Application of Seismic Methods For Geotechnical Site Characterization

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

International Workshop in “Geoenvironment and 1

Geotechnics”, September 2005, Milos island, Greece

Application of seismic methods for geotechnical site characterization

P.M. Soupios
Technological Educational Institute of Crete, Department of Natural Resources & Environment,
Laboratory of Geophysics & Seismology, Hania, Greece

C.B. Papazachos, G. Vargemezis and I. Fikos


Geophysical Laboratory, Aristotle University of Thessaloniki, Macedonia, Greece

ABSTRACT maximum depth of 75 m, approximately located


20 m from the Nestos River. The objective of
There is an increasing requirement for geo-
the survey was to obtain the high-resolution
physical surveys carried out during geotechnical
compressional and shear wave velocities of the
investigations to provide direct information
shallow subsurface for computation of elastic
about rock quality or other geotechnical para-
engineering properties of the unconsolidated
meters. With the paucity of information to cor-
material interposed between the investigated
relate geophysical results with actual rock prop-
boreholes. Joint interpretation of borehole logs
erties, this is still difficult to achieve and addi-
and velocity images obtained by P- and S-wave
tional research effort needs to be performed to
traveltime inversions outlined shallow geologi-
address this issue. Nowadays, several research-
cal anomalies.
ers have been described detailed studies which
designed specifically to investigate this matter.
Generally, geophysical techniques which are 1. INTRODUCTION
used to estimate seismic velocities in the subsur-
face focus on the low strain levels that are not The seismic method is a powerful geophysical
large enough to induce significant nonlinear, exploration technique that has been in wide-
nonelastic stress-strain behavior (Luna and Jadi, spread use in ground engineering for more than
2000), which is normally evaluated when ad- 40 years and has been increasingly used since
dressing liquefaction potential of soils in the 1996 in geotechnical and environmental
shallow subsurface. Soil parameters relevant to applications, usually at depths shallower than 40
seismic ground response (Rechtien, 1996) are meters. The applicability of seismic methods
density, relative density and void ratio, perme- depends on the presence of acoustical contrasts
ability, shear modulus, water saturation, soil in the subsurface. In many cases the acoustical
fabric and stress history. Detailed characteriza- contrasts occur at boundaries between geologi-
tion of elastic properties near a major construc- cal layers, although man-made boundaries such
tion site (e.g. a bridge) is useful for near field as tunnels and mines also present contrasts.
dynamic analysis of soil-structure interaction Seismic survey is the geophysical method,
(SSI) effects. The coupled effect of the bridge which is most closely related to rock and soil
foundation (piles, piers or abutments) and the mass properties, since seismic wave velocity
soil immediately surrounding these substructure varies with the main mechanical properties,
elements is essential for dynamic SSI analysis such as Poisson’s ratios and others modules.
of these critical bridge structures. The earliest applications of the method primar-
This paper presents a geophysical investiga- ily concerned the determination of the depth to
tion performed along the under construction bedrock beneath a soil cover. Later, the same
highway of Egnatia Odos, approximately 13 km method was used successfully for the location of
west from the town of Xanthi in N. Greece. “weak” zones, such as shear zones and faults.
Twelve geotechnical boreholes, drilled at the Nowadays, seismic methods have been used in
connection with planning of dams (e.g. Klimis
2 International Workshop in “Geoenvironment and
Geotechnics”, September 2005, Milos island, Greece

et al., 1999), tunnels and portals.


