723
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
1
Combined computational-experimental Laboratory Testing for Soil Behavior
Modeling
Combinaison d
’
essais numériques et expérimentaux pour la modélisation du comportement des
sols
.
Y. M.A. Hashash
Dept. of Civil and Environmental Engineering, Univ. of Illinois at Urbana-Champaign, Urbana, U.S.A
R. Asmar & S. Moon
Dept. of Civil and Environmental Engineering, Univ. of Illinois at Urbana-Champaign, Urbana, U.S.A
ABSTRACT: Solving complex boundary value problems in geotechnical engineering requires a soil constitutive model that reliably
captures soil behavior under general loading conditions. Laboratory testing has greatly contributed to the development of constitutive
models that reflect soil nonlinear and anisotropic behavior. Available laboratory tests are interpreted assuming uniform stress and
strain states within a tested specimen and therefore provide information on material behavior within a narrow range of stress
–
strain
paths and do not cover general loading conditions which occur in field problems. This paper presents the integration of self-learning
simulations (SelfSim) inverse analysis framework with laboratory testing to extract soil-behavior. Application of this framework to
Direct Simple Shear (DSS) tests shows that it is possible to characterize soil behavior over a wide range of stress paths from a single
test. The paper also describes the development of a modified triaxial testing device intended to impose non-uniform loading
conditions to increase the range of stresses and strains that can be extracted via SelfSim. The new device represents an important step
towards a tighter integration between laboratory testing and constitutive model development.
RÉSUMÉ :
Résoudre des problèmes complexes aux limites en géotechnique nécessite un modèle constitutif de sol qui capte de
manière fiable le comportement du sol dans des conditions générales de chargement
.
Les essais en laboratoire ont grandement
contribué à l'élaboration de modèles de comportement qui reflètent le comportement non-linéaire et
l’
anisotrope du sol.
Les essais
de
laboratoire disponibles sont interprétés en supposant
que les états
de contraintes
et de déformation sont uniformes au sein de
l’
éprouvette testée. Ceci
permet de fournir des informations sur le comportement du matériau dans une gamme étroite de
chemins
contrainte-déformation, et ne couvrent pas les conditions générales de chargement qui se produisent dans les problèmes réels
.
Cet
article présente l'intégration de l'auto-
apprentissage
des simulations (
SelfSim)
dans le cadre d’une
analyse inverse à partir d
’
essais en
laboratoire pour obtenir
le comportement
du sol. L'application de cette approche aux essais de cisaillement simple
direct
(
DSS)
montre qu'il est possible de caractériser le comportement du sol sur une large gamme de chemins de contrainte à partir d'un seul test.
Le document décrit également le développement d'un dispositif d'essai triaxial modifié destiné à imposer des conditions de
chargement non uniformes pour augmenter la gamme des contraintes et des déformations qui peuvent être obtenues par SelfSim
.
Le
nouveau dispositif représente une étape importante vers une intégration plus étroite entre les essais de laboratoire et
l’
élaboration d'un
modèle constitutif.
KEYWORDS: SelfSim, direct simple shear (DSS), triaxial shear, inverse analysis, constitutive modeling.
1 INTRODUCTION.
In geotechnical engineering problems,
soil behavior
interpretation is commonly based on laboratory tests, such as
triaxial, plane strain, and direct simple shear tests (Ladd and
Foott 1974, Jamiolkowski et al. 1985, Mesri and Choi 1985).
These tests or devices allow soil behavior to be evaluated under
a range of loading modes, and provide in-depth understanding
of soil’s stress–
strain
–
strength behavior (Bolton 1986,
Jamiolkowski et al. 1985, Ladd et al. 1977). Uniform stress and
strain states within the specimen are generally imposed in the
device designed for laboratory testing and the soil response
corresponding to a single loading path is provided. Measured
soil response is interpreted assuming the specimen is a single
element and is sheared uniformly even from devices such as the
Direct Simple Shear (DSS) device, which generates non-
uniform stresses and strains. This is due to the lack of means to
extract the complex stress-strain behavior with a specimen. Due
to this uniformity requirement or assumption, laboratory testing
can only reveal a narrow range of soil behavior, which is
significantly different from the general loading conditions
experienced by the soil in the field.
Regardless of the extent of non-uniform conditions within
the test itself, interpretation of stress
–
strain
–
strength response is
based on the assumption of uniform conditions. The design of
complex boundary value engineering problems whereby soils
are sheared under general loading conditions requires material
constitutive models that can represent soil behavior under these
loading conditions. The process of development of material
constitutive models is lengthy and requires numerous tests to
cover a broad range of loading paths. However, all available
models are developed based on limited behavior measured by
existing laboratory tests. This limited information generally
results in a model that may not be justifiable for representing
loading conditions that differ substantially from the ones in
laboratory tests.
A weak link clearly exists between laboratory testing and
material modeling. Hashash, Ghaboussi and co-workers, over
the last decade, successfully developed an integrated test-
analysis framework to build a stronger link between material
testing and material modeling. This is accomplished through the
use of a biologically inspired inverse analysis framework, self-
learning simulations (SelfSim), which uses a neural network
(NN)-based material model to extract non-uniform stress
–
strain
i
t ti
l-
ri
t l
r t r stin f r il ehi or
Combinaison d’essais numériques et expérimentaux pour la modélisation du comportement
des sols
Hashash Y.M.A., Asmar R., Moon S.
Dept. of Civil and Environnemental Engineering, Univ. of Illinois at Urbana-Champaign, Urbana, USA
