Transp Porous Med (2011) 86:537–557

DOI 10.1007/s11242-010-9638-0

Linear Elastic Wave Propagation in Unsaturated Sands,

Silts, Loams and Clays

Bettina Albers

Received: 26 October 2009 / Accepted: 23 July 2010 / Published online: 7 August 2010
© Springer Science+Business Media B.V. 2010

Abstract Due to its propitious material properties sandstone is the most studied porous
medium for the investigation of linear wave propagation. However, in practical applications
the behavior of other soil types, i.e., especially the three main soil types sand, silt, and clay,
are much more important. Therefore, the model for partially saturated soils introduced by
Albers (Habilitation Thesis, 2010a) is applied to 11 soil types classified in the German stan-
dard DIN 4220 to obtain information on the phase velocities and attenuations of the three
longitudinal waves and the shear wave appearing in such media.

Keywords Unsaturated soils · Wave propagation · Porous media · Main soil types

1 Introduction

The most investigated solid of partially saturated porous media is sandstone. In comparison
to other rock and soil types it possesses a high compressibility of the skeleton. Therefore,
sandstone is especially suitable for the use of elastic theories. Thus, the model used in this
paper has been developed originally for these media (for the development of the model see:
Albers (2010a,b), for the application to wave propagation (Albers 2009, 2010a)). However,
for practical applications the behavior of granular materials such as sands, silts, and clays is
even more important.
There exist some attempts of other authors to consider the wave propagation in granular
materials: Pham et al. (2002) investigate the wave velocities and the quality factors of clay-
bearing sandstones as a function of pore pressure, frequency, and partial saturation. They
use the theory developed by Carcione et al. (2000). This is a three-phase model which—as
the model used in this study—is based on Biot’s model but unlike many other theories it
considers the coexistence of two solids (sand grains and clay particles) and a fluid-mixture
in order to account for the clayey character of the soil. It yields the seismic velocities as a

B. Albers (B)
Institute for Geotechnical Engineering and Soil Mechanics, Technical University of Berlin,
Berlin, Germany
e-mail: [email protected]

123
538 B. Albers

function of clay (shale) content, porosity, saturation, dry-rock moduli, and fluid-solid prop-
erties. The authors recognize that the effect of partial saturation on the acoustic quantities
depends on the frequency range. Lo et al. (2005) derive a general set of coupled partial dif-
ferential equations to describe the dilatational wave propagation through an elastic porous
medium filled by two immiscible fluids. The authors proceed from a theory introduced by
themselves (Lo et al. 2002) which is based on the mixture theory (Truesdell 1984) and which
follows the approach of Gray (1983). It incorporates both inertial coupling and viscous drag
and is described in an Eulerian frame of reference. The field equations incorporate the cou-
pling among the fluid phases and the solid matrix caused by their relative accelerations. The
incorporation of relative accelerations is questionable. In fact, the authors state that “relative
acceleration ... possess(es) objectivity” and refer in this connection to Drew and Passman
(1999) and Truesdell (1972). However, it is easy to show that in principle a relative acceler-
ation is nonobjective. Thus, a change of the reference frame to a noninertial system yields
constitutive contributions in the momentum balances which follow from the presence of the
relative acceleration. They are additional to the usual centrifugal, Coriolis, Euler, and trans-
lational accelerations which are characteristic in noninertial frames (Liu 2002). Of course,
the violation of objectivity is a bothering feature, however, Wilmanski (2005) could show
that the Biot Model follows by linearization from a nonlinear poroelastic two-component
model and that the nonobjectivity in Biot’s equations follows from the linearization of such
objective nonlinear equations. The set of equations of Lo, Sposito, and Majer extends two
concepts for saturated porous media namely the linearized increment of fluid content (Biot
1956) and a closure relation for porosity change (de la Cruz et al. 1993). The latter is a linear
relation between porosity change and the dilatations of solid and fluid. After some simpli-
fications and under the assumption of constant volume fractions and material densities the
set of equations reduces to the model of Berryman et al. (1988). The theory is applied to the
example of Columbia fine sandy loam containing either an air-water or an oil-water mixture.
The influence of wave excitation frequency between 50 and 200 Hz and of saturation on
phase speeds and attenuation is considered. They observe three compressional waves. The
P1-wave which—as they claim—results from in-phase motions of the solid framework and
the two pore fluids, moves with a speed equal to the square root of the ratio of an effective
bulk modulus to an effective density of the fluid-containing porous medium, regardless of
fluid saturation and the type of mixture. The authors conclude that the effect of water satura-
tion on the P1-wave does not depend sensitively on the state of consolidation of the porous
medium. The other two waves are called diffusive modes and it is stated that the P2 -wave
results from out-of-phase motions of the solid framework and the fluids and the P3-wave
from capillary pressure fluctuations.
In a recent paper, Lo et al. (2007) theoretically study the wave propagation in several soil
textures. Based on the dispersion equation obtained in the Lo, Sposito, and Majer model (Lo
et al. 2005), the phase velocity and attenuation coefficient of P1- and P2-waves are deter-
mined as a function of water saturation for 11 soil texture classes. The slowest wave (P3)
has not been accounted for since it does not travel far due to very high attenuation. The cal-
culated phase velocity and attenuation coefficient for different soil textures were normalized
by those computed for sand. The normalization leads the resulting dimensionless parameters
to be frequency independent throughout the whole range (up to 500 Hz) with Darcy’s law
remaining valid for the description of each fluid flow under wave excitation. The normalized
phase velocity of the P1-wave was shown to have a substantially constant value at higher
water saturations, but in the lower saturation range it first increases to reach a certain max-
imum value for different soil types and then decreases. The maximum appears in a region
of saturation between 0 and 0.2. In my opinion this result is very artificial and in reality not

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 539

feasible because the authors did not account for the fact that for many soil textures which
they investigated this range of saturations does not exist at all. They calculated the velocities
and attenuations of the different soil types regardless the values of saturation appearing in
reality. For example, clays appear in reality in the range of saturations between around 60 and
100%. Thus, it does not make sense to investigate the behavior of waves in this soil texture
for saturations between 0 and 20%.
It is well known that it is a difficult task to determine general wave properties for dif-
ferent soil textures. For example one paragraph of Gebrande et al. (1982) is cited: “Elastic
wave velocities measured at atmospheric pressure on different samples of the same rock type
generally show large scatter, even when made by a single method. They depend not only on
petrological composition and rock fabric, but are influenced to a large degree by the presence,
size and orientation of pores and cracks, and by the presence or absence of interstitial fluids.
Without detailed information about these factors elastic wave velocities in a rock of known
composition can hardly be predicted.” Even if it is clear that the generalization which is
undertaken in this paper is not valid without exceptions it is attempted to find an approach
which is useful in practical applications.

