Computers and Geotechnics 158 (2023) 105381

Contents lists available at ScienceDirect

Computers and Geotechnics

journal homepage: www.elsevier.com/locate/compgeo

Cavity expansions in partially drained soils and a comparison to CPT results

Adrian R. Russell *, Yanzhi Wang , Nasser Khalili
Centre for Infrastructure Engineering and Safety, School of Civil and Environmental Engineering, UNSW Sydney, NSW, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Cylindrical and spherical cavity expansions in soils are of interest, especially when interpreting cone penetration
Stress analysis tests (CPTs) and pressuremeter tests (PMTs). Here the cavity expansion problem is solved semi-analytically,
Time dependence considering partial drainage of the soil, when the cavity wall velocity is constant. The solutions may assist
Pore pressures
with the interpretation of CPTs and PMTs when partially drained conditions prevail in the surrounding soils. The
Penetrometers
solution procedure involves writing the governing partial differential equations in dimensionless form. The
modified Cam-clay constitutive model is adopted for the soil to demonstrate the solution procedure, due to its
wide familiarity, although other models could be used by making simple modifications. The variables in the
partial differential equations are functions of a dimensionless time, T, and a dimensionless radial coordinate, H. T
depends on cavity radius, the time after expansion began and a coefficient of consolidation. The equations are
solved for when T is fixed, simultaneously as an initial value problem, as the partial differential equations
become ordinary differential equations. Important features captured by the analyses include pore water pressure
variations with soil volume changes, as well as the differential of the pore water pressure satisfying the no-flow
cavity wall boundary condition. Results for a cavity expansion reveal that initially, when T and the cavity are
small, drained conditions prevail. As T and the cavity become larger the soil becomes partially drained and
eventually undrained. T is analogous to the dimensionless cone penetration velocity used in CPT studies. Nor­
malised cavity wall pressures vary with T similar to that of normalised cone penetration resistances, demon­
strated using experimental data. Plots of the normalised cavity wall pressures against the state parameter show
that the two key fitting constants are heavily influenced by partial drainage. A similar dependence must exist
when plotting a normalised cone penetration resistance against the state parameter when partially drained
conditions prevail. The practical ramifications of not dealing with this partial drainage may be huge, especially
when assessing liquefaction susceptibility through estimations of the state parameter using correlations devel­
oped for when drained conditions prevail. Significant underestimations of state parameter, liquefaction sus­
ceptibility and post-liquefaction strengths may result.

radius.
The analyses provide a theoretical basis for the interpretation of the
1. Introduction
cone penetration test (CPT) (e.g. Vesic, 1972; Shuttle & Jefferies, 1998;
Salgado & Prezzi, 2007; Pournaghiazar et al., 2012; 2013) and the
The analyses of expanding cylindrical or spherical cavities in soils are
pressuremeter test (PMT) (e.g. Palmer, 1972; Yu, 1994; Yu et al., 1996),
of wide interest to geotechnical engineers. Most cavity expansion studies
two in situ tests used to characterise the engineering properties and
in the geotechnical engineering literature have been restricted to satu­
states of soils. The cone penetration resistance from CPTs may be related
rated soils under drained (constant pore water pressure) or undrained
to the wall pressure of a spherical cavity when expanded from an initial
(constant volume) conditions. They can also relate to dry soils for which
radius of zero. The expansion pressure versus cavity strain relationship
drained conditions usually apply. A wide variety of constitutive models
measured in a PMT can be interpreted using the cylindrical cavity
for the soil may be considered. The solution procedure required depends
expansion theory when expansions start from a non-zero initial radius.
on the complexity of the constitutive model, the assumed rate forms of
There are many other applications in geotechnical engineering as well,
the soil’s stresses and strains around the expanding cavity, whether a
related to pile foundations, earth anchors, tunnelling, wellbores, as
solution applies to cylindrical, spherical or both cavity types, and
demonstrated by Yu (2000).
whether the expansion starts with the cavity having a zero or non-zero

* Corresponding author.
E-mail address: [email protected] (A.R. Russell).

https://doi.org/10.1016/j.compgeo.2023.105381
Received 27 November 2022; Received in revised form 13 February 2023; Accepted 4 March 2023
Available online 20 March 2023
0266-352X/© 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Nomenclature qt
′
effective cone penetration resistance
Qp normalised cone penetration resistance
List of symbols Qp,ce normalised radial cavity wall pressure
a pore pressure parameter that depends on the cone r radius of a soil element
geometry Ss stress scale
Bqp,ce pore pressure ratio at the cavity wall t time since cavity expansion began
Bqp pore pressure ratio at the cone tip T non-dimensional time, equal to Tts and rccwc
c coefficient of consolidation, equal to Dk/γw Ts time scale
cv alternate coefficient consolidation used in some CPT u pore water pressure
kv0 σ v0
′

interpretations, equal to γw λ
v specific volume
dc cone diameter vv cone penetration velocity
wc
D 2m
constrained modulus of elasticity, equal to K + m+1 G V non-dimensional cavity wall velocity, equal to Vs and 1

G shear modulus of elasticity, equal to (m+1)(1− 2μ) VCPT non-dimensional cone penetration velocity, equal to vvcdv c
2(1+(m− 1)μ) K
Vs velocity scale
h hardening modulus
w radial velocity of the soil element at r
k soil permeability
0 subscript denoting the initial (far field) condition
k0 ratio between initial horizontal and vertical effective
subscript denoting association with the cavity wall
stresses c
subscript denoting association with the elastic–plastic
k fitting constant relating normalised cone resistance or
R
[ ( boundary
cavity wall pressure to state parameter in the ln Qp 1 − e
) ] superscript denoting the elastic component of strain
Bqp + 1 ∼ ψ plane p
superscript denoting the plastic component of strain
vp
′
K bulk modulus of elasticity, equal to κ
γw unit weight of water
L non-dimensional length, equal to Lrcs and rc wc
c
Γ constant defining the location of the critical state line in
Ls length scale the v ∼ lnp’ plane
m parameter to distinguish between cylindrical (m = 1) and εp volumetric strain, equal to εr + mεθ
m
spherical (m = 2) cavities εq deviatoric strain, equal to m+1 (εr − εθ )
m fitting constant relating normalised cone resistance or εr radial strain
[ (
cavity wall pressure to state parameter in the ln Qp 1 − εθ tangential strain
) ]
Bqp + 1 ∼ ψ plane ζ radial displacement
m unit vector controlling direction of plastic flow H dimensionless radial coordinate, equal to rrc
M slope of the critical state line in the q ∼ p’ plane, equal to κ constant representing the slope of the elastic unload-reload
2(m+1)sinφcs
′
line in the v ∼ lnp’ plane
(m+1)− (m− 1)sinφ′cs λ slope of the critical state line in the v ∼ lnp’ plane
n unit vector at current stress state on the yield surface μ Poisson’s ratio
N constant defining the location of the isotropic normal σr
′
radial effective stress
consolidation line in the v ∼ lnp’ plane σ v0 initial vertical stress
′ ′

p mean effective stress, equal to

σr +mσθ
tangential effective stress
′ ′
m+1 σθ
pc hardening parameter critical state friction angle
′ ′
φcs
q deviatoric stress, equal to σr − σ θ triaxial critical state friction angle
′ ′ ′
φtx
qc uncorrected total cone penetration resistance ψ state parameter
qt corrected total cone penetration resistance

In many cases the cylindrical and spherical cavity expansion prob­ (1992) and Collins & Stimpson (1994) clarified how there is a similarity
lems can be investigated simultaneously using one set of equations by to the soil response around a cavity that expands from a zero initial
introducing a ‘cavity type’ parameter in to the stress and strain defini­ radius. The soil response is independent of cavity size and the time since
tions and equilibrium equation (Carter et al., 1986). For an initially expansion began, and only depends on a dimensionless distance from
isotropic soil state, and ignoring self-weight and body force influences the cavity wall. They developed the so-called similarity technique, and
which are minor, spherical symmetry holds for an expanding sphere demonstrated it using a state parameter hardening model and modified
while axial symmetry and plane strain holds for an expanding cylinder. Cam-Clay. The similarity technique has been applied by many re­
The displacements in the soil are radial. searchers since, implementing more complex constitutive models. Rus­
Closed-form solutions may be obtained using analytical methods for sell & Khalili (2002) and Jiang & Sun (2012) addressed particle crushing
both cavity types, for example when the Mohr Coulomb elastic-perfectly in their constitutive models. Others have used bounding surface plas­
plastic model is used and the convected part of the stress rate is ignored ticity for when the soil is unsaturated and exhibits suction hardening
(Carter et al., 1986) and, more accurately, when it is considered (Yu & (Russell & Khalili, 2006; Yang & Russell, 2015), or thermoplasticity
Carter, 2002). Mo & Yu (2017) also obtained closed form expressions for (Zhou et al., 2018). Alternate semi-analytical techniques have been
both cavity types when a hardening soil model was used and the soil was presented by others, e.g. Chen & Abousleiman (2018) for drained cy­
undrained. lindrical cavity expansions, and Carter & Yu (2022) for a model with
More usually, when hardening of soil is permitted in the constitutive limited dilation. Numerical methods may be adopted as well, e.g. using
model, solutions are obtained using semi-analytical techniques, often for codes employing the Finite Element Method or the Finite Difference
both cavity types using one set of equations, and when the convected Method, with examples including Yu & Houlsby (1990) and Cudmani &
part of the stress rate is considered. For these conditions Collins et al. Osinov (2001).

