analyses of the cross-hole method for determining shear wave velocities and damping ratios
TRANSCRIPT
Analyses of the cross-hole method for determining shear wave velocitiesand damping ratios
L. Hall*, A. Bodare
Department of Civil and Environmental Engineering, Royal Institute of Technology, S-100 44 Stockholm, Sweden
Abstract
The cross-hole test is a common method used for determining shear wave velocities and, hence, to characterize the shear modulus. The
determination of the shear wave velocities can be greatly improved by signal processing techniques. With further signal processing the
damping ratios can also be determined. In this paper, the signal processing techniques to determine shear wave velocity and damping ratio is
explained. The techniques are then applied to a actual cross-hole test, and also veri®ed in a ®nite element analysis. q 2000 Elsevier Science
Ltd. All rights reserved.
Keywords: Cross-hole test; Shear wave velocity; Damping ratio; Signal processing; Field measurements; Finite element analysis
1. Introduction
Great progress has been made in recent years in the devel-
opment of analytical procedures, such as the ®nite element
method, for evaluating the response of soils under dynamic
loading. As these analytical methods are re®ned, procedures
to provide more realistic parameters for de®ning stress and
strain properties for use in these models must also be
improved. In cases where no permanent soil displacements
occur, the response to a speci®c input vibration is primarily
controlled by the shear modulus and the damping ratios of the
soil through which the vibrations travel. These properties are,
however, non-linear at larger strains with a shape and orienta-
tion that can only be determined in laboratory tests. At small
strains, the stress±strain characteristics are nearly linear elas-
tic, and better estimates of the low-strain soil properties are
made in situ by ®eld tests. Linear elastic stress±strain models
can be used for analyses of wave propagation and vibration
problems that usually are associated with small strains.
The easiest and most accurate way of determining the
low-strain shear modulus in situ is the cross-hole method.
The cross-hole method involves generating a seismic wave
in one bore-hole and measuring the vibration response by
geophones in two or more adjacent bore-holes at the same
depth. The shear wave velocity, which is closely related to
the shear modulus, can then be estimated by looking at the
average time for the wave to travel between two receivers.
However, using signal processing techniques such as the
Fourier transform algorithm, the determination of shear
wave velocity from cross-hole tests can be enhanced.
Now, the damping ratio can also relatively easy be esti-
mated. Fourier transforms can determine the phase velocity
and material damping by calculating the phase difference
between the measurements of two geophones. The phase
velocity is considered by many seismologists [8] to be a
better estimator of the shear wave velocity.
In this paper, analyses of measurements from in situ
cross-hole tests are presented together with analyses of
results from simulation of cross-hole tests from axisym-
metric ®nite element models. In these analyses, the shear
wave velocity (both phase and average time estimation) and
damping ratios were determined. The ®nite element simula-
tions showed that the shear wave velocity and material
damping could be accurately estimated. The in situ
measurements were, however, very sensitive to noise and
other disturbances. The results can be greatly improved by
®rst re®ning the measured signals by using signal proces-
sing techniques such as ®ltering and windowing.
2. Digital signal analysis
The development of digital data acquisition system has
increased the possibility of using signal processing techni-
ques on the collected data. The most important improve-
ment is the possibility of signal analyses in the frequency
domain. The transformation of data from the time domain
into the frequency domain is based on the Fourier transform.
Soil Dynamics and Earthquake Engineering 20 (2000) 167±175
0267-7261/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.
PII: S0267-7261(00)00048-8
www.elsevier.com/locate/soildyn
* Corresponding author. Present address: FB Engineering AB, S-402 41
Gothenburg, Sweden.
E-mail address: [email protected] (L. Hall).
In the frequency domain, many useful parameters describ-
ing the measured signals as well as soil properties can be
determined.
