m/s]
3.60 10.10 28.20 1 1.0 110
3.80 6.27 23.70 2 2.5 165
4.00 33.30 33.53 3 ∞ 300
4.20 48.00 36.35
4.40 43.00 35.85 SASW KUL1
4.80 8.50 28.10 [m] [m/s]
5.00 42.80 35.00 1 0.19 108
5.20 58.00 37.55 2 2.53 158
5.40 15.10 31.05 3 0.22 176
5.60 11.00 28.95 4 3.33 248
5.80 8.15 26.15 5 ∞ 249
6.00 23.95 29.20
6.20 44.15 32.50 SASW KUL2
6.60 60.00 36.80 [m] [m/s]
6.80 44.00 35.30 1 3.02 150
2 0.10 280
3 0.56 278
4 ∞ 282
Table 5.6: Numerical values of the averaged CPT, SCPT4, SCPT5 and SASW
results.
Chapter 6
Test site Waremme
6.1 Introduction
The site is situated in the garden of the house Rue de la Costale 55 in Waremme.
The property borders directly on the high speed railway track Brussels-Cologne.
Data from the elaborated soil testing campaign performed in preparation of
the construction works to the new build high speed track in 1993 and 1994
are provided by TucRail. They delivered conventional soil parameters of the
site. SASW tests, SCPT’s and a boring are performed by the Ghent University
and K.U. Leuven. Undisturbed and also disturbed material obtained by the
boring are tested by means of bender element, resonant column and free torsion
pendulum test to determine Gmax and D. Table 6.1 gives an overview of all
considered tests performed at this site. A site plan is given in Figure 6.1.
For the testing procedure of the SASW test and the visual interpretation
of the SCPT can be referred to the report Areias & Haegeman (2001). The
report gives the numerical values of wave velocities from SCPT1 and SCPT2.
The SASW test of K.U. Leuven is described in Pyl & Degrande (2001b).

The boring B1 is driven on the meadow on a direct line between the house
Rue de la Costale 55 and the train track. The description of the boring profile
is given in Table 6.2. It shows beneath a shallow top layer of 1 m thickness
consisting of topsoil mixed with gravel a quite homogeneous layer of a clayey
silt. The color of the silt layer changes from rust brown over gray to black. It
is resting on a cretaceous formation of chalk. The transition zone consists of a
mixture of silt with field-stones and starts at about 12 m depth.
By means of the boring B3064 driven in 1993 on the other side of the train
track and a boring, not closer described, available from the Flemish Subsoil
Database, the first 7.5 m from the surface is assigned geologically to a qua-
ternary deposit. The rest of the silt layer until the beginning of the chalk is
67
68 CHAPTER 6. TEST SITE WAREMME
Figure 6.1: Site location plan, Waremme.


[m]
01/08/1936 Boring close to the station in 52.0
Waremme, Flemish Subsoil Data
Base
20/09/1993 Boring No. 3064 provided by B3064 20.0
TucRail
27/02/2001 Boring with discontinuous samp- B1 12.5
ling
18/08/1993 1 CPT No. 3063 provided by CPT3063 12.0
TucRail
19/08/1993 1 CPT No. 3064 provided by CPT3064 12.2
TucRail
31/05/2001 2 SASW set-ups K.U. Leuven SASW KUL1
SASW KUL2
01/06/2001 2 SASW set-ups Ghent SASW UGent1
University SASW UGent2
cone SCPT2 12.0
13/08/2002 1 SCPT with a dual accelerome- SCPT3 11.4
ter cone
ter cone
ter cone
considered as belonging to the tertiary formation of Heers (member of Orb).

Boring B3064 described in Table 6.3 confirms the findings from boring B1.
It is driven until a depth of 20 m and shows that the chalk extends at least to
this depth. It is reasonably assumed that the surface at B1 and B3064 are on
the same level.
The boring provided by the Flemish Subsoil Database is located at the
station. This is situated in the center of Waremme in about 1.3 km distance
from the testing field Rue de la Costale. Its data are given in Table 6.4. It
shows the extension of the cretaceous bedrock from 14 m to 52 m depth (final
boring depth). The found rock material changes between chalk, flint and tuff.

Classification tests have been performed on two samples of boring B3064. Both
belong to the silt layer, one from a depth between 3.0 and 3.4 m, the other from
a depth between 5.0 and 5.4 m. Figure 6.2 shows the grain size distribution of
the two samples. A wet unit weight of 19.74 and 20.24 kN/m3 was determined;

[m] [m]
0.00 1.00 dark brown gravel -
1.00 7.50 rust brown silt clay
7.50 8.25 gray brown silt clay
8.25 11.50 gray silt clay
11.50 12.00 gray silt field-stone
12.00 12.50 black silt field-stone
12.50 12.75 black field-stone chalk
Lithographic Stratification: 0 to 7.5 m quaternary deposits, 7.5 to 12.5 m
member of Orb (formation of Heers), 12.5 to 12.75 m chalk (cretaceous
formation).
Table 6.2: Results of boring B1 at the Waremme site.
the dry unit weight is around 16.71 kN/m3 . More details and the plasticity
parameters are given in Table 6.5.
0
10
20
Sieve residue [%]
30
40
50
Depth 1 (3.0−3.4 m)
60
70
80
Depth 2 (5.0−5.4 m)
90
100
20 10 2 1 0.2 0.1 0.06 0.01 0.002 0.001
Grain size [mm]
Figure 6.2: Grain size distribution at the Waremme site, B3064.
CU-triaxial tests are performed on the samples. An interpretation of the

results leads to mean values for the effective shear resistance parameters: φ 0 =
36o and c0 = 4 kPa.
The undisturbed material of boring B1 is used for bender element tests,
resonant column and free torsion pendulum tests. The results are given later.
Groundwater table
The borehole of drilling B3064 was converted to a monitoring station for the
groundwater table. Form October 1993 to May 1994 the water level is mea-
sured. It changes in a range between 3.28 m to 4.23 m below the surface, the
mean value is 3.7 m.
During the boring B1, directly in the garden of the property, the groundwa-
ter table was found at a depth of 6.72 m. The latter value might be erroneous
because only a single level determination was done.

[m] [m] Component
0.00 1.40 brown silt fine sand, angular
small stones
1.40 3.40 brown ocher silt fine sand
3.40 4.50 brown, yellow ocher silt fine sand
4.50 7.40 yellow ocher silt fine sand
7.40 8.00 olive-green to silt fine sand
yellow ocher
8.00 9.40 dark green, grayly, silt fine sand
partly yellow ocher
9.40 11.40 dark green, grayly, silt clay, ferric oxide
rust brown
11.40 12.50 dark gray, greenish silt sand, small
angular flint
stones
12.50 14.00 dark green angular flint fine sand
stones (1-3 cm)
14.00 15.50 light gray chalk sand, small
angular flint
stones (2-5 cm)
15.50 16.00 whitish beige chalk sand
16.00 19.50 light gray, beige chalk sand, small
angular flint
stones (1-5 cm)
19.50 20.00 whitish beige chalk sand
Lithographic Stratification: 0 to 8.0 m quaternary deposits, 8.0 to 12.5 m
formation of Heers, 12.5 to 20.00 m chalk (cretaceous formation).
Table 6.3: Results of boring B3064 at the Waremme site.
6.3 In situ tests

6.3.1 Cone penetration test (CPT)
Two CPT profiles provided by TucRail are taken into consideration. CPT3064
was performed at the same place as B3064. CPT3063 was situated at about
85 m distance from the test site in the direction of Brussels. The profiles of the
cone resistance qc are given in the Figures 6.3 and 6.4. The average of both
is shown in Figure 6.5. CPT3063 is shifted about 1.5 m in depth to take the
different levels of the surface between both CPT’s into account so that they
can be directly compared.
The penetration depth of both tests is limited by the stiff transition layer
between the clayey silt and the chalk bedrock at 12 respectively 13 m.

[m] [m]
0.00 3.00 brown silt topsoil
3.00 10.00 brown silt lime
10.00 14.00 gray silt -
14.00 18.00 gray flint chalk
18.00 23.00 brown tuff -
23.00 47.00 yellow chalk -
47.00 52.00 gray tuff flint
Lithographic Stratification: 0 to 14.0 m quaternary deposits,
14.0 to 52.0 m chalk (cretaceous formation).
Table 6.4: Results of a boring at the station in Waremme.
Figure 6.6 gives the profile of the correlated undrained angle of internal
friction φ.
6.3.2 Seismic cone penetration test (SCPT)

Remarks on the testing setup
The 200 kN CPT truck fitted with additional tracks is employed for SCPT1
and SCPT2. The seismic single cone with the triaxial geophone array together
with the de-coupled seismic beam source is used for these tests. The beam is
placed under the end of the truck chassis.
SCPT3, SCPT4 and SCPT5 are performed by the 200 kN automotive
remote-controlled track vehicle and the seismic dual cone equipped with two
triaxial accelerometers. The de-coupling rolls for the source are not used. In
SCPT3 the source beam is placed directly under the tracks of the CPT vehicle,
approximately in the middle. Because by this means the vertical static loading
of the beam could not be adjusted to a reasonable level the obtained vibration
signal strength was not optimum and the results have not been further ana-
lyzed. In SCPT4 and SCPT5 the source beam is placed under the hydraulic
supporting legs of the vehicle leading to better results. Some closer details on
cone and source can be found in Section 4.4.1.
The penetration depth of the SCPT’s is limited by the hard strata of chalk
at depths between about 11 and 12 m.
Test results for the wave velocity

Figure 6.7 shows the results of SCPT1 and SCPT2, analysed using the visual
interpretation of the first shear wave arrival. Figure 6.8 gives the profiles of
SCPT4 and SCPT5. These are calculated by means of cross-correlation bet-
ween the two simultaneously recorded signals of the dual cone in one of the
horizontal axes. Each of the lines of SCPT4 and SCPT5 is the average of the
Unit Sample 1 Sample 2

Depth m 3.0-3.4 5.0-5.4
Grain size distribution
Fraction Grain diameter
I <0.002 mm
(clay) % 20.1 12.3
II 0.002-0.06 mm
(silt) % 69.0 76.4
III 0.06-0.2 mm
(fine sand) % 10.6 9.3
IV 0.2-2.0 mm
(medium/coarse sand) % 0.3 1.1
V 2.0-20.0 mm
(fine gravel) % 0 0.9
VI >20.0 mm
(coarse gravel/stones) % 0 0
Specific gravity, GS - 2.641 2.644
Water content, w % 17.9 22.1
Void ratio, e - 0.5764 0.5956
Degree of Saturation, Sr % 83.6 100.0
Unit weight, γ kN/m3 19.74 20.24
Dry unit weight, γd kN/m3 16.74 16.58
Saturated unit weight, γr kN/m3 20.34 20.23
Liquid limit, wL % 34.6 29.5
Plastic limit, wP % 19.9 23.3
Plasticity index, IP % 14.7 6.2
Classification CL CL/ML
silty silty clay to
clay clayey silt
Effective angle of friction, φ0 ◦
31.9 39.2
Effective cohesion, c0 kN/m2 3 5
Table 6.5: Properties of samples taken in Waremme at B3064.

0 0 0
2 2 2
4 4 4
6 6 6
Depth [m]
Depth [m]
Depth [m]
8 8 8
10 10 10
12 12 12
14 14 14
0 10 20 0 10 20 0 10 20
qc [MN/m²] qc [MN/m²] qc [MN/m²]
Figure 6.3: CPT3064, Figure 6.4: CPT3063, Figure 6.5: CPT3064

cone resistance profile cone resistance profile and 3063, mean qc .
qc . qc .
two independently obtained velocity profiles using a sledgehammer on one side

of the beam and the mechanical swing hammer on the other side. The results
of the shear wave velocity show a very good agreement with both types of
hammers.
The profiles obtained by visual interpretation show a higher scatter and

variation than those obtained by the cross correlation technique. Therefore
also in this case it is assumed that SCPT4 and SCPT5 deliver more reliable
values than the other SCPT’s.
However, SCPT4 and SCPT5 differ in the shallow rage down to 3.5 m not
insignificantly. A local difference in the soil properties might be an explanation
since the tests have been performed at some distance from each other. In
greater depths both tests agree very well.
If SCPT4 and SCPT5 are summarized a quite linear increase from 115 m/s
at 1 m depth to 265 m/s at 11 m depth can be concluded.
Vp values are not calculated accurately because of poor signal definition.

6
Depth [m]
10
12
14
0 10 20 30 40 50
φ [°]
Figure 6.6: Undrained angle of internal friction φ calculated from qc , Waremme.

0 0
2 2
4 4
6 6
Depth [m]
Depth [m]
8 8
10 10
12 12
No. 1 No. 4
No. 2 No. 5
14 14
0 200 400 600 0 200 400 600
Figure 6.7: S-wave velocity from Figure 6.8: S-wave velocity from
SCPT1 - SCPT2 using visual inter- SCPT4 - SCPT5 using cross corre-
pretation, Waremme. lation, Waremme.
SCPT 5
0
6
Depth [m]
10
12
Mechanical Hammer
Sledgehammer
14
0 2.5 5 7.5 10
Damping Ratio [%]
Figure 6.9: Values of damping ratio obtained from SCPT5, Waremme.
Test results for the damping ratio

The signals recorded in the SCPT5 are analyzed to obtain the damping ratio.
The spectral ratio slope method is used for this purpose. The results are
presented in Figure 6.9. The determined values focus in the range between 0
and 3 %. For closer details on the scattering it can be refered to Section 4.4.3
on the Retie site where a similar scatter is found.

Remarks on the testing setup and inversion assumptions
Two measurement lines are tested once by the K.U. Leuven and another time
by the Ghent University. The first line is situated parallel to the building No.
55 and is named setup 1. The other line is perpendicular to the rail way track
and named setup 2. Both lines are indicated in Figure 6.1.
SASW performed by K.U. Leuven (KUL) The falling weight device of

the K.U. Leuven is used as source in these tests.
In the first setup the accelerometers are installed at 3, 4, 6, 8, 12, 24, 32
and 48 m distance from the source; setup two uses accelerometers at distances
of 3, 4, 6, 8, 12, 16 and 24 m only, because the space in the direction of the rail
way track is limited.
Detailed information on the SASW tests of K.U. Leuven are given in Pyl
& Degrande (2001b). The inversion is performed initially assuming a layer on
6.4. LABORATORY TESTS TO OBTAIN GMAX AND D 77
a halfspace. An extension to two layers on a halfspace changed nearly nothing

on the determined s-wave profiles.
The coefficient of Poisson ν = 0.33, the density ρ = 1800 kg/m3 and the
damping ratio for p- and s-wave βs = βp = 0.5 % are kept constant in the in-
version procedure.
SASW performed by Ghent University (UGent) The two SASW tests

of Ghent University are performed on the same lines as the tests of K.U. Leuven.
The drop weight of the Ghent University and a hand-held hammer is used to
generate the surface waves. The testing procedure is the same as at the Retie
site.
The inversion calculation delivered already sufficient result with the simu-
lation of a halfspace only. The density and Poisson’s ratio are chosen with
ρ = 2000 kg/m3 and ν = 0.3, respectively.
Test results
The shear wave velocity profiles are given in Figure 6.10 for setup 1 and in
Figure 6.11 for setup 2.
The results of nearly all inversion calculations show a constant stiffness with
depth. The SASW test of the K.U. Leuven suggests a change of s-wave velocity
at a depth of 4.5 m in the case of setup 1.
6.4 Laboratory tests to obtain Gmax and dam-

ping ratio
6.4.1 Bender elements with time arrival interpretation
Two samples of boring B1 were installed in the triaxial bender element test
device in a way that the s-wave velocity in the vertical direction can be deter-
mined. Two other were installed turned by 90o . By this means the horizontal
s-wave velocity can be measured. The sample was consolidated under isotropic
stress conditions. The s-wave velocity is measured at the end of the consolida-
tion by a visual interpretation of the arrival signals. A back pressure of 100 kPa
was used in al tests.
Table 6.6 summarizes the results and gives also calculated values of the
maximum shear modulus Gmax . The necessary density are actual values at the
end of the consolidation step. Figure 6.12 and Figure 6.13 show the results in
graphical form.
It is obvious that the differences in the results of horizontal and vertical
propagation are small if samples from a similar depth are compared. This
means that no significant structural anisotropy is found.
The development of Gmax during consolidation and further details can be
found in the technical report Haegeman (2002).
Setup 1 Setup 2
0 0
1 1
2 2
3 3
Depth [m]
Depth [m]
4 4
5 5
6 6
7 7
UGent UGent
KUL KUL
8 8
0 100 200 300 0 100 200 300
Figure 6.10: SASW results at setup Figure 6.11: SASW results at se-
1, parallel to the house, Waremme. tup 2, perpendicular to the rail way
track, Waremme.
0
Sample / σ1,2,3 [kPa]
Depth 25 50 100 200 400 600
B1/2 Vs [m/s] 120.1 143.2 182.5 221.6 279.4 -
vert.
2.0-2.5 m Gmax [MPa] 26.7 40.9 66.9 99.4 159.6 -

B1/10 Vs [m/s] - 168.2 198.2 239.8 294.3 315.6
10.0-10.5 m Gmax [MPa] - 57.1 79.1 116.6 177.0 204.8
B1/3 Vs [m/s] 110.4 138.9 175.1 217.7 - -
horiz.
3.0-3.5 m Gmax [MPa] 27.3 43.8 70.6 111.0 - -

B1/11 Vs [m/s] - 173.1 211.7 245.5 293.1 -
11.0-11.5 m Gmax [MPa] - 55.5 82.7 111.1 159.1 -
Table 6.6: Results of isotropic bender element tests with vertical and horizontal
propagation direction.
400 250
350
200
300
250
150
Gmax [MPa]
Vs [m/s]
200
100
150
B 1/2, vertical
B 1/2, vertical
100 B 1/10, vertical
B 1/10, vertical
50 B 1/3, horizontal
B 1/3, horizontal
B 1/11, horizontal
50 B 1/11, horizontal
0 0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
σc [kPa] σc [kPa]
Figure 6.12: S-wave velocity by iso- Figure 6.13: G0 by isotropic ben-

tropic bender element tests with der element tests with vertical and
vertical and horizontal propagation horizontal propagation direction.
direction.
A sample from the boring B1 from a depth of 12.0 to 12.5 m was tested in
the free torsion pendulum device of the Ghent University. The sample consists
mainly of silt and is taken from the transition area between the silt layer and
the chalk bedrock.
The dimensions of the sample were: 20 cm height, 10 cm diameter. The
isotropic stress stages have been chosen to 50, 100, 200 and 400 kPa. The
modulus at the reference shear strain of 10−5 is 35.1, 62.1, 100.6 and 152.9 MPa
at 50, 100, 200 and 400 kPa confining stress respectively.
The obtained results of the 200 kPa consolidation stage show an exceptional
behavior. The G-modulus increases extremely from γ = 10−5 in the direction of
the lower shear strain. The damping ratio shows a peak at γ = 3 · 10−6 . Even
though the test was repeated several times with the same exceptional result,
both phenomena are considered to be caused by the testing device. Therefore
the given values for the 200 kPa stage should be trusted only above a shear
strain of 10−5 for the G-modulus and above 10−4 for the damping ratio.
The damping ratio at a reference shear strain was found in the range of 1.0
to 3.0 %, neglecting the 200 kPa stage. The smaller values correspond to the
higher confining stress stages and the larger ones to smaller stresses.
Shear modulus and damping ratio in function of the applied shear strain can
be found in Figure 6.14. The same parameters in dependence of the confining
stress at a shear strain of 10−5 are given in Figure 6.15.
G-Modulus
180
50 kPa
160 100 kPa
140 200 kPa
400 kPa
120
G [MPa]
100
80
60
40
20
0 -6 -5 -4 -3
10 10 Shear Strain, γ [-] 10 10
Damping Ratio
0.15
50 kPa
100 kPa
0.1 200 kPa
400 kPa
D [-]
0.05
0 -6 -5 -4 -3
10 10 Shear Strain, γ [-] 10 10
Figure 6.14: G-modulus and damping ratio versus applied shear strain obtained
by the Free Torsion Pendulum Test, sampling depth: 12.0 to 12.5 m, boring
B1, Waremme.
The tests were performed by the laboratory of the Norwegian Geotechnical

Institute (NGI) in Oslo on two undisturbed samples from the site of Waremme.
The sampling depth was 4.0-4.5 and 5.0-5.5 m. An isotropic confining stress of
50, 100, 200 and 400 kPa was applied to the soil sample. Shear modulus and
damping ratio were measured during consolidation at a strain level of 10−5 .
The results at the end of consolidation, after about two days, are given in
Figure 6.15.
The averaged values of the shear modulus for both samples increase with
the confining stress from 41.8 MPa at 50 kPa isotropic stress to 65.4 MPa at
100 kPa, to 103.2 MPa at 200 kPa and finally to 154.6 MPa at 400 kPa.
The damping ratio is for all stress stages between 1.0 and 1.5 %. The
damping ratio obtained by the free torsion pendulum test is in for the lower
stress stages higher than the results from the resonant column test. The shear
moduli agree very well with the findings from the free torsion pendulum test.
6.5. OVERVIEW OF THE TEST RESULTS WAREMME 81
G−Modulus Dampig Ratio

160 0.05
0.045
140
0.04
120
0.035
100
0.03
G [MPa]
D [−]
80 0.025
0.02
60
0.015
40
0.01
20
0.005
0 0
0 100 200 300 400 0 100 200 300 400
σ [kPa] σ [kPa]
c c
Resonant Column Test, 4.0−4.5 m

Resonant Column Test, 5.0−5.5 m
Free Torsion Pendulumn Test, 12.0−12.5 m
Figure 6.15: G-Modulus and damping ratio versus isotropic confining stress,
shear strain 10−5 , boring B1, Waremme.
6.5 Overview of the test results Waremme

Figure 6.16 summarizes the gathered geotechnical data for the site in Waremme.
It gives a simplified profile of drilling B1. The layers are labeled based on their
main components.
The profile of the shear wave velocity is concluded out of the results of
SCPT4 and SCPT5 for depths deeper than 1 m. It is assumed that both tests
deliver the most reliable values. They show a uniform increase of the velocity
with depth, which is considered by the assumption of a linear increase from
115 m/s at 1 m depth to 265 m/s at 11 m depth. Furthermore the results
obtained by the SASW tests fall in the same range. The s-wave velocity of
the very shallow layer is taken from the SASW tests on setup 1. Its value is
155 m/s.
Table 6.7 presents selected values for qc and φ of the averaged CPT-profile;
Table 6.8 gives the s-wave velocities determined by SCPT4, SCPT5 and the
four SASW tests.
82
Simplified Stratification Averaged CPT-Profiles Simplified S-Wave

of Drilling B1 Velocity Profile
Tip Internal Vs [m/s]
0.00 m 0 Resistance 0 Friction 0.00 m
0
Gravel 155
1.00 1.00
115
2 2 2
4 4 4
Deposits
Quaternary
Groundwater
Silt 6 6 6
6.72 m (Clayey Silt
to Silty Clay)
7.50
Depth [m]
Depth [m]
Depth [m]
8 8 8
10 10 10
265 11.00
11.50
Silt/Stone Mixture 12 12
(Tertiary Deposit)
12.50 12
12.50
Chalk
12.75
14 14 14
0 5 10 15 20 0 20 40 0 200 400
Site Waremme q c [MN/m²] φ [°] S-wave velocity [m/s]
formation
Creataceous Formation of Heers
Figure 6.16: Summary of the test results at Waremme.

CHAPTER 6. TEST SITE WAREMME
6.5. OVERVIEW OF THE TEST RESULTS WAREMME 83
Averaged CPT-Profile
Depth qc φ Depth qc φ
[m] [MN/m2 ] [◦ ] [m] [MN/m2 ] [◦ ]
0.20 4.69 41.10 6.40 1.68 16.50
0.40 5.47 38.70 6.60 1.49 14.35
0.60 5.08 36.50 6.80 1.49 11.25
0.80 2.77 32.20 7.00 1.30 12.45
1.00 3.54 32.30 7.20 1.49 12.10
1.20 2.77 30.00 7.40 1.50 11.90
1.40 3.17 29.90 7.60 1.88 13.60
1.60 2.00 31.65 7.80 1.89 13.95
1.80 2.59 30.85 8.00 2.08 14.55
2.00 2.97 30.05 8.20 4.20 20.95
2.20 2.40 27.05 8.40 4.97 22.30
2.40 1.83 24.20 8.60 5.36 22.40
2.60 1.25 20.05 8.80 6.13 23.85
2.80 0.87 14.70 9.00 5.36 22.50
3.00 1.25 19.25 9.20 6.33 23.45
3.20 1.25 18.10 9.40 5.95 22.70
3.40 1.65 20.30 9.60 4.41 20.35
3.60 1.26 16.75 9.80 4.22 19.85
3.80 1.46 17.15 10.00 5.18 21.35
4.00 1.27 16.15 10.20 4.41 19.90
4.20 1.27 15.55 10.40 6.54 22.60
4.40 1.47 16.35 10.60 5.96 21.05
4.60 1.47 15.95 10.80 7.31 20.80
4.80 1.28 13.55 11.00 7.31 20.65
5.00 1.47 14.20 11.20 6.93 21.80
5.20 1.66 15.95 11.40 7.32 22.30
5.40 1.67 16.05 11.60 5.97 21.45
5.60 2.05 17.60 11.80 4.24 18.20
5.80 1.87 15.95 12.00 28.68 25.65
6.00 1.87 14.30 12.20 30.99 26.75
6.20 1.87 16.05
Table 6.7: Numerical values of the averaged CPT results.

Profiles of SCPT’s SASW UGent Setup 1

SCPT4 SCPT5 Layer Thickness Vs
Depth Vs Vs [m] [m/s]
[m] [m/s] [m/s] 1 >7 155
0.7 52.55 138.88
1.20 65.75 193.76 SASW UGent Setup 2
2.20 116.95 200.76 [m] [m/s]
2.70 117.45 154.75 1 >7 165
3.20 135.10 190.88
3.70 145.55 170.25 SASW KUL Setup 1
4.70 145.15 176.11 [m] [m/s]
5.20 158.95 183.72 1 4.46 154
5.70 178.45 189.64 2 1.08 215
6.20 186.55 198.07 3 ∞ 216
6.70 191.75 199.98
7.20 193.60 199.84 SASW KUL Setup 2
8.20 219.25 212.40 [m] [m/s]
8.70 227.20 219.79 1 0.27 145
9.20 251.90 244.33 2 0.70 179
9.70 273.95 259.74 3 ∞ 179
10.20 264.20 267.79
10.70 264.90 252.49
Table 6.8: Numerical values of SCPT4, SCPT5 and SASW results.

Chapter 7
Test site
Sint-Katelijne-Waver
7.1 Introduction
The test location (I) is situated on a field on the property of the De Nayer
Institute in Sint-Katlijne-Waver. Two borings were driven and two SCPT’s
were performed. In the framework of a research project on Soil Displacement
Screw Piles a previous elaborate soil testing campaign has been performed on
a research site at a distance of about 150 m. The surface of this site, later
on labelled as II, is about 2.3 m below the surface of site I. The tests have
been performed at the end of the 1990th. They deliver beside conventional soil
parameters also information on shear and compression wave velocity obtained
by SASW and seismic refraction tests. A detailed summary on this results can
be found in Mengé (2001), but the most essential findings are also repeated in
the following. Table 7.1 gives an overview of the in situ tests which results are
needed later.
A sketch of the site is given in Figure 7.1. It shows the position of the in
situ test on location I.

The stratification of the site is quite simple. It consists of a disturbed top layer
of clayed sand with stones and gravel, followed by a homogeneous layer of clay
which extends to great depths. The interface between the sand and the clay is
for testing site I at about 3 m depth and for site II at 0.7 m depth.
The clay belongs geologically to the tertiary formation of Boom, the sand
of the top layer is a quaternary deposit.
85
86 CHAPTER 7. TEST SITE SINT-KATELIJNE-WAVER
Date Available In Situ Tests Site Abbr. Depth

[m]
14/05/2003 Boring with discontinuous samp- I B1I 11.0
ling
14/05/2003 Boring with continuous sampling I B2I 11.0
13/05/2003 SCPT I SCPT1 12.9
14/05/2003 SCPT I SCPT2 12.9
Boring with undisturbed samp- II B1II 14.0
ling
30 CPT using an electrical cone II CPT-E 14.0
4 Marchetti dilatometer tests II DMT 13.0
5 SASW profiles II SASW1-5
3 measurement lines tested by II SRT
seismic refraction test
Table 7.1: Overview of the used in situ tests.
Building of the
De Nayer Institute
3.60
Entrance
12.10 7.20 3.10
Parking Space
14.00
Access Road
4.00
5.20
SCPT 2
5.00
Tree SCPT 1
Tree Tree Tree
B2 B1
1.60 1.80
[m]
Figure 7.1: Site location plan, Sint-Katelijne-Waver, site I.
Groundwater table
The groundwater table is found at site I at depths between 0.5 and 1 m depth.
If a horizontal groundwater table between site I and II is assumed the table at
site II can be estimated 2.8 to 3.3 m deep.

Important physical parameters of the Boom clay found in boring B1II are given
in the following. Laboratory tests to obtain Gmax and the damping ratio are
discussed in Section 7.4.
At site II samples are taken in the Boom clay at four depths, 4.5 to 4.9 m
(sample a), 8.5 to 8.9 m (sample b), 10.5 to 10.9 m (sample c) and 13.5 to
13.9 m (sample d). The density, the water content, the degree of saturation
and the plasticity index are similar. Therefore Table 7.2 also summarizes the
mean values for all four samples. The range of the curves of the grain size
distribution is given in Figure 7.2.
CU-triaxial tests are performed on samples from the four depths. A global
interpretation of all results leads to the effective shear resistance parameters
φ0 = 28o and c0 = 23 kPa.
Parameter Unit Sample Mean

a b c d Value
Dry unit weight, γd kN/m3 15.7 15.3 15.5 15.4 15.5
Natural unit weight, γ kN/m3 19.7 19.3 19.5 19.4 19.5
Water content, w: % 25.3 26.0 25.8 26.1 25.8
Saturation degree, Sr : % (101.8) 99.2 (100.9) (101.0) 100.0
Liquid limit, wL : % 72.2 65.1 75.9 71.8 71.6
Plastic limit, wP : % 25.4 25.3 26.2 26.0 25.7
Plasticity index, IP : % 46.8 39.9 49.7 45.8 45.6
Table 7.2: Properties of Boom clay.
0
10
20
Sieve residue [%]
30
40
50
Range of grain size distribution
60
70
80
90
100
20 10 2 1 0.2 0.1 0.06 0.01 0.002 0.001
Grain size [mm]
Figure 7.2: Grain size distribution at Sint-Katelijne-Waver.
7.3 In situ tests

7.3.1 Marchetti dilatometer test (DMT)
Altogether four DMT’s were performed on site II. A profile of the overcon-
solidation ratio OCR and another for the K0 -value were correlated using the
0 0
2 2
4 4
6 6
Depth [m]
Depth [m]
8 8
10 10
12 12
14 14
0 20 40 60 80 1 2 3 4
Overconsolidation ratio K
0
Figure 7.3: Overconsolidation ra- Figure 7.4: K0 -value derived from

tio derived from dilatometer test re- dilatometer test results at site II.
sults at site II.
horizontal stress index KD , obtained directly from the test. Both are shown
in Figure 7.3 and Figure 7.4. Mengé (2001) gives details on the correlation
equation.
The OCR-value in the boom clay is in the range between 70 at shallow
depths and 15 at 14 m depth. K0 decreases in the same zone from 3 to 1.8.
Because the top layer at site I is 2.3 m thicker than at site II it is not possible
to convert the values directly to site I. However, it can be assumed that site I
would show similar results.
7.3.2 Cone penetration test (CPT)

CPT’s have been done during the research project on Soil Displacement Screw
Piles using electrical CPT-cones. The averaged profile of the tip resistance q c
of 6 of the CPT’s is drawn in Figure 7.5. A high reliability of the CPT results
is concluded from the small range of standard deviation. Some discontinuities
in the averaged profile can be explained by the presence of a few boulders at
several depths. Also in this case the different thick top layers at site I and II
have to be kept in mind if the profile should be used also for site I.
7.3.3 Seismic cone penetration test (SCPT)

Test setup
The two SCPT’s on site I are performed using the 200 kN automotive remote-
controlled track vehicle. The cone was driven to a final depth of about 13 m in
6
Depth [m]
10
12
14
0 1 2 3 4 5
q [MN/m²]
c
Figure 7.5: Averaged cone resistance qc and range of standard deviation, site II.
SCPT 1−2 SCPT 2

0 0
No. 1 Mechanical Hammer
No. 2 Sledgehammer
1 1
2 2
3 3
4 4
Depth [m]
Depth [m]
5 5
6 6
7 7
8 8
9 9
0 100 200 300 0 2 4 6 8 10
S−wave velocity [m/s] Damping Ratio [%]
Figure 7.6: S-wave velocity using Figure 7.7: Values of damping ratio
cross correlation based on SCPT1 obtained from SCPT2.
and SCPT2.
both tests. Because of a bad signal quality leading to unreasonable fluctuations

in the results of the shear wave velocity the results are only shown to a depth
of 7.5 and 8.5 m.
The dual cone with two triaxial accelerometers in 1 m distance is used
together with the de-coupled seismic source for SCPT1 and SCPT2. Some
closer details on cone and source can be found in Section 4.4.1.
Test results for the wave velocity

Figure 7.6 gives the profiles of the shear wave velocity obtained by the cross
correlation technique. Each profile is the mean curve of a profile calculated
using the sledgehammer and another one using the mechanical swing hammer.
At depths beneath about 4 m the profiles of SCPT1 and SCPT2 fall together
and are in the range of 150 to 180 m/s. The shallow layer of clayed sand is
indicated by higher velocities up to 200 and 250 m/s. The agreement between
the two tests is in this top layer not as perfect as in the deeper clay layer.
Test results for the damping ratio

The SCPT2 is analyzed to obtain the damping ratio using the spectral ratio
slope method. The results are presented in Figure 7.7. The values are in the
range between 0 and 5 % at all depths. The variation is qualitativelly similar
to the other test sites. For closer details it can be refered to Section 4.4.3 on
the Retie site.