2
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

The studies which incorporate further complexities like anisotropy, of the expanding cavity is denoted rc and the radius of the elastic–plastic
fabric or structure in to the constitutive model have been restricted to boundary is denoted rR . The radius of a particular soil element is
cylindrical cavities only, for example Chen & Liu (2019), Liu & Chen denoted r. Only cavities expanded from zero initial radius are consid­
(2019), Sivasithamparama & Castro (2018, 2020) and Chen et al. ered. The velocity at the cavity wall is wc and is assumed constant. It
(2019). The anisotropy, fabric or structure has a symmetry that is co­ follows that rc = wc t, where t is the time since expansion began. The
axial with that of the cylinder permitting a semi-analytical solution to be solution does not obey similarity as the velocity of the elastic–plastic
obtained. Extension to spherical cavities has not yet been achieved. boundary, denoted wR , varies during the expansion. A subscript c de­
Cheng et al. (2022) explored differences between constant mass and notes association with the cavity wall and a subscript R denotes asso­
constant volume conditions. ciation with the elastic–plastic boundary.
Very few attempts have been made to address partial drainage in the In the soil around an expanding cavity the radial stress (σ r ) repre­
′

soil around the cavity. Partial drainage means a combination of pore sents the major principal stress and the tangential stress (σ θ ) represents
′

pressure change and volume change occur. The amounts of change, and the minor principal stress. The mean stress (p ) and deviatoric stress (q)
′

the extent to which the partially drained condition differs from a fully are defined as
drained or undrained one, depends on a normalised quantity which in­
corporates the rate of cavity expansion, the size of the cavity and the
′ ′
σ r + mσ θ
(1)
′ ′ ′
p = , q = σr − σθ
soil’s coefficient of consolidation. Silva et al. (2006) solved the cylin­ m+1
drical cavity expansion problem in a partially drained soil using the
A prime symbol indicates the stress variable is an effective stress. m is
Finite Element Method, in which soil behaviour was described using
the cavity type parameter. For cylindrical cavities m = 1 and for
modified Cam-clay with altered elastic stiffness definitions. Suzuki &
spherical cavities m = 2. Plane strain conditions are assumed for cylin­
Lehane (2015) adopted the Finite Element Method to solve the spherical
drical cavities and the influence of the intermediate principal stress is
cavity expansion problem in partially drained soil with soil behaviour
accounted for, approximately, when defining some soil properties as
assumed to obey a non-linear strain hardening model. Also, Mo et al.
explained later.
(2020) estimated the responses that would prevail during cylindrical
These stress definitions are commonly used in cavity expansion an­
and spherical cavity expansions in partially drained soils by using an
alyses when the soil is treated as initially isotropic. The appeal is that a
empirical interpolation between drained and undrained responses. This
simple and single set of equations and definitions can be applied so that
approach violated the no-flow boundary condition at the cavity wall, nor
expansions of both cylindrical and spherical cavities can be solved.
did it capture the volume and pore pressure change in the elastic region.
The volumetric strain (εp ) and deviatoric strain (εq ) are related to
A few studies have also modelled the cone penetration problem in
radial strain (εr ) and tangential strain (εθ ) by
partially drained soils directly. Yi et al. (2012) conducted a large
deformation analysis, using the Finite Element Method, adopting an εp = εr + mεθ , εq =
m
(εr − εθ ) (2)
elastic-perfectly plastic model for the soil. Sheng et al. (2014) conduced m+1
a similar analysis although incorporated the more sophisticated modi­ ensuring work conjugacy.
fied Cam-clay constitutive model. Ceccato et al. (2016) and Ceccato & Here x is used to denote one of the variables at r and t. Its material
Simonini (2017) used the Material Point Method to model partial ◦
time derivative is denoted x and its local time derivative is denoted ẋ =
drainage around a penetrating cone, again with modified Cam-clay.
∂x/∂t. Around an expanding cavity these are related by (Collins et al.,
Monforte et al. (2021) used the Particle Finite Element Method and
1992; Collins & Stimpson, 1994)
adopted a hybrid sand and clay hardening model.
The main objective of this study is to solve the cavity expansion ◦ ∂x
x = ẋ + w (3)
problem in a partially drained soil semi-analytically. Advantages of a ∂r
semi-analytical method include the requirement to clearly identify the
where w is the radial velocity of the soil element at r. The radial and
governing differential equations and dimensionless groups which con­
tangential strain rates at any instant are defined in terms of w and r as
trol a solution. This is not necessarily the case when numerical methods
are used. Semi-analytical methods are more accessible to a wide range of ∂w w
ε˙r = − , ε˙θ = − (4)
users as the differential equations can be easily implemented in to a ∂r r
differential equation solving software. Specialised coding skills and It follows that
knowledge of the computational mechanics which underpin the nu­ ( )
merical method are not required. Also, once the differential equation ε˙p = −
∂w mw
− , ε˙q =
m
−
∂w w
+ (5)
solver is setup the recovery of a solution only takes a few seconds, while ∂r r m+1 ∂r r
a solution from a numerical method may require much longer.
The solution procedure presented is applicable to both cylindrical 3. Governing equations
and spherical cavities. A continuity equation that accounts for the
volumetric strain rate dependant pore pressure around the cavities will The problem is solved here for the case when the soil’s stress–strain
be incorporated. Soil behaviour will be described using modified Cam- behaviour in the elastic–plastic region it is described by modified Cam-
clay although the solution procedure presented may be applicable to clay, a constitutive model formulated using critical state soil mechanics.
other constitutive models. It is the simplest and most familiar of all the critical state–based
Results show how partial drainage, particularly cavity size, expan­ constitutive models. As modified Cam-clay is not well suited to some
sion velocity and the coefficient of consolidation, affect cavity wall and soils, including over consolidated clays and granular soils, a user may
pore water pressures. The results, for spherical cavities, reproduce wish to implement an alternate critical state constitutive model for those
normalised cone resistances measured from partially drained CPTs very materials. This is a simple task and will be commented on below.
well, and pore pressures less well. The elastic stress–strain relationship is
⎡ ⎤
⎡ ⎤ 1
2. Notations 0 [ ◦]
ε˙ep ⎢K ⎥ p′
⎣ ⎦=⎢
⎣
⎥ ◦
⎦ (6)
ε˙eq m
The expanding cavity is surrounded by soil in which deformation 0 q
2(m + 1)G
comprises a combination of elastic and plastic strains. The elastic–plastic
e
region is surrounded by soil which deforms purely elastically. The radius in which the superscript denotes the elastic component of strain.

3
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

The bulk modulus (K) and shear modulus (G) are defined as ′
∂p m ∂q mq ∂u
′ + + + =0 (15)
vp (m + 1)(1 − 2μ) ∂r m + 1 ∂r r ∂r
K= and G = K (7)
κ 2(1 + (m − 1)μ)
where u represents the pore water pressure.
where v denotes the specific volume, μ is Poisson’s ratio and κ is a The continuity equation which accounts for partial drainage, when
material constant and represents the slope of the elastic unload-reload water is incompressible, relates a change of u to the local time derivative
line in the v ∼ lnp’ plane. κ is taken to have the same value for cylin­ of the volumetric strain and is (Verruijt, 2016)
drical and spherical cavities. ( )
k ∂2 u m ∂u
The yield surface is an ellipse in the q ∼ p’ plane − ε˙p = + (16)
γ w ∂r2 r ∂r
[ ′( ′ ′)]
f = q2 − M 2 p pc − p =0 (8) Another continuity equation links the rate of volumetric change to
in which M is the slope of the critical state line the total strain rates ensuring the conservation of solid mass. It follows
′
that
2(m + 1)sinφcs
M= (9) ◦
(m + 1) − (m − 1)sinφ′cs v ∂w m
= − ε˙p = + w (17)
v ∂r r
and φcs is the critical state friction angle and is defined as 1.125φtx
′ ′

and φtx for cylindrical and spherical cavities, respectively, where φtx is
′ ′ In the elastic region u is not a constant when the soil is partially
drained. A closed form expression for u may be derived, as detailed in
the triaxial critical state friction angle. This accounts for the omission of
Section 4.3. The values of p′ , q, v and w throughout the elastic region
the intermediate stress influences on strength, approximately (Wroth,
depend on u, and may also be expressed in closed form. This is in
1984). Very similar relationships have been advocated by Hanna (2001)
contrast to completely drained and undrained conditions where p′ , u and
and Nanda & Patra (2015).
v remain constant in the elastic region, equal to the initial values.
The ellipse always passes through the origin. Its size is controlled by
In the elastic–plastic region closed form expressions for p′ , q, u, v and
the hardening parameter pc , which represents the intercept of the ellipse
′

w can not be derived. Their quantities are obtained as an output of a

with the p axis.
′
numerical integration procedure. Also in the elastic–plastic region, in
The normality condition applies, meaning that the plastic potentials
addition to the constitutive and governing equations listed above, it is
g are automatically given by the yield surfaces, i.e. g = f.
necessary to consider the consistency equation which causes the stress
It is assumed that the yield loci expand while maintaining constant
state to remain on the yield surface
shape, through changes in pc . An isotropic hardening law is found by
′

assuming a linear relationship between v and the logarithm of pc ,

′ ◦
∂f ′ ∂f ◦ ∂f ′ ◦
p + q+ ′ p = 0 (18)
defining the isotropic normal compression line (iso-ncl). It is assumed ∂p′ ∂q ∂pc c
that
as well as the hardening law which interrelates changes to pc and εpp
′

