Modeling Volume Change and Mechanical Properties with Hydraulic Models

Th. Baumgartl* and B. Kock

ABSTRACT

Volume Change by Internal and

External Stresses

Volume change of soils may be caused either by external (mechanical) or internal (hydraulic) stresses or a combination of both. A complete description of volume change must therefore include both mechanical and hydraulic stresses. By combining theories of mechanical
and hydraulic stress states, a hydraulic function, which predicts the
change of water volume as a function of the stress state parameter
soil water suction (water retention curve), is adopted to model volume
change. The utilization of such a continuous function also enables the
derivation of soil mechanical parameters (e.g., preconsolidation stress,
Youngs modulus) by determining mathematically the point of maximum curvature and inflection point. This information can then be
used to calculate the preconsolidation stress according to the method
of Casagrande. The presented calculation has considerable advantages
compared with the graphic method of Casagrande or other methods.
On the basis of stressstrain relationships of various textured and
structured soils and soil substrates and various test procedures (oedometer test, triaxial test, shrinkage test), volume change is modeled using
the described method. It is shown that modeling volume change by
the van Genuchten equation using the software RETC is possible with
high accuracy. Soil mechanical parameters are derived using the parameters of the van Genuchten equation. The comparison of results
of this method with the Casagrande and a statistical method shows that
these methods have deficiencies when the data sets have a high variability, the samples are not homogeneous, and when the stressstrain
curve is flat. The accuracy of the mathematical method in contrast is
very high and the calculated preconsolidation stress very reliable.

Processes of volume change in a porous system like

soil can either occur in a two-phase system (solidwater
or solidair) or a three-phase system (solidwaterair).
For reasons of simplification, it is generally assumed
that air-filled pores are interconnected to the atmosphere
and that the influence of pore air pressure buildup during
volume changes can therefore be neglected. The stress
state can be defined by two kinds of stresses within the
soil. It is a result of external (mechanical) stresses and
internal (hydraulic) stresses. Mechanical stresses are
transmitted via the solid phase. Volume change occurs
when maximum shear resistances between particles are
exceeded. Hydraulic stresses or internal stresses are
caused by the water phase in the pore system and influence the shearing resistances. When stresses are applied,
they will be positive in a two-phase system (water-saturated soil), but may also be positive in the short term
in a three-phase system (water-unsaturated soil), while
compaction caused by external stresses will redistribute
water into a pore system with a modified pore-size distribution. At the state of equilibrium of the water phase
within the pore system (equivalent with the pore size
distribution), hydraulic stresses will be generally negative and can be characterized by tensile stresses. Under
unsaturated conditions, the change of internal stresses
by wetting and drying will cause reduction or expansion
of a soil volume. Depending on the intensity of the stress
application, a deformation will be elastic or plastic.
For a complete description of the stress state, both
the external and internal stresses have to be combined
as it is formulated in the Terzaghi equation (Terzaghi
and Jelinek, 1954) for a two-phase system and in the extended form for a three-phase system, first described
by Bishop and Blight (1963). This relationship is expressed as follows:

redicting volume change of soils under stress application is important when calculating the stability
of soil substrates. This information is commonly used
in engineering or agriculture to derive mechanical parameters (e.g., preconsolidation stress), which explain
the stress history of a material. Preconsolidation stress,
for example, shows which stresses a soil substrate has
been exposed to in its history (mechanical, hydraulic)
and up to which stress level it can be loaded without further unrecoverable volume change. Information about
the kind of volume change is also important for the
understanding of shrinkage behavior, crack and structure formation, as well as resistance to (shearing) stresses
(Horn and Baumgartl, 2002). In the literature, various
methods can be found which describe stress-dependent
strain and the derivation of mechanical properties. However, many of these basic approaches lack in simplicity
and uniqueness. A method is presented here which allows the description of volume change as a function of
stresses based on hydraulic and mechanical stress state
models and a well-defined derivation of mechanical
properties using this theory.

( ua) (ua uw).

