You are on page 1of 88

R

modeling soil behavior


Cino Viggiani, laboratoire 3SR cino.viggiani@hmg.inpg.fr

soils – what are they (fundamentals)

soil behavior (from lab tests)


stress,
constitutive
strain
models
invarian
(elasticity +
ts
models of soil behavior plasticity)
stress
paths

Æ geotechnical engineering (FEM analysis)


soils are multiphase materials R

for the simple case of saturated (two-phase) soils


(the principle of) effective stresses R

the father of soil mechanics, Karl Terzaghi (1883 – 1963)


(the principle of) effective stresses R
drained / undrained R

in undrained conditions, no water movement takes place and,


therefore, excess pore pressures are built up
Δu ≠ 0, Δσ ≠ Δσ'

in drained conditions, no excess pore pressures are built up


Δu = 0, Δσ = Δσ'

• drained analysis appropriate when


– permeability is high
– rate of loading is low
– short term behavior is not of interest for problem considered

• undrained analysis appropriate when


– permeability is low and rate of loading is high
– short term behavior has to be assessed
triaxial test (NC soils) – drained / undrained R

drained undrained
triaxial test (OC soils) – drained / undrained R

drained undrained
stress paths in undrained triaxial test – NC / OC R
stress invariants R

in terms of principal stresses σ1, σ2, σ3

the stress state is completely defined by any of the following three triplets:

( σ1, σ2, σ3 ) or ( I1, I2, I3 ) or ( p, q, θ )

σ2 − σ3 2 ⋅ tan (θ )
b = =
σ 2 − σ3 allows to quantify the influence σ1 − σ 3 3 + tan (θ )
b= of σ2 (usually neglected), i.e.
σ1 − σ 3 the variation of Lode angle ⎛ b⋅ 3
θ = tan −1 ⎜


⎜ 2−b ⎟
⎝ ⎠
R

σ1
B
stress space diagonal
σ1
B
A θ
deviatoric plane
deviatoric plane

O A
O
σ3

σ2 σ2 σ3
OA = 3 ⋅ p
2 θ = Lode angle
AB = 2 ⋅ J 2 = q
3

the stress state is completely defined in terms of p, q and θ


stress and strain invariants R

we will use the following invariant quantities:

q=
1
2
[
(σ 1 − σ 2 ) + (σ 2 − σ 3 ) + (σ 3 − σ 1 )
2 2 2 1/ 2
]
deviatoric
εs =
2
(ε1 − ε 3 )2 + (ε 2 − ε 3 )2 + (ε 3 − ε1 )2
3

σ1 + σ 2 + σ 3
p=
volumetric 3
ε v = ε1 + ε 2 + ε 3

these invariants are work-conjugate, i.e., total work done in strain increment is :

δW = σ1δε1 + σ2δε2 + σ3δε3 = pδεv + qδεs = δWp + δWq


R

geomechanics convention (for stress and strain): compression is positive

principal stresses
• σ1 : major principal stress (the most compressive)
• σ2 : intermediate principal stress
• σ3 : minor principal stress (the least compressive)

common states of stress in laboratory tests:


• isotropic compression, if the three principal stresses are equal and positive:
σ1= σ2= σ3
• uniaxial (unconfined) compression, if the major principal stress is positive,
and the two others are zero: σ1>0, σ2= σ3=0
• uniaxial (unconfined) extension, if the minor principal stress is negative, and
the two others are zero: σ3<0, σ1= σ2=0
• axisymmetric compression, if the minor and intermediate principal stresses
are equal and positive: 0< σ3= σ2< σ1
• axisymmetric extension, if the major and intermediate stresses are equal
and positive (the minor one can be positive or negative): 0< σ2= σ1
R
R

Stress paths in real problems

• Stresses:
– Lambe → (pL,qL,p'L)
– Cambridge
• 3D → (pC,qC,p'C)
• 2D → (s,t,s')
• In the following examples the Lambe representation will be used:

