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 20042005

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

Contents
Preface

Frequently used symbols and units

xi

Introduction and background

1 Introduction
1.1 Purpose and scope . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . .

3
3
4

2 Dynamic soil properties

2.1 Dynamic shear modulus . . . . . . . . . . . . . . . . . . . . . .
2.2 Attenuation parameters of soils . . . . . . . . . . . . . . . . . .

7
7
11

3 Methods to determine G and D

3.1 Laboratory tests . . . . . . . . . . . . . . . . . . . .
3.1.1 Piezoelectric bender and compression element
3.1.2 Cyclic triaxial tests . . . . . . . . . . . . . . .
3.1.3 Cyclic simple shear tests . . . . . . . . . . . .
3.1.4 Cyclic torsional shear tests . . . . . . . . . .
3.1.5 Resonant column test . . . . . . . . . . . . .
3.1.6 Free torsion pendulum test . . . . . . . . . .
3.2 In situ tests . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Seismic reflection test . . . . . . . . . . . . .
3.2.2 Seismic refraction test . . . . . . . . . . . . .
3.2.3 Spectral analysis of surface waves (SASW) .
3.2.4 Seismic cross-hole test . . . . . . . . . . . . .
3.2.5 Seismic down-hole and up-hole test . . . . . .
3.2.6 Seismic cone penetration test . . . . . . . . .
3.2.7 Geotomography . . . . . . . . . . . . . . . . .
3.2.8 High-strain tests . . . . . . . . . . . . . . . .

19
19
21
22
23
23
23
24
28
28
29
30
31
32
33
33
34

vii

. . .
tests
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

viii

II

CONTENTS

Characterization of the testing sites

4 Test site Retie

4.1 Introduction . . . . . . . . . . . . . . . . .
4.2 Borings and undisturbed sampling . . . .
4.3 Cone penetration test (CPT) . . . . . . .
4.4 Seismic cone penetration test (SCPT) . .
4.4.1 Test description . . . . . . . . . . .
4.4.2 Test results for the wave velocities
4.4.3 Test results for the damping ratio
4.5 Spectral analysis of surface waves (SASW)
4.5.1 Test description . . . . . . . . . . .
4.5.2 Test results . . . . . . . . . . . . .
4.6 Overview of the test results Retie . . . . .

35
.
.
.
.
.
.
.
.
.
.
.

37
37
38
41
41
41
44
45
47
47
48
49

5 Test site Lincent

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Borings and undisturbed sampling . . . . . . . . . . . . . . . .
5.3 Cone penetration test (CPT) . . . . . . . . . . . . . . . . . . .
5.4 Seismic cone penetration test (SCPT) . . . . . . . . . . . . . .
5.4.1 Remarks on the testing Setup . . . . . . . . . . . . . . .
5.4.2 Test results for the wave velocity . . . . . . . . . . . . .
5.4.3 Test results for the damping ratio . . . . . . . . . . . .
5.5 Spectral analysis of surface waves (SASW) . . . . . . . . . . . .
5.5.1 Remarks on the testing setup and inversion assumptions
5.5.2 Test results . . . . . . . . . . . . . . . . . . . . . . . . .
5.6 Overview of the test results Lincent . . . . . . . . . . . . . . .

53
53
53
57
60
60
61
62
62
62
63
64

6 Test site Waremme

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 Borings and undisturbed sampling . . . . . . . . . . . .
6.3 In situ tests . . . . . . . . . . . . . . . . . . . . . . . . .
6.3.1 Cone penetration test (CPT) . . . . . . . . . . .
6.3.2 Seismic cone penetration test (SCPT) . . . . . .
6.3.3 Spectral analysis of surface waves (SASW) . . .
6.4 Laboratory tests to obtain Gmax and D . . . . . . . . .
6.4.1 Bender elements with time arrival interpretation
6.4.2 Free torsion pendulum test . . . . . . . . . . . .
6.4.3 Resonant column test . . . . . . . . . . . . . . .
6.5 Overview of the test results Waremme . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

67
67
67
71
71
72
76
77
77
79
80
81

7 Test site Sint-Katelijne-Waver

7.1 Introduction . . . . . . . . . . . . . . . . .
7.2 Borings and undisturbed sampling . . . .
7.3 In situ tests . . . . . . . . . . . . . . . . .
7.3.1 Marchetti dilatometer test (DMT)

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

85
85
85
87
87

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

CONTENTS

7.4

7.5

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

7.3.3 Seismic cone penetration test (SCPT) . . . . . .
7.3.4 Spectral analysis of surface waves (SASW) . . .
7.3.5 Seismic refraction test (SRT) . . . . . . . . . . .
Laboratory tests to obtain Gmax and D . . . . . . . . .
7.4.1 Bender elements with time arrival interpretation
7.4.2 Free torsion pendulum test . . . . . . . . . . . .
7.4.3 Resonant column test . . . . . . . . . . . . . . .
Overview of the test results St.-Katelijne-Waver . . . . .

8 Test site Ghent

8.1 Introduction . . . . . . . . . . . . . . . . .
8.2 Borings and sampling . . . . . . . . . . .
8.3 Cone penetration test (CPT) . . . . . . .
8.4 Seismic cone penetration test (SCPT) . .
8.4.1 Testing setup . . . . . . . . . . . .
8.4.2 Test results for the wave velocity .
8.4.3 Test results for the damping ratio
8.5 Overview of the test results Ghent . . . .

III

ix

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

97
.
97
.
97
.
99
.
99
.
99
. 100
. 100
. 101

Studies on testing methods

9 SCPT based techniques

9.1 Motivation and purpose . . . . . . . . . . . . . . . . . . . .
9.2 Applied equipment and selection criteria . . . . . . . . . . .
9.2.1 Seismic source . . . . . . . . . . . . . . . . . . . . .
9.2.2 Seismic cones . . . . . . . . . . . . . . . . . . . . . .
9.2.3 Data acquisition system . . . . . . . . . . . . . . . .
9.3 Methods for the shear modulus . . . . . . . . . . . . . . . .
9.3.1 Direct time methods . . . . . . . . . . . . . . . . . .
9.3.2 Indirect time methods . . . . . . . . . . . . . . . . .
9.4 Methods to obtain the Damping Ratio . . . . . . . . . . . .
9.4.1 Attenuation coefficient method . . . . . . . . . . . .
9.4.2 Spectral ratio slope method . . . . . . . . . . . . . .
9.4.3 Hysteresis stress-strain-curves . . . . . . . . . . . . .
9.5 Artificially generated signals . . . . . . . . . . . . . . . . . .
9.5.1 Calculated velocities from the simulated signals . . .
9.5.2 Calculated damping ratio . . . . . . . . . . . . . . .
9.5.3 Conclusions for the analysis of real SCPT data . . .
9.6 Measurements performed at the test sites . . . . . . . . . .
9.6.1 Campaign I (geophones) . . . . . . . . . . . . . . . .
9.6.2 Campaign II (HAMLIN accelerometers) . . . . . . .
9.6.3 Campaign III, Ghent (Br
uel & Kjr accelerometers)
9.7 Summary and remaining problems . . . . . . . . . . . . . .

88
88
90
90
91
91
92
93
94

105
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

107
107
108
108
111
117
119
121
121
125
125
128
131
142
151
152
156
160
160
160
166
169

CONTENTS

10 Bender element technique

10.1 Motivation and purpose . . . . . . . . . . . . . . . . .
10.2 Description of the equipment . . . . . . . . . . . . . .
10.2.1 Bender elements . . . . . . . . . . . . . . . . .
10.2.2 Signal generation and measurement apparatus
10.3 Techniques to determine Gmax . . . . . . . . . . . . .
10.3.1 Wave travel distance . . . . . . . . . . . . . . .
10.3.2 Selection of the input-signal shape . . . . . . .
10.3.3 Methods for determining the travel time . . . .
10.3.4 Difficulties in the arrival time determination . .
10.4 Methods to obtain the damping ratio . . . . . . . . . .
10.4.1 Resonant method . . . . . . . . . . . . . . . . .
10.4.2 Use of multiple shear wave arrivals . . . . . . .
10.4.3 Use of different travel paths . . . . . . . . . . .
10.5 The resonant method . . . . . . . . . . . . . . . . . . .
10.5.1 Introduction . . . . . . . . . . . . . . . . . . .
10.5.2 Site in Waremme . . . . . . . . . . . . . . . . .
10.5.3 Site in Sint-Katelijne-Waver . . . . . . . . . . .
10.6 Summary and remarks . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

173
173
174
174
176
178
178
179
179
182
186
186
204
210
213
213
213
215
216

11 Conclusions and summary

235

IV

239

Appendix

A Accuracy and costs of soil

A.1 Laboratory tests . . . .
A.2 Field tests . . . . . . . .
A.3 Selection conclusions . .
Bibliography

tests
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .

241
242
243
244
249

Frequently used symbols

and units
The following table presents the most frequently used symbols and abbreviations. 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
a
c
CA
D
d
Ds
E
Emax , E0
EI
e
f
fcut
Fe
FFT

G
Gmax , G0
H
h
IFFT

signal amplitude, amplitude spectrum;

area
sample radius, distance;
time history
cohesion; damping coefficient;
wave velocity
added mass coefficient
damping ratio
diameter, depth, distance
damping ratio of the s-wave
elasticity modulus; energy
dynamic elasticity modulus
bending stiffness
void ratio
frequency
cut-off frequency
void ratio function
Fast Fourier Transformation
as a synonym for a discrete
Fourier Transformation in general
shear modulus
dynamic shear modulus
transfer function; height
height
inverse transformation to the FFT
xi

m, m/s, m/s2 ;
m2
m;
m, m/s, m/s2
Pa; kg/s;
m/s

, %
m
, %
Pa; J
Pa
Nm2

Hz
Hz

Pa
Pa
; m
m

xii

FREQUENTLY USED SYMBOLS

I
Ip
J
K
k
ki
l, L
m
m
a
OCR
p
pa
Q
qc
R
r
S
s
s
T
t
u(t)
Vp
Vs
x, y, z
y
z

n , n+1
t
x

area moment of inertia; impedance

plasticity index
mass polar moment of inertia
bulk modulus
stiffness; wavenumber;
shear coefficient; permeability
intrinsic permeability
length
mass
added mass per unit length
over consolidation ratio
pseudo simultaneous signal
atmospheric pressure
quality factor
cone resistance
cross power spectrum; receiver
cross-correlated signal; receiver signal;
radius, distance
spectrum of signal s; source
time history
mean absolute deviation
oscillation period
time
displacement function
compression wave velocity
shear wave velocity
coordinates in space
mode shape (displacement)
derivative of the logarithmic
amplitude ratio; distance
attenuation coefficient
porosity
shear strain; unit weight
logarithmic decrement
amplitudes of an oscillation
time shift
distance
viscosity
wavelength
ratio of poisson
fraction of critical damping
density
normal stress
mean effective stress

m4 ; kg/(m2 s)
, %
kg m2
Pa
kg/s2 ; 1/m;
; m/s
m2
m
kg
kg/m

m, m/s, m/s2
Pa

N/m2
s m2 ; ...;
m
m, m/s, m/s2
s
s
m
m/s
m/s
m
m
s/m;
m
1/m

, %; N/m3

m
s
m
Pa s
m

, %
kg/m3
Pa
Pa

FREQUENTLY USED SYMBOLS

1,2,3

isotropic confining stress

shear stress; time shift
phase shift; angle of internal friction
circular frequency
circular frequency at resonance

xiii
Pa
Pa; s
; o
1/s
1/s

Subscripts
B
k
kin
max
min
p
pot
R
r
s
T

related to the bottom sensor of the seismic cone

related to bulk compression
kinematic
maximum
minimum
related to p-waves
potential
related to a receiver signal
related to Rayleigh-waves
related to s-waves; related to a sender signal
related to the top sensor of the seismic cone

Superscripts

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

related to an amplitude of a parameter

related to a complex conjugate

xiv

FREQUENTLY USED SYMBOLS

Part I

Introduction and
background

Chapter 1

Introduction
1.1

Purpose and scope

The evaluation of a number of civil engineering problems involve the transmission of waves through soil, such as seismic response under earthquake loading,
foundation response under dynamic loading and vibrations in buildings, induced 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- (secondary, 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 perpendicular 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 pwaves and the material damping ratios of these body waves. Material damping
in the soil represents energy dissipation caused by friction between solid particles 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

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 techniques 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 buildings initiated from the K.U. Leuven and Ghent University. Within the framework 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, SintKatelijne-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, Lincent, Waremme, Sint-Katelijne-Waver and Ghent. Apart from a general description 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

Chapter 9 is dedicated to the SCPT. The selected test equipment and developed 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. Procedures 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.

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 stressstrain 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 = 5 103 %. 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
5 103 %. 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 = Vs2

(2.1)

with the soil density () 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

CHAPTER 2. DYNAMIC SOIL PROPERTIES

Damping ratio, D

Shear modulus, G

Figure 2.1: Cyclic loading: load history.

Figure 2.2: Cyclic loading: hysteresis loops.

D
Shear strain,

Figure 2.3: Shear modulus and damping ratio in function of shear strain.

2.1. DYNAMIC SHEAR MODULUS

Laboratory tests have shown that soil stiffness is influenced mainly by cyclic
strain amplitude, void ratio, mean effective stress, plasticity index, overconsolidation 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
mean effective stress, 0
void ratio, e0
geological age, tg
cementation, c
overconsolidation ratio, OCR
plasticity index, Ip

Strain rate,

number of loading cycles, N

Gmax
increases with 0
decreases with e0
increases with tg
increases with c
increases with OCR
increases with Ip if OCR > 1;
stays approximately constant
if OCR = 1
no effect for non-plastic soils;
increases with for the case of
plastic materials
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 (0 = (1 + 2 + 3 )/3), overconsolidation ratio (OCR) and plasticity
index (Ip ):
Gmax = 625 Fe pa1n 0 n OCRk

(2.2)

with the atmospheric pressure (pa 100kPa) in the same units as Gmax and
0 , an empirical exponent k related to the plasticity index Ip
k

Ip0.72
0.5
50

(2.3)

the void ratio function Fe , varying somewhat in different studies, and a stress

10

CHAPTER 2. DYNAMIC SOIL PROPERTIES

exponent n. Hardin & Black (1969) introduced originally

Fe = 0.51

(2.973 e0 )2
1 + e0

(2.4)

this has later been improved to cover a wider range of void ratios in Hardin
(1978):
1
0.3 + 0.7e20

Fe =

(2.5)

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 individual 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 penetration 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 :
G=

Gmax
1 + /r

(2.7)

r =

max
Gmax

(2.8)

with

max represents the maximum shear stress before failure occurs. 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
as a function of mean effective stress and plasticity index only:
G
Gmax

= K(, Ip ) (0 ) m(,Ip )m0

(2.9)

with

m(, Ip ) m0 = 0.272 1 . . .
. . . tanh ln

0.000556

0.4

1.3

e0.0145 Ip

(2.10)

2.2. ATTENUATION PARAMETERS OF SOILS

11

K(, Ip ) = 0.5 1 . . .
. . . + tanh ln

0.000102 + n(Ip )

0.492

0.0
for Ip = 0

(sandy soils)

6 1.404

3.37

10
I
for
0
< Ip 15

(low plastic soils)

n(Ip ) =
7.0 107 Ip1.976 for 15 < Ip 70

(medium plastic soils)

5 1.115

2.7

10
I
for
I

p > 70
p

(high plastic soils)

(2.11)

(2.12)

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 dominating 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 t

(2.13)

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

F (t) = k u(t) + c u(t)

= k u0 sin t + c u0 cos t

(2.14)

The energy dissipated during the oscillation can be obtained using the relation du = u dt as:
ED =

c u du =

c u u dt

(2.15)

12

CHAPTER 2. DYNAMIC SOIL PROPERTIES

u(t)
c
F(t)

Q(t)

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:
T

Aloop, vis =E
D, vis =
0

c u u dt = u20 c

(2.16)

T = 2/ 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
AAOB =E
S=

1 2
ku
2 0

(2.17)

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
2 c
=
ES
k
and by using the relationsfor the natural frequency 0 =
damping cc = 2m0 = 2 k m:
ED, vis
c
= 4
= 4
ES
c c 0

(2.18)
k/m and the critical

(2.19)

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

ratio as /0 = . This leads to a formulation for the ratio of critical damping:
=

c
ED, vis
Aloop, vis
=
=
cc
4 ES
4 AAOB

(2.20)

is a constant for the viscous damped system. But the dissipated energy per
loading cycle ED, vis is proportional to the loading frequency 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.

2.2. ATTENUATION PARAMETERS OF SOILS

13

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:
ceq =

(2.21)

is the loss factor and independent of frequency. The fraction of critical

damping eq is then

ceq
k
k

=
eq =
=
=
(2.22)
cc
2
2
m
2 mk
eq is a function of and therefore also function of the loading frequency .
The loss factor can be expressed as energy ratio. The dissipated energy
of the hysteretic system can be given as
ED, hys = u20 ceq

(2.23)

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

ED, hys
ceq
= 4
ES
c c 0

(2.24)

and after introduction of Equation (2.21)

ED, hys
= 2
ES

(2.25)

This leads to a loss factor of

=

ED, hys
2 ES

(2.26)

The afterwards used material damping ratio D can be derived from the loss
factor by
D=

(2.27)

Therefore Equation (2.26) can be expressed as

D=

ED, hys
Aloop, hys
=
4 ES
4 AAOB

(2.28)

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.

14

CHAPTER 2. DYNAMIC SOIL PROPERTIES

In this case hysteretic damping can be approximated by an equivalent hysteretic 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 ( = 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
ij = 2Geij + 3K

m ij

+ 2G e ij + 3K

m ij

(2.29)

in which G is the shear modulus, K the bulk modulus, G the viscous shear
modulus, K the viscous bulk modulus, m the isotropic strain component, eij
the deviatoric strain tensor component and ij the Kronecker delta.
The equation can be divided into the isotropic stress component m and
the deviatoric component sij
ij = sij + m ij

(2.30)

m = 3K

(2.31)

with
m

+ 3K m

and
sij = 2Geij + 2G e ij

(2.32)

For cyclic deformations the Equations (2.31) and (2.32) can be written as
m = 3K m ei(t+m )

(2.33)

sij = 2G eij ei(t+ij )

(2.34)

and

in which m is the phase of the isotropic stress components, ij the tensor of

the phase of the deviatoric stress component and indicates the amplitude of
a parameter. G and K incorporate moduli and viscous moduli in complex
parameters
G = G + iG

(2.35)

K = K + iK

(2.36)

and

2.2. ATTENUATION PARAMETERS OF SOILS

15

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 volumetric 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
Ds =

G
2G

(2.37)

for the shear deformation and

Dk =

K
2K

(2.38)

for dilatation.
Since D should be frequency independent, G and K have to be inversely
proportional to . This concept is already used in connection with the introduction of the ceq in the SDOF system. G and K can be expressed by
converting of the Equations (2.37) and (2.38) as
G =

2Ds G

K =

2Dk K

with leads introduced in the Equations (2.35) and (2.36) to the complex moduli
G = G + 2iDs G

(2.39)

K = K + 2iDk K

(2.40)

and

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
G

Vs =

(2.41)

this is linked to the complex wave number by

k =

Vs

(2.42)

16

CHAPTER 2. DYNAMIC SOIL PROPERTIES

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

for k = k + k
Ds =

kk
k2 k 2

(2.43)

which can be simplified for a small attenuation (k << k) and writing the attenuation coefficient as = k to
Ds =

=
k
k

(2.44)

If is isolated the in Section 9.4.1 needed relation

= k Ds =

Ds
2 f Ds
=
Vs
Vs

(2.45)

is obtained.
Other measures of energy dissipation are the quality factor Q and the specific damping capacity related to the damping ratio D and the loss factor
by:
Q=

1
,
2D

= 2 D,

= 2D

(2.46)

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 + ) = 3 Dp (1 ) 2 Ds (1 2)

(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
Udas (1999) obtains the following relation
Dp
Qs
2 1 2
=
=
Ds
Qp
3 1

(2.48)

with the ratio of Poisson.

