The final publication is available at Springer via http://dx.doi.org/10.1007/s10694-016-0602-6

Numerical simulations have been widely applied, for the determination of the resistance of steel structural elements, when experimental analysis are not possible (due to cost or size limitations) or when parametric studies with high number of variables are needed. However, the numerical models must be properly validated with experimental tests in order to deliver reliable studies. With the purpose of studying the behaviour of stainless steel plate girders in fire situation, a total of 34 experimental tests from the literature have been numerically modelled. The tested girders had different configurations: rigid and non-rigid end posts, 2 and 4 panels, and transversal and longitudinal stiffeners were considered. Comparative analyses between those experimental and numerical results have been done. Good approximations to the experimental results at normal temperatures have been achieved with differences on average lower than 5%. Afterwards, the developed numerical model has been used to perform a sensitivity analysis on the influence of the initial geometric imperfections at both normal and elevated temperatures, considering different values for its maximum amplitudes, concluding that 10% of the web thickness is an appropriate value for the maximum amplitude of the geometric imperfections when modelling experimental tests. The effect of the residual stresses has also been analysed, being obtained differences lower than 2%. Finally, comparisons between the numerical results and the Eurocode 3 design procedures have been performed considering different uniform elevated temperatures.

Peer Reviewed