5 measurement lines were tested by the SASW method on site II. The resulting
profiles are presented in Figure 7.8, the numerical values are given in Table
7.3. The inversion calculation was performed assuming three layers on the
halfspace, for SASW 2 and 5 four layers of the halfspace were preferred because
the measured dispersion curve could be better approximated and because a fill
material was used at both measurement lines to replace the first 30 cm of the
top soil material. The very similar profiles from the five SASW tests show an
increasing stiffness with depth. Within the clay layer the shear wave velocity
increases from about 170 to 470 m/s. In comparison with the SCPT’s are these
noticeable higher values.
7.3.5 Seismic refraction test (SRT)

3 measurement lines on site II are also tested by seismic refraction. Figure 7.9
shows the obtained profiles. A s-wave velocity of VS = 145 m/s was found in
the top layer until a depth of 2 m. This corresponds to a G0 of 40 MPa. The
p-wave velocity gives values below 1000 m/s.
Table 7.4 gives the obtained velocity values for the clay layer beneath 2 m
depth. The s-wave velocity averages to about 350 m/s; the p-wave velocity
0 0
No. 1 Profile A
No. 2 Profile B
1 No. 3 1 Profile C
No. 4
No. 5
2 2
3 3
4 4
Depth [m]
Depth [m]
5 5
6 6
7 7
8 8
9 9
0 200 400 600 800 0 200 400 600 800
Figure 7.8: SASW tests performed Figure 7.9: Seismic refraction test
at site II. results at site II.
No. 1 No. 2 No. 3 No. 4 No. 5

Layer d Vs d Vs d Vs d Vs d Vs
[m] [m/s] [m] [m/s] [m] [m/s] [m] [m/s] [m] [m/s]
1 0.8 90 0.3 170 0.7 110 0.6 90 0.3 200
2 1.0 180 0.5 100 1.1 170 1.2 170 0.5 110
3 4.1 380 1.1 160 4.4 320 4.2 350 1.1 160
4 ∞ 450 4.3 340 ∞ 470 ∞ 470 4.4 300
5 ∞ 500 ∞ 700
Table 7.3: SASW Profiles, site II.
agrees with the propagation speed in water of about 1450 m/s or is even higher.
Closer details on these seismic refraction tests can be found in Mengé (2001).
However, the found s-wave velocity of the clay confirms the results of the
SASW tests but is higher than the results from the SCPT’s.
7.4 Laboratory tests to obtain Gmax and dam-

ping ratio
7.4.1 Bender elements with time arrival interpretation
The undisturbed samples from site II were tested in a triaxial cell under iso-
tropic confining stress. The results of the analysis of the arrival of the shear
Profile Vp Vs ν Edyn G0
[m/s] [m/s] [-] MPa MPa
A 1667 320 0.48 576 195
B 1457 348 0.47 675 231
C 1669 385 0.47 840 286
Table 7.4: Seismic refraction test results for the clay material beneath a depth
of 2 m, site II.
160 300
140
250
120
200
100
G0 [MPa]
VS [m/s]
80 150
60
100
40
50
20
0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
σc [kPa] σc [kPa]
Figure 7.10: Results of bender ele- Figure 7.11: Results of bender ele-
ment tests, shear modulus G0 ver- ment tests, s-wave velocity VS ver-
sus isotropic consolidation stress σc , sus isotropic consolidation stress σc ,
site II. site II.
wave are drawn in Figure 7.10 and Figure 7.11. Both diagrams show the same
results once in terms of shear modulus and the other in terms of shear wave
velocity. They can be linked if a density of 1988 kg/m3 is used.

A sample taken from the boring segment between 4.0 and 4.5 m depth of
boring B2I was installed in the free torsion pendulum test device of the Ghent
University. The dimensions of the sample were: 20 cm height, 10 cm diameter.
The isotropic stress stages have been chosen to 50, 100, 200 and 400 kPa. The
damping ratio at a reference shear strain of 10−5 was found in the range of 3.0 to
4.0 %. The smaller values correspond as expected to the higher confining stress
stages and the larger ones to smaller stresses. The modulus at the reference
shear strain is 35.0, 42.1, 53.4 and 69.2 MPa at 50, 100, 200 and 400 kPa
confining stress respectively.
Shear modulus and damping ratio in function of the applied shear strain can
be found in Figure 7.12. The same parameters in dependence of the confining
G−Modulus
90
80
70
60
G [MPa]
50
40
30
50 kPa
20 100 kPa
200 kPa
10 400 kPa
0
−6 −5 −4 −3
10 10 Strain [−] 10 10
Damping Ratio
0.08
0.07
0.06
0.05
D [−]
0.04
0.03
50 kPa
0.02 100 kPa
200 kPa
0.01 400 kPa
0
−6 −5 −4 −3
10 10 10 10
Strain [−]
Figure 7.12: G-Modulus and damping ratio versus applied shear strain obtained
by the Free Torsion Pendulum Test, sampling depth: 4.0 to 4.5 m, site I.
stress at a shear strain of 10−5 are given in Figure 7.13. The found increase
of the damping ratio from a shear strain of 2 · 10−6 to lower strains is not
expected. It is assumed that this is caused by limitations in the test equipment
to measure very small movements of the excitation beam.
A Boom clay sample has been sent to the private laboratory GEOLABO in
France to perform a resonant column test. The isotropic stress stages have
been again chosen to 50, 100, 200 and 400 kPa. Shear modulus and damping
ratio were measured during consolidation at a strain level of 10−5 . The values
of the shear modulus increase with the confining stress from 27.9 MPa at 50 kPa
isotropic stress to 32.3 MPa at 100 kPa, to 42.0 MPa at 200 kPa and finally
to 55.9 MPa at 400 kPa. The damping ratio is for all stress stage between 3.5
and 4.0 %. Both can be seen in Figure 7.13.
The damping ratio agrees very well with the findings from the free torsion
pendulum test. The shear modulus obtained by the free torsion pendulum test
is in all cases 20 to 30 % higher than the results from the resonant column test.
G−Modulus Damping Ratio

80 0.06
70
0.05
60
0.04
50
G [MPa]
D [−]
40 0.03
30
0.02
20
0.01
Resonant Column Test
10 Resonant Column Test Free Torsion Pendulum Test
Free Torsion Pendulum Test
0 0
0 100 200 300 400 0 100 200 300 400
σ [kPa] σ [kPa]
c c
Figure 7.13: G-Modulus and damping ratio versus isotropic confining stress,
shear strain 10−5 , sampling depth: 4.0 to 4.5 m, site I.
7.5 Overview of the test results Sint-Katelijne-

Waver
Figure 7.14 summarizes the geotechnical test results for the site in Sint-
Katelijne-Waver. The simplified profile of the shear wave velocity at site I
is based on both SCPT’s. But there are no SCPT data available for depths
smaller than 0.8 m. The velocity for this shallow layer is assumed out of the
SASW data and the seismic refraction tests at site II. The soil material close
to the surface has to be similar at site I and II to justify this assumption.
Therefore the reliability of the shear wave velocity values in the first meter is
limited. Table 7.5 presents numerical values for the SCPT’s.
Simplified Stratification Simplified S-Wave Simplified Stratification Averaged CPT-Profile
at Site I Velocity Profile at Site I at Site II at Site II
0.00 m Vs [m/s] 0.00 m
0
100
210 0.80
1
Clayey Sand
Groundwater, 2 210 2.30 m 0.00 m 0
Deposits
Clayey
Quaternary
max. 2.80 m
3.00 3 3.00 3.00 m 0.70 m Sand
min. 3.30 m 150
Deposits
Quaternary
4
Depth [m]
2
Groundwater
5 as at site I
6
4
Clay
7
8
180 8.50 6
Formation of Boom
9
0 100 200 300
S-wave velocity [m/s] Clay
Depth [m]
8
11.00
Formation of Boom
10
12
Site Sint-Katelijne-Waver
16.30 m 14.00 m
14
0 1 2 3 4 5
Reference level: surface site I surface site II q [MN/m²]
c
Figure 7.14: Summary of the test results at Sint-Katelijne-Waver.

7.5. OVERVIEW OF THE TEST RESULTS ST.-KATELIJNE-WAVER 95
Profiles of SCPT’s
SCPT1 SCPT2
Depth Vs Vs
[m] [m/s] [m/s]
0.83 - 204.27
1.34 142.06 172.30
1.86 234.22 181.08
2.38 244.37 208.20
2.89 255.15 166.95
3.41 173.68 140.83
3.93 145.93 151.18
4.44 151.52 153.12
4.96 155.63 151.81
5.48 157.93 160.79
5.99 162.15 163.29
6.51 164.29 174.00
7.03 166.16 169.56
7.54 164.36 165.63
8.06 - 179.59
8.58 - 184.12
Table 7.5: Numerical values of Vs from SCPT1 and SCPT2.

Chapter 8
Test site Ghent
8.1 Introduction
The test location in Ghent is situated close to a storage and garage building
of the Flemish Geotechnical Institute on the campus of the Ghent University
in Zwijnaarde. A SCPT is performed on this site. Boring and CPT data
are additionally available. They have been gathered in preparation of the
construction of the buildings of the geotechnical institute in the late 1960th.
A lay-out plan of the testing site is given in Figure 8.1 and the overview of
all used in situ test results can be found in Table 8.1.

[m]
15/07/1968 Boring inside the ground-plan of the B6 14.0
main building of the Flemish Geotech-
nical Institute
19/07/1968 Boring at the closest corner of the sto- B21 5.0
rage and garage building to SCPT1
19/07/1968 CPT at the place of boring B21 CPT21 14.75
25/11/2003 SCPT with a dual cone equipped with SCPT1 17.5
Brüel & Kjær accelerometers
8.2 Borings and sampling

The boring B21 in close proximity to the performed SCPT reached only to a
depth of 5 m. Therefore it is combined with the deeper, but also more far,
boring B6. The stratification of the combined profile at the place of boring
97
98 CHAPTER 8. TEST SITE GHENT
Parking Space
B6
Flemish Geotechnical Institute / Ghent University
Technologiepark 905, Zwijnaarde
Parking Space
CPT 21 SCPT 1
Storage and B 21
Garage Building
Figure 8.1: Site location plan, Ghent.
B21 is given in Table 8.2 and in the site overview in Figure 8.5. Because of the
extrapolation, the accuracy of the layer interface-levels at depths deeper than
5 m might be rather low.
However, the site has a 0.75 m thick top layer of a silty sand followed by
fine sand to a depth of about 8.25 m. The sand is mixed from this depth on
with an increasing amount of silt. The material is classified as silty sand to
sandy silt.
All described soil layers are deposits originated in the Quaternary. The first
tertiary layer at the site is referred to the formation of Tilt. It is deeper than
the final depth of the evaluated boring. Its presence is seen in boring profiles
available in the Flemish Geotechnical Database on other locations inside the
university campus.
Groundwater table
The groundwater table is found at the boring B21 at 1.85 m depth.


[m] [m] Component
0.0 0.75 dark brown fine sand silt
0.75 8.25 gray, brown fine sand -
8.25 10.80 gray fine sand silt
10.80 13.50 gray silt fine sand
Lithographic Stratification: 0 to 13.5 m quaternary deposits.
Table 8.2: Combined results of boring B21 and B6 for the location of B21.

A CPT sounding has been performed at the place of boring B21 at one corner
of the garage. The profile of the tip resistance of the cone qc is given in Figure
8.2.
0
6
Depth [m]
10
12
14
16
0 2 4 6 8 10 12 14 16
q [MN/m²]
c
Figure 8.2: Cone resistance qc , CPT21.

8.4.1 Testing setup
The SCPT was performed using a 200 kN CPT-tuck without track support.
Because of a very good signal quality the test was driven to a final depth of
about 17.5 m.
The dual cone with two triaxial accelerometers from Brüel & Kjær (Section
9.2.2), mounted in 1 m distance from each other, is used together with a de-
coupled seismic source. Instead of the steel beam two square shaped wooden
plates were applied. The horizontal hammer blows on the edge of the plates
were performed by the sledgehammer from both sides of the truck. The me-
chanical swing hammer was not used at this site. Some closer details on cone
and source can be found in Section 4.4.1.
8.4.2 Test results for the wave velocity

The shear wave velocity profiles obtained by cross correlation from both excita-
tion directions are very consistently. They are given in Figure 8.3. A simplified
profile can be found in the summary in Figure 8.5.

The data of the SCPT are analyzed to obtain the damping ratio using the
spectral ratio slope method. The results are presented in Figure 8.4. The
values are mainly in the range between 0 and 10 % at all depths. The variation
is qualitativelly similar to the other test sites. For closer details on the variation
in the results can be refered to Section 4.4.3.
SCPT 1
SCPT 1
0
0
2 2
4 4
6 6
8
Depth [m]
8
Depth [m]
10 10
12 12 Left side
Right side
Left side
14 Right side 14
16 16
18 18
0 100 200 300 400 0 10 20 30
S−wave velocity [m/s] Damping Ratio [%]
Figure 8.3: S-wave velocity from Figure 8.4: Values of damping ra-
SCPT using cross correlation, exci- tio obtained from SCPT, excitation
tation from both sides of the truck. from both sides of the truck.
8.5. OVERVIEW OF THE TEST RESULTS GHENT 101
8.5 Overview of the test results Ghent

Figure 8.5 summarizes the gathered geotechnical data for the site in Ghent.
The simplified s-wave velocity profile is only derived from the SCPT in Figure
8.3 because there are no other measurements of wave velocity performed at this
site. Table 8.3 presents numerical values for the SCPT.
102
Simplified Stratification Averaged CPT-Profile Simplified S-Wave Velocity Profile
at Drilling B21 at CPT21 from SCPT1
0.00 m Vs [m/s] 0.00 m
0
0
Fine Sand
0.75 (Silty Sand)
Groundwater
1.85 m 180
2 2
3.00
160
4
Fine Sand 4
(Sand)
6 6
Deposits
280
Depth [m]
7.80
Quaternary
8.25 8 8
Fine Sand
(Silty Sand)
Depth [m]
10 10
10.80
Silt 12 12
(Sandy Silt)
250
13.50
14 14
16 16
0 2 4 6 8 10 12 14 16
q [MN/m²] 17.00
c
18
0 100 200 300 400
Site Ghent, Technologiepark 905 S-wave velocity [m/s]
Figure 8.5: Summary of the test results at the site in Ghent.

CHAPTER 8. TEST SITE GHENT
8.5. OVERVIEW OF THE TEST RESULTS GHENT 103
Profiles of SCPT’s
SCPT1
Depth Vs, Left Vs, Right
[m] [m/s] [m/s]
1.34 228.02 170.23
1.86 198.58 192.48
2.38 179.91 173.01
2.89 156.71 164.08
3.41 172.16 176.04
3.93 182.39 181.69
4.44 203.71 195.25
4.96 214.44 220.84
5.48 231.68 227.22
5.99 242.09 260.00
6.51 269.30 272.48
7.03 265.46 286.80
7.54 287.33 284.05
8.06 261.24 261.19
8.58 251.82 258.82
9.09 247.12 250.81
9.61 238.09 245.00
10.13 246.17 242.83
10.64 242.77 239.01
11.16 248.12 249.57
11.67 256.85 251.50
12.19 228.94 238.74
12.71 243.60 248.02
13.22 257.02 256.12
13.74 261.63 268.89
14.26 259.69 279.15
14.77 267.13 270.41
15.29 256.53 260.21
15.81 245.56 260.25
16.32 252.19 264.30
16.84 230.26 241.43
Table 8.3: Numerical values of Vs from SCPT1; excitation from the left and
from the right side.
Part III
Studies on testing methods
105
Chapter 9
SCPT based techniques
9.1 Motivation and purpose

The seismic cone penetration test (SCPT) is developed from the cone penetra-
tion test (CPT) using electrical cones. The capabilities of an electrical cone are
extended by adding geophones or accelerometers to the usual transducers for
tip resistance, sleeve friction and pore pressure. Also cones that are limited to
the vibration sensors are in use. These cones are called ”dummy cones”. Waves
generated at the surface travel to the cone pushed into the ground. The test is
similar to a down-hole test except for the borehole, necessary in the down-hole
test to install seismic sensors. The main purpose of the method is to obtain the
velocity of shear and compression waves leading to values for Gmax and Emax .
Objectives and methodology

Aim of this part of the research is to obtain at first confidence in the deter-
mination of wave velocity profiles. Afterwards the possibilities to use hidden
information in the time histories to calculate the damping ratio of the soil
should be investigated. The incorporation of the damping ratio measurement
would mean a significant improvement of the SCPT method. This is especially
true considering the fact that the few other approaches (e.g. SASW with ex-
tended inversion, cross-hole test) to measure material damping in-situ are still
under research and not state-of-the-art.
In order to achieve the aims the following topics are treated and described
in this chapter:
- Description of the test equipment
- Selection of suitable sensors for the installation in the seismic cone
- Summary of available direct and indirect time methods to measure wave

velocity, including example calculations
107
108 CHAPTER 9. SCPT BASED TECHNIQUES
- Theoretical background of three approaches to determine the damping

ratio: Attenuation coefficient method, spectral ratio slope method and
the analysis of stress-strain curves
- Application of attenuation coefficient method and spectral ratio slope

method on artificially generated time records to study the influence of
signal windowing to the results
- Application of the spectral ratio slope method on all test sites (Retie,
Lincent, Waremme, Sint-Katelijne-Waver, Ghent)
- Analysis of stress-strain curves based on data from a SCPT in Waremme
- Comparison of the results from the SCPT with resonant column and free
torsion pendulum tests on samples from Waremme and Sint-Katelijne-
Waver concerning Vs and Ds
9.2 Applied equipment and selection criteria

An overview of a typical seismic cone set-up is given in Figure 9.1; the equip-
ment described in detail in the following paragraphs is shown in Figure 9.2.
Lateral Offset
Static
Mechanical Load
Sledge−
Hammer hammer
Beam
Depth
tT
LT tB
LB
Triaxial
Accelerometers
Seismic Cone
Penetrometer
Figure 9.1: Schematic sketch of a SCPT set-up using a dual cone.
9.2.1 Seismic source

The seismic source consists of a steel beam loaded by the weight of the CPT
truck and different types of hammers. A horizontal hammer blow at one end
of the beam produces a shear wave dominated pulse and a vertical blow on top
of the beam a compression wave dominated pulse.
9.2. APPLIED EQUIPMENT AND SELECTION CRITERIA 109
ENDEVCO ICP source/amplifier
6 channels
6 channels Fugro signal conditioner module

Termination box
3 channels
Brül & Kjaer accelerometers
X2, Y2, X2, Y2,

Hamlin accelerometers or
Z2 Z2
X1, Y1,
geophones
Z1
Geophones
PC with data−
Trigger acquisition card
channel
Hammer and beam

X1, Y1, X1, Y1, (shearwav e source)
Z1 Z1
Dual seismic cone Single

seismic cone
Figure 9.2: Functional overview of the seismic cone equipment.
Applied load
provided by the CPT truck
a)
Source beam Impact anvil
Applied load
b) provided by the CPT truck
Upper bearing beams
Roller system
Source beam Impact anvil
Applied load
c) provided by the CPT truck
Impact anvil
Wooden beams
Figure 9.3: a) conventional coupled source, b) de-coupled source, c) de-coupled

source with a wooden plate as designed by Areias et al. (2004).
The SCPT’s performed in the framework of this research use in most of the
cases a source beam with a system of rollers on top, under the loading weight of
the CPT truck. It is developed by Areias et al. (1999) and called a de-coupled
source. This increases the horizontal mobility of the beam resulting in a more
efficient transfer of the impact energy of the hammer to the actually generated
shear waves. Comparative studies between the de-coupled and the coupled
source without the roller system are found in Areias et al. (2004) showing the
positive effect of the rollers.
According to the same author there is an optimum vertical stress at the
boundary between beam and soil. Consequently, it is always attempted to
adjust the vertical force on the beam, applied by hydraulic cylinders, to its
optimum value. This is checked by comparing time records of additional ham-
mer blows performed in the beginning of each SCPT with the cone at a fixed
shallow depth.
In the SCPT test on the university campus in Ghent two wooden plates are
successfully used instead of the steel beam. Their size suits more the demands
of an easy transport than the steel beam. Also in this case rollers under the
loading piston are placed. Figure 9.3 shows principle sketches of the source
beams.
The impact energy is given by a sledge hammer or a mechanical swing
hammer with an adjustable falling height. With the latter is aimed to generate
repeatable shear pulses of the same intensity. However, comparisons of repea-
ted blows with the cone at a certain depth show that the assumption that the
mechanical swing hammer provides a repeatable source is questionable. Diffe-
rences in the amplitudes and shapes of the generated shear pulse are observed,
probably caused by a slight backwards movement of the hammer support and
beam from blow to blow. A fixing of the hammer frame with metal pins on
the ground could not improve the result. Furthermore it can be assumed that
the load on the source beam changes with time due to a slowly sinking of the
CPT truck additionally causing a changing shear amplitude.
The acceleration of the head of the mechanical swing hammer is measured
on the occasions of the SCPT5 in Retie and SCPT1 in Sint-Katelijne-Waver.
A small shock accelerometer is installed on the backside of the hammer head.
The weight of the whole moving arm of the hammer is approximately 25 kg.
Because of the unknown weight distribution in the hammer arm, the impact
force was not calculated.
Four selected hammer impacts at the site in Retie are shown in Figure 9.4.
Figure 9.5 provides the associated amplitude spectra. Even so the spectra are
partly different, it can be concluded that the main energy of the hammer impact
is located between 200 and 600 Hz. It has to be remarked that the measured
time histories characterize only the impact force applied to the source beam.
The energy distribution of the dynamic force transfered to the ground might
be different.
Another point concerns the measurement itself: the signal is sampled with
2000 Hz, more than sufficient for the ground vibrations, but the frequency
content of the impact pulse approaches the Nyquist frequency of 1000 Hz.
Because there is no anti-aliasing filter applied to the hammer channel, it is
not completely impossible that existing signal parts at frequencies higher than
1000 Hz are wrapped into the high frequency parts of the spectrum just below
1000 Hz. However, this disturbance is probably limited to the higher frequency
range.
4 4
x 10 x 10
8 8
No. 295 No. 309
Particle acceleration [m/s²]

6 6
4 4
2 2
0 0
−2 −2
−4 −4
−6 −6
−8 −8
0.48 0.5 0.52 0.54 0.56 0.48 0.5 0.52 0.54 0.56
Time [s] Time [s]
4 4
x 10 x 10
8 8
No. 321 No. 325
6 6
4 4
2 2
0 0
−2 −2
−4 −4
−6 −6
−8 −8
0.48 0.5 0.52 0.54 0.56 0.48 0.5 0.52 0.54 0.56
Time [s] Time [s]
Figure 9.4: Acceleration of the mechanical swing hammer during the SCPT
No. 5 in Retie, selected impacts.
9.2.2 Seismic cones

Two seismic cones are available at Ghent University manufactured by Fugro
Engineers, the Netherlands; one cone with a single triaxial array of geophones
and another cone with two triaxial arrays of accelerometers, installed 1 m or
0.5 m distance from each other. The single cone is additionally equipped with
transducers for the tip resistance, sleeve friction and pore pressure. The dual
cone is a dummy cone without conventional transducers. The cross-sectional
tip area of the single cone is 10 cm2 and of the dual cone 15 cm2 . Sketches
of the dual cone with 1 m sensor distance and of the single cone are given in
100
90
No. 321
80
Amplitude Spectrum [m/s²]
70
60 No. 325
50
40
No. 309
30
20
No. 295
10
0
0 200 400 600 800 1000
Frequency [Hz]
Figure 9.5: Amplitude spectra of the selected hammer impacts from Figure
9.4.
Figure 9.6.
In the frame of this research both cones are used. The seismic sensors
in the dual cone are replaced two times in the run of the project. The first
change replaced the originally installed geophones by VTI Hamlin accelerome-
ters because of insufficient frequency response characteristics of the geophones
as described later on. These accelerometers where later replaced by accelero-
meters from Brüel & Kjær to improve the signal-to-noise ratio leading to larger
testing depths.
In the following the characteristics of the sensors used are described.
GeoSpace geophones
The triaxial array consists of three single geophones GSC-14-L3 produced by
GeoSpace. The geophones are designed and built to maintain performance
characteristics even after being subjected to high shock forces. The main spe-
cifications of the sensors given by the manufacturer are:
• Natural frequency: 28 Hz ± 5 Hz
• Moving mass: 2.15 g
• Dimensions (diameter/height): 1.7 cm / 1.8 cm
• Open circuit damping: 18 %
• Intrinsic voltage sensitivity: 0.114 V/(cm/s), amplified by an internal

amplifier to 27 V/(cm/s)
Triaxial Sensor
Triaxial Sensor (top-part)
(A=18 cm²)
(A=15 cm²)
(A=15 cm²)
(bottom-part)
Ø 47.873
Ø 43.702
Ø 43.702
(A=10 cm²)
(A=10 cm²)
Ø 35.682
Ø 35.682
Cone-tip
60°
189 1033 260
392 720 370

1482
Load cell/Pore Dual Cone

pressure transducer Friction Triaxial (1 m configuration)
Sensor
(A=10 cm²)
transducer
Ø 35.682
60°
400 230
Cone-
tip 51 133 446
[mm] 630 Single Cone
Figure 9.6: Dual cone in 1 m configuration and single cone.
Because the suspension of the geophones in the housing of the cones might
have an influence on the natural frequency and to validate the given sensitivity,
a dynamical calibration of the single cone has been performed in the Laboratory
for Mechanical Engineering at Ghent University. The testing set-up to calibrate
the vertically orientated geophone is given in Figure 9.7 and the horizontally
orientated sensors in Figure 9.8.
In both set-ups the cone is excited by a mechanical shaker using a homoge-
neous sine function and the time records of the output signal of the geophones
are stored. The movement of the cone is additionally measured by a well cali-
brated ICP accelerometer mounted at the surface of the cone in close proximity
to the geophones. At each frequency of the sinusoidal excitation the amplitude
of the signals from the geophone output and from the accelerometer are measu-
red. After integrating the acceleration amplitude into velocity, the ratio to the
velocity amplitude of the geophone provides one point in the response curve,
describing the sensitivity dependency on the frequency. The whole procedure is
repeated with a changed excitation frequency until the desired frequency range
from 1 to 500 Hz is covered.
In an early stage a LVDT, an eddy current proximity transducer and a laser
interferometer are also used to provide reliable reference values for the cone
movement. However, the results show that the surface accelerometer delivered
the clearest signals easily to analyze. Figure 9.9 shows the obtained sensitivity
curves, Figure 9.10 the general calibration curve of the manufacturer. The
response curves of the horizontal geophones fall together in Figure 9.9, therefore
only one line is drawn.
Roller bearing
Triaxial geophone
(inside)
Single Cone
Accelerometer Accelerometer
Single cone
Leaf spring
Roller bearing
(rotation centre) Triaxial geophone (inside)
Shaker
Shaker
Figure 9.7: Calibration of Figure 9.8: Calibration of the single cone

the single cone with instal- with installed geophones, horizontal.
led geophones, vertical.
It is not possible to compare the calibration curve given by the sensor

manufacturer with the results of our calibration directly because the cone ma-
nufacturer has added an amplifier to the signal flow. It is impossible to switch
this amplification off during the calibration.
The results of these tests are: a) the geophones in vertical and horizontal
direction have a similar response behavior, b) the sensitivity over the investiga-
ted frequency range is not flat at any frequency, c) the natural frequency of the
sensors is about 33 Hz. Therefore the measurement sensors are not qualified for
tasks were an accurate amplitude is requested. Nevertheless they are sufficient
for analyses of time arrivals for wave velocity measurements.
VTI Hamlin accelerometers SCA 600
For afore mentioned reasons the geophones of the dual cone are in the frame-
work of this research replaced by accelerometers of the type SCA600-C13H1G
for the horizontal axes and by SCA600-C13V1G for the vertical axis.
The vertical Hamlin accelerometers of the dual cone are calibrated in the
mechanical engineering laboratory under similar testing conditions as the single
cone, using one reference accelerometer. The testing set-up for the horizontal
accelerometers is slightly modified to consider the two triaxial arrays. The
sketch of the set-up is given in Figure 9.11.
6
Vertical Geophone
Horizontal Geophones
5
Sensitivity [V / (mm/s)]
0
0 50 100 150 200 250 300 350 400 450 500
Frequency [Hz]
Figure 9.9: Calibration curves of the seismic single cone equipped with geo-
phones GeoSpace GSC-14-L3.
0
10
OUTPUT [V/(IN/SEC)]
−1
10 GeoSpace
GSC−14−L3
−2
10 1 2 3
10 10 10
FREQUENCY [Hz]
Figure 9.10: Calibration curve of the geophones GeoSpace GSC-14-L3 as pro-

vided by the manufacturer.
Reference Roller bearing Reference

Dual cone (rotation centre)
accelerometer accelerometer
Leaf spring
Triaxial Triaxial
accelerometer (inside) accelerometer (inside)
Shaker
Figure 9.11: Calibration of the dual cone with installed Hamlin accelerometers,
horizontal.
The main specifications of the Hamlin sensors are:
• Measuring range: ± 1.5 g
• Sensitivity: 1.116 V/g by own calibration, 1.333 V/g given by manufac-

turer
• linearity: until 400 Hz ± 150 Hz flat response with a deviation smaller

than -3 dB
• Resolution (Residual noise): 0.9 - 4.5 mg (1 mg = 9.81 · 10−3 m/s2 )
The accelerometers have in principle a true DC response. That means

the sensitivity should not drop down in the low frequency range. However,
the conditioner equipment adds a bandpass filter with corner frequencies of 1
and 350 Hz, that can partly explain the changing curve in the range below
approximately 20 Hz as shown in Figure 9.12.
The sensitivity of the accelerometers is according to our own calibration
slightly lower than expected from the manufacturer specification. In the in
situ tests the value of 1.116 V/g is always used.
Triaxial DeltaTron Brüel & Kjær accelerometers

The installed Hamlin accelerometers showed an unexpected high noise compo-
nent forcing in almost all cases to stop SCPT testing at depths of about 10 m
because of low signal-to-noise ratio, so the main wave pulse was not identifiable
anymore.
Therefore it is decided to change the accelerometers once more. This time
triaxial accelerometers of the type Brüel & Kjær 4506 B 002 are installed.
The sensors require a constant power supply compatible to the IsoTron or ICP
specifications. They have a noise level 30 to 150 times smaller than those of
the Hamlin accelerometers.
1.4
chosen characteristic sensitivity: 1.116 V/g

1.2 (exact at 23 and 50 Hz )
1
Sensitivity [V/g]
0.8
0.6
0.4
0.2
0
0 20 40 60 80 100 120 140
Frequency [Hz]
Figure 9.12: Calibration curves of the Hamlin accelerometers of type SCA600-

C13H1G/C13V1G.
The main parameters of the Brüel & Kjær 4506 B 002 accelerometers are:
• Measuring range: ± 7 g
• Sensitivity: ≈ 1 V/g (exact values specified by the manufacturer separa-
tely for each axis)
• Amplitude linearity (flat response with a deviation smaller than 10 %):
– Horizontal axes: 30 - 1600 Hz
– Vertical axis: 30 - 3500 Hz
• Resolution (Residual noise):
– Horizontal axes: 0.03 mg (1 mg = 9.81 · 10−3 m/s2 )
– Vertical axis: 0.06 mg
A calibration by means of the shaking apparatus is not performed anymore.
Figure 9.13 gives the response curve provided by Brüel & Kjær.
9.2.3 Data acquisition system

A PC-based data acquisition system is used to record waveform signals from
the SCPT. The set-up can be found in Figure 9.2.
For measurements with the dual cone, at least 6 channels are required. The
signal conditioner module provides the power for the sensors inside the cone
Figure 9.13: Typical frequency response of the Brüel & Kjær 4506 B 002.
and applies an analogue bandpass filter to the signals with the edge frequencies
1 and 350 Hz (high-pass 1 Hz, 6 dB/oct; low-pass 350 Hz, 12 dB/oct). This
conditioner module of the seismic cone is connected to the termination box of
the data acquisition card inside the PC by conventional BNC-cables. The first
channel is used to trigger the measurement. The hammer and the beam are
connected to this port. A voltage of 5 V is applied to the channel inside the
termination box. When hammer and beam get in contact, the two poles are
bypassed over a resister and the voltage on the channel drops. This can be
used to generate a software trigger in the acquisition program.
In the case that the Brüel & Kjær accelerometers are used the conditioner
unit is to be replaced by an ICP or Isotron source. At the test site in Ghent
a filter and amplifier system KEMO VBF35 provided by the K.U. Leuven is
used. The lowpass cut-off frequency is set to 1000 Hz. Alternatively the 9
channel source and amplifier of Ghent University ENDEVCO 2792B without
filter can be applied.
SCPT’s using the single cone are performed with a sampling frequency of
4000 Hz; vibration events measured with the dual cone are sampled at 2000
Hz. So the Nyquist frequency is 2000, respectively 1000 Hz, much higher than
the 350 Hz cut-off frequency of the bandpass filter, or equal to the 1000 Hz cut-
off frequency of the KEMO VBF35 system. Aliasing effects are consequently
impossible or in case of the KEMO amplifier sufficiently reduced. On the site
in Ghent at some depths measurements are repeated without analog low-pass
filter. The time histories show a similar quality as those with filter except for
some additional noise.
Main specifications of the data acquisition card manufactured by Microstar
Laboratories are:
• Type: DAP 4000a/112, PCI bus
• Number of analog inputs: 16

