You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228715935

Seismic velocities and Poisson's ratio of shallow unconsolidated sands

Article  in  Geophysics · March 2000


DOI: 10.1190/1.1444751

CITATIONS READS
152 4,286

3 authors:

Ran Bachrach Jack Dvorkin


Schlumberger Limited King Fahd University of Petroleum and Minerals
157 PUBLICATIONS   1,470 CITATIONS    191 PUBLICATIONS   12,609 CITATIONS   

SEE PROFILE SEE PROFILE

Amos Nur
Stanford University
183 PUBLICATIONS   14,807 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Hi, Natela. View project

Material and Stress Rotations: The Key to Reconciling Crustal Faulting Complexity with Rock Mechanics View project

All content following this page was uploaded by Ran Bachrach on 23 May 2014.

The user has requested enhancement of the downloaded file.


GEOPHYSICS, VOL. 65, NO. 2 (MARCH-APRIL 2000); P. 559–564, 8 FIGS.

Seismic velocities and Poisson’s ratio of shallow


unconsolidated sands

Ran Bachrach∗ , Jack Dvorkin‡ , and Amos M. Nur‡

discrepancy between measured and theoretical values of wave


ABSTRACT velocities and the Poisson’s ratio remains puzzling.
We determined P- and S-wave velocity depth pro- In this study, we are concerned with the acoustic properties
files in shallow, unconsolidated beach sand by analyzing of unconsolidated sands at very low pressure (below 0.1 MPa or
three-component surface seismic data. P- and S-wave 15 psi), which corresponds either to a depth of just a few feet be-
velocity profiles were calculated from traveltime mea- low the surface or to deeper but highly overpressured soft un-
surements of vertical and tangential component seismo- consolidated sands. Because of transducer-sand coupling prob-
grams, respectively. lems, laboratory measurements at these low pressures are hard
The results reveal two discrepancies between theory to conduct and in-situ measurements are very few.
and data. Whereas both velocities were found to be pro- Our approach was to conduct a three-component surface
portional to the pressure raised to the power of 1/6, as seismic experiment on a beach with an array of closely spaced
predicted by the Hertz-Mindlin contact theory, the ac- (0.3-m apart) geophones. The results enabled us to determine
tual values of the velocities are less than half of those seismic velocities versus depth at a very shallow depth. Our
calculated from this theory. We attribute this discrep- goal was to compare the results with the existing theory and, if
ancy to the angularity of the sand grains. Assuming that necessary, find a theoretical model to match these results.
the average radii of curvature at the grain contacts are Theoretical models for the elasticity of granular material
smaller than the average radii of the grains, we mod- usually involve the calculation of normal and shear contact
ify the Hertz-Mindlin theory accordingly. We found that stiffnesses of two adjacent grains which are then statistically
the ratio of the contact radius to the grain radius is about averaged to calculate effective moduli for a pack of grains. The
0.086. classical Hertz-Mindlin model (Mindlin, 1949) relates in addi-
The second disparity is between the observed tion the contact stiffnesses of two elastic spheres to confining
Poisson’s ratio of 0.15 and the theoretical value (0.008 for pressure via their radii and elastic moduli. This model, together
random pack of quartz spheres). This discrepancy can be with statistical averaging (e.g., Walton, 1987), fairly accurately
reconciled by assuming slip at the grain contacts. Because reflects experimental observations of velocity versus pressure
slip decreases the shearing between grains, Poisson’s ra- (Figure 1a) (A. Tutunco, Stanford Rock Physics Project, pri-
tio increases. vate communication, 1996).
Recently we (Bachrach et al., 1998) showed that the Hertz-
Mindlin model successfully reproduced the power-law depth
INTRODUCTION
dependence (depth to the power of 1/6) of the P-wave velocity
in shallow sand in situ. However, this model overestimates by
Understanding the acoustic properties of unconsolidated a factor of two the velocity values actually measured in beach
granular materials is important for interpreting sonic and seis- sand. This was found also for both V p and Vs in the experiments
mic measurements in sediments and especially in soils and presented here.
related geological environments. Although many studies of A second discrepancy between model and experimental
elastic-wave velocities in such materials have been conducted data is Poisson’s ratio (Figure 1b). The measured Poisson’s
in the lab and in situ, and theoretical models have been devel- ratio in glass beads does not match the one derived by the-
oped (see overviews in White, 1983; Wang and Nur, 1992), the ory (theoretical value of 0.033 versus measured value of about

Manuscript received by the Editor November 2, 1998; revised manuscript received May 26, 1999.

