You are on page 1of 12

GEOPHYSICS, VOL. 60, NO.1 (JANUARY-FEBRUARY 1995); P.108-119, 7 FIGS., 4 TABLES.

Ultrasonic velocity-porosity relationships for sandstone


analogs made from fused glass beads
Patricia A. Berge*, Brian P. Bonner*, and James G. Berryman*
ABSTRACT
Using fused glass beads, we have constructed a
suite of clean sandstone analogs, with porosities rang-
ing from about 1 to 43 percent, to test the applicability
of various composite medium theories that model
elastic properties. We measured P- and S-wave veloc-
ities in dry and saturated cases for our synthetic
sandstones and compared the observations to theoret-
ical predictions of the Hashin-Shtrikman bounds, a
differential effective medium approach, and a self-
consistent theory known as the coherent potential
approximation. The self-consistent theory fits the ob-
served velocities in these sandstone analogs because it
allows both grains and pores to remain connected over
a wide range of porosities. This behavior occurs
INTRODUCTION
We can construct theoretical estimates of the physical
properties of rocks and sediments by treating rocks and
sediments as composite materials having multiple solid and
fluid components. Physical properties of random composites
depend upon the properties of each component, the relative
concentrations of the components, and the microstructure.
Clays, microcracks, grain compositions and shapes, and
uncertainties in porosities and grain densities complicate
physical properties prediction.
Synthetic sandstones made from fused glass beads, having
uniform spherical grains, provide an ideal two-component
material for investigating theoretical relationships between
porosity, microstructure, and physical properties of compos-
ites (e.g., Wyllie et aI., 1956; Domenico, 1977; Sen et aI.,
1981; Johnson and Plona, 1982; Schwartz, 1984; Palciauskas,
1992; Blair et aI., submitted for publication). In this paper,
we compare measured elastic properties of fused glass-beads
and glass foams to estimates from effective medium theories.
because this theory treats grains and pores symmetri-
cally without requiring a single background (host)
material, and it also allows the composite medium to
become disconnected at a finite porosity. In contrast,
the differential effective medium theory and the
Hashin-Shtrikman upper bound overestimate the ob-
served velocities of the sandstone analogs because
these theories assume the microgeometry is repre-
sented by isolated pores embedded in a host material
that remains continuous even for high porosities. We
also demonstrate that the differential effective medium
theory and the Hashin-Shtrikman upper bound cor-
rectly estimate bulk moduli of porous glass foams,
again because the microstructure of the samples is
consistent with the implicit assumptions of these two
theoretical approaches.
Our results for these ideal composites demonstrate how
effective medium theories depend on microstructure.
A plethora of effective medium theories is available for
modeling elastic properties of composites. By ignoring mi-
crogeometry, many such theories use only the physical
properties and concentrations of the components to predict
effective elastic properties of the whole material (e.g., Voigt,
1928; Reuss, 1929; Hill, 1952; Wyllie et aI., 1956). Others
make simplified assumptions about microgeometry, such as
assuming the composite is approximately a homogeneous
background hosting isolated ellipsoidal inclusions (e.g.,
Eshelby; 1957; Wu, 1966).
We constructed a suite of sandstone analogs, with poros-
ities ranging from about 1 to 43 percent, to test the applica-
bility of composite medium theories. After measuring P- and
S-wave velocities in these dry and saturated synthetic sand-
stones, we compared the observations to theoretical predic-
tions of the Hashin-Shtrikman bounds (Hashin, 1962; Hashin
and Shtrikman, 1963; Watt et al., 1976), a differential effec-
tive medium approach (Walsh, 1980; Berge et al., 1992), and
Manuscript received by the Editor April 12, 1993; revised manuscript received June 17, 1994.
"Lawrence Livermore National Laboratory, Livermore, CA 94550.
1995 Society of Exploration Geophysicists. All rights reserved.
108
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
Velocity-porosity Relationships 109
a self-consistent theory (Gubernatis and Krumhansl, 1975;
Berryman, 1980a, b). We present the velocity measurements
and show that the self-consistent theory provides the best
estimates of observed velocities in these sandstone analogs
because it contains assumptions consistent with the actual
sample microgeometry.
The self-consistent theory allows a composite to become
disconnected at some finite porosity because it treats all
components, whether grains or pores, equally. Without a
background host material, cracks and pores can overlap and
percolate to make the composite discontinuous at suffi-
ciently high crack and pore concentrations. The self-consis-
tent theory is expected to provide good elastic properties
estimates for granular sedimentary rocks and other materials
that become unconsolidated at high porosities. The differen-
tial effective medium approach, however, is more appropri-
ate for materials such as glass foams (Walsh et al., 1965;
Walsh, 1980) and some igneous rocks (Berge et al., 1992)
whose microstructures can be approximated by a continuous
background host containing isolated cracks and pores (Berge
et al., 1993a).
Our results provide guidelines for using effective medium
theories to estimate elastic properties of natural materials,
such as rocks and sediments in geophysical exploration or
engineering applications. The bulk elastic properties and
density of the glass used to make our synthetic sandstone
samples are nearly identical to the properties of the glass
used by Walsh et al. (1965) to make foam samples. For all
practical purposes, the two suites of samples differ mechan-
ically only in microgeometry. This coincidence in the bulk
properties adds weight to our conclusions about the impor-
tance of microgeometry and enhances the utility of our
measurements. The tabulation of measured properties pre-
sented for our suite of synthetic sandstones will also be
useful to others investigating relationships between porosi-
ties and elastic properties.
METHOD
Samples
We constructed our sandstone analogs from beads of
soda-lime plate glass having composition 71-74 percent by
weight SiO
z,
12-15 percent NaOz, 8-10 percent CaO, 1.5-
3.8 percent MgO, 0.2-1.5 percent Al
Z03
, and 0-0.2 percent
KzO. Individual beads deviated less than 5 percent from a
uniform spherical shape, according to the manufacturer's
specifications. We used beads with diameters of 230 20 IJ.m
for most samples, but made five samples from beads with
548 48 IJ.m diameters, and one sample from 137 12 IJ.m
beads. The glass had a softening temperature of 722-730C
and an annealing temperature of 540-548C. To create our
samples, we placed beads inside 30 ml pyrex beakers and
sintered them to peak temperatures between 700 and 760C.
This was followed by slow cooling through the annealing
temperature to minimize microcrack development. Figure 1
shows the furnace temperature history for one sample. To
reduce thermal gradient effects, we surrounded the beakers
with fire bricks. Resulting samples were homogeneous and
isotropic.
Glass bead samples were cored out of the beakers and
ground to right circular cylinders with lengths and diameters
of about 2.54 em. Uncertainties in sample dimensions,
masses, and densities were about 0.1 percent, 0.01 percent,
and 0.2 percent, respectively, except for poorly consolidated
samples having uncertainties of -1 percent in their dimen-
sions. The manufacturer's specifications gave the glass den-
sity (p) as 2.42-2.50 g/cm
3
Since a sample sintered at 760C
had a 2.456 0.005 g/cm' density, we used 2.48
0.02 g/cm' for the glass density. Sample porosities (<I
ranged from about 0.01 to 0.43 (Table 1). For samples with
porosities <0.39, only the uncertainty in glass density con-
tributed significantly to porosity uncertainties. The more
friable, high porosity samples had higher uncertainties in
porosities because of sample dimension uncertainties. Ac-
cording to the manufacturer's specifications, loose beads can
have bulk densities between 1.36 and 1.55 g/cm
3
, implying
porosities of 0.36 to 0.46 for unconsolidated beads. The
theoretical porosity is 0.40 for random loose packing of
uniform spheres, and 0.36 for close packing (Bernal and
Mason, 1960). Our beads probably become unconsolidated
for porosities >0.40 for the smaller beads and >0.43 for the
larger beads.
Horizontal and vertical cross-sections were cut from the
ends of four fused-glass cores before the cores were ground
down to the desired length. These cross-sections were used
to make standard petrographic thin sections and scanning
electron microscope (SEM) images (Figure 2) for a separate
image processing study (Blair et al., submitted for publica-
tion). The image processing study found these samples to be
homogeneous and isotropic, and porosities determined from
the images agreed with porosities determined from sample
densities (Table 1) to within about 2 percent (Blair et al.,
submitted for publication). Images of low porosity samples
showed few microcracks, but the highest porosity samples
may contain many microcracks.
The grains in Figure 2 are in welded contact. Our synthetic
sandstones are thus analogs of cemented sandstones, where
cement and grains are both relatively stiff. The sample in
Figure 2d has few grain contacts and approaches the upper
8
00