1 1 1
pL = (σ 1 + σ 3 ) pL′ = (σ 1′ + σ 3′ ) qL = (σ 1 − σ 3 )
2 2 2
R

Example 1: Embankment

embankment
f

≡ α

Stress increments at point M due to load f :


f ⎫
Δσ 1 = (α + sin α ) ⎪⎪
π
⎬ elastic solution (independent of E, ν)
f ⎪
Δσ 3 = (α − sin α ) ⎪⎭
π
R

Example 1: Embankment

Δσ 1 + Δσ 3 f 1 f
ΔpL = = ( 2 ⋅α ) ⋅ = α
2 π 2 π
Δσ 1 − Δσ 3 f 1 f
ΔqL = = ( 2 ⋅ sin α ) ⋅ = sin α
2 π 2 π

ΔqL sin α
= constant for point M
ΔpL α

therefore the (total) stress path is a straight line of


slope sin α
tan β =
α
R

Example 1: Embankment

qL

very deep, β = π/4

intermediate, 0 < β < π/4

qini under the load, β = 0

pini pL
• Point M:
– At the surface: α = π ⇒ sin α = 0 ⇒ β = 0
– Very deep: α → 0 ⇒ (sin α)/α → 1 ⇒ β → π/4
R

Example 2: Excavation

excavation
≡ α

Stress increments at point M due to load f :


f ⎫
Δσ 1 = ( sin α − α ) ⎪⎪
π
⎬ elastic solution (independent of E, ν)
f
Δσ 3 = ( −α − sin α ) ⎪⎪
π ⎭
R

Example 2: Excavation

Δσ 1 + Δσ 3 f 1 f
ΔpL = = ( −2 ⋅ α ) ⋅ = − α
2 π 2 π
Δσ 1 − Δσ 3 f 1 f
ΔqL = = ( 2 ⋅ sin α ) ⋅ = sin α
2 π 2 π

ΔqL sin α constant for point M


=−
ΔpL α

therefore the (total) stress path is a straight line of


slope sin α
tan β = −
α
R

Example 2: Excavation

qL

very deep, β = -π/4

intermediate,
-π/4 < β < 0

qini
under the load, β = 0
pini pL
• Point M:
– At the surface: α = π ⇒ sin α = 0 ⇒ β = 0
– Very deep: α → 0 ⇒ (sin α)/α → 1 ⇒ β → -π/4
R
R

categories of laboratory tests

• triaxial compression

“classic” tests • oedometric compression

• direct shear

• plane strain compression


more complex tests • simple shear

• torsional shear

• true triaxial
uncommon tests
• “special” tests
the triaxial test R

symmetry axis

σv

σh σh
conventional triaxial apparatus
R

σr σr

⎛ a⎞ F
σ a = σ r ⋅ ⎜1 − ⎟ +
Tie bars
⎝ A⎠ A
F a
q = σa − σr = − σr
A A

F
=q if a<<A
A

F
= σa if a=A
A
conventional triaxial apparatus
R

σr σr

⎛ a⎞ F
σ a = σ r ⋅ ⎜1 − ⎟ +
⎝ A⎠ A
F a
q = σa − σr = − σr
A A

F
=q if a<<A
A

F
= σa if a=A
A
the triaxial test R

I σ + 2σ r σ1 = σ z
p= 1 = a
3 3
q = σa-σr
σ3 = σ x = σr
ε v = ε1 + 2ε r σ3 = σ y = σr

εs =
2
(ε a − ε r )
3

variations of cross-sectional area homogeneous deformation


Do
π π
A= D2 = ( Do + ΔD ) 2
4 4
⎛ ε − εa ⎞
ΔD = ⎜ v ⎟⋅D ho h
⎝ 2 ⎠

D
the triaxial test R

triaxial compression (σa>σr)

σ1
σ1=σ2= σ3
σ1=σa
triaxial compression plane

σ3 σ2=σr σr = σ3
45°

σ2
b = 0, θ = 0°
the triaxial test R

triaxial extension (σr>σa)