[1]

The stresses explained by the matric water potential of

Eq. [1] represent the tensile stresses (Snyder and Miller,
1985), whereas the effective stresses are represented by
the mechanical stress state.
The -factor of Eq. [1] accounts for the amount of
water-filled pores which is, when defined as stress states,
dependent on the matric water potential. Under saturated conditions, the parameter 1 and Eq. [1] reduces to the general Terzaghi equation. It could be
shown that this factor is explained within certain limits
by the water retention curve (Baumgartl, 2002). The
calculation of the tensile stresses is possible as long as
the pore size distribution for each state of water potential is known (in fact, it is often assumed that the pore
size distribution is constant). In principle, a calculation
is possible even in soils with volume change (shrinkage
or swelling, compaction), when the deformation behavior is known (Katou et al., 1987; OSullivan and Ball,

Th. Baumgartl, Inst. for Plant Nutrition and Soil Sci., Univ. of Kiel,
Olshausenstrasse 40, 24118 Kiel, Germany; B. Kock, School of Mathematics, Univ. of Southampton, Southampton, SO17 1BJ, U.K. Received 19 Sept. 2002. *Corresponding author (t.baumgartl@soils.unikiel.de).
Published in Soil Sci. Soc. Am. J. 68:5765 (2004).
Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

57

58

SOIL SCI. SOC. AM. J., VOL. 68, JANUARYFEBRUARY 2004

1993). Because volume change as a combination of compaction and shrinkage is very complex, deformation due
to mechanical or hydraulic stresses are in the following
viewed and modeled separately.

Adoption of the StressStrain Relationship

with a Hydraulic Model
In mechanical and agricultural engineering, the relationship of volume change with stress application is a
common feature to describe the mechanical characteristics of soil substrates. With a semilogarithmic plot of
normal stress vs. strain, a recompression curve at a lower
stress range can be distinguished from a virgin compression curve at higher stresses (Casagrande, 1936; Terzaghi
and Jelinek, 1954). Both curves are separated by the
preconsolidation stress (Fig. 1). Likewise, volume change
caused by increased internal stresses results in a relationship which behaves similarly (Fig. 1). At lower capillary stresses (i.e., less negative matric potentials), the
shrinkage is smaller, whereas after exceeding a certain
degree of dryness, the shrinkage rate increases with less
negative matric water potentials. Within the range of
small matric potentials, this relationship can be defined
as the preshrinkage line. It is separated from the virgin
shrinkage line by the preshrinkage stress (Baumgartl,
2002). The preshrinkage line at high matric water potentials is equivalent to the structural shrinkage of the wellknown shrinkage curve derived from the void ratiomoisture ratio relationship (Yong and Warkentin, 1975).
The above considerations show that processes and
the characteristic of volume change can be expressed
similarly. Additionally, when both stresses occur, the mechanical and hydraulic stresses have to be coupled to
define the complete stress state (Groenevelt and Bolt,
1972; Richards and Smettem, 1992; Richards et al.,
2000). Fredlund and Rahardjo (1993) point out that
several coefficients can be defined which formulate the
volumetric deformation with hydraulic and mechanical

stress. For reasons of normalization in the following,

pore volume is defined as the void ratio e (pore volume/
volume of solids) and water content as the moisture
ratio (water volume/volume of solids). Figure 2 shows
several possible scenarios.
The application of mechanical stress on a water-saturated substrate (with 0 at drained conditions) may
result in curve II (stressstrain relationship). The slope
is defined by de/d. A steep slope at higher stresses
(virgin compression curve) can be distinguished from a
less steep slope at low mechanical stresses (preconsolidation curve).
Taking the equivalent relation for hydraulic stress
() vs. void ratio (e), the void ratio will decrease less
with decreasing matric potential due to the emptying of
pore space and hence reduced tensile stress. A possible
relationship may be curve I in Fig. 2 and can be characterized as shrinkage-strain relationship. The slope at
each point is defined by de/d and has low values at
very low water potentials (preshrinkage curve) and very
high water potentials (residual shrinkage curve). The
curve in between these two ranges of water potential
depict the virgin shrinkage curve.
Referring the parameter moisture ratio to mechanical stress and assuming saturated conditions (with
0 at drained conditions), then this relationship is
equivalent to curve II because e. The slope is defined by d/d. When referring matric water potential
to moisture ratio , the result may be curve III and
describes the water retention curve (after Fredlund and
Rahardjo, 1993).
Similar considerations have been conducted by Toll
(1995). He compared the shrinkage behavior of normally consolidated soil and overconsolidated samples
with the moisture ratio and void ratio. He states that
as long as a soil sample is saturated, the e and
relationships are identical. Once the sample starts to
be desiccated, is reduced with further desaturation

Fig. 1. Stressstrain relationship as a consequence of compaction and shrinkage.