2 Soil Classification

The classification of soils is nonuniform. Nearly every country has its own standards in which
the characterization is regulated. In the German standards alone at least two different clas-
sifications of soil types do exist. On the one hand there is the German standard DIN 18196
(2006) which is called “Earthworks and foundations—Soil classification for civil engineering
purposes” and on the other hand there is DIN 4220 (2005) named “Pedologic site assess-
ment—Designation, classification and deduction of soil parameters (normative and nominal
scaling)”. The reason for two different classifications might be that in soil mechanics and
civil engineering—and thus in view of the soil as the foundation for the construction of
buildings—where DIN 18196 is applied rather a conglomeration of several basic soil types
is studied while in geophysics where DIN 4220 is used also details on several minerals are
of interest. In this regard in DIN 18196 groups of soils are denominated which are mainly
characterized by grain size and civil engineering properties like shear strength, ability to com-
paction and compression or frost resistance. On the other hand in DIN 4220 rather pedologic
properties are of interest. So, for example, the water and air permeability, root penetration
or cation exchange capacity are covered. However, the latter standard does not only contain
these pedologic quantities but also some which are needed in the wave analysis. The decisive
factor for using here the classification according to DIN 4220 is the fact that for each of
the soil types also the parameters appearing in the van Genuchten formula and thus also the
capillary pressure curves are given. Thirty-one soil types are given in the standard. They
consist of different fractions of the three main soil types sand, silt and clay. According to
these fractions their short cuts are explained in Table 1 and illustrated in the triangle shown in
Fig. 1. The soil classification in America by the US Soil Taxonomy is done in an equivalent
way but the assignment of several sub soil types is much finer.

3 Determination of Material Parameters of Some Soil Textures

Some of the soil mechanical characteristic values can also be found in, e.g., the chapters of
the Geotechnical Engineering Handbook on properties of soils and rocks by von Soos (2001)

123
540 B. Albers

Table 1 Denotations, short cuts and grain size fractions of 11 chosen soil types (extract of Table 6 of DIN
4220—German standard DIN 4220 (2005))

Main soil Soil type Soil type Short cut Grain size fraction
type group group percent by weight

Clay Silt Sand

Sands Pure sands Medium sand mS 0–5 0–10 85–100

Loam sands Low clayey sand St2 5–17 0–10 73–95
Silt sands Strong silty sand Su4 0–8 40–50 42–60
Loams Sand loams Silty loamy sand Slu 8–17 40–50 33–52
Normal loams Low sandy loam Ls2 17–25 40–50 25–43
Clay loams Sandy clayey loam Lts 25–45 15–30 25–60
Silts Sand silts Sandy silt Us 0–8 50–80 12–50
Loam silts Medium clayey silt Ut3 12–17 65–88 0–23
Clay silts Strong clayey silt Ut4 17–25 65–83 0–18
Clays Silt clays Medium clayey loam Lt3 35–45 30–50 5–35
Loam clays Pure clay Tt 65–100 0–35 0–35

Fig. 1 Soil subtypes consisting of different fractions of the three soil types sand, silt, and clay illustrated
in a triangle coordinate system (in German from Homepage “Archäologie & Graphik” (www.nbgraphik.de/
archaeologie/3eck.html))

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 541

Table 2 Porosity, initial mass density, Poisson’s number and grain size fraction of 11 chosen soil types

Short Porosity Initial mass  Poisson’s Grain size fraction

kg
cut density ρ0S 3 number Percent by weight
m

Clay Silt Sand

mS 0.42 1,537 0.32 3 5 92

St2 0.42 1,537 0.33 11 5 84
Su4 0.43 1,511 0.34 4 45 51
Slu 0.43 1,511 0.35 13 45 42
Ls2 0.43 1,511 0.36 21 45 34
Lts 0.43 1,511 0.37 35 23 42
Us 0.44 1,484 0.35 4 65 31
Ut3 0.43 1,511 0.37 15 77 8
Ut4 0.44 1,484 0.38 21 74 5
Lt3 0.44 1,484 0.38 40 40 20
Tt 0.46 1,431 0.41 76 12 12

and on soil dynamics and earthquakes by Klein (2001) or in text books on soil dynamics and
geotechnics (e.g., Studer and Koller 1997; Schmidt 1996). However, these references do not
classify the soil types into the same groups as is done in DIN 4220 but in more general groups.
Furthermore, the parameters concerning properties of unsaturated soils are not incorporated.
Several parameters are needed for the wave analysis. Their determination is discussed below.

3.1 Porosity, Mass Density of the Skeleton

In Table 31 of DIN 4220 air capacity, usable field capacity and field capacity for the soil
types for different values of the oven-dry density are given. The oven-dry density is classified
in Table 12 of DIN 4220. Further a medium value of this quantity is assumed, namely the
class ρt3. In the standard the calculation formula for the total void fraction and thus for the
porosity (as volume fraction in %) is proposed:

Total void fraction = Field capacity + Air capacity. (1)

In this manner, the porosity for different soil types has been determined. For the medium
value of the oven-dry density the porosity for the different soil types lies in the range 0.42 ≤
n 0 ≤ 0.46 in which the values for the sands are smallest and those of clays are biggest (see:
Table 2).
The notion of porosity is closely connected to those of compactness and grain density
commonly used in geotechnics, namely
Compactness
Void fraction = 1 – . (2)
Grain density
Relation (2) is used—with somewhat different denotations – to determine the reference
mass density of the skeleton, ρ0S , which is equivalent to the notion compactness of soil
mechanics. Namely, if we denote the grain density by its counterpart of continuum mechan-
ics—the initial true mass density of the skeleton, ρ0S R , then (2) yields with n 0 being the initial
porosity

123
542 B. Albers

ρ0S = (1 − n 0 ) ρ0S R . (3)

The mass densities of the different soil types are obtained under the assumption of the
same ρ0S R = 2,650 kg/m3 for all soil types (they are given in Table 2).