U
0
'-'
8
ci.
5
'<t
E-<
8
C'l
0
0 6 12 18 24
Time (hours)
FIG. 1. Temperature history for a sintered glass-bead sample.
Peak temperature determines sample porosity; plateau at
annealing temperature reduces microcracking.
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
110 Berge et at,
limit in porosity for a consolidated sample. Yet the welded
grain contacts are expected to make the high porosity
samples much stiffer than loose bead packings (e.g., Wyllie
et aI., 1956).
Velocity measurement technique
We dried the fused glass-bead samples for 8-10 hours at
80C and stored them over a dessicant. We then used the
pulse transmission technique (Hughes and Jones, 1950;
Sears and Bonner, 1981) to measure P- and S-wave veloci-
ties (Vp and V
s
) at room temperature and atmospheric
pressure. Placing the transducer on top of the sample pro-
vided a consistently reproduceable, small axial stress neces-
sary for coupling. Frequencies were 0.5 MHz for the P and
5 MHz for the S transducers. We used honey as the coupling
agent between the transducers and dry samples because its
water solubility allowed easy sample cleaning for later use.
Traveltime measurements made using aluminum calibration
samples determined traveltime corrections for the coupling
agent and apparatus. We measured traveltimes but not
waveforms in this study.
The dry samples had traveltime uncertainties of -1-1.5
percent for P-waves. S -wave traveltime uncertainties were
-1-1.5 percent for most samples, but were -3 percent for
samples with porosities >0.36. We repeated traveltime mea-
surements at least three times for consistency and then used
measured sample lengths to calculate P- and S -wave veloc-
ities for all the dry samples. Uncertainties in velocities were
caused mainly by uncertainties in the traveltimes.
After completing the dry measurements, we evacuated the
samples and backfilled them with filtered, distilled water.
Because permeabilities were -1-30 darcies for samples with
porosities of -0.17-0.40 (Blair et aI., submitted for publica-
tion), we achieved full saturation for all but our lowest
porosity samples. The -1 percent porosity sample probably
contained a few isolated dry pores.
When the samples were saturated, we repeated traveltime
measurements to determine P- and S-wave velocities. For
these measurements, water was the coupling agent. Travel-
time uncertainties for the saturated samples were compara-
ble to uncertainties for the dry case. We were unable to
measure velocities for high porosity saturated samples made
from large diameter beads because the signals were too weak
for accurate timing.
EXPERIMENTAL RESULTS
We present observedP- and S-wave velocities for our dry
and saturated synthetic rocks in Table 2 and Figure 3. Vp
values for wet and dry samples differ by less than 5 percent
for all but the most porous samples. V
s
is higher for dry than
for saturated samples, and the difference increases with
porosity. This trend implies that most of the sandstone
analogs contain round pores, without significant concentra-
tions of flat cracks (Kuster and Toksoz, 1974). The density
contrast between water and air in pores results in higher V
s
values for dry samples. The four highest porosity specimens
constructed from large beads have much higher dry Vp and
Vs values than comparable specimens constructed from
smaller beads. Possibly the larger beads were more resistant
to microcracking. The observed changes in velocity with
porosity are similar to findings for other sintered glass-bead
samples (Plona, 1980; Johnson and Plana, 1982; Johnson et
Table 1. Synthetic sandstone sizes and porosities. *
Length Diam. Mass DensiR'
Sample (cm) (cm) (g) (g/cm ) Porosity (</l)
---
Q 2.541 2.550 31.857 2.456 0.010 0.008
BQ 2.541 2.541 30.081 2.334 0.059 0.008
L3.1 2.064 2.544 24.226 2.309 0.069 0.008
1.1 2.544 2.543 26.513 2.052 0.173 0.007
2.1 2.539 2.540 22.985 1.787 0.280 0.006
1.2 2.542 2.539 22.806 1.772 0.285 0.006
BDA 2.539 2.524 22.355 1.760 0.290 0.006
A 2.543 2.531 21.891 1.711 0.310 0.006
1.3 2.543 2.534 21.669 1.689 0.319 0.006
2.2 2.543 2.541 21.527 1.669 0.327 0.006
2.3 2.541 2.541 21.506 1.668 0.327 0.006
3.6** 2.540 2.540 21.476 1.669 0.327 0.006
1.5 2.542 2.525 20.196 1.587 0.360 0.006
1.4 2.543 2.527 20.102 1.576 0.365 0.006
3.1t 2.540 2.536 20.239 1.577 0.365 0.006
B 2.542 2.511 19.094 1.517 0.39 0.01
2.6 2.538 2.523 19.202 1.513 0.39 0.01
