Académique Documents
Professionnel Documents
Culture Documents
constitutive models: an
overview
Dr. Alejo O. Sfriso
Professor, Soil Mech and Geology, University of
Buenos Aires
Principal Geotechnical Engineer, SRK Consulting
(Argentina)
asfriso@srk.com.ar
Table of contents
Introduction
Implications of the ingredients of a
constitutive model
Kinematics
Stress-strain equation
Yield function and strength
Flow rule
Hardening and softening
Conclusions
1352015 @ ISRM
Montreal
Introduction
Numerical modelling routine for predicting 2D
failure, now used for 3D problems and
serviceability estimates
For 3D problems, constitutive models
themselves (not just the parameters) have
big impact on the outcome
In this presentation we will overview some
challenges that occassional practitioners may
encounter when performing numerical
modelling
1352015 @ ISRM
for more info: asfriso@srk.com.ar
Montreal
We will focus in the implications of
Definition of a constitutive
model
Output
updated stress and state vars
1352015 @ ISRM
Montreal
p e
1352015 @ ISRM
Montreal
e p
E e
1352015 @ ISRM
Montreal
p
F 0 0
1352015 @ ISRM
Montreal
Small-strain kinematics
Stress-strain equation
Yield function
Flow rule:
F 0p0
evolution of plastic
strain and plastic
work dissipation
e p
E e
1352015 @ ISRM
Montreal
Small-strain kinematics
Stress-strain equation
Yield function
Flow rule
F 0p0
Hardening and
softening
e p
E e
1352015 @ ISRM
Montreal
Small-strain kinematics
Stress-strain equation
Yield function
c
Flow rule
Hardening and
E
softening
The definition of all these elements (and
their numerical implemmentation) will p e
v
1352015 @ ISRM
for more info: asfriso@srk.com.ar
Montreal
Table of contents
Introduction
Implications of the ingredients of a
constitutive model
Kinematics
Stress-strain equation
Yield function and strength
Flow rule
Hardening and softening
Conclusions
1352015 @ ISRM
Montreal
P A
Sample height before/after
Failure: difference
confining pressure applied
readily apparent
Stress on deformed
Intermediate strain (hardening
d Psample
A0
0
D 7012 (rocks)
D 4767 (soils)
1352015 @ ISRM
Montreal
d P A
A
1 v
A0
1 a
P A
Failure: difference
readily apparent
Stress-strain equation
Linear elasticity is universal in rock
engineering
When incorporating structures into the rockmass, elastic nonlinearity might improve
predictive capability
Nonlinear elasticity available
d
in codes or can be incorporatedIncreasing
d
stiffness due to
(e.g. fish function or .dll)
gap closure
Suggestion: Address the nonlinearity
of elastic response by
1352015 @ ISRM
for more info: asfriso@srk.com.ar
Montreal
(Chen 1990)
using a proper elastic formulation
Stress-strain equation
d
d
1352015 @ ISRM
Montreal
Phase2
1352015 @ ISRM
Montreal
UDEC
for more info: asfriso@srk.com.ar
Pressure
dependent
stiffness
1352015 @ ISRM
Montreal
(Chen 1990)
1352015 @ ISRM
Montreal
1352015 @ ISRM
Montreal
1352015 @ ISRM
Montreal
1352015 @ ISRM
Montreal
1352015 @ ISRM
Montreal
1352015 @ ISRM
for more info: asfriso@srk.com.ar
Montreal
(Bonnier 2015)
(Waterman
Flow rule
In practical terms, flow rule controls plastic
dilatancy
Most codes use the deBorst-Vermeer equation
for non-associativity of the Mohr-Coulomb
model
F 1 3 1 3 sin 2ccos 0
( replaced by )
G 1 3 1 3 sin 2ccos 0
1352015 @ ISRM
Montreal
Flow rule
For 3D applications, this formulation deserves
a review
Associativity: too much plastic volumetric
strain
vp
1352015 @ ISRM
Montreal
Flow rule
For 3D applications, this formulation deserves
a review
Associativity: too much plastic volumetric
strain
Volumetric
non-associativity: correct plastic
vp
qvol strain
1352015 @ ISRM
Montreal
Flow rule
For 3D applications, this formulation deserves
a review
Associativity: too much plastic volumetric
strain
Volumetric
non-associativity: correct plastic
vp
qvol strain
de Borst equation: different 2 predicted
1352015 @ ISRM
Montreal
Flow rule
Experimental evidence is scarce, not
necessarily supporting non-associativity in
the deviatoric plane
Difference more significative for J3 models
Experimental
1352015 @ ISRM
Montreal
HoekBrown
Flow rule
Using deviatoric associativity,
the Mohr-Coulomb model
has attractors in the corners
Experimental
1352015 @ ISRM
Montreal
Flow rule
Suggestion: When specifying dilatancy in 3D,
check various stress-paths using simple
numerical tests
Run a version with no dilatancy and check
differences
If not reasonable, check
if your combination of
yield function and flow
rule allows for the
development of realistic MatsuokaHoek1352015 @ ISRM
for more info: asfriso@srk.com.ar
Nakai
Brown
Montreal
plastic deformations
1352015 @ ISRM
Montreal
Displacements, PSC
simulation
for more info: asfriso@srk.com.ar
(Sfriso 2010)
(Sfriso 2010)
You must
consider the
contribution of
strains of all
these
points
You
refine:
(coarse)
more
points
1352015 @ ISRM
Montreal
must be
included
(Sfriso 2010)
1352015 @ ISRM
Montreal
This slope is
not a material
property but
proportional
(Sfriso 2010)
# of
elements
FoS
41
4
871
1.4
3
1352015 @ ISRM
Montreal
11
(Waterman 2010)
2D Mohr-Coulomb:
FoS=1.05
3D Mohr-Coulomb:
FoS=1.20
Table of contents
Introduction
Implications of the ingredients of a
constitutive model
Kinematics
Stress-strain equation
Yield function and strength
Flow rule
Hardening and softening
Conclusions
1352015 @ ISRM
Montreal
Conclusions
Numerical modelling addresses complex
geometries and nonlinear behaviour of rock
masses
Using popular numerical tools at full capacity
probable intermediate between routine 2D
models and involved discontinuum
mechanics formulations
Still, going 3D and into hardening plasticity
may pose several challenges that deserve
careful meditation
1352015 @ ISRM
for more info: asfriso@srk.com.ar
Montreal
Implications of selecting adequate models for
References
Most important
Potts et al (2002). Guidelines for the
use of advanced numerical analyses.
Thomas Telford, 177 p.
Four papers by Feng & Hudson @
IJRMMS (2004, 2007, 2010, 2010)
Please find the rest of the references in the
paper
1352015
@ ISRM
Acknoweldgment:
my colleagues at SRK
for moreto
info: asfriso@srk.com.ar
Montreal
END