A periodic sampled signal (e.g. velocity time history) can
be transformed into the frequency domain by the discrete
Fourier transform (DFT). For a variable x(tk), k � 0; N 2 1,
the discrete Fourier transform is given by:
X� fn� � DtSXN 2 1
k�0
�x�tk� cos �2pfntk�2 ix�tk� sin �2pfntk��
� DtSXN 2 1
k�0
x�tk�´e2i2pfntk for n � 0¼N 2 1 (2.1)
where tk � kDtS and fn � nDf � n=NDtS:
In general, the transform into the frequency domain will
be a complex valued function, which therefore is commonly
described with Fourier magnitude and phase angle. The
Fourier magnitude and phase angle from the Fourier trans-
form are determined as follows:
iX� fn�i �XN 2 1
k�0
�X�k�real´X�k�real 1 X�k�imag´X�k�imag�0:5; respectively;
�2:2�
fX� fn� �XN 2 1
k�0
tan21 X�k�imag
X�k�real
� �: �2:3�
A plot of the magnitude versus frequency is known as the
Fourier amplitude spectrum; and a plot of the phase angle
against frequency is called the Fourier phase spectrum. The
frequency resolution (Df ) in the spectra is obtained from the
number of samples N and the sampling time step DtS as
follow: Df � 1=T where T � NDtS:
The Fourier amplitude spectrum illustrates how a quan-
tity varies with the frequency. This information can also be
expressed in terms of power. The power spectrum also
called power spectral density function (PSD) is de®ned as:
SXX� fn� � Xp� fn�´X� fn�; �2:4�where Xp( f ) is the complex conjugate of X( f ).
Frequency analyses can also be used as an indication of
mutual frequencies from two records obtained simulta-
neously. This is done with the cross power spectrum (SXY)
that is the product of the linear spectrum of record 1 (i.e. X)
and the complex conjugate of the linear spectrum of record
2 (i.e. Yp) as:
SXY � fn� � Yp� fn�´X� fn�: �2:5�The cross power spectrum is a useful tool to determine
relative phase difference Df between two signals caused
by time delay, propagation delays and varying wave paths
between the receivers. The phase difference Df from the
cross power spectrum can be calculated as follows:
DfXY � fn� �XN 2 1
k�0
tan21 SXY �k�imag
SXY �k�real
� �; �2:6�
or just as the difference between phases of the two signals:
DfXY � fn� � fX� fn�2 fY � fn�k k: �2:7�The use of the phase difference is explained further in
Section 3.
3. Determination of material properties
When measuring two or more signals at different
distances, the dynamic material properties such as propaga-
tion velocities and material damping can be determined.
This can be done both in the time domain or in the frequency
domain. The most straightforward method to determine soil
properties is by analyzing the signals in the time domain.
However, if the signals are digitized it might be easier to
determine the signals in the frequency domain.
3.1. Time domain analyses
In time domain, the propagation velocity is determined
by looking at the arrival time of the waves and the known
distance between the receivers. Material damping can be
estimated by examining the amplitude decay between the
signals. The amplitude decay is, however, very dif®cult to
determine as the signals are corrupted by inhomogeneous
effects in the soils through which the ground motions travel.
3.2. Frequency domain analyses
In the frequency domain, the phase difference between
two signals from the cross power spectrum is used in order
to determine the propagation velocity and material damp-
ing. The phase difference is determined according to Eqs.