3.2t 2.255 2.513 16.934 1.514 0.39 0.01
3.4t 2.546 2.508 18.879 1.501 0.40 0.01
1.6 2.534 2.477 18.278 1.498 0.40 0.01
2.5 2.536 2.512 18.758 1.493 0.40 0.01
2.4 2.532 2.513 18.649 1.486 0.40 0.01
3.3t 2.575 2.451 17.301 1.423 0.43 0.01
3.5t 2.441 2.438 16.064 1.409 0.43 0.01
"Bead diameter = 230 20 urn unless otherwise indicated.
**Bead diameter = 137 12 urn.
tBead diameter = 548 48 urn,
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
Velocity-porosity Relationships 111
aI., 1982) and for sandstones having low clay contents (e.g.,
Han et aI., 1986; Blangy et at, 1993).
We used the densities and velocities to compute dynamic
bulk (K) and shear (u.) moduli for the synthetic sandstone
samples (Table 3). Bulk moduli are larger than shear moduli
for all samples (Figure 4a and 4b). K is smaller for dry than
for saturated samples, while fL values are about the same
now that the fluid density effect has been removed. Excep-
tions are the high porosity samples that may be weakened by
microcracks. Both the bulk and the shear moduli drop
rapidly with increasing porosity, approaching zero as the
porosity approaches the random packing point where the
glass beads would be unconsolidated. For our samples,
Gassmann's theory (Gassmann, 1951) cannot be applied to
estimate K for the saturated samples using K for the dry
samples, because dynamic moduli from low-pressure veloc-
ity measurements are much higher than the static moduli
required by the theory (Berge et aI., 1993b).
We included Vp/V
s
ratios and Poisson's ratios (IT) along
with the dynamic moduli in Table 3. The observed Vp/V
s
and IT values are slightly higher for saturated than for dry
samples, but do not vary greatly with porosity. For most
samples, observed IT values agree with the values of 0.2-0.3
generally observed in nonporous glasses (Barnes and
Hiedemann, 1956; Walsh et al., 1965). This suggests that if
V
s
measurements had been unavailable, reasonable esti-
mates could be found using the observed Vp values and the
assumption that Poisson's ratio is close to 0.25 for porous as
well as nonporous glasses. The Vp/V
s
ratios for the fused
glass-bead samples all lie in the range of 1.6--2.0 typically
FIG. 2. SEM images of four synthetic sandstones having porosities of (a) 0.17, (b) 0.285, (c) 0.365, and (d) 0.39. Scale bar is 100
um long. Porosity occupies the black space; the glass appears as white.
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
112 Berge et al,
observed for dry or saturated sedimentary rocks under low
pressure conditions (e.g., Toksoz et aI., 1976; Domenico,
1984; Wilkens et aI., 1984). Fusing the glass beads creates a
very stiff cement at the grain contacts, resulting in higher
Vp/V
s
ratios than the values near 1.5 typical of unconsoli-
dated sediments and loose bead packs (e.g., Domenico,
1976, 1977). This stiff cement precludes use of Hertzian
contact theory (e.g., Winkler, 1983; Palciauskas, 1992) for
modeling elastic properties of fused glass beads (Schwartz,
1984).
THEORETICAL ESTIMATES OF ELASTIC PROPERTIES
Hashin-Shtrikman model
We investigate microgeometry effects by comparing ob-
served velocities and dynamic moduli for our sandstone
analogs to theoretical predictions from effective medium
theories. The first method considered uses only volume
concentrations and elastic properties of the glass and the
pore fluid (air or water) to predict elastic properties of the
Table 2. Observed velocities for synthetic sandstone samples,"
DryV
f
Dry V
s
Wet Vp Wet V
s
Sample
<I>
(km/s (km/s) (kmls) (km/s)
Q 0.010 5.84 3.42 5.87 3.41
BQ 0.059 5.58 3.26 5.56 3.22
L3.1 0.069 5.53 3.36 ** **
1.1 0.173 5.29 3.09 5.24 2.98
2.1 0.280 4.68 2.76 ** **
1.2 0.285 4.55 2.70 4.55 2.54
BDA 0.290 4.51 2.67 4.38 2.50
A 0.310 4.20 2.62 4.23 2.36
1.3 0.319 4.20 2.54 4.24 2.36
2.2 0.327 4.42 2.51 ** **
2.3 0.327 4.42 2.54 ** **
3.6 0.327 4.25 2.53 ** **
1.5 0.360 3.31 2.08 3.44 1.89
1.4 0.365 3.52 2.21 3.57 2.01
3.1 0.365 3.67 2.25 ** **
B 0.39 2.84 1.69 3.08 1.62
2.6 0.39 2.77 1.66 ** **
3.2 0.39 3.56 2.20 ** **
3.4 0.40 3.63 2.19 ** **
1.6 0.40 2.68 1.61 2.90 1.51
2.5 0.40 2.59 1.57 ** **
2.4 0.40 2.48 1.51 ** **
3.3 0.43 2.76 1.63 ** **
3.5 0.43 2.76 1.57 ** **
*Vp uncertainties 1-1.5 percent for all samples. V
s
uncertainties 1-1.5 percent for samples with <I> :5 0.36 and 3 percent
for samples with <I> > 0.36.
**Not measured.
\0

