RISK ASSESSMENT OF STRUCTURAL COMPONENTS

Abstract

Great accuiaey and speed have been obtained in the computation of stresses, deflection and buckling loads with the evolution of high-speed high-precision computers. This improved structural analysis would have permitted significant reduction in factor of safety and consequently greater economy, if the loads and strengths were exactly known. But, for almost all the problems, loads and strengths at best can only be described as statistical variables, and therefore safety of structure must also be quantified as a variable, using the algebra of statistics. It is obvious that improved computer-aided structural analysis and design should be accompanied by improved methods of predicting structural safety. Therefore, one can say that the rational criterion for the safety of a structure is its reliability or probability of survival. This in turn calls for the use of statistical frequency distributions for loads and strengths in determining the reliability. In past, the computational difficulty in calculating reliability were too complex to be comprehended manually. So, the only feasible way to provide structure safety was deterministic safety factor format. But at present, it is easy to compute reliability or to be more practical to correlate the safety factors to the reliability. The Load and Resistance Factor Design (LRFD) based on a given reliability index is a way to achieve this objective. A correlated safety however, relates to the failure of a single member under a single load condition and a constant safety factor cannot be used to insure that structure will have the same overall reliability. Thus the entire structure must be examined to predict its probability of survival. In practice, two classes of problems arise in reliability analysis. In the first, the reliability of an existing structure at the 'present time' is required to be assessed. In the second, the likely reliability of some future, or as yet incomplete, structure must be predicted. It is here when reliability has a role to play as a decision tool. Any reliability theory rests on a mathematical formulation of the limit state