(10)
′
v = N − λlnpc ′
′
∂pc ˙p
in which N is a constant defining the location of the iso-ncl and λ is
p◦ = εp (19)
c ∂εpp
the slope, both taken to have the same values for cylindrical and
spherical cavities.
4. Solution procedure
The elastic–plastic stress–strain relationship is
⎡ ⎤
Prior to an expansion the soil state is given by p0 and v0 . The initial
′
[ ][ ′ ◦ ]
ε˙p
⎣ p ⎦ = 1 np mp nq mp p
◦ (11) deviator stress and pore pressure are zero, i.e. q0 = 0 and u0 = 0, and the
ε˙pq h np mq nq mq q
initial strains are also zero. The subscript 0 indicates the initial values.
in which the superscript p
denotes the plastic component of strain The initial values are also the far field values, i.e., the values at infinite
and distance from the expanding cavity.
⎡ ⎤T
⎢
⎢
∂f ∂f ⎥
⎥ 4.1. Scale definitions
∂p′
(12)
∂q
n = [np , nq ]T = ⎢
⎢√(
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
)2 ( )2̅ , √( ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
)2 ( )2̅
⎥
⎥
⎣ ∂f
+ ∂∂qf ∂f
+ ∂∂qf
⎦ Variables will be made dimensionless by dividing them by a scale for
∂p′ ∂p′
time (Ts ), length (Ls ), velocity (Vs ) or stress (Ss ), each being a constant
during a cavity expansion. The scales are defined as
⎡ ⎤T
c c Ls
⎢
⎢
∂g ∂g ⎥
⎥ Ts = , Ls = , Vs = = wc , Ss = D (20)
m = [mp , mq ]T = ⎢
∂p′ ∂q ⎥ (13) wc 2 wc Ts
⎢ ( )2 ( )2̅ , √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ( )2 ( )2̅ ⎥
⎣ ∂g
+ ∂q ∂g ∂g
+ ∂∂gq
⎦ in which
∂p′ ∂p′

c = Dk/γw (21)
is the soil’s coefficient of consolidation, γw is the unit weight of
′
∂f ∂pc mp
h= − √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ (14)
∂p′c ∂εpp ( ∂f )2 (∂f )2 water, k is the soil’s permeability and D is the constrained modulus of
+
∂p′ ∂q elasticity at the elastic–plastic boundary.
At the cavity wall the non-dimensional time (T), length (L) and ve­
where n is the unit vector normal to the yield surface at the current
locity (V) are then
stress state, pc hardens with changes to εpp only, m is the unit vector
′

normal to the plastic potential, and h is the hardening modulus. T=

t
=
twc 2 rc wc
= , L=
rc rc wc
= , V=
wc
=1 (22)
The stress equilibrium at any location around an expanding cavity, Ts c c Ls c Vs
whatever the drainage condition, obeys (Russell & Khalil, 2006) A dimensionless radial coordinate H (Eta) is defined as

4
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

r r r This, by introducing equation (28), becomes

H= = = (23)
r c r c wc t
d u (m
2 ) du
Notice that x is the non-dimensional form of variable x, for which x = 2
+ + TH =0 (30)
dH H dH
x/Xs and Xs is the scale relevant to x.
For drained conditions there is no change to u (or u). In the elastic
region the stress equilibrium, kinematics, and linearity between stress
4.2. Conversion of variables from rate form to dimensionless differential
and strain increments create a balance so that σ r and σθ vary while al­
′ ′

form ′ ′
ways satisfying rm+1 θ = p0 , and εr and εθ vary while always satisfying
σ +mσ ′

The variables in the above partial differential equations are functions εr + mεθ = εp = 0. For undrained conditions there is no volume change.
of r and t. They are made dimensionless and functions of H using the u must also remain constant in the elastic region, due to the balance
following conversions between stresses and strains. The balance is a characteristic of the cavity
expansion problem when drained and undrained conditions prevail and
∂x 1 dx 1 Xs dx 1 Xs dx
= = = (24) is well known (Carter et al., 1986; Collins & Stimpson, 1994). For
∂r rc dH rc Ls dH T Ls dH
partially drained conditions, however, a change to u occurs in the elastic
∂x H dx H Xs dx region and disrupts the balance, even when there is a linearity between
ẋ = =− =− (25)
stress and strain increments. As an expansion occurs p′ ∕= p0 and εp ∕
= 0.
′
∂t t dH T Ts dH
( ) Equation (30) can be integrated for a given T although the solutions
◦ w wc dx 1 Xs dx obtained are different for the two m values.
x= − H = (w − H) (26)
wc rc dH T Ts dH For m = 2
The derivation of equations (24) and (25) is detailed in Appendix A. ( (√̅̅̅̅̅̅ ) ( ))
√̅̅̅̅̅̅̅̅̅ 2T H 2 TH 2
Equation (26) is obtained by substituting equations (24) and (25) into u = C1(T) + C2(T) 2πT erf + exp − (31)
2 H 2
equation (3).
These conversions are valid for when T is fixed. Fixing T means the in which erf is the error function. The two T dependent constants,
partial differential equations may be rewritten as ordinary differential C1(T) and C2(T), emerge. They can be determined by imposing
equations and solved simultaneously as an initial value problem. A so­ boundary conditions: u = uR when H = HR for any T, u = 0 when H = ∞
lution reveals a variable’s dependence on the distance from the cavity for any T. Their substitution in to equation (31) yields
wall through H. A solution is relevant to the instant during the expansion ⎛
(√̅̅̅̅ ) ( )
⎞
√̅̅̅̅̅̅̅̅̅
process when the cavity has attained a certain size. The cavity radius, 2T H TH 2
HR ⎜
⎜ 2πT Herfc( 2 )− 2exp (− 2 ) ⎟
⎟
the time since expansion began and the coefficient of consolidation u = uR ⎜√̅̅̅̅̅̅̅̅̅ √̅̅̅̅ ⎟ (32)
H ⎝ 2πT H erfc 2T HR − 2exp − THR 2 ⎠
combine to give T. The fixed T constraint is only invoked after the in­ R 2 2

terdependencies between x, H, T, r and t are correctly accounted for

(Appendix A). in which erfc is the complementary error function.
For m = 1
4.3. Elastic region ( ( ))
TH 2
u = C3(T) + C4(T) Ei 1, (33)
Throughout the elastic region it is assumed, for simplicity, that the 2
bulk and shear moduli are constants and, for modified Cam-clay, are The two T dependent constants, C3(T) and C4(T), are again deter­
equal to mined by imposing the boundary conditions mentioned above. It follows
′ that
v R pR (m + 1)(1 − 2μ)
KR = and GR = KR (27) ( )
κ 2(1 + (m − 1)μ) Ei 1, TH2
2

u = uR ( ) (34)
in which vR is the specific volume at the elastic–plastic boundary and Ei 1, TH2R
2

pR is the mean effective stress at the elastic–plastic boundary. This in­

′

vokes a small violation of the elasticity assumptions of modified Cam- in which Ei is the exponential integral. uR and HR depend on T and
clay. A more correct approach would be to have the bulk and shear thus the size of the expanding cavity. They also depend on v0 , p0 and the
′

moduli vary with v and p′ throughout the elastic region. However, that solution in the elastic–plastic region, satisfaction of the boundary con­
prevents closed form expressions for the elastic region from being ob­ ditions at the cavity wall, and maintaining a smooth continuation of u
tained. The constant moduli assumption adopted here is pragmatic and across the elastic–plastic boundary. uR and HR are evaluated iteratively
maintains consistency of elastic stiffness across the elastic–plastic when solving the elastic–plastic region, for a particular T, as will be
boundary. The differences between results using this approach and the demonstrated later.
approach where the moduli vary are not significant, as demonstrated We note that equations (32) and (34) are approximate expressions
later. for u in the elastic region as the histories of the pore pressure dissipations
For constant moduli, and a small strain assumption, the equilibrium prior to time T being attained, and how historic pore pressure dissipa­
equation and continuity equation can be combined to attain a simple tions affect the current pore pressure regimes at T, were not accounted
expression that links differential forms of εp and u or u (as per Section 3.5 for in their derivations. The expressions would be exact if the partially
in Verruijt, 2016) drained cavity expansions obeyed similarity. However, the deviation
dεp du dεp du from similarity is not major.
− D = or − = (28) The expressions for εp , p′ and v throughout the elastic region can be
dH dH dH dH
obtained
noting Ss = D = KR + m+1
2m
GR . Equation (16) may then be rewritten in
terms of dimensionless variables as εp = − u (35)
( )
H 1 dεp 1 d2 u m du KR 1 + (m − 1)μ
= 2
+ (29) p’ = p’0 + εp = p’0 + εp (36)
T Ts dH TH dH H dH D (m + 1)(1 − μ)