4lt
(a)

V) "<t
..-.. ,.

(b)
~

g
.-
"<t
I
('f"l
Q..
I ~
~ ...
>-
++
s

(") dryVp
+
~ dry Vs
~
\-
+
+ wetVp + wetVs
"
,
~
-
0 10 20 30 40 50 0 10 20 30 40 50
Porosity (%) Porosity (%)
FIG. 3. Dependence of observed velocities (a) V
p
and (b) V
s
on porosity for dry and saturated synthetic sandstones.
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
Velocity-porosity Relationships 113
fused glass beads. Hashin and Shtrikman (1963) used varia-
tional principles to bound the change in strain energy caused
by introducing cracks and pores into a homogeneous me-
dium, providing rigorous upper and lower bounds on effec-
tive medium elastic moduli. These bounds are exact for a
particular microgeometry of concentric spheres (Hashin,
1962). Bounds on velocities can be computed from the
bounds on moduli, using volume averages of the solid and
fluid densities. Watt et al. (1976) and Berryman (1995) give
convenient mathematical expressions for the Hashin-Shtrik-
man bounds (see the Appendix).
In Figure 5, we plot measured velocities of our fused
glass-bead samples together with measured properties of
some other glass samples (Table 4) and the theoretical
Hashin-Shtrikman bounds. Figure 5a shows velocities for
the dry case, and Figure 5b shows velocities for the satu-
rated case. To calculate the Hashin-Shtrikman bounds, we
used the properties of our own glass given in Table 4, we
Table 3. Dynamic moduli for synthetic sandstone samples.
DryK Dry J..l. Dry WetK Wet J..l. Wet
Sample (GPa) (GPa) Vp/V
s
Dry a (GPa) (GPa) Vp/V
s
Wet a
--- ---
Q 45.4 28.8 1.71 0.238 46.4 28.6 1.72 0.244
BQ 39.8 24.8 1.71 0.242 40.7 24.8 1.72 0.247
L3.1 35.9 26.0 1.65 0.208 * * * *
1.1 31.2 19.6 1.71 0.240 34.7 19.8 1.76 0.260
2.1 20.8 13.7 1.69 0.231 * * * *
1.2 19.4 13.0 1.68 0.227 24.8 13.3 1.79 0.273
BDA 19.1 12.5 1.69 0.231 22.2 12.8 1.75 0.258
A 14.6 11.7 1.61 0.184 21.1 11.3 1.79 0.273
1.3 15.3 10.9 1.65 0.212 21.1 11.2 1.79 0.274
2.2 18.6 10.5 1.76 0.262 * * * *
2.3 18.2 10.8 1.74 0.253 * * * *
3.6 15.8 10.7 1.68 0.224 * * * *
1.5 8.25 6.89 1.59 0.173 13.7 6.94 1.82 0.283
1.4 9.19 7.73 1.59 0.172 14.1 7.85 1.77 0.266
3.1 10.6 8.01 1.63 0.197 * * * *
B 6.43 4.36 1.68 0.224 11.4 4.99 1.90 0.309
2.6 6.08 4.17 1.67 0.221 * * * *
3.2 9.40 7.30 1.62 0.192 * * * *
3.4 10.1 7.19 1.66 0.213 * * * *
1.6 5.53 3.89 1.66 0.215 10.1 4.28 1.92 0.315
2.5 5.12 3.67 1.65 0.211 * * * *
2.4 4.61 3.41 1.64 0.204 * * * *
3.3 5.84 3.77 1.70 0.234 * * * *
3.5 6.10 3.48 1.76 0.260 * * * *
*Not measured.
0
Irl

e
+
0
e
0
";'
~ ~

~
+

~
~
(a)
(*.
0
(b)

_N
,
....
~
' ...
......
0
edryK
,..
::10
dry J.L
- -
,
+wetK
,.
+ wet u
..
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Porosity(%) Porosity(%)
FIG. 4. Dependence of dynamic moduli (a) K and (b) J..l. on porosity for dry and saturated synthetic sandstones.
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
114 Berge et al.
assumed vanishing pore fluid moduli and density for the dry
case, and in the saturated case, we used a bulk modulus of
2.3 OPa, a vanishing shear modulus, and a density of
1.0 g/cm' for water. The Hashin-Shtrikman lower bound on
a given effective modulus, bulk or shear, vanishes if any of
the corresponding component moduli vanish; therefore these
curves are trivial for Vp and V
s
of dry samples (Figure 5a)
or for V
s
of saturated samples (Figure 5b).
P-wave velocities of unconsolidated bead packs (Wyllie et
al., 1956; Johnson and Plona, 1982) are in agreement with the
Hashin-Shtrikman lower bound predictions (Figure 5a, b)
because the Hashin-Shtrikman lower bound is exact for the
microgeometry of an assemblage of stiff spheres (beads)
surrounded by concentric weak spheres (air or water). The
Hashin-Shtrikman upper bound yields exact results for stiff
shells surrounding weak pores, and therefore predicts the
velocities of glass foam samples (Walsh et al., 1965)
(Figure 5a) and for fused glass-bead samples containing
isolated pores (Figure 2a), those having porosities less than
-0.20 (Figure 5a and 5b).
For higher porosities, in the saturated as in the dry case,
the microgeometry of fused glass-bead samples differs from
the concentric sphere model, and the Hashin-Shtrikman
upper bound greatly overestimates the velocities. Different
theoretical approaches containing different microgeometry
assumptions are required for improved estimates of the
velocities for the higher porosity fused glass-bead samples.
Self-consistent and differential effective medium theories
Next, we consider theoretical approaches that make sim-
ple assumptions about microgeometry to estimate velocities
'"
'=to"
........ +
.....
...... HS+
".
'"
..., ..It.
Vp
..... 1II.......... HS+
"- ..
."""
It. It. .......
It. It. It.......
.J It. It.
+
+*
'"
'"
'"
'"
'"
-
. t
..:--.---..
HS- ..
................t... +......
......
+
-
o -+----r----..,..---.,...--"""T-----,
O-+---""T--"""---T"""--"'T""---'
(b) saturated
o
(a) dry
10 20 30