σ1 σ1=σ2= σ3

σ3=σa
triaxial extension plane

45°
σ3 σ1=σr σr = σ2

σ2

b = 1, θ = 60°
warning: σa and σr or σ1 and σ3 ?
R

compression

σa σ1
σr1=σr2= σa

θ = 60
C
“triaxial” plane extension E
O
σr2
45°

σr1 σ2 σ3
σa

OC = triaxial compression
O OE = triaxial extension

E
σr1 σr2
warning: soil behavior depends on the 3rd invariant !
R

σ1
triaxial compression

non triaxial tests

Drucker-Prager

triaxial extension
Mohr-Coulomb

σ2 σ3

Saada e Bianchini, 1988, Constitutive Equations for granular non-cohesive soils, Balkema
stress paths in a triaxial test R

standard triaxial compression: σr = constant (dq/dp = 3, b = 0, θ = 0)

τ B
σa
A

A B σ σr

B
q=σa-σr
problems with such a path 3

• from a "constitutive" viewpoint 1

- both p and q do change


A p
• from a "characterization" viewpoint
- unrealistic path
stress paths in a triaxial test R

constant p triaxial path: p = constant (b = 0, θ = 0)

τ
σa B 1

2
dσr dσa
A

A
σ
σr
B

B
advantages with such a path q =σa-σr
• stress state remains into the
initial deviatoric plane
• allows to separate responses to
volumetric and deviatoric loading A p
Æ this requires capability of stress control
stress-controlled paths, "advanced " TX systems R

"stress-path" triaxial systems (Bishop & Wesley 1975) are now conventional

a control program monitors the pressures and displacements, controls the stresses
and strains, and allows the user to define the triaxial testing stages with automatic
changes of the stress paths
automated stress-path system: schematic diagram R

cell pressure + ram pressure + back


pressure are supplied by a compressor

air pressures provided by the


compressor are applied to the sample
as hydraulic pressure through air-water
interface systems

the pressures are managed by 3


stepper-motor driven manostats
controlled by a computer through relay
outputs of the data-logging and control
card

a stepper-motor driven pump


connected to the hydraulic system and
controlled directly by the computer
controls the water flow to the base
piston in strain-controlled tests
pressure controllers
probing soil response upon a variety of stress paths R

"stress rosette" : input stress paths …

all these stress paths still belong to the triaxial plane


probing soil response upon a variety of stress paths R

… and output strain paths

investigating the directional dependence of soil behavior (Costanzo et al. 2006)


the triaxial test: possibilities and limitations R

several different tests can be performed in the triaxial apparatus

for example: if σa and σr are increased together such that the radial
strain εr = 0, the sample will deform in one dimensional compression,
just like in oedometric conditions (so-called K0 compression)

BUT: don’t forget that we do not escape out of the triaxial plane (2
out of the 3 principal stresses are equal to each other)

AND: principal directions of stress (and strain) are fixed throughout a


test (except when a shear band forms)

PLUS: there are a few "troubles“ (sources of errors) with the test
• friction at specimen ends
• inaccuracies in stress and strain measurement
• strain non homogeneity
• ...
conventional triaxial test: main sources of errors R

low stiffness of the loading frame

external measurement of
axial load

external measurement of axial strain

friction ram - collar

low quality connection ram – top


platen

result: measured response strongly depends on the apparatus,


while it should only depend of the material
the triaxial test: troubles (2) R

errors in the measurement of axial displacement

JARDINE, R. J., SYMES, M. J. & BURLAND, J. B. (1984). Géotechnique


back to standard testing: the oedoemeter test R

• to determine the one-dimensional compression response of a soil


• to investigate time-dependent behavior of soil (one-dimensional consolidation + creep)

widely used in practice, classical in soil mechanics for obtaining parameters for calculation of
consolidation settlements and for assessing stress history of soils

soil lateral movement is prevented by containing the specimen in a “rigid“ (stainless steel) ring so
that only vertical movements are possible

drainage is only permitted from the horizontal faces so that the water in the pores is also
constrained to flow vertically

symmetry axis

σv

εr=0
oedometric compression R

what we do and what we do not control

• radial symmetry (two principal stresses are coincident, as in the triaxial test).
Direction of principal stresses is prescribed and constant throughout a test

• one can control either vertical (principal) stress or vertical displacement.


