The interaction of local buckling and stability loss of a thin-walled column under compression

The influence of input random quantities (imperfections) on the output random quantity (load-carrying capacity) variability is studied by the sensitivity analysis. The influence of the initial curvature shape of a column is discussed. The initial curvature is often supposed to have the sine function shape. The Euler method based on proportional loading in combination with the Newton-Raphson method was used. The girder symmetry and that of loading were made use of. The realizations of input imperfections is simulated using the Latin Hypercube method (improved Monte Carlo method) that makes it possible to simulate the realization of the data from the known statistics of the input values in a similar way as they could be obtained by direct measurements. The method LHS gives more accurate estimates of mean value and standard deviation for a smaller number of runs.