40 50 o 10 20 30 40 50
Porosity (%) Porosity (%)
FIG. 5. Observed and theoretical velocities for (a) dry and (b) saturated synthetic sandstone samples. Data (a) and (b) are from
our samples (dots and crosses). as well as for unconsolidated glass bead samfles of Wyllie et al. (1956)(squares) and Johnson
and Plona (1982) (diamond, just below square), and (a) glass foamsamples 0 Walsh et al. (1965) (triangles). (Poisson s ratio was
assumed to be independent of porosity, for determining glass foam Vp from compressibility and density measurements.) Bold
lines indicate theoretical Vp ; thin lines show HS + indicates Hashin-Shtrikman upper bound In (a) and (b); Hashin-
Shtrikman - is the HS lower bound for Vp in (b).
Table 4. Bulk glass properties.
Glass" K (GPa) I.L (GPa) o V
p
(km/s) V
s
(km/s) p (g/cm')
!!l
46.1 0.7t 29.2 O.4t 0.238 0.006t 5.86 0.04 3.43 0.05 2.48 0.02
49.9t 26.2t 0.277t 5.85 3.25 2.48-2.50
5.49 2.40
46.0 0.5 30.4:1: 0.23 0.01 5.87:1: 3.48:1: 2.51
"Sources are: (1) this study; (2) 177-210 I.Lm unconsolidated beads from Johnson and Plona (1982); (3) 28 I.Lm unconsolidated
beads from WyIfie et al. (1956); (4) foam from Walsh et al. (1965).
tCalculated from velocities.
:l:Calculated from moduli.
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
Velocity-porosity Relationships 115
of our fused glass-bead samples. Berryman's (1980a, b)
self-consistent (SC) theory for estimating elastic moduli of a
composite uses component moduli and volume concentra-
tions, and represents all component pores, cracks, and
grains as ellipsoids (Appendix). Although the SC approach
has been criticized for allowing rapid changes in potential
energy as porosity increases and for yielding vanishing
moduli values at finite porosities (e.g., Bruner, 1976; Cheng,
1978; Henyey and Pomphrey, 1982), this model is appropri-
ate for materials that become disconnected at finite porosi-
ties (O'Connell and Budiansky, 1976; Berryman, 1980a).
Most SC approximations are physically realizable for spher-
ical inclusions (e.g., Hill, 1965; Budiansky, 1965; Wu, 1966;
Berryman, 1980a). The formulation of Berryman (1980b) is
also physically realizable, in principle, for all ellipsoidal
inclusions (Milton, 1985). We also use a differential effective
medium (DEM) theory (Walsh, 1980; Berge et al., 1992) that
represents cracks and pores as isolated ellipsoidal inclusions
embedded in a continuous background material (Appendix).
For a solid material containing fluid-filled cracks or pores,
this method produces stiffer effective moduli estimates than
the SC method (Cheng, 1978; Cleary et aI., 1980; Walsh,
1980) because cracks and pores cannot overlap, and the
background medium remains continuous as porosity in-
creases.
In our SC and DEM modeling, we used properties for our
glass listed in Table 4 and the same pore fluid properties that
were used in the Hashin-Shtrikman bound modeling. We
represented the glass as a continuous background medium
for the DEM modeling, and as spherical grains in the SC
modeling. Both approaches represent the pore space by
using spheroids with assumed aspect ratios (see Appendix).
We used spheres (with aspect ratios of 1.0) for the pores, and
spheroids with aspect ratios of 0.001 to assess effects of
microcracks (see Appendix). We converted the SC and
DEM moduli estimates to velocity estimates, using glass
bead and pore fluid densities.
Figure 6 presents SC and DEM Vp and V
s
estimates for
our dry and saturated fused glass beads. Curves labeled
DEM and SC were calculated using spherical pores. Curves
labeled DEM + cracks and SC + cracks also included the
0.001 aspect ratio cracks. For spherical pores without
cracks, the DEM and SC theories yield similar results for
low porosities and fit observed velocities for fused glass-
bead samples having porosities less than -0.20, in the dry
(Figure 6a and 6b) and the saturated (Figure 6c and 6d) cases.
These low porosity results plot close to the Hashin-Shtrikman
upper bound (Figure 5).
DEM results calculated without using microcracks over-
estimate velocities for higher porosity samples. Adding
microcracks improves the DEM fit for saturated samples
with porosities near 0.30 (Figure 6c and 6d), while degrading
the fit for dry samples (Figure 6a and 6b). Adding different
amounts of microcracks would not solve the problem, be-
cause the DEM theory's continuous background medium
requirement forces estimated velocities to be too high for
samples with porosities near the random packing point,
where the bead packs become unconsolidated.
Clearly, the SC theory provides the best Vp and V
s
estimates for both wet and dry fused glass-bead samples with
porosities >0.20 (Figure 6). Microcracks are not required for
modeling samples with porosities less than about 0.35 to
0.40. Including microcracks in the SC modeling would
improve the fit for the highest porosity samples made from
small glass beads, although 0.09 percent is too great a
concentration. This low volume fraction of cracks is also
consistent with the image processing study (Blair et al.,
submitted for publication).
Microgeometry effects
In Figure 7, we show bulk moduli previously plotted in
Figure 4a for our dry fused glass-bead samples along with
bulk moduli determined from compressibility measurements
for the glass foam samples of Walsh et al. (1965). We
compare these observations to the Hashin-Shtrikman upper
bound and the SC and DEM estimates of bulk moduli
(Figure 7). The SC and DEM estimates resulted from the
same computations used to produce SC and DEM velocity
curves shown in Figure 6a, whereas the Hashin-Shtrikman
curve came from computations used to produce its bounds
on velocities shown in Figure Sa. Although we used the
bulk properties of our own glass to compute the Hashin-
Shtrikman upper bound, DEM, and SC curves, the glass for
the foam samples of Walsh et al. (1965) has nearly identical
bulk properties (Table 4).
For porosities below about 0.20, the glass foam and
glass-bead samples have similar microgeometries. Bulk
moduli predicted by the three theories lie close together and
agree with the data (Figure 7). As porosity increases, micro-
geometry assumptions in the three theoretical approaches
lead to different results, just as microgeometry differences in
the two suites of samples lead to different observed bulk
moduli.
The microgeometry of the glass foam samples is analogous
to the concentric sphere model for which the Hashin-
Shtrikman upper bound yields an exact result. Thus the
Hashin-Shtrikman upper bound in Figure 7 lies just above
the observed bulk moduli for glass foam samples having
porosities from 1 to 70 percent. The DEM curve also
provides good estimates of bulk moduli for all the glass foam
samples because the implicit assumption of a continuous
background material is valid for glass foam.
In contrast, our sandstone analogs have much lower bulk
moduli than the Hashin-Shtrikman upper bound and DEM
theory predict for porosities above -0.20. The SC theory
yields much better results for these fused glass-bead sam-
ples, because it allows voids to percolate so that the material
can become unconsolidated at a finite porosity. For the same
reason, the SC theory fails to predict good estimates of the
bulk moduli for high porosity glass foam samples, since it
inherently assumes an inappropriate microstructure.
CONCLUSIONS
We find that the SC theory is the preferred effective
medium theory for estimating velocities of our synthetic
sandstones because microgeometry assumptions are com-