3.2 Compressibility of the Skeleton, Shear Modulus, Poisson’s Number

Generally, for the compressibility modulus of the skeleton (regardless of the type) the value
K s = 35 GPa has been chosen. However, it turned out that this value may be too high if
the soil is clayey. Vanorio et al. (2003) state that the presence of clay minerals can alter
the elastic behavior of rock significantly. They report in their article of theoretical val-
ues for the bulk modulus of clay between 20 and 50 GPa. These values have been widely
adopted in the literature because measurements of clay minerals had proven to be diffi-
cult. Elastic properties of clay minerals are almost unknown, mainly due to their intrinsic
properties. Their small grain size makes it impossible to isolate an individual crystal and
to measure acoustic properties. According to Vanorio et al. (2003) the only experimental
measurement of the bulk modulus in a clay mineral gives a much lower value of 6.2 GPa.
The measurements described in Vanorio et al. (2003) provided a value in-between, namely
K s = 12 GPa and a shear modulus μ S = 6 GPa. Furthermore, Vanorio et al. (2003) report
that laboratory data from several investigators show a systematic decrease of both P- and
S-wave velocity with increasing clay content in both well and poorly consolidated sediments.
Klimentos and McCann (1988) argued in their paper on saturated porous media that also the
P2-wave is influenced by the clay content. They presented observations of the slow wave
in artificial rocks made of cemented sand grains and demonstrate that this wave becomes
attenuated very rapidly when the pore space contains clay particles. Gudehus and Külzer
(2002) showed that there is a practical importance of the fraction of finest particles: it can
be proven by model tests that in very fine soils with inclosed gas bubbles slides and mud
volcanoes may appear in the same way as they do in situ.
The value of Poisson’s number has been estimated according to the fractions of the three
main soil types. The fractions of clay, silt and sand are given in Table 6 of DIN 4220.
Poisson’s numbers for these main soil types have been proposed by Girsang (2001) to be
νsand = 0.3, νsilt = 0.366, νclay = 0.414. However, after comparison of values appearing in
books on soil mechanics the following values have been preferred in the present approach:
νsand = 0.31, νsilt = 0.37, νclay = 0.43. (4)
Using them and approximately the median of the fractions of the main soil types (but
taking into account that the fractions do have to sum up to 100%) from Table 1 the 11 soil
types possess the values of Poisson’s number given in Table 2.
In Albers and Wilmanski (2005) it has been discussed for a two-component porous medium
that the shear modulus, μ S , does not depend on any coupling parameters but only on Pois-
son’s number and on the compressibility modulus of the dry matrix, K d . Thereby, of course it
depends on the initial porosity n 0 since K d is used in the form proposed by Geertsma (1957):
K d = K s / (1 + 50n 0 ) . As done in almost all approaches on partially saturated porous media
it is assumed that the shear modulus does not depend on the degree of saturation. Thus, the
shear modulus can be expressed by
3 1 − 2ν Ks
μS = . (5)
2 1 + ν 1 + 50n 0
Of course, this form of the shear modulus would be not applicable if the linear elastic
behavior of the skeleton had not been presumed. Possibly, for granular media this assumption

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 543

Fig. 2 Range of capillary pressure curves for sands, silts, and clays. Modified figure from Hartge and Horn
(1999)

is oversimplifying (White 1983). Furthermore, since μ S is proportional to K s which is even-

tually chosen too high also the shear modulus may be too large. In contrast to the other soil
types the grains of clay are not nearly spherical but plate-like. Thus, they can be oriented and
this influences the compressibility and shearing properties.

3.3 Soil–Water Characteristic Curve

As indicated in Fig. 2 capillary pressure curves for several soil types differ considerably. The
most convenient method to obtain capillary pressure curves is their direct measurement. If
this is not possible, one has to use already existing data. In this paper, the approach of van
Genuchten is chosen because the determination of the three parameters appearing in this for-
mula has been already widely studied. The van Genuchten parameters have been incorporated
for the classified 31 soil types into the German standard DIN 4220; they are given in Table
37 of the standard. The van Genuchten (1980) approach is used in its original formulation:
θs − θr
θ (ψ) = φr +  m , (6)
1 + (α |ψ|)n
where θ is the water content given as a volume fraction in % as a function of the matrix
potential ψ in hPa. The residual water content is denoted by θr , the water content in
case of full saturation by θs . The parameters α, n and m are the van Genuchten param-
eters. The parameter m is not given in the tables of DIN 4220 but calculated from n by
m = 1 − 1/n. Additionally, in Table 34 of DIN 4220 measured values of the retention curves
are given. A combined illustration of the curves for the 11 chosen soil types, converted into
the present notation, is given in Fig. 3.
In the case that further soil types have to be considered for which retention data are avail-
able but no analytical functions, the paper of van Genuchten and Nielsen (1985) presents
several closed-form predictive equations for the unsaturated hydraulic conductivity in terms
of parameters that can be fitted to observed data. A further method to obtain the parameters
needed in the van Genuchten formula is the application of the R E T C code by van Genuchten,

123
544 B. Albers

Fig. 3 Capillary pressure curves for chosen soil types. Lines: calculated according to the van-Genuchten-
formula under consideration of the values given in Table 37 of DIN 4220, symbols are the measured values
given in Table 34 of DIN 4220

Leij, and Yates (van Genuchten et al. 1991). Of course, it is also possible to search through
the literature for appropriate experimental curves and to determine the parameters oneself.

