INFLUENCE OF CORRELATION LENGTH OF RANDOM .. SOIL PROPERTIES ON BEHAVIOUR OF MONOPILES SUBJECTED TO LATERAL LOAD

Abstract

Energy from renewable sources is the need of today's era and lot of research is going on to make the optimized use of renewable sources. One of the sources with great potential is wind energy, captured by wind turbines installed onshore and offshore. Due to many reasons such as environmental impact and lack of space available for onshore turbines, the number of offshore turbines steadily increases. Moreover, an offshore wind turbine can be run more efficiently than onshore, due to stronger and more constant wind. In this dissertation work a study is carried out to evaluate the influence of spatial variation of the soil stiffness as a function of correlation length on behavior of monopiles in sand subjected to lateral load. The effect of soil variability is studied by coupling a numerical soil pile model with the stochastic subsoil model. Abaqus is used for numerical modelling white MATLAB is used for stochastic modelling. Before performing the probabilistic analysis, deterministic analysis is carried out to determine the standard behavior of a monopile with soil stiffness corresponding to a constant density. The probabilistic study is carried out by modeling the random variable using three models, i.e. deterministic, single random variable and generalized random field model. Initially the analysis is performed incorporating geometric soil stiffness as a single random variable I. (autocorrelation length corresponding to infinity) and then investigation was carried out by modeling geometric soil stiffness as generalized random field. Probabilistic investigation is carried out for pile head displacement, pile head rotation, pile bending moment and pile deflection. 1-lere, the random field is generated using matrix decomposition method. Finally a study is carried out to find out whether a critical autocorrelation length is exist or not and how the distribution of considered output variable has to change if there any such critical autocorrelation length exists