Under consideration of the dependence of p-wave velocity Vp and s-wave
velocity Vs
Vp
=
Vs

1
1 2

(2.49)

Equation (2.48) can also be written in terms of the wave velocities

Dp
Qs
4
=
=
Ds
Qp
3

Vs
Vp

(2.50)

2.2. ATTENUATION PARAMETERS OF SOILS

17

The increase of the damping ratio with the shear strain can be estimated
under the same assumptions used for Equation (2.8):
D
/r
=
Dmax
1 + /r

(2.51)

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
mean effective stress, 0
void ratio, e0
geological age, tg
cementation, c
overconsolidation ratio, OCR
plasticity index, Ip
cyclic strain,
Strain rate,
number of loading cycles, N

Damping ratio, D
decreases with 0 ;
effect decreases with increasing Ip
decreases with e0
decreases with tg
may decrease with c
not affected
decreases with Ip
increases with
stays constant or
may increase with
not significant for
moderate 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.
0.333
D

1.3

1 + e0.0145Ip

2
. . . 1.547

G
+1
Gmax

0.586

G
Gmax

...
(2.52)

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.

18

CHAPTER 2. DYNAMIC SOIL PROPERTIES

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 [%]

Sort of Problem

104

103

101

102

10

seismic surv ey
v ibrations caused by traffic,
blastings and weak earthquakes
machine foundations
strong earthquakes,
farfield of explosions
nearfield of explosions

In situ Tests

refraction and reflection seismic

crosshole, uphole
and downhole seismic
free oscillation tests,
enforced oscillations
dynamically loaded plates,
shear tests

Laboratory Tests

ultrasonic tests
resonant column tests
(cylindrical specimen)
resonant column
tests (hollowcylinder)
cyclic torsional shear
tests (hollowcylinder)
cyclic triaxial test
shaking table

104

103

102

101

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 fullscale problem, more commonly in centrifuges.

Figure 3.2: Cyclic simple shear test

device from Airey & Wood (1987).

3.1.1

Figure 3.3: Cyclic triaxial test device from Kramer (1996).

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

22

CHAPTER 3. METHODS TO DETERMINE G AND D

Top cap
(receiv er)
Bender
elements

Sample

Bottom cap
(sender)

Figure 3.4: Bender elements installed in a triaxial cell.

Figure 3.5: Bender element, positive voltage causes the element to

bend one way, negative voltage causes 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
Emax = Vp2

(1 + )(1 2)
1

(3.1)

with the p-wave velocity (Vp ) and Poissons ratio ().

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 applied cyclically. Isotropic or anisotropic initial stress conditions are possible. A
sketch of the device in given in Figure 3.3. Bedding errors and system compliance effects generally limit the measurements to shear strains greater than
102 %, although local strain devices can produce accurate measurements at
strain levels as small as 104 %.

3.1. LABORATORY TESTS

3.1.3

23

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 specimens 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, gradually increases until the response is locally maximized and the phase shift
between driving signal and measured acceleration signal is equal to . 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).

24

CHAPTER 3. METHODS TO DETERMINE G AND D

Top cap and
mov ing parts
of the oscillator

J 0, m 0

J, m

Sample

Figure 3.6: SDOF system assumed for the behavior of sample and apparatus.

J
n h
n h
=
tan
J0
Vs
Vs

(3.2)

m
n h
n h
=
tan
m0
Vl
Vl

(3.3)

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 n 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 = Vl2

(3.4)

Alternatively to a harmonic sinusoidal excitation, the fundamental natural frequency 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 damping ratio by means of bender elements. Finally, if the applied dynamic force
is quantitatively known by a careful calibration of the apparatus, the amplification 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 inventor, is performed on a sample from the site in Waremme and Sint-KatelijneWaver. 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).

3.1. LABORATORY TESTS

25

Hanger
Dial gauge

Confinement
pressure

Air
Hardin
oscillator
Oil
Sample

Cell

Water

Figure 3.7: Resonant column test

device with Hardin oscillator of the
Drnevich type.

Air pressure chamber

Figure 3.8: Resonant column test

device of the Stokoe type, Kramer
(1996).

Horizontal arm
Counter weights
Sealing bus

Confinement
pressure

Contactless
displacement sensor

Torsion shaft

Cell

Sample

Air

Water

Figure 3.9: Test set-up of the free torsion pendulum test at Ghent University.

26

CHAPTER 3. METHODS TO DETERMINE G AND D

Beam displacement [mm]

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 applied hydrostatic pressure. Due to the lamellas, no slip between the caps and
the sample can occur when afterwards trough the upper cap a torsional moment 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
n

0.2

n+1

0
0.2
0.4
0

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:
G = S 2

2h JS
a4

(3.5)

3.1. LABORATORY TESTS

27

In this equation S is the undamped circular frequency of the equipment including 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 S can be calculated using the damped frequency D , obtained out of the measured vibration
period, and the fraction of critical damping :
S =

D
1 2

(3.6)

is calculated with the help of the logarithmic decrement method:

=

n
1
log
2
n+1

(3.7)

with n and n+1 successive amplitudes of the oscillation.

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 Equation (2.28) assuming ED, vis = ED, hys . It is obvious that for the resonance case,
= 1, the damping ratio D and the fraction of critical damping are equal,
that means:

D = = for = 1
(3.8)
2
The level of shear strain can be estimated by:
=

2 n,n+1 a
3
rh

(3.9)

n,n+1 is the mean oscillation amplitude between the two amplitudes n and
n+1 used to calculate the damping ratio, r is the distance between the measuring 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 103 %. Typical values for small
hammer blows are 102 %. 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 104 % and 101 % of strain. Because of the

28

CHAPTER 3. METHODS TO DETERMINE G AND D

2r
Horizontal arm
Ja
Jp
h 20 cm
2a 10 cm
2r = 99.7 cm

Sample

JS=J p+J a

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-KatelijneWaver.

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 functions. 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.

R
direct wave

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

30

CHAPTER 3. METHODS TO DETERMINE G AND D

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 Snells law given in Equation (3.10).
sin i1
sin i2
=
v1
v2

(3.10)

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 Udas (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.

3.2. IN SITU TESTS

31

x
t

R
head wave

direct wave

v1

v1

I
ic

II

ic

head wave

1/v2

direct wave
1/v1

v2>v 1

xc

Figure 3.13: Ray paths for the seismic refraction test.

(refracted wav e)

Figure 3.14: Time arrival

diagram as result of a seismic 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.

R1
d1

R2
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 experimental 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 & Nazarian (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.

32

CHAPTER 3. METHODS TO DETERMINE G AND D

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 desirable 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 between 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

a)

R1

R2

b)

c)

Figure 3.16: a) Seismic cross-hole test, b) seismic up-hole test, c) seismic downhole test.

3.2. IN SITU TESTS

3.2.6

33

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 predictions 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.

34

3.2.8

CHAPTER 3. METHODS TO DETERMINE G AND D

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 correlation 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 measuring 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. Mertens, 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

01/04/1936

Boring, creamery St. Martin,

Flemish Subsoil Database
Boring, Lageweg 19,
Flemish Subsoil Database
3 CPT performed by
Geologica NV, Bertem

01/08/1977
28/01/2000

08/11/2000

13/12/2000
10/04/2001
12/05/2003

Boring
3 SASW set-ups (two from KUL
and one from UGent, concluding
in an inversion calculation from
KUL and one from UGent)
2 SCPT with a single geophone
cone
2 SCPT with a single geophone
cone
1 SCPT with a dual accelerometer
cone

Abbr.

186.0
CPT1
CPT2
CPT3
B1
SASW KUL
SASW UGent

7.4
3.2
9.0
18.0

SCPT1
SCPT2
SCPT3
SCPT4
SCPT5

10.8
12.3
12.5
12.5
12.3

Table 4.1: Overview of the available in situ tests.

37

Depth
[m]
68.0

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
SCPTs 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
[m]
[m]
0.00
0.50 dark brown
0.50
1.50 light brown
1.50
2.50 beige
2.50
3.50 beige
3.50
4.00 beige
4.00
5.00 beige
5.00
7.00 beige, green
7.00
9.00 beige, green
9.00 14.00 beige, green
14.00 15.00 green
15.00 16.00 green
16.00 17.00 green
17.00 18.00 green
Lithographic Stratification: 0
mation of Mol.

Main Component

Admixtures

topsoil
sand
sand
fine sand
fine sand
fine sand
sand
fine sand
sand
sand
sand
silt
sand
to 5 m quaternary

silt
silt
gravel
silt
silt
silt
sand
deposits, 5 to 18 m for-

Table 4.2: Results of boring B1 at the Retie site.

4.2. BORINGS AND UNDISTURBED SAMPLING

Figure 4.1: Site location plan, Retie.

39

40
From
[m]
0.00
6.00
10.00
34.00
38.00

CHAPTER 4. TEST SITE RETIE

To
[m]
6.00
10.00
34.00
38.00
46.00

Color
yellow
yellow
white
green
gray, white

Main
Component
fine sand
fine sand
fine sand
fine sand
sand

Admixtures

quartz
partly lignite
quartz, partly lignite
clay, partly much glauconite
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 formation 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 sampling 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 Geotechnical 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 calculate 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)

Fraction
0

IV coarse to medium Sand

V Gravel

41

III fine Sand

II Silt

I Clay

10
Sieve residue [%]

20
30
40
50
60
70
80
90
100
10

0.2

0.06
0.1
Grain size [mm]

0.002

0.01

0.001

Figure 4.2: Grain size distribution of Mol sand.

Fraction
I+II
III
IV
V

Grain diameter
[mm]
<0.06
0.06-0.2
0.2-2.0
>2.0

Classification
Clay to Silt
fine Sand
medium to coarse Sand
Gravel and Stones

Mass
[%]
0.6
63.4
36.0
0.0

Table 4.4: Grain distribution of Mol Sand.

4.3

Cone penetration test

(CPT)

Three CPTs 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 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 CPTs is 1.15 m below
the surface.

4.4
4.4.1

Seismic cone penetration test

(SCPT)

Test description

SCPT1 to SCPT4 were performed using a 200 kN CPT truck fitted with additional 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-

42

CHAPTER 4. TEST SITE RETIE

Parameter
Median grain size, d50
Uniformity coefficient, dd60
10
d30 2
Degree of curvature, Cc = d10
d60
Specific gravity, Gs
Void ratio:
maximum emax
minimum emin
Dry unit weight:
minimum dmin
maximum dmax
Saturated unit weight: minimum rmin
maximum rmax

Unit
mm
kN/m3
kN/m3
kN/m3
kN/m3

Value
0.195
1.6
1.02
2.65
0.918
0.585
13.553
16.387
18.25
20.02

Table 4.5: Properties of Mol sand.

CPT 2

CPT 1
0

CPT 3

CPT 13

0
Mean q

Depth [m]

10
q [MN/m]
c

20

10
q [MN/m]
c

20

10
q [MN/m]
c

20

Range of
standard
deviation

10
q [MN/m]

20

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

Mean
Range of
standard deviation

Depth [m]

9
10

15

20

25

30

35

40

[]

Figure 4.4: Undrained angle of internal friction 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 interpretation; 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).

44

4.4.2

CHAPTER 4. TEST SITE RETIE

Test results for the wave velocities

The results are presented in graphical form for shear wave velocity Vs and compression 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 increasing 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 14

SCPT 12

0
No. 1
No. 2

No. 1
No. 2
No. 3
No. 4

6
Depth [m]

Depth [m]

10

10

12

12

14

200

400

Swave velocity [m/s]

600

14

500

1000

1500

2000

Pwave velocity [m/s]

Figure 4.5: S-wave and P-wave velocity, SCPT1-SCPT4, visual interpretation,

Retie.

SCPT1 to SCPT4 belong to the first SCPTs performed by Ghent University. 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.

4.4. SEISMIC CONE PENETRATION TEST (SCPT)

45

SCPT 5
0

Depth [m]

8
Mechanical Hammer
Sledgehammer

10

12

50

100

150

200

250

300

350

400

450

500

Swave 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 103 %, 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
0 , the plasticity index Ip and the shear strain 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 = 103 %.

46

CHAPTER 4. TEST SITE RETIE

Depth [m]

Depth [m]

Mechanical Hammer
Sledgehammer

10

10

12

10

20

30

Damping ratio [%]

40

12

Damping ratio [%]

Figure 4.7: Values of damping ratio found from SCPT5 (left) and profile obtained by Ishibashis 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)

Unit weight:
Angle of internal
friction :
K0 :
Plasticity index Ip :

above groundwater level

below groundwater level
0 to 4 m depth
deeper than 4 m
0 to 4 m depth
deeper than 4 m
all depths

47

Unit
kN/m3
kN/m3

Value
18.0
10.0

34.0
30.0
0.441
0.500
0.0

Table 4.6: Input parameters for Ishibashis equation.

The by the buoyancy corrected unit weight of the saturated sand below the
groundwater table 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 taken from the CPTs 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
4.5.1

Spectral analysis of surface waves

(SASW)

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

48

CHAPTER 4. TEST SITE RETIE

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 geophones 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 shorter 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 KUL
0

Depth [m]

Depth [m]

SASW UGent
0

100
200
Swave velocity [m/s]

300

100
200
Swave velocity [m/s]

300

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 calculation of K.U. Leuven and Ghent University. The density was chosen as
= 2000 kg/m3 . The coefficient of Poisson was assumed as = 0.48 from
K.U. Leuven and as = 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 influence 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 profiles from K.U. Leuven and Ghent University show a weaker layer with lower
propagation velocity. The profile from Ghent University has the soft layer between 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 CPTs 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 SCPTs. The averaged CPT
profiles for tip resistance qc and internal friction are repeated. Table 4.7 gives
the numerical values.

50

CHAPTER 4. TEST SITE RETIE

Figure 4.10: Summary of the test results at Retie site.

4.6. OVERVIEW OF THE TEST RESULTS RETIE

Averaged CPT-Profile
Depth
qc

[m] [MN/m2 ]
[ ]
0.20
0.60
26.70
0.40
1.53
29.96
0.60
2.90
31.57
0.80
3.47
31.55
1.00
5.70
33.37
1.20
8.07
34.83
1.40
9.33
34.97
1.60
10.60
35.31
1.80
10.63
34.90
2.00
10.77
34.64
2.20
10.00
33.95
2.40
10.50
33.92
2.60
10.07
33.37
2.80
9.50
32.65
3.00
11.00
33.53
3.20
10.50
33.03
3.40
9.60
32.27
3.60
10.10
32.33
3.80
7.15
29.13
4.00
4.80
27.23
4.20
5.20
27.53
4.40
3.75
24.80
4.60
3.00
22.32
4.80
3.20
20.23
5.00
9.60
30.58
5.20
8.50
29.67
5.40
7.60
28.62
5.60
9.65
30.09
5.80
9.75
29.98
6.00
8.20
28.58
6.20
8.00
28.19
6.40
7.60
27.48
6.60
8.35
27.87
6.80
8.20
27.58
7.00
9.90
29.00
7.20
11.35
29.68
7.40
15.75
31.42
7.60
15.20
31.03
7.80
15.50
31.02
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

Depth
[m]
1.34
1.86
2.38
2.89
3.41
3.93
4.44
4.96
5.48
5.99
6.51
7.03
7.54
8.06
8.58
9.09
9.61
10.12
10.64
11.16
11.67

Layer
1
2
3
4

Layer
1
2
3
4

51

Profile of
Vs[MH]
[m/s]
232.53
236.93
213.81
201.65
200.16
211.03
243.12
240.19
223.52
216.02
256.70
304.54
344.96
356.10
362.70
360.40
339.19
314.46
378.21
302.68
353.76

SCPT5
Vs[SH]
[m/s]
225.53
230.33
228.68
218.30
206.83
210.14
226.73
222.77
212.77
226.58
260.62
282.66
331.56
318.50
311.04
362.95
333.69
324.38
329.72
388.34
389.94

SASW UGent
Thickness
[m]
1.2
0.3
0.5

Vs
[m/s]
220
80
120
250

SASW KUL
Thickness
[m]
0.46
0.31
2.93

Vs
[m/s]
143
46
159
253

Vs[MEAN]
[m/s]
229.03
233.63
221.24
209.97
203.49
210.58
234.92
231.48
218.15
221.30
258.66
293.60
338.26
337.30
336.87
361.68
336.44
319.42
353.96
345.51
371.85

Table 4.7: Numerical values of the averaged CPT, SCPT5 and SASW results.

52

CHAPTER 4. TEST SITE RETIE

Chapter 5

Test site Lincent

5.1

Introduction

The testing site is located on a field between the Rue de la Bruy`ere 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 SCPTs 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).

5.2

Borings and undisturbed sampling

The drilling with the closest proximity to the locations of the SCPTs, 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
CPTs or SCPTs 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.

5.2. BORINGS AND UNDISTURBED SAMPLING

Date

Available In Situ Tests

01/02/1939

Boring at the Rue de Liege in

Lincent, from the Flemish Subsoil
Data Base
Boring at the corner Burgemeester
Dumont Straat to Lijsemstraat in
Waasmont, from the Flemish Subsoil Data Base
Boring No. 1108 provided by TucRail
3 CPT No. 1107, 1108, 1109
provided by TucRail

24/06/1991

18/08/1993
06/07/1993

09/04/2001

3 SCPT with a single geophone

cone

01/06/2001

3 SASW set-ups

15/05/2003

2 SCPT with a dual accelerometer

cone

55
Abbr.

Depth
[m]
34.0

74.0

B1108
CPT1107
CPT1108
CPT1109
SCPT1
SCPT2
SCPT3
SASW KUL1
SASW KUL2
SASW UGent
SCPT4
SCPT5

15.0
4.0
5.5
7.2
5.0
5.5
7.5

5.7
7.2

Table 5.1: Overview of the available in situ tests.

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 described in Table 5.3 and Table 5.4. The location of these borings is in the
center of Maasmont and in the Rue de Li`ege 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

56

CHAPTER 5. TEST SITE LINCENT

From
[m]
0.00
1.20
2.00
2.50

To
[m]
1.20
2.00
2.50
3.20

brown
gray, brown to olive-green
beige to gray beige
gray to dark gray

Main
Component
silt
fine sand
fine sand
fine sand

3.20

7.50
a)

Sequence of:
dark gray to green

arenite

b)

dark gray to green brown

clay

8.00
8.50
10.00
10.50
11.00
11.50
12.50
a)

dark gray, dark green

gray green to dark green
gray, dark green
brown gray to dark green
gray, dark green
brown gray to dark green
Sequence of:
dark brown to dark green

clay
clay
fine sand
clay
fine sand
clay

dark gray green to

brown green
dark gray
Sequence of:
gray, dark gray

sand

silt, fine sand,

glauconite
arenite

clay

silt, fine sand

7.50
8.00
8.50
10.00
10.50
11.00
11.50

b)
12.50
13.00

13.00
15.00
a)

Color

clay

Admixtures
gravel
clay, glauconite
clay, glauconite
silt, clay,
glauconite
fine sand,
glauconite
fine sand, silt,
glauconite
silt
silt, glauconite
glauconite
glauconite
glauconite
glauconite

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 member of Halen/Lincent (formation of Hannut), 2.5 to 13 m member of Waterschei (formation of Hannut), 13 to 15 m member of Gelinden/Orp (formation 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 .