3.4 Compressibilities of Fluid and Gas, Viscosity, Permeability

The microscopic compressibilities of water and air are assumed not to change by consider-
ation of different soil types. They are

K f = 2.25 × 109 Pa, K g = 1.01 × 105 Pa. (7)

For the wave analysis the incorporation of relative resistances is necessary. They are given
by

πF πG
π FS = , πGS = , (8)
kf kg
where the formulae for the relative permeabilities
1
   
1 m 2 1
 
1 2m
kf = S2 1− 1− Sm , k g = (1 − S) 3 1 − S m , (9)

again, are given by van Genuchten (1980). The parameter m is one of the van Genuchten
parameters which are arranged in Table 3 for the chosen soil types. The viscosities of water
and air are implicitly incorporated in the parameters π F and π G . While the permeability or
resistance with respect to the gas is not influenced by the soil type, the permeability with
respect to the fluid differs. Thus, the value π G = 1.82 × 105 Pa is the same for all soil types
while π F is different. This value can be deduced from the hydraulic conductivity in water-
saturated soil which is also given in Table 3 (part of Table 35 of DIN 4220). The permeability,
k, is related to the resistance, π F , by

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 545

Table 3 Van-Genuchten-parameters in dependence on soil type (extract of Table 37 of DIN 4220 (2005)
supplemented by the third van-Genuchten-parameter m = 1 − n1 ); medium permeability in water-saturated
soil for a medium value of the oven-dry density (see Table 35 of DIN 4220)

Soil type van Genuchten parameters Saturated water

short cut permeability k in m/s
θr θs α n m

Slu 0.0225 0.4152 0.0425 1.21240 0.17519 3.24 × 10−6

St2 0.0394 0.4022 0.09725 1.38158 0.27619 1.37 × 10−5
Su4 0 0.4158 0.03996 1.24244 0.195139 4.40×10−6
Ls2 0.1644 0.4157 0.02664 1.50608 0.33602 2.66 × 10−6
Lt3 0.2312 0.4274 0.03005 1.28574 0.22224 8.10 × 10−7
Lts 0.1739 0.4165 0.03713 1.25012 0.20008 1.16 × 10−6
Us 0 0.4025 0.0136 1.23252 0.18865 2.55 × 10−6
Ut3 0 0.4053 0.00781 1.23052 0.18734 1.39 × 10−6
Ut4 0 0.4115 0.00982 1.17848 0.15145 1.50 × 10−6
Tt 0 0.4301 0.00837 1.08011 0.07417 3.47 × 10−7
mS 0.036 0.4071 0.04732 1.75418 0.42993 5.67 × 10−5


k 1 n 0 · 1000 · 9.81 kg
∼ F, i.e. π F
= . (10)
ρg π k m3 s

4 Governing Equations

In this section, the macroscopic continuum model for three-component media is introduced
(see: Albers 2010b). The partial stress tensors satisfy the following constitutive relations

T S = T0S + λ S e1 + 2μ S e S + Q F ε F 1+Q G ε G 1, e := tr e S ,
T F = − p F 1, p F = p0F − ρ0F κ F ε F − Q F e − Q F G ε G , (11)
T = − p 1,
G G
p =G
p0G − ρ0G κ G ε G −Q e−Q
G
ε ,
FG F

where e S is the macroscopic deformation tensor, the volume changes of the skeleton,
fluid and gas are denoted by e, ε F and ε G . The constant tensor T0S is the initial partial
stress in the skeleton, p0F is the initial partial pressure in the fluid and p0G the initial par-
tial pressure in the gas. These quantities as well as the macroscopic material parameters
λ S , μ S , κ F , κ G , Q F , Q G , Q F G are functions of an initial porosity n 0 and an initial satura-
tion S0 . The parameters λ S and μ S correspond to the classical Lamé constants and κ F , κ G
correspond to compressibility coefficients of an ideal fluid and of the gas, respectively. The
macroscopic material parameters have to be obtained from experimentally measurable ones
by use of a micro-macro-transition procedure (see: Albers 2010a,b). In Albers (2010a) for
each of the 11 above mentioned soil types detailed information on microscopic material
parameters, the successive macroscopic material parameters and phase velocities in depen-
dence on frequency and saturation are given.

123
546 B. Albers

The linear constitutive relations for the sources have the following form
   
p̂ S = π F S v F − v S + π G S vG − v S ,
   
p̂ F = −π F S v F − v S , p̂G = −π G S vG − v S . (12)

The resistances π F S and π G S are also constants depending on the initial porosity n 0 and
the initial saturation S0 . The volume changes of the components are related to their mass
densities by

ρ0S − ρ S ρ0F − ρ F ρ0G − ρ G

e= , εF = , εG = . (13)
ρ0S ρ0F ρ0G

The linear thermodynamical model without memory effects can be described by the essen-
tial fields

v S , v F , vG , e S , ε F , ε G , (14)

which have to satisfy the field equations

∂v S
ρ0S = div λ S e1 + 2μ S e S + Q F ε F 1+Q G ε G 1
∂t    
+π F S v F − v S + π G S vG − v S ,
∂v F  
ρ0F = grad ρ0F κ F ε F + Q F e + Q F G ε G − π F S v F − v S , (15)
∂t
 
G ∂v
G
ρ0 = grad ρ0G κ G ε G + Q G e + Q F G ε F − π G S vG − v S ,
∂t
∂e S ∂ε F ∂ε G
= sym grad v S , = div v F , = div vG , e ≡ tr e S .
∂t ∂t ∂t

This set of equations coincides with the classical Biot model, if we neglect the third
component, i.e., the gas.

5 Wave Velocities and Attenuations

The linear wave analysis (see: Albers 2009) leads to the dispersion relations for the transversal
wave
   
μS k 2 F S G S ρ0 + ρ0 + ρ0
S F G
μS k 2
ω 1− S
2
−π π 1− S
ρ0 ω ρ0S ρ0F ρ0G ρ0 + ρ0F + ρ0G ω
⎡ ⎤
  π FS πGS 2
π +π
F S G S π F S π G S + μ S
⎢ ρ F ρ0
G k ⎥
+iω + F + G ⎣1 − F S G S0 ⎦ = 0,
ρ0S ρ0 ρ0 π +π π FS
+ F + G 0 πGS ρ S ω
S ρ0 ρ0 ρ0
(16)

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 547

and for the longitudinal waves

  
λ S + 2μ S k 2 λ S + 2μ S  F 
ω 1−
2
+ κ F
+ κ G
+ κ + κG + κ F κG
ρ0 S ω ρ0S

2 2 2
   2

ρ0S Q F G + ρ0F Q G + ρ0G Q F k 4 λ S + 2μ S Q F G
− + −κ κF G
ρ0S ρ0F ρ0G ω ρ0S ρ0F ρ0G
2 2
  