5
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

( ) The stress equilibrium equation is

v = v0 1 − εp = v0 (1 + u) (37)
’
dp m dq du mq
The equilibrium equation is then introduced and integrated, sub­ + + =− (48)
dH m + 1 dH dH H
jected to the constraint that q = qR at H = HR , to find the expressions
for.q The equation which introduces the elastic–plastic constitutive law is
For m = 2 ( )
1 np mp dp’ nq mp dq 1 dv
+ + + =0 (49)
qR HR 3 K h dH h dH v dH
q=
H3
⎡
( ) ( )
⎤ in which K = K/Ss and h = h/Ss .
2
exp − TH2R − exp − TH2
2
The continuity equation which ensures the stress state stays on the
2 u R HR 3
1 − 2μ ⎢
⎢
⎥
− ⎢ (√̅̅̅̅ ) ( )⎥
⎥ yield surface is
T H3 1 − μ ⎣√̅̅̅̅̅̅̅̅
̅
2πT HR 3 erfc 2T2 HR − 2HR 2 exp − THR 2
⎦
2
∂f dp’ ∂f dq ∂f dp’c
+ + =0 (50)
(38) ∂p’ dH ∂q dH ∂p’c dH
For m = 1 The differential equation for the hardening rule is
⎡ ⎤ ( )
( ) ( ) 1 dp’ λ − κ dp’c 1 dv
TH 2 THR 2
+ + =0 (51)
qR HR 2 2 uR HR 2 1 − 2μ ⎢
⎢ exp − − exp − ⎥
⎥ K dH vp’c dH v dH
(39)
2 2
q= − ⎢ ( ) ⎥
H2 T H2 1− μ ⎣ 2 THR 2 ⎦
HR Ei 1, 2 The differential form of the continuity equation which links volu­
metric change to strain rates is
In the elastic region the small strain definitions give εr = − ∂ζ/∂r and ( )
w − H dv dw m
εθ = − ζ/r, in which ζ is the displacement. It follows that ζ = − = w (52)
( εp εq ) v dH dH H
− r m+1 − m , in which εp = (p′ − p0 )/KR and εq = qm/(m + 1)/2/GR . The
′

velocity is obtained using w = ζ̇. Novelty lies in the other continuity equation which accounts for
At the elastic–plastic boundary it can be shown that partial drainage. In differential form it is
( )
εpR = − uR (40) T
w − H dv d(du/dH) m du
− − =0 (53)
v dH dH H dH
KR 1 + (m − 1)μ The seventh equation simply relates du/dH to the differential of u.
p’R = p’0 + εpR = p’0 + εpR (41)
D (m + 1)(1 − μ) The governing equations may be presented in matrix form as
⎡ ⎤
vR = v0 [1 + uR ] (42)
C11 C12 0 C14 0 0 0 ⎢ ⎥ − mq/H
The stress state at the elastic–plastic boundary must satisfy the yield ⎡
⎢
⎤⎢ dp’/dH ⎥
⎥ ⎡ ⎤
C21 C22 0 0 0 C26 0 ⎢ ⎥ 0
condition, which for modified Cam-clay means qR is given by ⎢ dq/dH ⎥
⎢ ⎥⎢ / ⎥ ⎢ ⎥
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ⎢ C31 C32 C33 0 0 0 0 ⎥⎢ dp’c dH ⎥ ⎢ 0 ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
(43)
′ ′ ′
qR = M 2 pR (pcR − pR ) ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ C41 0 C43 0 0 C46 0 ⎥⎢ du/dH ⎥=⎢ 0 ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ 0 0 0 0 0 C56 C57 ⎥⎢ d(du/dH)/dH ⎥ ⎢ mw/H ⎥
in which pcR is a function of vR and pR
′ ′
⎣ ⎦⎢ ⎥ ⎣ ⎦
⎢ ⎥
0 0 0 C64 C65 C66 0 ⎢ dv/dH ⎥ 0
( ′ ) ⎢ ⎥
1 N − vR − κln(pR D) ⎢ dw/dH ⎥
(44)
′
pcR = exp 0 0 0 C74 0 0 0 ⎣ ⎦ du/dH
D λ− κ

It is possible to express vR and pR in terms of the far field values v0 and (54)
′

p0 as well as uR
′
in which
1 + (m − 1)μ m
(45) (55)
′ ′
pR = p0 − uR C11 = 1, C12 = , C14 = 1
(m + 1)(1 − μ) m+1
( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) 1 1
1 1 4κuR (1 + (m − 1)μ) C21 = + np mp /h, C22 = nq mp /h, C26 = (56)
vR = v0 + 1+ ′ (46) K v
2 2 v0 ((m + 1)(1 − μ)p0 − (1 + (m − 1)μ)uR )
(57)
′
C31 = ∂f /∂p′ , C32 = ∂f /∂q, C33 = ∂f /∂pc
Also at the elastic–plastic boundary
qR 1 λ− κ 1
wR = HR (47) C41 = , C43 = ′ , C46 = (58)
2GR K vpc v

w− H
4.4. Elastic-plastic region C56 = , C57 = − 1 (59)
v
( )
In the elastic–plastic region the solution procedure involves solving m w− H
seven differential equations as an initial-value problem. Six of the var­ C64 = − , C65 = − 1, C66 = T (60)
H v
iables in the equations are p′ , q, pc , v, w, u. The seventh variable is dH ,
′
du

enabling the solution procedure to involve only first-order ordinary C74 = 1 (61)
differential equations. The equations are solved in terms of H. Additional
equations could be added in, for example involving variables εp and εq if 4.5. Relating quantities at the cavity wall and elastic–plastic boundary
the desire was to have those variables quantified, although that won’t be
done here. When generating results presented in Section 5 a range of v0 , p0
′

6
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

combinations were adopted. Solutions for each combination were path. When T→∞ the cavity is large and undrained conditions prevail
generated for a wide range of T. The solutions sought here are those around it due to the long drainage path.
which satisfy du/dH = 0 at the cavity wall, meaning no flow occurs Results are presented in the p’ , p ∼ q, v ∼ logp’, σ’r ∼ H and u ∼ H
across the cavity wall. This maintains an analogy to the CPT and PMT planes in Figs. 1-4, for initial values v0 = 1.9, p0 = 200 kPa, and v0 =
′

problems, where no flow occurs in to, or out of, the cone tip or pres­ 1.85, p0 = 400 kPa, and a variety of T, for cylindrical and spherical
′

suremeter wall. This boundary condition also enables conditions at the

cavities. The two sets of initial v0 and p0 correspond to lightly over
′

elastic–plastic boundary to be related to HR , by combining equations

consolidated and heavily over consolidated states, respectively.
(52) and (53), to give
The results represent the states in the soil around the expanding
( )
1 d(du/dH) m du dw m cavity at an instant in time when the cavity wall velocity, cavity radius
+ = + w (62)
T dH H dH dH H and coefficient of consolidation combine to give the inputted T. They do
not also represent a stress path followed by a soil element as an
Integrating equation (62), and imposing H = 1 when du/dH = 0,
expansion occurs. That would only be the case if similarity was obeyed
leads to
(Collins & Stimpson 1994). The diamond symbols represent points on
( )
du 1 the elastic–plastic boundary.
=T w− (63)
dH H m In the p’ ∼ q plane the effective stress states are indicted using
continuous lines, except for the drained case (T = 0) where medium-
Applying equation (63) at the elastic–plastic boundary, and intro­
dashed lines are used, and the undrained case (T = ∞) where long-
ducing the differential form of equations (32) or (34) to ensure
dashed lines are used. In the p ∼ q plane the total stress states are
smoothness of u across the elastic–plastic boundary, leads to an
indicated using short-dashed lines, except for the undrained case. At the
expression which relates HR , uR , wR and T.
cavity wall w = H = 1 and the soil is at the critical state.
For m = 2 it is
The stress states in the p’ ∼ q plane are noticeably influenced by T.
⎛ ⎞
( ) In the elastic region they deviate from constant p′ . As T increases the
(
1
) ⎜
⎜ 2exp − 2 THR 2 ⎟ elastic parts of the effective stress states become more inclined to the left
T wR − 2
= uR ⎜√̅̅̅̅̅̅̅̅̅ (√̅̅̅̅ ) ( )⎟
⎟ (64) meaning that p′ reduces. At larger T the states then move towards to
HR ⎝ 2πT H 2 erfc 2T HR − 2H exp − THR ⎠ 2
R 2 R 2 right and also become curved, approaching vertical lines as T→∞. The
intercepts with the CSL also occur at decreasing values of p′ and q as the
For m = 1 it is inputted T increases. There is a significant T influence in the total stress
⎛ ⎞ states in the p ∼ q plane as well.
( )
( ) TH
⎜2exp − 2 ⎟ R
2
In the v ∼ logp’ plane, due to the generation of u and the alteration
1
T wR −
⎜
= − uR ⎜ ( )⎟⎟ (65) of p′ in the elastic region, v changes from the initial value. Then as the
HR ⎝ H Ei 1, THR 2 ⎠
R 2 deformation becomes elastic–plastic v reduces. For the undrained cases v
remains constant at all stages. The uR increases from zero, then de­
Satisfaction of equation (64) or (65) is crucial to obtaining a correct creases and returns to zero, as the inputted T increases from 0 to ∞. The
solution. The search for a solution involves adjustment (iteratively) of associated changes of u within the elastic region, as H reduces from ∞ to
HR until H and w both become equal to unity at the cavity wall, with HR , are gradual for small values of T. As T increases the u alterations
parameters uR , wR , vR , pR , qR and pcR at the elastic–plastic boundary
′ ′ ′ become more and more concentrated within a small region near the
elastic–plastic boundary. In fact, for T = ∞, the changes become so
depending on the HR value as well as v0 , p0 and T. This simple iterative
′