Formerly Stanford University, Department of Geophysics, Mitchell Building, Stanford, California 94305-2215; currently Michigan State University,
Department of Geological Sciences, 206 Natural Science Bldg., East Lansing, Michigan 48824. E-mail: bachrach@gaea.glg.msu.edu.
‡Stanford University, Department of Geophysics, Mitchell Building, Stanford, California 94305-2215. E-mail: jack@pangea.stanford.edu; nur@
pangea.stanford.edu.
c 2000 Society of Exploration Geophysicists. All rights reserved.

559
560 Bachrach et al.

0.15). Poisson’s ratio values measured here and by others (e.g., 1990):
Spencer, 1994) in dry quartz sand typically are between 0.115
and 0.237. The Hertz-Mindlin model, however, predicts an 4π Rg2 P
effective value of about 0.008 for random pack of identical F= . (5)
n(1 − φ)
quartz spheres. Manificat and Gueguen (1998) argued that this
disparity may exist due to the weakening of the grain contacts
when roughness of the grain contacts is taken into account. In For a pack of perfect spheres (e.g., glass beads), Rc = Rg (e.g.,
this paper, we try to find a simple model which will predict the Dvorkin and Nur, 1996). However, for angular grains, the con-
elastic behavior of loose beach sand. tact radius Rc may be significantly smaller than the grain radius
Rg (Figure 2). The mechanics of this effect may be especially
THEORY pronounced at very low pressures in soft unconsolidated sands.
The above equations thus relate the effective elastic moduli
Walton (1987) shows that for a dry, dense, random pack of of a sand to the effective confining pressure (the difference
identical elastic spheres, the effective bulk (K H M ) and shear between the overburden and the pore pressure), porosity, and
(G H M ) moduli are the elastic moduli of the grains.
n(1 − φ) n(1 − φ) The effective moduli of dry sand versus depth Z can be cal-
KHM = Sn , GHM = (Sn + 1.5St ), culated by assuming that the effective pressure is given by
12π Rg 20π Rg
(1)
where Sn and St are the normal and shear stiffness of a two- P = ρb g Z , (6)
grain contact, respectively; n is the average number of con-
tacts per grain; φ is porosity; and Rg is the grain radius. Equa- where ρb is bulk density and g is gravitational acceleration.
tions (1) can be used with various contact models for the Sn For the Hertz-Mindlin model, the effective Poisson’s ratio
and St dependence. ν H M can be expressed using effective bulk (K H M ) and shear
The Hertz-Mindlin model yields (Mavko et al., 1998)
4aG 8aG
Sn = , St = , (2)
1−ν 2−ν
where a is the radius of the contact area between two grains,
and G and ν are the grain shear modulus and Poisson’s ratio,
respectively. The radius a is related to the confining force F
and the contact curvature radius Rc :
 1
3F Rc (1 − ν) 3
a= , (3)
8G
where Rc is related to the local radii of curvature of the two
grains, R1 and R2 :
 
Rc−1 = 0.5 R1−1 + R2−1 . (4)
The confining force F in equation (3) can be expressed through FIG. 2. Angular grains: the contact radius Rc defined by equa-
hydrostatic effective pressure P as follows (e.g., Marion, tion (4) compared to Rg .

FIG. 1. Compressional and shear-wave velocities (a) and Poisson’s ratio (b) versus confining pressure in a dry,
dense, random pack of identical glass beads versus confining pressure (from A. Tutuncu, Stanford Rock Physics
Project, private communication, 1996). Circles are measured values; lines represent Hertz-Mindlin theory.
Seismic Properties of Shallow Sands 561

