Professional Documents
Culture Documents
Hardening Soil Model PDF
Hardening Soil Model PDF
τ
γ
G τmax
1 elastic
γ σ‘
• Associated flow (φ’=ψ’) – simple, numerically
efficient, but too much dilation
• Non-associated flow (φ’ ≠ ψ’) – non-symmetric
stiffness matrix not numerically efficient, still far
from real soil behaviour
MC vs Real Soil Behaviour
Stress path in oedometer
loading-unloading
Critical State Models
q, τ
τ
yield surface
elastic
γ p′′, σ‘
Stress-Strain Curve
γ
Model characteristics:
• Hyperbolic stress-strain relationship in axial compression
• Plastic strain in mobilizing friction (shear hardening)
• Plastic strain in primary compression (volumetric hardening)
• Stress-dependent stiffness according to a power law
• Elastic unloading/ reloading compared to virgin loading
• Memory of pre-consolidation stress
• Mohr-Coulomb (MC) failure criterion
• Dilatancy below MC line
• Small strain stiffness (HS-small only)
Hyperbolic stress-strain relationship
Duncan-Chang or hyperbolic model in (tri)axial loading:
q
q asymptote
For q < qf: ε1 = ε 50 q
qa − q a
q f
2sin ϕ 1
where qf = (σ 3' + c cot ϕ ) E
1 − sin ϕ 50
qf ½ qf E
ur
and qa = ≥ qf
Rf
1
Rf = ‘failure ratio’ (standard value: 0.9) ε1
The hyperbolic model with Mohr-Coulomb failure criterion constitutes the basis for
the HS and HS-small models. In contrast to the Duncan-Chang model, the HS
models are elasto-plastic models.
Shear hardening
s qa q 2q
Yield function (cone): f = − − γ ps
E50 qa − q Eur
where γps is a state parameter that is tracking the opening of the cone:
Elastic
q MC failure line
q
plastic
Increasing
plastic
shear
strains
-a = c cot ϕ p’
γ
Evolution law for γps: d γ ps = d λ s where dλs is the plastic multiplier for the
cone type yield surface of the model.
Volumetric (density) hardening
q% 2
Yield function (cap): f = c
− p '2 − p p 2
α2
q
αpp’
K0’
cap
p’
pp’
σ1’
σc’
Volumetric or
vertical strain,
ε1
Volumetric (density) hardening
αpp
Therefore: q% = f (σ 1 , σ 2 , σ 3 , ϕ )
m
K s Kc σ 3 + a p
Evolution law: dp p = d ε v
K s − K c p ref + a
Eurref
where K s = and the bulk stiffness of the cap Kc is determined
3(1 − 2ν )
from Eoed and K0nc
Stress dependent moduli
q asymptote: qa
E50
qf=0,9 qa
1
m
σ '+ a
E50 = E50ref 3
pa '+ a
(Secant modulus)
εy
σc’ σy’
m
ref σ '3 / K +a
nc
Eoed 0
εy 1 Eoed = Eoed
p a '+ a
(Tangent modulus)
Eur >E50
1
σ 3 '+a
εy Eur = Eurref ( )m
pa '+ a
σc’ σy’
ν ur = low value
Eoed
εy 1
Preconsolidation is entered by
d d
e e OCR or POP relative to initial
p p vertical stress and is then
t σc’ t
converted to pp.
h h
σy0’ σy0’ σc’
σ’yy
Prestress
σ’c Initial horizontal stress:
Initial
CAP
POP 1 σ x 0 ' = K 0 '⋅σ c '−(σ c '−σ y 0 ' ) ⋅ 1ν−ν
ur
ν ur ur
σ’x0 σ’xx
Initial conditions for the HS model
Initial stresses:
q
MC failure line
Output:
pc
K0nc line
' OCR ' = OCRiso = eq
p
α pc Cap p eq = ( p' )
2
+ q2 /α 2
p’0, q0
peq0 pc,0 p
Dilatancy
~ e max
2sin ψ ψ
1-sin ψ
εy
εv
sin ϕ '− sin ψ
sin ϕ cv =
1 − sin ϕ ' sin ψ
Dilatancy formulation:
Rowe (1962) modified
σ '1 −σ '3
sin ϕ m =
σ '1 +σ '3 −2c' cot ϕ '
sin ϕ m − sin ϕ cv
sin ψ m =
1 − sin ϕ m sin ϕ cv
Dilatancy from cone (most important part):
q
Associated cap flow:
Increasing contractancy from
zero to a maximum value at
MC line, but only when cap
moves!
