Statistical accuracy of finite element method in predicting horizontal displacement of monopiles for offshore wind turbines

Abstract

Large-diameter steel pipe pile foundations are widely used to support offshore wind turbines subjected to horizontal loads during operation. The pile horizontal displacement must be restricted within a certain value to ensure its safety. Therefore, an accurate prediction of pile horizontal displacement is of great significance. The finite element method (FEM) has been prevailing in such prediction, with its accuracy remaining unquantified. This study compiles a large database containing 959 pile horizontal displacement measurements from 14 offshore wind turbines. Numerical models are then built using FEM to predict horizontal displacements of these piles. A bias defined as the ratio of measured to predicted horizontal displacement is used to quantify the accuracy of the FEM. Results showed that the FEM is moderately risky as it underestimates the pile horizontal displacement by about 40 % on average. The dispersion in prediction accuracy is about 40 % ranked as moderately dispersive. An empirical constant of 1.41 is introduced to the predicted displacement for model calibration, making the prediction unbiased on average. The probability density functions for the biases are characterized as 2-order Gaussian functions. Last, analysis of a monopile from a real project is presented to highlight the significance of the calibrated finite element model.