Landslides in quick clay are often large, may occur without warning and have catastrophic consequences. The physical process of failure in quick clay material is not fully understood, and we encounter problems when conducting numerical analyses. A failure zone (shear band) forms when quick clay fails. The thickness of this shear band determines the response of the quick clay to external loading. The problem with numerical analyses is that the shear band thickness becomes too thin, and hence analyses produce unrealistic results.
The objective of this project is to explore how two naturally occurring processes in quick clay, namely rate dependency and pore water flow, affect the thickness of the shear band in a numerical analysis. We hope that when these processes are accounted for, that the shear band will not thin to a small size.
The project is exploring the above theme by making a new tool (soil model) which is capable of reproducing rate dependency and pore water flow for a quick clay. The soil model is then used to analyse a range of cases: from one-dimensional shearing to two-dimensional landslide problems.
The results to date indicate that the newly introduced processes of rate dependency and pore water flow are sufficient in a one-dimensional problem to delay the onset of numerical problems. Analyses of more complex two-dimensional problems are underway.
The initial micro-structural fabric and the macro stress-strength response of an undisturbed fluvial sand has been investigated in this research. The fluvial deposited sand was obtained from a well characterized research site named Øysand near Trondheim, Norway. Efforts were made to obtain samples using state-of-the-practice and state-of-the-art piston techniques, but without much success; hence the ground freezing for sampling technique was assessed as an alternative. After performing series of frost-heave laboratory tests, the susceptibility of Øysand soils was investigated. The study concluded that ground freezing for sampling was a viable alternative to sampling undisturbed sand, but not silt. Ground freezing for sampling was then successfully carried out in the field and several metres of frozen soil were obtained. Using ?CT-scanning and image analysis, a frozen sample was then used to investigate the micro-fabric of the undisturbed soil. The initial fabric of the frozen soil was then compared against the fabric imposed by four different reconstitution methods, namely dry deposition, moist tamping, water spooning, and slurry deposition. Moreover, additional frozen samples were carefully handled, thawed, and tested under triaxial monotonic drained and undrained conditions. The monotonic behaviour of undisturbed sand was compared against test results of specimens reconstituted to the same void ratio and using the exact same original soil. The experimental results indicate that neither the initial fabric, nor the stress-strength response of fluvial deposited sand can be exactly replicated in the laboratory. However, from the four reconstitution methods used, the slurry deposition produced the closest results, both in terms of initial fabric and the stress-strength response, while moist tamping showed the greatest differences. In general, this study highlights the importance of the initial fabric on the micro- and macro-behaviour of granular materials and its implications to geotechnical engineering research and practice.