-a pp’ p’
HS input parameters
Parameter Description
E50ref Stiffness modulus for primary loading in drained triaxial test
Monotonic
loading
Example: Vertical displacement of
a retaining wall
Wall pulled up
Limitation: No small strain stiffness
1 Retaining walls
odulus G/G0 [-]
Foundations
Tunnels
Very
small
Shear mo
Larger strains
0 Shear strain γ[-]
-6 -5 -4 -3 -2 -1
10 10 10 10 10 10
Dynamic methods
Local gauges
Limitation: Unloading/reloading
stiffness is the same
q=σ1-σ3 [kPa]
ε1 [%]
• Truly elastic behaviour on for very small loops
• At small strains stiffness increases
• Hysteresis increases with increasing strains
Limitation: Non-monotonic loading in heavily OC clays
Need to use artificially low POP/OCR value to trigger plasticity within ‘yield
surface’ in order to represent different stiffness for loading/unloading for non-
monotonic loading.
However the stress path may still be wrong when approaching to failure.
OC clay
HSmodel
p’
Recommended procedure for
application
800
600
400
200
q' [kN/m2]
300
M-C
250
H-S
200
150
100
50
eps-v
9.00E-03
6.00E-03
3.00E-03 M-C
0.000
H-S
-3.00E-03
-6.00E-03
0 0.02 0.04 0.06 0.08
eps-1
Illustration by Brinkgreve, R.B.J.
Comparison HS models and MC model
sig'-yy [kN/m2]
1000
800
600
M-C
400
H-S
200
HS MC
ε1 [%]
• Truly elastic behaviour on for very small loops
• At small strains stiffness increases
• Hysteresis increases with increasing strains
Soil Stiffness at Small Strains
q=σ1-σ3 [kPa]
ε1 [%]
Small-strain stiffness or Gur vs. G0
Soil stiffness derived from laboratory testing is commonly plotted as (secant-) shear
modulus G over shear strain γ. G = G(γ) is a function of the applied shear strain after the
last load reversal.
E0 = 2 (1 + νur) G0
Gγ = ∆q/∆γ
q = σ1-σ3 G0
Gur
∆q
γ = ε1-ε3 ∆γ γ = ε1-ε3
Analysis results are also less sensitive to the choice of proper boundary conditions. Large
meshes no longer cause extensive accumulation of displacements, because marginally
strained mesh parts are very stiff.
Experimental evidence and data for small-strain stiffness
True elastic stiffness was first observed in soil dynamics. Back then, the apparent higher
soil stiffness in dynamic loading applications was attributed to the nature of loading, e.g.
inertia forces and strain rate effects. Nowadays, static small-strain measurements are
available as well. These show only little differences to dynamic measurements. Still, the
term dynamic soil stiffness is sometimes used when true elastic or small-strain stiffness is
meant.
Soil Stiffness at Small Strains
1 Retaining walls
odulus G/G0 [-]
Foundations
Tunnels
Very
small
Shear mo
Larger strains
0 Shear strain γ[-]
-6 -5 -4 -3 -2 -1
10 10 10 10 10 10
Dynamic methods
Local gauges
Experimental evidence and data for small-strain stiffness
Empirical relationships between G0 or E0 =2(1+νur)G0 and void ratio e:
Clay
p' 50
E0 = Eref
0
pref
140 MPa 0
ref
with E 0 = 0.5 1 1.5 2 2.5
e Void ratio e [-]
* J. Biarez, P.-Y. Hicher. Elementary Mechanics of Soil Behaviour. Balkema, 1994
Experimental data & empirical relationships (E0)
The relationship between E0 and Eur can be estimated from the chart by Alpan* assuming
Edynamic/Estatic ≅ E0/Eur (10 kg/m²=1 MPa):
E dynamic E
≈ 0
E static Eur
* H.B. Seed, I.M. Idriss, Soil moduli and damping factors for dynamic response analysis. Report 70-10,
EERC (Berkeley, Cali-fornia), 1970.