9.3. METHODS FOR THE SHEAR MODULUS 119
• Possible input ranges: ± 2.5 V, ± 5 V, ± 10 V
• Maximum sampling frequency (all channels together): 800 kHz
• Analog/digital resolution: 14 bit
• Programmable gain rates: 1, 10, 100, 500
For this research a software tool for data acquisition is developed in the
graphical programming environment LabVIEW. The main features of this pro-
gram are described below.
• possibility to handle 10 channels, 6 for the dual seismic cone, 3 for an

additional surface vibration measurement, 1 for a shock accelerometer
mounted at the impact hammer
• possibility to average different repetitions of an event at the time of mea-

surement
• free choice of sampling rate, pre- and post-trigger time (limited by the
acquisition card)
• measurement triggering by a contact signal or by exceeding an adjustable

voltage level on one of the channels
• possibility to apply a gain (1, 10, 100, 500) to each channel separately
• storage of the acquired data in an easy to read ASCII-file
• quick overview of the signals in time and frequency domain
• possibility to reject disturbed signals
• possibilities to store events manually, semi-automatically and automati-

cally
• advanced capabilities of further automated analyses using the power of

the LabVIEW environment
The program replaces the commercial software packet ProView from Nicolet
and is already used during the calibration tests of the single and dual cone. This
work is done with the help of the Department of Electronics and Information
Systems of the Ghent University.
9.3 Methods to obtain the dynamic shear mo-

dulus
The shear modulus is obtained by determining the shear wave velocity Vs .
Because the generated waves reach only very small strain levels below the
elastic threshold strain the modulus calculated with Equation (2.1) can be
assumed as Gmax or G0 .
The elasticity modulus Emax or E0 can be calculated out of the compression
wave velocity Vp using Equation (9.1) for an infinite elastic medium.
(1 + ν)(1 − 2ν)
Emax = Vp2 ρ (9.1)
1−ν
Poisson’s ratio ν can be directly determined out of the ratio between Vp
and Vs in Equation (9.2).
2
Vp
1 Vs − 2
ν = 2 (9.2)
2 Vp
Vs −1
The wave velocity is calculated by comparing time records from different

depths. If a single cone with a sensor array at a certain depth is used, the
excitation pulse can be only measured once. In this case the comparison is
performed with a time record obtained at another cone depth, associated with
another excitation pulse. The test depth interval in this research is approxi-
mately 50 cm. Other authors prefer 100 cm because this value matches the
standard length of an extension rod of the SCPT.
If a dual cone is used for the test, as in nearly all cases in this project,
the comparison can be done between the time records obtained from the two
triaxial arrays installed in the cone related to the same excitation pulse. The
most important advantages of a dual cone in comparison to a single-array cone
are: a) measurements of arrival times are independent of trigger accuracy, b)
measurements are not affected by the energy of the source and c) accuracy of
depth measurement between tests is not critical to the overall accuracy of the
results.
It is assumed that the waves travel on the shortest way from the center of
the source, in general for s-waves the middle of the used source beam or plate,
to the sensors installed in the cone. Because the source beam is placed at a
horizontal distance usually longer than 1 m, from the cone rod, this geometrical
offset has to be taken into account.
The assumption that the origin of the generated wave is in the middle of
the source beam might be questionable depending on the actual shape of the
interface between beam and ground surface. Another possible origin might be
the end of the beam on the side of the hammer strike. For a vertical hit on top
of the beam to generate p-waves, the last assumption seems more reliable and
is therefore chosen. However, one should bear in mind that the possible error
introduced by a wrong selected origin, decreases with an increasing penetration
depth of the cone.
The actual travel path can deviate from the direct line if the wave is re-
fracted on layer interfaces. Based on Snell’s law as given in Equation (3.10)
the bended travel path from source to receiver can be predicted. However,
Stewart (1992) has shown that the consideration of a bended path has nearly
no influence on the calculated wave velocity and is therefore not taken into
account.
The main focus to calculate the wave velocity is the determination of the
travel time between two measurement points. The methods to achieve this can
be divided into direct and indirect time methods.
The signals of the horizontal sensors in the direction of the hammer impact
(y-axis) are used for the s-wave velocity determination. The vertical sensors
(z-axis) are used for the p-wave velocity determination.
9.3.1 Direct time methods

Direct time methods are based on the visual interpretation of the time histories.
Characteristic points in the signals are identified and their shifts in time with
increasing depth are used to calculate the wave velocity.
The most obvious way is to determine the time of the first arrival of the p-
or s-wave, but also the first peak or the first trough might be used. Another
possibility is to draw two signals from hammer impacts from both sides of the
source beam on top of each other. The shear wave component of the two signals
will be polarized. This can be seen in the example time histories in Figure 9.14.
Components not related to the generated s-wave will show the same polarity in
both signals. In this way the s-wave component can be identified more easily.
The reference point in time is the first crossing point of the two signals close
to the time of first arrival. The method is therefore called cross-over method.
The direct time methods represent the most straightforward way to obtain
the wave velocity. However, there are cases where the identification of the cho-
sen reference point is difficult and requires some experience from the operator.
Beyond that is an automation of the first arrival or peak picking fault-prone.
Therefore the application of the methods is in general time consuming.
9.3.2 Indirect time methods

The most popular indirect methods are the cross correlation method, still per-
formed in the time domain, and the method based on the phase of the cross
power spectrum as a frequency domain technique.
It is found in this research that the cross correlation in the time domain
works most effectively and reliably. In addition it provides the opportunity for
automation. Therefore the applied technique is explained in more detail using
example time histories.
Cross correlation method
The time shift, ∆t = τ , between the signals of the bottom and the top acce-
lerometer in the cone, sB and sT , is obtained at the maximum of the cross
Top element, y−direction, Depth: 12.19 m

0.4
Hammer blow left
Hammer blow right
Particle velocity [m/s²]
0.2
−0.2
−0.4
0.5 0.52 0.54 0.56 0.58 0.6 0.62 0.64 0.66
Time [s]
Bottom element, y−direction, Depth: 13.22 m

0.4
Hammer blow left
Hammer blow right
Particle velocity [m/s²]
0.2
−0.2
−0.4
0.5 0.52 0.54 0.56 0.58 0.6 0.62 0.64 0.66
Time [s]
Figure 9.14: Example time histories from the site in Ghent, cone tip at a fixed
depth, hammer blows from left and right.
correlation function, rBT :

Z ∞
rBT (τ ) = sB (t) sT (t + τ ) dt (9.3)
t=−∞
The difference in the travel path length divided by the time shift gives the
shear wave velocity.
Because the sampling frequency is limited, the accuracy of the time shift
∆t, obtained by the cross correlation is also limited. In most cases the sampling
frequency is set to 2000 Hz, corresponding to a sampling interval of 0.5 ms.
The calculated time shifts are consequently multiples of 0.5 ms. This leads, in
consideration of the small travel path length between the two receivers of about
1 m, to an insufficient accurate velocity calculation. The obvious solution for
this problem is to increase the sampling frequency. This is done in reality by
changing the settings of the analogue/digital converter in the data acquisition
unit or, more conveniently, by interpolating between the measured data points.
The latter technique allows to increase the sampling frequency artificially to
virtually any value.
The interpolation is based on the sampling theorem of the digital signal pro-
cessing. The theorem implies that a band limited signal is completely described
if it is sampled with at least the double of the maximum signal frequency. This
is ensured by the selection of a sufficient high sampling frequency. Therefore
it is possible to reconstruct the signal between the sampling intervals unambi-
guously. With other words, it is possible to interpolate. Kammeyer & Kroschel
(1998) provide an equation for this interpolation for an arbitrary time t:
∞
X sin[π(fS · t − k)]
s(t) = s(k/fS ) · (9.4)
π(fS · t − k)
k=−∞
with the signal s(t) originally sampled as s(k/fS ) at times k/fS . fS is the original
sampling frequency and k the sample index.
The signals are processed by another algorithm more straightforward to
implement. The source can be found in Weinstein & Clifford (1979) and is
realized in a MATLAB script. The main steps are the following:
1. expanding of the input data vector by inserting zeros between the original
data values
2. design of a special symmetric FIR (finite impulse response) filter that
allows the original data to pass through unchanged and interpolates in
between
3. application of the filter to the input data vector to produce the interpo-
lated output vector
The time records obtained from SCPT’s are interpolated to a 50 times
higher sampling rate. This means, for the case of fS = 2 kHz, an artificial
sampling frequency of 100 kHz, corresponding to a sampling interval of 0.01 ms.
This leads consequently to a finer time shift and also wave velocity resolution.
Figure 9.15 gives an example time history from the testing site in Ghent.
The actual acquired data points are indicated as dots, the line in between is
interpolated and further on used for the cross correlation.
Figure 9.16 shows the cross correlation of the left side time records given
in Figure 9.14. The maximum of the curve appears at a ∆t of 4.23 ms. The
travel-path difference is 1.0304 m, giving a s-wave velocity of 243.60 m/s.
Phase of the cross power spectrum

The phase of the cross power spectrum can be used to analyze the variation of
the velocity with frequency. The technique is the basis of the SASW method.
After the calculation of the discrete Fourier transformation of both signals
sT (t) and sB (t) the cross power spectrum is obtained by Equation (9.7). SB (f)
is the complex conjugate spectrum of the bottom signal. The discrete Fourier
transformation is abbreviated for simplicity as FFT (Fast Fourier Transfor-
mation) even though the FFT is only a certain but most common numerical
implementation of the discrete Fourier transformation.
Top element, y−direction, depth: 12.19 m

0.4
Acquired data points
Interpolated time history
0.2
−0.2
−0.4
0.54 0.55 0.56 0.57 0.58 0.59 0.6
Time [s]
Figure 9.15: Example of the signal interpolation on the measurement signal.
−4
x 10
∆t = 4.23 ms
Cross−correlation [m²/s³]
−2
−4
−0.1 −0.05 0 0.05 0.1 0.15

Time shift [s]
Figure 9.16: The cross correlation of the signals from Figure 9.14; picked ma-
ximum at ∆t = 4.23 ms.
9.4. METHODS TO OBTAIN THE DAMPING RATIO 125
ST (f ) = F F T (sT (t)) (9.5)

SB (f ) = F F T (sB (t)) (9.6)
RT B = ST (f ) · SB (f ) (9.7)
The phase shift φ between the signals is calculated from RTB :
Im(RT B )
tan(φ) = (9.8)
Re(RT B )
The phase shift is wrapped, that means the values are limited by −π and π.
Before the time shift ∆t(f) between the two signal is calculated using Equation
(9.9), it has to be unwrapped.
φ(f )
∆t(f ) = (9.9)
2π f
Finally the wave velocity V(t) is obtained in consideration of the travel path
difference ∆x.
∆x
V (t) = (9.10)
∆t(f )
Figure 9.17 shows the application on signals from the site in Ghent as used
in Figure 9.14. The velocity in the range of 50 to 150 Hz is nearly constant,
confirming the assumption that the s-wave is not dispersive, and is in good
agreement with the result of the cross correlation in the time domain. Outside
the frequency range of 50 to 150 Hz, the velocity fluctuates. This is in consi-
deration of the frequency content of the signals also reasonable. Most of the
energy is concentrated in the range of 50 to 150 Hz. The calculation of the
coherence function would probably confirm this finding.
9.4 Methods to obtain the Damping Ratio

9.4.1 Attenuation coefficient method
This paragraph will introduce the equations describing body waves and discuss
factors affecting the amplitude of these waves. This leads to a method to
quantify the wave attenuation from seismic measurements.
First consider a simple sine wave with no attenuation traveling along a
string with wavelength λ, velocity c and source amplitude A0 . t marks the
time and x the distance of the current observation point from the source, then:
2π
A++ (x, t) = A0 sin (x − ct) (9.11)
λ
−3
x 10 Amplitude spectra, y−direction
2.5
Top element, depth: 12.19 m
Amplitude [m/s²]
2 Bottom element, depth: 13.22 m

1.5
1
0.5
0
0 50 100 150 200 250 300
Wrapped / unwrapped phase of the cross−power spectrum
10
Phase angle [rad]
Unwrapped phase
Wrapped phase
5
−5
0 50 100 150 200 250 300
S−wave velocity
500
S−wave velocity [m/s]
400
300
200
100 Result of time domain cross−correlation, V = 243.60 m/s
s
0
0 50 100 150 200 250 300
Frequency [Hz]
Figure 9.17: Analysis of the phase of the cross power spectrum; amplitude
spectra of the signals, phase shift and resulting s-wave velocity; travel-path
difference between the two signals is 1.0304 m.
By introducing the wave number k = 2π/λ and the angular frequency

ω = 2π/T with T = λ/c where T is the period and neglecting any phase shift
the following equation is obtained:
A++ (x, t) = A0 sin(kx − ωt) (9.12)
or, in terms of the complex exponential:
A++ (x, t) = A0 ei(kx−ωt) (9.13)
For spherical waves in a homogeneous medium, neglecting near field terms, the
amplitude decays inversely with the distance x:
A0 i(kx−ωt)
A+ (x, t) = e (9.14)
x
This attenuation is also called geometrical damping. Factors for other types
of waves can be found in Table 9.1.
Amplitude decay with distance x

Wave type Point source Line source
−1
Body waves at depth x x−1/2
−2
Body waves at the surface x x−1
Rayleigh waves x −1/2
x0
Table 9.1: Geometrical damping; the amplitude at a distance x from the source
is proportional to the tabulated values (Studer & Koller (1997)).
In order to introduce material damping caused by a not perfectly elastic

medium, the wave number is allowed to be complex: k∗ = k + iα. Equation
(9.14) becomes:
A0 −αx ik x −iωt
A+ (x, t) = e e e (9.15)
x
with α the frequency dependent attenuation coefficient. The real factor of
Equation (9.15) characterizes the spatial variation of amplitudes of the wave
and is repeated here:
A0 −αx
A(x) = e (9.16)
x
If we consider signals measured at two distances x1 and x2 from the source,
where x2 is greater than x1 , then the ratio of the amplitudes is:
A2 x1 −α(x2 −x1 )
= e (9.17)
A1 x2
and after taking the natural logarithm:

A2 x1
ln = ln − α(x2 − x1 ) (9.18)
A1 x2
Equation (9.18) can be transformed to:

A1 x 1
ln A2 x 2
α= (9.19)
x2 − x 1
which leads with Equation (2.45):
2π f D
α= (9.20)
c
to the formulation of the damping ratio

A1 x 1
c ln A2 x 2
D= (9.21)
2πf x2 − x 1
If the dual cone is used and the technique is applied to s-waves with velocity
Vs , Equation (9.21) changes to:

AT x T
Vs ln AB xB
Ds = (9.22)
2πf xB − xT
xT and xB represent the travel path length from the source to the top, respec-
tively the bottom sensor array; AT is the amplitude of the frequency spectrum
of the signal of the top sensor, AB is the corresponding amplitude of the bottom
sensor.
The time records of a dual cone SCPT given in Figure 9.14 are used to
give an example of the described theory. Figure 9.18 shows, based on Equa-
tion (9.19), the calculated attenuation coefficient α versus the frequency. The
highest signal energy is in the frequency range between 90 and 130 Hz. This
can be seen in Figure 9.17. The curve of α shows, neglecting some scatter, a
quite linear slope at the discussed frequencies.
This slope appears in the diagram of the damping ratio Ds given in Figure
9.19 as a region of an only slightly changing Ds . The mean value between 90
and 130 Hz is in this case 2.7 %.
9.4.2 Spectral ratio slope method

The attenuation coefficient method can be adapted to make the results inde-
pendent from the selection of a factor for the geometrical attenuation that does
not depend on frequency.
0.5
0.4
0.3
Attenuation Coefficient, α [m ]
−1
0.2
0.1
−0.1
−0.2
−0.3
−0.4
−0.5
0 50 100 150 200 250 300
Frequency [Hz]
Figure 9.18: Attenuation coefficient for the example signals from Figure 9.14.
20
15 Mean value 90 − 130 Hz:

D = 2.67 %
10
Damping Ratio, D [%]
−5
−10
−15
−20
0 50 100 150 200 250 300
Frequency [Hz]
Figure 9.19: Damping ratio based on attenuation coefficient method, mean

value for the range of highest signal energy.
It can be assumed that the attenuation coefficient α depends linearly on

the frequency f for the frequency range of interest:
α=zf (9.23)
After introduction of Equation (9.23), Equation (9.18) is differentiated with
respect to f:

A2
d ln A 1
= −z(x2 − x1 ) (9.24)
df
It can be noted that the factor to correct the geometrical damping from
Equation (9.14) is eliminated by differentiating. Any geometric term affecting
the amplitude that does not depend on frequency will also be eliminated. This
will include the transmission and reflection corrections on layer interfaces if the
velocities are independent of frequency, which is the case if the frequency range
for the analysis is properly selected. If x1 is held constant and x2 (or simply
x) is varied, differentiating with respect to x gives:

A2
∂ 2 ln A 1
= −z (9.25)
∂f ∂x
Techniques which are based on the Equations (9.24) and (9.25) are called
spectral ratio slope methods (SRS) and are described among others in Udı́as
(1999), Buttkus (1991), Kudo & Shima (1981) and Stewart (1992). z is linked
to the damping ratio D and the geophysical properties quality factor Q and
dissipation factor Q−1 :
1 Q−1 zc
D= = = (9.26)
2Q 2 2π
D is generally assumed to be independent of the amplitude of the particle
motion at small strains and of the frequency. The latter means α is a linear
function of frequency.
The determination of the damping ratio for the case of the dual cone and
a s-wave follows from the Equations (9.25) and (9.26). The differentiation to
the length is simply replaced by the difference of the travel path length to the
bottom and top accelerometer ∆x = xB − xT :

AT
1 Q −1
Vs Vs d ln AB
Ds = = = z= (9.27)
2Q 2 2π 2π df ∆x
where f is frequency, AT is the amplitude of the frequency spectrum of the
signal of the top sensor installed in the cone, AB is the corresponding spectrum
of the bottom accelerometer, z is the slope of the logarithmic spectral ratio
divided by ∆x and Vs is the shear wave velocity.
Figure 9.20 shows the natural logarithm of the spectral ratio ln(AT /AB ), as
used in Equation (9.27), calculated based on the time histories of Figure 9.14.
The slope of the line in the frequency range 90 to 130 Hz is equal to z ∆x and
results in a damping ratio Ds of 7.7 %.
0.5
0.4 Mean slope 90 −130 Hz

related to D = 7.68 %
0.3
Logarithmic Ratio, ln (AT / AB) [−]
0.2
0.1
−0.1
−0.2
−0.3
−0.4
−0.5
0 50 100 150 200 250 300
Frequency [Hz]
Figure 9.20: Logarithmic amplitude ratio of the signals from Figure 9.14, fitted
slope for the range of highest signal energy, corresponding damping ratio.
9.4.3 Ideas to obtain and analyze hysteresis stress-strain-

curves
Strong motion records of earthquakes gathered by vertically arranged sensor
arrays can be used for the evaluation of site stress-strain hysteresis curves. They
provide the source for a calculation of the shear modulus and the damping ratio
based on observations of the slope of the hysteresis curve and on the loop area
according to Figure 2.2 and Equation (2.28).
A vertical down-hole sensor array consists of an accelerometer placed at the
surface and another at a certain depth. Arrays with a couple of accelerometers
at different depths are also frequently in use as depicted in Figure 9.21.
Different authors provide information on the analyzing procedures and the
results of their application. Elgamal et al. (1995) give a summary based on
multi-sensor arrays in Taiwan (Lotung site), Japan (Port Island, Kobe) and on
a site in California (Wildlife Refuge) with a single accelerometer at depth. The
site in Taiwan is treated in Zeghal et al. (1995) and the site in California in
Zeghal & Elgamal (1994) more detailed. Ghayamghamian & Kawakami (2000)
present results on five single sensor sites in Japan. Data from a multi-sensor
array close to Mexico City are used in Martı́nez-Carvajal et al. (2001).
Principles of the method
Elgamal et al. (1995) developed a procedure based on the following assumpti-

ons. The shear stress τ (z, t) at any level z may be expressed using the stress-free
Free ground
surface Accelerometers
a1 1
a2 z 2 ∆ z1
a3 3 ∆ z2
4 ∆ z3
a4
5 ∆ z3
a5
6 ∆ z4
a6
ai−1 i−1
i ∆ zi−1
ai
∆ zi
ai+1 i+1
Figure 9.21: Multi-accelerometer array (Zeghal et al. (1995)).
surface boundary condition and a shear beam model as:

Z z
τ (z, t) = ρ ü(ζ, t) dζ (9.28)
ζ=0
in which ρ is the mass density and ü(ζ, t) is the lateral acceleration at level ζ.
Because the acceleration is only known at the discrete points zi in depth where
accelerometers are placed, an assumption for the intermediate sections has to
be made. The most simple way is to apply a linear interpolation. The shear
stresses at levels zi and (zi−1 + zi )/2 reduce then to:
üi−1 + üi
τi (t) = τi−1 (t) + ρ ∆zi−1 (9.29)
2
3üi−1 + üi
τi−1/2 (t) = τi−1 (t) + ρ ∆zi−1 (9.30)
8
i = 2, 3, . . .
in which subscripts i and (i-1/2) refer to levels zi (of the ith accelerometer)
and (zi−1 + zi )/2 (halfway between accelerometers i and (i-1)) respectively,
τi = τ (zi , t), üi = ü(zi , t) and ∆zi is the distance between the accelerometers.
The corresponding shear strains at levels zi and (zi−1 + zi )/2 may be ex-
pressed as:

1 ∆zi−1
γi (t) = (ui+1 − ui ) ···
∆zi−1 + ∆zi ∆zi

∆zi
. . . +(ui − ui−1 ) (9.31)
∆zi−1
ui − ui−1
γi−1/2 (t) = (9.32)
∆zi−1
i = 2, 3, . . .
in which ui = u(zi , t) is the lateral displacement.

The displacement can be calculated through double integration of the re-
corded acceleration history ü(zi , t). It could be necessary to remove the base
line drifts caused by the integration. This can be done by removing of the very
low frequency parts by means of a high-pass filter or a detrend algorithm.
The result of the Equations (9.30) and (9.32) are time histories of shear
stress and shear strain. If the shear stress is drawn against the shear strain
the typical hysteresis loops appear. Because the shape of the loops is strain
dependent it is useful to select cycles with similar maximum strain range for
the further analysis by using a rectangular window on the stress and strain
histories.
A disadvantage of the calculation of the shear stress history in multi-sensor
arrays by means of Equation (9.30) is that the calculation for a certain depth
needs all previous stress histories at smaller depth. Possible errors will therefore
be cumulated from depth to depth.
The accuracy of the calculated shear stress and strain amplitudes is a func-
tion of the distance between the recording points and the recorded acceleration
wavelength. The discrete formulas can only capture waves below a certain fre-
quency, determined by the distance between the recordings. From the sampling
theorem follows, in a complete analogue way to a discrete sampling in time,
that the smallest possible wavelength λmin has to be at least the double of
the distance ∆z between the sampling points. Ghayamghamian & Kawakami
(2000) claim even a stricter rule, the highest frequency fmax should correspond
to a wavelength λmin four times the distance ∆z between the recording points
c
λmin = 4 ∆z fmax = (9.33)
λmin
in which c is the wave propagation velocity.

Based on this criterion stress and strain time histories have to be low-pass
filtered to remove high frequency components. It has to be ensured that the
filter introduces no phase shift to the signals therefore FIR (finite impulse
response) filters are preferred.
Typical upper edges of the passband for earthquakes and common transdu-
cer spacings in seismic arrays are between 1.7 and 6 Hz.
Transfer of the technique to the SCPT
The setup of the seismic cone penetration test has a lot of similarities to a
seismic down-hole array to record strong motion earthquake events. However,
there are a couple of differences which have to be taken into account before the
method can be used for SCPT’s. The main differences are:
1. There is usually no surface time record available in a normal SCPT.
2. The dual-cone provides only two simultaneously recorded time histories.
3. The generated waves by the SCPT source travel downwards whereas the
earthquake waves travel upwards from the bedrock to the surface.
4. The typical frequencies of an earthquake signal are much lower than the
range generated by the SCPT source.
5. The distances between the seismic transducers is in case of the dual-cone

approximately 1 m; the accelerometers in the earthquake arrays have a
much larger spacing.
The highest usable frequency for a SCPT can be calculated by Equation

(9.33). Assuming a wave velocity c = Vs = 150 . . . 350 m/s and ∆z = 1.033 m
the usable frequency range ends at fmax = 36.3 . . . 84.7 Hz. Because the most
signal energy can be found between about 50 and 150 Hz it is obvious that only
a part of the signal energy is actually in the usable range.
Because during the test the seismic cone is pushed into the ground, recor-
ded signals at different depth are available. If the seismic source is sufficiently
repeatable all signals can be collected and it can be assumed that they are
identically the same as if they would be measured simultaneously. Unfortuna-
tely the used mechanical swing hammer cannot provide a highly reproducible
source. A solution to overcome this problem could be to calculate the transfer
function between the signals obtained from both cone sensors. By means of
these transfer functions it should be possible to recreate an array of signals with
the same properties as if they would have been recorded at the same excitation
blow.
Figure 9.22 gives a schematic view of the transfer function technique. A
transfer function H1/2 between two signals a1 and a2 is the complex ratio of
their spectra A1 and A2 :
F F T (a2 ) A2
H1/2 = = (9.34)
F F T (a1 ) A1
Fourier transformation and inverse Fourier transformation are abbreviated

with FFT and IFFT, respectively.
signals.
Depth− SCPT time records of cons ecutive meas urement depths Ps eudo s imultaneous s ignals
Index (m) Surface (dual cone) Free ground
sensor a1,1 p1 = a1,1 surface
m=1
z
a1, 2 p 2 = a1, 2
2 A1,3
1.033 H 2/3 =
A1, 2 a1,3 a 2,3 p3 = a1,3
3
A2, 4
1.033 H 3/ 4 = p 4 = IFFT (P3 ⋅ H 3 / 4 )
Dual cone A2,3 a 2, 4 a 3, 4
4 = IFFT (A1,3 ⋅ H 3 / 4 )
sensors A3,5
1.033 H 4/5 =
A3, 4 a 3, 5 p5 = IFFT (P4 ⋅ H 4 / 5 )
5
= IFFT (A1,3 ⋅ H 3 / 4 ⋅ H 4 / 5 )
An , m = FFT ( a n , m )
Pm = FFT ( p m )
i−1 ai − 2,i −1 pi −1 = IFFT (Pi − 2 ⋅ H i − 2 / i −1 )
Ai − 2,i = ...
1.033 H i −1 / i =
Ai − 2,i −1
i ai − 2 , i pi = IFFT ( Pi −1 ⋅ H i −1 / i )
9.4. METHODS TO OBTAIN THE DAMPING RATIO
= IFFT ( A1,3 ⋅ H 3 / 4 ⋅ H 4 / 5 ⋅ ... ⋅ H i −1 / i )

[m] i
= IFFT A1,3 ( ∏ k =4
H k −1 / k )
Index SCPT −
n=1 2 3 i−2
sub−setup (n)
Figure 9.22: Use of transfer functions (H) to generate pseudo simultaneous

135
Application of the hysteresis technique to Waremme data

The described method is evaluated at the site in Waremme on data from the
sledgehammer excitation of SCPT5. Additionally to the dual cone, an accele-
rometer array consisting of one sensor for each coordinate axis is placed at the
surface in close proximity to the pushing rod of the cone. The transfer function
technique is used based on the recorded lateral signals in the direction of the
hammer blow (y-axis). Figure 9.23 shows the applied algorithm.
The signals are first visually checked and averaged as in the spectral-ratio
slope and the attenuation coefficient method. In the following step the matrix of
acceleration signals as obtained from one excitation hammer strike at all SCPT
depths simultaneously, is calculated by means of transfer functions. To obtain
stable pseudo simultaneous signals it is necessary to eliminate all frequency
components below 10 Hz using a high-pass filter.
A low-pass filter with an edge frequency of 60 Hz is applied at the same time
as the 10 Hz high-pass filter. It is supposed to remove high frequency parts
not correctly captured by the sensor array because of the distances between
the accelerometers. Following the SCPT profile of the site in Waremme (e.g.
Figure 6.8) the s-wave velocities are between 150 and 260 m/s. If a maximum
frequency of 60 Hz is present the smallest wavelengths are in the range of 2.5 to
4.3 m. However, λmin for the accelerometer interval of 1 m is according to the
rule in Equation (9.33) 4 m. The low-pass filter edge frequency has to be set
to 36.3 Hz to fulfill the rule. But in the frequency band below 36.3 Hz remains
only a very few amount of signal energy. Therefore the edge frequency is set
to 60 Hz to include more signal energy. This way seems acceptable because
Equation (9.33) includes some kind of safety factor. λmin = 2 ∆z = 2 m only
follows from the sampling theorem directly, which is fulfilled.
Because only signal parts close to the mean shear wave pulse are qualified to
draw meaningful stress-strain curves and to further improve the stability of the
calculated signals, the acceleration signals are windowed by means of a tapered
window. The window is the same as shown in Figure 9.39 and described in
9.5 with a length of 32.5 ms to both sides of the maximum of the acceleration
signal. The filtering and windowing is done before the transfer function process
and is repeated once more afterwards.
In a next step the acceleration signals are integrated with depth, star-
ting at the surface, and multiplied with the density. An averaged density
of ρ = 2040 kg/m3 is obtained by the performed laboratory tests on samples
from the site. Shear stress time histories are the result.
The shear strain histories need first the calculation of the displacement
histories by double integration of the acceleration signals. A filter with a pass
band between 10 and 60 Hz is applied after each integration step. In this
way signal drifts can be successfully removed. Figures 9.24 to 9.27 give the
acceleration signals, the calculated displacement time histories, shear strain
and shear stress histories drawn against the depth.
In the final step shear stress, τ , is plotted versus shear strain, γ. Complete
loops are selected and G-modulus and damping ratio are calculated based on
Input data from LabVIEW

data acquis tition program
Data preparation as for spectral ratio slope

and attenuation coefficient method
Data Format Conversion
from ASCII format of LabVIEW
−consideration of sensor sensitiv ity and
applied gain factors
ascii2mat.m
Vis ual check of the time records and

averaging
−1 to 5 signals av eraged under manual control
av ershowex.m
autoav erfiles.m
Generation of pseudo s imultaneous s ignals

−band−pass filter with edge frequencies 10 and
60 Hz, tapered window (extended−cosine−window)
+/− 32.5 ms arround the maximum
−calculation of the transfer function between the
cone sensor positions
−repetition of filtering and windowing
maketrans.m
Calculation of shear stres s and s hear s train

time histories
−double integration of the acceleration records to

displacement time histories, application of the
bandpass filter after each integration step
−differentiating of the surface sensor records to
acceleration for the case that geophones are
used
−calculation of stress and strain time histories
lintrans.m
Drawing and evaluation of the s tres s −s train−

hys teresis−loops
−manual selection of certain loops
−calculation of the related G−modulus and
Damping ratio htrans.m
histloops.m
ellipse.m
G and D
for the selected loop
Figure 9.23: Application of the hysteresis method to data from the Waremme
site, SCPT3.
Figure 9.24: Normalized acceleration histories versus depth, Waremme,

SCPT3.
Figure 9.25: Normalized displacement histories versus depth, Waremme,

SCPT3.
Figure 9.26: Normalized shear strain histories versus depth, Waremme, SCPT3.
Figure 9.27: Normalized shear pstress histories versus depth, Waremme,

SCPT3.
the measurement of areas following Equation (2.28).
Hysteresis-Loop I Hysteresis-Loop II
150 150
Loop I Loop II
t = 0.505 ... 0.524 s, t = 0.523 ... 0.544 s,
G = 87.127 MPa G = 63.436 MPa
100
Vs = 207 m/s (with ρ = 2040 kg/m³),
100 Vs = 176 m/s (with ρ = 2040 kg/m³),
D = 25.75 % D = 27.57 %
50 50
Shear stress, τ [Pa]

0 0
Loop I t = 0.544 s
t = 0.523 s
-50 -50
t = 0.524 s Loop II
-100 t = 0.505 s -100
-150 -150
-2 -1 0 1 2 -2 -1 0 1 2
Shear strain, γ [-] -6 Shear strain, γ [-] -6
x 10 x 10
Figure 9.28: Stress-strain hysteresis Figure 9.29: Stress-strain hysteresis

curves related to the time-histories in curves related to the time-histories in
Figure 9.30, loop I. Figure 9.30, loop II.
Figure 9.28 and 9.29 give the stress-strain diagrams at a depth of 3.71 m.
Two complete loops could be distinguished and are marked by thick lines in
the two figures. The found damping ratio is 25.7 to 27.6 % and unexpectedly
high. The calculated shear wave velocity is 176 to 207 m/s and in the range of
the cross correlation measurement on the SCPT data.
Figure 9.30 shows the location of the loops in the form of strain and stress
versus time.
Obstacles and difficulties in applying the technique

It is quite difficult to find usable loops for analysis. A majority of the stress-
strain diagrams show also in the parts of the strongest signal energy comple-
tely unshaped loops and even if acceptable portions are found the calculated
damping ratio is often above 100 %. Because of the questionable results the
technique is not elaboratly tried at the other sites.
The difficulties related to this technique are probably caused by the ne-
cessary quite strict filter and windowing regime. Critical error introducing
steps of the algorithm are listed in the following.
Transfer functions Preliminary trials to use the transfer function method

without any filtering and windowing led with increasing depth to dissatisfying
time histories. They are dominated by noise and very low frequency com-
ponents. From a certain depth on, the shear wave pulse is not distinguishable
any more although the real measured time record at the corresponding depth
is quite clear. Especially the windowing is virtually indispensable for the tech-
nique.
−6 Shear strain
x 10
2
Loop I
1
Shear strain, γ [−]
−1
Loop II
−2
0.45 0.475 0.5 0.525 0.55 0.575 0.6
Time, t [s]
Shear stress
200
Loop I
100
−100
Loop II
−200
0.45 0.475 0.5 0.525 0.55 0.575 0.6
Time, t [s]
Figure 9.30: Shear strain and shear stress time histories at a depth of 3.71 m
after sledgehammer excitation, Waremme, SCPT3.
A comparison of the calculated time histories with the actually measured

ones shows a small time shift. Explanation for this might be the multiplication
of small numerical errors in the transfer function from the surface over all cone
depths to the depth of interest. Another reason is the fact that the distance
between the cone sensors is actually 1.033 m but the cone has been pushed into
the ground in intervals of 0.5 m. That means the pseudo simultaneous signals
are calculated for a slightly longer travel path length, 3.3 cm per 1 m depth, as
in reality. However, in the later SCPT’s the cone pushing interval was changed
to 51.65 cm instead of 50 cm to compensate the deviation. Calculations of the
transfer function techniques on these tests still show some signal shift between
calculated and measured signals at greater depths.
An improvement of the seismic source, so that it can be considered as fully
reproducible, could make the transfer function approach avoidable with positive
effects on the results.
Acceleration integration The double integration of the pseudo simulta-

neous acceleration time histories adds other difficulties to be handled. After
each integration step the signal has to be filtered to remove the appearing si-
gnal drifts. In this case a rectangular high-pass filter (FIR) is used. This filter
provides a unit gain factor over the desired frequency pass band and introduces
no phase shift in the filtering procedure. Also an IIR filter with smoother filter
flanks designed after Ormsby (1961) is tried. Both filters are only efficient in
removing the disturbing signal drifts if all components below about 10 Hz are
significantly reduced or removed. This means of course a loss of important
signal information.
Stress-strain calculation The receiver spacing between the dual cone sen-
sors is not optimum. A low-pass filter has to be applied to remove frequency
components above 60 Hz. Unfortunately, the main wave energy is with 50 to
150 Hz in this frequency range.
9.5 SRS method and ACM on artificially gene-

rated signals
Preliminary trials of the SRS method on SCPT data show a wide result scat-
tering concerning the damping ratio. Because other in situ testing techniques
were not available it was not clear if this variation is due to real changes in the
soil properties with depth or caused by the algorithm of analyzing.
Therefore the seismic cone test is simulated using the program SPECTRAL
providing artificial input signals for the analyzing procedures.
The algorithm of analyzing applied on the measured signals at the testing
sites is also used to analyze the simulated signals. Essentially three artificial
soil profiles as given in Figure 9.31 are evaluated, a half-space, a one layer and
a two layer profile. Damping ratio, shear wave and compression wave velocity
are kept constant within one layer.
The geometrical conditions for a typical SCPT are also used for the simula-
ted tests as given in Figure 9.32. In particular a horizontal off-set of the source
of one meter and a depth distance between the simulated signal records of 0.5
m are chosen.
The excitation hammer blow is simulated by a point load on the surface.
In the cases of profile a) and c) the direction of the force is once horizontal
and once vertical. For profile b) only the horizontal excitation is used. The
behavior of the excitation force follows a squared sine function and is given in
Figure 9.33.
The calculation of the p-wave velocity is performed using the vertical par-
ticle acceleration caused by the simulated vertical impact. The s-wave velocity
is obtained based on the horizontal particle acceleration due to the horizontal
impact force.
The damping ratio is calculated for the shear wave only, therefore the ho-
rizontal acceleration in connection with the horizontal impact is used.
The Figures 9.34, 9.35 and 9.36 show the time histories used for the s-wave
velocity and the damping ratio calculation. The direct s- and p-wave compo-
9.5. ARTIFICIALLY GENERATED SIGNALS 143
+/− 0.00 m +/− 0.00 m +/− 0.00 m
VS = 107.5 m/s
VS = 107.5 m/s − 3.00 m
I VP = 215.0 m/s
VS = 107.5 m/s
VP = 215.0 m/s D = 3.0 %
I VP = 215.0 m/s
D = 1.0 % D = 1.0 %
VS = 215.0 m/s
− 7.00 m
II VP = 430.0 m/s
I D = 2.3 %
− 10.00 m
VS = 152.1 m/s
II VP = 304.2 m/s
D = 2.3 % VS = 107.5 m/s
III VP = 215.0 m/s
D = 1.0 %
− 15.00 m − 15.00 m − 15.00 m
a) Half−Space b) 1 Layer Profile c) 2 Layer Profile
Figure 9.31: Stratification and parameters of the simulated sites.
Excitation Cone
Axis Axis
1m
v ertical
+/− 0.00 m
horizontal
Source
0.5 m
Calculation
Points 0.5 m
− 15.00 m
Figure 9.32: Set-up of the simulated SCPT’s.

