369
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
1
Soil structure in gravel-mixed sand specimen and its influence on mechanical
behavior
Structure du sol des échantillons de sable avec gravier et son influence sur le comportement
mécanique
Kodaka T., Cui Y., Mori S., Kanematsu Y.
Meijo University
Lee K.-T.
CTI Engineering
ABSTRACT: Soil parameters are very important to conducting stability inspections of earth structures, e.g., river levees. However,
the results of triaxial tests using reconstituted sand specimens change easily depending on the specimen preparation method. In the
present paper, a series of conventional undrained triaxial tests is performed using specimens with the same void ratio, but different
initial water contents. Even though the specimens are fully saturated before the consolidation process, the shear behavior varies
remarkably with the different initial water contents. It is first assumed that the initial suction in the soil sample during the specimen
preparation may produce various soil structures. Next, a numerical simulation by the SYS Cam-clay model is carried out to confirm
this assumption. It is seen that the numerical simulation can reproduce the various types of experimental shear behaviors of the gravel-
mixed sand derived from the different soil structures.
RÉSUMÉ : Les paramètres du sol sont très importants pour effectuer le contrôle de stabilité des structures en terre (ex : digue).
Cependant, les résultats des essais triaxiaux utilisant des échantillons de sable reconstitués changent facilement en fonction de la
méthode de préparation de ces échantillons. Dans cet article, une série d’essais triaxiaux non drainés conventionnels sont réalisés en
utilisant des échantillons de même indice des vides mais avec une différente teneur initiale en eau. Bien que les échantillons aient été
complètement saturés avant la phase de consolidation, le comportement des cisaillements change en fonction de la teneur en eau
initiale. On peut supposer que la succion initiale lors de la préparation de l’échantillon peut induire des structures de sol différentes.
Ensuite, une simulation numérique par modèle de SYS Cam-Clay est réalisée pour confirmer cette hypothèse. La simulation
numérique peut reproduire les différents comportements en cisaillement des échantillons de sable avec gravier dérivés de différentes
structures de sols.
KEYWORDS: soil structure, gravel-mixed sand, strengh coefficient, triaxial test, numerical simulation, river levee.
1 INTRODUCTION
Soil parameters, e.g., strength coefficients
c
and
, are very
important to the design and the inspection of earth structures.
However, practicing engineers do not necessarily note the
reliability of the soil parameters. In the stability inspections of
river levees consisting of gravel-mixed soils, the required soil
parameters have to be obtained through laboratory triaxial tests
using specimens reconstituted with mechanically stabilized soil
and without large-sized gravel. In such cases, the test results
change easily depending on the specimen preparation methods,
even when the specimens are reconstituted to have the same
common void ratio. In practice, however, there are no rules
regarding the specimen preparation methods. The first aim of
the present paper is to show the difference in the results
obtained by triaxial tests using specimens of reconstituted
gravel-mixed sand. Dry and wet soil samples, with various
water contents, were tamped in steel molds to the prescribed
degree of compaction. Specimens with the same void ratio, but
different initial water contents, are prepared. These specimens
are then fully saturated to conduct conventional triaxial tests
under undrained conditions. Ishihara (1993) showed that each
type of undrained shear behavior of the Toyoura sand
specimens, made by various specimen preparation methods, is
different from the others. In his study, the void ratios of all the
specimens are different due to the initial structure of the sand
particles. However, since the specimens in this study are
reconstituted using well-graded natural gravel-mixed sand, it is
possible for the initial void ratios of all the specimens to be the
same. This point is quite different from the results of Ishihara’s
experiment.
A series of the triaxial tests is performed using the
specimens, with three degrees of compaction and five initial
water contents for each degree of compaction. The shear
behavior, including both stress-strain and dilatancy relations, is
seen to vary remarkably with the different initial water contents.
The authors assume that the soil structure in the specimen
causes this difference in shear behavior. Namely, the initial
suction in the soil sample during the specimen preparation may
produce the various soil structures. The second aim of this paper
is to confirm the assumption using a numerical simulation by
the SYS Cam-clay model (Asaoka et al. 2002).
2 TEST PROCEDURE
Figure 1 shows the grain size distributions of the gravel-mixed
sand samples. The natural soil was taken from a real river levee.
Mechanically stabilized soil is used in this study, which
involved the removing of over 9.5mm gravels from the natural
soil.
0.1
1
10
100
0
20
40
60
80
100
9.5mm
Precentage by mass
passing(%)
Grain size(mm)
Natural soil
Mechanically
stabilized soil
Figure 1. Grain size distributions of gravel-mixed sand used for this
study.