The field measurements can be carried out on
the surface, in boreholes, or even on the seabed.
The necessity of a borehole controls the overall
geophysical cost which is increased unless bor-
ing is also needed for other geotechnical pur-
poses (CPT/SPT tests, etc). Recently, most sci-
entists prefer to apply Crosshole and Downhole
Seismics (CHS and DHS) tests, since it is highly
accurate method for determining material prop-
erties of rock and soil sites. Thin low-velocity
layers lying between high velocity layers can be
detected with this method, which may not be
possible with surface methods. In addition, the
accuracy and resolution of the CHS and DHS
methods is almost constant for all test depths,
whereas the accuracy and resolution of the sur-
face methods decreases with depth. A limitation
of these methods is to generate adequate energy Figure 1: Map of Greece showing the area of the founda-
without damaging the borehole casing. tion of Nestos Bridge. A geological map of the site with
the main formations is also presented.
The present work is a link in a series of ac-
tivities aiming at implementing geophysical me-
the ground surface (due to the adjacent Nestos
thods for geotechnical engineering. Specifically,
river) and the pore water pressure is hydrostatic
CHS and DHS measurements are performed in
from this level.
five selected places in the area of Nestos river in
For seismic tests 12 boreholes were con-
order to obtain information on dynamic soil and
structed using a water drilling machine. The
rock properties for earthquake design analyses
borehole setup is shown in Figure 2.
for bridge construction. Tests determine shear
In CH1 and CH3, three boreholes are used in
and compression wave velocity profiles versus
order to also obtain an estimation of the attenua-
depth and other crucial parameters such as Pois-
tion model with depth. The boreholes were 4.5
son’s ratios, Young and Rigidity modules.
inches in diameter, PVC cased and grouted ac-
Complementary, the attenuation model for two
cording the American Standards (ASTM
areas (where three boreholes are used) were also
D4428/D4428M-84) to ensure good transmis-
calculated applying spectral-ratio technique to
sion of the wave energy. The hole must usually
waveforms, as suggested by Sams and Goldberg
be cased and grouted to prevent the soil from
(1990) and Neep et al., (1996).
caving in during the testing. The distances be-
tween adjacent boreholes were of the order of
2. STUDY AREA & DATA ACQUISITION 3.5 meters.
The source and receiver boreholes are drilled
The investigation area is located in the eastern to the total depth of investigation. The source is
part of Greece, thirteen kilometers from the lowered to the measurements depth and one or
town of Xanthi. The main geological formations two receivers were lowered to the same depth in
of the area are shown in Figure 1. The general the others boreholes. To generate shear and
geology of the area consists of gravels, coarse compression wave energy, we used the BGS
sand, moraine (20+ meter thick) overlying semi- Cross hole shear-wave electromagnetic hammer
consolidated pleistocenic sediments consisting with inflated tube clamping system. To record
of cemented gravel and sand, occasionally with the generated pulse we used two similar tri-axial
big blocks of limestones. The basement consists GEOSTUFF geophones (BHG-2 model)
of fractured gneiss, gneiss schist and mica clamped to the borehole wall by means of a mo-
schist. Sedimentary layers become thicker and tor-driven bow-springs. The vertical component
more cohesive with depth. The free ground wa- of the receiver is used to capture the vertically
ter level is normally located 3.4 meters below
International Workshop in “Geoenvironment and 3
Geotechnics”, September 2005, Milos island, Greece

Figure 2: Topography map of the study area showing the location of the wells, presented with black circles. A total num-
ber of 12 boreholes were used for the experiment.
propagating shear waves (SV). The radial com- associated with particle motion perpendicular to
ponent senses the propagating compression the direction of propagation. Using this method
waves (P) and the tangential component senses we could distinguish the component which “re-
the horizontally propagating shear waves (SH). ceived” the S-wave motion by plotting compo-
The hammer input and the receiver outputs are nent pairs, as is shown in Figure 4, since the
recorded by a Geometrics seismograph (Strata- orientation of the horizontal components of the
View-24bit, 24 channels). At the same time, the recording geophones was unknown.
one of the geophones was also used to acquire
2.1 Preliminary data interpretation
DHS data set, when compression and shear
waves are generated in the surface by the use of When one receiver borehole is used, the travel
a sledgehammer with a triggering system. The time from source to the receiver is measured.
source and receivers were moved to the next This is referred to as direct travel time meas-
measurement depth and the process is continued urements. In the case that two receiver bore-
until all desired measurements were taken. holes are used, the travel time between the re-
We have used the SAC freeware interactive ceivers is measured, usually referred to as inter-
software (Seismic Analysis Code, developed by val travel time measurements. Note that interval
Lawrence Livermore Laboratory, University of travel times are normally more accurate than
California, 11/6/2000, Version 00.59.2) to pick direct travel times, since this method reduces
arrivals times. Picking of P arrivals is much eas- timing errors caused by differences in seismic
ier than S-arrival identification since it is always triggering, variations in source impulse charac-
the first wave usually sharply arriving at the teristics and errors arising from variations in
geophone. borehole size or mud-cake thickness near the
For the S-waves, the picking of the onset of transmitters. The wave velocities for compres-
shear-wave motion in the presence of source sion and shear waves at a specific depth are eas-
generated noise (later cycles of P-motion or tube ily determined by dividing the travel distances
waves) can sometimes be challenging. Two by the measured travel time. The travel dis-
methods of identification were applied: a) The tances were measured in the surface at the beg-
conventional method of overlapping waveforms ging of the survey, assuming that the boreholes
(from the “positive” and “negative” source po- were vertical.
larity records) for each of the two geophones
2.2 Tomographic interpretation of the Data
(usually vertical and a horizontal) is often ade-
quate to obtain a crossover onset of shear-wave We used an iterative, 3-D tomographic inver-
energy, as shown in Figure 3 and, b) alterna- sion routine to determine the velocity structure
tively we used the change of polarization direc- betweens the wells and the ground (Soupios
tion of the wave field using particle motion dia- et al., 2001). The forward routine calculates the
grams. In practice for the P-wave arrival a linear first arriving travel time using the revisited ray
particle motion along the direction of propaga- bending method, supplemented by an approxi-
tion is observed, whereas the S-wave arrival is mation of the first Fresnel zone at each point of
4 International Workshop in “Geoenvironment and
Geotechnics”, September 2005, Milos island, Greece

