PETROPHYSICS, VOL. 43, NO. 3 (MAY-JUNE 2002); P.

185-196; 15 FIGURES

Grain Sorting, Porosity, and Elasticity

Jack Dvorkin' and Mario A. Gutierrez'

ABSTRACT

This paper presents an effective-medium theoretical

description of the elastic properties of isotropic sandshale mixtures. The effective elastic moduli and velocities are related to the total porosity, shale content, and
the mixing mode (dispersed or laminar). The effectivemedium equations can be used for theoretical mixing of
sand and shale in dispersed or laminar modes and, ultimately, for seismic forward modeling and reservoir
characterization.

INTRODUCTION AND PROBLEM FORMULATION

Grain size and grain size distribution are basic characteristics of sediment texture (Figure 1). They are determined
by the depositional history and affect reservoir quality

FIG. 1

directly via porosity and permeability. They also affect the

elastic properties of the sediment and, ultimately, the seismic attributes that are analyzed to characterize the reservoir.
Here, the emphasis is on simple bimodal sand-shale mixtures for which the large difference between the sand and
shale particle size determines the sediment's heterogeneity.
To demonstrate the complex bulk- and elastic-property
behavior of a bimodal system consisting of small shale particles and large sand grains, we examine a water-saturated
section of a Gulf Coast well at a depth of about 900 m (Figure 2). The section includes several depositional cycles
whose vertical extent is apparent in the gamma-ray (GR)
curve. The porosity curve approximately mirrors the GR
curve, with high porosity matching the sandy (low-GR)
zones. The P-wave impedance curve approximately mirrors
the porosity curve with the low impedance matching high
porosity.
In spite of the congruency between different log curves
apparent in Figure 2, the cross-plots of sediment properties

Thin sections of unconsolidated sand showing deteriorating sorting (from left to right). Images courtesy Norsk Hydro.

Manuscript received by the Editor August 10,2001.

'Geophysics Department, Stanford University
02002 Society of Professional Well Log Analysts. All rights reserved.
May-June 2002

PETROPHYSICS

185

Dvorkin and Gutierrez

do not producc uniquc trcnds (Figurc 3). Thc incrcasing GR

acts first to incrcasc and then reducc tlic bulk dcnsity;
reducc and incrcasc porosity; and increasc and rcducc thc
impedance.
Thc clastic modulus vcrsus porosity cross-plot also has
two branchcs, in accord with thc changing GR values (Fig-

50

75 100
<;it

1.X2.02.22.4

.2 ..i .4 .5

llIlOl3

I'orosity

urc 4). Within thc satnc porosity range, tlic slialc-branch

scdiincnt is softer than thc sand-branch scdimcnt.
The V-shaped cross-plots, such as in Figure 3 , sccond
frame, and invcrtcd I!third frame, are well-known in powder technology (Cumberland and Crawford, 1 987). Thcy
rcsult from mixing particlcs of two distinctivcly diffcrcnt
incan sizcs such that thc small particlcs fill the pore space

4
3
0
Ip (kn1k g t c )

1.0
1.5
2.0
Is (km's g k c )

Itcsistivity

FIG.2 Well log curves for a Gulf of Mexico well. From left to right-GR; bulk density; density-derived porosity (PhiD) and neutron
porosity (NPHI); P-wave impedance (Ip), S-wave impedance (Is), and deep resistivity. Data courtesy Lake Ronel Oil Company.

FIG.3 Cross-plots of well log data from Figure 2. Left to right: bulk density versus GR; density-derived porosity (PhiD) versus GR;
f-wave impedance versus GR; and PhiD versus NPHI.

186

PETROPHYSICS

May-Junc 2002

Grain Sorting, Porosity, and Elasticity

between the large particles (dispersed mixing). In the above

field example these are the shale and sand particles, respectively.
This bimodal mixture effect can be observed in a synthetic data set shown by c . A. Estes (1 992, pers. comm.)
where the porosity and elastic-wave velocities have been
measured in mixtures of round glass beads of two
sizes-0.5 and 0.05 mm (Figure 5). Porosity is approximately the same, and so are the elastic-wave velocities, at
the end points where the pack consists of either large or
small grains. As the fraction of small grains increases,
porosity first decreases from 0.38 to its minimum of 0.26
and then increases again to 0.36. Both P- and S-wave velocity first increase and then decrease. The maximum of the
elastic-wave velocity corresponds to the minimum porosity.
The V-shaped cross-plots have been used to identify
clay-supported and framework-supported sediment
domains (Herron et al., 1992). The V-shaped cross-plots of
the density-derived porosity and neutron porosity (such as
in Figure 3, right panel) appear as early as 1974 in the
Schlumberger log interpretation manual.
Marion (1990) and Yin (1992) have developed a heuristic theory to explain the Vshapes and supported it by laboratory measurements on synthetic geomaterials. The

