| dc.description.abstract | This thesis studies the behaviour of axially loaded circular hollow cylindrical columns in aluminium and
is written as a part of the project Design of power pylons in aluminium customised for automatic
production . The behaviour of columns exposed to axial loading, was studied in the laboratory. The columns studied, had a length of 3450 mm and 4560 mm with three different compositions of material and cross-section. The Aluminium alloys used were, 6060-T6 and 6082-T6. A full testing procedure was developed and is presented in the appendix.
Material tests were conducted on a pre-notched specimen of the Aluminium alloy 6082-T6. An Edge Tracing method was used along with Digital Image Correlation, to retrieve the material data. The edge tracing method was able to retrieve the curvature of the notched specimen, and this was used in the Bridgman correction. Numerical simulations of the material test were run, and the results showed that a direct calibration was not capable of reproducing the correct engineering stress-strain curve from the material test. A reverse engineering approach was used to calibrate different material models, and a pressure sensitivity of the material was observed. No significant difference in the material specimens containing extrusion welds was detected.
The global buckling mode was initiated by an initial horizontal load at the mid-span of the columns. The results from the tests displayed a global buckling mode for every column, with a maximum capacity close to or above, the Euler load. Every column displayed a sudden buckling, despite having an initial deflection of L/250. The horizontal load did not follow the column as the mid-span horizontal deflection increased. This loading sequence was recreated in the numerical simulations, and the same buckling behaviour was displayed.
The choices made when modelling and simulating in the finite element method software Abaqus were presented and evaluated. A parameter study was conducted on the element type and element size, to find the best combination for the numerical models used to recreate the laboratory tests. Different material models calibrated from tension- and compression tests, were established for the aluminium alloy 6082-T6. These models were used in numerical simulations of columns of length L=200 mm, and compared to the laboratory results from 2016. These simulations resulted in a underestimation of the capacity of 6 percent. Global initial stress-free imperfections and imperfections initiated by a horizontal load, were further studied in numerical simulations, and a relationship between the two imperfection types were established for two different column lengths. Buckling curves for all cross-sections were established and compared to calculations from Eurocode 9. Characteristic capacity from Eurocode 9, using calibrated material parameters, was equal to the capacity from numerical simulations for lengths above 500 mm for the material 6060-T6. The same comparison showed a ~10 percent lower capacity for the alloy 6082-T6. | en |