** M. Vucetic, R. Dobry, Effect of soil plasticity on cyclic response, Journal of Geotechnical Engineering,
ASCE 117 (1991), no. 1, 89-107.
Empirical relationship for γ0.7
Based on statistical evaluation of test data, Darandeli* proposed correlations for a
hyperbolic stiffness reduction model, similar to the one used inside the HSS model.
Correlations are given for different plasticity indices.
Based on Darendeli‘s work, γ0.7 can be estimated to:
p'
IP = 0: γ 0.7 = 0.00015
pref
p'
IP = 30: γ 0.7 = 0.00026
pref
p'
IP = 100: γ 0.7 = 0.00055
pref
Note: The indicated stress dependency of γ0.7 is not implemented in the commercial HSS
model. If needed, the stress dependency of γ0.7 can be incorporated into boundary value
problems through definition of sub-layers.
*Darendeli,Mehmet Baris, Development of a New Family of Normalized Modulus Reduction and Material
Damping Curves. PhD Dissertation (supervisor: Prof. Kenneth H. Stokoe, II), Department of Civil
Engineering. The University of Texas at Austin. August, 2001.
HS-Small model
1.0
0.8
0.6
G/G0 [-]
0.4
0.2
0.0
-3 -2 -1 0 1 2 3
1e 1e 1e 1e 1e 1e 1e
Normalisierte Scherdehnung γ/γ0.7 [-]
Small-strain stiffness in the HS
model (HSsmall)
Strain(path)-dependent elastic overlay model:
G0
Gs =
1 + 0.385 γ / γ 0.7
G0
Gt = ≥ Gur
(1 + 0.385 γ / γ 0.7 )2
G starts again at G0
Gur after full strain reversal
Small-strain stiffness in the HS
model (HSsmall)
τ
Gt Cyclic loading
G0 leads to Hysteresis
Gs
G0
G0
CiTG, Geo-engineering,
http://geo.citg.tudelft.nl
Small-strain stiffness in the HS
model (HSsmall)
G0 γ0.7
Gt Gs
Gur
HS-small extension 1-dimensional
The 1-dimensional model by Hardin & Drnevich*:
τ
G0 Hardin & Drnevich:
τf 1 G 1
=
G0 1+ γ / γ r
Modified HS-Small:
G 1
γ =
G0 1 + (3γ ) /(7γ 0.7 )
* B.O. Hardin, V.P. Drnevich, Shear modulus and damping in soils: Design equations and curves,
ASCE Journal of the Soil Mechanics and Foundations Division 98 (1972), no. SM7, 667-692.
HS-small model – stiffness reduction
G0ref
40000 40000
30000 30000
20000 20000
Gurref
10000 10000
HS-Small
Hardin & Drnevich
0 0
1E-5 0.0001 0.001 0.01 1E-5 0.0001 0.001 0.01
Shear strain γ [-] Shear strain γ [-]
Difference HS and HS-small
Model response in a standard triaxial test.