5.3. CONE PENETRATION TEST (CPT)

57

From
To
Color
Main
Admixtures
[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 formation 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.
From
To
Color Main
Admixtures
[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 Li`ege.
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.

5.3

Cone penetration test

(CPT)

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 CPTs are adjusted in height to the level of CPT1108 and

58

CHAPTER 5. TEST SITE LINCENT

Depth
Grain size distribution
Frac- Grain
tion
diameter
I
<0.002 mm
(clay)
II
0.002-0.06 mm
(silt)
III
0.06-0.2 mm
(fine sand)
IV
0.2-2.0 mm
(medium/
coarse sand)
V
2.0-20.0 mm
(fine gravel)
VI
>20.0 mm
(coarse gravel/
stones)
Specific gravity, GS
Water content, w
Void ratio, e
Degree of Saturation, Sr
Unit weight,
Dry unit weight, d
Saturated unit weight, r
Liquid limit, wL
Plastic limit, wP
Plasticity index, IP
Classification

Unit
m

Sample 1
1.0-1.4

Sample 2
3.0-3.4

Sample 3
5.0-5.4

16

34

18

57

26

15

24

32

15

13

32

%
%
%
kN/m3
kN/m3
kN/m3
%
%
%

0
2.633
12.87

0
2.594
42.99
1.2864
86.7
15.9
11.1
16.7
71.1
39.8
31.3
OH
organic
clay

7
2.514
24.50

31.6
21.7
9.9
CL
silty
clay

59.0
38.3
20.7
GC
gravel-sandclay mixture

Table 5.5: Properties of samples taken in Lincent.

5.3. CONE PENETRATION TEST (CPT)

VI

V fine Gravel

59

II Silt

III fine Sand

IV coarse to medium Sand

I Clay

20
30
40
50
60
70
80

Depth 1 (1.01.4 m)

coarse Gravel to Stones

Sieve residue [%]

10
Depth 2 (3.03.4 m)

Depth 3 (5.05.4 m)

90
20

10

0.2
0.1
Grain size [mm]

0.01

0.06

0.002

Figure 5.2: Grain size distribution at the Lincent site.

CPT 1107

CPT 1109

CPT 1108

CPT 11071109
0
Mean q

Range of
standard deviation

Depth [m]

100

50
q [MN/m]
c

100

50
q [MN/m]
c

100

50
100
q [MN/m]
c

50
100
q [MN/m]
c

Figure 5.3: Cone resistance qc single and mean profiles, Lincent.

0.001

60

CHAPTER 5. TEST SITE LINCENT

CPT 1107, 1108, 1109
0
Mean
Range of
standard deviation

Depth [m]

10

15

20

25

30

35

40

45

[]

Figure 5.4: Undrained angle of internal friction calculated from qc , Lincent.

B1108 so they can be directly compared.

The penetration depth of the CPTs 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 .

5.4
5.4.1

Seismic cone penetration test

(SCPT)

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

5.4. SEISMIC CONE PENETRATION TEST (SCPT)

61

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.
SCPT 13

SCPT 45

0
No. 1
No. 2
No. 3

3
Depth [m]

Depth [m]

No. 4
No. 5

100
200
300
400
Swave velocity [m/s]

500

100
200
300
Swave velocity [m/s]

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
SCPTs. 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 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.
As already seen on the site of Retie, the profiles obtained by visual interpretation 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 SCPTs.
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

62

CHAPTER 5. TEST SITE LINCENT

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.

5.4.3

Test results for the damping ratio

SCPT 5
0
Mechanical Hammer
Sledgehammer
1

Depth [m]

5
6
Damping Ratio [%]

10

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
5.5.1

Spectral analysis of surface waves

(SASW)

Remarks on the testing setup and inversion assumptions

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 = 0.33 and the density = 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 measurement 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 Poissons ratio are chosen = 2000 kg/m3 and = 0.3, respectively.

5.5.2

Test results
SASW KUL1/KUL2

SASW UGent
0

3
Depth [m]

Depth [m]

Setup 1 (KUL1)
Setup 2 (KUL2)

100
200
300
Swave velocity [m/s]

400

100
200
300
Swave velocity [m/s]

400

Figure 5.7: SASW performed by Ghent University and K.U. Leuven, S-wave
velocity, Lincent.

64

CHAPTER 5. TEST SITE 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 swave 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.

Mean
Groundwater
10.40 m

1.00

(Silty Clay)

15.00

13.00

(Silty Clay)

Sequence of
Fine Sand / Clay

10.00

Fine Sand

8.50

Clay

7.50

(Sandy Clay
to Silty Clay)

Sequence of
Arenite / Clay

3.20

(Clayey Sand)

Fine Sand

1.20

Silt

0.00 m

Quaternary
Deposits
Formation of Hannut
(Tertiary Deposit)
Formation of Heers
(Tertiary Deposit)

7
0

20 40
phi []

310

160

140

Vs [m/s]

0 200 400
S-wave velocity [m/s]

Simplified S-Wave
Velocity Profile

Site Lincent

Remark: The s-wave velocity profile starts from a depth of

0.70 m with the surface of the boring B1108 as
reference-level.

0 20 40 60
qc [MN/m]

Internal
0 Friction

Tip
0 Resistance

Depth [m]

Depth [m]

Averaged CPT-Profiles

Depth [m]

Simplified Stratification
of Drilling B1108

7.20

1.70

the testing field)

0.70 (Surface altitude of

0.00 m

5.6. OVERVIEW OF THE TEST RESULTS LINCENT

65

Figure 5.8: Summary of the test results at the Lincent site.

66

CHAPTER 5. TEST SITE LINCENT

Averaged CPT-Profile
Depth
qc

[m] [MN/m2 ]
[ ]
0.20
12.20
40.00
0.40
9.75
40.00
0.60
9.37
37.80
0.80
6.50
34.10
1.00
9.03
34.67
1.20
10.90
34.97
1.40
8.33
34.40
1.60
4.17
29.47
1.80
1.37
20.80
2.00
2.53
23.53
2.20
13.53
30.93
2.40
14.37
32.37
2.60
1.30
17.13
2.80
2.33
20.87
3.00
3.83
24.10
3.20
40.77
36.07
3.40
47.07
37.50
3.60
10.10
28.20
3.80
6.27
23.70
4.00
33.30
33.53
4.20
48.00
36.35
4.40
43.00
35.85
4.60
19.60
28.45
4.80
8.50
28.10
5.00
42.80
35.00
5.20
58.00
37.55
5.40
15.10
31.05
5.60
11.00
28.95
5.80
8.15
26.15
6.00
23.95
29.20
6.20
44.15
32.50
6.40
52.50
35.75
6.60
60.00
36.80
6.80
44.00
35.30

Profiles of SCPTs
SCPT4
SCPT5
Depth
Vs
Vs
[m]
[m/s]
[m/s]
0.83
421.31
173.21
1.34
174.63
166.65
1.86
166.23
190.10
2.38
176.08
203.97
2.89
216.80
201.14
3.41
223.96
221.10
3.93
231.00
251.71
4.44
231.92
274.00
4.96
253.00
279.62
5.48
272.43
5.99
280.41
6.51
324.83
SASW UGent
Thickness
[m]
1
1.0
2
2.5
3

Vs
[m/s]
110
165
300

SASW KUL1
Thickness
[m]
0.19
2.53
0.22
3.33

Vs
[m/s]
108
158
176
248
249

SASW KUL2
Thickness
[m]
3.02
0.10
0.56

Vs
[m/s]
150
280
278
282

Layer

Layer
1
2
3
4
5

Layer
1
2
3
4

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, SCPTs 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).

6.2

Borings and undisturbed sampling

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 quaternary 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.

6.2. BORINGS AND UNDISTURBED SAMPLING

Date

Available In Situ Tests

01/08/1936

Boring close to the station in

Waremme, Flemish Subsoil Data
Base
Boring No. 3064 provided by
TucRail
Boring with discontinuous sampling
1 CPT No. 3063 provided by
TucRail
1 CPT No. 3064 provided by
TucRail
2 SASW set-ups K.U. Leuven

20/09/1993
27/02/2001
18/08/1993
19/08/1993
31/05/2001
01/06/2001
06/04/2001
13/08/2002
13/08/2002
18/09/2002

2 SASW set-ups Ghent

University
2 SCPT with a single geophone
cone
1 SCPT with a dual accelerometer cone
1 SCPT with a dual accelerometer cone
1 SCPT with a dual accelerometer cone

69

Abbr.

Depth
[m]
52.0

B3064

20.0

B1

12.5

CPT3063

12.0

CPT3064

12.2

SASW KUL1
SASW KUL2
SASW UGent1
SASW UGent2
SCPT1
SCPT2
SCPT3

12.5
12.0
11.4

SCPT4

11.4

SCPT5

11.4

Table 6.1: Overview of the available in situ tests.

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.
Properties of the sampled material
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;

70

CHAPTER 6. TEST SITE WAREMME

From
To
Color
Main Component Admixtures
[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

VI

V fine Gravel

IV coarse to medium Sand

III fine Sand

II Silt

I Clay

20
30
40
50
60
70
80

coarse Gravel to Stones

Sieve residue [%]

10

Depth 1 (3.03.4 m)

Depth 2 (5.05.4 m)

90
100

20

10

0.2
0.1
Grain size [mm]

0.06

0.01

0.002

0.001

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: =
36o and c = 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 measured. 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 groundwater table was found at a depth of 6.72 m. The latter value might be erroneous
because only a single level determination was done.

6.3. IN SITU TESTS

From
[m]
0.00

To
[m]
1.40

Color

1.40
3.40
4.50
7.40

3.40
4.50
7.40
8.00

8.00

9.40

9.40

11.40

11.40

12.50

brown ocher
brown, yellow ocher
yellow ocher
olive-green to
yellow ocher
dark green, grayly,
partly yellow ocher
dark green, grayly,
rust brown
dark gray, greenish

12.50

14.00

dark green

14.00

15.50

light gray

brown

71
Main
Component
silt

Admixtures

silt
silt
silt
silt

fine sand, angular

small stones
fine sand
fine sand
fine sand
fine sand

silt

fine sand

silt

clay, ferric oxide

silt

sand, small
angular flint
stones
fine sand

angular flint
stones (1-3 cm)
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
6.3.1

In situ tests
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 CPTs 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.

72

CHAPTER 6. TEST SITE WAREMME

From
To
Color Main Component Admixtures
[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 .

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 analyzed. 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 SCPTs 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 between 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

6.3. IN SITU TESTS

Depth
Grain size distribution
Fraction Grain diameter
I
<0.002 mm
(clay)
II
0.002-0.06 mm
(silt)
III
0.06-0.2 mm
(fine sand)
IV
0.2-2.0 mm
(medium/coarse sand)
V
2.0-20.0 mm
(fine gravel)
VI
>20.0 mm
(coarse gravel/stones)
Specific gravity, GS
Water content, w
Void ratio, e
Degree of Saturation, Sr
Unit weight,
Dry unit weight, d
Saturated unit weight, r
Liquid limit, wL
Plastic limit, wP
Plasticity index, IP
Classification

Effective angle of friction,

Effective cohesion, c

73

Unit
m

Sample 1
3.0-3.4

Sample 2
5.0-5.4

20.1

12.3

69.0

76.4

10.6

9.3

0.3

1.1

0.9

%
%
%
kN/m3
kN/m3
kN/m3
%
%
%

0
2.641
17.9
0.5764
83.6
19.74
16.74
20.34
34.6
19.9
14.7
CL
silty
clay
31.9
3

0
2.644
22.1
0.5956
100.0
20.24
16.58
20.23
29.5
23.3
6.2
CL/ML
silty clay to
clayey silt
39.2
5

kN/m2

Table 6.5: Properties of samples taken in Waremme at B3064.

CHAPTER 6. TEST SITE WAREMME

0

Depth [m]

Depth [m]

Depth [m]

74

10

10

10

12

12

12

14

10
qc [MN/m]

20

Figure 6.3: CPT3064,

cone resistance profile
qc .

14

10
qc [MN/m]

20

Figure 6.4: CPT3063,

cone resistance profile
qc .

14

10
qc [MN/m]

20

Figure 6.5: CPT3064

and 3063, mean 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 SCPTs.
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.

6.3. IN SITU TESTS

75

Depth [m]

10

12

14

10

20

[]

30

40

50

Figure 6.6: Undrained angle of internal friction calculated from qc , Waremme.

SCPT 45
0

Depth [m]

Depth [m]

SCPT 12
0

10

10

12

12
No. 4
No. 5

No. 1
No. 2
14

200
400
Swave velocity [m/s]

600

Figure 6.7: S-wave velocity from

SCPT1 - SCPT2 using visual interpretation, Waremme.

14

200
400
Swave velocity [m/s]

600

Figure 6.8: S-wave velocity from

SCPT4 - SCPT5 using cross correlation, Waremme.

76

CHAPTER 6. TEST SITE WAREMME

SCPT 5
0

Depth [m]

10

12
Mechanical Hammer
Sledgehammer
14

2.5
5
7.5
Damping Ratio [%]

10

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.

6.3.3

Spectral analysis of surface waves (SASW)

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 = 0.33, the density = 1800 kg/m3 and the
damping ratio for p- and s-wave s = p = 0.5 % are kept constant in the inversion 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 simulation of a halfspace only. The density and Poissons ratio are chosen with
= 2000 kg/m3 and = 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
6.4.1

Laboratory tests to obtain Gmax and damping ratio

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 determined. 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 consolidation 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).

78

CHAPTER 6. TEST SITE WAREMME

Setup 2
0

3
Depth [m]

Depth [m]

Setup 1
0

UGent
KUL

UGent
KUL
8

100
200
Swave velocity [m/s]

300

horiz.

vert.

Figure 6.10: SASW results at setup

1, parallel to the house, Waremme.

Sample /
Depth
B1/2
2.0-2.5 m
B1/10
10.0-10.5 m
B1/3
3.0-3.5 m
B1/11
11.0-11.5 m

Vs [m/s]
Gmax [MPa]
Vs [m/s]
Gmax [MPa]
Vs [m/s]
Gmax [MPa]
Vs [m/s]
Gmax [MPa]

25
120.1
26.7
110.4
27.3
-

100
200
Swave velocity [m/s]

300

Figure 6.11: SASW results at setup 2, perpendicular to the rail way

track, Waremme.

50
143.2
40.9
168.2
57.1
138.9
43.8
173.1
55.5

1,2,3
100
182.5
66.9
198.2
79.1
175.1
70.6
211.7
82.7

[kPa]
200
221.6
99.4
239.8
116.6
217.7
111.0
245.5
111.1

400
279.4
159.6
294.3
177.0
293.1
159.1

600
315.6
204.8
-

Table 6.6: Results of isotropic bender element tests with vertical and horizontal
propagation direction.

6.4. LABORATORY TESTS TO OBTAIN GMAX AND D

400

79

250

350
200
300

Gmax [MPa]

Vs [m/s]

250
200
150

50

100

200

300
400
c [kPa]

500

B 1/2, vertical
B 1/10, vertical
B 1/3, horizontal
B 1/11, horizontal

50

600

0
0

700

Figure 6.12: S-wave velocity by isotropic bender element tests with

vertical and horizontal propagation
direction.

6.4.2

100

B 1/2, vertical
B 1/10, vertical
B 1/3, horizontal
B 1/11, horizontal

100

0
0

150

100

200

300
400
c [kPa]

500

600

700

Figure 6.13: G0 by isotropic bender element tests with vertical and

horizontal propagation direction.

Free torsion pendulum test

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 105 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 = 105 in the direction of
the lower shear strain. The damping ratio shows a peak at = 3 106 . 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 105 for the G-modulus and above 104 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 105 are given in Figure 6.15.

80

CHAPTER 6. TEST SITE WAREMME

G-Modulus

180

50 kPa
100 kPa
200 kPa
400 kPa

160

G [MPa]

140
120
100
80
60
40
20
0 -6
10

10

-5

Shear Strain, [-]

10

-4

10

-3

Damping Ratio

0.15

50 kPa
100 kPa
200 kPa
400 kPa

D [-]

0.1

0.05

0 -6
10

10

-5

Shear Strain, [-]

10

-4

10

-3

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.

6.4.3

Resonant column test

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 105 .
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

GModulus

81

Dampig Ratio

160

0.05
0.045

140

0.04
120
0.035
0.03
D []

G [MPa]

100
80

0.025
0.02

60

0.015
40
0.01
20
0

0.005
0

100

200
300
[kPa]

400

100

200
300
[kPa]

400

Resonant Column Test, 4.04.5 m

Resonant Column Test, 5.05.5 m
Free Torsion Pendulumn Test, 12.012.5 m

Figure 6.15: G-Modulus and damping ratio versus isotropic confining stress,
shear strain 105 , 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 of the averaged CPT-profile;
Table 6.8 gives the s-wave velocities determined by SCPT4, SCPT5 and the
four SASW tests.

Groundwater
6.72 m

Chalk

Site Waremme

12.75

12.50

Silt/Stone Mixture

11.50

(Clayey Silt
to Silty Clay)

Silt

1.00

Gravel

0.00 m

Simplified Stratification
of Drilling B1

12.50

7.50

Quaternary
Deposits
Creataceous Formation of Heers
(Tertiary Deposit)
formation

Depth [m]

14

12

10

10
15
q c [MN/m]

20

14

12

[]

20

40

14

12

10

10

200
400
S-wave velocity [m/s]

Simplified S-Wave
Velocity Profile

Internal
Friction

Depth [m]

Depth [m]

Averaged CPT-Profiles
Tip
Resistance

265

115

155

Vs [m/s]
0.00 m

11.00

1.00

82
CHAPTER 6. TEST SITE WAREMME

Figure 6.16: Summary of the test results at Waremme.

6.5. OVERVIEW OF THE TEST RESULTS WAREMME

Depth
[m]
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
2.20
2.40
2.60
2.80
3.00
3.20
3.40
3.60
3.80
4.00
4.20
4.40
4.60
4.80
5.00
5.20
5.40
5.60
5.80
6.00
6.20

Averaged CPT-Profile
qc

Depth
qc
[MN/m2 ]
[ ]
[m] [MN/m2 ]
4.69 41.10
6.40
1.68
5.47 38.70
6.60
1.49
5.08 36.50
6.80
1.49
2.77 32.20
7.00
1.30
3.54 32.30
7.20
1.49
2.77 30.00
7.40
1.50
3.17 29.90
7.60
1.88
2.00 31.65
7.80
1.89
2.59 30.85
8.00
2.08
2.97 30.05
8.20
4.20
2.40 27.05
8.40
4.97
1.83 24.20
8.60
5.36
1.25 20.05
8.80
6.13
0.87 14.70
9.00
5.36
1.25 19.25
9.20
6.33
1.25 18.10
9.40
5.95
1.65 20.30
9.60
4.41
1.26 16.75
9.80
4.22
1.46 17.15
10.00
5.18
1.27 16.15
10.20
4.41
1.27 15.55
10.40
6.54
1.47 16.35
10.60
5.96
1.47 15.95
10.80
7.31
1.28 13.55
11.00
7.31
1.47 14.20
11.20
6.93
1.66 15.95
11.40
7.32
1.67 16.05
11.60
5.97
2.05 17.60
11.80
4.24
1.87 15.95
12.00
28.68
1.87 14.30
12.20
30.99
1.87 16.05

83

[ ]
16.50
14.35
11.25
12.45
12.10
11.90
13.60
13.95
14.55
20.95
22.30
22.40
23.85
22.50
23.45
22.70
20.35
19.85
21.35
19.90
22.60
21.05
20.80
20.65
21.80
22.30
21.45
18.20
25.65
26.75

Table 6.7: Numerical values of the averaged CPT results.