(G H M ) moduli: The compressional and shear-wave velocities V p and Vs are


given by
3K H M − 2G H M Sn − St ν  
νH M = = = . (7) K ∗ + 43 G H M GHM
2(3K H M + G H M ) 4Sn + St 10 − 6ν Vp = , Vs = , (9)
ρb ρb
The frame Poisson’s ratio is thus a function only of the grain
where ρb is the bulk density.
Poisson’s ratio. For quartz, ν = 0.08; ν H M is therefore 0.0084.
For grains with ν of 0.35, the frame will have effective ν H M of EXPERIMENT
0.04.
One basic assumption in the Hertz-Mindlin model is that We collected the compressional and shear-wave data at Moss
there is no slip between the grains (except at a small region at Landing beach, Monterey Bay, California. We used three-
the periphery of the contact which, under acoustic excitation, component Galperin-mount 40-Hz geophones (Steeples et al.,
is negligibly small). 1995) and a 60-channel acquisition system. The line included
When the tangential stiffness between the grains is negligible 20 three-component receiver locations. Station intervals were
(St = 0; e.g., in loose sand with water-lubricated grain contacts), 0.3 m, and the seismic line was parallel to the shore. As source,
equation (7) implies that the effective Poisson’s ratio is 0.25 we used a hammer applied to a metal block with two fins that
(Walton, 1987). were inserted into the ground. We stacked six hammer blows
Consider now sand that is a mixture of (1) a material with per direction in the vertical, seaward and landward direction
St = 0; and (2) a material that obeys the Hertz-Mindlin model. each. To generate adequate vertical and tangential (SH), we
For this mixture, we now calculate the effective elastic mod- applied vertical pressure of about 50 kg to the block.
uli as a function of their volumetric fractions by using the To improve the shear wave signal-to-noise ratio, we neg-
Hashin-Shtrikman (Hashin and Shtrikman, 1963; Mavko et al., atively stacked the seaward and landward shot gathers. This
1998) upper and lower bounds, and then the mixture’s effective technique, often used in land surveys, helps cancel monopole
Poisson’s ratio (Figure 3). It is apparent that by allowing zero radiation and enhances the dipole radiation of the shear com-
tangential stiffness in a fraction of the granular aggregate, we ponent of the source. After negatively stacking the shots, we
can match the high experimental Poisson’s ratio value. If this is rotated the recorded seismogram to get the tangential and ver-
true, a measured Poisson’s ratio is an indicator of the fraction tical components.
of grains in a pack with no tangential stiffness. Figure 4 shows (4a) a raw shot gather, (4b) a vertical compo-
The air compressibility is accounted for when calculating the nent shot gather, and (4c) a tangential component shot gather
dry sand bulk modulus K ∗ with Gassman’s (Gassman, 1951) of our data. The tangential seismogram shows a curved shear-
equation: wave arrival and a slower, high-amplitude Love wave train.
The S-wave travel time was obtained from the tangential
K∗ KHM K Air

= + , (8) seismogram by picking the peak of the wavelet at every trace.
Kg − K Kg − K H M φ(K g − K Air ) To correct for first break, we subtracted the quarter-cycle trav-
eltime (6.55 ms) from the picked traveltime values. The P-wave
where K Air and K g are the bulk moduli of air and grain, re- traveltime was derived directly from the first breaks at each
spectively. The shear modulus (G H M ) is not affected by pore trace shown in Figure 4.
fluid. To calculate the V p and Vs velocities versus depth, we as-
sumed that they depend only on depth and that they obey the
Hertz-Mindlin depth dependence:
1 1
Vs (Z ) = V0 + αs Z 6 , V pVacuum (Z ) = α p Z 6 , (10)

where αs , α p , and V0 are adjustable constants, and V pVacuum is


the P-wave velocity in the vacuum-saturated sand. To obtain
V p (Z ) for sand with a pore fluid (such as air), we insert V p from
equation (10) into equations (8) and (9).
The constants in equation (10) were obtained by match-
ing the measured traveltime (T ) versus offset (X ) data to the
velocity-depth profile using ray theory (Jakosky, 1950):
 Hmax
pV (Z ) d Z
X =2  ,
0 1 − p 2 V 2 (Z )
 Hmax
dZ
T =2  , (11)
0 V (Z ) 1 − p 2 V 2 (Z )
where p = 1/V (Hmax ) = sin θ0 /V (Z = 0). For details see
FIG. 3. The Poisson’s ratio of a quartz grain pack versus the Bachrach et al. (1998.) Note that the velocity-depth functional
volumetric fraction of the contacts that have no tangential stiff- forms, which accurately match the data, represent the actual
ness. velocity versus depth of the sand in situ (Figure 5).
562 Bachrach et al.