Figure 3: Detection of S-waves arrival using the conventional method of overlapping waveforms as recorded from hori-
zontals and vertical component. It is obviously that the vertical component, receive the S-wave energy better than the
horizontal components.

Figure 4: Picking of P and S-waves arrival applying the particle motion diagrams.

the ray, hence using physical and not only the updates the model. The program iterates until
mathematical rays (Fig. 5). Therefore, the algo- the root-mean-square travel time residual cannot
rithm computes the raypath, solves for the be further minimized.
slowness using either an LSQR routine (Paige We used a grid spacing of 2 meters in depth,
and Saunders, 1982) or an SVD algorithm and similar to the source and receiver spacing during
International Workshop in “Geoenvironment and 5
Geotechnics”, September 2005, Milos island, Greece

Figure 6: 1D velocity model for CH3 and its correlation


with geological formations identified from the borehole.
“down” records to increase the signal-to-noise
Figure 5: Ray coverage for the boreholes configuration in
CH3 location.
ratio before processing. Furthermore, we se-
lected various time windows, keeping always
data acquisition. The final velocity images (e.g. the time of shear wave arrival as the lower limit
Fig. 6) are obtained performing a constrained of the window gate. Figure 7A shows the vari-
least squares inversion scheme. ous windows (T1-T2, T1-T3, etc.) tested for the
2.3 Estimation of the attenuation model near offset geophone in both “up” and “down”
waveforms, as they were recorded from the ver-
The attenuation model was estimated using the tical component (SV waves) for borehole CH3
spectral ratio method as suggested by Neep at the depth of 10m, while the corresponding
et al. (1996). Specifically, the Q factor was cal- spectra for the near- and far-offset geophones
culated from the fit of a straight line to the spec- are presented in Figure 7B. It should be noted
tral ratio of receiver and source power spectra that the Q-estimates were practically stable for
recorded at different offset from the source. all time windows, as long as the main pulse of
The main assumptions of our approach are the S-waveform (window T1-T2 in Fig. 7A)
that: a) the receivers lie along a common ray- was used.
path from the source and, b) the two power The spectral ratio of the two spectra is pre-
spectra can be linearly related by a simple at- sented in Figure 8A, where a clear linear de-
tenuator operator. It is often necessary to apply crease is observed for frequencies between 150-
windowing to the first few circles of the shear 450Hz. Since the shear velocity for each depth
wave arrival before transform waveforms to the is known from the previous tomographic step,
frequency domain in order to avoid contamina- the Q estimation was performed by fitting a
tion from lower velocity arrivals. straight line to the slope of the spectral ratio in
In this work, we have overlapped waveforms the frequency range of 150-470 Hz. Further-
with different polarity and stacked “up” and more, in order to stabilize the estimation of the
6 International Workshop in “Geoenvironment and
Geotechnics”, September 2005, Milos island, Greece

Figure 8: A) Linear decreasing logarithmic spectral ratio


of the spectra presented in figure 7A. B) Average spectral
Figure 7: A) Time windows (T1-T2, T1-T3, etc.) selected ratio for the depth of 8 meters, as determined by the
for the S-wavetrains for the “up” and “down” source re- summation of spectral ratios for the depths of 4, 6, 8, 10
cords tested for the Q-estimation. B) Spectra of the far- and 12 meters. The linearity of the spectral slope is more
and near-offset geophones for borehole CH3 at the depth smooth and stable.
of 10m used for Q-estimation.