patible with the actual microstructures of our fused glass-
bead samples. Although the SC approximation does not
necessarily provide vanishing moduli precisely at the appro-
priate porosity, it correctly describes the general behavior of
materials that become disconnected at finite porosities
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
116 Berge et at
(O'Connell and Budiansky, 1976). At low porosities, our
synthetic sandstones have microstructures compatible with
the DEM assumption of a continuous background medium.
But the DEM approximation greatly overestimates the
strength of these sandstone analogs for higher porosities.
The DEM approach is appropriate for modeling the glass
foam (Walsh et al., 1965; Cheng, 1978; Walsh, 1980), since
the foam samples have microstructures that can be described
by isolated inclusions embedded in a continuous background
material (Berge et aI., 1993a).
We conclude that, since microgeometry controls the prop-
erties of porous composite materials, it is probably not
possible to find a single "best" effective medium theory for
modeling all such materials. Instead, it is preferable to pick
a theory whose assumptions about microgeometry match
those of the material being modeled. The glass beads and the
glass foam serve to illustrate this point, but porous real earth
materials will also show great dependence on microstruc-
ture. Our results provide guidelines for successful modeling
of natural materials. Well-cemented granular sedimentary
rocks share the key microgeometry characteristics that
determine the best theoretical approach for modeling the
fused glass-bead velocities (Berge et aI., 1993a). The micro-
geometry of rocks having isolated cracks and pores is
analogous to the glass foam microgeometry; thus theoretical
approaches that provide the best estimates of glass foam
50 40 30 20
SC+cracks
10
(b) dry
o
-+--...,r-----r---r--..... " T " - - ~
i
>
50
",
....
' .
40 30 20
.....
.....
.....
SC+cracks ""' .....
.....
,
,
10
~ pEM+cracks
.......::.: .....
.... .
" .
'"
'"
o
N -+--_--, ..,..__-,._.....__-.J......,
Porosity (%) Porosity (%)
50 40 30 20
............ OEM
'" .. ....................
SC+cracks
.........
<.
10
(d) saturated
o 50 40 30 20
....
....
10
....
SC+cracks .........
, . ~ : + -
'. ....
....... OEM
OEM+cracks '-f.....
.... .... ' ....
.... .
" ....
........ ' .
".
'.
(c) saturated
o
N-+---....,.---,.---"T'"'-"";''''''--'''''
Porosity (%) Porosity (%)
FIG. 6. SC and DEM velocity estimates compared to observed velocities. (a) Vp for dry synthetic sandstones. (b) V
s
for dry
samples. (c) Vp for saturated samples. (d) V
s
for saturated samples. Dot-dashed curves were calculated using spherical pores
and the DEM theory; solid curves came from spherical pores and the SC theory. Dotted curves also included cracks having
0.001 aspect ratios and 0.0009 volume concentrations, together with the spherical pores for the DEM theory; dashed curves had
similar cracks and pores for the SC theory.
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
Velocity-porosity Relationships 117
O+---,--,--....,.--r---=r---,----,
elastic media I. Spherical inclusions: J. Acoust. Soc. Am., 68,
1809-1819.
--- 1980b, Long-wavelength propagation in composite elastic
media II. Ellipsoidal inclusions: J. Acoust. Soc. Am., 68, 1820-
1831.
--- 1995, Mixture theories for rock properties, in Ahrens, T. J.,
Ed., American Geophys. Union handbook of physical constants:
Am. Geophys. Union, (scheduled for June).
Blangy, J. P., Strandenes, S., Moos, D., and Nur, A., 1993,
Ultrasonic velocities in sands-revisited: Geophysics, 58, 344-
356.
Bruner, W. M., 1976, Comment on 'Seismic velocities in dry and
saturated cracked solids' by Richard J. O'Connell and Bernard
Budiansky: J. Geophys. Res., 81, 2573-2576.
Budiansky, B., 1965, On the elastic moduli of some heterogeneous
materials: J. Mech. Phys. Solids, 13, 223-227.
Burns, D. R., Cheng, C. H., and Wilkens, R. H., 1990, Sandstone
pore aspect ratio spectra from direct observations and velocity
Inversion: Internal. J. Rock Mech. Mining Sci., 27, 315-323.
Cheng, C. H., 1978, Seismic velocities in porous rocks: Direct and
inverse problems: Ph.D. thesis, Massachusetts Institute of Tech-
nology.
Cheng, C. H., and Toksoz, M. N., 1979, Inversion of seismic
velocities for the pore aspect ratio spectrum of a rock: J. Geo-
phys. Res., 84, 7533-7543.
Cleary, M. P., Chen, I-W., and Lee, S.-M., 1980, Self-consistent
techni9ues for heterogeneous media: J. Eng. Mech. Div. Am.
Soc. CIV. Eng., 106, 861-887.
Domenico, S. N., 1976, Effect of brine-gas mixture on velocity in an
unconsolidated sand reservoir: Geophysics, 41, 882-894.
--- 1977, Elastic properties of unconsolidated porous sand
reservoirs: Geophysics, 42, 1339-1368.
--- 1984, Rock lithology and porosity determination from shear
and compressional wave velocity: Geophysics, 49, 1188-1195.
Eshelby, J. D., 1957, The determination of the elastic field of an
ellipsoidal inclusion, and related problems: Proc., Roy. Soc.
London, Ser, A, 241, 3('.6-396.
Gassmann, F., 1951, Uber die elastizitat poroser medien:
Veirteljahrsschrift der Naturforschenden Gesellschaft in Ziirich,
96,1-23.
Gubernatis, J. E., and Krumhansl, J. A., 1975, Macroscopic engi-
neering properties of polycrystalline materials: J. App!. Phys., 46,
1875-1883.
Han, D-H., Nur, A., and Morgan, D., 1986, Effects of porosity and
clay content on acoustic properties of sandstones and unconsoli-
dated sediments: Geophysics, 51, 2093-2107.
Hashin, Z., 1962, The elastic moduli of heterogeneous materials:
Am. Soc. Mech. Eng. Trans., J. App!. Mech., 29, 143-150.
Hashin, Z., and Shtrikman, S., 1963, A variational approach to the
theory of the elastic behavior of multiphase materials: J. Mech.
Phys. Solids, 11, 127-140.
Henyey, F. S., and Pomphrey, N., 1982, Self-consistent elastic
moduli of a cracked solid: Geophys. Res. Lett., 9, 903-906.
Hill, R., 1952, The elastic behaviour of a crystalline aggregate:
Proc., Phys. Soc. London, Ser, A, 65, 349-354.
---1965, A self-consistent mechanics of composite materials: J.
Mech. Phys, Solids, 13, 213-222.
Hughes, D. S., and Jones, H. J., 1950, Variation of elastic moduli of
igneous rocks with pressure and temperature: Bull., Geo!. Soc.
Am., 61, 843-856.
Johnson, D. L., and Plona, T. J., 1982, Acoustic slow waves and the
consolidation transition: J. Acoust. Soc. Am., 72, 556-565.
Johnson, D. L., Plona, T. J., Scala, C., Pasierb, F., and Kojima, H.,
1982, Tortuosity and acoustic slow waves: Phys. Rev. Lett., 49,
1840-1844.
Kuster, G. T., and Toksoz, M. N., 1974, Velocity and attenuation of
seismic waves in two-phase media: Part I. Theoretical formula-
tions: Geophysics, 39, 587-606.
Laws, N., 1980, A note on the yrediction of overall moduli for
composite materials: Quarterly . Mech. App!. Math., 33, 43-45.
McLaughlin, R., 1977, A study of the differential scheme for
composite materials: Internal. J. Eng. Sci., 15,237-244.
Milton, G. W., 1985, The coherent potential approximation is a
realizable effective medium scheme: Commun. Math. Phys., 99,
463-500.
Norris, A. N., 1985, A differential scheme for the effective moduli of
composites: Mech. Mater., 4, 1-16.
O'Connell, R. J., and Budiansky, B., 1976, Reply to William M.
Bruner: J. Geophys. Res., 81, 2577-2578.
Palciauskas, V. V., 1992, Compressional- to shear-wave velocity
ratio of granular rocks: Role of rough grain contacts: Geophys.
Res. Lett., 19, 1683-1686.
Plona, T. J., 1980, Observation of a second bulk compressional
70 60 50 40 30
Porosity (%)
20