Horizontal (radial) strain is zero.

• symmetry Æ Lode angle is constant, cannot be controlled

• stress state is unknown (radial stress?)

F F F
σv = − π ⋅ D ⋅ ∫ τ ⋅ dz ≈
A A
τ(z) τ(z)

A
σr = ? (k0 σv)
oedometric compression R

standard (fixed-ring) oedometer: the specimen alternative: floating-ring oedometer,


moves downward wrt the ring, which generates where compression occurs toward the
friction drag on the sidewall middle from both top and bottom

fixed-ring floating-ring
oedometric compression R

various test procedures


1. discrete loading steps (traditional)
2. CHG (Constant Hydraulic Gradient)
3. CRS (Constant Rate of Strain)
4. CRL (Constant Rate of Loading)

(1) is the simplest and most used procedure


BUT: it causes extremely high hydraulic gradients during consolidation Æ nonuniformities of
effective stress in the specimen
when an increment of stress is applied to the oedometer test specimen there is an instantaneous increase in
pore pressure equal to the stress increment. As drainage occurs, the pore pressures decrease and the effective
stresses increase, until when consolidation (i.e., excess pwp dissipation) is complete – the change in effective
stress is then everywhere equal to the applied total stress increment

0 1min 4min 9min 16min 25min 36min 49min 64min


30s 2min t 90

A
Initial reading P α/2
Calculated zero α/2
primary

Void Ratio e
consolidation
Dial Gauge Reading

σ′vm E

90% Primary
consolidation
χ
1.15χ
Vertical Stress σ′ (log scale)
oedometric compression R

upon loading (b=0, θ=0)


1
σv = σo
Ko

⎛ ⎞⎞
τ ⎛ 1− Ko
τ = σ ⋅ tan⎜ sen −1 ⎜⎜ ⎟⎟

⎝ ⎝ 1+ Ko
⎟⎟
⎠⎠ σv
dσv
effective stress path

σ σh
dσh

3 ⋅ (1 − K o )
q = σv-σh q=
1 + 2K o
p

stress state depends on K0


(K0 const, K0 < 1) effective stress path

earth pressure coefficient at rest p


oedometric compression R

upon unloading (b=0, θ=0 if K0<1)


1
σv = σo
Ko

τ
σv
dσv

σ σh
dσo

3 ⋅ (1 − K o )
q = σv-σh q=
1 + 2K o
p

stress state depends on K0, and K0


depends on OCR

p
non conventional oedometric compression R

so-called "soft oedometer" (Kolymbas & Bauer 1993)

+ a few variations on the theme in Delft (Molenkamp & Co.), Milano (Castellanza),
Grenoble (Flavigny)

the thin ring of the oedometer allows


very small lateral strain to occur so that
radial (horizontal) stress can be derived
from measurement of the circumferential
strain using strain gauges

measurement of both σa and σr

with a sufficient instrumentation, the soft oedometer is a good alternative to the triaxial test

e.g., Castellanza & Nova, Oedometric tests on artificially weathered carbonatic soft rocks, J. Geotech. & Geoenv. Engng ASCE 2004

Æ it was not advisable to use a servo-controlled triaxial apparatus. The acid solution employed to induce natural degradation in
the specimen would have damaged the specimen membrane and jeopardized proper operation of the delicate apparatus and the
instrumentation used for measuring local strain and correcting the confining pressure
R
R
soil, rock, concrete: which is which?
R
constitutive relations / general principles
R

the mechanical response of soils depends on:


• current state
• previous loading history
• load increment size and direction
• time

Æ an appropriate description of soil behavior requires


the stress σ to be a function of the deformation history

history up to time t of the deformation gradient at a point X

is a functional, not a function !

knowledge of strain at time t is not sufficient to uniquely determine the stress state
constitutive relations / general principles
R

differentiable Æ elastic material

Æ non-differentiable functionals should be employed whenever irreversible


(inelastic) behavior is of concern

working with non-differentiable functionals is a mess !