Here: Dense Hostun sand
-0.04 40000
0 0.00 0
0.00 0.02 0.04 0.06 0.08 ε1[-] 0.0001 0.001 0.01 ε1-ε3[-]
-0.004
-0.008
-0.012 MC (E50)
-0.016 MC (Eur)
Settlement [m]
-0.016
HSS
Settlement [m]
Excavation example
Limburg excavation: Horizontal wall displacement
MC-Model (E50) MC Model (Eur) Hardening Soil Model Hardening Soil Small
-0.04 -0.03 -0.02 -0.01 0 -0.04 -0.03 -0.02 -0.01 0 -0.04 -0.03 -0.02 -0.01 0 -0.04 -0.03 -0.02 -0.01 0
0 0 0 0
-5 -5 -5 -5
MC-Model (E50) MC Model (Eur) Hardening Soil Model Hardening Soil Small
-600 -400 -200 0 200 400 -600 -400 -200 0 200 400 -600 -400 -200 0 200 400 -600 -400 -200 0 200 400
0 0 0 0
-5 -5 -5 -5
-0.01
Measurement
HS (original)
-0.02
HS-Small
Settlement [m]
HS-Small model
1.0
0.8
0.6
G/G0 [-]
0.4
0.2
0.0
-3 -2 -1 0 1 2 3
1e 1e 1e 1e 1e 1e 1e
Normalisierte Scherdehnung γ/γ0.7 [-]
HS model (HSsmall)
Relevance of small-strain stiffness:
• Very stiff behaviour at very small strains (vibrations)
• Reduction of stiffness with increasing strain; restart after
load reversal
• Hysteresis in cyclic loading:
– Energy dissipation
– Damping
0
3.0
20
40
G m [-]
2.0 Gm=Gref /Gurref
60
80
1.0
100
The dark (blue) area is the strain area where G = Gur. The light gray
(yellow) area is the very-small-strain area with G ≈ G0. In between Gur
and G0 is the area where shear strains are small but not very small
according to the definition by Atkinson.
Limitation: Heavily OC clays
Need to use artificially low POP/OCR value to trigger plasticity within ‘yield
surface’ in order to represent different stiffness for loading/unloading for non-
monotonic loading.
However the stress path may still be wrong when approaching to failure.
200
180
160
140
120
q (kPa)
100
60
HSsmall Model Prediction
40
20
0
0 50 100 150 200 250
P` (kPa)
Limitation: Heavily OC clays
By default, the initial stiffness is set to G0.
Care needs to be taken when the geologic loading history of a soil is modeled.
If, for example, a vertical surcharge was applied and removed in order to model
OCR, the model remembers the vertical heave upon unloading including its
decreased small-strain stiffness. The initial stiffness at the onset of loading the
footing might then look as the one shown at the left-hand side of the above
figure. Here, the material should be exchanged or a reverse load step applied.
Selection of parameters for the
Hardening Soil model
Based on materials by
Dennis Waterman
Plaxis bv
Prof. Steinar Nordal - NTNU Norwegian University of
Science and Technology
Nubia González – UPC Universidad Politécnica de Cataluña
Input parameters - Hardening Soil
model
Parameter Description
E50ref Reference modulus for primary loading in drained triaxial test
Eoedref Reference modulus for primary loading in oedometer test
Eurref Reference modulus for unloading/reloading in drained triaxial test
m Modulus exponent for stress dependency
νur = 0.2 Poisson’s ratio for loading/unloading
c’ Effective cohesion at failure
φ’ Effective friction angle at failure
ψ Dilatancy angle at failure
POP: (σp – σvo’) Initial preconsolidation stress, σp
