Comparative review of possible alternatives for performing safety assessment of design rules for steel structures

Abstract

sem informação

Currently, safety assessment is not consistently considered throughout Eurocode 3, mainly due to lack of guidance and existing databanks containing information on distribution of relevant basic variables and steel properties. In addition, some rules are not thoroughly validated, leading to higher uncertainties. In establishing the partial safety factor - γM of a design procedure, scatter related to material and geometrical properties may be isolated from resistance model. Simplified safety assessment procedures are analysed and further tested in which the basic variables are assumed independent from each other, with basis on EN 1990 safety assessment procedure. In this dissertation, firstly, a theoretical overview is proposed in order to introduce the basic principles in the probability theory; secondly, the basic principles of design codes are presented, focusing on the current European design codes EN 1990 to EN 1999. The design methodologies are listed and their background is discussed, and furthermore focus is given to the procedures applicable to Eurocode 3. The safety assessment procedure, given in Annex D of EN 1990 is also presented and its theoretical background is clarified. Subsequently, an analytical review of existing methodologies for safety assessment of design rules for steel structures is carried out. The procedures are presented and their field of application is detailed. Furthermore, numerical validation of simplified procedures, based on assumed distributions of basic variables such as material and cross-section properties, is performed focusing on stability failure modes, in particular flexural buckling of columns. In addition, a numerical example is presented, which aims at further clarifying latter methodologies. Sensitivity analysis is performed in order to assess the influence of different basic variables. In this example, statistical distributions of the imperfections are also considered. Finally, the statistical dependence of basic variables is discussed, based on correlation between yield stress and plate thickness. In order to obtain a reasonable correlation coefficient, statistical data from real experiments is used. This dissertation is part of European research project SAFEBRICTILE RFS-PR-12103 – SEP nº 601596, where safety assessment procedure is developed for brittle to ductile failure modes. Therefore, it is considered a valuable contribution for achieving the project goals and further improvement of the design methodologies in the Eurocodes.