k
. .
x-.
x,
. : .... HS
'.,:.: .
SC J..
" ...
', .
,. '"
" ".
" .....
DEM'., J.
".
10 o
0
V'l
0
oq-
0
;f
r<"l
Q.
::.d
0
~
0
.....
ACKNOWLEDGMENTS
elastic properties can successfully model rocks such as
oceanic basalts (e.g., Berge et al., 1992).
REFERENCES
FIG. 7. Estimated and observed bulk moduli for dry synthetic
sandstone and glass foam samples. Dots indicate dynamic
moduli of our sintered glass beads; triangles indicate bulk
moduli determined from compressibility measurements for
glass foam samples of Walsh et al. (1965). Dotted line shows
Hashin-Shtriknian upper bound on bulk moduli for various
porosities; dot-dashed curve shows DEM estimates of bulk
moduli; solid curve shows SC estimates.
The authors are grateful to Carl Boro for help with sample
preparation, to Wunan Lin for lending us transducers, to
Jack Dvorkin for suggesting the glass bead manufacturer, to
Xingzhou Liu for providing some of the beads, and to Pal
Wessel for his graphics software. This work was performed
under the auspices of the U.S. Department of Energy by the
Lawrence Livermore National Laboratory under contract
W-7405-ENG-48 and supported specifically by the Geo-
sciences Research Program of the DOE Office of Energy
Research within the Office of Basic Energy Sciences, Divi-
sion of Engineering and Geosciences.
Barnes, J. M., and Hiedemann, E. A., 1956, Determination of the
elastic constants of optical glasses by an ultrasonic method: J.
Acoust. Soc. Am., 28, 1218-1221.
Berge, P. A., Berryman, J. G., and Bonner, B. P., 1993a, Influence
of microstructure on rock elastic properties: Geophys. Res. Lett.,
20, 2619-2622.
Berge, P. A., Fryer, G. J., and Wilkens, R. H., 1992, Velocity-
porosity relationships in the upper oceanic crust: Theoretical
considerations, J. Geophys. Res., 97, 15239-15254.
Berge, P. A., Wang, H. F., and Bonner, B. P., 1993b, Pore pressure
buildup coefficient in synthetic and natural sandstones: Internal.
J. Rock Mech. Mining Sci., 30, 1135-1141.
Bernal, J. D., and Mason, J., 1960, Co-ordination of randomly
packed spheres: Nature, 188, 910-911.
Berryman, J. G., 1980a, Long-wavelength propagation in composite
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
118 Berge et 81.
wave in a porous medium at ultrasonic frequencies: AppI. Phys.
Lett., 36, 259-26l.
Reuss, A., 1929, Berechnung der fliessgrense von mischkristallen
auf grund der plastizitatsbedingung fur einkristalle: Zeitschrift fur
Angewandte Mathematik and Mechanik, 9, 49-58.
Schwartz, L. M., 1984, Acoustic properties of porous systems: II.
Microscopic description, in Johnson, D. L., and Sen, P. N., Eds.,
Physics and chemistry of porous media: Am. Inst. Phys., Conf,
Proc. 107, 105-117.
Sears, F. M., and Bonner, B. P., 1981, Ultrasonic attenuation mea-
surement by spectral ratios utilizing signal processing techniques:
IEEE Trans. on Geoscience and Remote Sensing, GE-19, 95-99.
Sen, P. N., Scala, C,; and Cohen, M. H., 1981, A self-similar model
for sedimentary rocks with application to the dielectric constant
of fused glass beads: Geophysics, 46, 781-795.
Toksoz, M. N., Cheng, C. H., and Timur, A., 1976, Velocities of
seismic waves in porous rocks: Geophysics, 41, 621-645.
Voigt, W., 1928, Lehrbuch der kristallphysik: Teubner.
Walsh, J. B., 1965, The effect of cracks on the compressibility of
rock: J. Geophys. Res., 70, 381-389.
--1980, Static deformation of rock: J. Eng. Mech. Div. Am.
Soc. Civ. Eng., 106, 1005-1019.
Walsh, J. B., Brace, W. F., and England, A. W., 1965, Effect of
porosity on compressibility of glass: J. Am. Ceramic Soc., 48,
605-608.
Watt, J. P., Davies, G. F., and O'Connell, R. J., 1976, The elastic
properties of composite materials: Rev. Geophys. Space Phys.,
14, 541-563.
Wilkens, R., Simmons, G., and Caruso, L., 1984, The ratio Vp/V
s
as a discriminant of composition for siliceous limestones: Geo-
physics, 49, 1850-1860.
Winkler, K. W., 1983, Contact stiffness in granular porous materi-
als: Comparison between theory and experiment: Geophys. Res.
Lett., 10, 1073-1076.
Wu, T. T., 1966, The effect of inclusion shape on the elastic moduli
of a two-phase material: Int. J. Solids Structures, 2, 1-8.
Wyllie, M. R. J., Gregory, A. R., and Gardner, L. W., 1956, Elastic
wave velocities in heterogeneous and porous media: Geophysics,
21,41-70.
APPENDIX
EXPRESSIONS FOR CALCULATING MODUU AND VELOCmES
A composite material has effective elastic moduli K and IJ.,
and an effective density p, Within the material, the moduli
and density are functions of position, K(r), lJ.(r), and per),
which have discrete values K
i
, 1J.j, and Pi within the ith
component. In all of our modeling, we computed theoretical
estimates of elastic moduli and then calculated velocities
using
(A-6)
and
t+'"
Vp =
p
V s = ~ ,
(A-I)
(A-2)
The Hashin-Shtrikman upper and lower bounds Kiis and
Kiis on the effective bulk modulus are
KiIs == A[K(r), ~ IJ.min] S K s A[K(r), ~ IJ.max] == KtIs,
(A-5)
if IJ.min is the smallest and IJ.max is the largest of the
component shear moduli lJ.i (Berryman, 1995). The expres-
sion for the Hashin-Shtrikman bounds on the effective shear
modulus requires defining
IJ. (9K + 81J.)
F(K, IJ.) == - .
6 K+ 21J.
Then the upper and lower Hashin-Shtrikman bounds lJ.iis
and lJ.iis are
where p is the volume average of densities Pi of N solid and
fluid components having relative concentrations C i:
IJ.HS == A[IJ.(r), F(Kmin , IJ.min)]
S IJ. S A[IJ.(r), F(K
max
, IJ.max)] == IJ.tIs (A-7)
Calculated effective moduli can be expressed in terms of this
weighted average if X is replaced with either K or IJ. and if y
is replaced by some constant dependent upon component
moduli.
N = 2 for all our models. We use the bracket notation in
equation (A-3) for the volume average.
Although many papers give convenient expressions for the
Hashin-Shtrikman upper and lower bounds on moduli (e.g.,
Watt et al., 1976), we favor the compact notation of
Berryman (1995), which also allows us to express the SC
estimates in a similar form. Following Berryman (1994), we