*--

V-shaped plots resulting from Yin's (1992) measurements

of the total porosity, elastic-wave velocity, and permeability in synthetic mixtures of Ottawa sand and kaolinite are
shown in Figure 6.
In this data set, both the particle tjpe and its volumetric
proportion in the mixture (which is relevant to geologic
sorting), strongly affect the properties of the synthetic sediment in a non-linear way. The same is true in situ for real
sand/shale mixtures.
Worthington (1985) has analyzed the effect of sandhhale
mixing on resistivity. In this paper, we concentrate on the
elastic properties of the sediment.
The ability to quantify the effect of sorting and sand/
shale mixing on the bulk and elastic sediment properties is
important for the interpretation in situ of the seismic
response of sand and shale sequences. Synthetic models are
used to predict reservoir quality from measured seismic
attributes. One important target of such modeling are thin,
interbedded sandhhale layers in deep water sedimentary
environments. In present deepwater exploration, seismic
near-, far-, and full-offset substacks are inverted for P- and
S-impedances which are then correlated with net-to-gross
and porosity at calibration well locations to allow for lateral
reservoir prediction away from the wells. Consequently,
models are needed that can be used for upscaling from grain
to log to seismic scale and that relate the elastic properties to
porosity and shale content at the seismic scale.

Critical Concentration Point

Critical Concentration Point

#

90

TOPOLOGY OF BIMODAL
MIXTURES-DISPERSED MODE

80

Thomas and Stieber (1975) proposed an effectivemedium model, defining three types of sandhhale distribution and analyzing the total porosity as a function of the
component concentration. The tree types are: (a) dispersed
shale in a sand matrix (shale coating the grains, or pore fill-

70

$and
end-membe

60

50
1

0.2

0.3

0.4

GR

Porosity
FIG. 4 Compressional modulus (M-modulus), the product of
bulk density and P-wave velocity squared, versus total porosity
for data shown in Figure 2, color-coded (gray scale) by GR. The
curves are from the sand/shale mixing theory. The large
symbols represent the pure sand and pure shale end points and
the critical concentration point (upper left corner). The other
large symbol at the critical concentration point is from the
approximate equation (19). The two laminar mixture curves that
connect the pure sand and pure shale end points are from equations (22) and (23) and are practically identical to each other.

May-June 2002

.2

.4

.6

.8

Small Bead Fraction

.2

.4

.6

.8

Small Bead Fraction

FIG.5 Mixtures of glass beads of two sizes, 0.5 and 0.05 mm.
Porosity (left) and P- and S-wave velocity (right) versus the
volume fraction of small beads. After Estes (1992).

PETROPHYSICS

187

Dvorkin and Giitierrez

ing); (b) laminated sand/shale sequences; and (c) structural

shale (sand-sizcd shalc particlcs in load-bcaring positions
within the rock, rcplacing quartz grains).
Most common arc laminatcd and disperscd shales, and
naturally, combinations of both distributions. In this paper
we study the elastic properties of these two distributions,
assuming that the shale platelets can be approximated by
isotropic spheres with a known radius r which is much
smaller than the radius R of tlic quartz grains.
Thc two end members of inixturcs 0.f largc and sinall
grains arc the packs of only the large grains or only the small
grains (Figurc 7). The porosity of the former is @~s,y and that
of the latter is @,y,. The size ofthe large grains is much larger
than that of thc sinall grains ( R >> r.). As a result, the small
grain packs can fit within thc pore space of the large grain
pack and still retain their local porosity of @,y,. This mixing

inode is the dispersed mode. It is the inost coinpact way of

mixing grains of diffcrent sizcs.