QF κG QG κ F Q F QG Q FG k 6
+ S F + S G −2
ρ0 ρ0 ρ0 ρ0 ρ0S ρ0F ρ0G ω
  
π FS + πGS π FS πGS λ S + 2μ S κ G κF
+iω + F + G − + F
ρ0S ρ0 ρ0 ρ0S ρ0G ρ0
 FS     
π + πGS κ F + κG π FSκG + π GSκ F π FS QF π GS QG k 2
+ + +2 S F +2 S G
ρ0S ρ0F ρ0 ρ0 ρ0 ρ0 ω
   
λ S + 2μ S π F S κ G π GSκ F π FS + πGS κ F κG
+ + +
ρ0S ρ0F ρ0G ρ0S
π FS  2 2
 πGS  2 2

− S F G Q G + Q F G + 2Q G Q F G − S F G Q F + Q F G + 2Q F Q F G
ρ0 ρ0 ρ0 ρ0 ρ0 ρ0
  
4
π GSκ F QG π FSκG QF k
+ 2 +2
ρ0 ρ0
S G ρ0 ρ0
S F ω
ρ0S + ρ0F + ρ0G λ S + 2μ S π F S π G S
−π F S π G S +
ρ0S ρ0F ρ0G ρ0S ρ0F ρ0G
 
π FSπ GS  F F   k 2
+ S F G ρ0 κ + ρ0 κ + 2 Q + Q + Q
G G F G FG
= 0. (17)
ρ0 ρ0 ρ0 ω

The dispersion relations are solved for the complex wave number k. The results of k reveal
both the phase speeds of the body waves c = ω/ (Re k) and the corresponding attenuations
Im k. The analysis is carried out for the above chosen 11 types of soil containing an air-
water-mixture in their voids. In the following the phase velocities and attenuations of the
shear wave, S, and the three longitudinal waves, P1, P2, P3, are compared.

5.1 Dependence on the Frequency

5.1.1 Phase Velocities

First, the dependence of the phase velocities on the frequency is considered. Figure 4 shows
this dependence for the four waves appearing in the respective soil type whose voids are
partially filled by water and partially by air. In the upper left panel the phase velocity of the
shear wave, S, is illustrated. It is obvious, that for all soil types except for the medium sand
(mS) the frequency does not have any influence. The velocity is constant in the range of fre-
quencies. This is true even though a wide range of frequencies between 10−2 Hz ≤ ω ≤ 107
Hz has been considered. The sand—as it is the case for sandstone—shows different high and
low frequency limits. The value of this wave velocity is quite different for the several soil
types: it ranges in the low frequency limit from approximately 390 m/s for clay (Tt) to nearly

123
548 B. Albers

Fig. 4 Dependence on the frequency of the phase velocities of the four waves appearing in 11 different soil
types filled by an air–water-mixture. The degree of saturation is 80%

600 m/s for sand (mS). The shear wave velocity is mainly driven by the mass density of the
soil and by the shear modulus which is a function of both porosity and Poisson’s number
which are different for the various soil types. Thus, according to these values the velocities
of the other soil types are ranging in-between.
The same qualitative behavior is observable for the fastest compressional wave, the
P1-wave. However, the velocities of these waves are higher than these of the S-wave. They
range in the low frequency limit from a little bit more than 1,000 m/s for clay (Tt) and nearly
1,150 m/s for sand (mS). This wave is, additional to the just mentioned parameters which
influence the S -wave, driven by the elasticity of the skeleton, λ S . This parameter is constant
for almost all values of the saturation and, thus, also the P1-wave velocity behaves in this
way.
The P2-wave velocity for all soil types starts with zero for zero frequency and ends up at a
high frequency limit which is almost the same for all soil types (≈280 m/s) only for the sand
it is a little bit higher (≈310 m/s). The P2-wave is driven mainly by the compressibilities of
the pore fluids. Since in all soil types the same pore filling is considered the wave propagates
with similar velocity in all soil types.
The P3-wave is the slowest compressional wave. It is attributed to the capillary pressure.
The shape of the capillary pressure curve strongly varies for the different soil types and, thus,

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 549

Fig. 5 Dependence on the frequency of the attenuations of the four waves appearing in 11 different soil types
filled by an air–water-mixture. The degree of saturation is 80%

also the velocities are not the same. They have in common that for zero frequency they start
with zero velocity. However, in the high frequency limit they reach values between 0.2 m/s
(Tt) and 3.5 m/s (mS). It is obvious, that this wave is much slower than the three other waves.

5.1.2 Attenuations

Due to large differences in the order of magnitude of the attenuations in the limits of the
frequency which appear for all four waves in Fig. 5 the dependence of the attenuations on the
frequency is plotted logarithmically. In this scale, the attenuations of all soil types increase
linearly in the range from small frequencies to a very large value of the frequency. For
some soil types between 106 Hz ≤ ω ≤ 107 Hz the linearity is lost and these curves do not
increase further. Even if the behavior of the attenuation looks quite similar for the four waves,
their values are extremely different. While for the S- and P1-wave very low values appear
(approximately between 10−15 1/m and 102 1/m) the P2-wave is much stronger attenuated
(5 × 10−1 1/m to 104 1/m) and the P3-wave even more (5 × 101 1/m to 5 × 107 1/m).
Again, the curves for the P2-wave lie in a very narrow range while for P1-, S- and P3-wave,
respectively, differences are bigger. While for S-, P1-, and P2-wave the sand yields the
highest values and the clay the lowest for the P3-wave it is opposite.