 CK: HYDRAULIC MODELS FOR VOLUME CHANGE

BAUMGARTL & KO

following a virgin desiccation line with a steeper slope

(curve III in Fig. 2). The slope of the e/ relationship
(curve I in Fig. 2) will remain constant as long as the
shrinkage limit has not been reached. Beyond this point,
the slope will reduce to zero.
With the above given hypotheses, the similarity of
volume reduction and change of water content due to
hydraulic stress can be referred to volume change
due to mechanical stress . Additionally, it has been
described (Katou et al., 1987; Bruand and Cousin, 1995)
that mechanical stress will first decrease the size of less
stable coarse pores, which is comparable with draining
coarse pores of a water-saturated soil first by increasing
hydraulic stress. Furthermore, changes of the pore size
distribution by mechanical stresses can be compared
with the change of the pore size distribution by tensile
stresses, although different in magnitude (see Eq. [1]).
Combining these concepts of describing volume change
by hydraulic and mechanical stresses provides a basis
to model void ratio using an hydraulic model, which has
been adapted to de/d and de/d other than d/de.
As a model, the van Genuchten-equation (van Genuchten, 1980) for determining of the water retention
curve was chosen. The following expressions and parameters are used for different void ratio and moisture ratio relationships:
(i) Moisture ratio vs. matrix potential (water
retention curve)
r
[1 ()n]m
[2]
s r
(ii) Void ratio e vs. matric potential (shrinkagestrain relationship)
e er
[1 (e)ne]me
[3]
es e r
(iii) Void ratio e vs. total stress (stressstrain relationship)

e er
[1 (e)ne]me
es er

59
[4]

The parameters s and es describe moisture ratio at

water-saturated conditions and void ratio before stress
application, respectively. The parameter r (residual
moisture ratio) is dependent on the texture of the material, but is often defined as r 0. The minimum (residual) void ratio er is dependent on the maximum packing
density (i.e., minimum void ratio) of the material and
will be generally around er 0.27. The van Genuchten
parameters , n, and m have specific values (indicated
by indices) depending on being related to void ratio
or moisture ratio and mechanical stress or hydraulic
stress .
The advantage of this hypothesis is that it enables the
definition of volume change by a simple and continuous
function. Also, volume change due to external stresses
can be expressed by the same model as volume change
due to hydraulic stresses. In the case of shrinkage or
swelling by internal stresses, the stress state parameter
and the functional relationship with moisture ratio
and void ratio are the same.

Derivation of Soil Mechanical Properties

Stressstrain relationships provide the basis to describe
soil mechanical behavior. Such parameters can be, for
example, preconsolidation stress or Youngs modulus.
The preconsolidation stress is determined either graphically or statistically. The graphic determination after
Casagrande (1936) defines this stress by use of the point
of maximum curvature of the semilogarithmic stress
strain relationship. The preconsolidation stress is the
point of intersection of two lines. The first line is the
bisector line of the angle made by a horizontal and
tangent line drawn at the point of maximum. The second

Fig. 2. Void ratio e and moisture ratio as a function of the dependent variable mechanical stress or hydraulic stress .

60

SOIL SCI. SOC. AM. J., VOL. 68, JANUARYFEBRUARY 2004

line is the extended straight line of the virgin compression curve in direction to smaller stresses (Fig. 3). Although this method is commonly used, the determination of the point of maximum curvature and the decision
of which range of stresses determine the virgin compression line are often subjective.
Dias Junior and Pierce (1995) present a statistical
procedure with which the difficulties of exact and objective determination of the Casagrande values can be excluded. They also compare this method with a number
of different approaches. The method of Dias Junior
and Pierce (1995) fits the recompression line and virgin
compression line by a regression line starting with the
smallest and highest applied stress, respectively. Stepby-step, the adjoined stresses for calculation of both
regression lines are included and statistically compared.
One combination of stresses, which determines the recompression line and the virgin compression line, will
result in the greatest statistical difference of these two
lines. Dias Junior and Pierce (1995) state a good agreement of this statistical method with graphic determinations. The preconsolidation stress is then defined as the
point of intersection of these two lines. In practice, like
the graphic method, this method has disadvantages when
the number of observations is small and when the curvature of the log(stress)-strain relationship is small. Additionally, this method will only be an approximation dependent on the quality and number of data.
With use of a continuous function, the soil mechanical
parameters can be determined purely mathematically.
On the basis of this hydraulic model of van Genuchten,
the hydraulic/mechanical stressstrain relationship can
be modeled with the above given boundary values. Volume change should always be defined by void ratio rather
than relative strain. It simplifies calculations because
residual void ratio can be defined by a constant value
as explained above, whereas when using strain a specific
bulk density has to be related to each soil and hence a
specific minimum strain estimated. As the derived parameters of the van Genuchten curve respond very sensitively
to the shape of the curve, it is quite important to model
with a reliable residual pore volume (void ratio).