define
N
P = (p(r == L CiPi (A-3)
(Berryman, 1995).
Eshelby (1957) showed that for a uniform strain field
applied at an infinite distance, the strain inside an isolated
ellipsoidal inclusion in an infinite elastic medium is uniform
and the solution for displacement can be found in closed
form or evaluated in terms of tabulated integrals. If all cracks
and pores in a rock are represented by randomly distributed,
randomly oriented spheroids, effective elastic moduli and
velocities of the whole rock can be estimated using simple
algebraic expressions in terms of spheroid aspect ratios
(ratios of minor to major semiaxes) and known moduli of the
rock's solid and fluid components (Eshelby, 1957; Wu,
1966).
Many variations using static and single scattering approx-
imations have been developed from Eshelby's work. Back-
ground moduli used in the algebraic expressions can be
assumed to equal the solid grain moduli (e.g., Kuster and
Toksoz, 1974), or they can be equated to the rock's effective
moduli and the expressions can be solved iteratively (e.g.,
Wu, 1966). Iterative approaches are generally called self-
consistent (SC). All these methods will produce similar
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/
Velocity-porosity Relationships 119
results in the low porosity limit. Using spherical pores and
representing the background medium by the grain moduli
gives the Hashin-Shtrikman upper bound (Kuster and
Toksoz, 1974; Cheng, 1978).
Iterative solutions in which the roles of pore fluids and
grain materials can be interchanged without altering the
effective moduli formulas do not have an identifiable back-
ground host medium that remains continuous at all porosities
(Berryman, 1980a, 1995). Since the expressions are symmet-
rical for all component moduli, cracks and pores may
overlap and effective elastic moduli may vanish at some
finite porosity (O'Connell and Budiansky, 1976; Bruner,
1976; Cheng, 1978; Henyey and Pomphrey, 1982). Most SC
theories have been developed using a static approach and are
symmetrical for spherical inclusions, but the formulation of
Berryman (1980a, b) is dynamic and has no specified host
material even for ellipsoidal inclusions.
For spherical grains and pores, SC estimates for the
effective bulk and shear moduli can be written as
K
sc
= A[K(r), j ILsc] and ILsc = A[1L(r), F(K
sc,
ILsc )].
(A-8)
This result, given by Berryman (1980a), reduces multiple
scattering effects and treats all component moduli in the
same manner. For spherical inclusions the results match
estimates derived using a static approach (e.g., Hill, 1965;
Budiansky, 1965; Wu, 1966). Berryman (1980b) presents
corresponding expressions for spheroidal inclusions. We
solved iteratively, using moduli of the glass beads (Table 4)
as initial estimates of K
sc
and ILsc.
In estimating elastic properties of a porous composite, one
way to accommodate the energy change caused by addi-
tional cracks and pores is to allow each crack or pore to grow
slowly in the medium so that effective medium potential energy
changes slowly (Bruner, 1976; Henyey and Pomphrey, 1982).
An equivalent approach is to add small portions of the cracks
and pores gradually to the background material and iteratively
solve for effective elastic moduli (Cheng, 1978). Various similar
methods that incrementally change strain are termed the dif-
ferential effective medium (DEM) approach (McLaughlin,
1977; Laws, 1980; Cleary et al., 1980; Walsh, 1980; Norris,
1985). The DEM solution is physically realizable (Norris, 1985)
and lies between the Hashin-Shtrikman upper and lower
bounds (McLaughlin, 1977; Norris, 1985).
For a background medium with elastic moduli K
l
and ILl
and spherical inclusions with moduli K
z
and ILz, the DEM
estimates of the effective moduli K
DEM
and ILDEM are found
by computing
DEM 4 DEM]
M K m - l + j"lLm-l mc-;
K
DEM
-K = ~ (K _K
DEM
) [ -
1 LJ z m-l 4 DEM M
m=l K z + j"lLm-l
(A-9)
and
M
2: (l/5)(lLz - I L ~ : : ' Y )
m=l
[
DEM F(K
DEM
DEM)1
ILm-l + m-l , ILm-l me-
- (A-10)
+F(K
DEM
DEM) M
ILz m-l' ILm-l
(Berge et al., 1992). Here the pores slowly added over M
iterations cannot overlap. The background remains contin-
uous, even for very high porosities cz. This result is numer-
ically equivalent to those of Cheng (1978) and Walsh (1980).
Berge et aI. (1992) give expressions for multiple sets of
spheroidal inclusions.
We can see from Figure 2 that the voids in the fused
glass-bead samples are mostly large open pores with aspect
ratios close to unity. Although some of the pores may have
aspect ratios closer to 0.1, as long as the total concentration
of large aspect ratio spheroids is sufficient to account statis-
tically for most of the pore volume, the exact values used for
aspect ratios near unity do not greatly affect theoretical
estimates of velocities and moduli (Burns et al., 1990).
Therefore we used spheres to represent pores in our SC and
DEM modeling.
In Figure 2 some of the grains contain flat microcracks
with extremely small aspect ratios. Aspect ratios smaller
than -0.001 cannot be detectedreliably using SEM images,
and values smaller than -0.0001 are not generally needed in
velocity modeling (Walsh, 1965; Cheng and Toksoz, 1979;
Burns et al., 1990). Our DEM expressions become compu-
tationally unstable if the concentration of spheroids having a
given aspect ratio is not numerically smaller than the aspect
ratio (Berge et al., 1992). We could not constrain the aspect
ratios and concentrations of the microcracks, but did obtain
some qualitative information about their possible effects by
considering SC and DEM models without microcracks to-
gether with models containing concentrations of 0.0009 for
spheroids with the aspect ratio 0.001.
D
o
w
n
l
o
a
d
e
d

1
0
/
1
0
/
1
3

t
o

2
1
0
.
2
1
2
.
5
3
.
1
7
9
.

R
e
d
i
s
t
r
i
b
u
t
i
o
n

s
u
b
j
e
c
t

t
o

S
E
G

l
i
c
e
n
s
e

o
r

c
o
p
y
r
i
g
h
t
;

s
e
e

T
e
r
m
s

o
f

U
s
e

a
t

h
t
t
p
:
/
/
l
i
b
r
a
r
y
.
s
e
g
.
o
r
g
/

You might also like