alternative strategy: incremental (rate-type) formulation

rate of deformation
objective stress rate

state variables
is the spin tensor
(stress σ + internal variables)
is the spatial velocity account for the effects of
previous loading history
constitutive relations / general principles
R

constitutive relations in incremental form


+ evolution of internal variables in time

linear kinematics Æ Æ

rate-independence Æ

tangent stiffness at the current state depends on the strain rate


only through its direction
non-linearity and incremental non-linearity
R

initial state (σ0, q 0) Æ strain path ε from ε 0 to ε(t ) Æ stress state at t

the dependence of D on the current stress renders the function non-linear


(e.g., doubling the strain increment does not result in doubling the stress increment)

a (very) particular case

G linear in Æ D does not depend on η

this is called incremental linearity


non-linearity and incremental non-linearity
R

linear behavior Æ incremental linearity


Å

the response of an incrementally linear material remains completely reversible in any


closed loading-unloading program following the same path in two opposite directions
incremental non-linearity = irreversibility
R

we want to represent irreversible behaviors

the reversal of the loading path is associated


with a change in the tangent stiffness

for that, we need incremental non-linearity

in particular
constitutive relations / general principles
R

summing up:

incrementally linear models

(hypoelasticity, variable moduli models)

incrementally non-linear models

- bi-linear models Æ the prototype: classical plasticity

- multi-linear models

- thoroughly non-linear models

see Tamagnini & Viggiani 2002, 2004


plasticity theory for geomaterials
R

• incremental bi-linearity, the most simple form of incremental non-linearity

• in incrementally bi-linear models, there are two distinct tensorial zones (often
referred to as "loading" and "unloading" zones), with largely different stiffness
• the switch between the two zones is typically controlled by an algebraic equation,
the so-called loading/unloading condition

• the foundations of the classical theory of plasticity can be traced back to the
fundamental works of Hill (1950) and Koiter (1960)
• a thorough treatment of this subject can be found, e.g., in the treatises by Lubliner
(1990), Simo & Hughes (1997), Simo (1998). For applications to geomechanics, see
e.g. Desai & Siriwardane (1984), Zienkiewicz et al. (1999)
• applications of concepts of perfect plasticity (i.e., failure criterion and flow rule) to
geomechanics are as early as the works of Coulomb (1773) and Rankine (1853)
referring to the problem of evaluating the earth pressure on retaining structures.
These concepts still form the basis of limit equilibrium and limit analysis methods,
widely used in current practice
plasticity for geomaterials – historical perspective
R

• the application of plasticity as a general constitutive framework for describing


the stress-strain behavior of granular materials was pioneered by Drucker and
others (Drucker & Prager 1952, Drucker et al. 1957)

• a major impact to the application of the theory of plasticity to soil mechanics is


represented by the introduction of “Critical State Soil Mechanics” (isotropic
volumetric hardening), due to Roscoe and coworkers (see, e.g., Schofield &
Wroth 1968, Wood 1987)

• since the late ‘70, important developments have been introduced in the CSSM
framework, motivated by the need to improve the predictive capability of the
theory with respect to:
i) intrinsic and induced anisotropy
ii) cyclic behavior
iii) bonding and destructuration (natural soils)
R
you all remember elasticity… (1/3) R

simple example: uniaxial tensile stress applied to an elastic bar


a ε xx 1
b ε yy
ε xx = σ xx
E
σ xx 2b σ xx
ε yy E : Young’s modulus
=−v
ε xx ν : Poisson’s ratio
2a

stress-strain equations in three dimensions

σ ij = Eijkl ε kl ⇔ σ = E:ε
Eijkl = constants  linear elasticity Eijkl = Eijkl(strain)  non-linear elasticity
you all remember elasticity… (2/3) R

σ z′ 1
ε x = ⎡⎣σ x′ −ν (σ ′y + σ z′ ) ⎤⎦
τ zy E
τ zx τ yz 1
τ xz σ ′y ε y = ⎡⎣σ ′y −ν (σ x′ + σ z′ ) ⎤⎦
E
z
τ xy τ yx 1
y ε z = ⎡⎣σ z′ −ν (σ x′ + σ ′y ) ⎤⎦
σ x′ E
x
τ xy
γ xy =
isotropic elasticity (Hooke’s law) G E
G= =μ
τ xz 2 (1 + ν )
the material is fully specified when γ xz =
values of two parameters are given G
(e.g., E and ν) shear modulus
τ yz
γ yz =
G
you all remember elasticity… (3/3) R

