Non-Linear Hyperbolic Model & Parameter Selection

NON-LINEAR HYPERBOLIC MODEL & PARAMETER

SELECTION
(Introduction to the Hardening Soil Model)

(following initial development by

Tom Schanz at Bauhaus-Universitt Weimar, Germany)

Computational Geotechnics

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

Contents
Introduction
Stiffness Modulus
Triaxial Data
Plasticity
HS-Cap-Model
Simulation of Oedometer and Triaxial Tests on Loose and
Dense Sands
Summary

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

Introduction
Hardening Soils
Most soils behave in a nonlinear behavior soon after application of shear
stress. Elastic-plastic hardening is a common technique, also used in PLAXIS.
Usage of the Soft Soil model with creep
Creep is usually of greater significance in soft soils.
Hyperbolic stress-strain law for triaxial response curves

Fig. 1: Hyperbolic stress strain response curve of Hardening Soil model

Rf

qf
qa

Eur 3E50
(standard PLAXIS setting Version 7)

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

Stiffness Modulus
Elastic unloading and reloading (Ohde, 1939)
We use the two elastic parameters ur and Eur:

ref

E ur

c cot - 3'

ref
c
cot

+
p

Gur

1
Eur
2(1 ur )

p ref 100kPa

Initial (primary) loading

E50 E50ref

3' c cot
ref

p c cot

ref

3' sin c cos

E50

ref
p sin c cos

Fig. 2: Definition of E50 in a standard drained triaxial experiment

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

Stiffness Modulus
Oedometer tests

ref
Fig. 3: Definition of the normalized oedometric stiffness Eoed

Fig. 4: Values for m from oedometer test versus initial porosity n0

ref
Fig. 5: Normalized oedometer modulus Eoed
versus initial porosity n0

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

Stiffness Modulus
Triaxial tests
Fig. 6: Normalized oedometric stiffness for various soil classes (von Soos, 1991)

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

Stiffness Modulus

Fig. 7: Values for m obtained from triaxial test versus initial porosity n0

Fig. 8: Normalized triaxial modulus E50ref versus initial porosity n0

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

Stiffness Modulus
Summary of data for sand: Vermeer & Schanz (1997)

Fig. 9: Comparison of normalized stiffness moduli from oedometer and triaxial

tests

Eoed E

ref
oed

y'
p ref

E50 E

ref
50

x'
p ref

Engineering practice: mostly data on Eoed

Test data:

ref
Eoed
E50ref

(standard setting PLAXIS version 7)

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

p
1

Triaxial Data on p 2

Fig. 10: Equi-g lines (Tatsuoka, 1972) for dense Toyoura Sand

q
q
21 a
E50 qa q

qa

qf
Rf

E50 E50ref

M ( p c cot ) R f 1

3' sin c cos

ref

p sin c cos

6sin
3 sin

Fig. 11: Yield and failure surfaces for the Hardening Soil model

Course Computational Geotechnics

Non-Linear Hyperbolic Model & Parameter Selection

Plasticity
Yield and hardening functions

p 1p 2p 3p 21p 21 21e
f

qa
q
2q

E50 qa q Eur

qa
q
2q

p 0
E qa q Eur
1 50
4 2 43 {
21

21e

3D extension
In order to extent the model to general 3D states in terms of stress, we use a

% and the mobilized angle of internal

modified expression for q in terms of q
friction m

q% 1' ( 1) 2' 3'

with

3 sin m
3 sin m

f q% M%( p c cot m )

where

6sin m
M%
3 sin m

Course Computational Geotechnics

10

Non-Linear Hyperbolic Model & Parameter Selection

Plasticity
Plastic potential and flow rule

q* 1' ( 1) 2' 3'

with

3 sin m
3 sin m

where

M*

g q* M * ( p c cot m )

6 sin m
3 sin m

p
1

1
2

12 sin

g12 g13 1 1

sin

2
2

12 13 12
13

p
3

p
2

12 sin

0
12 12 sin
1
2

Flow rule

p
v

p
v

sin m sin m

with

sin m

sin m sin cv
1 sin m sin cv

cv p p

Table 1: Primary soil parameters and standard PLAXIS settings

C [kPa]
E50 [Mpa]
[o]
[o]
0
30-40
0-10
40
Eur = 3 E50
Vur = 0.2
Rf = 0.9
m = 0.5
Pref = 100 kPa

Course Computational Geotechnics

11

Non-Linear Hyperbolic Model & Parameter Selection

Plasticity
Hardening soil response in drained triaxial experiments

Fig. 12: Results of drained triaxial loading: stress-strain relations (s3 = 100 kPa)

Fig. 13: Results of drained triaxial loading: axial-volumetric strain relations (s3 =
100 kPa)

Course Computational Geotechnics

12

Non-Linear Hyperbolic Model & Parameter Selection

Plasticity
Undrained hardening soil analysis
Method A: switch to drained
Input:

c ' ; ' ; '

E50ref
0.2; E 3E ; m 0.5; p ref 100kPa
ur
50
ur

Method B: switch to undrained

Input:

cu ; u ; 0

E50ref
0.2; E 3E ; m 0.5; p ref 100kPa
ur
50
ur

Interesting in case you have data on Cu and not no C and

E50 E

3' sin u cu cos u

ref

p sin u cu cos u

ref
50

Eur Eurref

3' sin u cu cos u

ref

p sin u cu cos u

E50ref const .
Eurref const .