concentrated that they form a localisation in the u development at the

adjustment of HR is similar to what is done when solving the drained
elastic–plastic boundary. The variations of w with H approach the power
cavity expansion problem.
law w = wR (H/HR )− m as T→∞.
An alternate critical state–based constitutive model may differ from
Additional results (Figs. 5 and 7) include the effective and total radial
modified Cam-clay through the choice of the yield function f and plastic
cavity wall pressures, σ rc and σ rc , as well as pore pressure at the cavity
′

potential g. It may also assume hardening occurs due to distortional

wall, uc , plotted against the inputted T for the initial values p0 = 200
′
strain as well as volumetric strain, introduce critical state and isotropic
compression lines that are not linear in the v ∼ lnp’ plane, and even kPa and v0 = 2.0,1.9,1.8, and for p0 = 400 kPa and v0 = 1.85,1.75,1.65,
′

adopt multiple (loading and bounding) surfaces rather than a yield for cylindrical and spherical cavities.
surface to accurately capture a soil’s non linearity prior to yielding. Even The σrc value for a drained condition (denoted σrc,d ), i.e. T→0, is
′ ′

so the same general solution procedure resented here may be applied. larger than that its respective value for an undrained condition (σrc,u ), i.
′

Changes would be made to equation (8) for f, equation (10) for the shape e. T→∞. The effective radial cavity wall pressures, as well as the pore
of the isotropic compression line and equations (18) and (19) for the pressure at the cavity wall, are normalised using the undrained values to
hardening laws. Equations (43), (44) and (46) would change accord­
form σ rc /σ rc,u and uc /uc,u , and plotted against T in Figs. 6 and 8.
′ ′

ingly. Additional hardening parameters may be introduced as well, for

The results in Figs. 5 and 7 indicate the values of cavity wall pres­
example one controlling the size of a loading surface and one controlling
sures for a given T, whatever the combination of wc , rc and c used in its
the size of a bounding surface if the single surface model concept is
definition as long as wc is a constant during the expansion. They also
abandoned. Finally, a specific equation for g may be specified rather
represent the pressures at the cavity wall as it expands continually with a
than obeying f = g.
constant velocity, i.e. as T increases from 0. Tracing the results from left
to right, for a given v0 and p0 combination, indicates how the normalised
′

5. Results
pressures vary with T as the cavity expands.
Cavity expansion results were generated using the material constants In Fig. 9, for a spherical cavity, p0 = 200 kPa, v0 = 1.90 and T = 0.3,
′

for London clay (as also used by Collins & Stimpson, 1994), φtx = 22.75◦ ,
′
results for which constant elastic moduli are assumed in the elastic re­
λ = 0.161, κ = 0.062, N = 2.858 and μ = 0.3. gion, equal to those at the elastic–plastic boundary, are compared to
The results become the same as those for drained and undrained results for which the elastic moduli vary. The results are shown as σ r and
′

conditions as T→0 and T→∞, respectively. When T→0 the cavity is u versus H. The differences are not significant. For example, the differ­
small and drained conditions prevail around it due to the short drainage ences in σr and u at the cavity wall for the two approaches are only 1%
′

7
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Fig. 1. Cylindrical cavity expansion results for v0 = 1.9, p0 = 200 kPa and T = 0, 0.03, 0.1, 0.3, 1, ∞ in the (a) p’ , p ∼ q, (b) v ∼ logp’, (c) σ’r ∼ H and (d) u ∼
′

H planes.

and 2%, respectively. Importantly, use of T = 0.3 reveals what must be 6. Presentations relevant to CPT results
the largest differences for the two moduli definitions (for any of the T =
0, 0.03, 0.1, 0.3, 1, 3, ∞ considered). For T = 0.3 the u changes in the The spherical cavity expansion results are now interpreted to provide
elastic region are the largest and so must be the differences between the some relevance to the cone penetration problem. σrc is analogous to the
′

elastic moduli computed by the two approaches. effective cone penetration resistance qt . qt is defined as qt − u where
′ ′

Numerical difficulties were encountered for large values of T. There qt = qc +u(1 − a) is the corrected total cone resistance, u is the pore
were instabilities in the differential equation solver. MATLAB R2021b pressure measured on or directly behind the cone tip (and assumed here
software was used and various solvers were trialled, with ode23t being to be equal to the pore pressure everywhere around the cone tip), a is a
the most successful. It is an implementation of the trapezoidal rule but parameter that depends on the cone geometry and qc is the total un­
uses a “free” interpolant. It was found to be more accurate and faster corrected cone penetration resistance.
than ode45 and ode15s. The T = ∞ undrained solution was obtainable, A normalised cone penetration resistance is defined as
however. Details of the solver types are given by MathWorks® (2023).
( ) qt − u
In the study by Mo et al. (2020), rather than solve the partially Qp 1 − Bqp + 1 = ′ (66)
drained cavity expansion problem, it was assumed that drained, un­ p0
drained and partially drained cavity wall pressures lie on a straight line in which Qp = qt p− ′ p0 and Bqp = u− u0
. Note the rather than the
qt − p0 qp q
in the p ∼ q plane. Mo et al. (2020) used that straight line to estimate 0

subscript on the pore pressure ratio B, owing to the use of p0 and p0

′

the cavity wall pressure for a given value of wc rc /c through interpola­

rather than the vertical stresses σv0 and σv0 in the denominators as
′
tion. Here it has been revealed that the total stress states at the cavity
walls lie on curved lines, contrary to this. The Mo et al. (2020) as­ featured in the original version by Houlsby (1988). Qp (1 − Bqp ) +1
(qt − p0 )
sumptions also result in non-zero values of du/dr at the cavity wall, in correctly transitions to p0
′ +1 as undrained conditions transition to
violation of the boundary condition most relevant to the CPT and PMT
drained conditions. It has also been adopted by Shuttle & Cunning
problems. Also, the pore pressure changes through the elastic region
(2007) and Shuttle & Jefferies (2016).
which occur during partially drained expansions were not captured by
It so happens that plots of Qp (1 − Bqp ) +1 (on a log scale) against the
Mo et al. (2020), again in violation of the correct solution.
state parameter ψ exhibit (near) unique relationships for drained and
undrained penetrations, of the general form

Qp (1 − Bqp ) + 1 = kexp( − mψ ) (67)

8
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Fig. 2. Cylindrical cavity expansion results for v0 = 1.85, p0 = 400 kPa and T = 0, 0.03, 0.1, 0.3, 0.6, ∞ in the (a) p’ , p ∼ q, (b) v ∼ logp’, (c) σ’r ∼ H and (d) u ∼
′

H planes.

Analogously, a normalised cavity wall pressure is defined here as would be very useful for CPT interpretations when partially drained
σ rc − uc conditions prevail and when the state parameter is to be determined.
Qp,ce (1 − Bqp,ce ) + 1 = ′ (68)
p0
7. Comparison to CPT results
such that
Results of CPTs in normally consolidated and over consolidated
Qp,ce (1 − Bqp,ce ) + 1 = kce exp( − mce ψ ) (69)
kaolin, involving a range of drainage conditions, were presented by
in which Qp,ce = (σ rc − p0 )
, Bqp,ce = σurcc −− up00 and kce , mce are fitting parame­ Schneider et al. (2007). The results were generated using a centrifuge in
p0
′
which a soil sample was contained in a 325 mm deep strong box that was
ters for the cavity expansion problem.
accelerated to 40 g or 160 g. A 10 mm diameter cone penetrated a
The constants k and m, and kce and mce , depend on the soil type and sample at a velocity in the range 0.0004 mm/s to 3 mm/s. Relevant
soil properties (Jefferies & Been, 2016; Shuttle & Jefferies, 1998). They spherical cavity expansion results are generated here and compared to
are also different for drained and undrained penetrations. Little is those CPT results. The modified Cam-clay input parameters for kaolin
known about how k and m, or kce and mce , transition from one set of were taken to be M = 0.92 (φtx = 23.5◦ ), λ = 0.26, κ = 0.026, N = 3.519
′

values to another for partially drained conditions. and μ = 0.3. The initial stress states in the kaolin samples were not
In Fig. 10 the logarithm of Qp,ce (1 − Bqp,ce ) +1 is plotted against ψ for
isotropic. Rather, the initial vertical stresses were σ v0 = 88.8 kPa and
′