123
550 B. Albers

Fig. 6 Dependence on the initial saturation of the phase velocities of the four waves appearing in 11 different
soil types filled by an air–water-mixture. The frequency is 1,000 Hz

5.2 Dependence on the Initial Saturation

5.2.1 Phase Velocities

Next, the dependence of the phase velocities of the different soil types on the degree of
saturation is studied. It is illustrated in Fig. 6. It is obvious that the various soil types exist
in different ranges of saturation. For example, clay (Tt) does exist only in the range 0.58 ≤
S0 ≤ 1. For smaller degrees of saturation the soil particles cannot stick together and the
soil type is not defined anymore. The different ranges of existence of the 11 soil types are
especially obvious in the upper left panel of Fig. 6 in which the velocities of the shear wave
in dependence on the initial saturation are given. It was already mentioned that the shear
modulus is assumed to be constant in the range of saturation and only depends on Poisson’s
number, microscopic compressibility of the skeleton and porosity. Since these quantities are
constant for each soil type the mass densities are the only quantities by which the shear wave
velocity is driven. Thus, the velocities for the different soil types are linearly decreasing with
increasing saturation. The velocity for the sand is largest while clay has the lowest.
As it was already observed for the waves in sandstone (Albers 2009) the P1 -wave behaves
linearly slightly decreasing for almost all values of the saturation. Only for very high values
of the saturation there appears a very abrupt increase to nearly the double of the value for
the other degrees of saturation. P1-wave and P2-wave for some of the soil types exhibit a

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 551

Fig. 7 Dependence on the initial saturation of the attenuations of the four waves appearing in 11 different
soil types filled by an air–water-mixture. The frequency is 1,000 Hz

further increase of the velocity in the region of smaller saturations. While for the P2-wave
this is acceptable because also for the air–water-mixture in sandstone the velocity exposes a
deep minimum which is experimentally confirmed, for the P1-wave this result seems to be
rather questionable. Namely, for the other waves the wave velocity behaves linearly slightly
decreasing in the region of small saturations. The increase for the first mentioned soil types
(St2, Su4, Slu, Us, Ut3, Ut4, Tt) can be attributed to an increase of the elastic constants,
λ S + 23 μ S , in the region of these saturations. This, in turn, is caused by a sharp decrease of
the parameters Q G and Q F G . The reason for this diminution is unclear, however, it seems
to be observable for those soil types which consist of large fractions of particles with small
grain size (clay or silt). The velocity of the P3-wave does not exhibit such distinct features.
This wave for all soil types shows a maximum for a certain saturation and does not exist in
the limits of the saturation. The degree of saturation for which the maximum appears depends
on the soil type. It is shifted in the direction of water saturation the smaller the grain size is
(for sand, mS, it is around 90% while for clay, Tt, it is approximately 99%). Simultaneously,
the maximum velocity is smaller (ca. 0.16 m/s for mS, 0.023 m/s for Tt).

5.2.2 Attenuations

At the first sight, the dependence of the attenuations of the four waves on the saturation
appears somewhat confusing (Fig. 7). However, a closer look shows, that S- and P1-wave

123
552 B. Albers

behave similar. Both attenuations for all soil types show a clear minimum. The values of these
attenuations are very small. They range between 10−9 and 10−2 1/m. The curves for clay lie
below these for sand, i.e., predominantly the attenuations of soil types with smaller particles
are smaller than of those with larger particles. Again, the degree of saturation for which
the minimum appears is dependent on the soil type. For sand it is nearer to air saturation
(S0 ≈ 30%), for clay to the other limit—water saturation (S0 ≈ 90%). For clay the minimum
is much more pronounced than for sand. Loam sand (St2) shows a strange behavior for low
values of the saturation. There appears a second—strong—minimum which does not appear
for any of the other soil types. Possibly this is an artefact.
The qualitative behavior of the attenuation of the P3-wave is similar. It also shows a
minimum. However, the degree of saturation for which this minimum appears lies for all soil
types in the region of very high saturations whereas it is much nearer to water saturation for
clay (S0 ≈ 99%) than for sand (S0 ≈ 90%). It appears for the same value of the saturation at
which the maximum in the velocity of the P3 -wave appears. For this wave, admittedly, the
values for sand are lower than these for clay. However, the attenuation of the P3-wave is the
biggest of all four waves. Values are larger by order of magnitudes than the other attenuations.
This is consistent with the very low velocity of this wave which had been discussed above.
Due to its enormous high attenuation the observation of this wave in natural media, but
presumably also in artificial media, will be nearly impossible. However, P3-waves influence
the structure of surface waves and it cannot be ruled out that some surface modes caused
by P3-waves can be observed. This requires an extensive and complex numerical analysis.
Already the detection of the P2-wave whose attenuation is smaller by some degrees turned
out to be difficult. The attenuation of the P2-wave behaves opposite to the other attenuations.
For this wave a maximum instead of a minimum appears. It appears nearly at the same degree
of saturation for which the minimum in the P3-wave attenuation arises. This means the curve
is much steeper for clay than it is for sand. The decrease for some soil types in the region of
small saturations, again, is questionable.

5.3 Comparison of Three Chosen Soil Textures

Figure 8 provides a comparison of the results on wave velocities and attenuations in sandstone
filled by an air–water-mixture (black curves) with those on sand (mS, red curves) and silty
clay (Lt3, green curves). The figure shows the phase velocities (left) and attenuations (right)
of shear wave, P1-, P2-, and P3-wave in dependence on the initial saturation. Different
curves represent several geotechnically relevant frequencies between 10 and 5,000 Hz.
Phase velocities of S- and P1-wave are independent of the frequency and—except for the
rapid increase near water saturation—nearly constant with change of saturation, however,
for the three soil types they differ in size. For both waves the velocity in sandstone is biggest
while this of clay is the smallest. The shear wave velocity of sand is approximately one half
of the sandstone velocity, for silty clay it is around one fourth. For the P1-wave the curves
of sand and clay are closer. These waves propagate with approximately 65% (Lt3) and 70%
(mS) of the P1-velocity in sandstone. As mentioned before the attenuations of S- and P1-
wave are very small. In contrast to the velocities they do not differ in same extent between
the three soil types.
For the P2-wave this is the case both for the velocity and for the attenuation: they are
of same order for all three soil types. However, both for velocity and for attenuation it is
obvious that the curves for sandstone are much smoother than these for sand or even clay.
The minimum in the velocity and the maximum in the attenuation move in comparison to

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 553

Fig. 8 Comparison of the phase velocities (left column) and the attenuations (right column) of the four waves
S, P1, P2 and P3 appearing in sandstone (black), sand (red), and silt clay (green) filled by an air–water-
mixture in dependence on the initial saturation

123
554 B. Albers

Table 4 Extract of a table of

Soil type vT (m/s)
Schultze and Muhs (1950)
specifying shear wave velocity,
Sands, depending on condition 100–250
vT
Sand and gravel 180–550
Loam 150–200
Clays dep. on water content 120–700
Keuper sandstone, weathered 250
Bunter, nonweathered 1,100