⎡σ x′ ⎤ ⎡1 −ν ν ν 0 0 0 ⎤ ⎡εx ⎤
⎢σ ′ ⎥ ⎢ ν 1 − ν ν 0 0 0 ⎥ ⎢ε ⎥
⎢ ⎥ y ⎢ ⎥⎢ y ⎥
⎢σ z′ ⎥ E ⎢ ν ν 1 −ν 0 0 0 ⎥ ⎢ εz ⎥
⎢ ⎥= ⎢ ⎥⎢ ⎥
τ
⎢ xy ⎥ (1 + ν )(1 − 2ν ) ⎢ 0 0 0 (1 − 2ν ) 2 0 0 ⎥ ⎢γ xy ⎥
⎢τ xz ⎥ ⎢ 0 0 0 0 (1 − 2ν ) 2 0 ⎥ ⎢γ xz ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
τ
⎢⎣ yz ⎥⎦
14444444444444
⎢⎣ 0 0 0 0
4244444444444444
0 (1 − 2ν ) 2
3
⎥⎦ ⎢⎣γ yz ⎥⎦
E

shear stress ÅÆ shear strain isotropic stress ÅÆ volume strain

pure spherical states of stress result only in volume changes (no distortions – shape changes)
pure shear states of stress result only in distortions – shape changes (no volume changes)
let’s start with single-mechanism plasticity
R

Mohr-Coulomb Model
essential ingredients of perfect plasticity
as applied to the modeling of the behavior of geomaterials
contents R

• Motivations for plasticity (elasticity vs. plasticity)


• Basic ingredients of any elastoplastic model
Æ elastic properties (how much recoverable deformation?)
Æ yield surface (is plastic deformation occurring?)
Æ plastic potential (direction of plastic strain increment?)
Æ consistency condition (magnitude of plastic strain increment?)
Æ hardening rule (changes of yield surface?)
• Element tests: (drained) simple shear & triaxial tests
• Limitations Æ other (simple) perfect plasticity models
Elasticity vs. Plasticity (1)
R

In elasticity, there is a one-to-one relationship between stress and


strain. Such a relationship may be linear or non-linear. An essential
feature is that the application and removal of a stress leaves the
material in its original condition
Elasticity vs. Plasticity (2)
R

for elastic materials, the mechanism of deformation depends on the


stress increment

for plastic materials which are yielding, the mechanism of (plastic)


deformation depends on the stress

reversible = elastic irreversible = plastic

when building up an elastic-plastic model,


the first ingredient that we need is a yield surface
(is plastic deformation occurring?)
yield function (1) R
yield function (2) R

The yield surface bounds all elastically attainable states


(a generalized preconsolidation pressure)
yield function (3)
R

Basically:
changes of stress which remain inside the yield surface are
associated with stiff response and recoverable deformations,
whereas on the yield surface a less stiff response is obtained and
irrecoverable deformations are developed

Where do we get this function f ?

The dominant effect leading to irrecoverable changes in particle


arrangement is the stress ratio, or mobilized friction

The mean normal effective stress p ’ is of primary importance.


The range of values of q for stiff elastic response is markedly
dependent on p ’
ÆTresca & Von Mises yield functions are not appropriate
yield function (4)
R

the theory of plasticity was originally developed for metals and later modified for other
materials such as soils and rocks

Von Mises (1913) Tresca (1864)

ÆTresca & Von Mises yield functions are not appropriate


yield function (5)
R

Mohr-Coulomb Drucker-Prager

in both cases: explicit dependence on mean normal stress


be careful with DP!
R

problems with Drucker-Prager criterion

• circular cross-section in the deviatoric plane (unrealistic)


• gives unrealistic friction angles (e.g., φTE = 90° when φTC = 36.9°)