84

CHAPTER 6. TEST SITE WAREMME

Profiles of SCPTs
SCPT4 SCPT5
Depth
Vs
Vs
[m]
[m/s]
[m/s]
0.7
52.55
138.88
1.20
65.75
193.76
1.70
92.25
263.57
2.20
116.95
200.76
2.70
117.45
154.75
3.20
135.10
190.88
3.70
145.55
170.25
4.20
141.55
171.41
4.70
145.15
176.11
5.20
158.95
183.72
5.70
178.45
189.64
6.20
186.55
198.07
6.70
191.75
199.98
7.20
193.60
199.84
7.70
204.80
201.18
8.20
219.25
212.40
8.70
227.20
219.79
9.20
251.90
244.33
9.70
273.95
259.74
10.20
264.20
267.79
10.70
264.90
252.49

SASW UGent Setup 1

Layer Thickness
Vs
[m] [m/s]
1
>7
155
SASW UGent Setup 2
Layer Thickness
Vs
[m] [m/s]
1
>7
165
SASW KUL Setup 1
Layer Thickness
Vs
[m] [m/s]
1
4.46
154
2
1.08
215
3

216
SASW KUL Setup 2
Layer Thickness
Vs
[m] [m/s]
1
0.27
145
2
0.70
179
3

179

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 SCPTs
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 Menge (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.

7.2

Borings and undisturbed sampling

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.

14/05/2003

Boring with discontinuous sampling

Boring with continuous sampling
SCPT
SCPT
Boring with undisturbed sampling
30 CPT using an electrical cone
4 Marchetti dilatometer tests
5 SASW profiles
3 measurement lines tested by
seismic refraction test

B1I

Depth
[m]
11.0

I
I
I
II

B2I
SCPT1
SCPT2
B1II

11.0
12.9
12.9
14.0

II
II
II
II

CPT-E
DMT
SASW1-5
SRT

14.0
13.0

14/05/2003
13/05/2003
14/05/2003

Table 7.1: Overview of the used in situ tests.

3.60

Building of the
De Nayer Institute

Entrance

14.00

12.10

7.20

3.10

Parking Space

5.00

SCPT 2

5.20

4.00

Access Road

Tree
Tree
B2

SCPT 1

Tree

Tree

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.

7.3. IN SITU TESTS

87

Properties of the sampled material

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
= 28o and c = 23 kPa.
Parameter

Unit

Dry unit weight, d

Natural unit weight,
Water content, w:
Saturation degree, Sr :
Liquid limit, wL :
Plastic limit, wP :
Plasticity index, IP :

kN/m3
kN/m3
%
%
%
%
%

a
15.7
19.7
25.3
(101.8)
72.2
25.4
46.8

Sample
b
c
15.3
15.5
19.3
19.5
26.0
25.8
99.2 (100.9)
65.1
75.9
25.3
26.2
39.9
49.7

d
15.4
19.4
26.1
(101.0)
71.8
26.0
45.8

Mean
Value
15.5
19.5
25.8
100.0
71.6
25.7
45.6

Table 7.2: Properties of Boom clay.

VI

V fine Gravel

IV coarse to medium Sand

II Silt

III fine Sand

I Clay

20
30
40
50
60
70
80

coarse Gravel to Stones

Sieve residue [%]

10

Range of grain size distribution

90
100

20

10

0.2
0.1
Grain size [mm]

0.06

0.01

0.002

0.001

Figure 7.2: Grain size distribution at Sint-Katelijne-Waver.

7.3
7.3.1

In situ tests
Marchetti dilatometer test (DMT)

Altogether four DMTs were performed on site II. A profile of the overconsolidation ratio OCR and another for the K0 -value were correlated using the

CHAPTER 7. TEST SITE SINT-KATELIJNE-WAVER

0

Depth [m]

Depth [m]

88

10

10

12

12

14

20
40
60
Overconsolidation ratio

14

80

Figure 7.3: Overconsolidation ratio derived from dilatometer test results at site II.

Figure 7.4: K0 -value derived from

dilatometer test results at site II.

horizontal stress index KD , obtained directly from the test. Both are shown
in Figure 7.3 and Figure 7.4. Menge (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)

CPTs 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 CPTs 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 SCPTs on site I are performed using the 200 kN automotive remotecontrolled track vehicle. The cone was driven to a final depth of about 13 m in

7.3. IN SITU TESTS

89

Depth [m]

10

12

14

2
3
q [MN/m]

Figure 7.5: Averaged cone resistance qc and range of standard deviation, site II.

SCPT 12

SCPT 2

0
Mechanical Hammer
Sledgehammer

Depth [m]

Depth [m]

No. 1
No. 2

100
200
Swave velocity [m/s]

300

Figure 7.6: S-wave velocity using

cross correlation based on SCPT1
and SCPT2.

4
6
8
Damping Ratio [%]

10

Figure 7.7: Values of damping ratio

obtained from SCPT2.

90

CHAPTER 7. TEST SITE SINT-KATELIJNE-WAVER

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.

7.3.4

Spectral analysis of surface waves (SASW)

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 SCPTs 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

7.4. LABORATORY TESTS TO OBTAIN GMAX AND D

0

0
No. 1
No. 2
No. 3
No. 4
No. 5

200
400
600
Swave velocity [m/s]

1
2
3
4
5

No. 1
d
Vs
[m] [m/s]
0.8
90
1.0
180
4.1
380

450

800

Figure 7.8: SASW tests performed

at site II.
Layer

Profile A
Profile B
Profile C

Depth [m]

Depth [m]

91

No. 2
d
Vs
[m] [m/s]
0.3
170
0.5
100
1.1
160
4.3
340

500

200
400
600
Swave velocity [m/s]

800

Figure 7.9: Seismic refraction test

results at site II.
No. 3
d
Vs
[m] [m/s]
0.7
110
1.1
170
4.4
320

470

No. 4
d
Vs
[m] [m/s]
0.6
90
1.2
170
4.2
350

470

No. 5
d
Vs
[m] [m/s]
0.3
200
0.5
110
1.1
160
4.4
300

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 Menge (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 SCPTs.

7.4
7.4.1

Laboratory tests to obtain Gmax and damping ratio

Bender elements with time arrival interpretation

The undisturbed samples from site II were tested in a triaxial cell under isotropic confining stress. The results of the analysis of the arrival of the shear

92

CHAPTER 7. TEST SITE SINT-KATELIJNE-WAVER

Profile
A
B
C

Vp
[m/s]
1667
1457
1669

Vs
[m/s]
320
348
385

[-]
0.48
0.47
0.47

Edyn
MPa
576
675
840

G0
MPa
195
231
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
VS [m/s]

G0 [MPa]

100
80

150

60
100
40
50
20
0
0

100

200

300
c [kPa]

400

500

0
0

600

Figure 7.10: Results of bender element tests, shear modulus G0 versus isotropic consolidation stress c ,
site II.

100

200

300
c [kPa]

400

500

600

Figure 7.11: Results of bender element tests, s-wave velocity VS versus isotropic consolidation stress c ,
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.

7.4.2

Free torsion pendulum test

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 105 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

7.4. LABORATORY TESTS TO OBTAIN GMAX AND D

93

GModulus
90
80
70

G [MPa]

60
50
40
30
20
10

50 kPa
100 kPa
200 kPa
400 kPa

0
6
10

10

Strain []

10

10

Damping Ratio
0.08
0.07
0.06

D []

0.05
0.04
0.03
0.02
0.01
0
6
10

50 kPa
100 kPa
200 kPa
400 kPa
5

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 105 are given in Figure 7.13. The found increase
of the damping ratio from a shear strain of 2 106 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.

7.4.3

Resonant column test

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 105 . 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.

94

CHAPTER 7. TEST SITE SINT-KATELIJNE-WAVER

GModulus

Damping Ratio

80

0.06

70
0.05
60
0.04

D []

G [MPa]

50

40

0.03

30
0.02
20
0.01

Resonant Column Test

Free Torsion Pendulum Test

Resonant Column Test

Free Torsion Pendulum Test

10

100

200
300
[kPa]
c

400

100

200
300
[kPa]

400

Figure 7.13: G-Modulus and damping ratio versus isotropic confining stress,
shear strain 105 , sampling depth: 4.0 to 4.5 m, site I.

7.5

Overview of the test results Sint-KatelijneWaver

Figure 7.14 summarizes the geotechnical test results for the site in SintKatelijne-Waver. The simplified profile of the shear wave velocity at site I
is based on both SCPTs. 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 SCPTs.

Groundwater,
max. 2.80 m
min. 3.30 m

0.00 m

8.50

3.00

0.80

Reference level:

180

150

210

Vs [m/s]
100
210

0 100 200 300

S-wave velocity [m/s]

Site Sint-Katelijne-Waver

11.00

Clay

3.00

Clayey Sand

Quaternary
Deposits
Formation of Boom

0.00 m

Depth [m]

Simplified S-Wave
Velocity Profile at Site I

surface site I

16.30 m

Groundwater
as at site I

3.00 m

2.30 m

Clay

Clayey
Sand

surface site II

14.00 m

0.70 m

0.00 m

Simplified Stratification
at Site II

Quaternary
Deposits
Formation of Boom

Simplified Stratification
at Site I

14

12

10

2
3
q [MN/m]
c

Averaged CPT-Profile
at Site II

7.5. OVERVIEW OF THE TEST RESULTS ST.-KATELIJNE-WAVER 95

Figure 7.14: Summary of the test results at Sint-Katelijne-Waver.

Depth [m]

96

CHAPTER 7. TEST SITE SINT-KATELIJNE-WAVER

Profiles of SCPTs
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.
Date

Available In Situ Tests

Abbr.

15/07/1968

Boring inside the ground-plan of the

main building of the Flemish Geotechnical Institute
Boring at the closest corner of the storage and garage building to SCPT1
CPT at the place of boring B21
SCPT with a dual cone equipped with
Br
uel & Kjr accelerometers

B6

Depth
[m]
14.0

B21

5.0

19/07/1968
19/07/1968
25/11/2003

CPT21
SCPT1

14.75
17.5

Table 8.1: Overview of the available in situ tests.

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
B 21

SCPT 1

Storage and
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.

8.3. CONE PENETRATION TEST (CPT)

From
To
Color
[m]
[m]
0.0
0.75 dark brown
0.75
8.25 gray, brown
8.25 10.80 gray
10.80 13.50 gray
Lithographic Stratification: 0

99

Main
Admixtures
Component
fine sand
silt
fine sand
fine sand
silt
silt
fine sand
to 13.5 m quaternary deposits.

Table 8.2: Combined results of boring B21 and B6 for the location of B21.

8.3

Cone penetration test (CPT)

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

Depth [m]

10

12

14

16

6
8 10
q [MN/m]

12

14

16

Figure 8.2: Cone resistance qc , CPT21.

8.4
8.4.1

Seismic cone penetration test (SCPT)

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.

100

CHAPTER 8. TEST SITE GHENT

The dual cone with two triaxial accelerometers from Br

uel & Kjr (Section
9.2.2), mounted in 1 m distance from each other, is used together with a decoupled 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 mechanical 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 excitation directions are very consistently. They are given in Figure 8.3. A simplified
profile can be found in the summary in Figure 8.5.

8.4.3

Test results for the damping ratio

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

Depth [m]

Depth [m]

10

Left side
Right side

12

12
Left side
Right side

14

14

16

16

18

10

100
200
300
Swave velocity [m/s]

400

Figure 8.3: S-wave velocity from

SCPT using cross correlation, excitation from both sides of the truck.

18

10
20
Damping Ratio [%]

30

Figure 8.4: Values of damping ratio obtained from SCPT, excitation

from both sides of the truck.

8.5. OVERVIEW OF THE TEST RESULTS GHENT

8.5

101

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.

(Silty Sand)

Fine Sand

13.50

(Sandy Silt)

Silt

10.80

8.25

(Sand)

Fine Sand

Depth [m]

Quaternary
Deposits

16

14

12

10

Site Ghent, Technologiepark 905

Groundwater
1.85 m

0.75 (Silty Sand)

Fine Sand

0.00 m

8
10
q [MN/m]
c

12

14

Averaged CPT-Profile
at CPT21

16

18

16

14

12

10

100
200
300
400
S-wave velocity [m/s]

250

280

160

180

Vs [m/s]

Simplified S-Wave Velocity Profile

from SCPT1

Depth [m]

Simplified Stratification
at Drilling B21

17.00

7.80

3.00

0.00 m

102
CHAPTER 8. TEST SITE GHENT

Figure 8.5: Summary of the test results at the site in Ghent.

8.5. OVERVIEW OF THE TEST RESULTS GHENT

103

Profiles of SCPTs
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.

104

CHAPTER 8. TEST SITE GHENT

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 penetration 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 determination 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 extended 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-KatelijneWaver 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 equipment described in detail in the following paragraphs is shown in Figure 9.2.
Lateral Offset
Mechanical
Hammer

Static
Load

Sledge
hammer

tT
LT

tB

Depth

Beam

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

X1, Y1,
Z1

3 channels

X1, Y1,
Z1

Geophones

X2, Y2,
Z2

Hamlin accelerometers or
geophones

Brl & Kjaer accelerometers

Termination box
X2, Y2,
Z2

X1, Y1,
Z1

Dual seismic cone

PC with data
acquisition card

Trigger
channel
Hammer and beam
(shearwav e source)

Single
seismic cone

Figure 9.2: Functional overview of the seismic cone equipment.

Applied load
provided by the CPT truck

a)
Source beam

b)

Impact anvil

Applied load
provided by the CPT truck

Upper bearing beams

Roller system
Impact anvil

Source beam

c)

Applied load
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).

110

CHAPTER 9. SCPT BASED TECHNIQUES

The SCPTs 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 hammer 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 repeated blows with the cone at a certain depth show that the assumption that the
mechanical swing hammer provides a repeatable source is questionable. Differences 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

9.2. APPLIED EQUIPMENT AND SELECTION CRITERIA

111

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

x 10

8
No. 295

Particle acceleration [m/s]

Particle acceleration [m/s]

4
2
0
2
4
6
8
0.48

0.5

0.52
Time [s]

0.54

4
2
0
2
4
6
8
0.48

0.56

8
No. 321

Particle acceleration [m/s]

Particle acceleration [m/s]

0.5

0.52
Time [s]

0.54

0.56

x 10

4
2
0
2
4
6
8
0.48

No. 309

x 10

0.5

0.52
Time [s]

0.54

0.56

x 10

No. 325

6
4
2
0
2
4
6
8
0.48

0.5

0.52
Time [s]

0.54

0.56

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

112

CHAPTER 9. SCPT BASED TECHNIQUES

100
90
No. 321

Amplitude Spectrum [m/s]

80
70
60

No. 325

50
40
No. 309

30
20
No. 295

10
0
0

200

400
600
Frequency [Hz]

800

1000

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 accelerometers because of insufficient frequency response characteristics of the geophones
as described later on. These accelerometers where later replaced by accelerometers from Br
uel & Kjr 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 specifications 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)

 35.682
(A=10 cm)

43.702
(A=15 cm)

47.873
(A=18 cm)

Triaxial Sensor
(top-part)
35.682
(A=10 cm)

Triaxial Sensor
(bottom-part)

113

60

Cone-tip

43.702
(A=15 cm)

9.2. APPLIED EQUIPMENT AND SELECTION CRITERIA

189

1033
392

260

720

370

1482

400

Conetip
[mm]

51

230

133

Dual Cone
(1 m configuration)
35.682
(A=10 cm)

Triaxial
Sensor

Friction
transducer

60

Load cell/Pore
pressure transducer

446
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 homogeneous 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 calibrated 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 measured. 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.

114

CHAPTER 9. SCPT BASED TECHNIQUES

Roller bearing

Triaxial geophone
(inside)

Single Cone
Accelerometer

Accelerometer
Single cone

Shaker

Roller bearing
(rotation centre)

Leaf spring

Triaxial geophone (inside)

Shaker

Figure 9.7: Calibration of

the single cone with installed geophones, vertical.

Figure 9.8: Calibration of the single cone

with installed geophones, horizontal.

It is not possible to compare the calibration curve given by the sensor

manufacturer with the results of our calibration directly because the cone manufacturer 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 investigated 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 framework 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.

9.2. APPLIED EQUIPMENT AND SELECTION CRITERIA

115

6
Vertical Geophone
Horizontal Geophones

Sensitivity [V / (mm/s)]

50

100

150

200

250

300

350

400

450

500

Frequency [Hz]

Figure 9.9: Calibration curves of the seismic single cone equipped with geophones GeoSpace GSC-14-L3.

OUTPUT [V/(IN/SEC)]

10

10

GeoSpace
GSC14L3

10

10

10

10

FREQUENCY [Hz]

Figure 9.10: Calibration curve of the geophones GeoSpace GSC-14-L3 as provided by the manufacturer.

116

CHAPTER 9. SCPT BASED TECHNIQUES

Dual cone

Reference
accelerometer

Roller bearing
(rotation centre)

Reference
accelerometer
Leaf spring

Triaxial
accelerometer (inside)

Triaxial
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 manufacturer
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 = 9.81 103 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 Br
uel & Kjr accelerometers
The installed Hamlin accelerometers showed an unexpected high noise component 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 Br
uel & Kjr 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.

9.2. APPLIED EQUIPMENT AND SELECTION CRITERIA

117

1.4
chosen characteristic sensitivity: 1.116 V/g
(exact at 23 and 50 Hz )

1.2

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 SCA600C13H1G/C13V1G.

The main parameters of the Br

uel & Kjr 4506 B 002 accelerometers are:
Measuring range: 7 g
Sensitivity: 1 V/g (exact values specified by the manufacturer separately 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 = 9.81 103 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 Br
uel & Kjr.

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

118

CHAPTER 9. SCPT BASED TECHNIQUES

Figure 9.13: Typical frequency response of the Br

uel & Kjr 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 Br
uel & Kjr 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.
SCPTs 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 cutoff 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 program 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 measurement
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 automatically
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 modulus

The shear modulus is obtained by determining the shear wave velocity Vs .

Because the generated waves reach only very small strain levels below the

120

CHAPTER 9. SCPT BASED TECHNIQUES

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.
Emax = Vp2

(1 + )(1 2)
1

(9.1)

Poissons ratio can be directly determined out of the ratio between Vp

and Vs in Equation (9.2).
1
=
2

Vp
Vs
Vp
Vs

2
2

(9.2)

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 approximately 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 refracted on layer interfaces. Based on Snells law as given in Equation (3.10)
the bended travel path from source to receiver can be predicted. However,

9.3. METHODS FOR THE SHEAR MODULUS

121

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 por 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 chosen 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 performed 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 = , between the signals of the bottom and the top accelerometer in the cone, sB and sT , is obtained at the maximum of the cross

122

CHAPTER 9. SCPT BASED TECHNIQUES

Top element, ydirection, Depth: 12.19 m

Particle velocity [m/s]

0.4
Hammer blow left
Hammer blow right
0.2

0.2

0.4
0.5

0.52

0.54

0.56

0.58
Time [s]

0.6

0.62

0.64

0.66

Bottom element, ydirection, Depth: 13.22 m

Particle velocity [m/s]

0.4
Hammer blow left
Hammer blow right
0.2

0.2

0.4
0.5

0.52

0.54

0.56

0.58
Time [s]

0.6

0.62

0.64

0.66

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 :

rBT ( ) =

sB (t) sT (t + ) 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.