Time History Source Function

1
Normalised force [−]
0.8
0.6
0.4
0.2
0
10 15 20 25 30 35 40
t [ms]
Frequency Spectrum Source Function
1
Normalised force [−]
0.8
0.6
0.4
0.2
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
f [Hz]
Figure 9.33: Time history and frequency content of the simulated normalized
impact force.
nent is indicated. There is an unexpected component in profile b) between p-

and s-wave arrival assumed to be caused by an internal calculation instability
in the SPECTRAL program algorithm. Besides this, reflections on the layer
interfaces can be very well seen. Also nicely seen, at least in profile c), is
that the refraction of the s-wave component at a layer boundary generates an
additional wave component traveling with p-wave velocity.
The complex structure of the time records of the 2 layer profile c) emphasizes
the need to implement a possibility to select parts of the signals and not to
use the whole signal for the velocity and damping analysis. It is expected
that at least the common treatment of wave components of different speed
will cause significant errors in the calculation of the damping ratio as the wave
velocity is an essential part of the SRS method and the ACM. The wave velocity
calculation based on cross correlation should be not affected as long as the
amplitudes of the component of interest are higher than the others.
The analysis of the simulated signals is performed through a couple of MAT-
LAB script files. The algorithm is outlined in Figure 9.37. The SRS method
and the ACM are applied to signals from depths with one meter distance,
meter-by-meter approach. The SRS method is additionally tried on two si-
gnals from a depth of 5 and 12 m. The slope of the spectral logarithmic ratio
is obtained by a linear curve fit in a frequency range of 300 Hz always cente-
red at the frequency of the maximum spectral amplitude. The damping ratio
a) Half−Space
6
Depth [m]
S
8
10
12
14
16
0 0.05 0.1 0.15 0.2
t [s]
Figure 9.34: Simulated time histories for profile a) Half-Space, horizontal exci-
tation.
b) 1 Layer Profile
Layer I
4
P S
Depth [m]
10
Layer II
12
P S
14
16
0 0.05 0.1 0.15 0.2
t [s]
Figure 9.35: Simulated time histories for profile b) one layer on a half-space,
horizontal excitation.
c) 2 Layer Profile
Layer I
P S
2
Layer II
P S
Depth [m]
10
12
Layer III
P S
14
16
0 0.02 0.04 0.06 0.08 0.1 0.12
t [s]
Figure 9.36: Simulated time histories for profile c) two layers on a half-space,
horizontal excitation.
determined by the ACM is an average value for the same frequency range. To
study the influence of the different length of a cutting window, different shapes
of window functions and the meaning of additional noise to the results of the
half-space of profile a), some parameters are changed which will be explained
further on.
Input data from

SPECTRAL program
Data Format Adaption

from SPECTRAL format
makeSCPTfiles.m
Selection of s ignal parts

(rectangular window)
srsdatanew.m
Signal adaption
−Length adjustment of the windowed

signal part (E)
−Adding of noise (N)
−Applying of Hanning (H) or tapert (T) window
AdaptFFT.m
Meter−by−meter approach Selected depth area
Cros s −Correlation Cros s −Correlation
autoSCPTFromWinSig.m
autoSCPTFromWinSig3.m autoSCPTFromWinSig2.m
Vs for the selected
Vs v s. Depth
depth area
Spectral Ratio Attenuation Coefficient Spectral Ratio

Slope Method (SRS) Method (ACM) Slope Method (SRS)
dualmanu.m dualmanu3.m dualmanu2.m
D v s. Depth D v s. Depth D for the selected depth area
Figure 9.37: Flow chart for the calculations on the simulated SCPT’s.
Window-length
The signal from the half-space profile a) is analyzed using three different win-
dow lengths. First the complete signal without window is used, secondly a
signal of a length of the first 1.5 cycles of the s-wave component is used which
is virtually the whole s-wave component and finally a signal containing only
the first cycle of the s-wave component is analyzed. Figure 9.38 illustrates the
shape of the signal after the window is applied. The windowing is done manu-
ally by selecting the crossing points with the neutral position. In this way it
can happen that the two proceeded windowed signals have a slightly different
length. To investigate influences of this fact all calculations are done additio-
nally with signals of exactly the same length. For this case it is accepted that
one of the signals contains more than one respectively 1.5 cycles. Results of
exact equal signal length are indicated with ”E” in the summary Table 9.2.
1.5 Cycles
1 Cycle
Figure 9.38: Schematic view of the results after windowing of 1.5 cycles (1.5
periods) and one cycle (1 period) of the signal.
Window-function
Hanning window Additionally to the simple rectangular shaped window, a
tapered window and a Hanning, also called Hann window, is applied. The shape
of the window function is given in Figure 9.39. The Hanning window follows
a cosine function and is defined by Equation (9.35). n describes the length of
the window in number of samples; k is the sample index. The Hanning window
is a common window type and for instance used by Kudo & Shima (1981) for
damping measurements on a down-hole setup with two boreholes.

k
w(k + 1) = 0.5 1 − cos 2π , k = 0, . . . , n − 1 (9.35)
n−1
Tapered window The tapered window is a combination of Hanning window

and rectangular window. The abrupt jump from 0 to 1 of the rectangular
window is changed by the gentle slope of the Hanning window. The interval
between 0 on 1 is chosen to 2 ms. The tapered window called extended-cosine-
bell window is used by Mok et al. (1988) for the damping determination in
down-hole tests.
Results based on the Hanning window and the tapered window are indicated
with ”H” respectively with ”T” in the summary Table 9.2.
Benefits of windowing The reasons for applying a window to the signal

are essentially the following.
Most obvious and already mentioned is the need to select a certain wave
component if other components with high amplitudes are present. The origin
Time domain Frequency domain

50
Hanning
1
T apered
Rectangular
0.8
0
Magnitude (dB)
Amplitude
0.6
0.4
-50
0.2
0 -100
0 5 10 15 0 0.2 0.4 0.6 0.8
Time (ms) Frequency (kHz)
Figure 9.39: Hanning, tapered and rectangular window with a typical length
of 1.5 cycles in time and frequency domain.
of these wave components could be a reflection or refraction at a layer interface

or the surface, other wave sources in the neighborhood or simply different types
of wave propagation, like p- or s-waves.
A second reason to apply a window is the idea to eliminate portions in the
signal containing only noise. These are large parts of the signal before and after
the actual vibration event. In this way the disturbing spectral parts, related
to the noise in the amplitude spectrum, are reduced.
The third reason is related to the signal processing and the calculation
of the FFT spectrum. This is also the reason why windowing is of such an
importance in digital and analogue signal processing. If a periodic signal of
infinite length has to be analyzed to obtain its frequency content it is obvious
that a part of the signal has to be selected. If this is done completely with
arbitrary start and end points it is possible that at the borders of the selection
the signal is not close to the zero line. Because the FFT algorithm assumes
a periodic continuation of the signal after the cutting point, discontinuities
in the signal progression occur. This again leads in the frequency spectrum
to a ”leakage” of the peaks. That means parts of the energy of a frequency
component are not shown at the original frequency, they are included in the
broadened main peak and in ripples beside the main peak. This is the so called
Gibb’s phenomenon. To prevent the spectral leakage, virtually two ways can
be gone: a) the signal is cut (rectangular shaped window) at points of the
zero-crossing respectively parts where the signal is approximately zero or b) a
window function is applied which diminishes the signal amplitude towards the
borders of the window. Method a) uses an effect called ”picket fence”, so the
ripple effects are essentially multiplied with zero. Method b) is more general
to use, however the signal is altered by the window function which can cause
unpredictable effects.
One can imagine the application of a window on a signal in the frequency

domain by the convolution of the ideal spectrum of the infinite long signal
and the spectrum of the window function. Further details can be found in the
literature on digital signal processing as for instance in Best (1991).
The signals of the vibrations caused by the SCPT are not periodic, they
are transient. That means for analysis of the full signal length, windowing is
not necessary or more exact the selection of another window type than the
rectangular shaped window needs not to be done. But if for the first two
reasons parts of the signals should be blind out, the influence of the Gibb’s
phenomenon has to be considered. Therefore the signals are manually, with
best possible accuracy, cut at the zero-crossing points and additionally the
effects of a Hanning and a tapered window are studied.
Additional noise
Results marked with ”N” have been obtained based on signals where noise with
a maximum amplitude of 10 % of the maximum peak particle acceleration of
the signal was added.
9.5.1 Calculated velocities from the simulated signals
Simulated half−space
0
2.5
Vp
Depth [m]
7.5
Vs
10
12.5
15
0 50 100 150 200 250 300
Wave velocity [m/s]
Figure 9.40: Calculated wave velocities, profile a), half-space.
Figures 9.40, 9.41 and 9.42 give s- and p-wave velocity as calculated by the
cross correlation technique. The curves are virtually independent of the length
Simulated 1 layer profile Simulated 2 layer profile

0 0
2.5 2.5
Vs
5 5
Depth [m]
Depth [m]
7.5 7.5
V
s
10 10
V
p
12.5 12.5
15 15
0 50 100 150 200 250 300 0 100 200 300 400 500
Wave velocity [m/s] Wave velocity [m/s]
Figure 9.41: Calculated s-wave ve- Figure 9.42: Calculated wave velo-
locity, profile b), one layer on a half- city, profile c), two layers on a half-
space. space.
and shape of the applied window as long as the wave component of interest
has a higher amplitude than all others. The input velocities are nearly exactly
found except for the shallow depths smaller than 2.5 m in case of the profile
a). This is probably due to the fact that the difference in travel path length
between two measurement points close to the surface is rather small, so small
deviations in the time arrival determination have an unequal higher disturbing
influence in the resulting velocity.
9.5.2 Calculated damping ratio

The input parameter for the damping ratio is Ds = Dp = 1% over the whole
depth of profile a). Table 9.2 and Figure 9.44 summarize the results concerning
the damping ratio for the half-space profile a). The Figures 9.45 to 9.51 show
the results for selected parameter configurations. Table 9.2 gives the value of
the averaged damping ratio over all n obtained points of the profile between 5.5
and 11.5 m, D̄, and as a measure for scattering, the mean absolute deviation
s∗D0 from the target value D0 = 1%. The latter is defined as:
n
1 X
s∗D0 = |Di − D0 | . (9.36)
n i=1
In the case where the signals at 5 and 12 m are compared the single calcu-
lated value D is given. s∗D0 is then simply the deviation from the target input
value.
As an example Figure 9.43 shows for the calculation point at 10 m depth the
influence of the length of the rectangular window to the shape of the amplitude
spectra of the signals and the logarithmic ratio between both.
Spectrum at depth 10.5 m, A

B
1 Full signal (a)

Amplitude [−]
1.5 cycles (b)

Normalized
1 cycle (c)
0.5
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Spectrum at depth 9.5 m, AT
1
Amplitude [−]
Normalized
0.5
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Logarithmic Ratio, ln(A /A )
T B
1.5
Range of slope calculation,
Logarithmic
1 +/− 150 Hz from the peak−frequency

Ratio [−]
0.5
0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Frequency [Hz]
Figure 9.43: Influence of the length of a rectangular shaped window on the

amplitude spectra and the logarithmic amplitude ratio, half-space profile a),
calculation depth 10 m.
Evaluation of the results for the half-space

a) Full signal
The SRS method and ACM give for depths longer than about 2.5 m and
the rectangular window nearly perfect results. This can be seen in Figure
9.45. The application of a window function is nearly meaningless because
applied over the full signal almost all significant parts of the signal stay
virtually unchanged. The reason for the deviation in the case of the
combination Hanning window and ACM, D̄ ≈ 0.5% is not known.
cN
cET
cEH
1 cycle
cT
cH
cE
c
bN
bET
1.5 cycles
bEH
bT
bH
bE
b
aN Selected-Depth, SRS
Full signal
aT Metre-by-Metre, ACM
aH Metre-by-Metre, SRS
a
0.000 0.200 0.400 0.600 0.800 1.000

Mean absolute deviation s* [%]
Figure 9.44: Mean absolute deviation of the damping ratio profiles obtained
from the simulated SCPT’s.
Modified Parameters Metre-by-Metre Selected-Depth

Method Area Method
SRS ACM SRS
Equal Window length
Hanning Window
Tapered Window
Window length
Id-Code
Noise
D̄ s∗D0 D̄ s∗D0 D s∗D0

[%] [%] [%] [%] [%] [%]
a) Full signal
a 1.003 0.009 0.975 0.025 1.002 0.002

x aH 1.002 0.005 0.505 0.495 1.002 0.002
x aT 1.003 0.009 0.975 0.025 1.002 0.002
x aN 1.025 0.225 0.968 0.111 1.063 0.063
b 1.137 0.147 1.895 0.945 1.029 0.029

b) 1.5 cycles)
x bE 0.987 0.032 0.978 0.024 1.108 0.108

x bH 0.869 0.148 1.644 0.665 0.815 0.185
x bT 1.360 0.398 1.842 0.888 1.089 0.089
x x bEH 0.682 0.318 1.037 0.044 0.718 0.282
x x bET 0.910 0.092 0.983 0.022 1.174 0.174
x bN 1.139 0.432 1.896 0.985 1.078 0.078
c 0.580 0.566 1.281 0.476 0.570 0.430

x cE 0.469 0.645 1.173 0.179 0.774 0.226
c) 1 cycle
x cH 0.360 0.640 0.993 0.581 0.480 0.520

x cT 0.496 0.593 1.034 0.414 0.561 0.439
x x cEH 0.271 0.729 1.031 0.466 0.569 0.431
x x cET 0.329 0.671 1.108 0.395 0.761 0.239
x cN 0.584 0.586 1.273 0.483 0.638 0.362
Table 9.2: Evaluation of modified window parameters.
b) 1.5 cycles
The SRS method provides for this shorter window length the best results
if the rectangular shaped window is used. If care is taken for an exact
equal window length in both signals the result can be improved slightly.
The application of Hanning or tapered window is not able to better the
average value nor the mean absolute deviation. The ACM delivers in
general more inaccurate results than the SRS method only in the case of
an exact equal window length the 1 % damping can be approximatelly
found. Figures 9.46 and Figure 9.47 deliver selected calculated damping
profiles for 1.5 cycle window length.
c) 1 cycle
Finally for the shortest window, profiles can be found in Figure 9.48 and
Figure 9.49. The SRS methods gives, independent from the window type,
too low values for the damping ratio around 0.5 %. Better results come
from ACM but also here the scatter is high as seen in the SRS technique.
The closest proximity of the averaged damping ratio to the input damping
of 1 % is reached using an equal window length and a hanning window.
Evaluation of the results for the noise comparison

The introduced noise has some influence to the results. Example profiles com-
paring the results from the rectangular window with different window length
with and without the additional 10 % noise are given in the Figures 9.50 and
9.51. Differences are not really significiant.
Evaluation of the results for the profiles b) and c)

Figure 9.52 and 9.53 give the damping ratio profiles for the more complicated
soil profiles b) and c) for the SRS technique and ACM in case of a rectangular
window with and without equal length. Especially in profile c) in the range of
the layer change the deviation from the input damping ratios and the scatter
are high.
The direct traveling shear wave, indicated with ”S” in the Figures 9.35 and
9.36, is separated by the window. The length of the window is set as large as
possible.
9.5.3 Conclusions for the analysis of real SCPT data

This parameter study has shown that overall the influence of the type of window
is rather small. It seems useful to choose an exact equal window length for both
signals. However, the main conclusion is, that the length of the applied window
has the most significant influence on the result and its scatter. As shorter the
window is chosen as larger the deviation and as larger the scatter too.
Simulated half−space, full signal

0
2.5
5
Depth [m]
7.5
10
12.5
SRS a
ACM a
15
−0.5 0 0.5 1 1.5 2 2.5 3
Damping ratio [%]
Figure 9.45: Damping ratio, profile a), based on the full signal.
Simulated half−space, 1.5 cycles, SRS Simulated half−space, 1.5 cycles, ACM
0 0
2.5 2.5
5 5
Depth [m]
Depth [m]
7.5 7.5
10 10 ACM b
SRS b ACM bE
SRS bE ACM bH
SRS bH ACM bT
SRS bT
12.5 12.5
15 15
−0.5 0 0.5 1 1.5 2 2.5 3 −0.5 0 0.5 1 1.5 2 2.5 3
Figure 9.46: Damping ratio, profile Figure 9.47: Damping ratio, profile
a), based on a windowed signal of 1.5 a), based on a windowed signal of 1.5
cycle length, spectral ratio slope me- cycle length, attenuation coefficient
thod. method.
Simulated half−space, 1 cycle, SRS Simulated half−space, 1 cycle, ACM

0 0
2.5 2.5
5 5
Depth [m]
Depth [m]
7.5 7.5
SRS c ACM c
SRS cE ACM cE
SRS cH ACM cH
10 10 ACM cT
SRS cT
12.5 12.5
15 15
−0.5 0 0.5 1 1.5 2 2.5 3 −0.5 0 0.5 1 1.5 2 2.5 3
a), based on a windowed signal of 1 a), based on a windowed signal of 1 cy-
cycle length, spectral ratio slope me- cle length, attenuation coefficient me-
thod. thod.
Simulated half−space, noise, SRS Simulated half−space, noise, ACM

0 0
2.5 2.5
5 5
Depth [m]
Depth [m]
7.5 SRS a 7.5
SRS aN
SRS b
SRS bN ACM a
SRS c ACM aN
SRS cN ACM b
10 10
ACM bN
ACM c
ACM cN
12.5 12.5
15 15
−0.5 0 0.5 1 1.5 2 2.5 3 −0.5 0 0.5 1 1.5 2 2.5 3
a), noise influence, spectral ratio slope a), noise influence, attenuation coef-
method. ficient method.
Simulated 1 layer profile Simulated 2 layer profile

0 0
SRS
D = 3.0 %
D = 1.0 % SRS E
ACM
ACM E
2.5 2.5
5 5
D = 2.3 %
Depth [m]
Depth [m]
7.5 7.5
D = 2.3 %
10 SRS 10
SRS E
ACM
ACM E
12.5 12.5 D = 1.0 %
15 15
−5 0 5 10 15 −5 0 5 10 15
b), selected results in comparison with c), selected results in comparison with
the input values. the input values.
9.6 Measurements performed at the test sites

Details on the practical test setup on the five research sites can be found in the
Chapters 4 to 8. They include also results for the wave velocities and damping
ratio profiles. The SCPT’s are divided in three measurement campaigns: cam-
paign I) includes tests with the geophone equipped single cone, campaign II)
are tests with the dual cone and the Hamlin accelerometers and campaign III)
covers a SCPT with the dual cone and the Brüel & Kjær accelerometers.
Site Campaign
I II III
Retie SCPT1 to SCPT4 SCPT5
Lincent SCPT1 to SCPT3 SCPT4 to SCPT5
Waremme SCPT1 to SCPT2 SCPT3 to SCPT5
Sint-Katelijne-Waver SCPT1 to SCPT2
Ghent SCPT1
Table 9.3: Assignment of the performed SCPT’s to the measurement cam-

paigns.
9.6.1 Campaign I (geophones)

The first tests were performed using the single cone. The arrival time of the
s-wave component and exceptionally the p-wave velocity is visually determined.
Because the resonant frequency of the geophones lays in the frequency range
of interest the test data are not used for a damping calculation. It would also
be necessary to calculate the ratio of the spectra of time records related to
different hammer blows, because of only one sensor array in the cone. Because
of the limited reproducibility of the mechanical swing hammer this method is
less promising than measurements with two sensors at the same time
9.6.2 Campaign II (HAMLIN accelerometers)

The Hamlin accelerometers installed in the dual cone are used for these mea-
surements.
Algorithm of data processing

The analysis of the test data follows in general the flow chart in Figure 9.54.
The SRS method coupled with the meter-by-meter approach is applied ac-
cording to the algorithm used for the simulated SCPT’s. The attenuation
coefficient method is not tried.
Because the real measurements contain noise and other influences additional
steps for the data preparation are added.
9.6. MEASUREMENTS PERFORMED AT THE TEST SITES 161
Input data from LabVIEW

data acquistition program
Data Format Conversion

from ASCII format of LabVIEW
−consideration of sensor sensitivity and
applied gain factors
ascii2mat.m
Visual check of the time records and

averaging
−1 to 5 signals averaged under manual control
avershowex.m
autoaverfiles.m
Windowing / Filtering
−manual / semi−automatic signal−part−selection
(rectangular window)
−poss. low−pass filtering (200 Hz cut−off frequency)
srsdatanew.m
Meter−by−meter approach
Cross−Correlation
autoSCPTFromWinSig.m
autoSCPTFromWinSig3.m
Vs vs. Depth
Spectral Ratio
Slope Method (SRS)
dualmanu.m
D vs. Depth
Figure 9.54: Flow chart for the calculations on the data from the SCPT testing
sites, SRS method.
After the conversion of the ASCII data, gathered with the LabVIEW-
program, taking into account the sensors sensitivity and possibly applied gain
factors by the conditioner equipment, all signals are visually checked. Usually
5 repetition of excitation hammer blows at each cone depth are recorded. This
opens the possibility to skip obviously disturbed hammer blows that have been
overlooked during the measurement.
The remaining time records are averaged in the time domain to improve
the signal to noise ratio. A trial with averaging in the frequency domain led to
virtually the same averaged signals.
However, even after averaging of up to 5 signals the obtained time records
have still remarkable noise components. Figure 9.57 shows an averaged time
record from the site in Sint-Katelijne-Waver. Figure 9.55 shows the averaged
signals according to their depth.
0 0
2 2
4 4
6 6
Depth [m]
Depth [m]
8 8
10 10
12 12
14 14
16 16
18 18
0.45 0.5 0.55 0.6 0.65 0.7 0.45 0.5 0.55 0.6 0.65 0.7
t [s] t [s]
Figure 9.55: Sint-Katelijne-Waver, Figure 9.56: Ghent, bottom element,

bottom element, y-axis, sledge ham- y-axis, sledge hammer, SCPT1.
mer, SCPT2.
The high noise level suggests the application of a low-pass filter. In this
way it is more straightforward to find reasonable points for the beginning and
the end of the rectangular window. A corner frequency of 200 Hz is the best
compromise between smoothing the signal and keeping its frequency content.
The filtering is done before the signals are windowed.
Afterwards the main cycle, following the same definition given in Figure
9.38 (1 cycle), of every signal is cut by a rectangular window.
Because most of the signal energy is situated in the frequency range between
Unfiltered, averaged signal

0.1
0.05
−0.05
−0.1
0.55 0.555 0.56 0.565 0.57 0.575 0.58 0.585 0.59 0.595 0.6
Time [s]
200 Hz−low−pass filtered, averaged signal

0.1
0.05
−0.05
−0.1
0.55 0.555 0.56 0.565 0.57 0.575 0.58 0.585 0.59 0.595 0.6
Time [s]
Figure 9.57: Example of averaged time histories from the top accelerometer,
y-direction, Depth 11.67 m, Sint-Katelijne-Waver (SCPT2) before and after
filtering.
75 and 130 Hz, the slope of the logarithmic spectral ratio is obtained by fitting
a straight line in this range.
The s-wave velocity calculated by the cross correlation technique is found
to be independent of the application of the low-pass filter. Also if the signal is
not windowed the velocity profile stays the same.
The results of s-wave velocity and damping ratio are given in the Figures
9.58 to 9.65. The same results are already included in the previous chapters
on the test sites Retie, Lincent, Waremme and Sint-Katelijne-Waver, Chapter
4 to 7.
Retie The site in Retie is characterized by a sand profile. The soil changes at
about 5 to 6 m depth from a fine sand to the coarser sand of the formation of
Mol. This boundary layer is confirmed by a change in the shear wave velocity
from averaged 215 to 335 m/s.
The results of the damping ratio show major variation. The shallow layer
until about 6 m depth has a mean damping ratio of 4 %, below this layer the
value exceeds 10 %. This means the layer interface fine sand to sand is also
identifiable in the damping ratio measurements.
Although comparative damping values found in literature, based on labo-
ratory tests and a shear strain of smaller than 10−3 %, differ a lot, a damping
ratio of more than 10 % is not expected. Seed & Idriss (1970) designate a D
of 1 to 4 % as a resonable value for the low strain range in sands.
Retie Retie
0
0
Mechanical Hammer
Sledge Hammer
2 2
4 4
6
Depth [m]
Depth [m]
8 8
10 10
12 12
Mechanical Hammer (SCPT)
Sledge Hammer (SCPT)
SASW
14 14
0 50 100 150 200 250 300 350 400 0 10 20 30 40
S−Wave Velocity [m/s] Damping Ratio [%]
Figure 9.58: Shear wave velocity Vs , Figure 9.59: Damping ratio Ds , site
site in Retie, campaign II. in Retie, campaign II.
Lincent The site in Lincent shows a quite complicated stratification reaching

from clay to fine sand. Extraordinary is the stiff layer between 3.2 and 7.5 m
consisting of the rock residual arenite embedded in clay. The shear wave pro-
file shows a nearly uniform increase with depth. The stiff layer cannot be
recognized.
The calculated damping ratio is spread between 0 and 6 % over the whole
testing depth. There is no specific trend with the depth observable.
Lincent
0 Lincent
0
Mechanical Hammer
Sledge Hammer
1
1
2
2
3 3
Depth [m]
Depth [m]
4 4
5 5
6 6
7 7

8 Sledge Hammer (SCPT) 8
SASW
0 50 100 150 200 250 300 350 400 0 2 4 6 8 10
site in Lincent, campaign II. in Lincent, campaign II.
Waremme The testing site in Waremme has a homogeneous layer of silt

on top of an underlaying bedrock of chalk. The shear wave velocity increases
monotonously with the depth after some disturbances in the top 3 m.
The damping ratio is mostly between 0 and 3 %. These are values typical
for a silt according to some authors, for instance Kudo & Shima (1981). A
slight increase of the damping ratio with depth can be recognized.
The indicated results of resonant column and free torsion pendulum tests at
the sampling depths, a reference shear strain of 10−5 and an isotropic confining
stress equal to the mean effective stress in situ, show a reasonable agreement
with the SCPT results. The SASW profile gives a slightly lower Vs -value than
the SCPT.
Sint-Katelijne-Waver The site Sint-Katelijne-Waver consists of a clay layer

below a shallow layer of clayey sand. The boundary between both is at about
Waremme Waremme
0 0
2 2
4 4
6 6
Depth [m]
Depth [m]
8 8
10 10
12 12

Sledge Hammer (SCPT) Mechanical Hammer
14 14
SASW Sledge Hammer
Resonant Column Test Resonant Column Test
Free Torsion Pendulum Test Free Torsion Pendulum Test
16 16
0 50 100 150 200 250 300 350 0 2 4 6 8 10
site in Waremme, campaign II. in Waremme, campaign II.
3 m depth. It is clearly recognizable in the Vs profile by a sudden decrease in

Vs .
The damping ratio is in the top layer between 0 and 2 %. The deeper layer
exceeds this range until approximately 5 %. A typical range for a clay is given
for instance by Mok et al. (1988). This range is 4 to 7 % agreeing with the
SCPT results.
A resonant column test and a free torsion pendulum test, performed on
samples from 4.0 to 4.5 m depth, show a good agreement to the Vs and the D
calculated by the SCPT data.
9.6.3 Campaign III, Ghent (Brüel & Kjær accelerome-

ters)
The SCPT on the campus of the Ghent University is done with the Brüel &
Kjær accelerometers, providing a much higher resolution than the Hamlin sen-
sors and so an improved signal-to-noise ratio. Figure 9.56 shows the normalized
time records of the test in Ghent. The difference to the time histories obtained
with the Hamlin sensors on the site in Sint-Katelijne-Waver given on the left
side is quite obvious. While the test in Sint-Katelijne-Waver had to be stopped
at a depth of 13 m because of a bad signal quality, the test in Ghent provides
until the final depth of about 18 m signals with very small noise components.
Because of this good signal quality no low-pass filter has to be applied. Re-
garding to the experiences with small window lengths on the simulated signals,
Sint Katelijne−Waver Sint Katelijne−Waver

0 0
1 1
2 2
3 3
4 4
Depth [m]
Depth [m]
5 5
6 6
7 7
8 8
9 9
Mechanical Hammer (SCPT) Mechanical Hammer (SCPT)
Sledge Hammer (SCPT) Sledge Hammer (SCPT)
10 10
Resonant Column Test Resonant Column Test
Free Torsion Pendulum Test Free Torsion Pendulum Test
11 11
0 50 100 150 200 250 300 0 2 4 6 8 10
site in Sint-Katelijne-Waver, cam- in Sint-Katelijne-Waver, campaign
paign II. II.
different window lengths are tried. The procedure is the same as for campaign
II given in Figure 9.54. The averaging of up to 5 signals is done in the same way
as in campaign II. The slope of the logarithmic spectral ratio is obtained by
fitting a straight line in a range of 150 Hz on both sides of the amplitude peak
of the frequency spectrum. This is the same range as used in the simulated
signal analysis. Because the frequency of maximum energy is in general below
150 Hz, the application of the rule leads practically to a range from 0 to about
250 Hz.
The results of the s-wave velocity given in Figure 9.68 (identical with Figure
8.3) are again independent from the applied window. The damping ratio in
Figure 9.69 (identical with Figure 8.4) is calculated with the SRS method
with meter-by-meter approach and no window is used, so averaged signals are
analyzed without further modifications on the full signals.
For comparative reasons the SRS method is also applied with rectangular
windowed signals of one cycle length and on the full signal. Additionally a
rectangular window is tried with 0 from the beginning of the time record to the
first appearance of the shear wave component and 1 for the rest of the signal.
This window aims to exclude components arriving before the shear wave. Such
fast components, possibly p-waves, with remarkable amplitudes are found at
some depths.
As an example Figure 9.67 shows the amplitude spectra used for the dam-
ping calculation at the depth of 12.71 m. Figure 9.66 gives the resulting profiles
of the damping ratio obtained after applying the three window lengths.
Ghent, SCPT 1
0
2.5
7.5
10
Depth [m]
12.5
15
17.5
Full signal
2nd part
20
1 cycle
22.5
−10 0 10 20 30
Damping ratio [%]
Figure 9.66: Influence of the length of the rectangular window to the damping
profile on the Ghent site.
The profile based on the full signal agrees essentially with the profile based
on the second part of the signal, at least for greater depths. The profile of the
one-cycle-window shows some deviation.
The site in Ghent, the only one covered by campaign III, has an extended
layer of fine sand. From about 8 m depth to greater depths it changes stepwise
to a silt. The shear wave velocity increases with depth in the superficial fine
sand. In the underlaying slightly coarser material it is almost constantly 250
m/s.
The damping ratio is again unexpectedly high. Even though the majority of
the calculated values is below 10 % some of them fall also in the range between
10 and 20 %. The damping ratio seems to be not depth dependent.
It is remarkable that the damping ratio at both sites dominated by sand,
Retie and Ghent, has unexpected high values, not agreeing with values given
by other authors. However, the results at the sites characterized by silt and
clay give reasonable result ranges.
Beside this it cannot be hidden that the result scattering is not satisfying.
9.7. SUMMARY AND REMAINING PROBLEMS 169
Spectrum at depth 13.22 m, AB