Desrues (2002) - Limitations du choix de


l'angle de frottement pour le critère de
plasticité de Drucker-Prager, Revue
Française de Génie Civil

Bishop already back in 1972 !

repetita iuvant ?
Mohr-Coulomb model, yield function R

To most engineers the phrase "strength of rocks" conjures


up images of Mohr-Coulomb failure criteria

frictional
resistance

independent of
normal stress

Classical notions of Mohr-Coulomb failure can be


reconciled with the patterns of response that we are
modeling here as elasto-plastic behavior
Mohr-Coulomb model, yield function R

σ1 and σ3 : major and minor principal stresses


Mohr-Coulomb model, yield function R

σ1 , σ2 , σ3 : principal stresses
plastic potential (1) R

Summing up:

Plastic strain increment arises if:


1) the stress state is located on the yield surface (f = 0)
AND
2) the stress state remains on the yield surface after a stress increment

knowledge of function f tells us whether plastic strain is occurring or not

But, this is only one part of the story:


We would also like to know direction and magnitude of plastic strain
• will we get plastic volume changes?
• and plastic distortion?
Æ for that, we need another concept (another function: g)
plastic potential (2) R

flow rule
Recall: plastic deformations depend
on the stress state at which yielding
is occurring, rather than on the route
by which that stress is reached

we have now two functions, f and g


Æ the question is: where do we get g ?
associated and non associated flow rules R

it would be clearly a great advantage if, for a given material, yield locus
and plastic potential could be assumed to be the same
f = g Æ only 1 function has to be generated to describe plastic response
also advantageous for FE computations:
• the solution of the equations that emerge in the analyses is faster
• the validity of the numerical predictions can be more easily guaranteed

is f = g a reasonable assumption?

for metals, it turns out that YES, it is


Æ for geomaterials, it is NOT
Where is the problem? ÆThe assumption of normality of plastic strain
vectors to the yield locus would result in much greater plastic volumetric
dilation than actually observed
Mohr-Coulomb model, plastic potential R

dilatancy angle
plastic dilatancy R

how to understand dilatancy


i.e., why do we get volume changes when applying shear stresses?

ϕ = ψ + ϕi
the apparent externally mobilized angle of friction on horizontal planes (ϕ) is larger
than the angle of friction resisting sliding on the inclined planes (ϕi)

strength = friction + dilatancy


• (shear) dilatancy of granular materials was first discussed by Reynolds (1885)
• it is observed in all geomaterials (sand, clay, concrete, rock)
• the dilatancy angle was introduced by Bent Hansen (1958)
consistency condition (our last ingredient)
R

Æ yield surface (is plastic deformation occurring?)


Æ plastic potential (direction of plastic strain increment?)
Æ consistency condition (magnitude of plastic strain increment?)

magnitude of plastic strain increment

during plastic deformation, the stress state must remain on the yield surface

Ö value of i.e.
MC model for element tests R

Δε yy
tan ψ =
Δε xy
MC model for element tests R
MC model for element tests R
MC model – difficulties with a simple model
R
beyond Mohr-Coulomb (1) R

alternative yield surfaces

both the MN and LD criteria fit experimental predicted variations with b-value of
data well shear stress ratio q/p from different
models
in the absence of any 3D data, the MC criterion (WM=Willam-Warnke, AG=Argyris-Gudehus)
is a reasonable choice for 3D models for soils
Æ MC underestimates the experimentally
it must be noted, however, that the MC criterion observed higher friction angles in plane
has vertices which could create difficulties in strain conditions
numerical simulation
beyond Mohr-Coulomb (2) R

intrinsic limitations of Mohr-Coulomb model


• bi-linearity (const. E, ϕ)
• unlimited dilation
• isotropy
• perfect plasticity Å Æ elastic response far from the limit state
•…

Æ more advanced EP models of rock behavior:


hardening plasticity – the elastic domain does change (it can either expand or contract)

double-mechanism plasticity – the expansion or contraction of the elastic domain


depends on both volumetric and deviatoric plastic strains
soil, rock, concrete: which is which?
R

You might also like