(2.6) or (2.7). A phase shift equal to 360 degrees (2 rad) is
equivalent to one period. As such, the travel time of the
waves with different frequencies can be determined by
tph� fn� � 1
fn
DfXY � fn�2p
� �: �3:1�
Now the propagation velocity can be determined by dividing
the distance between the receivers by the travel time obtained
from the phase difference of the two signals. As the deter-
mined propagation velocity depends on the frequency, it is
more commonly called the phase velocity. The phase velo-
city, Cph, for a given frequency can be calculated as
Cph� fn� � Dr
tph� fn� ; �3:2�
where Dr is the distance between the receivers �Dr �r2 2 r1�:
The coef®cient of attenuation, a , can be determined by
the ratio of the magnitude spectra between two signals and
L. Hall, A. Bodare / Soil Dynamics and Earthquake Engineering 20 (2000) 167±175168
the geometrical damping as follows [2]:
aph� fn� �ln
X� fn�´r1
Y� fn�´r2
� ��Dr� : �3:3�
The coef®cient of attenuation varies depending on soil type
and vibration frequency according to the following
equation:
a � 2pf´j
C; �3:4�
where f is the frequency, j the damping ratio and C the wave
propagation velocity
A more common way is to express material damping in
terms of the damping ratio. By using the phase velocity and
Eq. (3.4), the damping ratio becomes:
jph� fn� �ln
X� fn�´r1
Y� fn�´r2
� �´Cph� fn�
2pfn´Dr�
lnX� fn�´r1
Y� f �´r2
� �2pf´tph� fn� : �3:5�
It should, however, be noted that the phase velocity and
material damping are determined with the following
assumptions [7]: (1) the damping ratio is independent of
frequency and strain amplitude; (2) measured wave ampli-
tudes are not affected by re¯ected or refracted waves; (3)
any additional near ®eld effects on amplitude are negligible;
and (4) particle motions are accurately tracked at both
measurement points.
When analyzing the soil properties obtained (phase velo-
city and material damping) from the phase difference, the
power spectra are used to determine the frequencies in the
signals that are of poor quality. A schematic description of
determination of the phase velocity and material damping is
shown in Fig. 1.
4. Cross-hole measurements
The cross-hole measurements took place in LedsgaÊrd just
outside Kungsbacka, Sweden, in conjunction with a project
supervised by the Swedish Rail Administration (Banverket)
and with participants from Swedish Geotechnical Institute
(SGI), Norwegian Geotechnical Institute (NGI) and others.
The project was initiated because very large vibrations had
L. Hall, A. Bodare / Soil Dynamics and Earthquake Engineering 20 (2000) 167±175 169
Fig. 1. Schematic description for determining phase velocity and damping ratio by analysis of the signals in the frequency domain.
been measured in the embankment for trains passing at high
speeds. The soil conditions in LedsgaÊrd consisted of a very
deep (,75 m) and soft clay layer. In order to investigate the
cause of these large ground vibrations, Banverket performed
extensive measurements of the embankment's response due
to passing high-speed trains. Soil investigations of the clay
layer were also performed. For this reason, seismic tests (i.e.
cross-hole tests) were performed by KTH (Royal Institute of
Technology) in order to determine the dynamic soil proper-
ties. KTH also performed measurements on train-induced
ground vibrations on the ground surface just outside the
embankment, which were used for comparing results from
®nite element analyses of train-induced ground vibrations
[4].
Standard geotechnical soil investigations were also
performed [1,6] and soil samples were taken and routinely
investigated for soil type, density, water content, plastic
index, sensitivity, and shear strength as well as odometer
modulus. The seismic in situ tests by KTH were performed
the same week as the measurements of the train-induced
ground vibrations in the fall of 1997. The focus on these
measurements was on cross-hole tests, though down-hole
tests and fall weight tests also were performed. The cross-
hole method applied in LedsgaÊrd is described in the follow-
ing section.
4.1. Description of the cross-hole tests
The instruments used for the cross-hole and down-hole
measurements were developed at KTH [5]. This instrument,
shown in Fig. 2, consisted of a cone of solid steel containing
a geophone with a resonance frequency of 10 Hz.
In the cross-hole tests the cones (two or more) were
forced down by hand to the desired depth with aid of the
connecting steel tubes. This was possible as the ground
consisted of soft clay. The cones were then released from
the steel tubes by lifting up the steel tubes about 10 cm. In
addition, the cones were kept vertical by steel wires, which
were also used to retain the cones when all the measure-
ments were complete.
In order to induce a ground motion at the desired depth,
the same excitation equipment as used in Bodare [3] was
applied. This equipment consisted of a screw-plate
connected to a steel tube (see Fig. 2a). By rotating the
tube, the plate was lowered to the same depth as the cones
containing the geophones. Then, by releasing an inner rod
on the screw-plate it was the possible to induce the waves at
the desired depth.