a range of T values (0, 0.1, 0.3, 1, 3, ∞). For undrained conditions the
26.8 kPa, and k0 = σ h0 /σ v0 values were taken to be 0.601 and 1.442, for
′ ′

equation (69) fit is exact and kce,u = (2M + 3)/3 = 1.59 and mce,u = 1/
the normally and over consolidated cases, respectively. The initial mean
λ = 6.21. For drained conditions the fit is approximate with kce,d = 4.33 stresses of the cavity expansion analyses were then determined using
and mce,d = 2.90. The fits are also approximate for partially drained
p0 = (1+2k 0)
σ v0 , i.e. 65.2 kPa and 34.6 kPa, respectively. The initial
′ ′
3
conditions. The dependence on T is shown in Fig. 11, where the kce and specific volumes were v0 = 2.433 and 2.337, for the normally and over
mce values have been normalised using the drained quantities, kce,d and consolidated cases.
mce,d , and plotted against the logarithm of T (and against T (for T < 1) or Modified Cam-clay parameters were not given by Schneider et al.
2 − 1/T (for T > 1) in the inset). Such a presentation enables kce and mce (2007), nor were profiles of v within the tested samples immediately
to be determined as long as T is known and as long as kce,d and mce,d are prior to the cone penetrations. The M and μ were chosen as they are
known. typical for kaolin, the λ matched the slope of the normal compression
An analogously equivalent presentation to those of Figs. 10 and 11 line observed in Rowe cell tests, and N, κ and v0 were chosen so the one-

9
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Fig. 3. Spherical cavity expansion results for v0 = 1.90, p0 = 200 kPa and T = 0, 0.03, 0.1, 0.3, 1, 3, ∞ in the (a) p’ , p ∼ q, (b) v ∼ logp’, (c) σ’r ∼ H and (d) u ∼
′

H planes.

dimensional over consolidation ratios in the tested samples were the considered in this cavity expansion analysis. Otherwise there is good
same as those given by modified Cam-clay. agreement between theory and experiment, despite the initial stress
The dimensionless velocity notation of Schneider et al. (2007) will be states in the experiments being anisotropic and contrary to the initially
adopted in the comparison, specifically isotropic state in the cavity expansion analysis, and despite modified
Cam-clay not being well suited to clays when over consolidated.
vv dc
VCPT = (70) Fig. 13 presents the comparison between pore pressure ratios Bqp and
cv
Bqp,ce , divided by the undrained values, when plotted against VCPT . The
with cv defined as agreement is not as good as that in Fig. 12, as is common when
′ comparing pore pressure measurements to a theoretical prediction (e.g.
kv0 σv0
cv = (71) Silva et al., 2006) for the CPT. The cavity expansion theory produces a
γw λ
higher rate of pore pressure rise as VCPT increases. Also, for the over
The cv values for each test are listed by Schneider et al. (2007). It can consolidated case the spike in pore pressure rises at VCPT values between
be shown that about 100 and 200 was not reproduced, possibly due to the absence of
′
viscous effects in the theory.
VCPT (1 − μ) λ pR vR
= 2(1 + 2k0 ) ′ (72)
T 1 + μ κ p0 v0
8. Comments on permeabilities and soil types for which partial
It can also be shown that v0 ≈ vR for all T while p0 /pR has a slight
′ ′
drainage may exist
dependence on T with p0 /pR ≈ 1.2 being a reasonable approximation. It
′ ′

Permeabilities for which partial drainage may exist during a CPT can
follows that VCPT ≈ 20T and VCPT ≈ 32T for the normally and over
be estimated from the results in Figs. 7 and 8, assuming those results are
consolidated cases, respectively, assuming vv dc = 2wc rc .
approximately representative of a variety of soil types. The relevant T
Fig. 12 presents the comparison between cone penetration re­
range is from about 0.02 to 50. Then, for the standard cone diameter dc
sistances and spherical cavity wall pressures, defined by equations (66)
= 35.7 mm and velocity vv = 20 mm/s, and vv dc = 2wc rc , partial
and (68) and divided by the undrained values, when plotted against
drainage is possible when k lies between 2 × 10-5 m/s and 4 × 10-9 m/s.
VCPT . Close agreement is obtained. For the over consolidated case notice
The significance of the partial drainage, that is its closeness to a fully
that the partially drained cone penetration resistances are less than the
drained or undrained condition, depends on where the permeability lies
undrained values for VCPT values between about 100 and 200, possibly
in this range as well as the initial stress state and void ratio. Susceptible
due to viscous rate effects (Lehane et al., 2009), a phenomenon not
materials may include gravels and sandy gravels if they contain some silt

10
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Fig. 4. Spherical cavity expansion results for v0 = 1.85, p0 = 400 kPa and T = 0, 0.03, 0.1, 0.3, 1.0, 3.0, ∞ in the (a) p’ , p ∼ q, (b) v ∼ logp’, (c) σ ’r ∼ H and (d)
′

u ∼ H planes.

or clay (GM and GC), well graded sands and gravelly sands with little or compacted silts and clays (ML-MH, CL-CH), organic clays of high plas­
no silt or clay (SW), silty or clayey sands (SM and SC), and many silty ticity (OH), and some clayey tailings. It may also include some silty
tailings. They may also include some low plasticity silty clays and sandy sands (SM) or silty tailings. The permeabilities (and soil/tailings types)
clays (CL). for which partial drainage influences a result at smaller pressuremeter
The results for cylindrical cavities in Figs. 5 and 6 are not so easily expansions can not be inferred until a technique like that of Collins & Yu
related to the PMT as pressuremeter expansions start from a non-zero (1996) and Russell & Khalili (2002) is applied.
radii whereas the cavity expansions start from zero initial radii. For
undrained conditions, Collins & Yu (1996) developed a technique to 9. Conclusions
interrelate cavity expansions starting from zero and non-zero radii as
similarity is obeyed, exploiting the exact power law relationship be­ The cylindrical and spherical cavity expansion problem in a partially
tween w and r. Russell & Khalili (2002) adapted and applied that tech­ drained soil can be solved semi-analytically when the cavity wall ex­
nique to drained conditions, again where similarity is obeyed, by pands with a constant velocity wc . The condition that there is no water
introducing a small approximation to the way w relates to r. It may be flow across the cavity wall, akin to the conditions at the tip of a pene­
true that this adapted technique and approximation also applies to trating cone in a CPT or at the wall of a pressuremeter, is imposed.
partially drained conditions, even though similarity is not obeyed, The procedure involves solving separate sets of equations for the
although a targeted study is required to investigate this. Even so, if a elastic and elastic–plastic regions, with conditions at the elastic–plastic
pressuremeter is expanded to about twice its initial radius then the soil boundary being common to the two solutions. The variables in the
response around it resembles that around a created cylindrical cavity. equations are made dimensionless by dividing them by length, time,
The so called limit pressure of a PMT can be recovered at that amount of velocity and stress scales, with each scale depending on input constants
expansion (ASTM, 2020). It is for this condition that we take guidance which include wc , the coefficient of consolidation c and a constrained
from the results in Figs. 5 and 6. The relevant T for partial drainage to modulus of elasticity D. The time t after expansion commences controls
exist range from about 0.008 to 20. Then, for a pressuremeter expanded the cavity size rc through rc = wc t.
from a radius rc of 37 mm to 74 mm, and an expansion velocity wc = 3.1 When partial drainage occurs the laws of similitude are not obeyed.
× 10-5 mm/s (being a 2.5% cavity volume increase (relative to the initial With the constant wc assumption the velocities at locations around the
volume) every 15 s as per ASTM (2020)), partial drainage may exist cavity, other than the cavity wall, are not constant.
when k lies between 3 × 10-7 m/s and 7 × 10-11 m/s. Materials with A key dimensionless quantity that influences the cavity expansion
permeabilities in this range include organic silts and silty clays of low results is wc rc /c = w2c t/c. It is referred to here as a dimensionless time T,
plasticity (OL), inorganic silts and clays of high plasticity (MH and CH), although resembles a dimensionless velocity used by others, including in

11
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Fig. 6. Cylindrical cavity expansion results: σ rc /σrc,u , σ rc /σrc,u and uc /uc,u

′ ′

Fig. 5. Cylindrical cavity expansion results: σ rc , σrc and uc against T for p0 =

′ ′

against T for p0 = 200 kPa, v0 = 2.0, 1.9, 1.8, and for p0 = 400 kPa, v0 = 1.85,
′ ′

200 kPa, v0 = 2.0, 1.9, 1.8, and for p0 = 400 kPa, v0 = 1.85, 1.75, 1.65. The
′

1.75,1.65. The numbers in the parentheses in (a) indicate the relevant p0 and v0
′

numbers in the parentheses in (a) indicate the relevant p0 and v0

′

becomes partially drained. At very large rc , t and T the response becomes

CPT investigations. Treating the wc rc /c = w2c t/c quantity as a dimen­ undrained. The time-dependant changes to conditions at the cavity wall
sionless time makes sense when understanding how the soil response are revealed in plots of σrc , σrc , uc versus T. Other results show how the
′

changes around a cavity as it expands in time at a constant velocity.