Dcpcnding on thc rclative conccntrations of thc large and
small grains, various mixturc configurations arc possiblc, as
shown in Figure 7. The critical point in this figure is shown
in the central panel, whcrc the sinall grains completely till
the porc space of the large grain pack and the large grains
arc still in contact with each other. Yin (1 992) calls this configuration critical concentrution
Thc importance of this middlc point is that it separates
two different structural domains. The domain on the left is
whcrc the external load applied to the mixture is supported
by the large grain framework. In the context of sandhhale
mixtures, this is shaly sand. The domain on thc right is
whcrc the large grains arc suspcnded in thc small particlc
framework which is now load-bearing. This is sandy shalc.

1000

1000

6-

s?100

Y
100

10

10

Clay Volume Fraction

Clay Volume Fraction

Clay Volume Fraction

Total Porosity

FIG. 6 Mixtures of sand and clay. Measurements were conducted on room-dry samples at varying hydrostatic confining pressure
(shown in the plots). From left to right-total porosity; P-wave impedance; air permeability versus volumetric clay content; and air
permeability versus total porosity (after Yin, 1992).

(P = (Pss

(PSS

@SS @SH

= (PSS @SH

(PSH

(PSS

(P

@SS (PSH

FIG. 7 Dispersed mixing mode of large and small grains. The large-grain end member is on the far left and the small-grain end
member is on the far right. The critical concentration point is in the middle. The total porosity is given above the respective frames.
The fourth from the left frame shows a sub-volume of the small particles that retains the porosity of the small grain pack end member.

I88

PETROPHYSICS

May-June 2002

Grain Sorting, Porosity, and Elasticity

Let the number of the large and small grains in the mixture be L and I, respectively. Then the total volume of the
mixture Vrin the domain to the left of the critical concentration point (shaly sand) is that of the large grain pack (V,>

= V --xR3

t1-3

L
l-@ss

-,

V,m=sXr

@SH

(8)

l-@SH

As a result, the total porosity is

@=@SH

(1)

/[l+(l-@SS)

/PI.

(9)

The volume fraction of the small grain pack in the mixture is, in the sandshale mixture context, the volume fraction of shale C in the whole rock. It is

The pore volume of the large grain pack is

c = [1+ (1 -) @,
and the total volume of the small grains, counting the pore
space between them, is

/ PI-'

Then equation (9) becomes @ = @sHC.

The summary of the above equations in the sand/shale
mixture context is

(3)
In the context of sand/shale mixtures, the latter is the total
shale volume in the sediment.
If vt, 5 Vpl,the small grains can fit into the pore space of
the large grains without distorting the initial large grain
framework. The pore volume of the mixture V,, is the pore
volume of the large grain pack V,, minus the small grain
material volume

where @ssand @sH are the porosities of the pure sand and
pure shale end-members, respectively.
The dry-rock bulk density pDRy
of the mixture is
= (1 - @sslpSs + ~ ( 1 -@sH

PDRY

=(~-C)PSS +C(~-@SH)PSH
,

If3!, > @ss, the large grains are suspended in the small
grain pack. The total volume of the mixture now is the sum
of the small grain pack volume and the large grain material
volume

4
4
1
V, = - x R 3 L + - n r 3 -.
3
l-@SH

May-June 2002

lPsH , c

PDRr

@ss; (12)

C>@ss;

-I

(7)

PETROPHYSICS

189

Ihorkin and Cntierrez

crs 0.f sand (Thomas and Sticbcr, 1975; Schlumbcrgcr,

1989).
The total porosity of this mixture is simply the wciglicd
average of the sand and shale porosities

cp = CcpSl, + (1 - C)cps,s .

(13)

The bulk density is

P/j

= ( ~ - c ) [ ( I- ~S.S>P.S.S + ~ . y . ~ ~ ~ : y , s l (14)

+ C[(1 - cpw ) P S I / + c p S / l

Pl3I1

of two clastic clcincnts--the softer clcmcnt that is porous

shale and the stiffer element that is the sand grain material.
The softer element envelops the stiffer clcmcnt thus crcating the topology that is a realization of the HashinShtrikman lower bound (IHSLB). Dvorkin ct al. (1999)
show that if the elastic contrast between the two clcnicnts is
large, HSLB accuratcly predicts cxperiincntal tncasurcmcnts. The resulting expressions for the mixture's elastic
moduli arc

1,

whcrcpl,55andplLyllarc the densities of the pore fluids in the

sand and shale bodics, respectively.
The porosity dependence on the small grain (shale) content is non-linear in the dispersed mode and linear in the
laminar inodc (Figure 9). Thc dispersed inode curve rcproduces the V-shapes apparent in Figures 3 , 5 , and 6.
ELASTICITY OF BIMODAL
MIXTURES-DISPERSED MODE