Table 5 P-wave velocities of

Medium Velocity [m/s]
some types of geologic media
Min Max

Air (dep. on temp.) 310 360

Weathered soil horizon 100 500
Gravel, dry sand 100 600
Loam 300 900
Wet sand 200 1,800
Clay 1,200 2,500
Water (dep. on temp.) 1,430 1,590
Friable sandstone 1,500 2,500
Dense sandstone 1,800 4,000
Chalk 1,800 3,500
Limestone 2,500 6,000
Marl 2,000 3,500
Gypsum 4,500 6,500
Ice 3,100 4,200
Granite 4,000 5,700
From Pavlovic and Velickovic Metamorphosed rock 4,500 6,800
(1998)

sandstone towards higher values of the saturation for sand and even more for clay. The
more the extremum scrolls towards water saturation the more pronounced it is. The reason
for this seems to be the bigger difference in the resistances of water and gas for clay and
sand (π G =1.82×105 mkg3 s ; sandstone: π F = 107 mkg3 s , mS: π F = 7.27 × 107 mkg3 s , Lt3:
π F = 5.33 × 109 mkg3 s ).
The remark concerning the smoothness of the curves and the intensity and location of the
extrema is also valid for both velocity and attenuation of the P3-wave. These are shown in
the last row of Fig. 8. However, for this wave the order of the values for the three soil types
are different. If we check, for example, the extremum P3-velocities for ω = 10 Hz, it is
obvious that it is approximately 0.09 m/s for sandstone, 0.016 m/s for mS and 0.003 m/s for
Lt3. I.e., they nearly differ in one order of magnitude. For the attenuation similar relations
are obtained. Since the P3-wave is attributed to the capillary pressure this seems to be an
effect of different retention curves for different soil types.

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 555

Table 6 Sound speed in gassy

Sediment Gas Sound speed
sediment
type content (m/s)

Clay 0 1,488
0.001 488
0.01 359
0.1 355
Silt 0 1,552
0.001 622
0.01 524
0.1 528
Fine sand 0 1,749
0.001 900
0.01 826
0.1 837
Coarse sand 0 1,907
0.001 1,106
0.01 1,039
Extract of Table III, Part II by 0.1 1,054
Anderson and Hampton (1980)

6 Final Remarks

In this paper the phase velocities and attenuations of some unconsolidated soils have been
investigated and compared to the acoustic behavior of sandstone filled by a water–air-mixture.
It turned out that the S- and P1-waves in sandstone are the fastest waves followed by the
waves in sand. In clay these two waves are the slowest. The reason is that both waves are
dependent on the shear modulus which in turn varies with porosity, Poisson’s number and
mass density and the P1-wave additionally to the elasticity of the skeleton. In order to
demonstrate that the calculated values for the speeds of the waves are reasonable, they will
be compared to measurements. Of course, the measurements concern only the P1- and the
S-wave since the other two waves are hard to observe in real media. Possibly, the conditions for
measurements and calculations do not agree exactly and the soil types are not entirely equally
composed but it is indicated that the calculated values for the shear wave speeds of sandstone
(c S ≈ 900 m/s), sand (c S ≈ 600 m/s) and clay (c S ≈ 400 m/s) are qualitatively concordant
with the values of Table 4 (bunter: up to 1,100 m/s, sand: up to 550 m/s and clay: 120–700 m/s).
Also the P1-velocities of Table 5 are qualitatively satisfied (loam: up to 900 m/s, calculated:
≈1,050 m/s; sandstone: 1,500–2,500 m/s, calculated: 2,000 m/s; sand: up to 1,800 m/s, here:
1,700 m/s). Only the velocity of clay (1,200-2,500 m/s, calculated: ≈ 1,100 m/s) seems to be
very high. However, in Table 6 in the limit of small water saturations much lower values of
the P1-velocity in clay (≈360 m/s) are quoted. This means, that also in the measurements
there are big differences. If the compressibility of the skeleton, K s , would be lowered or
another approach for the shear modulus would be used, in the calculations for clay also lower
values of the P1-velocity would be achieved. But also the application of a model assuming
isotropy and spherical particles can only yield a rough approximation of the behavior of clay
since the particle form of this soil type is plate-like.

123
556 B. Albers

The P2- and P3-wave which have been estimated to be hardly measurable are attributed
to the two pore fluids and behave in the opposite way. Where in the P2-wave a minimum
appears, in the P3-wave a maximum occurs. The values of velocities and of the attenuations
depend mainly on the resistances and on the capillary pressure. The role of P2- and P3-waves
may become clear first in the investigation of surface waves.

Acknowledgment I am very grateful for the financial support of the German Research Foundation (DEG).