From the V p and Vs values versus depth, we calculate The effect of grain angularity on velocity
Poisson’s ratio versus depth. We find that the calculated
Poisson’s ratio is depth independent and equals approximately In Figure 6, we compare the observed versus computed ve-
0.15 ± 0.03. locity profiles V p and Vs versus depth, (velocity calculated from
Hertz-Mindlin theory). For these calculations, we assumed
quartz sand grains (K = 36.6 GPa and G = 45 Gpa), porosity of
ANALYSIS OF THE RESULTS 0.4 (as measured), and a 50/50 mixture of slipping and nonslip-
ping grains. The number of contacts per grain n ranged from
Poisson’s ratio
four to eight. This yields a computed Poisson’s ratio of 0.15 for
As described above, Poisson’s ratio νeff can be used to es- the aggregate, close to the value measured in the field. The the-
timate the fraction of loose contacts in a sediment pack. As oretical P- and S-wave velocities are significantly larger than
shown in Figure 3, the experimentally derived Poisson’s ratio the measured values. We explain the discrepancy between the
of 0.15 can be modeled as a random pack of spheres, where half measured and theoretical V p and Vs values by assuming that
of its volume fraction has zero tangential stiffness at grain con- the contact curvature radius Rc is smaller (due to angularity)
tacts and half has zero slip at the grain contacts. Theoretically, than the average grain radius Rg . We use equations (1–9) to
such a sand will have a Poisson’s ratio of 0.15 at any depth or calculate the ratio of Rc /Rg that matched our P-wave data.
pressure, and for any coordination number. This is remarkably We find that for n = 5 (appropriate for loose sand), this ratio is
consistent with another observation (Figure 1). 0.086.
Consistency of the model is checked as follows. We used the
volumetric fraction of the grains with no tangential stiffness and
the Rc -to-Rg ratio thus defined to calculate Vs directly from the
Hertz-Mindlin theory. The Vs values thus calculated are very
close to the measured values (Figure 6).

Actual grain contacts


As we have shown, the observed Poisson’s ratio can be mod-
eled by a 50/50 mixture of slipping and nonslipping grains, and
the observed velocity profile can be derived from the grain
elastic moduli when angularity is taken into account. Because
the Poisson ratio estimation is not dependent on the angularity,
it can be used to estimate the grain-contact strength directly
from Figure 3 for any Hertzian material. The condition of zero
tangential stiffness can be physically caused by liquid at the
grain contacts.
Using the grain shapes and Powers’s visual and verbal
roundness scale (Boggs, 1987), we have optically determined
the beach sand to be angular (Figure 7). Thus, the average

FIG. 4. Seismic shot gathers from the beach: (a) 3-component


60-channel seismogram, (b) vertical and tangential compo-
nents, (c) Interpretation. The shear wave on the tangential
seismogram is slower then the P-wave. The residual energy
in the tangential seismogram before the shear-wave arrival is FIG. 5. Traveltime versus offset for both P- and S-waves.
residual P-wave energy. Markers are measured data; line is the best fit for α P and α S .
Seismic Properties of Shallow Sands 563

angularity of our sand using the Wadell class interval classi- for the density effect. The results in Figure 8 show that the
fication (Boggs, 1987) is RW ≈ 0.17–0.25, where RW is given S-wave velocity profile derived from our beach measurements
by is similar to the one measured by Hunter (1998) in a different
 depositional environment.
Rc
RW = . (12)
N Rg

We derived a lower angularity parameter of 0.086 from our


data. This discrepancy is expected, since our derived angular-
ity represents a mechanical average (i.e., stiffness average),
whereas the one derived from equation (12) is a simple mean.
Our smaller angularity represents a more compliant structure,
which is likely to be affected by the weaker part of the system
(i.e., sharper grain contacts).

COMPARISON TO OTHER MEASUREMENTS

We compare our model prediction to a shear-wave profile


measured by Hunter (1998) in shallow unconsolidated sand in
the Fraser River delta, Vancouver, British Columbia Canada.
This velocity profile is measured by a downhole log and by FIG. 7. Microscopic image of Moss Landing beach sand. This
seismic cone penetrometer measurements (SCPT). Since the sand can be classified as angular, with Waddell class interval of
sediments are saturated, we use Gassman’s equation to correct 0.17–0.25 (Boggs, 1987).

