2667
Analysis of Full-Scale Random Vibration Pile Tests in Soft and Improved Clays
Analyses à grande échelle de vibrations aléatoires sur pieux dans un sol argileux
Ashlock J.C., Fotouhi M.K.
Iowa State University, Ames, IA, USA
ABSTRACT: Full-scale pile vibration test results are analyzed for steel HP piles installed to a depth of 6 m in a soft clay profile, with
one pile surrounded by a cement-deep-soil-mixed (CDSM) improved zone. Multi-modal tests with vertical and coupled lateral-
rocking vibrations were conducted using a shaker mounted on a rigid pile cap. The improved soil zone significantly increased the
stiffness of the measured vertical response, but had little effect on the lateral-rocking mode. Results of the forced vibration tests are
analyzed using methods reported in the literature, including impedance functions and an approximate computational method which
incorporates variation of soil properties with depth. The simplified model is able to capture the vertical response reasonably well in
both the improved and native unimproved soil profiles, as well as the lateral response in unimproved soil. For the pile in improved
soil, however, calibration of the model to the observed vertical mode results in a greatly stiffened lateral-rocking response which was
not observed experimentally. To improve the simulation results, more sophisticated computational solutions are proposed for
modeling the dynamic interaction of the pile and improved soil.
RÉSUMÉ : Les résultats d’essais de vibrations à grande échelle sur des pieux en acier (HP) et un pieu renforcé en tête par un mélange
sol-ciment, installés sur 6 mètres de profondeur dans de l’argile, ont été analysés. Des tests multimodaux avec vibrations verticales et
balancements latérales ont été réalisés à l’aide d’un actionneur monté en tête des pieux. Nous montrons que le renforcement du sol
améliore la réponse verticale de manière significative mais n’a que peu d’effet sur la réponse latérale. Les résultats des essais ont été
interprétés en utilisant des méthodes publiées dans la littérature, notamment une méthode qui prend en compte des fonctions
d’impédance et une méthode numérique qui prend en compte les variations des propriétés du sol en fonction de la profondeur. Le
modèle simplifié utilisé est capable de décrire correctement la réponse verticale pour les deux types du sol, avec ou sans
renforcement, ainsi que la réponse latérale pour un sol non renforcé. Cependant, l’ajustement du model à partir de la réponse verticale
rend compte d’une plus grande raideur latérale que celle observé expérimentalement. Ainsi, afin d’améliorer les résultats de nos
simulations, nous proposons des modèles plus sophistiqués qui prennent en compte l’interaction dynamique des pieux avec le sol
renforcé.
KEYWORDS: soil-pile interaction, soil dynamics, random vibration, pile, soil improvement, soft clay, impedance.
1 INTRODUCTION
Accurate characterization of the dynamic interaction between
foundations and layered soils is an important issue for the
design and analysis of foundations under seismic or vibratory
loading. To date, several solutions ranging from simplified 2D
approximations to 3D numerical models have been developed
and employed to analyze soil-foundation interaction. However,
validation and calibration of the various methods against full-
scale field tests is essential for an understanding of their relative
capabilities and limitations. While analytical and computational
studies in the literature are numerous, the volume of full-scale
field testing studies is comparatively limited. To help bridge the
knowledge gap between theory and experimentation in soil-pile
interaction problems, the current study investigates a series of
full-scale dynamic field tests of two identical steel HP 250x63
(English HP 10x42) piles installed to a depth of 6 m in a soil
profile featuring soft clay. The influence of local soil
improvement on the dynamic pile response is also examined
experimentally using a 1.2 m diameter, 4 m deep cement-deep-
soil-mixed (CDSM) zone installed in the soft clay layer
surrounding one of the piles. The piles were subsequently used
for a reaction frame in a related study, which limited the pile
spacing and diameter of the improved zone. The pile in the
native unimproved soil profile is referred to as pile U, and the
pile in the improved CDSM soil as pile I. The random vibration
test procedures employed are described below, followed by
analyses of the experiments via simplified 2D numerical models
developed for dynamic interaction of piles with layered soils.
A newly developed servo-hydraulic shaker system was used
to deliver three types and various levels of excitation. The
excitation types used were chaotic impulse, random and swept-
sine, denoted C, R and S, respectively. Theoretically, the
broadband random (R) signal has a uniformly distributed energy
over all frequencies while the swept-sine signals (S) concentrate
the excitation’s energy at a single frequency which is
continually changing within a predefined interval. The chaotic
impulse (C) excitation consists of a series of randomly timed
impulses with randomly distributed amplitudes. In all tests,
accelerations of the shaker and pile cap were measured in the
horizontal (x) and vertical (z) directions in the plane of motion
of the pile cap using six uniaxial accelerometers. Additionally,
seven triaxial accelerometers were buried 6 inches below the
soil surface at selected locations to record the near-surface
vertical and horizontal motion. Figure 1 details the test setup
including the soil profile, improved zone and sensor
arrangement.
For data acquisition and real-time analysis in the time and
frequency domains, a 20 channel dynamic signal analyzer was
programmed in LabVIEW to record time histories and spectral
quantities including FFTs, auto- and cross-spectral densities,
transfer functions, and coherence functions. The stimulus for the
transfer functions was taken as the force applied by the shaker’s
inertial mass in the direction of excitation, and all other
accelerations of the pile cap and soil were treated as response
quantities. Additionally, the full time histories of all sensors
were simultaneously recorded on the nees@UCLA Kinemetrics
Granite seismic recording systems to enable further
interpretations such as time-domain analyses or different
stimulus-response combinations.
