AbstractTo reduce carbon emissions, the rationale for applying high factors of safety in civil engineering should be challenged by academics and practitioners by considering new design and management processes for buildings and infrastructure. An assessment of foundation safety requires the prediction of reserve capacity of the geotechnical structure under extreme loading. However, since loads are difficult to measure, assessing the real safety margin is usually achieved by examining the extent of ground displacements and building damage after an extreme event has taken place. Even under safe conditions, operating a structure may provoke excessive ground movements which cause functional and economic loss in nearby assets. Performance-based methods such as Mobilisable Strength Design (MSD) may present an opportunity to improve design efficiencies if the uncertainty associated with settlement predictions can be reduced.
This thesis presents a method to develop and analyse soil test databases for the purpose of characterising stress-strain variability in soils. Various strength mobilisation models are evaluated for their suitability to characterise nonlinear stress-strain behaviour of soil within the moderate stress range. Behavioural influences are studied using the parameters of the chosen model in the context of a statistical framework using a large laboratory database of reconstituted fine-grained materials collected from various publications. Sensitivity of the parameters to shear mode anisotropy, stress history, strain rate, liquid limit and plastic limit is examined by employing single and multiple linear regression analysis techniques. Results of the database analysis are compared with observations from previously reported tests on intact Bothkennar clay and a laboratory programme of isotropically consolidated triaxial compression and extension tests on Kaolin and Bothkennar clay. The new tests validate the relationships between mobilisation strain and overconsolidation ratio identified by analysis of the database and provide evidence for the causes of parameter variability, including differences in procedure and measurement uncertainty.
Using the framework presented in this study, informed decisions can be made about whether to use an empirical correlation calibrated with a database, accepting that the error range in the prediction represents parameter variability of the various soils in the database, or to invest in more tests to achieve the estimated reduction in variability for a particular soil. The results presented for reconstituted soils and a single deposit of soft clay demonstrate the potential value in using test databases for geotechnical variability analysis which could assist ground characterisation assessments on large-scale infrastructure projects.
|Date of Award||23 Jan 2020|
|Supervisor||Paul J Vardanega (Supervisor) & Erdin Ibraim (Supervisor)|