833
Validation of computational liquefaction in plane strain
Validation de liquéfaction simulée en déformation plane
Wanatowski D.
Nottingham Centre for Geomechanics, University of Nottingham, United Kingdom
Shuttle D.A., Jefferies M.G.
Golder Associates Ltd, Nottingham, United Kingdom
ABSTRACT: This paper presents a validation of computational liquefaction in plane strain for a user-defined model that closely
replicates liquefaction across a wide spectrum of soils. The results of triaxial and plane strain compression tests on Changi sand (used
in large reclamation projects in Singapore) are utilized for property determination and validation respectively. The stress-strain
behaviour is computed using a user-defined implementation of the NorSand general plasticity model in the FLAC numerical platform.
Reasonable matches are obtained between the plane strain data and the computed responses, both in terms of stress-path and stress-
strain behaviour. This accomplishes one necessary step towards allowing engineering practice for liquefaction damage
assessment/mitigation to use a convenient computational platform anchored in a proper mechanics-based representation of soil
behaviour.
RÉSUMÉ: Cet article présente une validation de liquéfaction simulée en déformation plane dans le cas d’un modèle original qui
reproduit de près la liquéfaction dans une large gamme de sols. Les résultats d’essais triaxiaux et de compression en déformation
plane sur du sable Changi (employé dans les grands projets de remblaiement à Singapour) sont utilisés respectivement pour la
détermination de ses propriétés et pour la validation du modèle. Le comportement contrainte-déformation est calculé en introduisant
le modèle de plasticité général NorSand dans le code de calcul FLAC. Les tests de validation et les simulations équivalentes
présentent des résultats plutôt proches, que ce soit en termes de parcours de contrainte ou de comportement contrainte-déformation.
Les pratiques de construction concernant les estimations/atténuations des dégâts dus à la liquéfaction peuvent ainsi être facilitées par
l’utilisation de cette approche numérique associée à une représentation mécanique correcte du comportement des sols.
KEYWORDS: sand, liquefaction, plane-strain, constitutive modelling.
1 INTRODUCTION
Earthquakes remain an ever-present hazard, with soil
liquefaction continuing to be a dominant mechanism in the
consequent infrastructure damage and losses (e.g. the recent
Chirstchurch events). Although the geologically-based
“NCEER” method (Youd et al. 2001) underlies most current
earthquake hazard reduction practice in geotechnical
engineering, the limitations and flawed physics of the NCEER
aproach are becoming increasingly recognized. These
limitations can be overcome by adopting advanced constitutive
models, possible in engineering practice with the ‘user defined
model’ facility of commercial numerical analysis platforms (e.g.
as available in the popular FLAC and Plaxis platforms).
The past decade has also witnessed a greatly increased
demand for metals, an economic trend that seems unlikely to
change as ‘BRIC’ group living standards continue to approach
those in the ‘developed’ world. Metals must be mined, and one
result of mining is vast quantities of ground rock – ‘tailings’,
which are produced as part of ore extraction. Usual mining
practice is to impound tailings using dams. A new trend with
tailings is to reduce their water content during deposition and
“stack” the tailings above the retaining dam – economically
attractive, but with the obvious potential for large scale release
of these waste materials to the environment if a liquefaction-
driven flowslide develops (e.g. failure of the Merriespruit
Tailing Dam analysed by Fourie et al. 2001).
Liquefaction can be triggered by various mechanisms
(Jefferies and Been 2000, Chu et al. 2003) but, regardless of
trigger, the greatest damage (deformation) arises when the post-
liquefaction strength is less than the pre-liquefaction stress state
– a situation captured in laboratory tests that focus on static
liquefaction.
This paper presents a plane strain validation using a critical
state based model that closely replicates liquefaction across a
wide spectrum of soils. Plane strain approximates the conditions
that arise in most geotechnical construction, certainly far more
so than the triaxial paths that underlie current geotechnical
understanding. Here, conventional triaxial compression tests are
used for property determination, and subsequently static
liquefaction tests in plane strain are used as the validation case.
The tested material is Changi sand, a sand used at large
reclamation projects in Singapore (Wanatowski and Chu 2007,
2012). The stress-strain behaviour is computed using a user-
defined implementation of the NorSand general plasticity model
(Jefferies 1993, Jefferies and Shuttle 2002) in the FLAC
numerical platform (Shuttle and Jefferies 2005). Reasonable
matches are obtained between the plane strain data and the
computed responses, both in terms of stress-paths and stress-
strain behaviour. This accomplishes one necessary step to allow
engineering practice for liquefaction damage assessment/
mitigation to use a convenient computational platform anchored
in a proper mechanics-based representation of soil behaviour.
2 LABORATORY EXPERIMENTS
2.1
Changi sand
Changi sand is a subangular marine dredged silica sand used for
the Changi land reclamation project in Singapore. The Unified