9.3. METHODS FOR THE SHEAR MODULUS

123

The interpolation is based on the sampling theorem of the digital signal processing. 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 unambiguously. With other words, it is possible to interpolate. Kammeyer & Kroschel
(1998) provide an equation for this interpolation for an arbitrary time t:

s(t) =
k=

s(k/fS )

sin[(fS t k)]
(fS t k)

(9.4)

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 interpolated output vector
The time records obtained from SCPTs 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 Transformation) even though the FFT is only a certain but most common numerical
implementation of the discrete Fourier transformation.

124

CHAPTER 9. SCPT BASED TECHNIQUES

Top element, ydirection, depth: 12.19 m

Particle acceleration [m/s]

0.4
Acquired data points
Interpolated time history
0.2

0.2

0.4
0.54

0.55

0.56

0.57
Time [s]

0.58

0.59

0.6

Figure 9.15: Example of the signal interpolation on the measurement signal.

Crosscorrelation [m/s]

x 10

t = 4.23 ms

4
2
0
2
4
0.1

0.05

0
Time shift [s]

0.05

0.1

0.15

Figure 9.16: The cross correlation of the signals from Figure 9.14; picked maximum 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 between the signals is calculated from RTB :

tan() =

Im(RT B )
Re(RT B )

(9.8)

The phase shift is wrapped, that means the values are limited by and .
Before the time shift t(f) between the two signal is calculated using Equation
(9.9), it has to be unwrapped.
t(f ) =

(f )
2 f

(9.9)

Finally the wave velocity V(t) is obtained in consideration of the travel path
difference x.
V (t) =

x
t(f )

(9.10)

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 consideration 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 , velocity c and source amplitude A0 . t marks the
time and x the distance of the current observation point from the source, then:
A++ (x, t) = A0 sin

2
(x ct)

(9.11)

126

CHAPTER 9. SCPT BASED TECHNIQUES

Amplitude [m/s]

2.5

Amplitude spectra, ydirection

x 10

Top element, depth: 12.19 m

Bottom element, depth: 13.22 m

2
1.5
1
0.5
0

50

Phase angle [rad]

10

300

Unwrapped phase
Wrapped phase
5
0

Swave velocity [m/s]

100
150
200
250
Wrapped / unwrapped phase of the crosspower spectrum

50

100

500

150
Swave velocity

200

250

300

250

300

400
300
200
Result of time domain crosscorrelation, V = 243.60 m/s

100
0

50

100

150
Frequency [Hz]

200

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.

9.4. METHODS TO OBTAIN THE DAMPING RATIO

127

By introducing the wave number k = 2/ and the angular frequency

= 2/T with T = /c where T is the period and neglecting any phase shift
the following equation is obtained:
A++ (x, t) = A0 sin(kx t)

(9.12)

or, in terms of the complex exponential:

A++ (x, t) = A0 ei(kxt)

(9.13)

For spherical waves in a homogeneous medium, neglecting near field terms, the
amplitude decays inversely with the distance x:
A+ (x, t) =

A0 i(kxt)
e
x

(9.14)

This attenuation is also called geometrical damping. Factors for other types
of waves can be found in Table 9.1.
Wave type
Body waves at depth
Body waves at the surface
Rayleigh waves

Amplitude decay with distance x

Point source
Line source
1
x
x1/2
2
x
x1
1/2
x
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 + i. Equation
(9.14) becomes:
A+ (x, t) =

A0 x ik x it
e
e
e
x

(9.15)

with 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:
A(x) =

A0 x
e
x

(9.16)

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 (x2 x1 )
=
e
A1
x2

(9.17)

128

CHAPTER 9. SCPT BASED TECHNIQUES

and after taking the natural logarithm:

A2
A1

ln

= ln

x1
x2

(x2 x1 )

(9.18)

Equation (9.18) can be transformed to:

ln
=

A1 x 1
A2 x 2

x2 x 1

(9.19)

which leads with Equation (2.45):

=

2 f D
c

(9.20)

to the formulation of the damping ratio

A1 x 1
c ln A2 x2
D=
2f x2 x1

(9.21)

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 =
2f xB xT

(9.22)

xT and xB represent the travel path length from the source to the top, respectively 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 Equation (9.19), the calculated attenuation coefficient 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 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 independent from the selection of a factor for the geometrical attenuation that does
not depend on frequency.

9.4. METHODS TO OBTAIN THE DAMPING RATIO

129

0.5
0.4

Attenuation Coefficient, [m ]

0.3
0.2
0.1
0
0.1
0.2
0.3
0.4
0.5
0

50

100

150
200
Frequency [Hz]

250

300

Figure 9.18: Attenuation coefficient for the example signals from Figure 9.14.

20
15

Mean value 90 130 Hz:

D = 2.67 %

Damping Ratio, D [%]

10
5
0
5
10
15
20
0

50

100

150
200
Frequency [Hz]

250

300

Figure 9.19: Damping ratio based on attenuation coefficient method, mean

value for the range of highest signal energy.

130

CHAPTER 9. SCPT BASED TECHNIQUES

It can be assumed that the attenuation coefficient depends linearly on

the frequency f for the frequency range of interest:
=zf

(9.23)

After introduction of Equation (9.23), Equation (9.18) is differentiated with

respect to f:
A2
A1

d ln

df

= z(x2 x1 )

(9.24)

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:
2 ln

A2
A1

f x

= z

(9.25)

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 Udas
(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 Q1 :
D=

1
Q1
zc
=
=
2Q
2
2

(9.26)

D is generally assumed to be independent of the amplitude of the particle

motion at small strains and of the frequency. The latter means 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
Q1
Vs
Vs d ln AB
Ds =
=
=
z=
2Q
2
2
2 df x

(9.27)

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 %.

9.4. METHODS TO OBTAIN THE DAMPING RATIO

131

0.5

Logarithmic Ratio, ln (AT / AB) []

0.4

Mean slope 90 130 Hz

related to D = 7.68 %

0.3
0.2
0.1
0
0.1
0.2
0.3
0.4
0.5
0

50

100

150
200
Frequency [Hz]

250

300

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-straincurves

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 Martnez-Carvajal et al. (2001).
Principles of the method
Elgamal et al. (1995) developed a procedure based on the following assumptions. The shear stress (z, t) at any level z may be expressed using the stress-free

132

CHAPTER 9. SCPT BASED TECHNIQUES

a1

Free ground
surface

Accelerometers

1
2

z1

a3

z2

a4

z3

z3

z4

a2

a5
a6

ai1
ai
ai+1

i1
i

i+1

zi1
zi

Figure 9.21: Multi-accelerometer array (Zeghal et al. (1995)).

surface boundary condition and a shear beam model as:

z

(z, t) =

u
(, t) d

(9.28)

=0

in which is the mass density and u

(, t) is the lateral acceleration at level .
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 (zi1 + zi )/2 reduce then to:
u
i1 + u
i
zi1
2
3
ui1 + u
i
i1/2 (t) = i1 (t) +
zi1
8
i = 2, 3, . . .
i (t)

= i1 (t) +

(9.29)
(9.30)

in which subscripts i and (i-1/2) refer to levels zi (of the ith accelerometer)
and (zi1 + zi )/2 (halfway between accelerometers i and (i-1)) respectively,
i = (zi , t), u
i = u
(zi , t) and zi is the distance between the accelerometers.
The corresponding shear strains at levels zi and (zi1 + zi )/2 may be expressed as:

9.4. METHODS TO OBTAIN THE DAMPING RATIO

i (t)

1
zi1 + zi

(ui+1 ui )

. . . +(ui ui1 )
ui ui1
zi1
i = 2, 3, . . .

i1/2 (t)

133

zi1

zi

zi
zi1

(9.31)
(9.32)

in which ui = u(zi , t) is the lateral displacement.

The displacement can be calculated through double integration of the recorded acceleration history u
(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 function of the distance between the recording points and the recorded acceleration
wavelength. The discrete formulas can only capture waves below a certain frequency, 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 min 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 min four times the distance z between the recording points
min = 4 z

fmax =

c
min

(9.33)

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 transducer spacings in seismic arrays are between 1.7 and 6 Hz.

134

CHAPTER 9. SCPT BASED TECHNIQUES

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 SCPTs. 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, recorded 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. Unfortunately 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 :
H1/2 =

F F T (a2 )
A2
=
F F T (a1 )
A1

(9.34)

Fourier transformation and inverse Fourier transformation are abbreviated

with FFT and IFFT, respectively.

[m]

1.033

1.033

1.033

1.033

i1

m=1

Index SCPT
subsetup (n)
n=1

Dual cone
sensors

A1, 2

A1,3

Surface
sensor

H 2/3 =

Depth
Index (m)

An , m = FFT ( a n , m )

Pm = FFT ( p m )

A2,3

a1,3
A2, 4

H 3/ 4 =

a1, 2

H 4/5 =

a 2, 4

a 2,3

A3, 4

A3,5

H i 1 / i =
Ai 2,i 1

Ai 2,i

a 3, 5

a 3, 4

i2

ai 2 , i

ai 2,i 1

SCPT time records of cons ecutive meas urement depths

(dual cone)
a1,1

= IFFT A1,3

k =4

H k 1 / k

= IFFT ( A1,3 H 3 / 4 H 4 / 5 ... H i 1 / i )

pi = IFFT ( Pi 1 H i 1 / i )

= ...

pi 1 = IFFT (Pi 2 H i 2 / i 1 )

Free ground
surface

= IFFT (A1,3 H 3 / 4 H 4 / 5 )

p5 = IFFT (P4 H 4 / 5 )

= IFFT (A1,3 H 3 / 4 )

p 4 = IFFT (P3 H 3 / 4 )

p3 = a1,3

p 2 = a1, 2

p1 = a1,1

Ps eudo s imultaneous s ignals

9.4. METHODS TO OBTAIN THE DAMPING RATIO

135

Figure 9.22: Use of transfer functions (H) to generate pseudo simultaneous

signals.

136

CHAPTER 9. SCPT BASED TECHNIQUES

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 accelerometer 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, min 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. min = 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, starting at the surface, and multiplied with the density. An averaged density
of = 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, , is plotted versus shear strain, . Complete
loops are selected and G-modulus and damping ratio are calculated based on

9.4. METHODS TO OBTAIN THE DAMPING RATIO

137

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

Data preparation as for spectral ratio slope

and attenuation coefficient method

Input data from LabVIEW

data acquis tition program

Generation of pseudo s imultaneous s ignals

bandpass filter with edge frequencies 10 and
60 Hz, tapered window (extendedcosinewindow)
+/ 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 teresisloops
manual selection of certain loops
calculation of the related Gmodulus 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.

138

CHAPTER 9. SCPT BASED TECHNIQUES

Figure 9.24:
SCPT3.

Normalized acceleration histories versus depth, Waremme,

Figure 9.25:
SCPT3.

Normalized displacement histories versus depth, Waremme,

9.4. METHODS TO OBTAIN THE DAMPING RATIO

139

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

Figure 9.27: Normalized shear pstress histories versus depth, Waremme,

SCPT3.

140

CHAPTER 9. SCPT BASED TECHNIQUES

the measurement of areas following Equation (2.28).

100

Hysteresis-Loop II

Hysteresis-Loop I

150

150
Loop II
t = 0.523 ... 0.544 s,
G = 63.436 MPa
100 Vs = 176 m/s (with = 2040 kg/m),
D = 27.57 %

Loop I
t = 0.505 ... 0.524 s,
G = 87.127 MPa
Vs = 207 m/s (with = 2040 kg/m),

50

Shear stress, [Pa]

Shear stress, [Pa]

D = 25.75 %

0
Loop I

-150
-2

0
t = 0.544 s
t = 0.523 s

-50

-50

-100

50

t = 0.505 s

-1

t = 0.524 s

0
Shear strain, [-]

Loop II

-100

2
x 10

-6

-150
-2

-1

0
Shear strain, [-]

2
x 10

-6

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 stressstrain diagrams show also in the parts of the strongest signal energy completely 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 necessary 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 components. 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 technique.

9.4. METHODS TO OBTAIN THE DAMPING RATIO

141

Shear strain

x 10

Shear strain, []

Loop I
1

1
Loop II
2
0.45

0.475

0.5

0.525

0.55

0.575

0.6

Time, t [s]
Shear stress
200
Shear stress, [Pa]

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 SCPTs 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 simultaneous acceleration time histories adds other difficulties to be handled. After
each integration step the signal has to be filtered to remove the appearing si-

142

CHAPTER 9. SCPT BASED TECHNIQUES

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 sensors 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 generated signals

Preliminary trials of the SRS method on SCPT data show a wide result scattering 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 simulated 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 particle 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 horizontal 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

+/ 0.00 m

+/ 0.00 m

VS = 107.5 m/s
VP = 215.0 m/s
D = 1.0 %

143

+/ 0.00 m

VS = 107.5 m/s
VP = 215.0 m/s
D = 1.0 %

3.00 m

7.00 m

VS = 107.5 m/s
VP = 215.0 m/s
D = 3.0 %

II

VS = 215.0 m/s
VP = 430.0 m/s
D = 2.3 %

III

VS = 107.5 m/s
VP = 215.0 m/s
D = 1.0 %

10.00 m

II

15.00 m

VS = 152.1 m/s
VP = 304.2 m/s
D = 2.3 %

15.00 m

a) HalfSpace

15.00 m

b) 1 Layer Profile

c) 2 Layer Profile

Figure 9.31: Stratification and parameters of the simulated sites.

Excitation
Axis

Cone
Axis

1m

v ertical
horizontal

+/ 0.00 m

Source
0.5 m
Calculation
Points

0.5 m

15.00 m

Figure 9.32: Set-up of the simulated SCPTs.

144

CHAPTER 9. SCPT BASED TECHNIQUES

Normalised force []

Time History Source Function

1
0.8
0.6
0.4
0.2
0
10

15

20

25
t [ms]

30

35

40

Normalised force []

Frequency Spectrum Source Function

1
0.8
0.6
0.4
0.2
0
0

200

400

600

800

1000
f [Hz]

1200

1400

1600

1800

2000

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 pand 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 MATLAB 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 signals 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 centered at the frequency of the maximum spectral amplitude. The damping ratio

9.5. ARTIFICIALLY GENERATED SIGNALS

145

a) HalfSpace

Depth [m]

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 excitation.

146

CHAPTER 9. SCPT BASED TECHNIQUES

b) 1 Layer Profile

Layer I

Depth [m]

6
S

Layer II

10

12

16
0

14

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.

9.5. ARTIFICIALLY GENERATED SIGNALS

147

c) 2 Layer Profile

Layer I

Depth [m]

Layer II

10

Layer III

12

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.

148

CHAPTER 9. SCPT BASED TECHNIQUES

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

Selected depth area

Meterbymeter approach
Cros s Correlation

Cros s Correlation

autoSCPTFromWinSig.m
autoSCPTFromWinSig3.m
Vs v s. Depth

autoSCPTFromWinSig2.m
Vs for the selected
depth area

Spectral Ratio
Slope Method (SRS)
dualmanu.m

Attenuation Coefficient
Method (ACM)
dualmanu3.m

Spectral Ratio
Slope Method (SRS)
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 SCPTs.

Window-length
The signal from the half-space profile a) is analyzed using three different window 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

9.5. ARTIFICIALLY GENERATED SIGNALS

149

shape of the signal after the window is applied. The windowing is done manually 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 additionally 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.
w(k + 1) = 0.5 1 cos 2

k
n1

, k = 0, . . . , n 1

(9.35)

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-cosinebell 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

150

CHAPTER 9. SCPT BASED TECHNIQUES

Time domain

Frequency domain
50

Hanning
T apered
Rectangular

Magnitude (dB)

Amplitude

0.8

0.6

0.4

-50

0.2

5
Time (ms)

10

15

-100
0

0.2

0.4

0.6

0.8

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
Gibbs 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.

9.5. ARTIFICIALLY GENERATED SIGNALS

151

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 Gibbs
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 halfspace
0

2.5

Depth [m]

5
Vp
7.5

Vs

10

12.5

15

50

100
150
200
Wave velocity [m/s]

250

300

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

152

CHAPTER 9. SCPT BASED TECHNIQUES

Simulated 1 layer profile

Simulated 2 layer profile

2.5

2.5

Vs
5

Depth [m]

Depth [m]

7.5

7.5
V

10

10

12.5

12.5

15

50

100
150
200
Wave velocity [m/s]

250

15

300

Figure 9.41: Calculated s-wave velocity, profile b), one layer on a halfspace.

100

200
300
Wave velocity [m/s]

400

500

Figure 9.42: Calculated wave velocity, profile c), two layers on a halfspace.

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 as a measure for scattering, the mean absolute deviation
and 11.5 m, D,

sD0 from the target value D0 = 1%. The latter is defined as:

sD0 =

1
n

n
i=1

|Di D0 | .

(9.36)

9.5. ARTIFICIALLY GENERATED SIGNALS

153

In the case where the signals at 5 and 12 m are compared the single calculated value D is given. sD0 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.

Normalized
Amplitude []

Spectrum at depth 10.5 m, A

1

0.5

Normalized
Amplitude []

200

400

600

800

1000

1200

1400

1600
1800
2000
Spectrum at depth 9.5 m, AT

200

400

600

800

1000

1200

1400

1600
1800
2000
Logarithmic Ratio, ln(A /A )

0.5

1.5
Logarithmic
Ratio []

Full signal (a)

1.5 cycles (b)
1 cycle (c)

Range of slope calculation,

+/ 150 Hz from the peakfrequency

1
0.5
0

200

400

600

800

1000
1200
Frequency [Hz]

1400

1600

1800

2000

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
0.5% is not known.
combination Hanning window and ACM, D

154

CHAPTER 9. SCPT BASED TECHNIQUES

cN

1 cycle

cET
cEH
cT
cH
cE
c

bN
1.5 cycles

bET
bEH
bT
bH
bE

Full signal

aN

Selected-Depth, SRS

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 SCPTs.

9.5. ARTIFICIALLY GENERATED SIGNALS

Modified Parameters

Metre-by-Metre
Method

SRS

Id-Code

Tapered Window

Hanning Window

Equal Window length

ACM

D
[%]

sD0
[%]

D
[%]

sD0
[%]

D
[%]

sD0
[%]

a
aH
aT
aN

1.003
1.002
1.003
1.025

0.009
0.005
0.009
0.225

0.975
0.505
0.975
0.968

0.025
0.495
0.025
0.111

1.002
1.002
1.002
1.063

0.002
0.002
0.002
0.063

b
bE
bH
bT
bEH
bET
bN

1.137
0.987
0.869
1.360
0.682
0.910
1.139

0.147
0.032
0.148
0.398
0.318
0.092
0.432

1.895
0.978
1.644
1.842
1.037
0.983
1.896

0.945
0.024
0.665
0.888
0.044
0.022
0.985

1.029
1.108
0.815
1.089
0.718
1.174
1.078

0.029
0.108
0.185
0.089
0.282
0.174
0.078

c
cE
cH
cT
cEH
cET
cN

0.580
0.469
0.360
0.496
0.271
0.329
0.584

0.566
0.645
0.640
0.593
0.729
0.671
0.586

1.281
1.173
0.993
1.034
1.031
1.108
1.273

0.476
0.179
0.581
0.414
0.466
0.395
0.483

0.570
0.774
0.480
0.561
0.569
0.761
0.638

0.430
0.226
0.520
0.439
0.431
0.239
0.362

x
x

x
x
x
x
x

x
x
x
x
x

Selected-Depth
Area Method

Noise

c) 1 cycle

b) 1.5 cycles)

a) Full signal

Window length

SRS

x
x

155

x
x

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

156

CHAPTER 9. SCPT BASED TECHNIQUES

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 comparing 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.

9.5. ARTIFICIALLY GENERATED SIGNALS

157

Simulated halfspace, full signal

0

2.5

Depth [m]

7.5

10

12.5
SRS a
ACM a
15
0.5

0.5

1
1.5
2
Damping ratio [%]

2.5

Figure 9.45: Damping ratio, profile a), based on the full signal.