Normalized Amplitude [−]
1
Full signal
2nd part
1 cycle
0.5
0
0 50 100 150 200 250 300
Spectrum at depth 12.19 m, AT
Normalized Amplitude [−]
0.5
0
0 50 100 150 200 250 300
Logarithmic Ratio, ln(AT/AB)
1
Logarithmic Ratio [−]
Range of slope calculation,

0 until the peak−frequency + 150 Hz
0.5
−0.5
0 50 100 150 200 250 300
Frequency [Hz]
Figure 9.67: Influence of the length of an applied rectangular shaped window

on the amplitude spectra and the logarithmic amplitude ratio, Ghent SCPT1,
calculation depth 12.71 m.
9.7 Summary and remaining problems

The selection of a suitable sensor equipment is more complex than expected.
A geophone shows to be not capable to measure vibration levels accurately be-
cause of its frequency dependent response. The first accelerometer selected has
a resolution not sufficient to reach greater depths. This problem can be solved
by another change of the sensors. An accelerometer with a higher resolution is
installed improving the signal quality significantly.
Amongst the studied methods to determine the s-wave arrival the cross
correlation in the time domain is quite fast selected as the preferable method.
In connection with an artificial increase of the sampling frequency accurate s-
wave velocities can be calculated by a fast automated procedure. This is shown
by SCPT’s on the five test sites.
Three methods are considered to calculate the damping ratio out of the
SCPT time records. Two techniques are based on the ratio of spectral ampli-
tudes of time records of the same vibration event gathered at different source
distances (attenuation coefficient method (ACM), spectral ratio slope method
Ghent
0 Ghent
Left side 0
Right side
2
2
4 4
6 6
Depth [m]
Depth [m]
8 8
10 10
Left side
12 12 Right side
14 14
16 16
18 18
0 50 100 150 200 250 300 350 400 0 5 10 15 20 25 30
site in Ghent, campaign III. in Ghent, campaign III.
(SRS method)). The third approach is based on the analysis of stress-strain

curves derived by the recorded time histories assuming a shear beam model of
the soil.
Artificially generated time records are generated for three fictive site stra-
tifications. The procedures for the ACM and the SRS method are applied on
these simulated SCPT data. A variation of the used window length and type
to select the direct shear wave components indicates that the window length
should be chosen as long as possible. An influence of the type of window
function cannot be found.
The spectral ratio slope method is applied on the five test sites. Concerning
the damping ratio the results of the field tests show some scatter. This still
has to be improved before the method can be used as a reasonable source for
numerical models to predict and evaluate vibrations. However, by taking in
consideration that there are nearly no other, at least no cost-equivalent, field
tests available to determine the damping ratio, it seems desirable to develop
this techniques further in the future.
A trail to calculate stress-strain hysteresis loops out of the signals from a
SCPT from the site in Waremme shows limited success. Even though a number
of loops are well shaped, the obtained damping ratios are beyond the expected
range. Reasons are assumed to be a to large receiver distance and the algorithm
of data analysis, which needs double integration of the time records and adds
small errors caused by signal noise from the surface to the calculation depth.
9.7. SUMMARY AND REMAINING PROBLEMS 171
Caparison tests using the resonant column and the free torsion pendulum
tests on samples from Waremme and Sint-Katelijne-Waver show a good agree-
ment with the SCPT’s concerning the s-wave velocity. The range of damping
ratios obtained by the SRS method matches with the results from the labora-
tory tests.
An outline of the main findings and conclusions:

• Accelerometers have because of their flat response in the low frequency
range in contrast to miniature geophones advantages if accurate vibration
measurements are needed. The finally installed Brüel & Kjær accelero-
meters are highly suitable for this purpose.
• The averaging of time histories is an important tool to improve the signal
to noise ratio.
• Vs can be accurately determined by the cross correlation technique.
• It is questionable if a mechanical swing hammer with an adjustable fal-
ling height can be considered as a highly repeatable source. An increasing
settlement of the source beam during a SCPT changes the coupling stress
beam-soil and therewith the energy transfer characteristic from the ham-
mer to the soil.
• An artificial increase of the sampling frequency to 100 kHz improves
the Vs determination. By this way the minimum sampling frequency
of 20 kHz, requested by the guideline for SCPT of the ISSMGE TC 10
(2004), can be fulfilled even if the physical sampling frequency is far
below.
• The application of a window function to a time record should be restric-
ted. The s-wave velocity determination on real measured signals needed
no windowing in the case of the testing sites. The application of the SRS
method to calculate the damping ratio needs a windowing but it should
be as long as possible. It is useful to choose exactly the same length for
both compared time histories.
• A window function other than the rectangular window brings no further
improvement.
Chapter 10
Bender element technique
10.1 Motivation and purpose

To characterize soils dynamically, piezoceramic elements have been increasingly
used in the last 30 years. In an early stage piezoceramics were mainly used to
generate and receive compression waves. Since little information about the soil
structure can be obtained and the p-wave velocities are highly influenced by the
presence of pore fluid, the piezoceramics have been combined in different forms
to generate and receive shear waves. Such combined forms of piezoceramics are
known as bender elements.
In most of the papers presented in literature bender elements are used to
measure wave velocity in the time domain. In these methods a pulse is emitted
by a bender element and the travel time is determined when the pulse arrives
at the second bender element, with the tip located at a known distance from
the tip of the emitter.
The shear strain of a pulse generated by a bender element is less than
10−3 %, falling in the elastic strain range of soil.
Objectives and methodology

The objectives of the part of the research devoted to the bender elements is at
first to select a reliable method to measure the s-wave velocity. Based on that,
procedures should be investigated to measure also the damping ratio by means
of bender elements. Such a possibility would be an important enhancement of
the bender element technique.
In order to achieve the objectives the following topics are treated and des-
cribed in this chapter:
- Description of bender elements and test equipment
- Summary of methods to determine the time of wave travel using time

and frequency domain techniques
173
174 CHAPTER 10. BENDER ELEMENT TECHNIQUE
- Discussion on problems which may affect the identification of the arrival

time (p-wave, components, near field effects, inductive coupling between
sender and receiver element)
- Theoretical background of a modal test of a soil sample installed in a
triaxial cell to determine the damping ratio (The test is later called bender
element resonant technique.)
- Analytical and numerical prediction of the natural frequencies of the soil
sample
- Experimental and theoretical studies on the influence of the confining
medium on the natural frequencies of the specimen
- Tests on undisturbed material from the sites in Waremme and Sint-
Katelijne-Waver using the bender element resonant method
- Discussion of a method using the properties of waves reflected on the end
caps of the specimen to determine the damping ratio, including a test on
a silt sample from Waremme
- Study of a method to eliminate the influence of the transfer characteristic
of the bender elements enabling the application of the spectral ratio slope
method on bender element time histories
10.2 Description of the equipment

10.2.1 Bender elements
Bender elements consist of two thin piezoceramic plates rigidly bonded to a cen-
tral metallic plate. Two thin conductive layers, electrodes, are glued externally
to the bender.
The polarization of the ceramic material in each plate and the electrical
connections are such that when a driving voltage is applied to the element, one
plate elongates and the other shortens. The net result is a bending displace-
ment. On the other hand, when an element is forced to bend, an electrical
signal can be measured through the wires leading to the element.
There are two possible versions, differing only in the electrical connection
of the two polarized plates: a series or a parallel connection (Figure 10.1). As
the series version gives a higher output for a given distortion, it is better to
use it as a receiver. The parallel version has an additional electrode between
the piezoceramic plates and is better used as transmitter because it gives the
largest distortion for a given input.
A transmitter and receiver element can be placed in various soil testing
devices, such as oedometer, simple shear test device or, as in this research, in
the bottom and top cap of a conventional triaxial device.
Although a combination of a parallel element as sender and a serial element
as receiver would improve the signal quality, in the frame of this research only
10.2. DESCRIPTION OF THE EQUIPMENT 175
a) b)
Figure 10.1: (a) Series and (b) parallel connected piezoceramic element (Dyvik
& Madshus (1985)).
series connected elements are used. This gives the opportunity to change the
pulse direction in the sample without the element configuration influencing the
receiving signal. It means the received signals should have the same shape
independent of the sending direction.
The electromechanical properties of the series connected bender elements
are given in Table 10.1.
Material: Lead Zirconate

Titanate (PZT-5HN)
Dimensions,
length: 20 mm
width: 10 mm
thickness: 0.5 mm
Resonant frequency,
unmounted: 610 Hz
capacitance: 6.5 nF
Maximum drive voltages,
low frequency: 50 V
at resonance: 15 V
used range 0.1 - 15 kHz: 20 V
Sample penetration,
bottom cap: 1.6 mm
top cap: 2.4 mm
Table 10.1: Properties of the bender elements.

10.2.2 Signal generation and measurement apparatus

The equipment to perform pulse bender element tests is given in Figure 10.2.
The setup for a test with a continuous excitation with a sample of a small
height, as it has been used in the later explained resonant method, is depicted
in Figure 10.3.
Signal Analyzer
Top cap
Signal Analyzer
Source Ch.1 Ch.2
Bender
elements Sample
Amplifier Source Ch.1 Ch.2
IN OUT TRIGG.
Top cap Amplifier

Var.: trigger on IN OUT TRIGG.
Bottom cap step pulse Bender Sample
elements
Bottom cap
Var.: trigger on
sending signal
Figure 10.2: Schematic test setup for Figure 10.3: Schematic test setup for
the pulse test with the two variations the resonant test.
of triggering.
Signal generation
A HP 3562A Dynamic Signal Analyzer provides a continuous sine output signal.
The frequency can be step-less chosen between 0 and 100 kHz. Because the
amplitude of the output is limited to 5 V, an additional power amplifier is
necessary.
Alternative signal generation Instead of the signal analyzer, not able to

generate a sine pulse directly, a sound card of a computer is also tried as signal
source with success. It provides a very cost-effective way to produce signals of
arbitrary characteristics, including sine pulses. The sound card is steered by a
MATLAB script but also a control by LabVIEW can be implemented. Because
the output amplitude of the sound card is also limited to about 5 V a subsequent
amplification is unavoidable. However, to obtain the results presented in the
frame of this research the sound card method is not used, it is stuck to the
signal analyzer as source.
Signal amplification
The driving signal is amplified by a specially developed power amplifier to an
amplitude of 20 V (40 V peak-to-peak amplitude). The amplifier is also able
to cut single sine pulses from the continuous sine to provide a pulse source.
Simultaneously to the sine pulse it also generates a function of rectangular
10.2. DESCRIPTION OF THE EQUIPMENT 177
shape with the same length as the circle duration of the sine pulse. This signal
is intended to trigger the data acquisition, because it is supposed to be perfectly
in phase with the sine pulse. It is shown later that the analyzer for the data
acquisition also can directly trigger on the sine pulse making the trigger pulse
unnecessary and eliminating the incorporation of triggering errors caused by
observed small deviations of the starting times of the rectangular trigger pulse
and the sine pulse.
If a continuous signal with amplified amplitudes is desired the pulse cutting
function of the device can be switched off. The signal analyzer and the amplifier
are shown in Figure 10.4.
Figure 10.4: Signal analyzer and source am- Figure 10.5: Triaxial cell with
plifier. sample.
Signal acquisition
The receiver signal is sampled by the same dynamic analyzer used already for
the generation of the continuous sine function. It is able to capture signals at
two channels. The first channel is in general connected in parallel to the signal
that is driving the sender element and is used to trigger the measurement.
The second channel is connected to the output of the receiver element. If the
rectangular pulse of the amplifier is used as trigger signal, it is connected to
the first channel instead of the driving signal. The collected data is transferred
to a computer by means of a floppy disk.
A function of the analyzer to average repetitions of the receiver signal is
used to reduce the signal noise during pulse tests. Usually 20 single pulses are
averaged and summarized to one signal for the further interpretation.
To calculate a response function between the output and the input signal
of the bender elements, the swept sine mode is used. If the dynamic analyzer
is set to this mode and a frequency span with a step width is given, the device
changes the source frequency automatically, calculates the ratio between source
and receiver amplitudes and draws the response curve. Such a test takes about
1.5 minutes.
Triaxial cell
The test triaxial cell, subjecting the sample to an isotropic confining water
pressure, is shown in Figure 10.5. It is a conventional triaxial cell connected to
a cell- and back-pressure supply panel and has a graduated burette to measure
the volume change. Bender elements are mounted at the center of the bottom
and top cap.
A diagram of the bender element mounted in the base of the cell is presented
in Figure 10.6.
8 4
3 1
5 2
9
7
Figure 10.6: Bottom cap with bender element and compression wave element;
1) compression wave transducer, 2) isolation support, 3) epoxy, 4) bender ele-
ment (shear wave transducer), 5) latex, 6) wire leads, 7) bottom cap, 8) porous
stone, 9) drainage.
The mass of a standard metal upper cap (about 600 g) is several times larger
than the mass of the soil specimen. Therefore it was necessary in application of
the later described resonant method, to redesign the metal cap and to replace
it by a new one, made of plastic material with a total mass of about 51 g. For
the test in resonant mode the plunger is disconnected from the top cap to allow
its free movement in all directions.
It was very useful to connect the electrical ground of all components of the
measuring system to the metal parts of the cell housing.
10.3 Techniques to determine Gmax

10.3.1 Wave travel distance
The smallest distance between the bender elements is assumed to be the length
of the wave travel path needed for the calculation of the wave velocity. This
10.3. TECHNIQUES TO DETERMINE GMAX 179
is the distance between the tip of the transmitter element and the tip of the
receiver element, also called the tip-to-tip distance. This assumption is based
on the work of Dyvik & Madshus (1985), confirmed by studies using different
element penetration depths by Brignoli et al. (1996), for instance.
10.3.2 Selection of the input-signal shape

Although pulse tests with bender elements started with a step function, as seen
in Dyvik & Madshus (1985), a sine pulse with a shape of a single sine cycle
is now most commonly used. It opens the opportunity to study the influence
of the frequency on the signals. Occasionally also driving signals consisting
of several sine cycles and deformed sine cycles, as suggested by Jovičić et al.
(1996), are used to simplify the identification of the s-wave arrival. For special
signal analyzes also continuous sine signals are applied.
10.3.3 Methods for determining the travel time

Figure 10.7 shows a typical receiver signal obtained from a bender test using a
single sine pulse of 4.5 kHz for the silt material of the testing site in Waremme.
As it will be explained later the arrival time corresponds with the first inflexion
at point D’. The example time record shows beside the first arrival also clearly
a second and a third wave arrival. However, in the majority of bender element
measurements it is not possible to identify such multiple arrivals.
0.5 20
Input A
Signal
A’
Output voltage [mV]
0.25 10
Input voltage [V]
A’’
C’ A’’’
0 0
O D’’’ B’’’
D’ D’’ B’’
−0.25 Near−Field Effect −10
B’
−0.5 B −20
−0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time [ms]
Figure 10.7: Bender element record on silt from the Waremme site, depth:
11.0 -11.5 m, 200 kPa isotropic effective stress, pulse frequency f = 4.5 kHz,
Vs = 229 m/s, arrival time at point D’: 371 µs, travel path length 8.5 cm, input
and output signal.
In the following, methods for the determination of the travel time of the
shear wave through the sample are summarized. Because the s-wave velocity
is not changing with frequency, the simple visual interpretation of the receiver
signal is still the most commonly used technique. Sanches-Salinero et al. (1986)
developed an analytical solution for the time record at a monitoring point
that would result from a transverse sine pulse of a point source within an
infinite isotropic medium. Jovičić et al. (1996) made finite element calculations
confirming the analytical calculation. The essential findings are the definition
of a phenomenon called near field effect, explained in detailed later in this
section, which is the reason for a deflection, as labeled in Figure 10.7, direct in
front of the arrival point. Viggiani & Atkinson (1995) and Jovičić et al. (1996)
concluded from their laboratory work and from the numerical study that the
most consistent results are obtained by assuming that the shear wave arrival
is marked by this first inflexion labeled as D’. Therefore in the framework of
this research this point is chosen in all cases, if it is identifiable. Additionally
the frequency of the sine pulse is varied to check if the selected point is not
moving with frequency. So the test is repeated at 3, 6, 9 and 15 kHz sine pulse
frequency.
First direct arrival in the output signal

The travel time can be estimated as the time between the start of the input
voltage pulse to the sending bender (point O) and the first deflection in the
output signal from the receiver bender (between point C’ and D’). Depending
on the bender element signals, the identification of the first deflection point
can be straightforward or difficult. The arrival point of the shear wave front
can be masked by faster traveling p-wave components, the described near field
phenomenon and by an inductive coupling between the sender and receiver
channels.
Travel time between characteristic points of input and output signals

The travel time of an impulse wave between two points in space may be taken
as the time between characteristic points in the signals recorded at these two
points. Two measured receiver signals are needed for this approach.
Because in the case of a bender element test only one actual receiver signal
is available, it has to be assumed that the input signal to the transmitting
element has the same shape as a fictive receiver signal at the source. The input
signal is then used as the first receiver signal.
This assumption is questionable because of two reasons. The first is the ca-
pability of the bender elements to follow the driving signal. Studies of different
authors as for instance by Jovičić et al. (1996), applying self-monitoring ele-
ments, show that the ability of the bender elements to follow the driving signal
decreases in the very high frequency range. This is no problem for single and
multiple excitation sine pulses since the commonly used frequencies are below
this range. But if rectangular step function are used, such high frequencies are
reached.
The second problem is related to the transfer characteristic element soil
and the attenuation in the soil itself. Even though the transfer characteristic
mechanism is not well understood it can be seen from practical tests that
both phenomena cause a change of the frequency content of the signal. The
compared signals, i.e. driving signal and one receiver signal, show therefore
a different shape also in the time domain. This again makes the picking of
comparable points at the two signals often erroneous. The method is therefore
not used in this work.
It has to be remarked that this doubt is also valid for other methods com-
paring a driving and a receiver signal, especially the cross correlation of the
two signal types in the time domain.
The most commonly used characteristic points are the first peak (point A
and A’), first trough point (B and B’) or zero crossings of the input and output
signals.
Travel time out of cross correlation of input and output signal

Another way for the determination of the travel time between sender and re-
ceiver element is the calculation of the cross correlation between sender and
receiver signal. The equation for the cross correlation is given, analogue to the
technique used in the SCPT, by:
Z ∞
rRS (τ ) = sR (t) sS (t + τ ) dt (10.1)
t=−∞
where sS is the sender signal, sR the receiver signal and t the time. rRS will
reach a maximum value for the time shift τ that equals the travel time of the
impulse from source to the receiver.
If reflected or refracted waves are strong and the cross correlation fails, a
windowing of the receiver signal can be useful.
Travel time out of the phase of the cross power spectrum of input
and output signal
Instead of the cross correlation in the time domain also cross correlation in the
frequency domain might be applied. This method leads, again very analogue to
the described technique for the SCPT, using the wrapped and unwrapped phase
of the cross-power spectrum to the wave velocity as function of the frequency.
Because the shear wave velocity is in general not frequency dependent, that
means not dispersive, an average value of the velocity over the frequency range
with a reasonable high signal energy, can be calculated. Viggiani & Atkinson
(1995) describe the steps involved in this method and apply it to bender ele-
ment tests. Their results show that the travel times calculated using the cross
correlation and the cross-power-spectrum method are almost identical.
Travel time using multiple arrivals in the output signal

Output signals show in some cases a clear second arrival or even multiple ar-
rivals of the input wave. The second arrival is the input wave after it reflects
from the receiver cap, travels back to the transmitter cap where it reflects again
and then returns to the receiver cap a second time. Arrivals of a higher order
are caused by successive refections on receiver and top cap. Assuming plane
wave propagation, the time between the multiple arrivals in the output signal
is equal to multiples of the double of the travel time from cap to cap.
The travel time may be determined using either characteristic peaks or
the cross correlation method. Possible characteristic points in the example
time record are A’-A”-A”’, B’-B”-B”’ or D’-D”-D”’. The cross correlation of
the output signal with itself, also called an auto-correlation, is in principle
sufficient to determine the travel time. If the different arrivals are not very
well separated it is useful to create, by windowing, for each wave arrival a
dummy signal, where all parts outside the time window containing the arrival,
are set to zero. The dummy signals are than cross-correlated which each other.
Arulnathan et al. (1998) found this method helpful in their studies.
The advantage using multiple arrivals in opposite to the comparison of
sender and receiver signal is the avoidance of the assumption that the electrical
input signal is transfered perfectly into a mechanical oscillation of the bender
tip. That means for instance that the method is also open for a step pulse
excitation.
However, multiple arrivals are not very often observed in the bender element
output signal. Therefore the field of application is quite restricted.
Travel time out of the phase shift of continuous cycled bender tests
To remove uncertainties in the identification of the travel time, recent innovati-
ons have led to continuous wave cycles being used. The technique is described
by Blewett et al. (1999) and a FE interpretation is given in Hardy et al. (2003).
In this method a continuously cycled sine voltage is applied to the trans-
mitter element at a low frequency. The output from the receiver is displayed
together with the driving voltage. The frequency of the input voltage is in-
creased until the input and the output signals come into phase. If the initial
frequency was low enough, under this conditions exactly one complete wave-
length is created between the bender element tips. The frequency is again
increased until a couple of full wavelengths is obtained. Knowing the frequency
of the input voltage, the number of full wavelengths and the tip-to-tip distance,
the shear wave velocity can be calculated. In the case that it is uncertain whe-
ther input and output are fully in phase or not, the cross correlation can help
to determine the phase shift and to correct the result.
10.3.4 Difficulties in the arrival time determination

There are effects which make the identification of the s-wave arrival difficult, if
not impossible. These are shortly described now. It is often possible to reduce
their influence by adapting the measurement configuration.
P-wave components
Bender elements generate besides the transversely polarized shear waves also
a certain amount of compression waves, traveling with p-wave velocity. This
velocity is faster than the s-wave and can reach, if the sample is fully satura-
ted, the compression wave velocity of the pore water. Therefore these wave
components arrive before the s-wave and might overlap with the actual s-wave
arrival.
Brignoli et al. (1996) show a collection of typical receiver signal shapes,
including partly significant p-wave components. It is noticed that this kind of
disturbances are more frequent in measurements with shear-plate transducers
and rather seldom seen if bender elements are used.
Near field effects
This phenomenon was first treated by Sanches-Salinero et al. (1986) by means of

an analytical model. A transverse sine pulse of a point source within an infinite
isotropic elastic medium is modeled and the time record at a monitoring point
at some distance is calculated. It was found that the resulting wave consists
of three coupled wave components with different characteristics. All three
components represent transverse motion, but they propagate with different
velocities; the first two (Γ1 and Γ2 ) with the velocity of a shear wave and the
third (Γ3 ) with the velocity of a compression wave. For the three components
the attenuation arising from the geometrical damping occurs at different rates,
Γ2 and Γ3 attenuating an order of magnitude faster than the Γ1 . Γ1 represents
the far field pure shear wave. Γ2 and Γ3 are the near field components, because
they have only significant amplitudes in close proximity to the source.
The initial polarity of Γ3 is opposite to Γ2 and more important also to
Γ1 . This agrees very well with the observations in typical time records of
experimental data as shown in Figure 10.7. The deflection caused by the near
field component is always opposite to the first deflection of the s-wave. The
arrival of the near field component Γ3 is almost identical with the p-wave
arrival. This is confirmed experimentally by Brignoli et al. (1996) for dry
specimens. In saturated soils the arrival of the near field component does,
following Brignoli et al. (1996), not show a strong initial excursion from the zero
axis so the arrival time cannot be clearly determined. Nevertheless, the near
field component arrival gives much lower velocities than p-waves in the fluid-
soil system, measured with compression transducers. Therefore it is assumed
that the near field component travels with the speed of the p-wave of the dry
specimen.
Even so the analytical solution is based on a single sine pulse the results
are transferable to other excitation shapes such as step pulses.
Brignoli et al. (1996) conclude from the findings of Sanches-Salinero et al.
(1986) that it is in general useless to reverse the electrical impulse to the trans-
mitter element for the purpose of reversing only the shear wave arrival (Γ1 )
since also Γ2 and Γ3 reverse.
In order to separate the near field from the far field Sanches-Salinero et al.
(1986) introduced the parameter Rd . The amplitude of the near field compo-
nent in relation to those of the first s-wave arrival becomes less significant with
an increasing Rd . Rd is defined by:
d df
Rd = = (10.2)
λ Vs
where d is the travel distance, λ is the wavelength and f is the frequency of the
input signal. Because of the increasing deviation of the frequency of the recei-
ved s-wave arrival from the input pulse frequency with increasing frequency,
Brignoli et al. (1996) use the frequency of the first sine cycle of the received
s-wave arrival.
While Sanches-Salinero et al. (1986) suggested Rd ≥ 2, Brignoli et al. (1996)
obtain results without near field component from Rd > 4 to 5 only. Viggiani &
Atkinson (1995) specify that the near field effect may mask the arrival of the
s-wave when Rd = 0.25 to 4. Arroyo (2001) calculates a lower limit for Rd of
0.625, based on an allowed error in the Vs determination of less than 5 %. The
minimum frequency in function of the wave velocity and the Rd value for a
sample in a typical triaxial cell with 11.7 cm tip-to-tip distance can be selected
using Figure 10.9. Figure 10.8 gives the same diagram for a sample with 2.6 cm
height, later used to obtaine the damping ratio.
Rd - Ratio for a sample with 26 mm tip to tip distance

30
10
7 6 5
Rd=3
25
Rd=2
Rd=4
20
R =1
d
Limits for the appearance of the near-field
f mean [kHz]
effect by Viggiani & Atkinson (1995)

15
10
Rd = d / λ
5
R =0.25
d
0
0 100 200 300 400 500 600
Vs [m/s]
Figure 10.8: Mean frequency of the received signal fmean in function of Vs . and
Rd for a sample with 26 mm distance between the bender tips.
If signal interpretation is difficult due to the near field phenomenon, an

increase of the excitation frequency will make the arrival point identification
easier. A high excitation frequency has also the advantage that the received
Rd - Ratio for a sample with 117 mm tip to tip distance

30
10
7
6
25
R =d/λ 5
d
20
R =4
d
f mean [kHz]
15
Rd=3
Rd=2
10
Limits for the appearance of the near-field Rd=1

5 effect by Viggiani & Atkinson (1995)
Rd=0.25
0
0 100 200 300 400 500 600
Vs [m/s]
Figure 10.9: Diagram like Figure 10.8 for 117 mm tip-to-tip distance.
signal will be shorter in time. Therefore the maximum possible error by a

wrong identification of the arrival point is smaller.
Different authors give also an upper limit for Rd . This is not based on the
near field effect. It is simply due to the fact that bender elements become less
effective at high frequencies. The driving voltage has to be increased with an
increase of the pulse frequency to obtain a similar movement. Furthermore
the elements cannot perfectly follow the input signal at high frequencies. This
phenomenon is known as overshooting. Additionally, the deviation between the
frequency of the driving signal and the mean frequency of the actually generated
waves becomes larger with higher frequencies. (The driving frequency is always
higher than the frequency of the transmitted wave.)
Inductive coupling
In some test set-ups it is observed that the output signal is superimposed with
a signal of the same shape as the input signal and with no shift in time to the
input signal. This behavior is caused by capacitive coupling between input and
output signal inside the measurement apparatus. The phenomenon is found at
all kinds of shapes of the input signal. Especially the results from the set-up
of Dyvik & Madshus (1985) with a step-pulse excitation show the effect quite
significantly.
A careful grounding of all involved devices, including the housing of the
testing cell, can reduce the effect partly.
10.4 Methods to obtain the damping ratio

10.4.1 Resonant method
The work of Brocanelli & Rinaldi (1998) describes a method to measure the
damping ratio and shear wave velocity, using bender elements while working
in the frequency domain.
The idea of the method is to bring a short cylindrical soil sample in a shear
movement. For that reason the bottom bender element is excited with a steady
sine signal of constant voltage and the amplitude is measured at the receiver
element. To make this value independent from the source amplitude it is nor-
malized by this amplitude. This process is repeated at different frequencies
until the whole spectrum of sample response is defined. The damping ratio is
estimated at the points of the curve around the natural frequency of the shear
mode. For this purpose different techniques are available such as the half-power
bandwidth and the more general circle-fit method, the latter is also using the
phase components of the resonant curves. Preliminary tests on a clay sample
showed that it could be difficult to find the correct peak in the response spec-
trum corresponding with the shear mode. So it is useful to have knowledge of
the frequency range were this mode dominates.
The main tasks to apply this method are: a) finding a suitable way to
pinpoint the shear mode frequency and b) the calculation of the damping value
from the response spectrum at that frequency.
Natural frequency of the shear movement

Analytical formulations A cylindrical soil sample placed in the triaxial
cell can be considered in a simplified assumption as a fixed-free beam with the
mass of the top cap attached at the free end. Brocanelli & Rinaldi (1998) have
shown by means of a finite element analysis of a 3 cm high sample that, with
a relatively light mass at the top, the sample deforms in perfect shear in the
first mode and with an increasing mass at the top, the shape of the deformed
specimen tends to be flexural. For a pure shear deformation an expression
for the angular natural frequency of the first shear mode (ωs ) can be derived
from the wave equation for a shear wave propagating in a rod and the force
equilibrium at the boundaries:
k mT ωs L ωs L
= tan (10.3)
m Vs Vs
where m is the mass of the top cap, mT is the mass of the sample, L is the
sample length, k is the shear factor and Vs is the shear wave velocity. The latter
is connected to the shear modulus G and the unit mass ρ by the expression
already mentioned:
s
G
Vs = (10.4)
ρ
A detailed formulation can be found for instance in Graff (1991) and Bro-
canelli & Rinaldi (1998). For long samples or large masses of the top cap, the
first resonant mode of the beam will be predominately of the bending. For
the case of a pure bending deformation Cascante et al. (1998) obtained the
following equation from the Rayleigh approximation:
3 EI
ωf2 = (10.5)
h 2
h h ii
33 h 9

L3 140 mT + m 1 + 3L + 4 L
E is the modulus of elasticity, I the area moment of inertia and h the distance
between the end of the sample and the centroid of the mass of the top cap.
The basic concept in the Rayleigh method is the principle of conservation of
energy. To apply the Rayleigh procedure, it is necessary to assume the shape of
the system in its fundamental mode of vibration. This assumption of a shape
function effectively reduces the system to a SDOF system. Thus the frequency
of vibration can be found by equating the maximum strain energy developed
during the motion to the maximum kinetic energy. Detailed information con-
cerning Rayleigh’s method can be found in Clough & Penzien (1993).
Because the real behavior of the sample is always influenced by shearing
and bending it could be difficult to find criteria to decide what expression is
more suitable. Therefore the Rayleigh expression was extended by a term for
the horizontal shear displacement. This led to the following equation:
60 EI L2 + 5 GA kη 2
ω 2 = 42 (10.6)
mT γ + m δ
with
γ = 198L5 + 231L3η + 70Lη 2
and
δ = 840L5 + 2520L4h + 840L3η . . .

. . . + 1260L2hη + 1890L3h2 + 210Lη 2
The derivation of Equation (10.6) can be found at the end of this section.
The actual shape of the mode is supposed as the superposition of a bending
and a shearing displacement. η depends on the ratio between these two parts.
The unit of η is m2 . A is the area of the cross section. The resonant frequency
ω can be calculated at those η where ω becomes a minimum. This follows from
the consideration that any shape other than the true vibration shape would
require the action of additional external constraints to maintain equilibrium.
These extra constrains would stiffen the system, adding to its potential strain
energy, and thus cause an increase in the computed frequency. Consequently,
the true vibration shape will yield the lowest frequency obtainable by Rayleigh’s
method. The solution of the extreme value problem is possible but leads to a
very large expression and is therefore not given here.
Another approach to calculate the natural frequency of the soil sample is
to use the thick beam theory, the so-called Timoshenko beam theory. The
Timoshenko beam includes the bending effects, and shear and rotary inertia
effects. A suitable solution of the system of differential equations of a cantilever
beam with a mass elastically mounted at the free end can be found in Rossit
& Laura (2001). Simplified to the case of a rigid mounted mass, this solution
is given by the following determinant expression:
δ
0 0

L2 L2

δ 2 2

− αL + αL −Ω mr ηλ βL − βL −Ω mr ηλ
0 = (10.7)
sin αL cos αL sinh βL cosh βL
− δ cos αL δ
αL sin αL βL cosh βL βL sinh βL

αL
with
m ωT2 L3 mT I 2(1 + ν)
mr = Ω2 = η= λ=
mT EI AL2 k
Ω2 η 2 λ − 1 Ω2 η
F = Ω2 H = (1 + λ)
L4 L2
= Ω2 ηλ + β 2 L2 δ = Ω2 ηλ − α2 L2
p √ p √
H + H 2 − 4F −H + H 2 − 4F
α= β=
2 2
The natural circular frequency ωT can be extracted from this expression.
mT, A, I, ν, k L
Figure 10.10: Simplified system with parameters of Equation (10.7).
Derivation of Equation (10.6), Rayleigh’s method The mode shape

for the horizontal bending displacement yB of the specimen at elevation x is
assumed to be a third-order polynomial, as shown in Figure 10.11:
yB (x) = a0 + a1 x + a2 x2 + a3 x3 (10.8)
At the lower plate (corresponding to x = 0) the displacement and the tangent

are y(0) = 0 and y’(0) = 0, respectively. Hence, a0 = 0 and a1 = 0. Neglec-
ting the moment at the top end x = L, EI y00 (L) = 0. Then, Equation (10.8)
becomes
yB (x) = αx2 [3L − x] for x ≤ L (10.9)
where α is a constant given by α = a2 /3L = −a3 . The horizontal displacement

of a rigid mass placed above the specimen is estimated from the horizontal
displacement y(L) and the tangent y’(L) at the top of the specimen:
yB (x) = αL2 [2L + 3(x − L)] for x > L (10.10)
The mode shape for the shearing displacement yS is assumed to be linear. This
is described by:
yS = βx for x ≤ L (10.11)
and
yS = βL for x > L (10.12)
The superposition of bending and shearing follows then with:
y(x) = αx2 [3L − x] + βx for x ≤ L (10.13)
and
y(x) = αL2 [2L + 3(x − L)] + βL for x > L (10.14)
The maximum internal potential energy Epot is computed by taking into con-
sideration the internal energy of the bending and shearing component of the
deformed specimen:
L L
1 1
Z Z
Epot = EI (y 00 )2 dx + GAk (y 0 )2 dx (10.15)
2 0 2 0
Introducing Equation (10.13) in Equation (10.15) gives:
Epot = 6EIα2 L3 . . .