L. Hall, A. Bodare / Soil Dynamics and Earthquake Engineering 20 (2000) 167±175170
Fig. 2. Cross section of the instruments used in the cross-hole measure-
ments in LedsgaÊrd. (a) Excitator±Screwplate on which an inner rod was
released. (b) Receiver±cone containing a geophone.
Fig. 3. Cross-hole setup used in LedsgaÊrd.
The signals measured by the geophones were sampled
digitally with a sampling frequency of 1000 Hz to a compu-
ter, where they were stored for future analyses. This enabled
the results also to be viewed immediately after the measure-
ments were performed. If the results were satisfactory, the
cones and the screw-plate were lowered to the next depth to
be evaluated. The applied cross-hole setup is shown in
Fig. 3.
L. Hall, A. Bodare / Soil Dynamics and Earthquake Engineering 20 (2000) 167±175 171
Fig. 4. Results from a cross-hole test at a depth of 3 m below the ground surface (v2� vertical particle velocity). #1, #2 and #3 refer to location of geophones
as numbered in Fig. 3. The estimated arrival times for the P-wave (Cp) and S-wave (Cs) are also shown in the ®gure.
Fig. 5. Results from signal analyses of windowed signals as shown in Fig. 4. (a) Normalized power spectra (PSD) of signal #1, #2 and #3. (b) Phase velocity
(Cph). (c) Damping ratio (j) obtained from the phase difference between signals #1±#2, #1±#3 and #2±#3.
4.2. Analyses of cross-hole measurements
Results from a cross-hole test are shown in Fig. 4. As seen
in Fig. 5, three cones with geophones were used in the tests.
Totally, cross-hole tests were performed at ®ve different
elevations from 2 to 6 m below the ground surface.
When analyzing the cross-hole tests, the signals were ®rst
improved by baseline correction and noise removals by a
lowpass ®lter (100 Hz). The S-wave propagation velocity
was then estimated by studying the time of arrival of the
geophones and using the known spacing of the cones. This
was, however, a very subjective and cumbersome process.
The S-wave propagation velocity was therefore also esti-
mated from the phase velocity in the frequency domain by
®rst determining the phase difference between the signals.
When determining the phase velocity, it was found that the
result could be improved if the signals were windowed (see
windows in Fig. 4) before the Fourier transformation into
the frequency domain. After windowing had been
performed, it was also possible to estimate the damping
ratio. Fig. 5 shows the normalized power spectrum, phase
velocity and damping ratio estimated from windowed
signals of a cross-hole test. The theory for estimating mate-
rial properties in the frequency domain is given in Section 5.
In Fig. 6, the results from all the cross-hole measurements
and other seismic ®eld and laboratory tests performed in
LedsgaÊrd are summarized. Cross-hole tests were available
from 2 to 6 m. At 6 m below the ground surface, the shear
wave velocity was correlated with the shear strength
obtained from laboratory tests (see Fig. 6(a)). The ratio
between the maximum shear modulus and the undrained
shear strength was estimated to be 335. This also agrees
with the values obtained by others [9] considering that the
plasticity index PI for the clay in LedsgaÊrd was very high
(PI < 139, [6]). The shear strength was measured down to
15 m below the ground surface. The shear wave velocity in
the crust layer was estimated from fall weight tests. In the
mud layer, the results from the cross-hole measurements
were con®rmed with the results from down-holes tests and
shear wave velocity measurements with bender elements.
The damping ratios were obtained from cross-hole tests
and dynamic triaxial tests. As seen in Fig. 6b, the low strain
measurements with the dynamic triaxial tests con®rms the
results from the cross-hole measurements. The obtained
damping ratio seemed to be around 4.8 in the layers of
mud and around 2.5% in the layers of clay.
5. Simulation of a cross-hole test in a ®nite elementmodel
In order to validate the appropriateness of the determina-
tion of the soil properties, and especially the damping ratio,
from cross-hole tests, a cross-hole test was simulated in a
®nite element analysis. The dynamic ®nite element program
ABAQUS was used in the analysis using direct time inte-
gration in the time domain.