The mechanical parameter preconsolidation stress

can be determined using the second and third derivation
of this function. The virgin compression line is determined by the tangent through the inflection point (zeropoint of the 2nd derivation) of this function. The point
of maximum curvature (1st zero-point of the 3rd derivation) is the starting point for the calculation of the bisector line of the angle made by a horizontal and a tangent line of the stressstrain relationship. When the
zero-point of the 2nd and 3rd derivation is known, the
point of intersection of the bisector and the extended
straight line of the virgin compression curve can be
calculated easily using a spread sheet. In the following,
the derivation of the van Genuchten equation is described in detail.
To simplify notations, the van Genuchten equation
(see Eq. [2], [3], and [4]) is rewritten in the following
uniform way:
f(h) (s r)[1 (h)n]m r,

[5]

with s and r being the water content at saturation and

residual water content, respectively, and , n, and m
being the van Genuchten parameters. The variable h
represents the matric water potential or mechanical
stress . To take account for the logarithmic stress, f(h)
is replaced by the following function:
g(x) f(10x) (s r)[1 (10x)n]m r .

[6]

Furthermore, the notation k: k(x): 10 will be

used. With use of k(x) k(x)ln(10) and the chain rule,
the first derivative g(x) of g(x) can be calculated as
follows:
x

g(x) (s r)(m)(1 kn)m1nkn1k ln(10)

ckn(1 kn)m1,

[7]

where c denotes the constant mn(s r) ln(10). With

the addition of the product rule, the second derivative
g(x) and third derivative g(x) of g(x) can be calculated
as follows:
g(x ) cnkn1k ln(10)(1 kn)m1
ckn(m 1)(1 kn)m2nkn1k ln(10)
cn ln(10)kn(1 kn)m2[(1 kn) (m 1)kn]
c(kn mk2n)(1 kn)m2,

[8]

where c denotes the constant cn ln(10).

g(x ) c[nkn1k ln(10) m2nk2n1k ln(10)](1 kn)m2
c(kn mk2n)( m 2)(1 kn)m3nkn1k ln(10)
cn ln(10)kn(1 kn)m3[(1 2mkn)(1 kn)
(m 2)(kn mk2n)]
cn ln(10)kn(1 kn)m3[(1 2mkn kn 2mk2n
(m 2)kn m(m 2)k2n]
cn ln(10)kn(1 kn)m3[1 (3m 1)kn m2k2n]
Fig. 3. Determination of the preconsolidation stress according to the
method of Casagrande (1936).

[9]
The zero

61

CK: HYDRAULIC MODELS FOR VOLUME CHANGE

BAUMGARTL & KO

hi 10xi 1/(1/m)1/n

[10]

of the factor 1 mk of g(x) is the inflection point of

g(x). The zeros of g(x) can be calculated by solving
the discriminante of the quadratic polynomial
n

1 (3m 1)Y m2Y 2,

[11]

where Y : k (10 ) h . The first zero of this

quadratic polynomial is
n

Ymc

x n

n n

3m 1 5m 6m 1
2m2

[12]

1 1/n
1 3m 1 5m2 6m 1 1/n
Y mc

2m2

Hydraulic parameters/relationships

Void ratio e
Mechanical stress
Distribution of Youngs modules
E /e
Stressstrain relationship e f()
Stressstrain relationship e f()

moisture ratio /water content

water potential
pore size distribution /

Preconsolidation stress p

water retention curve f()

shrinkage-strain relationship
e f()
preshrinkage stress s

g(xmc) (1/2)g(xmc)(xp xmc) g(xi) g(xi)(xp xi)

[14]
Hence,
g(xi) g(xmc)

1
g(xmc)xmc g(xi)xi
2

1
g(xmc) g(xi)
2

[15]

where the variables g(xi) and g(xi) are given by

g(xi) f(hi) (s r)[1 (hi)n]m r

It describes the elastic behavior of a material and is

defined by
E /.