spectral slopes and improve the reliability of the


attenuation results, we also estimated cumula- 3. CONCLUSIONS
tive spectral ratios for overlapping depth inter- The use of engineering geophysics provided an
vals of 8 meters. The resulting spectral ratios are efficient way to derive a geologic cross section
more stable and linear, as is shown in the exam- (Fig. 9) to aid bridge engineers in their design of
ple of Figure 8B where the average spectral ra- the Nestos bridge foundation.
tio for the depth of 8 meters (5 cross-hole wave- The tomographic interpretation of the seismic
forms from the depths of 4-12 are used) when data provides an explanation of the lateral varia-
compared to Figure 8A. tions in geology and allows the detailed descrip-
Finally, for each borehole several additional tion of not only elastic but also anelastic depth
elastic moduli (Poisson ratio, Young modulus behavior, necessary for geotechnical site charac-
and rigidity values) were also estimated at all terization.
depths for which data were available, in order to The three-borehole approach allows not only
provide additional information to engineers for the reliable estimation of attenuation parameters
the calculation of the final dynamic response of (Q-factor) but also the more reliable estimation
the structure. A typical example of the finally of seismic velocities.
obtained results for a selected borehole (CH1) is
presented in the Table 1.
International Workshop in “Geoenvironment and 7
Geotechnics”, September 2005, Milos island, Greece

Table 1: Final Vp-Vs-QS model and variation of several REFERENCES


elastic moduli such as Poisson ratio Young modulus and
rigidity value for borehole CH-1, as well as Q (attenua- ASTM D4228/D4228M –91, Standard Test Methods for
tion) values for several depths either individually esti- Crosshole Seismic Testing, July 91, Committee D-18
mated [Q(AV)] or from stacked spectral ratios [Q(AV- on Soil and Rock.
sm)] for different depth ranges. Klimis, N.S., C.B. Papazachos and Ch.F. Efremidis, 1999.
Determination of the behavior of a sedimentary rock
mass: comparison of measured static and dynamic
properties, Proc. of the 9th Int. Congress on Rock Me-
chanics, Paris, France, August 25-28, 1999.
Luna, R. and H. Jadi, 2000. Determination of dynamic
soil properties using geophysical methods. Proceed-
ings of the First International Conference on the Ap-
plication of Geophysical and NDT Methodologies to
Transportation Facilities and Infrastructure—
Geophysics 2000. Federal Highway Administration,
Saint Louis, MO. Paper No. 3 –1, 15 pp.
Neep, J.P., M.S. Sams, M.H. Worthington and K.A.
O’Hara-Dhand, 1996. Measurement of seismic at-
tenuation from high-resolution crosshole data, Geo-
physics, Vol. 61, No. 4, pp 1175-1188.
New, B.M., 1985. An example of tomographic and fourier
microcomputer processing of seismic records, Quar-
terly Journal of Engineering Geology 18 (4): 335-344.
Paige, C.C. and M.A. Saunders, 1982. LSQR: An algo-
rithm for sparse linear equations and sparse least
squares, A.C.M. Trans. Math. Software, 8, 43-71.
Rechtien, R.D., 1996. In Situ Liquefaction Investigation

Figure 9: A cross section of the area from the combined interpretation of the borehole logs and geophysical measure-
ments is obtained. Three stromatographic units are found, a) a formation with poorly–graded gravels, gravel-sand mix-
tures and little or no fines (Vp=1705m/s, Vs=254m/s), b) a weathered and fractured basement consist of gneiss
(Vp=2737m/s, Vs=681m/s) and c) the rigid gneiss basement (Vp=3379m/s, Vs=1184m/s).
8 International Workshop in “Geoenvironment and
Geotechnics”, September 2005, Milos island, Greece

of Liquefaction Potential of Soils. Report GL-96-1-US


Army Corps of Engineer, Waterways Experiment Sta-
tion. 46 pp.
Sams, M. and D. Goldberg, 1990. The validity of Q esti-
mates from borehole data using spectral ratios, Geo-
physics, 55, No. 1, pp 97-101.
Soupios, P.M., C.B. Papazachos, C. Juhlin and G.N. Tso-
kas, 2001. Non-linear three dimensional travel-time
inversion of crosshole data with an application in the
area of middle Urals, Geophysics, 66, 627-636.

You might also like