| Summary: | <p>Monopiles are the major current foundation type for offshore wind turbine foundations. The geotechnical design of monopiles usually employs conventional ‘<em>p-y</em>’ approaches, originally developed for the design of long slender piles for the oil and gas industry. It has been recognised, however, that the p-y approaches may not be entirely appropriate when applied to the design of monopiles for offshore wind turbines. Evidence from full scale field measurements indicates that they can underestimate the stiffness of the soil-pile response. A joint industry project – PISA (Pile Soil Analysis) was established to address these limitations and to provide improved design methods for monopiles. The new PISA methodology addresses the pile-soil interaction with a distributed lateral load and moment along the pile shaft, along with a horizontal force and moment at the pile base. Two design methods, incorporating these soil reaction curves, the rule-based method and the numerical-based method, have been proposed. The PISA project demonstrated that the two methods provided better predictions of pile response, as shown in medium scale field tests, compared to the conventional p-y approaches. The PISA project focused on two reference soils. Further work is needed to apply the new methods to different design scenarios and to assess their sensitivity to different constitutive models and boundary conditions.</p> <p>This thesis presents a series of 3D and 1D finite element (FE) analyses to examine the applicability of the numerical-based approach, considering a wide variation of pile geometries, load eccentricities, soil profiles and layered soils. The 3D FE analyses, which were performed in Abaqus 6.13, employed elastic perfectly plastic constitutive models and considered pile-soil interaction. Load-displacement response and soil reaction components were produced by a small number of calibration analyses in homogeneous clay and sand. A systematic procedure was developed to obtain soil reaction curves from the calibration analyses. These were subsequently used for input to a 1D model. Further 3D analyses for new design scenarios were performed, and the results computed were compared with the results predicted by the calibrated 1D model. The predictability of the calibrated 1D model was comprehensively evaluated. In addition, the soil reaction curves of rigid piles in lateral translation and rotation about the pile base were analysed.</p> <p>The results demonstrated that the numerical-based method and the 1D model are capable of predicting monopile lateral response in design scenarios involving variations of pile geometries and soil properties. Compared with 3D FE results, the errors of overall load-displacement response and ultimate capacity predicted by the 1D model were less than 10%. The calibrated 1D model was also applicable in layered soil cases, with the errors being less than 15%. The rigid pile analyses indicated that load eccentricity and boundary conditions have greater influences on soil reaction curves in sand than in clay. Inspection of base soil reaction components demonstrates that a coupling effect existed between base horizontal force and base moment, which should be evaluated in future monopile design methods.</p>
|
|---|