[13]

The preconsolidation stress p 10 p can be calculated

by equating the regression line through the inflection
point with the bisecting regression line between the
tangent at the point of maximum curvature and horizontal line (its slope is one half of the slope of the
tangent regression line):

xp

Mechanical parameters/relationships

Thus, the maximum curvature of the function g is at

hmc

Table 1. Comparison of mechanical and hydraulic parameters used.

[16]

and
g(xi) mn(s r)ln(10)(hi)n[1 (hi)n]m1, [17]
respectively, and the variables g(xmc) and g(xmc) are
given analogously.
Additional to preconsolidation stress, the Youngs
modulus E is also an important mechanical parameter.

[18]

The strain can be replaced by the void ratio. According to this definition, the Youngs modulus describes
the equivalent property mechanically as the pore size
distribution (/) hydrologically.
Table 1 summarizes the hydraulic and mechanical
parameters and relationships which are treated equivalently or are related to each other.
MATERIALS AND METHODS
Results of oedometer and triaxial tests from laboratory
experiments and literature were used to model stressstrain
relationships. The examples include stressstrain relationships
caused by compaction as well as by shrinkage. In the case of
compaction, a variety of differently textured and structured
substrates with various bulk densities and utilization were
investigated. Table 2 lists the main properties of the tested substrates.
The oedometer tests were performed either with multiple
replicates and a specific applied stress (oedometer test) or
one sample was loaded continuously with increasing stresses
(multistep test).
The volume change with stress was modeled using the van
Genuchten equation. The parameters and n of this model
were calculated by RETC (U.S. Salinity Laboratory, 1999).
The parameter m was fixed using m 1 1/n. As the unit
for mechanical stress is kilopascal, the -value of the van Genuchten equation of the presented calculations has the unit
1/kPa.

Table 2. Investigated material.

Substrate

Texture

Test

Soil

sandy loam

oedometer

Spoil

sandy loam

oedometer

Glacial till
Palsa
Clay-I
Clay-II

loamy sand
organic material
loamy clay
silty clay

oedometer
oedometer
oedometer
oedometer

Clay-III

silty clay (30%)

oedometer

Clay-IV-triax

clay (74%)

Kaolin K2

triaxial test
(hydrostatic pressure)
shrinkage

Kaolin K3

shrinkage

Remarks

Location

homogenised; natural bulk density; soil used for rehabilitation of a mine site
homogenised; natural bulk density; spoil material used for
rehabilitation of a mine site
B-horizon, depth 40 cm; undisturbed
undisturbed
undisturbed; B-horizon of a Vertisol
homogenized; maximum bulk density; base liner of a
municipal landfill
homogenized; maximum bulk density; base liner of a
municipal landfill
homogenized; low bulk density; base liner of a municipal
landfill
preconsolidation (28 kPa) and desiccation; data after
Biarez et al. (1988) taken from Toll (1995)
preconsolidation (55 kPa) and desiccation; data after
Biarez et al. (1988) taken from Toll (1995)

Australia

Indicates oedometer test with multiple increasing stress application of one sample (multistep-test).
Indicates replicates applied with one specific load only.

Australia
Germany
Finland
Germany
Germany
Germany
Germany

62

SOIL SCI. SOC. AM. J., VOL. 68, JANUARYFEBRUARY 2004

Table 3. Summary of results of modeling the stressstrain relationships with the van Genuchten equation and derivation of mechanical properties.
Parameter
Soil
Spoil
Glacial till
Palsa
Clay-I
Clay-II
Clay-III
Clay-IV-triax
Kaolin K2
Kaolin K3

e s

er

RMSE

NO

hmc#

hi

0.705
0.941
0.481
5.364
0.980
1.084
0.655
0.839
1.579
1.408

0.270
0.270
0.270
0.270
0.270
0.270
0.270
0.270
0.839
0.732

1/kPa
0.0314
0.0255
0.0074
0.1013
0.0212
0.0669
0.0191
0.0076
0.0094
0.0083

1.6419
1.9137
1.8116
1.3241
1.5397
1.3831
2.0921
1.2302
1.6318
1.5395

0.0118
0.0056
0.0020
0.0004
0.0236
0.0003
0.0239
0.0013
0.0003
0.0002

5
5
8
7
8
7
7
26
16
14

20
25
87
7
30
10
34
93
67
77

kPa
56
58
211
29
93
38
71
514
189
238

25
29
103
9
38
13
39
130
82
96

es, modeled void ratio at stress 0 kPa.

er, minimum void ratio.
and n, van Genuchten parameters.
NO, number of observations.
# hmc, stress at maximum curvature.
hi , stress at inflection point.
p, preconsolidation stress or preshrinkage stress.
fixed value during modeling.