FIG. 6. (a and b): Comparison of velocity versus depth for contact model of different coordination number in comparison with the
beach field data. The calculated values are for a 50/50 mixture of slipping and none slipping grains. (c and d) V p and Vs as calculated
from a model with Rc = 0.086Rg and mixture of slipping and nonslipping grains. The model fit to V p agrees with Vs .
564 Bachrach et al.

diagnosing sediment properties (e.g., the fraction of contacts


without tangential stiffness and the angularity of grains).

ACKNOWLEDGMENTS

We thank Don Steeples (University of Kansas) for the use


of the 3-component Galperin geophones. Thanks also to Ranie
Lynds and Erminia Millia-Zarb for their help throughout the
experiment. We also thank the reviewers of this paper for their
constructive comments. This project was supported by DOE
grant DE-FG07-96ER14723 and the Stanford Rock Physics
Project.

REFERENCES
Bachrach, R., Dvorkin, J., and Nur, A., 1998, High-resolution seismic
experiments in sands, Part II: Velocities in shallow unconsolidated
sand: Geophysics, 63, 1234–1240.
Boggs, J. S., 1987, Principles of sedimentology and stratigraphy: Merrill
Publ. Co.
Dvorkin, J. P., and Nur, A., 1996, Elasticity of high-porosity sand-
FIG. 8. Shear wave velocity in unconsolidated sands from stones: Theory for two North Sea datasets: Geophysics, 61, 1363–
Fraser River delta (data after Hunter, 1998) and shear wave 1370.
velocity in Moss Landing beach sands. The velocity profiles Gassmann, F., 1951, Uber di elastizitat poroser median: Vier. der Natur
are quite similar. Gesellschaft in Zurich, 96, 1–23.
Hunter, J., 1998, Shear wave measurements for earthquake hazards
DISCUSSION AND CONCLUSIONS studies, Fraser river delta, British Colombia: Proceedings of the sym-
posium on the application of geophysics to environmental and engi-
neering problems (SAGEEP 98), Environmental and Eng. Geophys.
We have shown in this paper the following: Soc., 459–469.
Hashin, S., and Shtrikman, S., 1963, A variational approach to the
1) The in-situ Poisson’s ratio of beach sand is 0.15. This value elastic behavior of multiphase materials: J. Mech. Phys. Solids, 11,
127–140.
can be modeled if we assume that in the granular aggre- Jakosky, J. J., 1950, Exploration Geophysics: Trija Publ. Co.
gate 50% of grains have zero tangential contact stiffness Manificat, G., and Gueguen, Y., 1998, What does control V p /Vs in gran-
and the other 50% of grains have zero tangential slip at ular rocks: Geophys. Res. Lett., 25, 381–384.
Mindlin, R. D., 1949, Compliance of elastic bodies in contact: J. Appl.
their contacts. Mech., 16, 259–268.
2) The measured shear-wave velocity is lower than the the- Marion, D., 1990, Acoustical, mechanical, and transport proper-
oretical values for round grains. This can be explained ties of sediments and granular materials: Ph.D. thesis, Stanford
Univ.
by the angularity of the grains. The difference between Mavko, G., Mukerji, T., and Dvorkin, J., 1998, The rock physics hand-
the computed and the measured velocities can be used to book: Cambridge Univ. Press.
Spencer, J. W., 1994, Frame moduli of unconsolidated sands and sand-
estimate grain angularity. stones: Geophysics, 59, 1352–1361.
3) The dependence of the S-wave velocity on the depth Steeples, D. W., Macy, B., and Schmeissner, C., 1995, S-waves and
in the site examined here is similar to that observed 3-component seismology: Soc. Expl. Geophys. Near-surface Seis-
mology short course.
in other depositional environments, such as the Frasier Walton, K., 1987, The effective elastic moduli of a random pack of
River delta in British Colombia. spheres: J. Mech. Phys. Sol., 35, 213–226.
Wang, Z., and Nur, A., 1992, Seismic and acoustic velocities in reservoir
rocks. Volume 2, Theoretical and model studies: Soc. Expl. Geophys.
We suggest that the theoretical models presented here can White, J. E., 1983, Underground sound: Application of seismic waves:
be used not only for matching experimental data but also for Elsevier.

View publication stats

You might also like