References

Albers, B.: Analysis of the propagation of sound waves in partially saturated soils by means of a macroscopic
linear poroelastic model. Transp. Porous Mater. 80(1), 173–192 (2009)
Albers, B.: Modeling and Numerical Analysis of Wave Propagation in Saturated and Partially Saturated
Porous Media. Veröffentlichungen des Grundbauinstitutes der Technischen, Universität Berlin, vol. 48.
Habilitation thesis, Shaker Verlag, Aachen (2010)
Albers, B.: On a micro–macro transition for a poroelastic three-component model. ZAMM (2010). Early View.
doi:10.1002/zamm.201000061.
Albers, B., Wilmanski, K.: On modeling acoustic waves in saturated poroelastic media. J. Eng.
Mech. 131(9), 974–985 (2005)
Anderson, A.L., Hampton L.D.: Acoustics of gas-bearing sediments., I. Background, II. Measurements and
models. J. Acoust. Soc. Am. 67(6), 1865–1898, 1890–1903 (1980)
Berryman, J.G., Thigpen, L., Chin, R.C.Y.: Bulk elastic wave propagation in partially saturated porous solids.
J. Acoust. Soc. Am. 84(1), 360–373 (1988)
Biot, M.A.: Theory of propagation of elastic waves in a fluid saturated porous solid, I. low frequency range,
II. higher frequency range. J. Acoust. Soc. Am. 28(2), 168–178, 179–191, (1956)
Carcione, J.M., Gurevich, B., Cavallini, F.: A generalized Biot-Gassmann model for the acoustic properties
of shaley sandstones. Geophys. Prospect. 48, 539–557 (2000)
de la Cruz, V., Sahay, P.N., Spanos, T.J.T.: Thermodynamics of porous media. Proc. Royal Soc.
A 443(1917), 247–255 (1993)
DIN 18196: Earthworks and foundations—soil classification for civil engineering processes. DIN Deutsches
Institut für Normung e.V., Beuth Verlag GmbH (2006) (in German), German title: Erd- und Grundbau—
Bodenklassifikation für bautechnische Zwecke
DIN 4220: Pedologic site assessment—Designation, classification and deduction of soil parameters (norma-
tive and nominal scaling). DIN Deutsches Institut für Normung e.V., Beuth Verlag GmbH (2005) (draft,
in German), German title: Bodenkundliche Standortbeurteilung—Kennzeichnung, Klassifizierung und
Ableitung von Bodenkennwerten (normative und nominale Skalierungen)
Drew, D., Passman, S.L.: Theory of Multicomponent Fluids. Springer, New York (1999)
Gebrande, H., Kern, H., Rummel, F.: Elasticity and inelasticity. In: Hellwege K.-H. (ed.) Landolt-Brnstein.
Numerical Data and Functional Relationships in Science and Technology, New Series; Group V. Geo-
physics and Space Research. Vol. 1b: Physical Properties of Rocks. Springer, Berlin, pp. 1–233 (1982)
Geertsma, J.: The effect of fluid pressure decline on volumetric changes of porous rocks. Trans. AIME 210,
331–340 (1957)
Girsang, C.H.: A numerical investigation of the seismic response of the aggregate pier foundation system.
Master’s thesis, Virginia Polytechnic Institute and State University (2001)
Gray, W.: General conservation equations for multi-phase systems: 4. constitutive theory including phase
change. Adv. Water Resour. 6(3), 130–140 (1983)
Gudehus, G., Külzer, M.: Einfluss von Ionen und Gasblasen auf die Kollapsneigung feinstkrniger Bden.
Geotechnik 25(1), 12–20 (2002) (in German)
Hartge, K.H., Horn, R.: Einführung in die Bodenphysik. Schweizerbart, Stuttgart (1999) (in German)
Klein, G.: Bodendynamik und Erdbeben. In Smoltczyk, U. (ed.) Grundbau-Taschenbuch Teil 1—Geotech-
nische Grundlagen, vol. 1, Chap. 1.8. Ernst & Sohn (2001) (in German); also available in English:
Geotechnical Engineering Handbook, volume 1
Klimentos, T., McCann, C.: Why is the Biot slow compressional wave not observed in real rocks?. Geo-
physics 53(12), 1605–1609 (1988)
Liu, I.-S.: Continuum Mechanics. Springer, Berlin (2002)
Lo, W.-C., Sposito, G., Majer, E.: Immiscible two-phase flows in deformable porous media. Adv. Water
Resour. 25, 1105–1117 (2002)

123
Linear Elastic Wave Propagation in Unsaturated Sands, Silts, Loams and Clays 557

Lo, W.-C., Sposito, G., Majer, E.: Wave propagation through elastic porous media containing two immiscible
fluids. Water Resour. Res. 41, W02025 (20 pp) (2005)
Lo, W.-C., Yeh, C.-L., Tsai, C.-T.: Effect of soil texture on the propagation and attenuation of acoustic wave
at unsaturated conditions. J. Hydrol. 338, 273–284 (2007)
Pavlovic, V.D., Velickovic, Z.S.: Measurement of the seismic waves propagation velocity in the real
medium. Facta Universitatis: Phys. Chem. Technol. 1(5), 63–73 (1998)
Pham, N., Carcione, J., Helle, H., Ursin, B.: Wave velocities and attenuation of shaley sandstones as a function
of pore pressure and partial saturation. Geophys. Prospect. 50, 615–627 (2002)
Schmidt, H.-H.: Grundlagen der Geotechnik. Teubner, Stuttgart (1996) (in German)
Schultze, E., Muhs, H.: Bodenuntersuchungen für Ingenieurbauten. Springer, Berlin, Göttingen, Heidelberg
(1950) (in German)
Studer, J.A., Koller M.G.: Bodendynamik. Springer, Berlin, Heidelberg, New York (1997) (in German)
Truesdell, C.A.: A First Course in Rational Continuum Mechanics. The Johns Hopkins University, Baltimore,
Maryland (1972)
Truesdell, C.A.: Rational Thermodynamics. 2nd edn. Springer, Berlin (1984)
US. Soil Taxonomy, A Basic System of Soil Classification for Making and Interpreting Soil Surveys. Download
via http://soils.usda.gov/technical/classification/taxonomy/
van Genuchten, M.T.: A closed-form equation for predicting the hydraulic conductivity of unsaturated
soils. Soil Sci. Soc. Am. J. 44, 892–898 (1980)
van Genuchten, M.T., Nielsen, D.R.: On describing and predicting the hydraulic properties of unsaturated
soils. Annales Geophysicae 3(5), 615–628 (1985)
van Genuchten, M.T., Leij, F.J., Yates, S.R.: The RETC code for quantifying the hydraulic functions of unsat-
urated soils. Tech. rep., U.S. Salinity Laboratory, Riverside, CA (1991)
Vanorio, T., Prasad, M., Nur, A.: Elastic properties of dry clay mineral aggregates, suspensions and sand-
stones. Geophys. J. Int. 155(1), 319–326 (2003)
von Soos, P.: Eigenschaften von Boden und Fels—ihre Ermittlung im Labor. In: Smoltczyk, U. (ed.) Grundbau-
Taschenbuch Teil 1—Geotechnische Grundlagen, vol. 1, chap. 1.4. Ernst & Sohn (2001) (in German);
also available in English: Geotechnical Engineering Handbook, volume 1
White, J.E.: Underground sound: application of seismic waves, Methods in Geochemistry and Geophysics,
vol. 18. Elsevier, Amsterdam, New York (1983)
Wilmanski, K.: Tortuosity and objective relative accelerations in the theory of porous materials. Proc. R. Soc.
A 461, 1533–1561 (2005)

123