12 2 5 1
. . . + GAk α L + 2αL3 β + β 2 L (10.16)
5 2
The maximum kinetic energy Ekin is computed by presuming harmonic oscil-

lation with frequency ω. The component for the specimen is
L
1 2
Z
Ekin, mT = ρω A y 2 dx (10.17)
2 0
or using Equation (10.13):

2 33 2 6 11 1
Ekin, mT = mT ω α L + αβL4 + β 2 L2 (10.18)
70 20 6
The kinetic energy for a concentrated mass added at a distance h above the
specimen can be evaluated using Equation (10.14) at x = h + L:
1 2
Ekin, m = m yh+L ω2 (10.19)
2
and
1
Ekin, m = mω 2 (4α2 L6 + 12α2 L5 h + 4αβL4 . . .
2
. . . + 6αβL3 h + 9α2 L4 h2 + β 2 L2 ) (10.20)
Finally, the circular resonant frequency for the flexural mode is computed by
equating the maximum internal potential energy Epot and the maximum kinetic
energy Ekin, mT + Ekin, m :
60 EI L2 + 5 GA kη 2
ω 2 = 42 (10.21)
mT γ + m δ
with
γ = 198L5 + 231L3η + 70Lη 2
and
δ = 840L5 + 2520L4h + 840L3η . . .

. . . + 1260L2hη + 1890L3h2 + 210Lη 2
At the same time α and β are replaced by η = β/α.
Top Cap h m m
E ,G=∞ E ,G=∞
L mT G=0 mT E =0
Sam ple
x a) b)
Figure 10.11: Assumed mode shapes a) bending and b) shearing.

The soil sample as a cylindrical waveguide The term waveguide is used

to describe situations where a wave is propagated in structures, like rods, plates
or geological strata, whose shape directs the motion along a favored dimension
of the structure, for instance the length of a rod or the plane of the strata.
When this dimension is assumed to be infinite and the cross-section has a cy-
lindrical symmetry the situation enables the mathematical analysis based on
the Pochhammer-Chree frequency equation originally published in Pochham-
mer (1876) and later independently by Chree (1889). This leads to the clas-
sification of waves in three families (torsion, longitudinal and flexural). These
waves show almost always dispersion, that means a dependence of the phase
velocity on frequency.
The equation is used mainly in techniques in the field of non-destructive
testing of pipes, cables and optical fibers. Publications on this topic focus
therefore on frequencies in the sound and ultrasonic range.
The sample in the triaxial cell is cylindrical and therefore axial symmetric
but clearly not infinite long, especially if, as presented in the tests later, a
sample of only 3 cm height with a diameter of 5 cm is used. Another obstacle
in applying the theory is, that the Pochhammer-Chree equation is a solution
for the far-field, that means the calculated mode shapes claim to be valid for
any section of the cylinder. This is not fitting with the reality if the excitation
in the sample center is considered.
Nevertheless, this approach enables insight into more complicated modes of
wave propagation in cylinders than treated so far.
Arroyo (2001) is applying the waveguide approach to the situation of bender
element pulse tests with the objective to explain geometrical dispersion effects
causing difficulties in the identification of the arrival time.
Theoretical background and mode description Waves propagating

along the waveguide in the positive z direction are assumed to be time harmonic
and can be described by:
~u(r, θ, z, t) = ~u(r) ei n θ ei(ω t−k z) (10.22)
where ~u = ur ~er + uθ ~eθ + uz ~ez is the displacement vector in cylindrical coor-

dinates r, θ, z and ω and k are the circular frequency and the wavenumber,
respectively. The circumferential order n is an integer that describes the sym-
metry of the waveform around the z-axis. The modes with n=0 (axisymmetric)
can be decoupled in two families: torsional (T(0,m)), with azimuthal displace-
ment only, and longitudinal (L(0,m)), with both axial and radial displacements.
The modes without axial symmetry are called flexural (F(n,m), with n ≥ 1) and
exhibit all three displacements coupled. The three mode families are shown in
Figure 10.12.
The circumferential order n specifies for n > 0 the number of cycles around
the z axis. The index m is used to sort the modes for a given family, in ascending
phase speed. It gives the number of counterrotating sections along the radius.
ur
uθ ur uθ uz
uz
Torsional Longitudinal Flexural
Figure 10.12: Torsional, longitudinal and flexural modes in a cylindrical wave-

guide (Seco et al. (2002)).
When the assumed waveform of Equation (10.22) is combined with the

boundary conditions, that the surfaces are traction-free, we arrive at the fre-
quency equation for the waveguide in the form of:
det Df (k, ω, n, b) = 0 (10.23)
where Df is a 6x6 matrix. The cylinder

p parameters are the outer radius b, the
p-wave velocity in a rod c0 (c0 = E/ρ) and the Poisson ratio ν. For given
cylinder parameters and circumferential order n, the result is a relationship
between the frequency and the wavenumber, which can be solved to find a
finite number of propagation modes, with real wavenumbers, and an infinite
number of evanescent modes, which have complex wavenumbers and do not
propagate. This leads to the conclusion that all modes except the fundamental
modes of each mode family are only present above a cut-off frequency, different
for each mode.
Because the practical computation is done using a set of MATLAB scripts
developed by Fernando Seco and because the Pochhammer-Chree approach
plays only a secondary roll in this research, the reader is referred to Eringen
& Şuhubi (1975), Thurston (1992) and Thurston (1978) for closer details in
the mathematical theory. The latter article provides also some insights in the
effects appearing if the waveguide is not of infinite length.
Calculation of cut-off frequencies and dispersion curves for the

silt from the Waremme site A computation based on the properties of
the tested reconstituted silt sample S1 from the Waremme site is done using
a set of MATLAB scripts (PCDISP) provided by Fernando Seco. Informa-
tion on these tools can be found for instance in Seco et al. (2002). The in-
put parameters for the program are a cylinder radius b = 2.5 cm, a density
ρ = 1984 kg/m3 , a rod p-wave velocity Vp = c0 = 348.4 m/s and a Poisson ra-
tio ν = 0.29. The parameters correspond to a s-wave velocity measured in the
specimen of Vs = 216.9 m/s.
Objective in the calculation are the dispersion curves of the modes below
10 kHz and the cut-off frequencies of every mode. Flexural modes are calculated
for n = 1 and n = 2 only.
Figure 10.13 shows the obtained frequency dependent phase velocities of
every identified mode and Table 10.2 gives the related cut-off frequencies.
Modes
Torsional Longitudinal Transversal
Type fcut Type fcut Type fcut
[kHz] [kHz] [kHz]
T(0,1) 0.0 L(0,1) 0.0 F(1,1) 0.0
T(0,2) 7.0916 L(0,2) 5.2910 F(1,2) 2.5424
L(0,3) 5.3678 F(1,3) 3.8733
L(0,4) 9.6875 F(1,4) 7.3620
F(1,5) 8.7852
F(1,6) 9.7086
F(2,1) 3.2417
F(2,2) 4.2175
F(2,3) 6.0311
F(2,4) 9.2602
Table 10.2: Calculated cut-off frequencies, fcut , for the example silt material
S1.
Even so Figure 10.13 is quantitatively valid only for the used input values,
the qualitative findings are valid also for other soil properties. Some essential
findings are therefore pointed out in the following:
- The fundamental torsional mode T(0,1) is the only mode showing no

dispersive behavior. It travels with the velocity of a shear wave in a
rod, equal to the shear wave velocity in an infinite medium. This is the
typical mode used in a resonant column device with torsional excitation
and agrees with the elementary theory of torsional vibration in a rod.
- Higher torsional modes start asymptotic at a certain cut-off frequency

and at high frequencies they approach asymptotic the s-wave velocity.
- The basic longitudinal mode L(0,1) agrees at low frequencies with the
velocity of a compression wave (c0 = Vp (rod)) in the thin rod model.
However, at high frequencies the velocity approaches the Rayleigh-wave
velocity Vr .
- Higher longitudinal rods have a cut-off frequency and approach at infinite

high frequencies the s-wave velocity Vs .
Dispersion curves (phase speed) 194
1000
L
T(0,2) (0,4)
900
800
F(2,4)
F(1,3) F(1,4)
700
F(2,2)
F(2,3) F(1,4)
600
L(0,2)
500 L(0,3)
reconstituted silt from the Waremme site.

V (infinite medium)
p
400
Phase speed [m/s]

c0 =Vp (rod)
L(0,1) F(1,2)
300 F(2,1)
V (infinite medium/rod)
s T(0,1)
200 V (half space)
r
F(1,1)
100
0
0 1 2 3 4 5 6 7 8 9 10
Frequency [kHz]
CHAPTER 10. BENDER ELEMENT TECHNIQUE
Figure 10.13: Dispersion curves of an infinite cylinder with 5 cm diameter

- F(1,1) is the fundamental flexural mode present at all frequencies. It

starts with a velocity of zero and then increases asymptotic to the
Rayleigh-wave velocity at high frequencies. It can be shown (Eringen
& Şuhubi (1975)) that the dispersion curve is coincident with that pre-
dicted by simpler theories of beam flexure. For the whole frequency range
the Timoshenko theory for thick beams is applicable and for the low fre-
quency range a Bernoulli-Euler beam is sufficiently exact.
- Higher flexural modes appear above their cut-off frequencies and approach
the s-wave velocity from above.
Conclusions for the further work Because above described models

of the guided waves show deviations to the real conditions of the laboratory
test, particularly concerning the geometry of the excitation elements and the
shortness of the sample, only incomplete information concerning the behavior
of the sample in the laboratory test is possible.
Longitudinal and torsional waves are most probably not excited because
there shape is less akin to the load shape generated by the bender elements.
This leaves only the flexural modes F(n,m) as candidates to describe the dy-
namic behavior of the soil sample. If the excitation frequency stays below the
cut-off frequency of the first higher flexural mode F(1,2) only the fundamental
mode F(1,1) is left.
Beside the numerical calculation, the cut off-frequency of mode F(1,2) can
be also approximated by this simple equation:
Vs
fcut, F (1,2) = 0.5681 (10.24)
d
where d is the cylinder diameter. This equation is based on Thurston (1992)
providing also cut-off frequencies for even higher flexural modes, longitudinal
and torsional modes.
The guided wave theory further confirms that the application of the given
Equation (10.7), based on the Timoshenko beam theory, is suitable to describe
the dynamic behavior of the soil sample because the formulation agrees with
the fundamental flexural mode F(1,1).
Finite element model The given analytical expressions describe only the
first flexural mode. Even though the kind of excitation suggests that this mode
is dominating, a finite element model is used to study what other modes appear
and especially to check how they interact with the flexural mode of interest.
Furthermore the validity of the Rayleigh and the Timoshenko formulation can
be evaluated.
The system of the cylindrical sample and the top cap is modeled in the three
dimensional space with the FE-program ABAQUS. All nodes in the bottom face
of the sample were fixed. The discretization of sample and cap is done by nearly
cubic brick elements of about 4 mm length as shown in Figure 10.14. Linear
z y
x
Figure 10.14: Discretization of sample and Figure 10.15: Axes orienta-

top-cap. tion.
elasticity is chosen for the behavior of the soil sample and the cap. The latter
is modeled with the actual properties of the plastic material, i.e. a mass of 51
g, an elasticity modulus of 3400 MPa, a height of 19 mm and a diameter of
50 mm. The frequency extraction procedure of ABAQUS is used to calculate
the natural frequencies and mode shapes. This procedure uses the subspace
iteration eigenvalue extraction technique. The modeling of the bender elements
itself and the excitation is not necessary for this purpose.
The first five modes together with a verbal description are given in Table
10.3. This table is valid for a tested reconstituted silt sample S1 at 100 kPa
confining pressure, as described in Section 10.5.2, and for the later defined
added mass coefficient CA = 0.
From the FEM-calculation it is seen that the lowest mode is indeed a flexural
mode dominated by shear deformation. The frequency of the second mode
seems to be sufficiently higher so not interacting with the first mode.
Mode 1 Mode 2 Mode 3 Mode 4 Mode 5
875 Hz 1301 Hz 2171 Hz 2354 Hz 4251 Hz

flexural torsional longitudinal flexural torsional,
2nd order
Table 10.3: Natural frequencies found by finite element calculation for sample
S1, 100 kPa and CA = 0.
Poisson’s ratio The shear wave velocity is determined by time arrival

measurements on the real sample so the shear modulus is known. However the
analytical expressions as well as the finite element model need the elasticity
modulus too. An experimental measurement using pulses of compression wa-
ves was not done. Therefore, the elasticity modulus was calculated assuming
a Poisson’s ratio of ν = 0.29. The finite element analysis is performed with
different ν values to evaluate the influence of ν on the natural frequencies. The
differences between the natural frequencies for the lowest mode, dominated
by shear deformation, were not crucial. Therefore an assumption of ν = 0.29
seems to be acceptable and is used in all calculations.
Shear coefficient k A shear coefficient k is needed in the Equations

(10.6) and (10.7). The coefficient is a dimensionless quantity, dependent on
the shape of the cross section and is introduced to account for the fact that
the shear stress and shear strain are not uniformly distributed over the cross
section. According to the definition, k is the ratio of the average shear strain on
a section to the shear strain at the centroid. The use of the shear stress instead
of the shear strain gives equivalent results. Besides the simple, Poisson’s ratio
independent assumption k = 0.9, Cowper (1966) gives the following expression
for a circular cross section:
6(1 + ν)
k= (10.25)
7 + 6ν
This gives for a Poisson’s ratio of ν = 0.29, k = 0.886.

The k coefficient can be derived from the FE calculation too. Therefore
the shear stress distribution in steady state condition caused by the first mode
is computed at 10 cross-sections of the sample-top-cap-model. The elastic
properties of the model are based on the properties of a reconstituted silt
sample used in a preliminary test and is slightly different from those of sample
S1. The shear wave velocity is Vs = 224.5 m/s, the density ρ = 2051 kg/m3
and Poisson’s ratio ν = 0.29. The sample hight is exactly 30 mm.
The system is loaded by a sinusoidal point force of 1 N with the frequency
of the first flexural mode in a node at 1.6 mm height from the bottom side
of the sample. This is the place of the bender element tip. The value of the
excitation force has no influence on the obtained k factor, because it is defined
as a ratio between two values of shear stress or shear strain. The excitation is
done in direction of the y-axis of the system. A sketch of the coordinate axes in
relation to the sample dimensions can be seen in Figure 10.15. The Tables 10.5
and 10.6 show the shear stress distribution at five selected cross-sections and
the shear stress along the x-axis of these sections. The shape of the distribution
changes quite obviously with the distance from the bottom of the sample.
The shear stress at each node in a cross section is taken and averaged under
consideration of its area of influence. This value is divided by the shear stress
in the middle of the section and gives a k factor. All k factors are given in Table
10.4. As already expected from the stress distribution diagrams the k value
decreases with the distance from a fixed bottom of the sample. The k values
for the different sections are again averaged taking into account its height of
influence, hinflu. .
Section z hinflu. k
[mm] [mm] [-]
1 0.40 0.40 1.0487
2 0.80 0.80 1.0254
3 1.60 2.43 0.9551
4 5.66 4.06 0.8552
5 9.72 4.06 0.7824
6 13.77 4.06 0.7556
7 17.83 4.06 0.7609
8 21.89 4.06 0.7781
9 25.94 4.06 0.7642
10 30.00 2.03 0.7412
Average: k̄ = 0.8039
Table 10.4: k-factors on all computed cross-sections.
The final k factor calculated at the frequency of the 1st mode is k = 0.8039,
somewhat less than Cowper’s shear coefficient. This seems reasonable since
Cowper points out that his values of k are most satisfactory for static and low-
frequency deformations of beams and not for high-frequency vibrations as in
our case.
Other authors, also summarized in Cowper’s paper, advocate that k should
be adjusted arbitrary to calculate the high-frequency spectrum of vibrating
beams and have successfully used lower values of k too. A deviation from the
analytical shear coefficient can also be expected from the fact that the assumed
evenness of the cross section is not guaranteed and also the local influences of
the boundary conditions at both ends of the sample cannot be negligible.
The estimation of the natural frequencies in the experimental part of this
work is always done with the FEM-value k = 0.8039.
Influence of the confining water pressure on the natural frequency of

the sample The water in the triaxial cell used to apply an isotropic stress
to the sample, is considered as an incompressible and inviscid fluid. An open
water surface does not exist in the closed cell. In this case the influence of the
fluid to the natural frequencies of a rigid cylinder, surrounded by the fluid, can
be taken into account by introducing an additional mass to the system. This
mass results from some of the fluids particles being permanently displaced by
the intruding body and can be quantified following Wilson (1984):
d2
ma = C A ρ F π (10.26)
4
Cross-section Center-line
Shear stress, section at z = 0 mm
30
100
20 z = 0, k = 1.0487
150
200
200
150
10 250
250
y [mm]
0 y↑→ Shear stress, line at y = 0, z = 0 mm

x 300
−10 Shear stress [N/m²]

150
200
200
150
−20 100
Isolines 100
labelled in N/m²
−30 0
−30 −20 −10 0 10 20 30 −30 −20 −10 0 10 20 30
x [mm] x [mm]
Shear stress, section at z = 5.66 mm

30
120
20 0 140130 1110 z = 5.66 mm, k = 0.8552
11010130 150 1020
0
90
90 140 160
10
160150 0
15 16
2
120 1
10400
y [mm]
130
10090
y↑→ Shear stress, line at y = 0, z = 5.66 mm

10090
110
0
150
300
110
x
14
130
Shear stress [N/m²]

120
0
−10 160 200

90
160 0
150
12 1150 140 90
−20 110 13400 130
0 1 100
Isolines 120 1
labelled in N/m²
−30 0
−30 −20 −10 0 10 20 30 −30 −20 −10 0 10 20 30
x [mm] x [mm]

30
140
20 z = 13.77 mm, k = 0.7556
140 120
100
100
80
0
10
60
16
80
160
140
120
y [mm]

160
40
40
0 300
x
121000 0
60
−10
8000
200
1
160
14
8
40
0
12
−20 1 100
Isolines
labelled in N/m²
−30 0
−30 −20 −10 0 10 20 30 −30 −20 −10 0 10 20 30
x [mm] x [mm]
Table 10.5: Shear stress distribution over the sample cross-section and at the
x-axis at z=0/5.66/13.77 mm, k-value for each height.
Cross-section Center-line
30
20 z = 21.89 mm, k = 0.7781

120
120
100
100
60
60
10
80
80
y [mm]

40
0 300
x
40
120
120
−10
100
100
200
60
80
80
−20 100
Isolines
labelled in N/m²
−30 0
−30 −20 −10 0 10 20 30 −30 −20 −10 0 10 20 30
x [mm] x [mm]
Shear stress, section at z = 30 mm

30
120
20 z = 30.00 mm, k = 0.7412
0
10
60
60
80
100
40
40
10
80
y [mm]
y↑→ Shear stress, line at y = 0, z = 30 mm

20
0 300
x
20
100
60
−10 200
60
80
100
80
−20 100
Isolines 120
labelled in N/m²
−30 0
−30 −20 −10 0 10 20 30 −30 −20 −10 0 10 20 30
x [mm] x [mm]
Table 10.6: Shear stress distribution over the sample cross-section and at the
x-axis at z=21.89/30.00 mm, k-value for each height.
ma denotes the added mass per unit length, CA is a non-dimensional added

mass coefficient, ρF is the fluid density and d the diameter of the cylinder. It
is observed that, as the cylinder length becomes much larger than its diameter,
the value of CA approaches a theoretical limit of unity. For shorter cylinders
different authors give coefficients depending on the length to diameter ratio
l/d. For instance Hafner (1977):
1
CA = (10.27)
1 + (d/l)2
Equation (10.27) is given for the case of a moving cylinder with two free ends.
Even though the considered system of sample and top cap is fixed at one side
it can be assumed that this equation gives a reasonable approximation. The
length l is assumed to be the total length of sample and top cap.
So resonance frequency calculations, including the effects of the confining
water, can be easily performed using a virtual mass composed of the actual
body mass and the above given added mass.
Viscous material damping

The basis for the analysis of the frequency response of the soil sample is the
identification of different modes of vibration at resonance. The damping ratio
D is calculated at these points of the response spectrum in the neighbourhood
of a resonance peak.
Amplitude of the response spectrum The most common method of mea-

suring damping uses the relative width of the response spectrum. Using the
variables indicated on the curve in Figure 10.16, the logarithmic decrement δ
and the damping ratio D, can be calculated from (Richart et al. (1970)):
s √
2πD π f22 − f12 A2 1 − 2D2
δ=√ = 2
(10.28)
1 − D2 2 fm A2max − A2 1 − D2
When D is small, the last term and (1 − D2 )0.5 can be taken as equal to
1.0. Further simplification is possible, if A is chosen equal to Amax /20.5 :
f22 − f12
D= 2
(10.29)
4fm
The application of this expression is usually called the half-power bandwidth

method.
Circle-fit method The circle-fit method, described in Ewins (1988) is able

to calculate the damping ratio with very few points around the resonance peak
and the amplitude of the peak has only little influence on the result. This is an
advantage in cases were different modes have frequencies close to each other.
Figure 10.16: Resonant curve with variables for half-power bandwidth method.
The Nyquist plot of the response spectrum of a single degree of freedom

system leads to a circle as shown in Figure 10.17. Even though the sample is
not such a system it behaves for selected frequency sections in the same way.
The material damping can be calculated from points close to the maximum
amplitude using the following expression:
ω22 − ω12
D= (10.30)
ω2 tan α22 + ω1 tan α21

2ω0
with:
- ω0 , angular frequency corresponding to the maximum sweep angular ve-

locity;
- ω1 , ω2 , angular frequencies;
- α1 , α2 , angles at both sides of ω0 .
A circle is fitted to the points of the response curve close to the resonant
frequency to find the center. Knowing this point makes it possible to determine
the necessary angles α.
Multi-degree of freedom methods The frequency response function from

a linear system is defined as the ratio between the frequency domain represen-
tations of the signal leaving the system Y(ω) and the input signal X(ω). The
function can be described further as a ratio of two polynomials, as shown in
Equation (10.31).
Pm
Y (ω) ak (iω)k
H(ω) = = Pk=0
n k
(10.31)
X(ω) k=0 bk (iω)
Im
Re
c
ω1 α1
α2
ω0 ω2
at Resonance
Figure 10.17: Nyquist plot used in the circle-fit-method.
b0 can be chosen as one without loss of generality. In a process of curve

fitting the analytical form to the measurement data, the unknown coefficients
of both the numerator and denominator polynomials, ak (k = 0, . . . , m) and
bk (k = 1, . . . , n), can be obtained.
If the rational fraction form is transformed into the partial fraction form
the poles ω∞ and the roots ω0 can be determined. Equation (10.32) gives such
a partial fraction form.
Qm Qm
am (iω − iω0,k ) a m im k=1 (ω − ω0,k )
H(ω) = Qnk=1 = n
Q n (10.32)
bn k=1 (iω − iω ∞,k ) b n i k=1 (ω − ω∞,k )
For a model with N-degrees-of-freedom the frequency response function con-

tains N-pole pairs. The order of nominator, m, and denominator, n, polyno-
mial has not to be different. Iglesias (2000) shows that both can be chosen as
n = m = 2N. This leads to the following simplified equation:
2N
a2N Y ω − ω0,k
H(ω) = (10.33)
b2N ω − ω∞,k
k=1
The eigenfrequencies fk and the damping ratio Dk can be calculated based

on the poles of the response function using these relations:
|ω∞,k |
fk = (10.34)
2π
Re(ω∞,k )
Dk = − (10.35)
|ω∞,k |
The practical difficulties in applying this approach are essentially the calcu-
lation of appropriate curve fitting parameters ak and bk . Fortunately the pro-
blem is almost identical with the design of a digital infinite-impulse-response
filter (IIR-filter) based on a given filter frequency response function. The ob-
tained parameters ak and bk are in this case called filter coefficients and could
be directly used to filter a signal in the time domain. For the filter design a
wide number of software packages are available.
In this research a MATLAB script developed by Iglesias (2000) for the
parameter estimation based on an article of Richardson & Formenti (1982) is
used.
10.4.2 Use of multiple shear wave arrivals

It is possible to identify a second or higher order arrival in the time record
of a bender element pulse test under special circumstances. This is caused
by a reflection of the generated waves on the receiver and the sender cap. A
time record with this characteristic has the advantage that the same wave train
is measured at several times with the same bender element. This overcomes
the problem that the transfer characteristic, particle movement in the soil to
electrical output voltage, is not quantitatively known. If the same element is
used, amplitude comparisons of the different arrivals can be performed.
Based on this idea a typical time record showing multiple arrivals in a
Waremme-site-silt-sample is analyzed using the spectral ratio slope method
(SRS).
It has to be emphasized that multiple arrivals cannot often be identified in
bender element time records. The tests on the Boom clay of Sint-Katelijne-
Waver for instance show no distinguishable second arrival.
Time records with multi-arrival characteristic
Figure 10.18 shows time records obtained from a sample of the site in Waremme
from a depth of 3.0 to 3.5 m (Sample: B1/3). Arrivals up to the third order
can be distinguished. The consolidation is performed with a confining stress of
200 kPa. Bender elements are mounted in the top and the bottom cap of the
triaxial cell. The top cap is connected to the loading piston of the cell. The
sample diameter is 50 mm, its height is L = 85 mm and Vs = 230m/s. The
bender elements are driven by sinusoidal pulses with frequencies of 3000, 4500
and 6000 Hz.
Spectral amplitude ratio
The attenuation coefficient αS is calculated based on the ratio between the

spectral amplitudes of signal sections (r1st , r2nd , r3rd ) related to arrivals of
different order.
This means for the first and the second arrival:
1 |F F T (r1st )|
αS = ln (10.36)
2L |F F T (r2nd )|
Pulse frequency: o
3000 Hz o o
Pulse frequency: o
4500 Hz o
o
Pulse frequency: o
6000 Hz o o
1st arrival 2nd arrival 3rd arrival

−0.8 −0.5 0 0.5 1 1.5 2 2.5 3 3.5
t in s −3
x 10
Figure 10.18: Time histories of BE measurements on the silt from Waremme,

sample height: L = 8.5 cm, Vs = 230m/s, selected window ranges are indicated.
and for the first and the third arrival:

1 |F F T (r1st )|
αS = ln (10.37)
4L |F F T (r3rd )|
A rectangular window is used to separate the arrivals. Its center is set to
the maximum signal peak related to the arrival. The length of the window is
always 586 µs.
The Figures 10.19 to 10.21 give the attenuation coefficient with frequency
for the three excitation frequencies. Two curves are included in each figure.
One is related to the wave travel from the first to the second arrival and the
other to the interval between the first and the third arrival.
A slope z is fitted to the curves in a frequency range of 2000 Hz centered
at the mean dominant frequencies and the damping ratio Ds is calculated by
Equation (9.26):
z · Vs
Ds = (10.38)
2π
The results are summarized in Table 10.7. The calculated damping ratios are
between 6.3 and 9.3 %.
Amplitude ratio at the frequencies with the highest energy

Because, in opposite to a SCPT, the frequency content of the BE-excitation
can be varied easily by changing the length, respectively the frequency, of the
10
1st to 2nd arrival
1st to 3rd arrival
Attenuation coefficient [1/m]

8
−3
z = 2.0352 ⋅ 10 s/m,
DS = 7.45 %
6
4 z = 2.0041 ⋅ 10−3 s/m,

DS = 7.34 %
0
0 1000 2000 3000 4000 5000 6000
Frequency [Hz]
Figure 10.19: Attenuation coefficient for a pulse of 3000 Hz.
10
8 z = 1.7205 ⋅ 10−3 s/m,

DS = 6.30 %
−3
z = 2.3415 ⋅ 10 s/m,
4 D = 8.57 %
S
2
1st to 2nd arrival
1st to 3rd arrival
0
0 2000 4000 6000 8000
Frequency [Hz]
20
1st to 2nd arrival
1st to 3rd arrival
15
z = 2.5276 ⋅ 10−3 s/m,

DS = 9.25 %
10
5 −3
z = 2.1814 ⋅ 10 s/m,
D = 7.99 %
S
0
0 2000 4000 6000 8000
Frequency [Hz]

Excitation Parameter Unit Considered travel path

frequency 1st to 2nd 1st to 3rd
arrival arrival
3000 Hz fmean [Hz] 3074 2922

αS [m−1 ] 5.13575 3.89520
z [s/m] 2.0352 · 10−3 2.0041 · 10−3
Ds [%] 7.45 7.34
4500 Hz fmean [Hz] 4427 3830

αS [m−1 ] 7.25377 5.91632
z [s/m] 1.7205 · 10−3 2.3415 · 10−3
Ds [%] 6.30 8.57
6000 Hz fmean [Hz] 4953 4166

αS [m−1 ] 8.18255 6.96369
z [s/m] 2.5276 · 10−3 2.1814 · 10−3
Ds [%] 9.25 7.99
Table 10.7: Parameters of the spectral ratio slope method applied to bender
element time histories showing multiple arrival; data based on the signals of
Figure 10.18; fmean : mean dominant frequency, αS : attenuation coefficient, z:
rise of αS with frequency, Ds : damping ratio.
sinusoidal pulse, the spectral ratio slope method can also be applied in another
way.
The attenuation coefficient at the mean dominant frequency, fmean , for each
excitation frequency is drawn versus the frequency in a separate diagram, as
in Figure 10.22. It is then again possible to fit a line to these points and to
calculate the damping ratio based on the slope z of this line.
A damping ratio of 5.9 % is obtained if first and second arrival are compared.
First and third arrival lead to a damping ratio of 8.9 %. The deviation of these
values to the mean values of Table 10.7 for each considered travel path is slightly
higher than the deviation of the single values in the table from its mean values.
A particular reason for that cannot be given yet.
Correction for the boundaries

Until now it is assumed that during the refections no loss of energy appears.
The SRS method would be able to compensate a frequency independent change
of amplitudes caused by a reflection but no change depending on the frequency.
A frequency independent amplitude decay appears in terms of an αS -f diagram
as a parallel upward shift of the curve. Only a frequency dependent decay
changes the slope z and as a consequence Ds .
The loss at the caps can be considered by the introduction of a reflection
10

8
z = 1.6108 ⋅ 10−3 s/m,
DS = 5.90 %
6
z = 2.4163 ⋅ 10−3 s/m,

4 DS = 8.85 %
2
1st to 2nd arrival
1st to 3rd arrival
0
2500 3000 3500 4000 4500 5000 5500
Frequency [Hz]
Figure 10.22: Attenuation coefficient at the mean dominant frequencies of the

three excitation pulses.
factor c(ω) in the Equations (10.36) and (10.37). This means for the first and
the second arrival:
F F T (r1st ) · c(ω)2

1
αS = ln
2L |F F T (r2nd )|
1 |F F T (r1st )| 1
ln c(ω)2

= ln + (10.39)
2L |F F T (r2nd )| 2L
and for the first and the third arrival:
F F T (r1st ) · c(ω)4

1
αS = ln
4L |F F T (r3rd )|
1 |F F T (r1st )| 1
ln c(ω)4

= ln + (10.40)
4L |F F T (r3rd )| 4L
The exponent of c(ω) indicates the number of reflections the signal has to
undergo. Two reflections are considered for the travel path between 1st and
2nd arrival and four reflections between 1st to 3rd arrival.
The simplest way to model the boundary effects is to assume at the interface
soil to metal cap a two media boundary as shown in Figure 10.23. The factor c
can be expressed based on a constant cross section of both media. The equation
follows Clough & Penzien (1993) and Fratta & Santamarina (1996):
ISample
ICap − 1
c= I
(10.41)
1 + Sample
ICap
Loading
Piston
Cap Sample Cap

s1 s2 = IFFT(FFT(s1) · c(ω))
Sample
c1 Top/Bottom
Cap
Base Plate
Figure 10.23: Schematic view of wave refection Figure 10.24: Triaxial

and refraction on top and bottom cap. cell with installed sam-
ple.
The impedances I can be calculated by the material density ρ and the shear
wave velocity Vs :
I = ρ Vs (10.42)
According to the equation, the reflection factor c is not depending on fre-

quency and therefore has no influence on the damping calculation using the
spectral ratio slope technique. For the used testing device and the tested sam-
ple |c| = 0.9586. This value is based on the material parameters in Table 10.8.
Parameter Unit Sample Caps

Density, ρ [kg/m3 ] 2342 8009
S-wave velocity, Vs [m/s] 230 3180
Impedance, I [kg/(m2 s)] 5.3866 · 105 2.5470 · 107
Reflection coefficient, |c| [-] 0.9586
Table 10.8: Calculation of the reflection coefficient c for the case of the silt
sample Waremme B1/3.
Fratta & Santamarina (1996) provide also an equation for a three media
boundary. The base plate of the triaxial cell and the loading piston, as shown in
in Figure 10.24, could be considered as the third medium. In this way the effects
of reflection and refraction on the interface cap to base plate respectively piston
could be taken into account. However, there are some difficulties in determining
a suitable impedance for the third medium. Fratta & Santamarina (1996)
found, based on their sample and apparatus parameters, a reflection coefficient
c(ω) depending on frequency in a range below 1000 Hz. At higher frequencies
c(ω) is nearly constant.
It is assumed that this frequency dependent component has only minor

effects on the damping ratio obtained by the spectral ratio slope method. It
could cause a slight underestimation of the actual damping ratio.
10.4.3 Use of different travel paths

The direct application of the spectral ratio slope method or alternatively the at-
tenuation coefficient method on the setup of two bender elements in a triaxial
cell sample is not possible because there is no sufficient information on the
transfer characteristics soil - BE - output voltage, respectively driving voltage
- BE - soil. This transfer characteristic might be expressed as a transfer func-
tion κ(f), potentially highly frequency dependent. κ(f) incorporates effects of
different resonance frequencies of the elements, different penetration depths of
the elements and the coupling of the element tip and the specimen.
However, there is a theoretical approach to eliminate the influence of κ(f)
if two measurements with opposite wave propagation directions are combined.
The realization of this idea needs two receiver elements and two transmitter
elements installed on one sample.
55
10 10
7 35 48 35 7
76
Ground level of the box
Figure 10.25: Frame with mounted sender (S1/S2) and receiver elements
(R1/R2), unit: mm, Wang et al. (2004).
A preliminary test using this technique is published by Wang et al. (2004).

