Evaluation of undrained response from drained triaxial shear tests: DEM simulations and Experiments

The undrained shear response of monotonically loaded isotropically

consolidated saturated sands can be characterised

by a change in the excess pore pressure generation in the

sample. The generation of positive and negative excess pore

water pressures is related to contractive and dilative responses.

The increase or decrease in pore water pressure

continues until it reaches critical state (known as steady state

for undrained tests). In general, Casagrande’s definition of

critical state for sands has been utilised in most commonly

used liquefaction analyses, and is referred to as the steadystate

procedure (Castro, 1969; Casagrande, 1977; Castro &

Poulos, 1977; Poulos et al., 1985). In the laboratory, consolidated

undrained triaxial tests on both reconstituted and

undisturbed samples are generally carried out to evaluate the

steady state of sands (Poulos et al., 1985). However, it is

very clear that steady-state determination from consolidated

undrained tests with pore pressure measurements is sensitive

to parameters such as initial effective confining pressure and

initial fabric (Poulos, 1981; Been & Jefferies, 1985; Been et

al., 1991; Castro et al., 1992). The most recent findings of

De Gregorio (1990) indicate that critical state is influenced

by the method of soil sample preparation (moist tamping,

moist vibration or dry pluviation). This may be due to the

volume change tendency caused by the difference in the

fabric of the sand, which affects the critical-state response.

Furthermore, such behaviour also depends on the loading

system equipment’s capability to keep up with the potential

for sample deformation, an important point with regard to

potential differences in testing equipment from one laboratory

to another. In this regard, Norris et al. (1997) developed

a methodology to predict the undrained shear response of

sands from drained triaxial tests carried out from isotropic

rebound paths based on the effective stress concept. This

method makes it possible for the majority of geotechnical

firms to participate in the prediction of static liquefaction

and residual strength by performing traditional drained tests

with volume change measurements. Furthermore, this method

also provides the condition and logic for the development

of complete as against limited liquefaction (Norris et al.,

1997).

In the laboratory, drained triaxial shear tests were used to

predict undrained behaviour using samples consolidated to

the desired confining pressure and then rebounded to lower

pressures. However, it is seldom possible in the laboratory to

consolidate the assemblies along identical paths owing to the

difficulty of preparing samples with the same initial fabric.

In this technical note, the method proposed by Norris et

al. (1997) is revisited using discrete element methods

(DEM) (Cundall & Strack, 1979), by which means the

sample preparation problem can be avoided. In addition,

laboratory experiments on clean sands were carried out to

validate the numerical simulation results using DEM.