Assume that the effective bulk ( K ) and shear (G) moduli

of the pure sand and pure shale end-members arc known and
arc, respectively, K,ysandGss for sand, and KSll and G5,, for
shale. Also known arc the grain material elastic moduli that
arc, respectively, K , and GI for the sand (large) grains, and
K2 and G2for the shale (small) grains.
Consider first sandy shale (the three right-hand frames in
Figure 7) where the pack of the shale particles envelops the
sand grains. Thc mixture undcr examination is a composite

(15)
and where C is the volume shale content as given by equation ( I 0).
Among effective-mcdiiun bounds, those of the HashinShtrikman model arc the tightest clastic bounds for an isotropic mixture of several elastic components. The VoigtRcuss bounds arc inore rclaxcd. The lower (Reuss) bounds
for the bulk and shear moduli arc (Mavko ct al., 1998)
C2

.5

..

.2

.4

.6

.8

Volumetric Shale Content

FIG. 9 Total porosity versus shale content for dispersed and
laminar modes of mixing. In this example, the pure-sand
porosity is 0.3 and the pure-shale porosity is 0.5. The symbols
indicate the end members (pure sand and pure shale) and the
critical concentration point.

I 90

KM/,y =[cKG: + ( I - C ) K , ' ] - ' ,

GM/,. =[CG,:

$sLy;

(16)

+(I-C)G,'IP1.

The P- and S-wave velocity (ultrasonic pulsc transmission) ineasurcd by Yin ( I 992) in water-saturated pure Iaolinite at 10 MPa hydrostatic effective pressure arc V p= I .94
and Y5= 0.99 kids, respectively. The corresponding bulk
density is pR= 1.83 g/cm'. The resulting bulk and shear
moduli arc K.o, =pB[V$ (4/3)1!?] = 4.5 GPa and G.sll=
pnV+ = 1.8 GPa, rcspcctively. The bulk and shear moduli of
pure quartz grains are K I = 36.6 GPa and GI = 45 GPa,
rcspectively. Using these inputs, the sandshale mixture
elastic moduli arc computed according to cquations ( I 5)
and ( I 6) and plotted in Figure 10.
The difference between the HSLB and Rcuss results is
substantial, especially at the critical concentration point
(C = @5.y =: 0.4). HSLB equations arc slightly inore complicated than thc Rcuss equations, however they arc inore
appropriate for an isotropic mixture of elastic elements and

PETROPHYSICS

May-dunc 2002

Grain Sorting, Porosity, and Elasticity

thus recommended for estimating the elastic moduli of

sandy shale.
In cases where the S-wave data are not available, equation (15) cannot be used because neither the bulk nor the
shear modulus can be calculated from Vp only. Let us
assume for the purpose of algebraic derivation that Poisson's ratio for the sand grain material and for the pure shale
are the same, v1 = vsH 5 v * (normally, v1 < vsH). Then by
substituting isotropic elasticity equation

K = M-

1-2v
4
l+v
G = MM=K+-G
2(1- v) '
3
3(1- V ) '

weighted average (lower bound) for the compressional

modulus
c2q5ss: MMIX =

c
-+MsH

l-c
MI

-1
'

(19)

The relative error of using this equation instead of the exact

equation (15) may be unacceptably large (Figure 11, right).

12

(17)

10

into equation (15), HSLB for the compressional modulus of

the mixture MMlxcan be expressed in terms of the compressional modulus of the sand grain material, A41 = K I +
(4/3)G1,that of pure shale, MSH= K ~ +H(4/3)Gs~,and v*

8
6
4

2
I

I
GMIX

[[

MSH

15(l'v*)

'

3(1- v*)

0
.4

1 - 2v*
4( 1 - *)(4 - 5 ~ * )

1
.6
.8
Volumetric Shale Content

FIG. 10. The effective bulk ( K ) and shear (G) moduli of the
mixture of water-saturated kaolinite and quartz grains computed
according to equations (15) and (16). The results are for the
sandy shale domain where the volumetric shale content
exceeds the porosity of pure sand (about 0.4).

I].