RESULTS AND DISCUSSION

The principal results of the modeling procedure with
RETC and the calculation of the preconsolidation stresses
using the point of highest curvature and the inflection
point of the van Genuchten equation are listed in Table 3. The results show that the volume change of all

Fig. 4. Data and model of the stressstrain relationship (compression)

of the investigated substrates.

samples can be modeled with high accuracy as indicated

by the RMSE value. Only the soil samples clay-I and
clay-III have a higher deviation between data and fitted
values. This is because of the relatively small number
of replicates, but also because of the fact that each
observation is represented by an individual sample. The
heterogeneity between the replicates is high, particularly in structured samples, and the accuracy is therefore
reduced. Regarding the oedometer tests with increasing
stress on the same-sample (multistep) and the shrinkage
test, the accuracy is very high. The results show that
even when the heterogeneity of the data set is high, it
is possible to find a well-fitted curve.
Figures 4a, 4b, 4c, and 5 show graphically the modeled
results. In Fig. 5, which shows the volume change of
a shrinking soil, the calculated mechanical parameters
(preshrinkage stresses) and the Youngs modules are included.
The results of the tests depict very well the parameters
structure and water potential, which influence soil mechanical properties and stability (Baumgartl and Horn,
1991). The combination of homogenized soil and low

Fig. 5. Data, model, and Youngs modulus of the stressstrain relationship (shrinkage) of the substrates Kaolin K2 and Kaolin K3 and
preshrinkage stresses.

63

CK: HYDRAULIC MODELS FOR VOLUME CHANGE

BAUMGARTL & KO

water potentials at test conditions result in small preconsolidation stresses (soil, spoil, clay-I, clay-II, clay-III;
Fig. 4a, 4c). The organic material of the substrate Palsa
(Fig. 4b) with very high initial void ratio also presented
a very low preconsolidation stress.
The triaxial test was performed on homogenized material under isotropic conditions (Fig. 4c). This kind of
compaction reduces the volume to a lesser extent. The
steepness of the curve is lower and the calculated preconsolidation stress higher.
The test on the structured glacial till (Fig. 4c) was
performed at water conditions drier than field capacity.
The sample originated from a site which was exposed
to high stresses resulting from heavy agricultural machinery. The preconsolidation stress as a result of both
the higher mechanical and hydraulic stresses, has to be
classified as very high.
The influence of both the mechanical compaction as
well as the hydraulic impact of shrinkage can be observed in the samples Kaolin K2 and Kaolin K3 (Fig. 5;
data taken from Toll, 1995, after Biarez et al., 1988).
The sample Kaolin K2 was exposed to about 50% of the
mechanical stress of Kaolin K3 (55 kPa). Both samples
were then water saturated and dried. The preshrinkage
stress which can be calculated from the shrinkage characteristic is higher than the mechanical preconsolidation
stress. The shrinkage capacity of the kaolinitic substrate
exceeds the mechanical compressing stresses when the
material is dried and therefore increases the preconsolidation/shrinkage stress to higher values. Nevertheless,
a memory effect can be recognized as the sample Kaolin
K3 with higher mechanical preconsolidation stress,
shows as well higher preshrinkage stresses after drying.
The difference in the pretreatment is also identifiable
in the elastic properties. The course of the Youngs modulus shows a higher width at lower water suctions for
the substrate Kaolin K2 (Fig. 5).
To value the mathematical method of deriving mechanical parameters from a functional relationship, the
herewith calculated preconsolidation stresses are compared with the commonly used graphic method (Casagrande) and statistical method (Dias Junior and Pierce,
1995). The preconsolidation stress according to the Casagrande method was determined by a manual line of
best fit. The statistical method consists of a four-parameter fit. As described earlier, two regression lines are
calculated using an increasing number of measured data
for the first regression line and a equivalently decreasing
number of observed data for the second regression line.
The regressions with the highest statistical difference
were used to calculate the preconsolidation stress.
The preconsolidation stress of the three methods are
summarized in Table 4. The results of the methods differ
partly to a great extent. The variation between the physical and mathematical method is higher with greater heterogeneity of the data set; that is, the less a true line
could be drawn through the points (e.g., clay-I, clay-II).
The statistical method has a high deviation when the
slopes of the recompression line and the virgin compres-

Table 4. Comparison of preconsolidation stresses using different methods.

