Assessing and achieving structural safety

Structures seldom fail, so it is reasonable to conclude that they are ‘safe’ and that this state was attained in the processes of designing and building them. The process of achieving structural safety and the process of assessing whether structural safety has actually been achieved for a given structure are, however, two very different matters. This is the essence of this paper and the issues are discussed in terms of the four related topics of responsibility, failure, uncertainty and decision. The first deals with the domain of responsibility of structural engineers: what it is and what it should be and how this relates to structural failure. Four sources of failure are distinguished: technical, process errors, technical ignorance and various non-technical matters. The relative contributions of these are then discussed. Recent codes of practice have focused narrowly on technical uncertainty and a notional probability of failure, with little attempt to incorporate non-technical matters and the ontological uncertainty, which is shown to be the origin of most failures. Finally, the nature of decision making is discussed for both code development and structural design and it is concluded that in both cases the decision process is one of satisficing rather than optimising. The whole process is less rational than is generally supposed and requires a broader view in the outlook and training of structural engineers.