Associated and non-associated flow rule based elastic- viscoplastic models for soft clays

摘要

Soft clays are associated with low shear strength, low permeability and high water content and they are mostly highly compressible. When a geotechnical structure for example road embankment is founded on such soft soil, it exhibits time-dependent deformations due to external loading. The excessive deformation in soft clay deposit may lead to damage of the structure or result in high maintenance cost. Nevertheless, due to increase of population and urbanization it is sometimes difficult to avoid construction of infrastructure on such clay deposit. Therefore, any structure founded on problematic soft clay requires accurate prediction of deformation with time, which includes prediction of time-dependent deformational behaviour of clay, to mitigate post construction damage. Predicting the time-dependent viscous behaviour of soft clay is a geotechnical challenge due to heterogeneity and complex formation of soils. To cope with this challenge, different elastic-viscoplastic (EVP) models are developed in this thesis adopting associated flow rule and non-associated flow rule in general stress space considering time dependent viscous behaviour of soft clay. These models are implemented in fully coupled consolidated non-linear finite element code AFENA involving composite bounding surface as well as Modified Cam Clay (MCC) equivalent single surface. Seven parameters are used in models that involve composite surface while six parameters for the single surface model, which are defined according to soil mechanics principles. Parameters of EVP models can be extracted from conventional oedometer and triaxial tests. Performances of the models in capturing the behaviours of natural clays: (Osaka clay, Shanghai clay) and reconstituted clays: (Kaolin clay, Hong Kong Marine Deposit clay, and Fukakusa clay), under drained and undrained conditions; different initial states (void ratio, mean pressure, over-consolidation ratio) and loading conditions (stress or strain controlled, strain rate) are also presented in this thesis. To account for softening behaviour of natural soft clay, a modification is also introduced for the EVP models developed in this thesis.The time-dependent viscous behaviour of soft clay is incorporated in these models adopting creep co-efficient. Implementation of constant creep in EVP model is common, whereas from long term laboratory tests it is revealed that linear or constant approximation of creep co-efficient may lead to misleading prediction. However, existing non-linear creep functions are either limited to specific EVP models or anchored in the clay for which it is derived. To avoid the complexity associated with non-linear creep function for EVP models, a general non-linear function is presented which is neither tied to any specific model nor any specific soil group. Moreover, this creep function does not involve any fitting parameters as it is evident in some non-linear creep functions proposed in the literature.To verify the predictability of the EVP models developed in this thesis for long-term performances, an instrumented Nerang Broadbeach Road (NBR) embankment’s field monitored data are compared with predictions obtained using the EVP models as well as that using the MCC model. The NBR embankment’s complete subsoil details are presented in this thesis and the model parameters are extracted from laboratory tests like oedometer tests, triaxial tests and field tests such as vane shear tests, cone penetration tests (CPT) and piezocone dissipation tests (CPT-u). The interpreted model parameters obtained from CPT and CPT-u tests are also compared with laboratory obtained data. Another important aspect of this thesis is that a simplified observational approach is proposed based on field monitored data which is capable to capture the time-dependent viscous response. The method is employed to predict the ultimate settlement as well as time-settlement relation for NBR embankment. The proposed observational approach prediction capabilities are compared with Asaoka and Hyperbola methods and coupled consolidated finite element analyses adopting MCC model as well as associated and non-associated flow rule based EVP models.