Page 175 - Numerical Analysis and Modelling in Geomechanics

156 C.L.RAMSHAW AND A.R.SELBY
Vibrodriving at the Second Severn Crossing
A detailed study was made of the installation of steel 1050 mm diameter casings
for one of the foundations for the approach viaduct to the main cable-stayed
spans of the Second Severn Crossing. The ground conditions consisted of soft to
firm clays to a depth of 13.8 m, overlying firm to stiff marl. These conditions are
far from ideal for Vibrodriving, since liquefaction is unlikely. However,
penetration to 15.5 m was achieved using a PTC 50H3 vibrodriver with an
eccentric moment of 50 m.kg, and running at 15.8 Hz. Ground surface vibrations
were measured using geophone triaxial sets at distances of 5.7 m, 14.5 m and 32.
9 m. Comparisons are made between measured and computed ground vibrations
in Figure 5.19 for the radial components, and in Figure 5.20 for the vertical
components.
Overall, close agreement is obtained in terms of form, amplitude and
attenuation, for both radial and vertical particle velocities.

Ground wave modelling and applications
In the previous two sections, methods have been presented for computing
outgoing ground waves, taking as the starting point the driving force from either
an impact hammer or a vibrodriver. For effectiveness of computation, and for
identification of the effects of parametric variation, the impact model has been
separated into three stages, while the vibrodriving has been split into two stages.
The three-stage model for impact hammers requires a number of parameters to
be evaluated. Of particular significance are the spring and damper values
ascribed to the dolly or packing; low stiffness and damping imply a force-time
relation with a lower peak and longer duration. Soil parameters for both shaft and
toe are also significant with respect to displacements of interface nodes as input
to the third stage of the model.
The two-stage model of the vibrodriver has a well-defined input excitation,
since the cyclic vertical force is known as a function of the eccentric masses and
their frequency of rotation, and in particular because the vibrodriver is firmly
clamped to the pile head. However, the shaft and toe interaction with the soils is
less predictable, with liquefaction as a key mechanism.
With both procedures, the above uncertainties are present, together with
ground non-uniformities, and imperfections in the driving process in terms of
non-axial forces and guide-frame weakness. In consequence, the modelling
procedures are shown to be highly effective when used in a back-analysis mode,
but their ability to predict ground waves is less secure. On-going work to back-
analyse a number of site cases with measured ground surface vibrations will
improve the predictive capability.
A major advance arising from the modelling capability is the potential for
computing the dynamic response of structures or buried services adjacent to the
pile driving, with full soil-structure interaction. When undertaking computer