1-c
+2(4-5v*)(M, / MSH) + 7-5v*
- ( 7 - 5v*)

(18)
It is not likely that v1 = v s ~ .Therefore, equations (18)
have to be treated as an approximation. MMIx computed
from these approximate equations is compared to that computed from the exact equation (15) in Figure 11, left. It
appears that although the Poisson's ratio of sand grains may
be as low as 0.07 (pure quartz) and that of the watersaturated shale may be as high as 0.45, the error in using
equation (18) does not exceed 5% if the common Poisson's
ratio v, is set close to VSH.
The correct choice for v, should be based on the knowledge of the elastic properties of shale in the region and
depth range of interest. Very shallow shales typically have a
very large Poisson's ratio that approaches 0.45. In deep
shales, Poisson's ratio may be close to 0.3 (Figure 12). In
uplifted sedimentary sequences, relatively low (- 0.3) Poisson's ratio in shales may appear at shallow depths.
An alternative for Vp-only modeling is to use the Reuss
May-June 2002

62

I5

a,

.->

-&j
.I
a,

0
.4

.6

.8

C = Shale Content

.4

.6

.8

C = Shale Content

FIG. 11 (left) The relative error of using equation (18) instead

of equation (15) for the effective compressional modulus (M) of
sandy shale. (right) The relative error of using equation (19)
instead of equation (15) for the effective compressional modulus
(M) of sandy shale. The triplets of numbers stand forv,, VSH, and
21,. For example, the triplet (0.07 0.45 0.45) meansvl = 0.07; VSH
= 0.45; and v,= 0.45.

PETROPHYSICS

191

Dvorkin and Gutierrez

Consider ncxt shuly sand (thc thrcc Icft-hand framcs in

Figurc 7) where the packs of thc shale particles are located
within thc pore space of the undisturbcd sand grain framcwork. Shaly sand can bc treated as a mixture of two end
mcmbcrs which arc the pure sand and thc critical concentration sandlshalc mixturc (Figurc 13). Then thc shaly sand
elastic moduli vary betwecn thosc of the end membcrs
which arc the moduli of purc sand (K,Tsand G.ys) and of thc
critical concentration mixture (Kc(. and Gc,c)as given by
cquation ( I 5 ) at C = @.y,y. The total porosity varics bctwcen
@,-,sand@,yy@,511, respcctively. The volumctric conccntration
of the purc sand end member in shaly sand is 1 - C/@,7.7
whilc that of the critical conccntration mixture is Cl@,5,y
(Figurc 13).
In shaly sand, thc shale particlcs fall in the pore spacc of
thc purc sand framework and do not significantly affcct its
stiffncss. Therefore, in order to calculate the cffcctivc elastic moduli of shaly sand, it is logical to conncct the clastic
moduli of thc end membcrs by the HSLB curve where the
soft end member is purc sand

(20)

where Kcc and Gcc are KMlxand GMIX,

respectively, as given
by equation (1 5) at C = @s.y. The Reuss wcighted average for
Vp-only modeling, similar to equation ( I 9 ) is

where Mc = Kc,. + ( 4 / 3)G,.,. .

The Rcuss lower-bound may significantly differ from
HSLB if the elastic contrast between thc cnd members is
large (e.g., the contrast betwecn pure quartz and porous
shale). However, if thc clastic contrast is relatively small (as
bctwccn purc sand and thc critical conccntration mixturc),
cquations (20) and (21) provide rcsults that arc close to each
other. Thcrcforc, thc use of the Rcuss bound is justified for
thc purpose of calculating thc clastic modulus of shaly sand.
ELASTICITY OF BIMODAL
MIXTURES-LAMINAR MODE

In the laminar mode, the effcctivc elastic moduli of thc

sandlshalc mixture vary monotonically between thosc of
thc pure sand and pure shale cnd mcmbcrs. If thc purc sand
and pure shale bodies arc arranged in an elastically isotropic
configuration, thc cffective elastic moduli of the laminar
mixture lie between thc lower and upper Hashin-Shtrikman
bounds (Mavko ct al., 1998).
In traditional geophysical applications, an clastic wavc
propagatcs approximately pcrpcndicularly to sand and
shale layers. Such a laycrcd configuration is anisotropic.
Its elastic constants arc givcn by the Backus averagc
(Backus, 1962). The exact solution for thc compressional
clastic modulus in thc dircction pcrpcndicular to thc laycrs
is thc Reuss averagc of the moduli of thc laycrs

Thc compressional modulus is the product of density and

P-wave velocity. Therefore, equation (22) is not equivalent
to the Wyllic's popular travel timc average equation
r

where Vp
I

7 - 1

Vp-ss, and Vp-Lsll arc for the P-wave vclocity in

Uplifted Shales

1000

2000

3000

Depth (m)

FIG. 12 Poisson's ratio (as calculated from P- and S-wave log

data) in shale versus depth for selected wells. Different shades
of gray corresponds to different wells.