conditions away from the cavity wall also vary with T.
Initially, when rc and t are small, T is also small and (near) drained
An alternate interpretation of the results, also correct and more in
conditions prevail. As rc and t become larger, so does T, and the response

12
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Fig. 7. Spherical cavity expansion results: σrc , σ rc and uc against T for p0 = 200 Fig. 8. Spherical cavity expansion results: σ rc /σrc,u , σ rc /σrc,u and uc /uc,u against
′ ′ ′ ′

kPa, v0 = 2.0,1.9,1.8, and for p0 = 400 kPa, v0 = 1.85,1.75,1.65. The numbers

′
T for p0 = 200 kPa, v0 = 2.0, 1.9, 1.8, and for p0 = 400 kPa, v0 = 1.85, 1.75,
′ ′

in the parentheses in (a) indicate the relevant p0 and v0

′
1.65. The numbers in the parentheses in (a) indicate the relevant p0 and v0
′

accord with the view that wc rc /c = w2c t/c is a dimensionless velocity, is cone penetration resistances, from CPTs performed using a wide variety
that they apply to when cavities are expanded at different yet constant of penetration velocities, vary with the normalised velocity vv dc /cv , after
wc values to attain a certain rc . making the two dimensionless quantities comparable. The normalised
The way normalised effective cavity wall pressures vary with wc rc / pore pressures at the cavity wall, however, differ significantly from
c = w2c t/c is shown to be very similar to the way normalised effective those at the cone tips, as observed by others as well.

13
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Fig. 9. Spherical cavity expansion results for p0 = 200 kPa, v0 = 1.90 and T =
′

0.3 in the σr and u versus H plane. Continuous lines are for when constant
′

elastic moduli are assumed in the elastic region, equal to those at the elas­
Fig. 11. Spherical cavity expansion results: Normalised parameters kce and mce
tic–plastic boundary. Dashed lines are for when the elastic moduli vary. ( )
which define the exponential fits between Qp,ce 1 − Bqp,ce +1 and ψ for a given
T, plotted against the logarithm of T (and in the inset against T and 2 − T1). The
parameters for drained and undrained conditions are kce,d = 4.33, mce,d = 2.90
and kce,u = 1.59,.mce,u = 6.21.

( )
Fig. 10. Spherical cavity expansion results: Qp,ce 1 − Bqp,ce +1 against ψ for
T = 0, 0.1, 0.3, 1, 3, ∞, with exponential fits shown for each T. Solid symbols
were recovered using p0 = 200 kPa while hollow symbols were recovered using
′

p0 = 400 kPa. The equation part on the vertical axis title is not displayed
′

properly. see accepted paper for correct version.

Plots of the normalised spherical cavity wall pressures against the

state parameter also show that the two key fitting constants, kce and mce ,
are heavily influenced by partial drainage. A similar dependence must
exist when plotting a normalised cone penetration resistance against the
state parameter when partially drained conditions prevail, as is often the
case in many soils and tailings for example. One way to deal with partial
drainage may be to convert CPT results to drained equivalent values
prior to the establishment of relationships with the state parameter.
The practical ramifications of not dealing with this partial drainage
may be huge, especially when assessing liquefaction susceptibility
through estimations of the state parameter. For example, it may not be
correct to determine k and m from λ10 (≈ 2.3λ) using Fg 4.29 in Jefferies
& Been (2016) and expect those values to apply to the CPT if partially
drained or undrained conditions exist. Most of the data used to establish ( )
the trend in Fg 4.29 is for CPTs conducted when drained conditions Fig. 12. Normalised cone penetration resistances Qp 1 − Bqp +1 and spherical
( )
prevailed. A significant underestimation of state parameter, liquefaction cavity wall pressures Qp,ce 1 − Bqp,ce + 1, divided by the undrained values,
susceptibility and post-liquefaction strength may result. There may be a plotted against VCPT for (a) normally and (b) over consolidated kaolin. Note
hidden, and so far unknown, correlation between VCPT (or T) and λ10 , as VCPT ≈ 20T and VCPT ≈ 32T for the normally and over consolidated cases,
respectively.
more compressible soils with larger λ10 are generally finer grained soils
with lower permeability and larger VCPT (or T). Fg 4.29 could have wider

14
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

focussed study to confirm.

CRediT authorship contribution statement

Adrian R. Russell: Conceptualization, Methodology, Software,

Formal analysis, Investigation, Writing – original draft, Project admin­
istration, Funding acquisition. Yanzhi Wang: Software, Validation,
Formal analysis, Investigation. Nasser Khalili: Conceptualization,
Methodology, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence
the work reported in this paper.

Data availability

Data will be made available on request.

Acknowledgments

This work forms part of TAILLIQ (Tailings Liquefaction), which is an

Australian Research Council (ARC) Linkage Project (LP160101561)
supported by financial and in-kind contributions from Anglo American,
BHP, Freeport-McMoran, Newmont, Rio Tinto and Teck. The TAILLIQ
project is being carried out at The University of New South Wales, The
University of South Australia, The University of Western Australia (lead
organisation) and The University of Wollongong. We acknowledge the
support and contributions of project personnel at each of the supporting
organisations. The work also forms part of an ARC Future Fellowship
(FT200100820) awarded to the first author and that funding is grate­
fully acknowledged.

Fig. 13. Pore pressure ratios Bqp and Bqp,ce (for spherical cavities), divided by
the undrained values, plotted against VCPT for (a) normally and (b) over
consolidated kaolin. Note VCPT ≈ 20T and VCPT ≈ 32T for the normally and
over consolidated cases, respectively.

applicability than to just drained conditions, but that would need a

Appendix A

Here the differential equations relating a variable x to Н, T, r and t are presented.

x is a function of variables r and t so we can write x = x(r, t). Also, r and t are functions of variables H and T so we can write r = r(H, T) and t =
t(H, T). The functions for r and t are
r = r(H, T) = wc tH = Vs Ts TH = Ls TH (A.1)

t = t(H, T) = cT/wc 2 = Ts T (A.2)

in which Ls and Ts are constants.
x is then a function of variables H and T so we can write x = x(H, T). From the chain rule
∂x ∂x ∂r ∂x ∂t
= + (A.3)
∂H ∂r ∂H ∂t ∂H
and
∂x ∂x ∂r ∂x ∂t
= + (A.4)
∂T ∂r ∂T ∂t ∂T
From equations (A.1) and (A.2) we obtain the partial derivatives
∂r
= Ls T (A.5)
∂H

15
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

∂r
= Ls H (A.6)
∂T
∂t
=0 (A.7)
∂H
∂t
= Ts (A.8)
∂T
Substituting equations (A.5) to (A.8) in to (A.3) and (A.4) gives
∂x ∂x
= Ls T (A.9)
∂H ∂r
and
∂x ∂x ∂x
= Ls H + Ts (A.10)
∂T ∂r ∂t
Equation (A.9) is easily rearranged to obtain
∂x 1 ∂x Xs ∂ x
= = (A.11)
∂r Ls T ∂H Ls T ∂H
We can substitute equation (A.9) in to equation (A.10) to obtain
∂x H ∂x ∂x
= + Ts (A.12)
∂T T ∂H ∂t
In this study the constraint that T is fixed is imposed to obtain a solution. The means ∂∂Tx = 0 in the solution. We can then rearrange equation (A.12)
to obtain
∂x H ∂x HXs ∂x
=− =− (A.13)
∂t TTs ∂H TTs ∂H
Since ∂∂Tx = 0 we are effectively evaluating x(H, T) in terms of H only, i.e. at a fixed value of T. x(H, T) makes a three-dimensional surface in the x :
H : T space. The solution we obtain for x(H, T) in terms of H is on the plane normal to the T axis cut through the surface.
Because we do not treat T as a variable the partial derivative ∂∂Hx can then be treated as a total derivative dH dx
. Equations (A.11) and (A.13) then
become
∂x 1 dx Xs dx
= = (A.14)
∂r Ls T dH Ls T dH

∂x H dx HXs dx
=− =− (A.15)
∂t TTs dH TTs dH
which are equations (24) and (25). The interdependencies between x, H, T, r and t are correctly accounted for in the derivations prior to ∂∂Hx being
interchanged with dH
dx
.

