No abstract

As is common in literature on the subject, the Authors seem reluctant to define exactly what limit state design is or is not. Thus in Q 12 it is defined as 'design with partial factors (reliability theory, level l)', whereas Q 14 accepts that the idea of limit states is not new and is implicit in current design methods; this is confirmed in Q 42, which also concedes that the extent to which partial factors should be used in real day-to-day design is 'debatable'. Similarly, while the definition in Q 12 describes partial factor design as 'reliability theory, level 1 ' and probabilistic assumptions underlie the comparison of methods in 5 14 onwards, Q 40 states emphatically that 'statistics and probability are not prerequisites of limit state methodology'. The status of Appendix 1 on reliability theory is completely unclear: after lengthy calculations, the Authors then say these are not intended for direct numerical comparison and suggest calibrating against old codes instead. Vagueness of this sort seems to be a persistent weakness in papers about limit state theory, and it is unfortunate that the Authors did not feel able to define more clearly exactly what they are proposing.60. Issues which need to be considered are (a) the principle of subdividing the overall safety factor into partial factors (6) the adoption of statistical definitions as a basis for determining design (c) the practical effects of partial factor systems in real designs.loads, strengths, etc.Critical reveals that there are fundamental problems with limit state theory in all these areas, and this is beginning to be recognized in structural codes. Therefore, although the latest BSI codesz0*" still nominally adhere to the limit state format, they have been simplified (BS 5950 actually reverts to a single safety factor) and there has been a considerable retreat on the role proclaimed for probability theory.61. Firstly, the notion that the overall safety factor should really be considered as the mathematical product of a series of sub-factors+ach acting independently to cover safety in one area-and that failure occurs when peak variations in all these areas coincide, may appear superficially attractive but it is not necessarily logical. These separate sub-factors exist only on the engineer's calculation sheets