Degree Name

Doctor of Philosophy


School of Civil, Mining and Environmental Engineering


The loading conditions in railway substructures are cyclic, unlike the steady seepage under static loading that usually occurs in dams and levees. The mechanisms of seepage and filtration, and between the subgrade and subballast layers, including the spatial and temporal changes in their drainage characteristics warrant further investigation to improve the existing procedures for designing filters. In practice, a drainage layer (subballast filter) is expected to perform two major functions in rail track environments; namely (1) cope with the sustainable transfer of stress from ballast to the subgrade, and (2) protect subgrade fines from upward pumping due to seepage and the build-up of pore pressure in subgrade due to cyclic loading. An effective subballast filter mitigates ballast fouling from the substructure and reduces the risk of track settlement due to clay pumping, hence making the rail track foundations safer. In contrast, a badly designed filter subjected to strong seepage forces and agitation induced by cyclic loading could lose its finer fractions to erosion and experience changes in its particle size distribution. This phenomenon is known as internal instability that may result in a highly porous and ineffective filter layer.

This research reports the results from 67 hydraulic tests carried out to examine the potential for internal instability of a selected range of 10 different soils which conform to the typical subballast gradations commonly used in Australia. Hydraulic tests were carried out using a modified hydraulic apparatus designed to capture the response of soils subjected to an upward flow under static and cyclic loading. This apparatus can monitor various factors that influence the inception of instability in soils, e.g. spatio-temporal variations in porosity, average and local hydraulic gradients, and the effective stress distribution with depth. An analysis of the test results revealed that the existing criteria are insensitive to any variations in the level of compaction of soils, and tend to be unsafe when applied under cyclic conditions. As a consequence, a new constriction size distribution based criterion for assessing the potential of internal stability is proposed that can accurately differentiate between internally stable and unstable soils. A large dataset of 95 tests are used to validate this technique, which is also sensitive to the level of compaction of soils, an area that was ignored in most existing criteria.

Furthermore, cyclic loading promotes premature washout failures in internally unstable and marginally stable soils, while constant agitation and the internal development of pore pressure due to cyclic loading affects the geometrical arrangement of stable constriction network enough to allow internal erosion to evolve. Erosion becomes excessive at higher frequencies, to the extent whereby marginally stable specimens also become increasingly unstable. The existing (static) criteria could not capture the effects of cyclic loading to correctly assess the internal stability of some of the soils tested herein, so the constriction based criterion proposed in this study is modified to include the effects of cyclic loading and then demarcate a more distinct boundary between stable and unstable soils under cyclic conditions, for a large experimental dataset of 87 results.

Internally stable, marginal, and unstable soils are characterised by heave, composite heave-piping, and suffusion that develops immediately after instability commences. The stable specimens exhibited heave at larger hydraulic gradients than the unstable specimens which failed by suffusion at relatively smaller hydraulic gradients. Under no external load (i.e. self-weight only), the relative density (Rd) and particle size distribution (PSD) together controlled the internal stability of soils, although the effective stress magnitude (𝜎′𝑣𝑡) also played a role in static and cyclic loading conditions. Instability in soils was governed by specific combinations of their geo-hydro-mechanical characteristics such as PSD, Rd, the stress reduction factor, critical hydraulic gradients and associated levels of effective stress. These factors are combined to model the development and inception of instability as well as developing visual guides as a practical tool for practitioners. Each soil possesses a unique critical envelope related to its PSD and Rd, and a critical path with its inclination that depends on the hydro-mechanical conditions. The current results of internal erosion tests conducted by the authors under static and cyclic loading, plus those adopted from literature, are used to verify the proposed models and demonstrate their practical implications.

This thesis is unavailable until Thursday, November 02, 2017