The author uses a container of the dimensions 25 cm (H) x 50 cm (L) x 33.2
cm (W) filled with sand. The bender elements are mounted on a frame, shown
in Figure 10.25, and buried in the container. The damping ratio is calculated
by means of the spectral ratio method.
The background of the technique is described based on a setup of a triaxial
cell with horizontally mounted elements. The Equation (9.19) for the atte-
nuation coefficient α is adapted using the parameters defined in Figure 10.26.
Equation (9.19) is repeated here:

A1 x 1
ln A 2
· x2
α(f ) =
x2 − x 1
The seismic pulse is transmitted from an element placed on the sample. It
is received simultaneously from two receiver elements on the other site of the
sample (setup A). After that, a pulse is sent from the other transmitter and is
received from both receiver elements (setup B).
Setup A Setup B
Receiver 1 Sender 1 Receiver 1 Sender 1
FFT(r1A) FFT(r1A)⋅κ1
z1A FFT(r1B) FFT(r1B)⋅κ1
BE BE BE BE
z2A z1B
Receiver 2 Sender 2 Receiver 2 FFT(r2B)⋅κ2 Sender 2

FFT(r2A) FFT(r2B)
BE BE BE BE
FFT(r2A)⋅κ2 z2B
Soil Specimen Soil Specimen
Figure 10.26: Sketch of a sample with horizontal bender elements.
Four time records are obtained, r1A and r2A for setup A and r1B and r2B for
setup B. Their spectral amplitudes as a function of the frequency are indicated
as FFT(r... ). z... gives the travel path length between sender and receiver for
the different combinations.
The attenuation coefficient for setup 1 can be expressed as:

ln F F T (r1A ) κ1 z1A
·
F F T (r2A ) κ2 z2A
α(f ) = (10.43)
z2A − z1A
α(f) for setup 2 is:

F F T (r2B ) κ2 z2B
ln F F T (r1B ) κ1 · z1B
α(f ) = (10.44)
z1B − z2B
The Equations (10.43) and (10.44) can be transformed into:

α(f ) (z2A − z1A ) = ln + ln + ln (10.45)
and

α(f ) (z1B − z2B ) = ln + ln + ln (10.46)
If the sum of both equations is calculated the transfer functions κ1 and κ2

are eliminated and an expression for α(f) using the information of all four time
records can be derived as:

F F T (r1A ) F F T (r2B ) z1A z2B
ln F F T (r2A ) · F F T (r1B ) · z2A · z1B
α(f ) = (10.47)
z2A − z1A + z1B − z2B
Equation (10.47) is independent of the unknown BE-transfer characteristics
and can be further developed to an expression according to the spectral ratio
slope method, which is not done here. Wang et al. (2004) found for dry Ottawa
sand a damping ratio of 3.6 % if this method is applied, admittedly with another
element configuration. However, this value is quite high compared with other
values given by the same authors based on resonant column tests.
10.5. THE RESONANT METHOD 213
10.5 Measurements of damping ratio on samp-

les taken at the test sites based on the re-
sonant method
10.5.1 Introduction
The bender element test is performed on samples from the site in Waremme
and Sint-Katelijne-Waver. The specimens are subjected to different isotropic
stress levels by an external water pressure. The back pressure applied to the
sample is set to 100 kPa. The external isotropic pressure is therefore chosen
always 100 kPa above the value of the target effective stress. The absorption of
water by the sample during consolidation is measured and taken into account
for the calculation of the sample mass. The shear wave velocity is determined
by the analysis of the wave arrival at each stress level. The properties of the
free top cap are: height 19 mm, mass 50.74 g and diameter 50 mm.
The damping ratio is calculated based on half-power bandwidth and circle-
fit method.
10.5.2 Site in Waremme

Altogether three samples taken from the drilling B1 in Waremme are tested.
Specimen S1 is a reconstituted sample; S2 and S3 are undisturbed samples. All
samples have a length of 29 to 30 mm. The 300 kPa consolidation stage of the
sample S3 is repeated with air as confining medium to study its influence on
the response spectrum.
The Figures 10.27 to 10.29 present the measured response spectra, the Ta-
bles 10.9 and 10.10 the predicted and measured natural frequencies including
the physical parameters of the samples.
The theoretical estimation of the natural frequencies of the first mode is
done with an added mass coefficient of CA = 0.48, which corresponds to a
length to diameter ratio of 0.96, and as well with no added mass (CA = 0) to
allow comparisons. The coefficient is calculated by means of Equation (10.27).
As expected, the frequencies including the added mass component are lower
than the values without the additional mass. For the same CA , the results from
the Rayleigh expression show a good agreement with the FEM-values. The
frequencies calculated by the Timoshenko equation are always higher.
The test results show that the natural frequencies and the shear wave ve-
locities increase with an increasing cell pressure. The amplitude ratios of the
observed peaks decrease with increasing stress level.
The predicted frequencies are in all cases higher than the measured ones.
The closest agreement to the experimental results is achieved at CA = 0.48 by
the Rayleigh equation and the FEM. A frequency comparison for the case of
sample S2 can be seen in Figure 10.30
The damping ratio versus effective confinement stress is shown in Figure
10.31. D is in the range of 7.2 to 9.9 % for S1, between 4.7 and 7.3 % for S2
and between 2.3 and 7.8 % for S3. The deviation between the values from the
half-power method and the circle-fit method are in nearly all cases less than
1 %. A dependence of the damping ratio on the confining stress could be not
noticed in the stress range applied. The damping values are high in comparison
with the resonant column and free torsion pendulm tests.
Example: half-power bandwidth and circle-fit method

Half-power and circle-fit method are used to obtain the damping ratio. The
Figures 10.32 shows an example of the half-power bandwidth technique applied
on the frequency response curve of the undisturbed silt sample S2 at 100 kPa
confining pressure. The calculation of the damping ratio D is based on Equation
(10.29).
Figure 10.33 gives insides in the circle-fit technique based on the same re-
sponse spectrum. The following algorithm is used:
• Drawing of the Nyquist plot from the experimental frequency response

curve.
• Manual selection of some points of the response curve close to the reso-
nance peak. These data points are indicated in the Nyquist plot of Figure
10.33 as ”data points used for curve fitting”.
• Based on these selected points a circle is fitted.
• The center point of this circle is the origin of the angles α. The data
point in the Nyquist plot with the largest distance to its adjoining points
is the point of the natural frequency.
• Based of the selected range of data points for the curve fitting many
combinations of α- and ω-values are possible. They are all evaluated by
Equation (10.30) and the damping ratio is obtained.
• All D-values are finally averaged and this average value provides the result
of the test.
Influence of the confining medium

For the evaluation of the influence of the water pressure to the response curves
in the experimental bender element test, the manner of applying the confining
stress is changed. The triaxial cell is connected to a source of controllable air
pressure and the 300 kPa consolidation stage of sample S3 is repeated using
air pressure instead of the usual water pressure.
The response curve of this stage, given in Figure 10.34, has a similar shape
as this found using water pressure. The resonant peak of the first mode is
shifted towards higher frequencies but not as much as predicted by analytical
and FEM calculations.
10.5. THE RESONANT METHOD 215
Comparison with other test methods

The Figures 10.35 and 10.36 summarize the results of the laboratory and in
situ tests concerning Vs and D.
S-wave velocity The SCPT-profile is taken from SCPT5; the s-wave velocity
of the shallow 7 m is 165 m/s based on the results of SASW UGent Setup 2.
The results of the BE-time arrival method, the resonant column test and the
free torsion pendulum test are drawn according to the depth of sampling. The
given velocities of the BE-time arrival method are, because of a better accuracy,
not taken from the short samples used for the BE-resonant method. Instead,
the results shown in Figure 6.12 obtained from standard sized triaxial samples
are used. The isotropic stress related to each measured value is indicated at the
data point. The actual mean effective stress assumed at the sampling depth is
given by an arrow enabling a raw ”graphical interpolation” between the data
points. The tip of the arrow gives therefore the calculated s-wave velocity out
of the laboratory tests at the sampling depth.
The effective stresses and the density of the silt increase with depth, accom-
panied by an increase of the s-waves velocity. The results of the SCPT and the
laboratory tests confirm this increase in Vs . The s-wave velocity determined
by the in situ tests is higher than the results of the laboratory tests. This is
valid for the BE test, working under very low shear strain amplitudes, and also
for the resonant column and free torsion pendulum test, operated at a shear
strain of 10−5 . A possible reason for the difference of field and laboratory tests
can be a loosening of the silt during the drilling and sampling process, leading
to a decrease in density. Another reason might be a lower shear strain in the
field tests.
Damping ratio The profile of the damping ratio in Figure 10.36 is arranged
in the same way as the Vs -profile. The damping ratio by the SCPT is based
on the spectral ratio slope method performed on the time records of SCPT5.
The results of the BE-resonant method is the mean value of half-power and
circle-fit method of sample S2 calculated at each consolidation stage.
The values of the SCPT show major variation with the depth but their mean
value falls in the same range than indicated by the resonant column and the
free torsion pendulum test. The damping ratio calculated by the BE-resonant
test is above the range of the SCPT results and the other laboratory tests. A
possible reason could be a frequency dependence of the damping ratio. More
information on this topic is given in Section 10.6.
10.5.3 Site in Sint-Katelijne-Waver

Two Boom clay samples from the drilling B1I in Sint-Katelijne-Waver are tested
by the bender element resonant technique. The length of sample C4 is about
30 mm and sample C5 is 60 mm long. The additional sample C5 is selected to
prove if the deviated equations for the resonance frequency can properly predict
the shift of the resonance frequency between samples of different lengths. If

this is the case, the association of the experimentally found natural frequency
with the modeled lowest vibration mode would be confirmed once more.
The isotropic confining stress is applied in stages of 50, 100, 200 and
400 kPa. The results can be found in detail in the Tables 10.11 and 10.12;
the response spectra are shown in Figure 10.37 and Figure 10.38. The added
mass coefficient is CA = 0.71 is case of sample C5, corresponding to a length
to diameter ratio of 1.57.
The evaluation of the response function of sample C5 is not possible because
of the relatively low natural frequency of the longer sample. It lays between 250
to 400 Hz following the Rayleigh equation. However, there is no corresponding
peak in the response spectrum of Figure 10.38. It is assumed that the bender
elements are not able to excite the sample sufficiently at such low frequencies.
Therefore in the following only results of sample C4 are discussed.
The frequencies of resonance are in general below the values of the tests on
the silt material from Waremme agreeing with a lower s-wave velocity of the
clay. The damping ratio calculated by the two analyzing methods are drawn
versus confining stress in Figure 10.39. The ratio lays between 7.3 and 15.7 %.
This is again higher than the values obtained by resonant column and free
torsion pendulum test and higher than the values found for the silt samples.
Comparison with other test methods

The profiles in the Figures 10.40 and 10.41 summarize the test results in the
same way as done for Waremme. The SCPT data are taken from SCPT2.
S-wave velocity The resonant column test indicates a slightly lower s-wave
velocity than the SCPT. The result of the free torsional pendulum test is almost
identical with the SCPT. The BE element test gives the highest velocities.
Their differences from the other laboratory tests can be explained by a lower
shear strain level during the BE tests.
Damping ratio An indication of the in situ mean effective stresses is for the
D-profile not useful because the test data of the different confining stresses are
to close to each other or because, in case of the BE-resonant method, D does
not change monotonic with the confinement stress.
The result comparison for the damping ratio comes to the same conclusions
as at the site of Waremme. The BE-resonant results are the highest of all
test. The other laboratory tests values fall in the range of the scattered SCPT
results.
10.6 Summary and remarks

The determination of the s-wave velocity is performed by a visual picking of
the first wave arrival. This technique has shown to be most straightforward
10.6. SUMMARY AND REMARKS 217
and flexible.
In the following methods to calculate the damping ratio are treated. An
approach to analyze multiple refections by the spectral ratio slope method is
applied on a sample of the site in Waremme. Damping ratios calculated by
this technique for the silt material are in the same range as obtained by the
BE resonant test.
A self correcting method is described to eliminate the influence of the trans-
fer characteristic of the bender elements. This enables the application of the
spectral ratio slope method.
The determination of the damping ratio focuses on the bender element
resonant method. At first the behavior of the sample-top-cap system is modeled
analytically, using the Timoshenko beam theory and Rayleigh’s approximation,
and numerically, using the FE program ABAQUS. This provides information
on the natural frequency of the first flexural mode to be used in the modal test.
The influence of the confinement medium water is considered by an added mass
component and experimentally validated.
The first flexural mode is identified in experimental response curves by
means of the natural frequency prediction. The response curves are analyzed
using circle-fit and half-power method to obtain the damping ratio. Undistur-
bed samples from Waremme and Sint-Katelijne-Wave are tested.
Two basic problems appear looking at the test results. The first is the
deviation of the calculated and the measured resonant frequencies. The ass-
umption of a beam with top-mass at its one end, even if shear effects are taken
into account, might be a relatively crude way to describe the very short, about
3 cm long, sample and the top cap. The finite element model can partly over-
come this geometrical problem, but also in this case, influences of the rubber
membrane and the not fully rigid mounting of the bottom cap are not taken
into account. Furthermore the consideration of the water in the cell by an
additional mass might not be sufficiently correct. If the viscosity of the water
is taken into account, a frictional (viscous) drag force, proportional to the ve-
locity of the moving cylinder, has to be introduced (Wilson (1984)). The used
coefficient CA for the added mass depends, besides the cylinder geometry, also
on the Reynolds number and the cylinder roughness. A more detailed investi-
gation could help to find a more exact theoretical solution. On the other hand
the prediction of the natural frequency is only used to find the right resonant
peak among several, so the calculations are already sufficient for a successful
selection.
The second problem concerns the difference between the results of the BE-
resonant test and the other laboratory tests. Possible explanations for this
phenomenon can be external factors as the supply cables to the top-cap, the
rubber membrane and the confining medium affecting slightly the oscillation.
Another factor could be a possible frequency dependence of the damping ratio.
Remarks on the frequency dependence of the damping ratio A possi-

ble reason of the high damping values obtained by the bender element resonant
technique might be the different range of working frequency of the tests. Cyclic
tests are usually performed at frequencies below 2 Hz; free torsion pendulum
test show commonly natural frequencies of sample and device below 10 Hz.
Even resonant column tests devices reach, dependent on the sample geometry
seldom resonance frequencies higher than 250 Hz, in the most cases much lower.
In contrast to that, samples subjected to a BE-resonant test have natural fre-
quencies, crucial for the damping determination, in the range between 400 to
900 Hz.
A frequency dependence of the damping ratio could possibly explain the
deviation in the results. Unfortunately, this question is not completely solved
yet. Dobry & Vucetic (1987) write in their summary, given in table Table 2.2,
on influence factors of the damping ratio, that with an increasing strain rate,
i.e. frequency, the damping ratio ”stays or may increase”.
Mok et al. (1988) suggested based on the work of others that, for dry
rocks and dry soils, D is independent of frequency. On the other hand, water-
saturated rocks, sands and silts would show a definite dependency of D on fre-
quency. For saturated clays the authors conclude a relatively small frequency
dependency.
Several authors compare results from cyclic torsional shear tests and reso-
nant column tests. Zavoral & Campanella (1994) are doing this for a marine
silt to clay deposit and found no frequency dependence. The cyclic torsional
apparatus was driven at frequencies between 0.01 and 1 Hz; the resonance was
reached in the range of 10 to 30 Hz. Bolton & Wilson (1989) come to the same
finding for a dry sand. The cyclic testing frequency was fixed at 0.001 Hz,
the authors call it a pseudo-static test; the resonant conditions were obtained
between 45 and 95 Hz.
Lin et al. (1996), tested two types of sand at various frequencies in the range
from almost static to 20 Hz using again a cyclic torsional device. They found
an increase of damping ratio from 2.6 to 6.4 % for the first sand and between
1.9 to 6.2 % for the other sand. These tests were performed at a shear strain of
4 · 10−3 %. Remarkably no significant change in the G-modulus was observed.
A few in situ tests give also some indication of the influence of the frequency
on the damping ratio. Hall & Bodare (2000) describe a cross-hole test perfor-
med in layers of mud and clay where the attenuation coefficient method is used
to analyze the attenuation. In the range between 20 and 30 Hz the damping
ratio is found to be frequency independent. Mok et al. (1988) came to the same
result for frequencies between 170 and 300 Hz and a site dominated by a clay.
Kudo & Shima (1981) confirm an almost frequency independent behavior for
an even larger range from 5 to 100 Hz based on a down-hole set-up in sand.
Mok et al. (1988) investigated the damping of p-waves in dry sand based on
a miniaturized cross-hole set-up and the attenuation coefficient method. They
found in the range of 2200 to 3000 Hz an almost constant damping ratio.
A mathematical model including a frequency dependent damping can be
obtained by Biot’s theory concerning the interaction of the pore water and the
soil fabric. If the soil is assumed to be a saturated two-phase medium the
damping ratio of the shear wave, following Biot’s theory, has to be frequency
dependent. The damping ratio increases form the static case with the frequency
until a local maximum is reached after which the ratio decreases again. Figure
10.42, taken from Stoll (1979), shows the logarithmic decrement as function of
frequency according to Biot’s theory for the parameters given in Table 10.13.
Beside self-explanatory values, a pore-size parameter (a) and a structure con-
stant (c) are influencing the shape of the curves. The pore-size parameter
(a) depends on the size and shape of the pores and can be estimated by the
mean grain diameter. However, the structure constant (c) is an experimentally
determined parameter not easy to quantify.
Curve No. 1 shows the results of the calculation for a less permeable mate-
rial. The logarithmic decrement stays constant in the whole observed frequency
range and is equal to the decrement of the frame. All other parameter sets with
higher permeability lead to the described frequency dependent curves.
Further information on Biot’s theory can be found in the basic papers Biot
(1956a) and Biot (1956b).
As a conclusion there is evidence that comparing free torsion pendulum
tests, resonant column tests and BE tests on the level of damping ratio can
give different results due to frequency dependence of the ratio.
−5
x 10
Amplitude ratio receiver − transmitter [−]
582 Hz 100 kPa

1 200 kPa
300 kPa
0.8
0.6 708 Hz
807 Hz
0.4
0.2
0
100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
Figure 10.27: Frequency response spectra for sample S1 (reconstituted silt).
−5
x 10
2
636 Hz
100 kPa
200 kPa
1.5 300 kPa
1
766 Hz
0.5
820 Hz
0
100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
Figure 10.28: Frequency response spectra for sample S2 (undisturbed silt).

−5
x 10
1.2
100 kPa
631 Hz 200 kPa
1 300 kPa
0.8
735 Hz
0.6
811 Hz
0.4
0.2
0
100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
Figure 10.29: Frequency response spectra for sample S3 (undisturbed silt).
1200
Frequency flexural mode [Hz]
Timoshenko Equation
1000
800 Rayleigh FEM

Experiment
Equation
600
400
200
0
100 200 300
Confinement stress [kPa]
Figure 10.30: Calculated and measured resonant frequencies for the first flexu-
ral mode of sample S2, added mass effects are included (CA = 0.48).
10
S1 Half−Power Method
Circle−Fit Method
9
S1
8
S1 S3
Sampling Depths:
7
Damping ratio, D [%]
S1, S2 −> 4.0−4.5 m,

S3 S3 −> 5.0−5.5 m
S2 S2
6
S2
5
3 Free Torsion Pendulum (12.0−12.5 m)

S3
2 Resonant Column (5.0−5.5 m)
1
Resonant Column (4.0−4.5 m)
0
0 100 200 300 400 500
Stress, σ1, 2, 3 [kPa]
Figure 10.31: Damping ratio’s obtained from the samples S1, S2 and S3 in
comparison with resonant column test and free torsion pendulum test on other
samples of the same material (reference shear strain 10−3 %).
−5
x 10
2.5
f = 636 Hz
m
A
2 max
1.5 A = Amax / 20.5
D =(f −f ) / (2 ⋅ f ) = 6.2 %
1 HPM 2 1 m
0.5 f1 = 593 Hz
f = 672 Hz
2
0
100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
Figure 10.32: Example of the half-power method to the frequency response

spectrum of sample S2 at 100 kPa isotropic stress.
−6
x 10
5
Nyquist plot of
DCFM = 5.9 % the respose spectrum
0
Imaginary Component
fitted circle
−5
center point
−10
α α
1 2
−15 ω1
ω2
−20 resonance ω data points used
0
frequency for curve fitting
−2 −1.5 −1 −0.5 0 0.5 1 1.5
Real Component x 10
−5
Figure 10.33: Circle-fit technique on the Nyquist plot of the response spectrum
of Figure 10.32.
−6
x 10
5
811 Hz 300 kPa − water
300 kPa − air
930 Hz
2
0
0 500 1000 1500 2000
Frequency [Hz]
Figure 10.34: Comparison of water and air pressure, experimental results at

300 kPa and sample S3.
actual mean effective stress at a certain depth Waremme

confining stress stages used in laboratory tests indicates the calculated velocity out of a
laboratory test at sampling depth
0
100 kPa
33 kPa
2
BE−Time arrival method
400 kPa
200 Pa
50 kPa
25 kPa
62 kPa
4
Resonant Column Test 77 kPa
50 kPa
100 kPa
200 kPa
400 kPa
Depth [m]
8 Mechanical Hammer (SCPT)

SASW
126 kPa
10 BE−Time arrival method

50 kPa
200 kPa
100 kPa
400 kPa
600 kPa
Free Torsion Pendulum Test 141 kPa

12
50 kPa
200 kPa
100 kPa
14
0 50 100 150 200 250 300 350
S−Wave Velocity [m/s]
Figure 10.35: Vs profile of the site in Waremme including the results of all
performed laboratory and in situ tests; the SCPT data are based on SCPT5;
SASW results are based on SASW UGent setup 2.
actual mean effective stress at a certain depth Waremme

confining stress stages used in laboratory tests indicates the calculated damping ratio out
of a laboratory test at sampling depth
0
Mechanical Hammer
Sledge Hammer
62 kPa
62 kPa
4
77 kPa Resonant Column Test
BE−resonant method
Depth [m]
10
141 kPa
12 Free Torsion Pendulum Test

50 kPa
100 kPa
14
0 2 4 6 8 10
Damping Ratio [%]
Figure 10.36: Ds profile of Waremme based on SCPT5 and including the other
laboratory tests.
−6
x 10
7
397 Hz 460 Hz 50 kPa
100 kPa
6 200 kPa
400 kPa
5 609 Hz
735 Hz
1
0
100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
Figure 10.37: Frequency response spectra for sample C4 (undisturbed Boom

clay).
−6
x 10
6
50 kPa
100 kPa
5 200 kPa
400 kPa
0
100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
Figure 10.38: Frequency response spectra for sample C5 (undisturbed Boom

clay).
16
Half−Power Method
Circle−Fit Method
14
12
Sampling Depth:
Damping ratio, D [%]
C4 −> 4.0−4.5 m
10
6 Free Torsion Pendulum (4.0−4.5 m)
Resonant Column (4.0−4.5 m)

4
0
0 100 200 300 400 500
Stress, σ1, 2, 3 [kPa]
Figure 10.39: Damping ratio’s obtained from the sample C4 in comparison

with resonant column test and free torsion pendulum test on other samples of
the same material (reference shear strain: 10−3 %).
actual mean effective stress at depth Sint−Katelijne−Waver

confining stress stages used in laboratory tests indicates the calculated damping ratio out
of a laboratory test at sampling depth
0
1
3
100 kPa
200 kPa
400 kPa
50 kPa
Depth [m]
4 Resonant Column Test

BE−resonant method
Free Torsion Pendulum Test
100 kPa
50 kPa
200 kPa
400 kPa
121 kPa
5
121 kPa
9
0 50 100 150 200 250 300
S−Wave Velocity [m/s]
Figure 10.40: Vs profile of the site in Sint-Katelijne-Waver including the results

of all performed laboratory and in situ tests; the SCPT data are based on
SCPT 2.
Sint−Katelijne−Waver
confining stress stages used in laboratory tests
0
Mechanical Hammer
Sledge Hammer
1
a
kP
3 400
00,
,2
00
,1
50
Depth [m]
Resonant Column Test

4 BE−resonant method
Free Torsion
Pendulum Test
100 kPa
200 kPa
400 kPa
50 kPa
9
0 5 10 15
Damping Ratio [%]
Figure 10.41: Ds profile of Sint-Katelijne-Waver based on SCPT2 and including

the other laboratory tests.
Sample Description:
Sample index S1 S2
Site Waremme Waremme
Description Silt Silt
Sampling reconst. sample undist. sample
Depth [m] 4.0-4.5 4.0-4.5
Initial test parameters:

Length, L [mm] 29.0 29.5
Diameters, d [mm] 50.0 50.0
Distance BE tips [mm] 25.0 25.5
Mass, mT [g] 101.53 119.49
Unit mass, ρ [kg/m3 ] 1783.10 2062.90
Consolidation parameters:
Effective stress, peff [kPa] 100 200 300 100 200 300
Water absorption [g] 11.4 10.9 10.2 -3.2 -5.6 -7.0
Effective mass, mT,eff [g] 113.0 112.5 111.8 116.3 113.9 112.5
S-wave velocity, Vs [m/s] 216.9 224.5 246.3 197.7 246.1 271.3
Resonant frequencies:
CA = 0
Rayleigh Eq., fRay [Hz] 888 918 1005 799 988 1084
Timosh. Eq., fTBT [Hz] 962 995 1090 864 1070 1175
FEM, fFEM [Hz] 875 904 989 787 972 1066
CA = 0 .48
Rayleigh Eq., fRay [Hz] 773 799 875 696 860 943
Timosh. Eq., fTBT [Hz] 843 871 954 757 937 1028
FEM, fFEM [Hz] 763 788 862 686 847 929
Exp. result, fmeas [Hz] 582 708 807 636 766 820
Damping ratio:
Half-Power, DHPM [%] 7.9 9.1 8.0 6.2 7.3 5.4
Circle-fit, DCFM [%] 7.2 9.9 8.9 5.9 4.7 4.9
Table 10.9: Test data for sample S1 (reconstituted silt) and S2 (undisturbed
silt).
Sample Description:
Sample index S3
Site Waremme
Description Silt
Sampling undist. sample
Depth [m] 5.0-5.5

Length, L [mm] 30.0
Diameters, d [mm] 50.0
Distance BE tips [mm] 26.0
Mass, mT [g] 116.54
Unit mass, ρ [kg/m3 ] 1978.44
Effective stress, peff [kPa] 100 200 300 300
air
Water absorption [g] 1.3 0.7 0.4 0.4
Effective mass, mT,eff [g] 117.8 117.2 116.9 116.9
S-wave velocity, Vs [m/s] 195.1 230.6 255.0 257.0
CA = 0
Rayleigh Eq., fRay [Hz] 774 913 1009 1017
Timosh. Eq., fTBT [Hz] 837 988 1092 1100
CA = 0 .48
Timosh. Eq., fTBT [Hz] 766 904 999 1007
Exp. result, fmeas [Hz] 631 735 811 930
Damping ratio:
Half-Power, DHPM [%] 7.8 6.8 (10.0) (17.5)
Circle-fit, DCFM [%] 7.3 6.2 2.3 2.1
Table 10.10: Test data for sample S3 (undisturbed silt).

Sample Description:
Sample index C4
Description Boom-Clay
Depth [m] 4.0-4.5

Length, L [mm] 29.8
Mass, mT [g] 114.60
CA = 0 .48
Exp. result, fmeas [Hz] 397 460 609 735
Damping ratio:
Half-Power, DHPM [%] 15.7 12.1 14.9 13.5
Circle-fit, DCFM [%] 10.7 7.3 12.8 13.4
Table 10.11: Test data for sample C4 (undisturbed clay), 30 mm height.

Sample Description:
Sample index C5
Description Boom-Clay
Depth [m] 4.0-4.5

Length, L [mm] 59.5
Mass, mT [g] 221.62
CA = 0 .71
Exp. result, fmeas [Hz] - - - -
Damping ratio:
Half-Power, DHPM [%] - - - -
Circle-fit, DCFM [%] - - - -
Table 10.12: Test data for sample C5 (undisturbed clay), 60 mm height.

1
Curve No.
1
2
Log. Decrement [−]
3
4
5
10−1
10−2
1 10 102 103 104
Frequency [Hz]
Figure 10.42: Logarithmic decrement in function of the frequency for s-waves

according to Biot’s theory, Stoll (1979).
Values for different curves

Material property Unit No. 1 No. 2 No. 3 No. 4 No. 5
Mass density of grains, ρr kg/m3 2650
Mass density of fluid, ρf kg/m3 1000
Viscosity of fluid, η Pa s 0.001
Shear modulus of frame, G MPa 56.7
Structure constant, c - 1.25
Intrinsic permeability, ki m2 2.5 1 1 2.3 5
·10−14 ·10−10 ·10−11 ·10−11 ·10−11
Permeability, k = ki · ρf g/η m/s 2.45 9.81 9.81 2.26 4.91
with g = 9.81m/s2 ·10−7 ·10−4 ·10−5 ·10−4 ·10−4
Pore-size parameter, a m 3.8 3.9 1.2 1.9 2.8
·10−6 ·10−5 ·10−5 ·10−5 ·10−5
Porosity, β - 0.49 0.36 0.36 0.36 0.36
Log. decrement of frame
for shearing motion, δ - 0.08 0.03 0.03 0.03 0.02
Damping ratio, D = δ/(2π) % 1.27 0.48 0.48 0.48 0.32
Table 10.13: Material properties of the Biot theory example, Stoll (1979).
Chapter 11
Conclusions and summary
Main purpose of this research is to investigate to what extent the seismic cone
penetration test (SCPT) and the bender element test (BE-test) primary deve-
loped for the measurement of shear and compression wave velocities in soils can
additionally provide information on the material damping behavior of the soils.
This chapter gives a summary of the research and points out main findings and
conclusions.
Determination of wave velocities

At first, methods for the calculation of the wave velocities are studied and the
most suitable techniques for both tests are selected. Because of its flexibility
and also considering the limited number of time histories to analyze, a visual
interpretation is chosen for the BE-test.
Under the conditions of a SCPT a visual arrival picking, cross correlation
in the time domain and also a calculation of the cross power spectrum is con-
sidered. The cross correlation in the time domain is preferred because it can
be automatically performed. This is necessary in view of the large amount of
time records during a SCPT. Furthermore it eliminates arbitrary influences by
the person performing the analysis regarding picking of the arrival time or the
selection of a suitable frequency range.
The determination of the wave velocities provides for both types of tests
reliable results and can be considered as sufficiently developed. This is proved
on several test sites.
Determination of the damping ratio

Seismic cone penetration test After an extended selection process, a qua-
lified sensor equipment for the damping measurements is found and installed in
a seismic dual cone. The selected triaxial accelerometers meet the requirements
concerning a small size, a flat frequency response, a high sensitivity and a high
resolution.
235
236 CHAPTER 11. CONCLUSIONS AND SUMMARY
Two groups of approaches to calculate the damping ratio are investigated.