158

CHAPTER 9. SCPT BASED TECHNIQUES

Simulated halfspace, 1.5 cycles, ACM

0

2.5

2.5

Depth [m]

Depth [m]

Simulated halfspace, 1.5 cycles, SRS

0

7.5

10

12.5

0.5

1
1.5
2
Damping ratio [%]

2.5

15
0.5

Figure 9.46: Damping ratio, profile

a), based on a windowed signal of 1.5
cycle length, spectral ratio slope method.

0.5

1
1.5
2
Damping ratio [%]

2.5

Simulated halfspace, 1 cycle, ACM

0

2.5

2.5

Depth [m]

Depth [m]

Simulated halfspace, 1 cycle, SRS

7.5
SRS c
SRS cE
SRS cH
SRS cT

10

Figure 9.47: Damping ratio, profile

a), based on a windowed signal of 1.5
cycle length, attenuation coefficient
method.

7.5
ACM c
ACM cE
ACM cH
ACM cT

10

12.5

15
0.5

ACM b
ACM bE
ACM bH
ACM bT

10

SRS b
SRS bE
SRS bH
SRS bT

12.5

15
0.5

7.5

12.5

0.5

1
1.5
2
Damping ratio [%]

2.5

Figure 9.48: Damping ratio, profile

a), based on a windowed signal of 1
cycle length, spectral ratio slope method.

15
0.5

0.5

1
1.5
2
Damping ratio [%]

2.5

Figure 9.49: Damping ratio, profile

a), based on a windowed signal of 1 cycle length, attenuation coefficient method.

9.5. ARTIFICIALLY GENERATED SIGNALS

159

Simulated halfspace, noise, ACM

0

2.5

2.5

Depth [m]

Depth [m]

Simulated halfspace, noise, SRS

0

SRS a
SRS aN
SRS b
SRS bN
SRS c
SRS cN

7.5

10

10

12.5

15
0.5

7.5

ACM a
ACM aN
ACM b
ACM bN
ACM c
ACM cN

12.5

0.5

1
1.5
2
Damping ratio [%]

2.5

15
0.5

Figure 9.50: Damping ratio, profile

a), noise influence, spectral ratio slope
method.

0.5

Simulated 1 layer profile

Simulated 2 layer profile

0

2.5

Depth [m]

2.5

7.5
D = 2.3 %

10

D = 2.3 %
7.5

10

SRS
SRS E
ACM
ACM E

12.5

12.5

5
Damping ratio [%]

SRS
SRS E
ACM
ACM E

D = 3.0 %

D = 1.0 %

Depth [m]

2.5

Figure 9.51: Damping ratio, profile

a), noise influence, attenuation coefficient method.

15
5

1
1.5
2
Damping ratio [%]

10

15

Figure 9.52: Damping ratio, profile

b), selected results in comparison with
the input values.

15
5

D = 1.0 %

5
Damping ratio [%]

10

15

Figure 9.53: Damping ratio, profile

c), selected results in comparison with
the input values.

160

9.6

CHAPTER 9. SCPT BASED TECHNIQUES

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 SCPTs are divided in three measurement campaigns: campaign 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 Br
uel & Kjr accelerometers.
Site
Retie
Lincent
Waremme
Sint-Katelijne-Waver
Ghent

I
SCPT1 to SCPT4
SCPT1 to SCPT3
SCPT1 to SCPT2

Campaign
II
SCPT5
SCPT4 to SCPT5
SCPT3 to SCPT5
SCPT1 to SCPT2

III

SCPT1

Table 9.3: Assignment of the performed SCPTs to the measurement campaigns.

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 measurements.
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 according to the algorithm used for the simulated SCPTs. 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 / semiautomatic signalpartselection
(rectangular window)
poss. lowpass filtering (200 Hz cutoff frequency)
srsdatanew.m

Meterbymeter approach
CrossCorrelation
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.

162

CHAPTER 9. SCPT BASED TECHNIQUES

Depth [m]

Depth [m]

After the conversion of the ASCII data, gathered with the LabVIEWprogram, 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.

8
10

8
10

12

12

14

14

16

16

18
0.45

0.5

0.55

t [s]

0.6

0.65

0.7

Figure 9.55: Sint-Katelijne-Waver,

bottom element, y-axis, sledge hammer, SCPT2.

18
0.45

0.5

0.55

t [s]

0.6

0.65

0.7

Figure 9.56: Ghent, bottom element,

y-axis, sledge hammer, SCPT1.

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

9.6. MEASUREMENTS PERFORMED AT THE TEST SITES

163

Unfiltered, averaged signal

Particle acceleration [m/s]

0.1

0.05

0.05

0.1
0.55

0.555

0.56

0.565

0.57

0.575
Time [s]

0.58

0.585

0.59

0.595

0.6

0.59

0.595

0.6

200 Hzlowpass filtered, averaged signal

Particle acceleration [m/s]

0.1

0.05

0.05

0.1
0.55

0.555

0.56

0.565

0.57

0.575
Time [s]

0.58

0.585

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.

164

CHAPTER 9. SCPT BASED TECHNIQUES

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 laboratory tests and a shear strain of smaller than 103 %, 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

Depth [m]

Depth [m]

Mechanical Hammer
Sledge Hammer

10

10

12

12

Mechanical Hammer (SCPT)

Sledge Hammer (SCPT)
SASW
14
0

50

100

150

200

250

300

350

400

SWave Velocity [m/s]

Figure 9.58: Shear wave velocity Vs ,

site in Retie, campaign II.

14
0

10

20

30

40

Damping Ratio [%]

Figure 9.59: Damping ratio Ds , site

in Retie, campaign II.

9.6. MEASUREMENTS PERFORMED AT THE TEST SITES

165

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 profile 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

Lincent

0
Mechanical Hammer
Sledge Hammer

Depth [m]

Depth [m]

Mechanical Hammer (SCPT)

Sledge Hammer (SCPT)
SASW
50

100

150

200

250

300

350

400

SWave Velocity [m/s]

Figure 9.60: Shear wave velocity Vs ,

site in Lincent, campaign II.

10

Damping Ratio [%]

Figure 9.61: Damping ratio Ds , site

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 105 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

166

CHAPTER 9. SCPT BASED TECHNIQUES

Waremme

Waremme
0

Depth [m]

Depth [m]

10

10

12

12

Mechanical Hammer (SCPT)

Sledge Hammer (SCPT)
SASW
Resonant Column Test
Free Torsion Pendulum Test

14

16
0

50

100

150

200

250

300

Mechanical Hammer
Sledge Hammer
Resonant Column Test
Free Torsion Pendulum Test

14

350

SWave Velocity [m/s]

Figure 9.62: Shear wave velocity Vs ,

site in Waremme, campaign II.

16
0

10

Damping Ratio [%]

Figure 9.63: Damping ratio Ds , site

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 (Br

uel & Kjr accelerometers)

The SCPT on the campus of the Ghent University is done with the Br
uel &
Kjr accelerometers, providing a much higher resolution than the Hamlin sensors 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. Regarding to the experiences with small window lengths on the simulated signals,

9.6. MEASUREMENTS PERFORMED AT THE TEST SITES

Sint KatelijneWaver
0

Depth [m]

Depth [m]

Sint KatelijneWaver
0

5
6

5
6

9
Mechanical Hammer (SCPT)
Sledge Hammer (SCPT)
Resonant Column Test
Free Torsion Pendulum Test

10
11
0

167

50

100

150

200

250

Mechanical Hammer (SCPT)

Sledge Hammer (SCPT)
Resonant Column Test
Free Torsion Pendulum Test

10

300

SWave Velocity [m/s]

Figure 9.64: Shear wave velocity Vs ,

site in Sint-Katelijne-Waver, campaign II.

11
0

10

Damping Ratio [%]

Figure 9.65: Damping ratio Ds , site

in Sint-Katelijne-Waver, campaign
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-

168

CHAPTER 9. SCPT BASED TECHNIQUES

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

Depth [m]

7.5

10

12.5

15

17.5
Full signal
2nd part
1 cycle

20

22.5

10

0
10
Damping ratio [%]

20

30

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.

Logarithmic Ratio []

Normalized Amplitude []

Normalized Amplitude []

9.7. SUMMARY AND REMAINING PROBLEMS

169
Spectrum at depth 13.22 m, AB

Full signal
2nd part
1 cycle
0.5

50

100

150

200

250
300
Spectrum at depth 12.19 m, AT

50

100

150

200

250
300
Logarithmic Ratio, ln(AT/AB)

0.5

0
1

Range of slope calculation,

0 until the peakfrequency + 150 Hz

0.5
0
0.5

50

100

150
Frequency [Hz]

200

250

300

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 because 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 swave velocities can be calculated by a fast automated procedure. This is shown
by SCPTs 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 amplitudes of time records of the same vibration event gathered at different source
distances (attenuation coefficient method (ACM), spectral ratio slope method

170

CHAPTER 9. SCPT BASED TECHNIQUES

Ghent

Ghent

Left side
Right side

Depth [m]

Depth [m]

10

10

12

12

14

14

16

16

18
0

50

100

150

200

250

300

350

400

SWave Velocity [m/s]

Figure 9.68: Shear wave velocity Vs ,

site in Ghent, campaign III.

18
0

Left side
Right side

10

15

20

25

30

Damping Ratio [%]

Figure 9.69: Damping ratio Ds , site

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 stratifications. 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 agreement with the SCPTs concerning the s-wave velocity. The range of damping
ratios obtained by the SRS method matches with the results from the laboratory 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 Br
uel & Kjr accelerometers 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 falling 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 hammer 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 restricted. 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.

172

CHAPTER 9. SCPT BASED TECHNIQUES

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
103 %, 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 described 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 SintKatelijne-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 central 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 displacement. 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

b)

a)

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:
Dimensions,
length:
width:
thickness:
Resonant frequency,
unmounted:
capacitance:
Maximum drive voltages,
low frequency:
at resonance:
used range 0.1 - 15 kHz:
Sample penetration,
bottom cap:
top cap:

Lead Zirconate
Titanate (PZT-5HN)
20 mm
10 mm
0.5 mm
610 Hz
6.5 nF
50 V
15 V
20 V
1.6 mm
2.4 mm

Table 10.1: Properties of the bender elements.

176

10.2.2

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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
Source
Bender
elements

Signal Analyzer

Ch.1 Ch.2

Sample
Amplifier
IN OUT TRIGG.

Source

Top cap

Bottom cap

Var.: trigger on
step pulse

Bender
elements

Ch.1 Ch.2

Amplifier
IN OUT TRIGG.

Sample

Bottom cap

Var.: trigger on
sending signal

Figure 10.2: Schematic test setup for

the pulse test with the two variations
of triggering.

Figure 10.3: Schematic test setup for

the resonant test.

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 amplifier.

Figure 10.5: Triaxial cell with

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

178

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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.
4

5
3
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 element (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 Jovicic 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.
20

Input
Signal

0.25

10
A

C
0

A
0

0.25 NearField Effect

B

0.5
0.5

10

Input voltage [V]

Output voltage [mV]

0.5

20
0.5

1.5

2.5
Time [ms]

3.5

4.5

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 s, 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

180

CHAPTER 10. BENDER ELEMENT TECHNIQUE

infinite isotropic medium. Jovicic 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 Jovicic 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 capability of the bender elements to follow the driving signal. Studies of different
authors as for instance by Jovicic et al. (1996), applying self-monitoring elements, 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

10.3. TECHNIQUES TO DETERMINE GMAX

181

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 comparing 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 receiver 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:

rRS ( ) =

sR (t) sS (t + ) 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 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 element 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 arrivals 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

182

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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 innovations 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 transmitter element at a low frequency. The output from the receiver is displayed
together with the driving voltage. The frequency of the input voltage is increased until the input and the output signals come into phase. If the initial
frequency was low enough, under this conditions exactly one complete wavelength 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 whether 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

10.3. TECHNIQUES TO DETERMINE GMAX

183

velocity is faster than the s-wave and can reach, if the sample is fully saturated, 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 (1 and 2 ) with the velocity of a shear wave and the
third (3 ) with the velocity of a compression wave. For the three components
the attenuation arising from the geometrical damping occurs at different rates,
2 and 3 attenuating an order of magnitude faster than the 1 . 1 represents
the far field pure shear wave. 2 and 3 are the near field components, because
they have only significant amplitudes in close proximity to the source.
The initial polarity of 3 is opposite to 2 and more important also to
1 . 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 3 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 fluidsoil 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 transmitter element for the purpose of reversing only the shear wave arrival (1 )
since also 2 and 3 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 component in relation to those of the first s-wave arrival becomes less significant with

184

CHAPTER 10. BENDER ELEMENT TECHNIQUE

an increasing Rd . Rd is defined by:

Rd =

d
df
=

Vs

(10.2)

where d is the travel distance, is the wavelength and f is the frequency of the
input signal. Because of the increasing deviation of the frequency of the received 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.

30

Rd - Ratio for a sample with 26 mm tip to tip distance

10

7 6

5
Rd=3

25
Rd=2

Rd=4
20

R =1

f mean [kHz]

Limits for the appearance of the near-field

effect by Viggiani & Atkinson (1995)
15

10
Rd = d /
5
R =0.25
d

0
0

100

200

300
Vs [m/s]

400

500

600

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

10.3. TECHNIQUES TO DETERMINE GMAX

185

Rd - Ratio for a sample with 117 mm tip to tip distance

30

10
7
6

25
5

R =d/
d

20
R =4
f mean [kHz]

15
Rd=3
Rd=2

10

Limits for the appearance of the near-field

effect by Viggiani & Atkinson (1995)

Rd=1
Rd=0.25

0
0

100

200

300
Vs [m/s]

400

500

600

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.

186

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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 normalized 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 spectrum 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 (s ) can be derived
from the wave equation for a shear wave propagating in a rod and the force
equilibrium at the boundaries:
k mT
s L
s L
=
tan
m
Vs
Vs

(10.3)

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 by the expression
already mentioned:
Vs =

(10.4)

10.4. METHODS TO OBTAIN THE DAMPING RATIO

187

A detailed formulation can be found for instance in Graff (1991) and Brocanelli & 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:
f2 =

3 EI
L3

33
140 mT

h
+ m 1 + 3L
+

9
4

h 2
L

(10.5)

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 concerning Rayleighs 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:
2 = 42

60 EI L2 + 5 GA k 2
mT + m

(10.6)

with
= 198L5 + 231L3 + 70L 2
and

= 840L5 + 2520L4h + 840L3 . . .

. . . + 1260L2h + 1890L3h2 + 210L 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. depends on the ratio between these two parts.
The unit of is m2 . A is the area of the cross section. The resonant frequency
can be calculated at those where 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 Rayleighs

188

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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:

0=

+ L
sin L

cos L
L

mr =

m
mT

L2
2

L
L
sinh L
L cosh L

mr
cos L

L sin L

L2

2 mr
cosh L
L sinh L

(10.7)

with

F = 2

2 =

2 2 1
L4

= 2 + 2 L2
=

H+

T2 L3 mT
EI

H = (1 + )

I
AL2

2(1 + )
k

2
L2

= 2 2 L2

H 2 4F
2

H +

H 2 4F

The natural circular frequency T can be extracted from this expression.

m
mT, A, I, , k

Figure 10.10: Simplified system with parameters of Equation (10.7).

Derivation of Equation (10.6), Rayleighs 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)

10.4. METHODS TO OBTAIN THE DAMPING RATIO

189

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. Neglecting the moment at the top end x = L, EI y (L) = 0. Then, Equation (10.8)
becomes
yB (x) = x2 [3L x] for x L

(10.9)

where is a constant given by = 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) = L2 [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 = x for x L

(10.11)

yS = L for x > L

(10.12)

and

The superposition of bending and shearing follows then with:

y(x) = x2 [3L x] + x for x L

(10.13)

y(x) = L2 [2L + 3(x L)] + L for x > L

(10.14)

and

The maximum internal potential energy Epot is computed by taking into consideration the internal energy of the bending and shearing component of the
deformed specimen:
Epot =

1
EI
2

L
0

1
(y )2 dx + GAk
2

(y )2 dx

(10.15)

Introducing Equation (10.13) in Equation (10.15) gives:

Epot

= 6EI2 L3 . . .
12 2 5
1
L + 2L3 + 2 L
5
2

. . . + GAk

(10.16)

The maximum kinetic energy Ekin is computed by presuming harmonic oscillation with frequency . The component for the specimen is
Ekin, mT =

1 2
A
2

L
0

y 2 dx

(10.17)

190

CHAPTER 10. BENDER ELEMENT TECHNIQUE

or using Equation (10.13):

33 2 6 11
1
L + L4 + 2 L2
70
20
6

Ekin, mT = mT 2

(10.18)

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:
Ekin, m =

1
2
m yh+L
2
2

(10.19)

and

Ekin, m

1
m 2 (42 L6 + 122 L5 h + 4L4 . . .
2
. . . + 6L3 h + 92 L4 h2 + 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 :
2 = 42

60 EI L2 + 5 GA k 2
mT + m

(10.21)

with
= 198L5 + 231L3 + 70L 2
and

= 840L5 + 2520L4h + 840L3 . . .

. . . + 1260L2h + 1890L3h2 + 210L 2

At the same time and are replaced by = /.

Top Cap

Sam ple

mT
x

E ,G=
G=0

m
mT

E ,G=
E =0

a)

Figure 10.11: Assumed mode shapes a) bending and b) shearing.

b)

10.4. METHODS TO OBTAIN THE DAMPING RATIO

191

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 cylindrical symmetry the situation enables the mathematical analysis based on
the Pochhammer-Chree frequency equation originally published in Pochhammer (1876) and later independently by Chree (1889). This leads to the classification 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, , z, t) = u(r) ei n ei( tk z)

(10.22)

where u = ur er + u e + uz ez is the displacement vector in cylindrical coordinates r, , z and and k are the circular frequency and the wavenumber,
respectively. The circumferential order n is an integer that describes the symmetry 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 displacement 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.

192

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Torsional

ur
uz

Longitudinal

ur
uz

Flexural

Figure 10.12: Torsional, longitudinal and flexural modes in a cylindrical waveguide (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 frequency equation for the waveguide in the form of:
det Df (k, , n, b) = 0

(10.23)

where Df is a 6x6 matrix. The cylinder parameters are the outer radius b, the
p-wave velocity in a rod c0 (c0 = E/) and the Poisson ratio . 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
& Suhubi (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. Information on these tools can be found for instance in Seco et al. (2002). The input parameters for the program are a cylinder radius b = 2.5 cm, a density
= 1984 kg/m3 , a rod p-wave velocity Vp = c0 = 348.4 m/s and a Poisson ratio = 0.29. The parameters correspond to a s-wave velocity measured in the
specimen of Vs = 216.9 m/s.

10.4. METHODS TO OBTAIN THE DAMPING RATIO

193

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.

Torsional
Type
fcut
[kHz]
T(0,1)
0.0
T(0,2) 7.0916

Modes
Longitudinal
Type
fcut
[kHz]
L(0,1)
0.0
L(0,2) 5.2910
L(0,3) 5.3678
L(0,4) 9.6875

Transversal
Type
fcut
[kHz]
F(1,1)
0.0
F(1,2) 2.5424
F(1,3) 3.8733
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 .

0
0

100

200

300

400

500

600

700

800

900

1000

V (half space)
r

F(1,1)

F(2,1)
T(0,1)

L(0,1)

V (infinite medium/rod)
s

c0 =Vp (rod)

V (infinite medium)
p

F(1,2)

L(0,2)

F(2,2)

5
6
Frequency [kHz]

F(1,3)

F(2,3)

T(0,2)

Dispersion curves (phase speed)

F(1,4)

L(0,3)

F(1,4)

F(2,4)

10

L
(0,4)

194
CHAPTER 10. BENDER ELEMENT TECHNIQUE

Figure 10.13: Dispersion curves of an infinite cylinder with 5 cm diameter

reconstituted silt from the Waremme site.

