Influence of various boundary conditions on local buckling of steel tubular piles

Abstract

The main focus of this thesis is the influence of different real boundary conditions on the failure mechanism of local buckling of steel tubular piles. The research was started with the theoretical method of describing boundary condition by specifying the degrees of freedom per boundary condition. The method makes use of the radial, meridional and rotational degrees of freedom. This method is also used in the current design code for steel shell structures, the NEN-EN 1993-1-6 \cite{NEN-EN_1993-1-6}. By varying the amount and types of constrained degrees of freedom, eight possible boundary conditions can be generated. Looking to the practical applications, a ninth variant with a non-homogeneous boundary completes a list of nine. From the nine boundary conditions, eight conditions are selected to be included in a FEM model study. The current design code describes only five of the nine conditions, what allows for a comparison of the known and unknown conditions with the prescribed rules. Two types of parameters that influence the local buckling failure mechanism are material and geometrical imperfections. In the study to geometrical imperfections, initial ovalisation, accidental eccentricity and dimples are considered. The effect on the design yield stress of the different imperfections is determined following the calculation methods described by the current design code. The calculated reduction of the design yield stress by specific boundary condition and an increasing imperfection is compared to results of ABAQUS FEM model tests. Residual stresses are investigated on the side of material imperfections. The used residual stresses are validated with a simple FEM ring model test that shows deformations due to the introduced residual stresses. The observed displacements could be compared with measured data of real experiments. A range of 36 different tubes is selected to investigate the influences of the eight different boundary conditions on the buckling failure behaviour. This test range covers a wide range of tubes with cross-section class 3 and 4 known for their sensitivity for local buckling. A base model is generated for all the 288 model tests, 36 tubes with all 8 boundary conditions per tube. The results from all tests are displayed in normalized moment-curvature diagrams. Because of a lack of test data of real experiments with boundary conditions involved, a validation with an analytical method is used that describes a linear and bi-linear part of the moment-curvature relation. Finally the results of the FEM model tests are analyzed per parameter but also the overall behaviour is discussed. From the model results could be concluded that residual stresses hardly influence the moment capacity but causes an increase in the curvature of most boundary condition. Geometrical imperfections cause a decreasing moment capacity and an increasing critical curvature whit an increasing imperfection. The trend of the influence of dimple imperfections is decreasing slightly faster compared to the trend observed from the code calculations. A safety of approximately 20\% is observed between the failure stresses of the test results and the design stresses from the code. With an increasing diameter over thickness (D/t) ratio or yield stress, the moment capacity decreases and the critical curvature increases. The relative influence of the D/t ratio within the tested range is larger compared to the yield stress. To investigate the influence of the different boundary conditions, the effect of all parameters is investigated per boundary condition. Following the test results, all boundary conditions except BC4F show the same influence on the maximum moment and corresponding curvature. Boundary condition BC4F shows a moment resistance reduction of 14\% and a critical curvature reduction of 26\% compared to the other conditions. Small observed differences in the post buckling behaviour between the other different boundary conditions are reported to complete this study.