OCR: σp/σvo’
Ko Earth pressure coefficients at rest
Konc = 1 – sinφ’
G0ref Reference shear stiffness at very small strain levels, (HSsmall only)
γ0.7 Shear strain at which G has reduced to 70% of G0ref , (HSsmall only)
Input parameters - HS model
• Important:
– The Hardening Soil is completely defined in
effective stresses and therefore needs both
effective stiffness parameters and effective
strength parameters
– A total stress analysis may be performed with
the Hardening Soil model using both
undrained strength (φ=0 and c=cu) and
undrained stiffnesses but with the following
limitations:
– No stress dependent stiffness, only constant stiffness
– No compression hardening, only elastic compression
Stiffness parameters - Hardening Soil
model
Stress dependent stiffness for primary shear, primary compression and
unloading/reloading behaviour
m
Plastic Cone hardening σ 3 '+ a
E50 = E50ref a = ccot(ϕ)
secant modulus: p + a
ref
m
Plastic Cap hardening, ref
σ '1 + a
tangent modulus: Eoed = Eoed
p
ref + a
m
Elastic, unloading, reloading σ 3 '+ a
Eur = Eurref
tangent modulus: p + a
ref
Parameters of the HS model
Parameters:
q
σ3=pref σc σ1=pref σ1
qult
(ϕ, c)
E50ref qf=Rf qult
Eurref
0.5 qf 1
Eoedref
ε1 εv
Triaxial test Oedometer test
Stiffness of Sand
Parameters of the HS model
Schanz (1998)
Stiffness of sand, drained
triaxial testing:
σy - σx σy - σx σy - σx E
50
E50
E50
εy εy εy
∆σy’
E50 σ 'x
E50 = E50ref
ref
50
pref
E
Loose sands: E50ref ≈ 15 MPa
pref = 100kPa σx’
Dense sands: E50ref ≈ 50 MPa
Stiffness of sand, K0 test:
σy’
Eoed
εy
ref σ 'y
E oed = E oed
p ref
ref
Loose sands: Eoed ≈ 15 MPa
Oedometer results confirm: ref
Dense sands: Eoed ≈ 50 MPa
Stiffness of sand
ref ref
E ≈E
50 oed
ref
Eoed ≈ RD • 60MPa Correlation by Lengkeek
for pref=100 kPa
emax − e
RD =
emax − emin
Stiffness of sand
E50ref ≈ Eoed
ref
How can this be true?
p ref
-σ’1= p ref
-σ’3 = p ref
-σ’3 -σ’1
Eoed Cone tip Soil type
at σ’v = σ’v0 Stiffness of sand
resistance
Eoed = 4qc qc < 10 MPa Loose sands:
Eoed = 2qc+20 10 MPa < qc < 50 Unaged and
MPa MPa uncemented,
Eoed = 120 MPa qc > 50 MPa predominantly
silica
Eoed = 5qc qc < 50 MPa Dense sands
Eoed = 250 MPa qc > 50 MPa
Triaxial tests:
Unloading is purely
elastic in HS model
Stiffness for unloading-reloading loops
From oedometer tests for elastic behaviour with low Poisson’s Ratio:
(1−νur )
Eur,oed = Eur ≈1.1⋅ Eur
(1−2νur )(1+νur )
Triaxial test
Alternatively:
dense 40
D e v ia t o r ic s t r e s s , q [ k P a ]
200
150
dense 40
100
50
pref + a
E50ref = E50
0 σ 'x + a
0 1 2 3 4 5
100kPa
Axial strain [%] = 20000kPa = 32MPa
40kPa
EXAMPLE
Triaxial test results by Shaoli (2004)
Dense Hokksund sand at 40 kPa,
n = 35,9% (initial) – 39,6% (end of test)
Dense 40
-3 1-sin ψ = = 1,2
-2
2sin ψ Dense 40 2 sinψ 4,2
-1
0 1 2 3 4 5 sinψ = 0,29
0
1 ψ = 17°
EXAMPLE
Oedometer test dense Hokksund sand, n = 39% , (Moen, 1975)
0
Loading:
-0,2
Test data
Vertical strain [%]
σ 3 '+ a
Unloading: Eur oed ≈ 1.1⋅ Eur , Eur = Eurref
Low Poisson Ratio pa '+ a
pa '+ a 850kPa 100