I92

FIG. 13 The sum of two end members (pure sand and critical
concentration mixture) produces shaly sand.

PETROPHYSICS

May-June 2002

Grain Sorting, Porosity, and Elasticity

the laminar sandshale mixture, pure sand, and pure shale,

respectively. However, if the elastic contrast between sand
and shale is not large (which is very often the case), equations (22) and (23) give essentially the same result.

where K1 and KF are the bulk moduli of the grain material

and pore fluid, respectively.
The elastic moduli of the grain material can be calculated
from those of the mineral constituents via ad-hoc Hills
average

ELASTICITY OF PURE END MEMBERS

In many applications, the geophysicist will pick the elastic properties of pure sand and pure shale from well logs
(see, e.g., curves in Figure 2 where the pure sand values and
pure shale values correspond to the lowest and highest GR
values in the interval, respectively). Afterwards, these endmember values can be used in the mixing equations given
above.
In case where the elastic properties of the pure sand and
shale end members are unknown, the uncemented (friable)
sand equations (Dvorkin and Nur, 1996) based on the
Hertz-Mindlin contact theory can be used to estimate them.
The elastic moduli of the dry frame of sand are

where Mis either bulk or shear modulus; subscript i stands

for i-th mineral constituent; and m is the number of mineral
constituents.
Shale is not a granular composite such as sand. Therefore, the validity of applying equation (24) to pure shale is
not obvious. However, there is evidence that these equations provide reasonable elastic property estimates (see
Gutierrez et al., 2001, and example below).
To use equation (24) in the pure shale case, the subscript
SS has to be replaced by SH and the shale grain elastic
moduli G2,u2,and K 2 have to be used instead of GI, u1, and
K1 .
APPLYING ROCK PHYSICS THEORY

The first example is based on data from the Gulf of Mexico. Figure 4 shows the results of applying the sandshale
where GI andq are the shear modulus and Poissons ratio of
the grain material, respectively, P is the effective pressure
that is the difference between the overburden and pore pressure, and nss is the coordination number (the average
number of contacts per grain). The coordination number
depends on porosity. Its upper bound can be estimated from
an empirical equation (after Murphy, 1982)

nss = 20 - 34#ss

+ ~ 4 # : ,~

(25)

where porosity is in fractions of unity. The spread of data

below this curve may reach two coordination numbers (Figure 14). Eventually, the coordination number has to be Calibrated by adjusting model results to site-specific data with
equation (25) serving as a guideline.
The elastic moduli of saturated sand are calculated from
those of the dry frame via Gassmanns (195 1) equations
#SSKSS-D,

Kss =K1

- (1 + $ S S I K F

KSS-Diy

I K1

.-6 8
.&

g 6

.3

+ KF

.4

.5

Porosity

= GSS-Diy ,

(26)

f 10
3
2

(1 - #SS)KF+ # s s 4 - KF K s s - D ~1 ~
KI
GSS

May-June 2002

&

FIG. 14 Coordination number versus porosity. The solid black

curve is from equation (25). The gray domain shows possible
spread in coordination number values below the upper bound
given by equation (25).

PETROPHYSICS

193

iiiixturc cquations to thc log data from Figurc 2. Both thc

pure sand and purc shale end incinbers wcrc pickcd from tlic
coinprcssioiial modulus vcrsus porosity cross plot. Thc
dispcrscd-shalc thcorctical curvcs wcre calculated from
cquations (I8) and (2 1 ). l'hc valuc of MI was 100 GPa (purc
quartz) and Ihc valuc ofv, was 0.45, according to the original log data.
Equation ( 1 9) was also uscd to calculatc tlic coinprcssioiial modulus at thc critical conccntration point. Thc rcsult
falls below the value givcn by cquation ( I 8) but still b''lVCS a
rcasonablc cstimatc for tlic clastic modulus at critical conccntration and, in principle, can bc uscd for modcling.
Tlic laminar shalc curvc was calculated from thc Backus
avcragc, a s givcn by cquation (22), and also from Wyllie's
time avcragc, as given by cquation (23). Thc two rcsults are
practically idcntical.
Thc ncxt cxaniplc is from a vcrtical well in La Cira Ficld
in Colombia (Guticrrcz, 2001). Thcsc wcll log data span a
deptli intcrval from I50-to-600in that includes shalc
scqucnccs and fluvial sand bodies. The comprcssional
modulus that is plottcd vcrsus total porosity in Figiu-c I5
cxhibits thc familiar dispcrscd-shalc V-shapc.
I n this examplc, cquations (24) through (26) wcrc
applicd dircctly, without picking thc cnd-mcinbcr clastic

