Structural safety prediction of steel systems

Abstract

Eurocodes and ISO standards are based on the limit state concepts and on the design methods of partial safety and the load and resistance factors. These standards and JCSS directions require that structural systems and their components shall be suited for their working life in a safe and economical way. The semi-probabilistic limit state analysis cannot be reliable method in the design of system the safety indices depending on their structural configuration and redundancy, on deteriorating resistances of elements and time-dependent extreme situations. Despite a fairly developed theory of reliability, it is difficult to apply probabilistic approaches to structural safety analysis. The probabilistic computation of series and parallel systems involves the integration of multivariate distributions. The probabilistic analysis of systems is very difficult and sometimes not feasible due to a stochastic dependence of their failure modes. The main task of our researches is to separate dynamic (time-dependent) probabilistic models as autosystems for steel elements from static probabilistic models for structural systems consisting of m autosystems with highly correlated components and avoid sophisticated multistate integration formats in the analysis of probabilistic models. The survival probability of the random decreasing sequence of ranked cuts of particular elements (normal or oblique sections, deflections and so on) at extreme situations ti, ..., t,, may be modelled as random distributions and idealized as dynamic stochastic series autosystems [...].