Structural safety, technical reliability, and optimal redundancy allocation in structural systems

Structural Reliability theory provides the means for managing uncertainties in modern codified structural design. Yet, historical records show that most structural collapses occur due to uncertainties not explicitly considered in reliability calculations, such as gross errors in design and execution, unanticipated loading, poor understanding of phenomena, new technology, etc. Recently, it has been shown that neglecting the above 'non-technical' factors in reliability-based optimisation leads to non-redundant optimal designs, with negative impacts on structural robustness. Surprisingly, the observed collapse frequency of structures was recently found to be smaller than calculated failure probabilities. This reveals that structural reliability calculations are conservative with respect to technical factors, while grossly overlooking non-technical factors. Population wise, the two independent errors may be partially offsetting each other. The discussion herein covers significant 'open issues' to be addressed by the Structural Safety community, before risk-based performance metrics can be incorporated into structural design practice.