150-600 m

M Pa

B lo1

--

0.1

0.2

DEPTH TRENDS

Lct us usc tlic sand/shale inixturc model to cxplorc how

thc porosity and clastic properties of sand/shalc inixturcs
cvolve with dcplli and compaction. The cffcct of compaction on porosity @ can bc approxiinatcd by an cxponcntial
function

where Z is dcptli; @o is porosity at Z = 0; and a is a fitting

cocfficien t .
Thc coefficients in cquation (28) are site-spccific. Also,
in upliftcd and crodcd cnvironments, Z = 0 may not corrcspond to the currcnt-limc zcro depth and may be, in fact,
ncgativc.
According to Allcn and Allen ( 1 990), compaction cocfficients appropriate for North Sea basins are = 0.63 and a
= 0.5 1 kin I for shalc, and @() = 0.49 and a = 0.27 kin I for
sand. These paraineters wcrc used to construct the porosity
vcrsus shalc contcnt and dcptli trend shown in Figurc 16,
Icft, for both dispcrscd and laminar mixing modes.
Thc corrcsponding vclocity trends for shale and sand cnd
niembcrs can bc detcrinincd froin site-specific wcll log data
or thcorctically cstiinatcd from cquation (24). The result of
tlic latter approach is shown in Figurc 16, right.

,Pal

CONCLUSION

0.3

Porosity
FIG. 15 Compressional modulus versus total porosity for a La
Cira well. The depth interval spans from 150 to 600 m. The two
dispersed shale V-shaped theoretical curves are shown for the
effective pressure of 2 MPa and 10 MPa.

194

propertics from tlic cross-plot, to cxplore thc accuracy of

cquation (24).
It was assuincd that thc sand grains wcrc quartz, with Kl
= 37 GPa and G I = 45 GPa, and thc shale grains wcre clay
with K2 = 2 1 CPa and G2 = 8 GPa (SCCMavlto ct al., 1998,
for mineral clastic moduli). The purc sand and purc shale
end-inembcr porosity valucs havc been sclcctcd as @,y.~ = 0.3
and @.sf, = 0.2, rcspectivcly. Thc corresponding coordination numbcr values from equation (25) are 17,~,5= 1 1 and n ~ l=/
14, rcspectivcly. Thc bull<modulus of thc watcr in the pore
space was 2.5 GPa, calculatcd according to the sitc-specific
salinity.
Thc rcsults of inodeling using the cffcctivc prcssurc o f 2
MPa (for thc shallow part ofthc interval undcr examination)
and 10 MPa (for thc dccp part) arc supcrimposed on thc data
in Figurc 15. Thc dispcrsed shalc V-sliapcd curves mimic
the data.

Data show that the clastic and bulk propertics of

sand/shalc niixturcs depend on the mixing inodc which may
inakc the mixturc very unsiinilar to thc initial cnd members,
in all rcspccts. Thc above examplcs of the effcct of this
co1nplcxity on density, porosity, and elastic-wave vclocity

PETROPHYSICS

May-June 2002

Grain Sorting, Porosity, and Elasticity

are supplemented by sandlshale permeability data (Figure

6, last frame), where the permeability of the shale end member is much smaller than that of the sand end member,
although the porosity of the shale is larger than that of the
sand.
An effective medium model introduced in this paper
quantitatively relates the elastic properties of sandhhale
mixtures to porosity and mixture topology. The effective
elastic moduli of a mixture can be calculated from those of
the pure sand and shale end members. An approximate solution is given where the compressional-wavemodulus of the
mixture can be calculated from the compressional moduli of
the end members without the knowledge of the shear
moduli.
The equations presented in this paper can be used for
designing rock physics transforms between the elastic and
bulk properties of sandhhale depositional sequences and,
eventually, creating porosity and lithology volumes from
volumes of seismic data.
ACKNOWLEDGMENT

This work was supported by Phillips Petroleum and the

Stanford Rock Physics Laboratory. The data were provided

by Lake Ronel Oil Company, Ecopetrol, and Norsk Hydro.