Phase speed [m/s]

10.4. METHODS TO OBTAIN THE DAMPING RATIO

195

- 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
& Suhubi (1975)) that the dispersion curve is coincident with that predicted by simpler theories of beam flexure. For the whole frequency range
the Timoshenko theory for thick beams is applicable and for the low frequency 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 dynamic 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:
fcut, F (1,2) = 0.5681

Vs
d

(10.24)

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

196

CHAPTER 10. BENDER ELEMENT TECHNIQUE

z y
x
Figure 10.14: Discretization of sample and
top-cap.

Figure 10.15: Axes orientation.

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
flexural

1301 Hz
torsional

2171 Hz
longitudinal

2354 Hz
flexural

4251 Hz
torsional,
2nd order

Table 10.3: Natural frequencies found by finite element calculation for sample
S1, 100 kPa and CA = 0.

10.4. METHODS TO OBTAIN THE DAMPING RATIO

197

Poissons 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 waves was not done. Therefore, the elasticity modulus was calculated assuming
a Poissons ratio of = 0.29. The finite element analysis is performed with
different values to evaluate the influence of 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 = 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, Poissons ratio
independent assumption k = 0.9, Cowper (1966) gives the following expression
for a circular cross section:
k=

6(1 + )
7 + 6

(10.25)

This gives for a Poissons ratio of = 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 = 2051 kg/m3
and Poissons ratio = 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

198

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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
1
2
3
4
5
6
7
8
9
10

z
[mm]
0.40
0.80
1.60
5.66
9.72
13.77
17.83
21.89
25.94
30.00

hinflu.
[mm]
0.40
0.80
2.43
4.06
4.06
4.06
4.06
4.06
4.06
2.03

k
[-]
1.0487
1.0254
0.9551
0.8552
0.7824
0.7556
0.7609
0.7781
0.7642
0.7412

Average:

=
k

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 Cowpers shear coefficient. This seems reasonable since
Cowper points out that his values of k are most satisfactory for static and lowfrequency deformations of beams and not for high-frequency vibrations as in
our case.
Other authors, also summarized in Cowpers 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):
ma = C A F

d2
4

(10.26)

10.4. METHODS TO OBTAIN THE DAMPING RATIO

Cross-section

199

Center-line

Shear stress, section at z = 0 mm

30
100

150

150
200

20

100

Isolines
labelled in N/m

30
30

Shear stress, line at y = 0, z = 0 mm

Shear stress [N/m]

y
x

0
10

250

250

200
150

10
y [mm]

z = 0, k = 1.0487

200
150

20

20

10

0
10
x [mm]

20

300
200
100
0
30

30

20

10

0
x [mm]

10

20

30

Shear stress, section at z = 5.66 mm

30
120

20

Shear stress [N/m]

0
150
140 90
130
1
120 1

Isolines
labelled in N/m

30
30

120

160

12 1150
110 13400
0

20

Shear stress, line at y = 0, z = 5.66 mm

110
10090

10090
120 1
2
160150 0

160

90

10

130
10400
15 16

y
x

150
0
14
130
110

z = 5.66 mm, k = 0.8552

90

10
y [mm]

140130 1110
1020
150
0
160

0
11010130
90 140

20

10

0
10
x [mm]

20

300
200
100
0
30

30

20

10

0
x [mm]

10

20

30

Shear stress, section at z = 13.77 mm

30
140

Shear stress [N/m]

40

1
8000
60

16
0

20

Shear stress, line at y = 0, z = 13.77 mm

160
40

12
0

y [mm]

160

14

121000 0
8

10

y
x

120
140

40

80
100

10

z = 13.77 mm, k = 0.7556

160

60
80
100
140 120

20

Isolines
labelled in N/m

30
30

20

10

0
10
x [mm]

20

30

300
200
100
0
30

20

10

0
x [mm]

10

20

30

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.

200

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Cross-section

Center-line

Shear stress, section at z = 21.89 mm

30

60

120

100
80

80

Shear stress [N/m]

120

100
80

80

60

100

120

10

Shear stress, line at y = 0, z = 21.89 mm

40

y
x

0
40

y [mm]

100

60

10

z = 21.89 mm, k = 0.7781

120

20

20
Isolines
labelled in N/m

30
30

20

10

0
10
x [mm]

20

300
200
100
0
30

30

20

10

0
x [mm]

10

20

30

Shear stress, section at z = 30 mm

30
120
10

100

60
100

80

20

60
80

10

Shear stress, line at y = 0, z = 30 mm

Shear stress [N/m]

20

y
x

100

80

40

80
60
40

60

10
y [mm]

z = 30.00 mm, k = 0.7412

20

20
Isolines
labelled in N/m

30
30

20

10

120

0
10
x [mm]

20

30

300
200
100
0
30

20

10

0
x [mm]

10

20

30

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.

10.4. METHODS TO OBTAIN THE DAMPING RATIO

201

ma denotes the added mass per unit length, CA is a non-dimensional added

mass coefficient, F 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):
CA =

1
1 + (d/l)2

(10.27)

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 measuring damping uses the relative width of the response spectrum. Using the
variables indicated on the curve in Figure 10.16, the logarithmic decrement
and the damping ratio D, can be calculated from (Richart et al. (1970)):
2D
f22 f12
=
=
2
2
fm
1 D2

A2
A2max A2

1 2D2
1 D2

(10.28)

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 :
D=

f22 f12
2
4fm

(10.29)

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.

202

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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:
D=

20

22 12
2 tan 22 + 1 tan 21

(10.30)

with:
- 0 , angular frequency corresponding to the maximum sweep angular velocity;
- 1 , 2 , angular frequencies;
- 1 , 2 , angles at both sides of 0 .
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 .
Multi-degree of freedom methods The frequency response function from
a linear system is defined as the ratio between the frequency domain representations of the signal leaving the system Y() and the input signal X(). The
function can be described further as a ratio of two polynomials, as shown in
Equation (10.31).
H() =

Y ()
=
X()

m
k=0 ak
n
k=0 bk

(i)k
(i)k

(10.31)

10.4. METHODS TO OBTAIN THE DAMPING RATIO

203

Im

Re
c
1

1
0

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 and the roots 0 can be determined. Equation (10.32) gives such
a partial fraction form.
H() =

am
bn

m
k=1 (i i0,k )
n
k=1 (i i,k )

a m im
b n in

m
k=1 ( 0,k )
n
k=1 ( ,k )

(10.32)

For a model with N-degrees-of-freedom the frequency response function contains N-pole pairs. The order of nominator, m, and denominator, n, polynomial has not to be different. Iglesias (2000) shows that both can be chosen as
n = m = 2N. This leads to the following simplified equation:
a2N
H() =
b2N

2N

k=1

0,k
,k

(10.33)

The eigenfrequencies fk and the damping ratio Dk can be calculated based

on the poles of the response function using these relations:
fk =

|,k |
2

Dk =

Re(,k )
|,k |

(10.34)
(10.35)

The practical difficulties in applying this approach are essentially the calculation of appropriate curve fitting parameters ak and bk . Fortunately the problem is almost identical with the design of a digital infinite-impulse-response

204

CHAPTER 10. BENDER ELEMENT TECHNIQUE

filter (IIR-filter) based on a given filter frequency response function. The obtained 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-KatelijneWaver 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 S 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:
S =

1
|F F T (r1st )|
ln
2L |F F T (r2nd )|

(10.36)

10.4. METHODS TO OBTAIN THE DAMPING RATIO

Pulse frequency:
3000 Hz

Pulse frequency:
4500 Hz

Pulse frequency:
6000 Hz

1st arrival
0.8 0.5

0.5

2nd arrival

3rd arrival

1.5
t in s

205

2.5

3.5
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:

S =

|F F T (r1st )|
1
ln
4L |F F T (r3rd )|

(10.37)

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 s.
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):
Ds =

z Vs
2

(10.38)

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

206

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Attenuation coefficient [1/m]

10

1st to 2nd arrival

1st to 3rd arrival

8
6

z = 2.0352 10
DS = 7.45 %

s/m,

z = 2.0041 103 s/m,

DS = 7.34 %

2
0
0

1000

2000 3000 4000

Frequency [Hz]

5000

6000

Figure 10.19: Attenuation coefficient for a pulse of 3000 Hz.

Attenuation coefficient [1/m]

10
8

z = 1.7205 103 s/m,

DS = 6.30 %

6
3

z = 2.3415 10
D = 8.57 %

s/m,

2
1st to 2nd arrival
1st to 3rd arrival

0
0

2000

4000
6000
Frequency [Hz]

8000

Figure 10.20: Attenuation coefficient for a pulse of 4500 Hz.

Attenuation coefficient [1/m]

20

1st to 2nd arrival

1st to 3rd arrival

15

10

z = 2.5276 103 s/m,

DS = 9.25 %

z = 2.1814 10
D = 7.99 %

s/m,

0
0

2000

4000
6000
Frequency [Hz]

8000

Figure 10.21: Attenuation coefficient for a pulse of 6000 Hz.

10.4. METHODS TO OBTAIN THE DAMPING RATIO

Excitation
frequency

Parameter

Unit

3000 Hz

fmean
S
z
Ds

[Hz]
[m1 ]
[s/m]
[%]

3074
5.13575
2.0352 103
7.45

2922
3.89520
2.0041 103
7.34

4500 Hz

fmean
S
z
Ds

[Hz]
[m1 ]
[s/m]
[%]

4427
7.25377
1.7205 103
6.30

3830
5.91632
2.3415 103
8.57

6000 Hz

fmean
S
z
Ds

[Hz]
[m1 ]
[s/m]
[%]

4953
8.18255
2.5276 103
9.25

4166
6.96369
2.1814 103
7.99

207

Considered travel path

1st to 2nd
1st to 3rd
arrival
arrival

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, S : attenuation coefficient, z:
rise of S 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 S -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

208

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Attenuation coefficient [1/m]

10
8

z = 1.6108 103 s/m,

DS = 5.90 %

6
z = 2.4163 103 s/m,
DS = 8.85 %

4
2
0
2500

1st to 2nd arrival

1st to 3rd arrival

3000

3500 4000 4500

Frequency [Hz]

5000

5500

Figure 10.22: Attenuation coefficient at the mean dominant frequencies of the

three excitation pulses.

factor c() in the Equations (10.36) and (10.37). This means for the first and
the second arrival:

=
=

F F T (r1st ) c()2
1
ln
2L
|F F T (r2nd )|
|F F T (r1st )|
1
1
ln
+
ln c()2
2L |F F T (r2nd )| 2L

(10.39)

and for the first and the third arrival:

=
=

F F T (r1st ) c()4
1
ln
4L
|F F T (r3rd )|
1
|F F T (r1st )|
1
ln
+
ln c()4
4L |F F T (r3rd )| 4L

(10.40)

The exponent of c() 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):

c=

ISample
ICap 1
I
1 + Sample
ICap

(10.41)

10.4. METHODS TO OBTAIN THE DAMPING RATIO

209

s1

Cap Sample Cap

Loading
Piston

s2 = IFFT(FFT(s1) c())
Sample
c1

Top/Bottom
Cap

Base Plate

Figure 10.23: Schematic view of wave refection

and refraction on top and bottom cap.

Figure 10.24: Triaxial

cell with installed sample.

The impedances I can be calculated by the material density and the shear
wave velocity Vs :
I = Vs

(10.42)

According to the equation, the reflection factor c is not depending on frequency and therefore has no influence on the damping calculation using the
spectral ratio slope technique. For the used testing device and the tested sample |c| = 0.9586. This value is based on the material parameters in Table 10.8.
Parameter
Density,
S-wave velocity, Vs
Impedance, I
Reflection coefficient, |c|

Unit
[kg/m3 ]
[m/s]
[kg/(m2 s)]
[-]

Sample
Caps
2342
8009
230
3180
5.3866 105 2.5470 107
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() depending on frequency in a range below 1000 Hz. At higher frequencies
c() is nearly constant.

210

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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 attenuation 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 function (f), potentially highly frequency dependent. (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 (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

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 attenuation coefficient is adapted using the parameters defined in Figure 10.26.

10.4. METHODS TO OBTAIN THE DAMPING RATIO

211

Equation (9.19) is repeated here:

ln
(f ) =

A1
A2

x1
x2

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
FFT(r1A)

Sender 1
FFT(r1A)1

BE

z1A

BE

Receiver 1
FFT(r1B)
BE

Sender 1
FFT(r1B)1

z2A

Receiver 2
FFT(r2A)
BE

BE

z1B

Sender 2
BE

FFT(r2A)2

Receiver 2
FFT(r2B)
BE

Soil Specimen

FFT(r2B)2

Sender 2
z2B

BE

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 F T (r1A ) 1
F F T (r2A ) 2

(f) for setup 2 is:

ln
(f ) =

z1A
z2A

z2B
z1B

(10.43)

z2A z1A

F F T (r2B ) 2
F F T (r1B ) 1

(10.44)

z1B z2B

The Equations (10.43) and (10.44) can be transformed into:

(f ) (z2A z1A ) = ln

F F T (r1A )
F F T (r2A )

+ ln

1
2

+ ln

z1A
z2A

(10.45)

(f ) (z1B z2B ) = ln

F F T (r2B )
F F T (r1B )

+ ln

2
1

+ ln

z2B
z1B

(10.46)

and

212

CHAPTER 10. BENDER ELEMENT TECHNIQUE

If the sum of both equations is calculated the transfer functions 1 and 2

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

F F T (r1A )
F F T (r2A )

F F T (r2B )
F F T (r1B )

z1A
z2A

z2A z1A + z1B z2B

z2B
z1B

(10.47)

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 samples taken at the test sites based on the resonant 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 circlefit 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 Tables 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 velocities 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

214

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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 response 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 resonance 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 . 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 - and -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, accompanied 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 105 . 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

216

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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 transfer 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 Rayleighs 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. Undisturbed 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 assumption 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 overcome 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 velocity 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 investigation 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 BEresonant 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 possible reason of the high damping values obtained by the bender element resonant

218

CHAPTER 10. BENDER ELEMENT TECHNIQUE

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 frequencies, 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, watersaturated rocks, sands and silts would show a definite dependency of D on frequency. For saturated clays the authors conclude a relatively small frequency
dependency.
Several authors compare results from cyclic torsional shear tests and resonant 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
4 103 %. 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 performed 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 Biots 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

10.6. SUMMARY AND REMARKS

219

damping ratio of the shear wave, following Biots 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 Biots theory for the parameters given in Table 10.13.
Beside self-explanatory values, a pore-size parameter (a) and a structure constant (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 material. 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 Biots 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.

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Amplitude ratio receiver transmitter []

220

x 10

582 Hz

100 kPa
200 kPa
300 kPa

0.8

708 Hz

0.6

807 Hz
0.4

0.2

0
100

200

300

400

500

600

700

800

900

1000

Frequency [Hz]

Amplitude ratio receiver transmitter []

Figure 10.27: Frequency response spectra for sample S1 (reconstituted silt).

x 10

636 Hz
100 kPa
200 kPa
300 kPa

1.5

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).

Amplitude ratio receiver transmitter []

10.6. SUMMARY AND REMARKS

221

1.2

x 10

100 kPa
200 kPa
300 kPa

631 Hz
1

0.8
735 Hz
0.6

811 Hz

0.4

0.2

0
100

200

300

400

500
600
700
Frequency [Hz]

800

900

1000

Figure 10.29: Frequency response spectra for sample S3 (undisturbed silt).

Frequency flexural mode [Hz]

1200
Timoshenko Equation

1000

FEM

800 Rayleigh
Equation

Experiment

600
400
200
0

100

200

300

Confinement stress [kPa]

Figure 10.30: Calculated and measured resonant frequencies for the first flexural mode of sample S2, added mass effects are included (CA = 0.48).

222

CHAPTER 10. BENDER ELEMENT TECHNIQUE

10

HalfPower Method
CircleFit Method

S1

9
S1

8
S1

S3

Damping ratio, D [%]

7
S3
S2

S2

Sampling Depths:
S1, S2 > 4.04.5 m,
S3 > 5.05.5 m

S2

5
4
3

Free Torsion Pendulum (12.012.5 m)

S3
Resonant Column (5.05.5 m)

2
1

Resonant Column (4.04.5 m)

0
0

100

200
300
Stress, 1, 2, 3 [kPa]

400

500

Amplitude ratio receiver transmitter []

Figure 10.31: Damping ratios 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 103 %).

2.5

x 10

f = 636 Hz
m

max

A = Amax / 20.5

1.5

=(f f ) / (2 f ) = 6.2 %

HPM

f1 = 593 Hz

0.5

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.

10.6. SUMMARY AND REMARKS

223

x 10

Nyquist plot of
the respose spectrum

Imaginary Component

DCFM = 5.9 %
0
fitted circle
5

center point
10

15

20

1.5

resonance
frequency

data points used

for curve fitting

0.5

0.5

Real Component

1.5
5

x 10

Amplitude ratio receiver transmitter []

Figure 10.33: Circle-fit technique on the Nyquist plot of the response spectrum
of Figure 10.32.

x 10

811 Hz

300 kPa water

300 kPa air

3
930 Hz
2

0
0

500

1000
Frequency [Hz]

1500

2000

Figure 10.34: Comparison of water and air pressure, experimental results at

300 kPa and sample S3.

224

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Waremme

actual mean effective stress at a certain depth

indicates the calculated velocity out of a

laboratory test at sampling depth

confining stress stages used in laboratory tests

100 kPa

33 kPa

200 Pa

50 kPa

25 kPa

400 kPa

BETime arrival method

62 kPa

400 kPa

100 kPa

200 kPa

77 kPa
50 kPa

Depth [m]

Resonant Column Test

Mechanical Hammer (SCPT)

Sledge Hammer (SCPT)
SASW

14
0

50

100

600 kPa

400 kPa

100 kPa

50 kPa
100 kPa

150

200 kPa

141 kPa

Free Torsion Pendulum Test

50 kPa

12

BETime arrival method

200 kPa

126 kPa

10

200

250

300

350

SWave 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.

10.6. SUMMARY AND REMARKS

225

Waremme

actual mean effective stress at a certain depth

confining stress stages used in laboratory tests

indicates the calculated damping ratio out

of a laboratory test at sampling depth

Mechanical Hammer
Sledge Hammer
2

62 kPa

62 kPa

Depth [m]

77 kPa

Resonant Column Test

BEresonant method

10

141 kPa

12

14
0

50 kPa

100 kPa

Free Torsion Pendulum Test

10

Damping Ratio [%]

Figure 10.36: Ds profile of Waremme based on SCPT5 and including the other
laboratory tests.

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Amplitude ratio receiver transmitter []

226

x 10

397 Hz

50 kPa
100 kPa
200 kPa
400 kPa

460 Hz

6
5

609 Hz

4
3
2
735 Hz

1
0
100

200

300

400

500
600
700
Frequency [Hz]

800

900

1000

Amplitude ratio receiver transmitter []

Figure 10.37: Frequency response spectra for sample C4 (undisturbed Boom

clay).

x 10

50 kPa
100 kPa
200 kPa
400 kPa

0
100

200

300

400

500
600
700
Frequency [Hz]

800

900

1000

Figure 10.38: Frequency response spectra for sample C5 (undisturbed Boom

clay).