Graphical
Statistical
Mathematical
(Mohr-Coulomb) (four-parameter fit) (van Genuchten)
Soil
Spoil
Glacial till
Palsa
Clay-I
Clay-II
Clay-III
Clay-IV-triax
Kaolin K2
Kaolin K3

20
18
80
18
68
32
22
145
105
85

kPa
21
17
73
6
46
71
112
49
35
37

25
29
103
9
38
13
39
130
82
96

Not statistically significant from second best fit.

sion line are similar as a cause of either a small number

of points (e.g., clay-II, clay-III) or a shallow stressstrain
relationship in case of mechanical stress application
(glacial till, clay-IV-triax) or reduced volume change in
case of shrinkage (Kaolin K2, Kaolin K3).
The comparison shows that the quality of the preconsolidation stresses of the graphic and the statistical
method are very much dependent on the range of the
applied stresses, which later define the recompression
and virgin compression line. The accuracy also improves
with greater homogeneity of the samples, that is, the
more the slopes differ from each other. Regarding shallow stressstrain relationships with only little difference
between the slopes of the recompression and the virgin
compression line, the risk of small accuracy will always
be high.
The mathematical method allows a continual fit of
the stressstrain relationship including the boundary
conditions initial and residual void ratio and improves
the accuracy of the derived preconsolidation stress (according to the Casagrande method).
The shape of the stressstrain relationship modeled
with the van Genuchten equation is based on the parameters and n. A sensitivity analysis shows that generally
the preconsolidation stress is mainly influenced by the
-value of the van Genuchten equation. For n values
1.6, the influence of n on the preconsolidation stress is
negligible (Fig. 6). Only for stressstrain relationships
with a small change in void ratio with increasing mechanical or hydraulic stress, that is, a small steepness of
the van Genuchten equation, will n values have an influence on the value of the preconsolidation stress. As the
shape of the curve has a sensitive influence on the calculated value of the preconsolidation stress, void ratio
should be preferably used instead of, for example, an
absolute or relative strain value.
In the hydraulic use of the van Genuchten model, the
-value represents the air-entry value, which may be defined as the elastic property of the pore system in respect
to emptying of water filled pores (Hillel, 1998). Hence,
for high n values in very deformable substrates, the preconsolidation stress is equivalent to the air entry-value.
The results of the modeling procedure show that it
is possible to predict stressstrain relationships by a hy-

64

SOIL SCI. SOC. AM. J., VOL. 68, JANUARYFEBRUARY 2004

APPENDIX
List of Parameters

ua
uw

Fig. 6. Influence of the van Genuchten parameters and n on the

value of the preconsolidation stress (isobars in kPa; in 1/kPa).

draulic function (van Genuchten-equation) and to derive

mechanical parameters like preconsolidation stress and
Youngs modulus from derivatives of that equation.

CONCLUSION
The combination of the theories which describe stress
state functions of mechanical and hydraulic stresses and
the utilization of a model, which describes the behavior
of volume change due to either mechanical or hydraulic
stresses, bears great advantages when modeling volume
change. The application of a continuous function enables the inclusion of soil mechanical information in
respect to stressstrain behavior into soil compaction
models and to model stresses and stress distribution in
soils. In regard to shrinkage, the stress state parameter
soil water suction controls volume change. Shrinkagerelated parameters like preshrinkage stress can also be
easily derived from a continuous function as soil mechanical parameters such as preconsolidation stress. In
comparison with existing methods, the mathematical
method has the advantage to easily derive soil characteristics from this continuous function. Furthermore, the
modeling of the stressstrain relationship is based on
the boundary conditions initial and final void ratio and
increases the accuracy of the result in case of data sets
with high variability. Furthermore, the utilization of the
same function for mechanical and hydraulic processes
simplifies the coupling of both these processes and enables one to relate the stress state parameters mechanical stress to hydraulic stress . This is of great advantage when the total stress state of soils has to be defined.

