Pile foundations used on and offshore in soft rock encounter complex installation and loading conditions which are not well captured by existing pile design methodology. Inspired by the goals of ICE-PICK, this research aimed to address rock-structure interactions controlling pile foundation performance in chalk and other calcareous bonded materials. The study included an experimental programme comprising element tests and small-scale 1g model pile tests. The latter were configured to study the subtleties of pile installation in soft rock (and their surrounding processes), their impact on axial/lateral monotonic and cyclic performance and the effects of long-term processes such as corrosion-induced set-up. The test series was facilitated by a new multi-axis loading frame developed as part of this doctoral work, designed specifically for tandem use with X-ray CT (XCT). Sub-surface rock-structure interaction observations drawn from the pile and element test series supported a better understanding of the short- and long-term performance of piles in soft rock and the development of practical design tools. The author's work supported the following overarching findings/conclusions: (a) the hydro-mechanical-fabric characterisation of various low-medium density soft rocks via conventional element testing revealed how inter-particle bond strength, porosity and saturation level controlled their pre-yield compressive- and tensile-strength. Compressive stress paths studied using a radially instrumented oedometer (1D-k0) depicted three distinct phases. These were controlled by the soft rock's pre-yield elasticity, post-yield progressive bond breakage (strain localisation/compaction banding) and subsequent destructurisation to an intrinsic assembly of cohesionless soil. Under the same stress path, measured permeability evolved from that controlled by the intact porous matrix (high permeability) to that of the destructured soil (low permeability) describable through a modified Kozeny-Carman formulation. The constitutive description of the soft rock was satisfied by a Structured-Modified Cam Clay model, thereafter employed for numerical simulations as part of the ICE-PICK project. (b) the development and fabrication of a multi-axis loading apparatus capable of performing high load small-scale geotechnical model tests. Its compactness enabled operation from within conventional laboratory XCT scanners at both the University of Dundee (SMART Lab) and Southampton (MUVIS); (c) model driven/jacked open- and closed-ended pile tests performed in the rock characterised in (a) demonstrated how installation destructures the rock close to the pile shaft, impacting subsequent lateral and axial pile performance. Live radiography and XCT captured installation processes including plug state evolution and associated end-bearing mechanisms. Insights demonstrated analogies to unplugged pile installation in the field whilst also identifying new risks relating to pile plugging. Further insights indicated that pile shaft, base capacity and rock damage depend on pile geometry and rock fabric. Rock damage promoted phenomena like pile gapping during lateral cyclic testing which subsequently curtailed axial capacity and exacerbated low-load lateral stiffness. Meanwhile, significant annuli disturbance, excess pore-pressure generation and confinement loss around pile tips appeared to be the probable cause of capacity losses during high amplitude short-term axial cyclic loading. Finally, long-term capacity growth was attributed to fabric (chemo-mechanical) changes and strength increases of the same destructured rock, which appeared to be influenced by both the corrodibility of the pile fabric and extent of disturbance. (d) observations from (a) and (c) inspired a review of existing weak-rock lateral pile design approaches (p-y) which prompted the development of a simple stress-strain based p-y method to capture pile installation effects. The new concept-level/first-iteration design approach was trialled on (c) and a field-scale database established as part of the work, showing promisingly low calculated-to-measured ratios. The method was simplified to a first-order elastic-plastic spring for high-level design applications, requiring either intact dry density or local CPT data.
T. Riccio (Thu,) studied this question.