Assume E50 0.7 Eu and use graph by Duncan & Buchignani (1976) to estimate
Eu
Eu 1.4 E50

2cu

Fig. 14: Undrained Hardening Soil analysis

Course Computational Geotechnics

13

Non-Linear Hyperbolic Model & Parameter Selection

Plasticity
Hardening soil response in undrained triaxial tests

Fig. 15: Results of undrained triaxial loading: stress-strain relations (s3 = 100 kPa)

Fig. 16: Results of undrained triaxial loading: p-q diagram (s3 = 100 kPa)

Course Computational Geotechnics

14

Non-Linear Hyperbolic Model & Parameter Selection

HS-Cap-Model
Cap yield surface
2
q%
f c 2 p 2 pc2
M

Flow rule

gc f c

(Associated flow)

Hardening law
For isotropic compression we assume

p
v

p
p
1

p
Kc K s H
With

Kc
Ks
K s Kc

For isotropic compression we have q = 0 and it follows from p pc

gc
pc H H c
2H c p
p

p
v

For the determination of, we use another consistency condition:

T

f c f c
fc

pc 0

pc

Course Computational Geotechnics

15

Non-Linear Hyperbolic Model & Parameter Selection

HS-Cap-Model
Additional parameters
The extra input parameters are K 0 ( 1 sin ) and Eoed / E50 ( 1.0)
The two auxiliary material parameter M and Kc/Ks are determined iteratively from
the simulation of an oedometer test. There are no direct input parameters. The
user should not be too concerned about these parameters.
Graphical presentation of HS-Cap-Model
I: Purely elastic response
II: Purely frictional hardening with f
III: Material failure according to Mohr-Coulomb
IV: Mohr-Coulomb and cap fc
V: Combined frictional hardening f and cap fc
VI: Purely cap hardening with fc
VII: Isotropic compression
1

Fig. 17: Yield surfaces of the extended HS model in p-q-space (left) and in the
deviatoric plane (right)

Course Computational Geotechnics

16

Non-Linear Hyperbolic Model & Parameter Selection

HS-Cap-Model
1 = 2 = 3

Fig. 18: Yield surfaces of the extended HS model in principal stress space

Course Computational Geotechnics

17

Non-Linear Hyperbolic Model & Parameter Selection

Simulation of Oedometer and Triaxial Tests on Loose and

Dense Sands

Fig. 19: Comparison of calculated () and measured triaxial tests on loose Hostun
Sand

Fig. 20: Comparison of calculated () and measured oedometer tests on loose

Hostun Sand

Course Computational Geotechnics

18

Non-Linear Hyperbolic Model & Parameter Selection

Simulation of Oedometer and Triaxial Tests on Loose and

Dense Sands

Fig. 21: Comparison of calculated () and measured triaxial tests on dense Hostun
Sand

Fig. 22: Comparison of calculated () and measured oedometer tests on dense

Hostun Sand

Course Computational Geotechnics

19

Non-Linear Hyperbolic Model & Parameter Selection

Summary
Main characteristics
Pressure dependent stiffness
Isotropic shear hardening
Ultimate Mohr-Coulomb failure condition
Non-associated plastic flow
Additional cap hardening

HS-model versus MC-model

c, ,

As in Mohr-Coulomb model

E50ref
ur

Normalized primary loading stiffness

Eurref

Normalized unloading / reloading stiffness

Unloading / reloading Poissons ratio

Power in stiffness laws

Rf

Failure ratio

Course Computational Geotechnics

20

Non-Linear Hyperbolic Model & Parameter Selection

Exercise 1: Calibration of the HS-Cap-Model for Loose

and Dense Sand
Oedometer and triaxial shear experimental data for both loose and dense sands are
given in Figs. 23 26.

loose
dense

vur
0.25
0.25

Table 2: Parameters for loose and dense sand

ref
Eoed
/ E50ref
Eurref / E50ref
m

0.65
34o
0o
1.0
3.0
o
o
0.65
41
14
0.9
3.0

E50ref

16
35 MPa

Proceed according to the following steps:

Use Ko = 1 sin and Eoed/E50 according to Table 2 in the advanced material
parameter input in PLAXIS.
For both simulations use an axis-symmetric mesh (1x 1 [m]) with a coarse
element density. Change loading and boundary conditions according to the
test conditions.
Simulation of oedmoter tests with unloading for unloading for maximum axial
stress.
Loose sand: 1max 200kPa
Dense sand: 1max 400kPa
If necessary improve given material parameters to obtain a more realistic
response.
Check triaxial tests with the parameters obtained from the oedometer
simulation.

Course Computational Geotechnics

21

Non-Linear Hyperbolic Model & Parameter Selection

Exercise 1: Calibration of the HS-Cap-Model for Loose

and Dense Sand
Results for loose sand

Fig. 23: Triaxial tests on loose Hostun Sand

Fig. 24: Oedometer tests on loose Hostun Sand

Course Computational Geotechnics

22

Non-Linear Hyperbolic Model & Parameter Selection

Exercise 1: Calibration of the HS-Cap-Model for Loose

and Dense Sand
Results for dense sand

Fig. 25: Triaxial tests on dense Hostun Sand

Fig. 26: Oedometer tests on dense Hostun Sand

Course Computational Geotechnics

23