It is proposed that the provision of adequate safety in structures depends on the satisfaction of three independent requirements: adequate safety factors, adequate control of the design and construction process, and adequate robustness. Only the first of these is currently susceptible to quantitative treatment. A proposal for the more rigorous, quantitative treatment of robustness is developed. It is suggested that most failures result from errors and that the risks to health and safety from failures result largely from insuficient robustness. The risks from failure may not be greatly
changed by changes in safety factor.
Professor A.W. Beeby