5.1. Finite element model and analysis
The ®nite element model used was 16 £ 15 m2 and
consisted of axisymmetric elements. The elements had the
dimensions 0.1±0.3 m, with a smaller element closer to the
source. The left boundary was locked in the horizontal
direction simulating the symmetry line. The right and
bottom boundaries consisted of in®nite elements (absorbing
boundaries) simulating in®nite regions. The ®nite element
model is shown in Fig. 7(a).
In this analysis, a cross-hole test at 6 m below the ground
L. Hall, A. Bodare / Soil Dynamics and Earthquake Engineering 20 (2000) 167±175172
Fig. 6. Summary of results from soil investigation in LedsgaÊrd concerning: (a) shear wave velocity and (b) damping ratio.
surface was simulated. The model was given material prop-
erties that would give S-wave and P-wave propagation velo-
cities of 56 and 1050 m/s, respectively. The material models
were linear elastic and de®ned by the Young's modulus and
Poisson's ratio. The density was set to 1450 kg/m3 and the
damping ratio of approximately 3% was chosen. In direct
time integration analyses, the material damping is modeled
using the Rayleigh damping model. The damping ratio for
the Rayleigh damping model is, however, not frequency
independent as the actual material damping. The parameters
L. Hall, A. Bodare / Soil Dynamics and Earthquake Engineering 20 (2000) 167±175 173
Fig. 7. (a) Axisymmetric ®nite element model used in the cross-hole test simulation. The ®gure shows the deformed mesh (magni®ed with a factor of 20 000)
at time 60 ms after initiation of load. Numbers in mesh (#1, #2 and #3) indicate location of nodes simulating geophones. (b) Vertical load applied in the ®nite
element analysis. Arrow shows location of load. (c) The damping ratio used in the ®nite element analysis.
Fig. 8. Results from ®nite element simulation of cross-hole tests of uniform soil (v2� vertical particle velocity), #1, #2 and #3 refer to location in ®nite
element mesh as shown in Fig. 7. The arrival times with corresponding P-wave (Cp) and S-wave (Cs) are also shown in the ®gure.
of the Rayleigh damping model were therefore chosen so
that the damping ratios would be relatively constant over the
frequency span of interest. The applied Rayleigh damping
model used in the analysis is shown in Fig. 7(c).
The source, modeled as half-sinus pulse force with a
duration of 20 ms in the vertical direction was applied in
the symmetry line (see Fig. 7(b)). This was found to agree
with the source used in the cross-hole tests in LedsgaÊrd. The
®nite element analysis was performed with a time step of
0.001 s. The vertical particle velocity was then obtained: 3,
6 and 9 m from the source (see numbering #1, #2 and #3 in
Fig. 7(a)). The results from the ®nite element analysis were
analyzed in the same manner as the actual cross-hole
measurements.
5.2. Analyses of results from ®nite element analysis
The particle velocities obtained from the ®nite element
analysis simulating the cross-hole test are shown in Fig. 8.
Windows were also applied here to minimize the effect of
the P-wave and re¯ection of waves from the ground surface
(see windows in Fig. 8). From the windowed signals, the
phase velocity and damping ratio were calculated according
to theory described in Section 3. The results, shown in Fig.
9, were also compared with the true values of the shear wave
velocity and damping ratio as given in the ®nite element
model. This indicated that the phase velocity and damping
ratio could be estimated using the phase difference from the
cross-power spectrum. The phase velocity seemed to give a
slightly dispersive (the velocity varies with frequency)
behavior despite the homogenous ªsoilº (see Fig. 9(b)).
The dispersive behavior is probably caused by some numer-
ical errors when calculating the phase difference. Neverthe-
less, the agreement with the true shear wave velocity was
very good. The damping ratio estimation also agreed very
well with the true damping ratio (see Fig. 9(c)).
