Quantitative electrical imaging of slope moisture dynamics

Abstract

Future climate change is likely to affect the factors which determine the stability of engineered slopes, which constitute a third of UK transport infrastructure. The cost of remediation works being approximately ten times that of preventative action, accurate methods of stability assessment are of increasing importance to stakeholders. Electrical resistivity tomography allows high resolution volumetric time-lapse imaging of subsurface resistivity distribution, which, combined with proxy relationships, offers the potential to quantitatively investigate slope moisture dynamics. Therefore, it is essential that the relationships between resistivity and the geotechnical parameters which impact slope stability (water content, suction and shear strength) are fully resolved, particularly where soils undergo significant cycles of drying and wetting, which progressively weaken fill material. This PhD thesis presents a study to establish these relationships for a remoulded clay taken from a purpose-built test embankment in Northumberland, United Kingdom. A rigorous, multi-scalar laboratory testing programme was combined with a three year geoelectrical field monitoring experiment, supported by a network of point sensors. To verify the resolution of the geoelectric method, the test site was subjected to series of hydrodynamic perturbations. To simulate seasonal effects, drying and wetting cycles were imposed on laboratory specimens. Results indicated an inverse power relationship between soil resistivity and water content, and showed the more conventional, two point method to over-estimate values due to the inclusion of contact resistances. Linear hysteretic relationships were established between undrained shear strength and water content, demonstrating suction loss between drying and wetting paths due to hysteretic soil water retention. Laboratory relationships were observed to evolve with ongoing seasonal cycling, due to soil fabric deterioration associated with drying beyond the continuity of the pore water phase. Trends observed in the laboratory were supported by images obtained from scanning electron microscopy. Waxman-Smits and Van Genuchten modelling parameters were applied to resistivity – water content and soil water retention curves respectively, defining proxy relationships which were then used to translate field resistivity data into estimates of both water content and suction. ii These images captured general seasonal trends in subsurface moisture processes, in response to sustained environmental conditions, allowing a structured hydrological model of the test embankment to be resolved. Localised damping of the geotechnical response was observed, as a function of depth, aspect and compaction. On a weekly scale, the development of near-surface cracks was captured during the summer months. Rainwater ingress following rapid rainfall events was investigated via daily time-lapse imaging, and highlighted seasonal differences in infiltration processes. Proxy-derived water content estimates compared very well with those measured using the point sensor network, both qualitatively and quantitatively, providing invaluable information on the dynamic moisture processes which precede slope failure, with particular reference to soil fabric deterioration. Ultimately, this thesis describes a methodology for translating ERT-derived resistivity data into information directly relevant to slope stability. It highlights the importance of considering soil water retention when estimating in situ soil suctions, employing a localised saturation history-based approach to account for spatial variation. The issue of ambiguity inherent to the nature of inverse theory is discussed, with suggestions for its minimisation, including the use of three- rather than two-dimensional resistivity data, temperature correction and correlation with point sensors. This methodology has been incorporated into a set of computer programmes which read in raw ERT data and automatically convert to geotechnical data, furthering the development of a fully-automated slope stability monitoring system.