Constructive comments of Dr. S. Gelinsky helped improve
the manuscript.
REFERENCES
Allen, P. and Allen, J., 1990, Basin analysis: Principles and applications, Blackwell.
Backus, G. F., 1962, Long-wave elastic anisotropy produced by
horizontal layering: Journal of Geophysical Research, vol. 67,
p. 44274441.
Cumberland, D. J. and Crawford, R.J., 1987, The packing of particles, Handbook of powder technology, Elsevier.
Dvorkin, J. and Nur, A., 1996, Elasticity of high-porosity sandstones: Theory for two North Sea datasets: Geophysics,vol. 61,
p. 1363-1370.
Dvorkin, J., Berryman, J., and Nur, A., 1999, Elastic moduli of
cemented sphere packs: Mechanics of Materials, vol. 31, p.
461469.
Gassmann, F., 1951, Elasticity of porous media-Ueber die Elastizitaet poroeser Medien: KerteQahrsschrft der Naturforschenden Gessellschaft,vol. 96, p. 1-23.
Gutierrez, M. A., Dvorkin, J., and Nur, A., 2001, Textural sorting
effect on elastic velocities, Part I: Laboratory observations,
rock physics models, and application to field data: SEG 2001,
Expanded Abstracts.

Shale Content 0,2y0.5

0 0

bepth (km)

FIG. 16 Total porosity (left) and P-wave velocity (right) versus depth and shale content. Inputs for the elastic property modeling are
the same as in the example shown in Figure 16. The two-branch surfaces are for the dispersed shale mode. The intersections of
these surfaces with the vertical planes of zero and 100% shale content give the pure sand and pure shale compaction curves,
respectively. These curves are connected by single-branch laminar mode surfaces.
May-June 2002

PETROPHYSICS

195

Dvorkin and Guticrrez

Guticrrez, M. A,, 2001, Rock physics and 3D seismic charactcrization of reservoir hctcrogeneitics to iinprove rccovcry efficiency: Ph.D. thesis, Stanford University.
IHerron, S. L., Hcrron, M. M., and Plumb, R. A., 1992, Idcntification of clay-supported and framework-supportcd domains rroin
geochemical and geophysical well log data: SPE24726, p.
667-680.
Marion, D., 1990, Acoustical, mechanical, and transport propertics of scdiincnts and granular materials: Ph.D. thcsis, Stadord
University,
Mavko, G., Mukcrji, T., and Dvorkin, J., 1998, The rock physics
handbook, Tools ,jOr .seismic analysis in p o t m ~ sniedin: Cambridge University Press.
Murphy, W. F., 1982, Effects of microstructure and pore fluids on
the acoustic properties of granular scdiinentary materials:
Ph.D. thcsis, Stanford University.
Schlumbcrgcr, 1974, Log Intcrprctation, Voluine Il---Applications, Schluinberger Liinitcd.
Schlumbcrger, 1 989, Log Interpretation Principles/Applications,
Schlumbcrgcr Wirclinc and Tcsting.
Tlioinas, E. C. and Sticlson, S. J., 1975, The distribution of shalc in
sandstones and its effcct upon porosity, paper T in the 16th
Annual Logging Symposium Transactions: Society of Proressional Well Log Analysts.
Worthington, P. F., 1985, The evolution of shaly-sand conccpts in
reservoir evaluation: The Log Anu/yst, vol. 26, number I , p.
23-40.
Yin, H., 1992, Acoustic velocity and attenuation of rocks: Isotropy, intrinsic anisotropy, and strcss induced anisotropy: P1i.D.
thcsis, Stanford University.

196

ABOUT THE AUTHOR

Jack Dvorkin rcccivcd his PhD in 1980 and MS in 1974, both
i n Continuum Mechanics froin Moscow University. From 1989
until prcscnt he has been a Senior Research Scientist at the Rock
Physics Laboratory at Stanford Univcrsity. He founded PetroSoft
Inc. i n 1991 (currcntly Rock Solid Images) and Petrophysical Consulting Inc. in 1995 wherc he is a principal. Currently he is principal advisor in Rock Physics for Rock Solid Iinagcs. Dr. Dvorkiii is
author ofabout 9Ojournal articles, one invention, and two books.
contact: jack@rockphysics.com

PETROIHYSICS

May-June 2002