A ®nite element analysis was also performed with layered
ªsoilº with an increasing shear wave velocity pro®le with
depth. In the result from this analysis, the damping ratio was
shown to be very sensitive to re¯ection and refraction of
waves. Small windows were required in order to obtain
reasonably good results. The phase velocity did not seem
as sensitive to re¯ection and refraction of waves. Here
again, the results improved if windows were applied in the
time domain before the phase differences were calculated.
6. Conclusions
The cross-hole method using cones of solid steel contain-
ing geophones was shown to be a simple in situ technique to
determine low strain soil properties such as the shear wave
velocity and the damping ratio in soft soils. The determina-
tion of the soil properties was improved by calculations of
the phase differences between the measured signals. From
the phase difference, shear wave velocity (phase velocity)
and damping ratio could be estimated. The method of deter-
mining soil properties from the phase difference of the
L. Hall, A. Bodare / Soil Dynamics and Earthquake Engineering 20 (2000) 167±175174
Fig. 9. Results from signal analyses of the windowed signals as shown in Fig. 8. (a) Normalized power spectra (PSD) of signals #1, #2 and #3. (b) Phase
velocity (Cph), (c) Damping ratio (j) obtained from the phase difference between signals #1±#2, #1±#3 and #2±#3.
signals was veri®ed in a ®nite element analysis. The ®nite
element analyses showed that the true shear wave velocity
and damping ratio could be obtained with this method. The
analyses showed that the results could be improved, espe-
cially for the damping ratio, if windows were applied to the
signals in the time domain before the phase differences were
calculated. The results from actual ®eld tests with the cross-
hole method were also con®rmed with other ®eld and
laboratory tests.
Acknowledgements
This work was supported by the Swedish Railway
Administration (Banverket) and the Swedish Council for
Building Research (ByggfoskningraÊdet) under Grant #93-
0768-9.
References
[1] Andreasson B. Geotechnical investigation in LedsgaÊrd, Appendix 1. In:
Bengtsson PE, editor. High speed lines on soft ground, evaluation and
analyses from the West Coast Line, Dnr 2-9710-502, Swedish Railway
Administration, BorlaÈnge, Sweden, 1999.
[2] Barkan DD. Dynamics of bases and foundations. New York: McGraw
Hill, 1962.
[3] Bodare A. Dynamic screw plate for determination of soil modulus in
situ. PhD dissertation. Institute of Technology, Uppsala University,
Uppsala, Sweden, 1983.
[4] Hall L. Simulations and analyses of train induced ground vibrations; a
comparative study of two- and three-dimensional calculations with
actual measurements. PhD dissertation. Department of Civil and Envir-
onmental Engineering, Royal Institute of Technology, Stockholm,
Sweden, 2000.
[5] Ljunberg R, SoÈrensen N. Seismic ®eld measurements in soft clay at
Grums, Sweden. Masters thesis. Department of Civil and Environmen-
tal Engineering, Royal Institute of Technology, Stockholm, Sweden,
1997.
[6] Madshus C, HaÊrvik L. Laboratory test results. LedsgaÊrd and Peppared,
Appendix 5. In: Bengtsson PE, editor. High speed lines on soft ground,
evaluation and analyses from the West Coast Line, Dnr 2-9710-502,
Swedish Railway Administration, BorlaÈnge, Sweden, 1999.
[7] Mok YJ, Sanches-Salinero I, Stokoe KH, Roesset JM. In Situ damping
measurements by cross-hole seismic method. Earthquake Engineering
and Soil Dynamics II. ASCE Specialty Conference, Park City, Utah,
US, 1988. p. 305±20.
[8] Sahay SK, Kline RA, Mignogna R. Phase and group velocity consid-
erations for dynamic modulus measurements in anisotropic media.
Ultrasonics 1992;30(6):373±6.
[9] Weiler WA. Proceedings, ASCE Conference on Earthquake Engineer-
ing and Soil Dynamics II: Recent Advances in Ground Motion Evalua-
tion, 20. 1988. p. 331±5 (ASCE Geotechnical Special Publication).
L. Hall, A. Bodare / Soil Dynamics and Earthquake Engineering 20 (2000) 167±175 175