Eurref ≈ 0.9 ⋅ Eur oed = 0.9 ⋅ = 195MPa
σ 3 '+ a 0,0028 200
HS Material parameters
for dense Hokksund sand from fitting PLAXIS
results to experimental data
(after trial and error, starting with estimated parameters):
200
180
Deviatoric stress, q [kPa]
160
140
120
Plaxis 40
100 dense 40
80
60
40
20
0
0,00 1,00 2,00 3,00 4,00 5,00
Axial strain [%]
Triaxial test results and PLAXIS
simulation
Dense Hokksund sand at 40 kPa,
n = 35,9% (initial) – 39,6% (end of test
-3
-2,5
-2 from PLAXIS 40
Dense 40
-1,5
-1
-0,50,00 1,00 2,00 3,00 4,00 5,00
0
0,5
1
Oedometer test and PLAXIS simulation
dense Hokksund sand, n = 39% , (Tore Ingar Moen, 1975)
0
Test data
-0,2
Vertical strain [%]
-0,4 Plaxis
-0,6
-0,8
-1
-1,2
-1,4
0 200 400 600 800 1000 1200
Vertical effective stress [kPa]
Stiffness of Clay
Stiffness of clay
Drained stiffness from oedometer tests
σc’ σy’ Basically
ref
σ 'y
Eoed = Eoed
p
ref
εy 1
ref 1
Eoed E oed = for pref = σ y '
σy’ mv
σy - σx E50 = f ⋅ Eu50
cu: undrained 1+ ν
f = CUR 195 (2000)
Eu50 shear strength
1+ νu
1 cu f = 1/ 3 COB (1997)
0.25 < f < 0.35 Lambe & Whitman (1969)
εy
Stiffness for unloading-reloading loops
From oedometer tests for elastic behaviour with low Poisson’s Ratio:
Clay (m=1):
NC : Eur ,oed ≈ 10Eoed ⇒ Eurref ≈ 10Eoed
ref ref
/ K0 ≈ 20Eoed
OC : Eur ,oed ≈ 3Eoed ⇒ Eurref ≈ 3Eoed
ref ref
/ K0 ≈ 3Eoed
Stiffness of clay
Other estimations for stiffness of normally consolidated clays (m=1):
ref 1 ref
E ≈ E
oed 2 50 Order of magnitude (very rough)
Oedometer modulus
1 + e0
Drained stiffness from ref
Eoed = 230 ⋅ (
Cc
)
3
oedometer tests on
intact lean Norwegian
2
NC clays for σy’ > σc’
1
Janbu :
m 103
σ′
Eoed = Eoed ⋅
ref
sandy gravel
pref
sand
more general: 10
m
σ′ + a Norwegian
Eoed = Eref ⋅
oed
clays
pref + a 1
ref 1 ref
Eoed ≈ 2 E50 Order of magnitude (very rough)
GW
σyy σxx High “oedometer” cut out as
a vertical column from the
site in question.
10 m γ = 20kN/m3 Start the “test” from in situ
stresses and specified
preconsolidation for the
Sample sample studied.
pw σ’yy pw σ’xx
5m
Soft clay by
ref
E 50 = 2 MPa m=1 ϕ = 25o
Hardening soil ref
E oed = 2 MPa νur = 0,2 Ψ = 0o
model: c = 5 kPa K0NC = 0.577
E urref = 10 MPa
Oedometer test results, soft clay
Strain
0,00
-0,10
0 -100 -200 -300 -400 -500 -600 σ’yy
Stress [kN/m2]
paths:
160 160
120 120
80 80
Initial cap
40 40
OCR = 1 OCR = 2
0 0
0 -40 -80 -120 -160 0 -50 -100 -150 -200 -250
180
160
q Top curve OCR = 3
140
OCR = 2
Effect of 120
OCR = 1
100
OCR: 80
60
40
20
εy
0
E50
α pc
Eur E oed
β p´
double hardening E50 / Eoed > 2 difficult to input
Small-strain stiffness in the HS model
(HSsmall)
Strain (path)-dependent elastic overlay model:
G starts again at G0
after full strain reversal
160
140
120
HSsmall E0
ref
E
q [kN/m²]
100 HS
G0ref = 0
80 2(1+νur )
60
40
20
Et ≥ Eur
0 0 -0.002 -0.004 -0.006 -0.008 -0.01 -0.012 -0.014
εyy
Small-strain stiffness parameters in the
HSsmall model
Order of magnitude:
ref Eurref
G0ref = (2.5 to10)Gurref where G
ur =
2(1 + ν ur )
γ 0.7 = (1 to 2) ⋅10 −4
Recommended procedure for
application