s
r

r
s
e
er
es

n
m
h
E

p
s

effective stress
total stress
water potential
pore air pressure
pore water pressure water suction [ (matric
water potential)]
parameter related to the degree of saturation of the
soil (0 1)
water content
water content at saturation
residual water content
moisture ratio
residual moisture ratio
moisture ratio at saturation
void ratio
residual (minimum) void ratio (e 0.27)
maximum void ratio
van Genuchten parameter (subscripts relate to the
dependent variables e, and , )
van Genuchten parameter (subscripts relate to the
dependent variables e, and , )
van Genuchten parameter (subscripts relate to the
dependent variables e, and , )
variable representing or
Youngs modulus
strain
preconsolidation stress
preshrinkage stress
REFERENCES

Baumgartl, T. 2002. Prediction of tensile stresses and volume change

with hydraulic models. p. 550. In M. Pagliai and R. Jones (ed.)
Sustainable land managementEnvironmental protection: A soil
physical approach. Catena Verlag, Reiskirchen, Germany.
Baumgartl, T., and R. Horn. 1991. Effect of aggregate stability on
soil compaction. Soil Tillage 19:203213.
Biarez, J., J.M. Fleureau, M.I. Zerhouni, and B.D. Soepandji. 1988.
Variations de volume des sols argileaux lors de cycles de drainagehumidification. Rev. Fr. Geotechnique 41:6371.
Bishop, A.W., and G.E. Blight. 1963. Some aspects of effective stress
in saturated and partly saturated soils. Geotechnique 13:177197.
Bruand, A., and I. Cousin. 1995. Variation of textural porosity of a
clay-loam soil during compaction. Eur. J. Soil Sci. 46:377385.
Casagrande, A. 1936. The determination of pre-consolidation load
and its practical significance. p. 6064. In Int. Soil Mech. Found.
Eng. Harvard University, Cambridge, MA.
Dias Junior, M.S., and F.J. Pierce. 1995. A simple procedure for
estimating preconsolidation pressure from soil compression curves.
Soil Technol. 8:139151.
Fredlund, D.G., and H. Rahardjo. 1993. Soil mechanics for unsaturated soils. A. Wiley, New York.
Groenevelt, P.H., and G.H. Bolt. 1972. Water retention in soils. Soil
Sci. 113:238245.
Hillel, D. 1998. Environmental soil physics. Academic Press, San
Diego, CA.
Horn, R., and T. Baumgartl. 2002. Dynamic properties of soils. p. 389.

 CK: HYDRAULIC MODELS FOR VOLUME CHANGE

BAUMGARTL & KO

In A.W. Warrick (ed.) Soil physics companion. CRC Press, Boca

Raton, FL.
Katou, H., K. Miyaji, and T. Kubota. 1987. Susceptibility of soils
to compression as evaluated from the changes in the soil water
characteristic curves. Soil Sci. Plant Nutr. (Tokyo) 33:539554.
OSullivan, M.F., and B.C. Ball. 1993. The shape of the water release
characteristic as affected by tillage, compaction and soil type. Soil
Tillage Res. 25:339349.
Richards, B.G., T. Baumgartl, and R. Horn. 2000. Modeling coupled
processes in structured unsaturated soilsTheory and examples.
p. 175190. In R. Horn et al. (ed.) Subsoil compaction: Distribution,
processes, consequences. 32nd ed. Catena Verlag, Reiskirchen,
Germany.
Richards, B.G., and K.R.J. Smettem. 1992. Modeling water flow in
two- and three-dimensional applications. I. General theory for nonswelling and swelling soils. Trans. ASAE 35:14971504.

65

Snyder, V.A., and R.D. Miller. 1985. Tensile strength of unsaturated

soils. Soil Sci. Soc. Am. J. 49:5865.
Terzaghi, K., and P. Jelinek. 1954. Theoretische Bodenmechanik.
Springer Verlag, Heidelberg, Germany.
Toll, D.G. 1995. A conceptual model for the drying and wetting of
soil. p. 805810. In E. E. Alonso and P. Delage (ed.) First Int.
Conf. on Unsaturated Soil, Paris. 68 Sept. 1995. Balkema, Rotterdam, the Netherlands.
U.S. Salinity Laboratory. 1999. Code for quantifying the hydraulic
functions of unsaturated soils. Release 6.0. U.S. Salinity Lab.,
USDA, ARS, Riverside, CA.
van Genuchten, M.T. 1980. A closed-form equation for predicting
the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am.
J. 44:892898.
Yong, R.N., and B.P. Warkentin. 1975. Soil properties and behaviour.
Elsevier, Amsterdam.