References Collins, I.F., Pender, M.J., Yan, W., 1992. Cavity expansion in sands under drained
loading conditions. International Journal of Numerical and Analytical Methods in
Geomechanics. 16 (1), 3–23.
ASTM (2020). D4719–20. Standard test methods for prebored pressuremeter testing in soils.
Collins, I.F., Stimpson, J.R., 1994. Similarity solutions for drained and undrained cavity
ASTM International. doi:10.1520/D4719-20.
expansions in soils. Géotechnique. 44 (1), 21–34.
Carter, J.P., Booker, J.R., Yeung, S.K., 1986. Cavity expansion in cohesive frictional soil.
Cudmani, R., Osinov, V.A., 2001. The cavity expansion problem for the interpretation of
Géotechnique. 36 (3), 349–358.
cone penetration and pressuremeter tests. Canadian Geotechnical Journal. 38 (3),
Carter, J.P., Yu, H.S., 2022. Cavity expansion in cohesive-frictional soils with limited
622–638.
dilation. Géotechnique. https://doi.org/10.1680/jgeot.21.00141.
Hanna, A., 2001. Determination of plane-strain shear strength of sand from the results of
Ceccato, F., Beuth, L., Simonini, P., 2016. Analysis of piezocone penetration under
triaxial tests. Canadian Geotechnical Journal. 38 (3), 1231–1240.
different drainage conditions with the two-phase material point method. Journal of
Houlsby, G.T., 1988. Introduction to papers 14–19. In: Penetration testing in the UK.
Geotechnical and Geoenvironmental Engineering, ASCE. 142, 4016066. https://doi.
Thomas Telford, London, pp. 141–146.
org/10.1061/(ASCE)GT.1943-5606.0001550.
Jefferies, M.G., Been, K., 2016. Soil liquefaction: A critical state approach, 2nd Edition.
Ceccato, F., Simonini, P., 2017. Numerical study of partially drained penetration and
CRC Press.
pore pressure dissipation in piezocone test. Acta Geotechnica. 12 (1), 195–209.
Jiang, M.J., Sun, Y.G., 2012. Cavity expansion analyses of crushable granular materials
Chen, S.L., Abousleiman, Y.N., 2018. Cavity expansion in strain hardening frictional soils
with state-dependent dilatancy. International Journal for Numerical and Analytical
under drained condition. International Journal for Numerical and Analytical
Methods in Geomechanics. 36 (6), 723–742.
Methods in Geomechanics. 42 (1), 132–142.
Lehane, B.M., O’Loughlin, C.D., Gaudin, C., Randolph, M.F., 2009. Rate effects on
Chen, H., Li, L., Li, J., Wang, H., 2019. Stress transform method to undrained and drained
penetrometer resistance in kaolin. Géotechnique. 59 (1), 41–52.
expansion of a cylindrical cavity in anisotropic modified cam-clay soils. Computers
Liu, K., Chen, S.L., 2019. Analysis of cylindrical cavity expansion in anisotropic critical
and Geotechnics. 106, 128–142.
state soils under drained conditions. Canadian Geotechnical Journal. 56 (5),
Chen, S.L., Liu, K., 2019. Undrained cylindrical cavity expansion in anisotropic critical
675–686. https://doi.org/10.1139/cgj-2018-0025.
state soils. Géotechnique. 69 (3), 189–202. https://doi.org/10.1680/jgeot.16.P.335.
MathWorks® (2023). https://www.mathworks.com/help/matlab/math/choose-an-ode-
Cheng, W., Chen, R.P., Pereira, J.M., Cui, Y.J., 2022. Undrained cylindrical cavity
solver.html. Accessed on 20 January 2023.
expansion/contraction in stiff clays using a two-surface plasticity model.
Mo, P.-Q., Gao, X.-W., Yang, W., Yu, H.-S., 2020. A cavity expansion–based solution for
International Journal for Numerical and Analytical Methods in Geomechanics. 46
interpretation of CPTu data in soils under partially drained conditions. International
(3), 570–593. https://doi.org/10.1002/nag.3312.
Journal for Numerical and Analytical Methods in Geomechanics. 44 (7), 1053–1076.
Collins, I.F., Yu, H.S., 1996. Undrained cavity expansions in critical state soils.
Mo, P.Q., Yu, H.S., 2017. Undrained cavity expansion analysis with a unified state
International Journal of Numerical and Analytical Methods in Geomechanics. 20 (7),
parameter model for clay and sand. Géotechnique. 67 (6), 503–515.
489–516.
Monforte, L., Gens, A., Arroyo, M., Mánica, M., Carbonell, J.M., 2021. Analysis of cone
penetration in brittle liquefiable soils. Computers and Geotechnics. 134, 104123.

16
A.R. Russell et al. Computers and Geotechnics 158 (2023) 105381

Nanda, S., Patra, N.R., 2015. Determination of soil properties for plane strain condition Sivasithamparama, N., Castro, J., 2018. Undrained expansion of a cylindrical cavity in
from the triaxial tests results. International Journal for Numerical and Analytical clays with fabric anisotropy: theoretical solution. Acta Geotechnical. 13 (3),
Methods in Geomechanics. 39 (9), 1014–1026. 729–746.
Palmer, A.C., 1972. Undrained plane-strain expansion of a cylindrical cavity in clay - Sivasithamparama, N., Castro, J., 2020. Undrained cylindrical cavity expansion in clays
simple interpretation of pressuremeter test. Géotechnique. 22 (3), 451–457. with fabric anisotropy and structure: Theoretical solution. Computers and
Pournaghiazar, M., Russell, A.R., Khalili, N., 2012. Linking cone penetration resistances Geotechnics. 120, 103386 https://doi.org/10.1016/j.compgeo.2019.103386.
measured in calibration chambers and the field. Géotechnique Letters. 2 (2), 26–35. Suzuki, Y., Lehane, B.M., 2015. Analysis of CPT end resistance at variable penetration
Pournaghiazar, M., Russell, A.R., Khalili, N., 2013. Drained cavity expansions in soils of rates using the spherical cavity expansion method in normally consolidated soils.
finite radial extent subjected to two boundary conditions. International Journal for Computers and Geotechnics. 69, September, 141–152.
Numerical and Analytical Methods in Geomechanics. 37 (4), 331–352. Verruijt, A. (2016). Theory and problems of poroelasticity. Unpublished book at https://
Russell, A.R., Khalili, N., 2002. Drained cavity expansion in sands exhibiting particle geo.verruijt.net/ pp. 274.
crushing. International Journal for Numerical and Analytical Methods in Vesic, A.S., 1972. Expansion of cavities in infinite soil mass. Journal of Soil Mechanics
Geomechanics. 26, 323–340. https://doi.org/10.1002/nag.203. and Foundations Division, ASCE. 98 (SM3), 265–290.
Russell, A.R., Khalili, N., 2006. On the problem of cavity expansion in unsaturated soils. Wroth, C.P., 1984. The interpretation of in situ soil tests. Géotechnique. 34 (4), 449–489.
Computational Mechanics. 37, 311–330. https://doi.org/10.1007/s00466-005- Yang, H., Russell, A.R., 2015. Cavity expansion in unsaturated soils exhibiting hydraulic
0672-7. hysteresis considering three drainage conditions. International Journal for
Salgado, R., Prezzi, M., 2007. Computation of cavity expansion pressure and penetration Numerical and Analytical Methods in Geomechanics. 39 (18), 1975–2016.
resistance in sands. International Journal of Geomechanics. 7 (4), 251–265. Yi, J.T., Goh, S.H., Lee, F.H., Randolph, M.F., 2012. A numerical study of cone
Schneider, J.A., Lehane, B.M., Schnaid, F., 2007. Velocity effects on piezocone penetration in fine-grained soils allowing for consolidation effects. Géotechnique. 62
measurements in normally and over consolidated clays. International Journal of (8), 707–719.
Physical Modelling in Geotechnics. 7 (2), 23–34. Yu, H.S., 1994. State parameter from self-boring pressuremeter tests in sand. Journal of
Sheng, D., Kelly, R., Pineda, J., Bates, L., 2014. Numerical study of rate effects in cone Geotechnical Engineering, ASCE. 120 (12), 2118–2135.
penetration test. In: Proceeding of the 3rd International Symposium on Cone Yu, H.-S. (Ed.), 2000. Cavity Expansion Methods in Geomechanics. Springer Netherlands,
Penetration Testing. Las Vegas, Nevada, USA, pp. 419–428. Dordrecht.
Shuttle, D.A., Cunning, J., 2007. Liquefaction potential of silts from CPTu. Canadian Yu, H.S., Carter, J.P., 2002. Rigorous similarity solutions for cavity expansion in
Geotechnical Journal 44 (1), 1–19. cohesive-frictional soils. International Journal of Geomechanics. 2 (2), 233–258.
Shuttle, D., Jefferies, M., 1998. Dimensionless and unbiased CPT interpretation in sand. Yu, H.S., Houlsby, G.T., 1990. A new finite element formation for one-dimensional
International Journal for Numerical and Analytical Methods in Geomechanics. 22 analysis of elastic-plastic materials. Computes and Geotechnics. 9 (4), 241–256.
(5), 351–391. Yu, H.S., Schnaid, F., Collins, I.F., 1996. Analysis of cone pressuremeter tests in sands.
Shuttle, D., Jefferies, M., 2016. Determining silt state from CPTu. Geotechnical Research Journal of Geotechnical Engineering, ASCE. 122 (8), 623–632.
3 (3), 90–118. Zhou, H., Kong, G., Liu, H., Laloui, L., 2018. Similarity solution for cavity expansion in
Silva, M.F., White, D.J., Bolton, M.D., 2006. An analytical study of the effect of thermoplastic soil. International Journal for Numerical and Analytical Methods in
penetration rate on piezocone tests in clay. International Journal for Numerical and Geomechanics. 42 (2), 274–294.
Analytical Methods in Geomechanics. 30 (6), 501–527.

17