10.6. SUMMARY AND REMARKS

227

16

HalfPower Method
CircleFit Method

14

Damping ratio, D [%]

12
Sampling Depth:
C4 > 4.04.5 m

10
8
6

Free Torsion Pendulum (4.04.5 m)

Resonant Column (4.04.5 m)

4
2
0
0

100

200
300
Stress, 1, 2, 3 [kPa]

400

500

Figure 10.39: Damping ratios 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: 103 %).

228

CHAPTER 10. BENDER ELEMENT TECHNIQUE

SintKatelijneWaver

actual mean effective stress at depth

confining stress stages used in laboratory tests

indicates the calculated damping ratio out

of a laboratory test at sampling depth

Mechanical Hammer (SCPT)

Sledge Hammer (SCPT)

400 kPa

200 kPa

50 kPa

100 kPa

Resonant Column Test

400 kPa

100 kPa

121 kPa

200 kPa

BEresonant method

Free Torsion Pendulum Test

50 kPa

Depth [m]

121 kPa

9
0

50

100

150

200

250

300

SWave 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.

10.6. SUMMARY AND REMARKS

229

SintKatelijneWaver
confining stress stages used in laboratory tests

0
Mechanical Hammer
Sledge Hammer
1

00

00

,2

Resonant Column Test

200 kPa

50 kPa

400 kPa

BEresonant method
Free Torsion
Pendulum Test

100 kPa

Depth [m]

,1
50

a
kP

4
0,

9
0

10

15

Damping Ratio [%]

Figure 10.41: Ds profile of Sint-Katelijne-Waver based on SCPT2 and including
the other laboratory tests.

230

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Sample Description:
Sample index
Site
Description
Sampling
Depth

[m]

S1
Waremme
Silt
reconst. sample
4.0-4.5

S2
Waremme
Silt
undist. sample
4.0-4.5

29.0
50.0
25.0
101.53
1783.10

29.5
50.0
25.5
119.49
2062.90

Initial test parameters:

Length, L
[mm]
Diameters, d
[mm]
Distance BE tips
[mm]
Mass, mT
[g]
Unit mass,
[kg/m3 ]
Consolidation parameters:
Effective stress, peff
[kPa]
Water absorption
[g]
Effective mass, mT,eff
[g]
S-wave velocity, Vs
[m/s]

100
11.4
113.0
216.9

200
10.9
112.5
224.5

300
10.2
111.8
246.3

100
-3.2
116.3
197.7

200
-5.6
113.9
246.1

300
-7.0
112.5
271.3

Resonant frequencies:
CA = 0
Rayleigh Eq., fRay
[Hz]
Timosh. Eq., fTBT
[Hz]
FEM, fFEM
[Hz]

888
962
875

918
995
904

1005
1090
989

799
864
787

988
1070
972

1084
1175
1066

CA = 0 .48
Rayleigh Eq., fRay
Timosh. Eq., fTBT
FEM, fFEM

[Hz]
[Hz]
[Hz]

773
843
763

799
871
788

875
954
862

696
757
686

860
937
847

943
1028
929

Exp. result, fmeas

[Hz]

582

708

807

636

766

820

Damping ratio:
Half-Power, DHPM
Circle-fit, DCFM

[%]
[%]

7.9
7.2

9.1
9.9

8.0
8.9

6.2
5.9

7.3
4.7

5.4
4.9

Table 10.9: Test data for sample S1 (reconstituted silt) and S2 (undisturbed
silt).

10.6. SUMMARY AND REMARKS

Sample Description:
Sample index
Site
Description
Sampling
Depth
Initial test parameters:
Length, L
Diameters, d
Distance BE tips
Mass, mT
Unit mass,

231

[m]

S3
Waremme
Silt
undist. sample
5.0-5.5

[mm]
[mm]
[mm]
[g]
[kg/m3 ]

30.0
50.0
26.0
116.54
1978.44

Consolidation parameters:
Effective stress, peff

[kPa]

100

200

300

Water absorption
Effective mass, mT,eff
S-wave velocity, Vs

[g]
[g]
[m/s]

1.3
117.8
195.1

0.7
117.2
230.6

0.4
116.9
255.0

300
air
0.4
116.9
257.0

Resonant frequencies:
CA = 0
Rayleigh Eq., fRay
Timosh. Eq., fTBT

[Hz]
[Hz]

774
837

913
988

1009
1092

1017
1100

CA = 0 .48
Rayleigh Eq., fRay
Timosh. Eq., fTBT

[Hz]
[Hz]

674
766

796
904

879
999

886
1007

Exp. result, fmeas

[Hz]

631

735

811

930

Damping ratio:
Half-Power, DHPM
Circle-fit, DCFM

[%]
[%]

7.8
7.3

6.8
6.2

(10.0)
2.3

(17.5)
2.1

Table 10.10: Test data for sample S3 (undisturbed silt).

232

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Sample Description:
Sample index
Site
Description
Sampling
Depth

[m]

C4
Sint-Katelijne-Waver
Boom-Clay
undist. sample
4.0-4.5

Initial test parameters:

Length, L
[mm]
Diameters, d
[mm]
Distance BE tips
[mm]
Mass, mT
[g]
Unit mass,
[kg/m3 ]

29.8
50.0
25.8
114.60
1961.86

Consolidation parameters:
Effective stress, peff
[kPa]
Water absorption
[g]
Effective mass, mT,eff
[g]
S-wave velocity, Vs
[m/s]

50
5.7
120.3
157.0

100
8.6
123.2
178.2

200
8.5
123.1
253.5

400
8.1
122.7
299.6

Resonant frequencies:
CA = 0 .48
Rayleigh Eq., fRay
[Hz]

553

634

902

1064

Exp. result, fmeas

[Hz]

397

460

609

735

Damping ratio:
Half-Power, DHPM
Circle-fit, DCFM

[%]
[%]

15.7
10.7

12.1
7.3

14.9
12.8

13.5
13.4

Table 10.11: Test data for sample C4 (undisturbed clay), 30 mm height.

10.6. SUMMARY AND REMARKS

Sample Description:
Sample index
Site
Description
Sampling
Depth

[m]

233

C5
Sint-Katelijne-Waver
Boom-Clay
undist. sample
4.0-4.5

Initial test parameters:

Length, L
[mm]
Diameters, d
[mm]
Distance BE tips
[mm]
Mass, mT
[g]
Unit mass,
[kg/m3 ]

59.5
50.0
55.5
221.62
1885.65

Consolidation parameters:
Effective stress, peff
[kPa]
Water absorption
[g]
Effective mass, mT,eff
[g]
S-wave velocity, Vs
[m/s]

50
8.1
229.7
175.4

100
7.3
228.9
205.9

200
5.5
227.1
232.9

400
2.8
224.4
290.0

Resonant frequencies:
CA = 0 .71
Rayleigh Eq., fRay
[Hz]

255

299

337

418

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.

234

CHAPTER 10. BENDER ELEMENT TECHNIQUE

Log. Decrement []

Curve No.
1
2
3
4
5

101

102
1

102
Frequency [Hz]

10

103

104

Figure 10.42: Logarithmic decrement in function of the frequency for s-waves

according to Biots theory, Stoll (1979).

Material property
Mass density of grains, r
Mass density of fluid, f
Viscosity of fluid,
Shear modulus of frame, G
Structure constant, c
Intrinsic permeability, ki

Unit
kg/m3
kg/m3
Pa s
MPa
m2

Permeability, k = ki f g/
with g = 9.81m/s2
Pore-size parameter, a

m/s

Porosity,
Log. decrement of frame
for shearing motion,
Damping ratio, D = /(2)

2.5
1014
2.45
107
3.8
106
0.49

0.08
1.27

No. 1

Values for different curves

No. 2
No. 3
No. 4
2650
1000
0.001
56.7
1.25
1
1
2.3
1010
1011
1011
9.81
9.81
2.26
104
105
104
3.9
1.2
1.9
105
105
105
0.36
0.36
0.36
0.03
0.48

0.03
0.48

0.03
0.48

No. 5

5
1011
4.91
104
2.8
105
0.36
0.02
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 developed 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 considered. 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 qualified 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 eliminates 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 BEs 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 formulations (Timoshenko beam theory, Rayleighs 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 repeatable 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 reflected 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 multiple 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 recommended 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 simulated 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 development 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 alternative field test method. The technique is found in Lai (1998). Results
of these SASW tests can be used to evaluate the profiles of the SCPTs.
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 eliminate 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 torsional 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

A.1

APPENDIX A. ACCURACY AND COSTS OF SOIL TESTS

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, except for the cyclic simple shear test, based on a triaxial apparatus extended
with a loading facility to apply a dynamic loading. Cyclic tests are performed 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 instead of a cylindrical sample because it has a more uniform shear strain distribution 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 techniques, 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-ups 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.

244

APPENDIX A. ACCURACY AND COSTS OF SOIL TESTS

The reflection and refraction test provide, dependent on the actual soil stratification, 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-, downand 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.

Currency of
determination of

small

large

Applicability at
G / Gmax /

+
+
++
++
++
+++
++++
+
+++
+
++++
++
++
++
++
++

+
+
?
+++
++
+++
+
+ (?)
+
+
+
++
+
+
-

Costs

+++
++
+
+
+
+
+++
++
+++
+
++++

+++
++

Accuracy / Reliability

G
D
strains
strains
V s / Vp
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
+++
i
g
+++
Ultrasonic tests (only Vp )
Shaking table
r
r
g
++
In situ tests
f
g
++
SASW - Vs profile
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 < 103 %, large strain > 103 %
Applicability: g good, m medium, l less, - not suitable
Accuracy/Costs: + low, ++ medium, +++ high, ++++ very high, - not applicable, ? unknown

Testing method

246
APPENDIX A. ACCURACY AND COSTS OF SOIL TESTS

Table A.1: Comparison of the test methods for G (Gmax ) and D.

A.3. SELECTION CONCLUSIONS

Testing method

Laboratory tests
Resonant column test

Laboratory

Conditions and remarks

Costs
(excl.
VAT)

a)
b)

c
c

one sample and four consolidation stages

same as a)

2300 e
1850 e

one sample and one consolidation stage

650 e

a)

250 e

b)

single test on one sample and one consolidation stage

together with a conventional triaxial test
to determine the shear parameters, including: three samples with one consolidation stage each

a)

2100 e

b)

one sample, one consolidation stage, including on-specimen measurement of the

sample dimensions and vertical arrivaltime-BE-test, staggered increase of cyclic
loading until sample failure
one sample, one consolidation stage, several cyclic loading amplitudes

profile to 12 m depth, seismic registrations every meter, s-wave profile; exclusive

the costs for the CPT-truck, transport to
the site and working hours of the technician operating the CPT-truck
CPT-truck and transport to a site in
Flanders (price dependent on the site situation)
operating on the site by one person, price
per hour

550 e

Free torsion pendulum

test
Bender element test
(pulse test)

Cyclic triaxial test

Field tests
SCPT

Cross-hole test

SASW test

247

a)

b)

a)
b)

n
n

680 e

900 e

100 to
250 e
60 e

profile to 12 m depth, seismic registrations every meter, application of a pand a s-wave source, p- and s-wave velocity determination, including one source
borehole and two receiver holes
same as a)

3000 e

6000 e

one profile
one day of site work and one day of signal
processing and reporting

200 e
1500 to
1800 e

commercial expert, n noncommercial institute (university, research institute)

Table A.2: Typical costs for laboratory and in situ tests.

248

APPENDIX A. ACCURACY AND COSTS OF SOIL TESTS

Bibliography
Airey, D. W. & Wood, D. M., 1987. An evaluation of direct simple shear tests
on clay. Geotechnique, 37(1):2535.
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 Environmental and Engineering Geophysics, 4:8795.
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 643645.
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):120131.
Best, R., 1991. Digitale Mewertverarbeitung. 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):168178.
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):179191.
Blewett, J., Blewett, I. J. & Woodward, P. K., 1999. Measurement of shearwave velocity using phase-sensitive detection techniques. Canadian Geotechnical Journal, 36:934939.
Bolton, M. D. & Wilson, M. R., 1989. An experimental and theoretical comparison 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):384397.
Brocanelli, D. & Rinaldi, V., 1998. Measurement of low-strain material damping and wave velocity with bender elements in the frequency domain. Canadian Geotechnical Journal, 35:10321040.
Buttkus, B., 1991. Spektralanalyse und Filtertheorie in der angewandten Geophysik. Springer, Berlin.
Cascante, G. & Santamarina, C., 1997. Low Strain Measurements Using Random Noise Excitation. Geotechnical Testing Journal, 20(1):2939.
Cascante, G., Santamarina, C. & Yassir, N., 1998. Flexural excitation in a
standard torsional-resonant column device. Canadian Geotechnical Journal,
35:478490.
Chree, C., 1889. The equations of an isotropic elastic solid in polar and cylindrical 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 Timoshenkos Beam Theory.
Journal of Applied Mechanics, pages 335340.
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 5187. 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):16291653.
Dyvik, R. & Madshus, C., 1985. Lab Measurements of Gmax using Bender
Elements. In Advances in the Art of Testing Soils Under Cyclic Conditions,
pages 186196. ASCE, New York.
Elgamal, A.-W., Zeghal, M., Tang, H. T. & Stepp, J. C., 1995. Lotung Downhole Array. I: Evaluation of Site Dynamic Properties. Journal of Geotechnical
Engineering, 121(4):350362.
Eringen, A. C. & Suhubi, 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):130140.
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 Geoenvironmental Engineering, 126(6):543555.
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 Kr
afte auf Bauk
orper
im Tiefwasserbereich des Meeres. Institut f
ur Wasserbau der Universit
at
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, 20:167175.
Hardin, B. O., 1978. The Nature of the Stress-Strain Behavior for Soils. In
Earthquake Engineering and Soil Dynamics, volume 1, pages 390. ASCE,
New York.
Hardin, B. O. & Black, W. L., 1969. Vibration Modulus of Normally Consolidated Clay. Journal of the Soil Mechanics and Foundation Division, 95:1531
1537.
Hardy, S., Zdravkovic, L. & Potts, D. M., 2003. Numerical interpretation of
continuously cycled bender element tests. Department of Civil & Environmental Engineering, Imperial College, London.
Iglesias, A. M., 2000. Investigating Various Modal Analysis Extraction Techniques to Estimate Damping Ratio. Masters 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):182191.
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 ISC2 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 141.
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 techniques. In Specialty Workshop on Site Characterization and Exploration, pages
322334. Northwestern University, Evanston, Illinois.
Jovicic, V., Coop, M. R. & Simic, M., 1996. Objective criteria for the determining Gmax from bender element tests. Geotechnique, 46:357362.
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 Characterization, volume 2, pages 16551662. Porto.
Karl, L., Haegeman, W., Pyl, L. & Degrande, G., 2003. Measurement of material damping with bender elements in triaxial cell. In Proceedings of the Third
International Symposium on Deformation Characteristics of Geomaterials,
Vol. 1, pages 311. Lyon.
Kramer, S. L., 1996. Geotechnical earthquake engineering. Prentice Hall.
Kudo, K. & Shima, E., 1981. Attenuation of shear waves in soil. In M. Toks
oz
& D. Johnston, editors, Seismic Wave Attenuation, Geophysics reprint series
No. 2, pages 325338. Society of Exploration Geophysicists.
Lai, C. G., 1998. Simultaneous inversion of Rayleigh phase velocity and attenuation for near-surface site characterization. Ph.D. thesis, Georgia Institute
of Technology.
Lai, C. G., Rix, G. J., Foti, S. & Roma, V., 2002. Simultaneous measurement and inversion of surface wave dispersion and attenuation curves. Soil
Dynamics and Earthquake Engineering, 22:923930.
Lin, M.-L., Huang, T.-H. & You, J.-C., 1996. The effects of frequency on
damping properties of sand. Soil Dynamics and Earthquake Engineering,
15:269278.
Lings, M. L. & Greening, P. D., 2001. A novel bender/extender element for
soil testing. Geotechnique, 51(8):713717.
Lytle, R. J., 1978. Geophysical characterization using advanced data processing. In Specialty Workshop on Site Characterization and Exploration, pages
291300. Northwestern University, Evanston, Illinois.

BIBLIOGRAPHY

253

Martnez-Carvajal, H., Taboada-Urtuzu

astegui, V. M. & Romo, M. P., 2001.
Analysis of some downhole acceleration records from CENTRAL DE ABASTO OFICINAS site at Mexico City. In Fourth International Conference
on Resent Advances in Geotechnical Earthquake Engineering and Soil Dynamics and Symposium in Honor of Professor W.D. Liam Finn, pages 16,
paper no. 9.10. San Diego.
Menge, P., 2001. Soil investigation results at Sint-Katelijne-Waver. In Holeyman, editor, Screw Piles - Installation and Design in Stiff Clay, pages 1962.
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 305320. ASCE, Park City, Utah.
Molenkamp, F. & Smith, I. M., 1980. Hysteretic and viscous material damping.
International journal for numerical and analytical methods in geomechanics,
4:293311.
Nakagawa, K., Soga, K. & Mitchell, J. K., 1996. Pulse Transmission System for
Measuring Wave Propagation in Soils. Journal of Geotechnical Engineering,
122(4):302308.
Nazarian, S. & Desai, M. R., 1993. Automated Surface Wave Method: Field
Testing. Journal of Geotechnical Engineering, 119(7):10941111.
Ormsby, J. F. A., 1961. Design of numerical filters with applications to
missile data processing. Journal of the Association for Computing Machinery, 8(3):440466.

Pochhammer, L., 1876. Uber

die Fortpflanzungsgeschwindigkeiten kleiner
Schwingungen in einem unbegrenzten isotropen Kreiszylinder. Journal f
ur
die reine und angewandte Mathematik, 81:324336.
Pyl, L. & Degrande, G., 2001a. Determination of the dynamic soil characteristics with the SASW method at a site in Lincent. Internal Report BWM2001-02, K.U. Leuven.
Pyl, L. & Degrande, G., 2001b. Determination of the dynamic soil characteristics 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 Frequency Response Measurements using Rational Fraction Polynomials. In
International Modal Analysis Conference, pages 167181.

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):28372840.
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., Martn, J. M., Jimenez, A., Pos, J. L., Calder
on, 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, Department of Civil Engineering, Vancouver.
Stoll, R. D., 1979. Experimental studies of attenuation in sediments. Journal
of the Acoustical Society of America, 66(4):11521160.
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 265273. 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):137.
Thurston, R. N., 1992. Elastic Waves in Rods and optical Fibers. Journal of
Sound and Vibration, 159(3):441467.
Udas, A., 1999. Principles of Seismology. Cambridge University Press, Cambridge, 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 Belgie, pages 118.
Viggiani, G. & Atkinson, J. H., 1995. Interpretation of Bender Element Tests.
Geotechnique, 45(1):149154.

BIBLIOGRAPHY

255

Wang, Y. H., Yan, W. M. & Lo, K. F., 2004. Laboratory and in-situ measurement of attenuation in soil. In Viana da Fonseca & Mayne, editors, 2nd
Conference on Geotechnical and Geophysical Site Characterization, volume 2,
pages 18831889. 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):14.
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):11121126.
Zavoral, D. Z. & Campanella, R. G., 1994. Frequency effects on damping/modulus of cohesive soil. In R. J. Ebelhar, V. P. Drnevich & B. L.
Kutter, editors, Dynamic Geotechnical Testing II, ASTM STP 1213, pages
191201. ASTM, Philadelphia.
Zeghal, M. & Elgamal, A.-W., 1994. Analysis of site liquefaction using earthquake records. Journal of Geotechnical Engineering, 120(6):9961017.
Zeghal, M., Elgamal, A.-W., Tang, H. T. & Stepp, J. C., 1995. Lotung Downhole Array. II: Evaluation of Soil Nonlinear Properties. Journal of Geotechnical Engineering, 121(4):363377.