Constitutive modelling of cyclically loaded soils for application in offshore engineering

Abstract

The increasing demand of energy combined with the aim for reduction of carbon emissions has driven the development of alternative energy sources such as offshore wind. In this sense, significant research efforts have focused and focus today in the reduction of the installation costs of these super-structures. The research presented in this thesis, carried out under the umbrella of the PISA (PIle Soil Analysis) research project, focuses on the development of constitutive models capable of reproducing soil behaviour subjected to cyclic conditions as those encountered in offshore wind turbines. This thesis presents a comprenhensive study on the applicability of non-linear kinematic hardening rules to soils under cyclic conditions. While non-linear kinematic hardening rules are widely used to describe ratchetting and cyclic behaviour of rubbers and metals, they are less popular in the field of geomaterials. In this sense the work included in this thesis is novel as it presents a new framework for soil modelling covering the theory, its implementation and also application to real engineering problems. The presented constitutive models uses the lessons learned from the endochronic theory from the 70-80’s, combined with the mathematical structure of hypoplasticity and finally, uses several acknowledged modelling strategies from cyclic plasticity used in metals. The framework is presented as general as possible as a unified framework capable of reproducing both sand and clay with simple modifications. The thesis also studies comprehensively the computational performance of various numerical integration strategies for the rate form equations comprising the framework. Two models are tested under a wide range of directions where a semi-implicit discretisation using an augmented variational form of the closest return method showed excellent capabilities in terms of precision, global convergence and performance. Extensions to the basic models for sand and clay are presented using the principle of backstress superposition with the objective of enhancing their response under cyclic conditions. The proposed extensions are tested for the basic sand version with real laboratory data from the Karlsruhe soil database, where the predicitons of 82 laboratory tests are presented. Other type of possible extensions are also presented through the use of “numerical artifices” to include perturbation of other history variables, in particular, a method for introducing a jump-in-cycles 6 method as a backstress perturbation is presented in order to include the modelling of large batches of cycles. The last part of the thesis focuses on the application of the previous proposed extensions to different boundary value problems, where all the capabilities of the model are tested and discussed at a global and local level. Finally, the sand model with various superimposed hardening rules is applied to reproduce the monotonic and cyclic large scale field tests carried out in Durkirk sand during the PISA research project, where the response of the pile is described in detail, covering both: the soil response arround the pile and its structural response.