The first methods are based on the spectral ratio of the amplitude spectra
of two recorded signals. These are the attenuation coefficient method (ACM)
and the spectral ratio slope method (SRS). The second group is studied to a
lower extend. It transfers a technology used for vertical seismic measurement
arrays to record earthquakes to the SCPT. The time histories of the particle
acceleration are converted using a shear beam model of the soil to stress-strain
hysteresis curves. These are analyzed to obtain the damping ratio.
SRS method and ACM give a significant scatter of the damping ratio with
depth but also for the same depth, repetitions of the SCPT give results with
some deviation. It is conjectured that the scatter is caused by the pretreatment
of the measured signals before the spectral ratio is calculated. Pretreatment
means mainly the way of cutting the signal part of interest by the application
of a window. This point is intensively studied by the variation of window
length and type on measured and on simulated time histories. Especially by
analyzing the simulated signals it could be quite clearly shown that a too
short cutting window is indeed the most dominating reason for the scattering,
whereas the shape of the window function has very low influence. Unfortunately
this finding cannot conclude in an improvement of the windowing technique
since the separation of the main shear wave arrival, especially in signals showing
reflected wave components and in noisy environments, is unavoidable and forces
often to short windows.
The SRS method has the advantage in opposite to the ACM that it eli-
minates the need of introducing of geometrical corrections and is therefore in
principle preferable.
The technique based on the analyses of stress-strain time histories is applied
to data of one site. Even though a number of well shaped stress-strain loops
could be calculated the damping ratio (25 to 30 %) is quite far above the
expected range. Four main reasons can be identified: a) The receiver distance
of 1 m of the dual cone is to large to avoid spatial aliasing. b) The transfer
function approach, used to bypass the insufficent repeatability of the source,
addes small errors caused by signal noise from the surface to the calculation
depth. c) The calculation of the shear stress time histories by an integration
with depth starting at the stress-free surface collects likewise small signal errors.
d) A double integration is needed for the shear strain calculation.
The reasons a) and b) can be eliminated when the receiver distance is
chosen closer and a repeatable source is used. It is expected that in this way
the obtained damping ratio is lower and more credible.
Bender element test The most elaborated method studied is the bender
element resonant method. It is based on a modal test of a soil sample in
a triaxial cell using BE’s for the excitation and the capturing of the response
movement. Some modification on the standard triaxial equipment are necessary
to allow the upper BE cap a free oscillation.
The experimental natural frequencies of the sample have to be assigned to
237
certain mode shapes. Because the first flexural mode is used to obtain the
damping ratio, its frequency is predicted by means of two analytical formu-
lations (Timoshenko beam theory, Rayleigh’s method) and a FE calculation
(ABAQUS). The influence of the confining medium is taken into account by
the introduction of an added mass component. A reasonable approximation of
measured and predicted natural frequencies is obtained.
Results based on bender element resonant method are gathered on samples
of a silt (5 − 10 %) and a clay (10 − 16 %) material. These results are repeata-
ble but higher than the results from resonant column and free torsion pendulum
tests (1 − 4 %) performed for comparative purposes. There is some evidence
that this deviation is caused by the higher frequency used in the BE-test in
comparison with the two other laboratory tests.
Another interpretation technique uses the property of shear wave pulses re-
flected on the end caps of the triaxial device. The SRS method can be applied
based on the spectral ratio between reflections of different order. Damping
ratios calculated in this approach are between 7 and 9 % for the silt material.
This is the same range as obtained by the BE resonant test. Unfortunately mul-
tiple reflections could be not identified in samples of the clay material probably
because of the higher attenuation in the clay.
A SRS based method to eliminate the influence of the transfer characteristic
of the bender elements by means of a combined analysis of different travel paths
is discussed.
Practical relevance The SCPT and BE test methods for the determination
of the damping ratio need still a further refinement until they can be recommen-
ded for the practical application. Especially the scattering seen in the ACM
and SRS method should be reduced. However, by taking in consideration that
there are nearly no other, at least no cost-equivalent, field tests available to
determine the damping ratio, it seems obvious to spend additional effort in
this technique.
Reasons for the high damping values of the BE-resonant method should
also be closer investigated.
Recommendations for further research

Seismic cone penetration test
• Application of the method to analyze stress-strain time histories on si-
mulated SCPT data.
• Analysis of stress-strain time histories using shorter receiver distances

and a repeatable source.
• Further extensive SCPT testing may open the possibility for the develop-
ment of correlations between damping and other cone measurements.
238 CHAPTER 11. CONCLUSIONS AND SUMMARY
• SASW tests with an advanced inversion algorithm to determine s-wave

velocity and damping ratio profiles simultaneously can provide an alter-
native field test method. The technique is found in Lai (1998). Results
of these SASW tests can be used to evaluate the profiles of the SCPT’s.
Bender elements
• The frequency dependency of the damping ratio can be investigated by a
resonant column device with electronically adjustable working frequency
or cyclic triaxial tests.
• Application of the self correcting method of Wang et al. (2004) to elimi-
nate the influence of the transfer characteristic of the bender elements on
bender element configurations as described in Section 10.4.3.
• Replacement of the flexural excitation of the BE-resonant test by a tor-
sional excitation. Nakagawa et al. (1996) give suggestions of a triaxial
set-up using some bender elements radially orientated. Such a set-up can
be used to excite the first torsional shear mode of the sample. This is the
mode also used in standard resonant column devices. The mode has the
advantage, in opposite to the flexural mode, that it shows a pure shear
deformation and the kind of confinement medium has almost no influence
on the natural frequency.
• Efforts to use the Pochhammer-Chree approach for infinite long cylinders
for the calculation of natural frequencies of finite cylinders can possibly
improve the natural frequency prediction.
Part IV
Appendix
239
Appendix A
Accuracy and costs of soil

tests
This chapter provides some help in the selection of a suitable testing method
for the s-wave velocity, respectively Gmax , and the damping ratio if a certain
accuracy is needed, taking into account the available financial and time budget.
It has to be emphasized that the testing methods related to the s-wave
velocity are of great variety and mostly well tried. Their application is relatively
widespread and some of them can be considered as standard methods.
This is in contrast to methods to determine the damping ratio. The selection
bandwidth of methods is limited. Furthermore a large number of techniques is
only poorly approved and has its use only in a research environment.
Table A.1 gives an overview of the common techniques indicating the shear
strain range of their applicability. In addition it gives relative ratings of their
costs, their accuracy and their currency of practical application. As a mean
to compare the techniques the given classification of accuracy and costs are
relative values.
Some more details are provided in the following for the different groups of
tests. All findings concerning the accuracy based on this research have to stay
quite general and qualitative since either, as for the SCPT, no reference values
of comparable tests are available or the data pool is too limited for statistical
analyzes. Also the information on the costs has to stay qualitatively because
its reliable quantification requires an elaborate market analysis which is not
performed.
However, in the frame of this research orders for resonant column tests are
placed with external laboratories. Additionally a few prices for other tests are
available based on recent quotations of external companies and not at least
based on the price list for tests performed by the Laboratory of Soil Mechanics
in Ghent. The individual test prices are summarized in Table A.2 together with
a description of the testing conditions. They can provide an indication for the
situation on the local market but might not be complete and representative.
241
242 APPENDIX A. ACCURACY AND COSTS OF SOIL TESTS
A.1 Laboratory tests

Cyclic and resonant tests
Cyclic tests (cyclic triaxial, cyclic torsional shear and cyclic simple shear tests)
and resonant tests (resonant column and free torsion pendulum test) are, ex-
cept for the cyclic simple shear test, based on a triaxial apparatus extended
with a loading facility to apply a dynamic loading. Cyclic tests are perfor-
med commonly at frequencies around 1 Hz. Resonant column tests have to
achieve the resonant frequency of the sample going up to a few hundred Hz.
The resonant methods can work at lower shear strains than the cyclic tests.
If higher strain levels are needed a hollow-cylindrical specimen is used ins-
tead of a cylindrical sample because it has a more uniform shear strain distri-
bution over the cross-section.
Since the conditions in the triaxial stress cell are well controlled and these
tests are long time established the accuracy and the reproducibility of the
results is in general high. This is true for both G-modulus and damping ratio.
Two resonant column tests performed on samples of comparable depths from
Waremme provided identical G and D values. The difference in the results of
the free torsion pendulum test to the resonant column test can be considered
as small. The maximum deviation is found in one case to be 1.5 % damping
ratio. The maximum deviation concerning the G-modulus is about 15 % at the
site in Waremme and up to 30 % at the site Sint-Katelijne-Waver. Both can
be tracked in the Figures 6.15 and 7.13.
However, if the in situ values should be approximated, the results depend
also on the success of the undisturbed sampling. Specimen of less cohesive soils
like sand or gravel can regularly not be taken in undisturbed conditions and
have to be reconstituted. It is often questionable if the results of a reconstituted
sample can be considered as representative for the in situ conditions especially
in the small strain region.
The costs for laboratory testing include always the costs for the test itself
and also the costs for the drilling to sample the soil specimen.
In the frame of this research resonant column and free torsion pendulum
test are applied. It is found that the resonant column test provides the greater
reliability of both tests.
Bender element based tests

Time arrival method The bender element methods using the arrival time to
determine the s-wave velocity have the advantage that they can be incorporated
in a wide range of classical laboratory testing devices, in most cases triaxial
test devices, oedometers and simple shear apparatuses. Since the BE testing is
considered as non-disturbing the basic purpose of the devices is not influenced.
The combination of bender element tests to obtain the shear wave velocity and
the classical tests make the bender element technique very cost effective.
Despite of occasional problems concerning the identification of the arrival
A.2. FIELD TESTS 243
time, the method is able to provide results of high accuracy and reproducibility.
Since the technique is in use for several years a lot of experience is available
and its application is widely spread.
The comparability to the resonant and the cyclic tests suffers sometimes
under the unknown level of shear strain reached. But it is assumed that the
shear strain is always lower during a bender element test than in the other
laboratory tests.
Damping ratio determination In comparison to the time arrival techni-

ques, the determination of the damping ratio by the described resonant method
is not a commercial testing technique. It is only used in a research environment.
In the frame of this research the BE-resonant method and, to a smaller
extend, also the multiple reflection approach is applied. Both techniques lead
to significantly (1 to 10 %) higher damping ratios than the resonant techniques
(resonant column and free torsion pendulum test). The results vary also with
the used analyzing technique, in case of the BE-resonant method, and the used
excitation frequency, in case of the multiple reflection approach. This variation
is in the range of +/– 0.25 to 2.5 % damping ratio. Based on these results the
accuracy of this techniques is considered as comparatively small.
The costs of the tests are similar to the time arrival method. It has to be
ensured only that the top cap of the triaxial apparatus is light in weight and
can move freely.
Ultrasonic tests / shaking table

These tests use ultrasonic acoustic waves to determine the p-wave velocity only.
Since they are based on the same time arrival techniques as the bender element
test, the accuracy of this method is considered as high. Application to obtain
the damping ratio of soil materials are not known.
Shaking table set-up’s are always very specifically orientated to a certain
problem. They are therefore rather expensive and the accuracy depends on the
chosen configuration.
A.2 Field tests

The principal advantage of field tests is that they avoid the need to obtain
undisturbed samples by drilling. This reduces on one side the testing costs
and on the other side effects caused by sample disturbances. Furthermore field
tests can be performed rapidly in comparison to laboratory tests where it needs
some consolidation time to adjust the in situ stress conditions.
Surface methods
Surface techniques need no costly equipment to penetrate the ground and are
therefore the most inexpensive in situ testing techniques. Seismic reflection
and refraction tests but also the SASW tests belong to this group.
The reflection and refraction test provide, dependent on the actual soil stra-
tification, wave velocities of medium to high accuracy. Approaches to determine
also the damping ratio are not common.
The accuracy of the SASW technique depends on the complexity of the soil
layering but can be considered as medium to high. Experiences concerning the
accuracy of the obtained damping ratio, if a combined Vs and D inversion is
used, are very limited in literature, but promising.
Penetrating methods
The SCPT belongs to this group . Its price is situated between a purely surface
technique and a borehole method. The accuracy concerning the s-wave velocity
has shown to be very high coequal to a cross- or down-hole test.
However, the determination of the damping ratio by spectral ratio slope
or attenuation coefficient method is much less accurate and produces quite
scattered profiles.
Borehole methods
These are the classical seismic methods to obtain Vs and Vp , the cross-, down-
and up-hole methods. The determination of the wave arrival is no questionable
issue anymore. Therefore the accuracy concerning the stiffness parameters is
high.
The determination of the damping ratio is also based on spectral ratio slope
or attenuation coefficient method. Published results indicate a slightly better
accuracy than found with SCPT. But also here the experiences are limited.
The borehole tests are quite expensive since the sender and / or the receiver
are to be lowered in a cased borehole. In case of a cross-hole test at least two
holes are necessary, for a damping ratio determination even three holes are
needed.
A.3 Conclusions for the selection of a testing

method
Concerning the determination of the shear wave velocity
• Expensive boreholes tests can be replaced by SCPT and SASW tests

without a loss of accuracy.
• If the testing costs are crucial, the SASW technique is suggested.
• The SCPT has its field of application at sites of a complex stratification

where the SASW inversion fails or the results are questionable.
A.3. SELECTION CONCLUSIONS 245
Concerning the determination of the damping ratio

• An exact and reliable determination of the damping ratio is best possible
with the classical laboratory resonant test.
• Cyclic tests deliver also accurate results but the application concentrates
on higher ranges of shear strain.
• In situ tests, especially the here considered SCPT, are less expensive, but
their results are sometimes questionable. However, relative information
on the difference of the attenuation characteristics can be found.
• Tests using bender elements to obtain the damping ratio cannot yet be
suggested as a reliable method, at least if they are not accompanied by
other testing techniques.
• SASW tests with a combined Vs and D inversion might be an accurate
source in the future. Until now, there is not much experience available.
Testing method Currency of Applicability at Accuracy / Reliability Costs 246
determination of small large G / Gmax /
G D strains strains V s / Vp D
Laboratory tests
Resonant column test (cylinder / hollow-cylinder) f f g/g l/m +++ +++ +++
Free torsion pendulum test i i m m ++ ++ ++
BE - time arrival method f - g - +++ +
BE - resonant method r r g - + + +
BE - multiple reflection approach r r g - +++ + +
BE - combination of different travel paths r r g - +++ ? +
Cyclic triaxial test f f l m +++ +++ +++
Cyclic simple shear test i i l m ++ ++ ++
Cyclic torsional shear test f f g m +++ +++ +++
Ultrasonic tests (only Vp ) i - g - +++ - +
Shaking table r r - g ++ + ++++
In situ tests
SASW - Vs profile f - g - ++ - +
SASW - combined inversion for Vs and D profile r r g - ++ + (?) +
SCPT - time arrival f - g - +++ - ++
SCPT - spectral ratio slope / attenuation coefficient method - r g - - + ++
SCPT - hysteresis approach r r g - + + ++
Down- and up-hole tests i r g - +++ + +++
Cross-hole tests i r g - +++ ++ ++++
Reflection-/refraction tests f - g - ++ - +
Free oscillation tests, enforced oscillations r r g m + + +++
Dynamically loaded plates f - l g + - +
In situ shear tests r r l g + + ++++
Geotomography i - g - ++ - ++
Dilatometer test (DMT) i - - g ++ - ++
Pressuremeter test (PMT) i - - g ++ - ++
Standard penetration test (SPT) with correlation f - - l + - ++
Cone penetration test (CPT) with correlation f - - l + - ++
Legend:
Currency: f → frequently, i → infrequent, r → under research conditions
Shear strain level: small strain → γ < 10−3 %, large strain → γ > 10−3 %
Table A.1: Comparison of the test methods for G (Gmax ) and D.

Applicability: g → good, m→ medium, l→ less, -→ not suitable
Accuracy/Costs: +→ low, ++→ medium, +++ → high, ++++ → very high, - → not applicable, ? → unknown
APPENDIX A. ACCURACY AND COSTS OF SOIL TESTS
A.3. SELECTION CONCLUSIONS 247
Testing method Labor- Conditions and remarks Costs

atory∗ (excl.
VAT)
Laboratory tests
Resonant column test a) c one sample and four consolidation stages 2300 e
b) c same as a) 1850 e
Free torsion pendulum n one sample and one consolidation stage 650 e
test
Bender element test a) n single test on one sample and one conso- 250 e
(pulse test) lidation stage
b) n together with a conventional triaxial test 680 e
to determine the shear parameters, inclu-
ding: three samples with one consolida-
tion stage each
Cyclic triaxial test a) n one sample, one consolidation stage, in- 2100 e
cluding on-specimen measurement of the
sample dimensions and vertical arrival-
time-BE-test, staggered increase of cyclic
loading until sample failure
b) c one sample, one consolidation stage, se- 900 e
veral cyclic loading amplitudes
Field tests
SCPT n profile to 12 m depth, seismic registrati- 550 e
ons every meter, s-wave profile; exclusive
the costs for the CPT-truck, transport to
the site and working hours of the techni-
cian operating the CPT-truck
c CPT-truck and transport to a site in 100 to
Flanders (price dependent on the site si- 250 e
tuation)
c operating on the site by one person, price 60 e
per hour
Cross-hole test a) n profile to 12 m depth, seismic registra- 3000 e

tions every meter, application of a p-
and a s-wave source, p- and s-wave velo-
city determination, including one source
borehole and two receiver holes
b) c same as a) 6000 e
SASW test a) n one profile 200 e

b) n one day of site work and one day of signal 1500 to
processing and reporting 1800 e
∗ c→ commercial expert, n→ noncommercial institute (university, research institute)
Table A.2: Typical costs for laboratory and in situ tests.

Bibliography
Airey, D. W. & Wood, D. M., 1987. An evaluation of direct simple shear tests
on clay. Geotechnique, 37(1):25–35.
Areias, L. & Haegeman, W., 2001. In Situ Tests Retie-Waremme-Lincent.

Interim Report IV.1.16.3 (19/09/2001), Ghent University.
Areias, L., Van Impe, W. F. & Haegeman, W., 1999. Variation of shear wave
energy with coupling stress in the SCPT method. European Journal of En-
vironmental and Engineering Geophysics, 4:87–95.
Areias, L., Haegeman, W. & Van Impe, W. F., 2004. New de-coupled shear
wave source for the SCPT test. In Proceedings of the 2nd International
Conference on Geotechnical Site Characterization, volume 1, pages 643–645.
Porto.
Arroyo, M., 2001. Pulse tests in soil samples. Ph.D. thesis, University of
Bristol.
Arulnathan, R., Ross, W. B. & Riemer, M. F., 1998. Analysis of Bender

Element Tests. Geotechnical Testing Journal, 21(2):120–131.
Best, R., 1991. Digitale Meßwertverarbeitung. Oldenbourg, Munich, Vienna.
Biot, M., 1956a. Theory of propagation of elastic waves in a fluid saturated

porous solid. I. Low frequency range. Journal of the Acoustical Society of
America, 28(2):168–178.
Biot, M., 1956b. Theory of propagation of elastic waves in a fluid saturated

porous solid. II. High frequency range. Journal of the Acoustical Society of
America, 28(2):179–191.
Blewett, J., Blewett, I. J. & Woodward, P. K., 1999. Measurement of shear-

wave velocity using phase-sensitive detection techniques. Canadian Geotech-
nical Journal, 36:934–939.
Bolton, M. D. & Wilson, M. R., 1989. An experimental and theoretical compari-

son between static and dynamic torsional soil tests. Geotechnique, 39(4):585–
599.
249
250 BIBLIOGRAPHY
Brignoli, E. G. M., Gotti, M. & Stokoe, K. H. II, 1996. Measurement of

shear waves in laboratory specimens by means of piezoelectric transducers.
Geotechnical Testing Journal, 19(4):384–397.
Brocanelli, D. & Rinaldi, V., 1998. Measurement of low-strain material dam-

ping and wave velocity with bender elements in the frequency domain. Ca-
nadian Geotechnical Journal, 35:1032–1040.
Buttkus, B., 1991. Spektralanalyse und Filtertheorie in der angewandten Geo-

physik. Springer, Berlin.
Cascante, G. & Santamarina, C., 1997. Low Strain Measurements Using Ran-
dom Noise Excitation. Geotechnical Testing Journal, 20(1):29–39.
Cascante, G., Santamarina, C. & Yassir, N., 1998. Flexural excitation in a

standard torsional-resonant column device. Canadian Geotechnical Journal,
35:478–490.
Chree, C., 1889. The equations of an isotropic elastic solid in polar and cy-
lindrical coordinates, their solutions and applications. Transactions of the
Cambridge Philosophical Society, 14:250.
Clough, R. W. & Penzien, J., 1993. Dynamics of structures. McGraw-Hill,

New York, 2nd edition.
Cowper, G. R., 1966. The Shear Coefficient in Timoshenko´s Beam Theory.

Journal of Applied Mechanics, pages 335–340.
Dobry, R. & Vucetic, M., 1987. Dynamic properties and seismic response of
soft clay deposits. In International Symposium on Geotechnical Engineering
of Soft Soils, volume 2, pages 51–87. Mexico City.
Duncan, J. M. & Chang, C. Y., 1970. Nonlinear analysis of stress and strain
in soils. Journal of the Soil Mechanics and Foundation Division, ASCE,
96(SM5):1629–1653.
Dyvik, R. & Madshus, C., 1985. Lab Measurements of Gmax using Bender
Elements. In Advances in the Art of Testing Soils Under Cyclic Conditions,
pages 186–196. ASCE, New York.
Elgamal, A.-W., Zeghal, M., Tang, H. T. & Stepp, J. C., 1995. Lotung Down-
hole Array. I: Evaluation of Site Dynamic Properties. Journal of Geotechnical
Engineering, 121(4):350–362.
Eringen, A. C. & Şuhubi, E. S., 1975. Elastodynamics, Linear Theory, volume 2.

Academic Press, New York, San Francisco, London.
Ewins, D. J., 1988. Modal Testing: Theory and Practice. Research Studies
Press, Taunton.
BIBLIOGRAPHY 251
Fratta, D. & Santamarina, J. C., 1996. Wave Propagation in Soils: Multi-Mode,

Wide-Band Testing in a Waveguide Device. Geotechnical Testing Journal,
19(2):130–140.
Geologica N.V., 2000. Soil Expertise Retie Molsebaan 43. Technical Report
D23923, Geologica N.V., Tervuursesteenweg 200, 3060 Bertem, Belgium.
Ghayamghamian, M. R. & Kawakami, H., 2000. On-site nonlinear hysteresis

curves and dynamic soil properties. Journal of Geotechnical and Geoenvi-
ronmental Engineering, 126(6):543–555.
Graff, K. F., 1991. Wave Motion in elastic Solids. Dover Publications, New
York.
Haegeman, W., 2002. Studie van de dynamische glijdingsmoduli via bender

elementproeven op de testsite te Waremme. Technical Report IV.1.16.3,
Ghent University.
Hafner, E., 1977. Untersuchung der hydrodynamischen Kräfte auf Baukörper

im Tiefwasserbereich des Meeres. Institut für Wasserbau der Universität
Stuttgart, Stuttgart.
Hall, L. & Bodare, A., 2000. Analyses of the cross-hole method for determining
shear wave velocities and damping ratios. Soil Dynamics and Earthquake
Engineering, 20:167–175.
Hardin, B. O., 1978. The Nature of the Stress-Strain Behavior for Soils. In
Earthquake Engineering and Soil Dynamics, volume 1, pages 3–90. ASCE,
New York.
Hardin, B. O. & Black, W. L., 1969. Vibration Modulus of Normally Consolida-

ted Clay. Journal of the Soil Mechanics and Foundation Division, 95:1531–
1537.
Hardy, S., Zdravković, L. & Potts, D. M., 2003. Numerical interpretation of

continuously cycled bender element tests. Department of Civil & Environ-
mental Engineering, Imperial College, London.
Iglesias, A. M., 2000. Investigating Various Modal Analysis Extraction Tech-

niques to Estimate Damping Ratio. Master’s thesis, Virginia Polytechnic
Institute and State University, Blacksburg, Virginia.
Ishibashi, I. & Zhang, X., 1993. Unified dynamic shear moduli and damping
ratios of sand and clay. Soils and Foundations, 33(1):182–191.
ISSMGE TC 10, 2004. Seismic Cone Downhole Procedure to Measure Shear

Wave Velocity - A Guidline (Draft Version 4). Meeting of TC 10 on occasion
of ISC’2 in Porto.
252 BIBLIOGRAPHY
Jamiolkowski, M., Leroueil, S. & Lo Presti, D. C. F., 1991. Theme Lecture:

Design Parameters from Theory to Practice. In Geo-Coast ’91, pages 1–41.
Yokohama.
Johnson, S. A., Greenleaf, J. F., Ritman, E. L., Harris, L. D. & Rajagopolan,
B., 1978. Three-dimensional analysis and display of rock and soil masses for
site characterization: potential contributions from medical imaging techni-
ques. In Specialty Workshop on Site Characterization and Exploration, pages
322–334. Northwestern University, Evanston, Illinois.
Jovičić, V., Coop, M. R. & Simić, M., 1996. Objective criteria for the determi-
ning Gmax from bender element tests. Geotechnique, 46:357–362.
Kammeyer, K. D. & Kroschel, K., 1998. Digitale Signalverarbeitung, Filterung
und Spektralanalyse. Teubner, Stuttgart, 4th edition.
Karl, L. & Haegeman, W., 2004. In situ measurement of material damping
with seismic cone penetration test. In Viana da Fonseca & Mayne, editors,
Proceedings of the 2nd International Conference on Geotechnical Site Cha-
racterization, volume 2, pages 1655–1662. Porto.
Karl, L., Haegeman, W., Pyl, L. & Degrande, G., 2003. Measurement of mate-
rial damping with bender elements in triaxial cell. In Proceedings of the Third
International Symposium on Deformation Characteristics of Geomaterials,
Vol. 1, pages 3–11. Lyon.
Kramer, S. L., 1996. Geotechnical earthquake engineering. Prentice Hall.
Kudo, K. & Shima, E., 1981. Attenuation of shear waves in soil. In M. Toksöz
& D. Johnston, editors, Seismic Wave Attenuation, Geophysics reprint series
No. 2, pages 325–338. Society of Exploration Geophysicists.
Lai, C. G., 1998. Simultaneous inversion of Rayleigh phase velocity and atte-
nuation for near-surface site characterization. Ph.D. thesis, Georgia Institute
of Technology.
Lai, C. G., Rix, G. J., Foti, S. & Roma, V., 2002. Simultaneous measure-
ment and inversion of surface wave dispersion and attenuation curves. Soil
Dynamics and Earthquake Engineering, 22:923–930.
Lin, M.-L., Huang, T.-H. & You, J.-C., 1996. The effects of frequency on
damping properties of sand. Soil Dynamics and Earthquake Engineering,
15:269–278.
Lings, M. L. & Greening, P. D., 2001. A novel bender/extender element for
soil testing. Geotechnique, 51(8):713–717.
Lytle, R. J., 1978. Geophysical characterization using advanced data proces-
sing. In Specialty Workshop on Site Characterization and Exploration, pages
291–300. Northwestern University, Evanston, Illinois.
BIBLIOGRAPHY 253
Martı́nez-Carvajal, H., Taboada-Urtuzuástegui, V. M. & Romo, M. P., 2001.

Analysis of some downhole acceleration records from ”CENTRAL DE AB-
ASTO OFICINAS” site at Mexico City. In Fourth International Conference
on Resent Advances in Geotechnical Earthquake Engineering and Soil Dy-
namics and Symposium in Honor of Professor W.D. Liam Finn, pages 1–6,
paper no. 9.10. San Diego.
Mengé, P., 2001. Soil investigation results at Sint-Katelijne-Waver. In Holey-
man, editor, Screw Piles - Installation and Design in Stiff Clay, pages 19–62.
Swets & Zeitlinger.
Mok, Y. J., Sanchez-Salinero, I., Stokoe, K. H. II & Roesset, J. M., 1988.
In situ damping measurement by crosshole seismic method. In Earthquake
Engineering and Soil Dynamics II. ASCE Spec. Conferences, volume No. 20,
pages 305–320. ASCE, Park City, Utah.
Molenkamp, F. & Smith, I. M., 1980. Hysteretic and viscous material damping.
International journal for numerical and analytical methods in geomechanics,
4:293–311.
Nakagawa, K., Soga, K. & Mitchell, J. K., 1996. Pulse Transmission System for
Measuring Wave Propagation in Soils. Journal of Geotechnical Engineering,
122(4):302–308.
Nazarian, S. & Desai, M. R., 1993. Automated Surface Wave Method: Field
Testing. Journal of Geotechnical Engineering, 119(7):1094–1111.
Ormsby, J. F. A., 1961. Design of numerical filters with applications to
missile data processing. Journal of the Association for Computing Machi-
nery, 8(3):440–466.
Pochhammer, L., 1876. Über die Fortpflanzungsgeschwindigkeiten kleiner
Schwingungen in einem unbegrenzten isotropen Kreiszylinder. Journal für
die reine und angewandte Mathematik, 81:324–336.
Pyl, L. & Degrande, G., 2001a. Determination of the dynamic soil characte-
ristics with the SASW method at a site in Lincent. Internal Report BWM-
2001-02, K.U. Leuven.
Pyl, L. & Degrande, G., 2001b. Determination of the dynamic soil charac-
teristics with the SASW method at a site in Waremme. Internal Report
BWM-2001-11, K.U. Leuven.
Pyl, L., Lombaert, G. & Degrande, G., 2000. Determination of the dynamic
soil characteristics with the SASW method at a site in Retie. Internal Report
BWM-2000-12, K.U. Leuven.
Richardson, M. H. & Formenti, D. L., 1982. Parameter Estimation from Fre-
quency Response Measurements using Rational Fraction Polynomials. In
International Modal Analysis Conference, pages 167–181.
254 BIBLIOGRAPHY
Richart, F. E., Hall, J. R. & Woods, R. D., 1970. Vibration of soils and
foundations. Prentice-Hall, Englewood Cliffs, N. J.
Rossit, C. A. & Laura, P. A. A., 2001. Transverse, normal modes of vibration

of a cantilever Timoshenko beam with a mass elastically mounted at the free
end. Journal of the Acoustical Society of America, 6(110):2837–2840.
Sanches-Salinero, I., Roesset, J. M. & Stokoe, K. H. II, 1986. Analytical studies

of body wave propagation and attenuation. Report GR 86-15, University of
Texas, Austin.
Seco, F., Martı́n, J. M., Jiménez, A., Pos, J. L., Calderón, L. & Ceres, R.,
2002. PCDISP: A tool for the simulation of wave propagation in cylindrical
waveguides. In 9th International Congress on Sound and Vibration. Orlando,
Florida.
Seed, H. B. & Idriss, I. M., 1970. Soil moduli and damping for dynamic response
analysis. Report EERC 70-10, UCB.
Stewart, W. P., 1992. Insitu measurements of dynamic soil properties with

emphasis on damping. Ph.D. thesis, University of British Columbia, Depart-
ment of Civil Engineering, Vancouver.
Stoll, R. D., 1979. Experimental studies of attenuation in sediments. Journal

of the Acoustical Society of America, 66(4):1152–1160.
Storrer, O., Van Den Broeck, M. & Van Impe, W. F., 1986. A numerical model
analysing free torsion pendulum results. In 2nd International Symposium
on Numerical Models in Geomechanics, pages 265–273. Ghent University,
Ghent.
Studer, J. A. & Koller, M. G., 1997. Bodendynamik; Grundlagen, Kennziffern,

Probleme. Springer, Berlin, 2 edition.
Thurston, R. N., 1978. Elastic waves in rods and clad rods. Journal of the
Acoustical Society of America, 64(1):1–37.
Thurston, R. N., 1992. Elastic Waves in Rods and optical Fibers. Journal of
Sound and Vibration, 159(3):441–467.
Udı́as, A., 1999. Principles of Seismology. Cambridge University Press, Cam-

bridge, 1 edition.
Van Impe, W., 1977. De vrije wringingsproef: een belangrijke bijdrage tot
de studie van het gedrag van funderingen onder invloed van grondtrillingen.
Tijdschrift der Openbare Werken van België, pages 1–18.
Viggiani, G. & Atkinson, J. H., 1995. Interpretation of Bender Element Tests.

Geotechnique, 45(1):149–154.
BIBLIOGRAPHY 255
Wang, Y. H., Yan, W. M. & Lo, K. F., 2004. Laboratory and in-situ measu-
rement of attenuation in soil. In Viana da Fonseca & Mayne, editors, 2nd
Conference on Geotechnical and Geophysical Site Characterization, volume 2,
pages 1883–1889. Porto.
Weinstein & Clifford, 1979. Programs for Digital Signal Processing. IEEE
Press, New York.
Wilson, J. F., 1984. Dynamics of offshore structures. John Wiley & Sons, New
York.
Winkler, K. & Nur, A., 1979. Pore Fluids and Seismic Attenuation in Rocks.
Geophysical Research Letters, 6(1):1–4.
Yoon, Y.-W., 1991. Static and dynamic behavior of crushable and non-crushable
sands. Ph.D. thesis, Ghent University, Laboratory of Soil Mechanics.
Yuan, D. & Nazarian, S., 1993. Automated Surface Wave Method: Inversion
Technique. Journal of Geotechnical Engineering, 119(7):1112–1126.
Zavoral, D. Z. & Campanella, R. G., 1994. Frequency effects on dam-
ping/modulus of cohesive soil. In R. J. Ebelhar, V. P. Drnevich & B. L.
Kutter, editors, Dynamic Geotechnical Testing II, ASTM STP 1213, pages
191–201. ASTM, Philadelphia.
Zeghal, M. & Elgamal, A.-W., 1994. Analysis of site liquefaction using earth-
quake records. Journal of Geotechnical Engineering, 120(6):996–1017.
Zeghal, M., Elgamal, A.-W., Tang, H. T. & Stepp, J. C., 1995. Lotung Down-
hole Array. II: Evaluation of Soil Nonlinear Properties. Journal of Geotech-
nical Engineering, 121(4):363–377.

Bender Thesis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bender Thesis

Uploaded by

Copyright:

Available Formats

Ghent University

Dynamic Soil Properties out of SCPT

A thesis submitted in accordance with the requirements of

Academic year 2004–2005

Laboratory of Soil Mechanics

Tel. +36 9 264 57 23

Frequently used symbols and units xi

I Introduction and background 1

2 Dynamic soil properties 7

3 Methods to determine G and D 19

II Characterization of the testing sites 35

5 Test site Lincent 53

6 Test site Waremme 67

7 Test site Sint-Katelijne-Waver 85

7.3.2 Cone penetration test (CPT) . . . . . . . . . . . . . . . 88

8 Test site Ghent 97

III Studies on testing methods 105

10 Bender element technique 173

11 Conclusions and summary 235

Symbols and units

A signal amplitude, amplitude spectrum; m, m/s, m/s2 ;

I area moment of inertia; impedance m4 ; kg/(m2 s)

α attenuation coefficient 1/m

1.1 Purpose and scope

damping. Since the velocity of p-waves is highly affected by the groundwater

1.2 Outline of the thesis

Dynamic soil properties

2.1 Dynamic shear modulus

Gmax = ρ Vs2 (2.1)

Increasing parameter Gmax

Table 2.1: Influence of different parameters on Gmax for normally to moderately

A large amount of laboratory test results, primarily from resonant column

Gmax = 625 Fe pa1−n σ00 n OCRk (2.2)

exponent n. Hardin & Black (1969) introduced originally

This empirical formulation is based on a large amount of data and covers a

2.2 Attenuation parameters of soils

u(t) = u0 sin ωt (2.13)

whereas ω is the excitation frequency. The reaction force F(t) is:

F (t) = k u(t) + c u̇(t) = k u0 sin ωt + c u0 ω cos ωt (2.14)

Figure 2.4: Damped SDOF system subjected to an external displacement u(t).

After introduction of integration limits the dissipated energy in one cycle of

T = 2π/ω is the period. At maximum displacement, the velocity is zero and

The fraction of critical damping is abbreviated as c/cc = ξ and the frequency

c ED, vis Aloop, vis

The aim can be achieved by changing the frequency independent damping

ED, hys = πu20 ω ceq (2.23)

If Equation (2.17) is used one can write in analogy to Equation (2.19)

In this case hysteretic damping can be approximated by an equivalent hys-

Isotropic visco-hysteretic elastic material model The further treatment

σij = 2Geij + 3Km δij + 2G0 ėij + 3K 0 m δij (2.29)

σij = sij + σm δij (2.30)

σm = 3Km + 3K 0 ˙m (2.31)

σm = 3K ∗ ˆm ei(ωt+φm ) (2.33)

A material is considered as hysteretic when the dissipated energy per cycle

It will be useful in the methods for determination the damping ratio of

this is linked to the complex wave number by

The Equations (2.35), (2.37), (2.41) and (2.42) conclude in an expression

In most problems, it is assumed that there is no dissipation of energy in pure

with ν the ratio of Poisson.

Increasing parameter Damping ratio, D

Table 2.2: Influence of different parameters on D for normally to moderately

3.1 Laboratory tests

Shear Strain Amplitude γ [%]

v ibrations caused by traffic,

σij = 2Geij + 3Km δij + 2G0 ėij + 3K 0 m δij (2.29)

σm = 3Km + 3K 0 ˙m (2.31)

σm = 3K ∗ ˆm ei(ωt+φm ) (2.33)