Numerical Methods in Geotechnical Engineering – Hicks, Brinkgreve & Rohe (Eds)
© 2014Taylor & Francis Group, London, 978-1-138-00146-6
Predicting the behavior of sand columns in soft clays
using hypoplastic finite element modeling
A. Riman, S. Sadek & S. Najjar
The American University of Beirut, Beirut, Lebanon
ABSTRACT: The use of the finite element method (FEM) in analyzing the behavior of clay deposits reinforced
with sand columns is not new. In previous studies conventional constitutive models were used to represent the
sand. These models do not account for the post-peak strain softening that generally occurs in compacted sands.
The main objective of this paper is to investigate the use of the recently developed hypoplastic model which
provides the ability to account for the post-peak strain softening in applications involving soft clays that are
reinforced with sand columns. This objective was achieved in two stages. In the first stage, the hypoplastic
model for both materials (sand and clay) was derived and built from the results of laboratory tests that were
carried out on Ottawa sand and Kaolin clay. The numerical analyses of triaxial tests on these sand and clay
materials, using Plaxis 2D and the hypoplastic models, closely matched the observed laboratory behavior. In the
second stage, laboratory tests that were conducted on sand-reinforced clay specimens were analyzed numerically
using the calibrated hypoplastic models. The numerical analyses were based on the same geometries and testing
conditions as those used in the actual laboratory tests with sand columns having diameters of 3 cm and 4 cm,
penetrating 10.65 cm into a 7.1 cm diameter clay specimen with a height of 14.2 cm. The consolidated drained
triaxial tests were carried out at confining pressures of 100 kPa, 150 kPa and 200 kPa. The FEM results using the
hypoplastic models closely matched the laboratory results, indicating that the hypoplastic model could be used
in FEM analyses to accurately predict the behavior of floating sand columns in soft clays under drained loading
conditions. The hypoplastic model presents an interesting and promising alternative to conventional models,
which fail to capture the full range of response of the complex clay-sand system.
1 INTRODUCTION
This paper investigates the use of hypoplasticity within
the context of the Finite Element Method (FEM) to
model the response of floating (partially penetrating)
sand columns in soft clays. The FEM has been uti-
lized previously to study the behavior or corroborate
experimental and field results of clays that are rein-
forced with granular columns. In all of the published
FEM work on granular columns in clays (Raithel and
Kempfert, 2000; Murugesan and Rajagopal, 2006;
Ambily and Gandhi, 2007; Elshazly et al. 2007;
Elshazly et al. 2008; Tan et al. 2008; Chen et al.
2009; Lo et al. 2010; Zahmatkesh and Choobbasti,
2010; Castro and Sagaseta, 2011; and Shahu and
Reddy, 2011) the sand was modeled using consti-
tutive models that do not account for the post-peak
strain softening that is characteristics of dense sands.
Most of the researchers used the Mohr coulomb model
and a few such as Raithel and Kempfert (2000) and
Murugesan and Rajagopal (2006) used hyperbolic
models to represent the response of the sand inclu-
sions. The hypoplastic model which accounts for the
post-peak strain softening in sands has not been used
in modeling sand columns in soft clays and presents
a promising avenue and tool to reliably model the
full range of the complex response of the composite
system.
The performance of the hypoplastic model was stud-
ied by comparing the predictions of the model with
experimental test results that were reported in Maalouf
(2012) and BouLattouf (2013) as part of a compre-
hensive experimental program that is based on triaxial
testing of soft Kaolin clay that is reinforced with
Ottawa sand columns (Najjar et al. 2010). Maalouf
(2012) and BouLattouf (2013) carried out consoli-
dated drained (CD) triaxial tests on normally consol-
idated Kaolin specimens, reinforced with partially or
fully penetrating single sand columns.
Hypoplastic soil models were built from the
mechanical and physical properties of the tested mate-
rials (Ottawa Sand and Kaolin Clay). The soil models
were then used in single-element analyses for calibra-
tion verification in reference to the triaxial tests carried
out in the laboratory. The calibrated hypoplastic mod-
els for the sand and the clay were then used to predict
the performance of the composite samples (reinforced
Kaolin). The main focus was on predicting the defor-
mation patterns and the stress strain behavior of the
composite specimens under fully drained conditions.
A comparison between the FEM and Experimental lab
test results was carried out to assess the reliability of
565