Towards a more rational approach of the serviceability limit states design of industrial steel buildings

Serviceability design criteria derived from reliability analysis could improve the efficiency of single storey industrial steel building design because serviceability is often the limiting design consideration. However, the difficulty of obtaining the necessary data to form the basis of reliability modelling is demonstrated by the lack of experimental results documented in the literature. Due to the practical imperative of designing for serviceability performance, considerable expertise is available from experienced designers. This expertise was used as a source of data in order to construct a reliability model from which design criteria could be derived. A probabilistic method to extract subjective expertise was applied to obtain the parameters of the basic variables for reliability based serviceability performance functions. The method, which is analogous to the classical method of experimental design, makes provision for expert calibration. Resulting values for the basic variables are presented in the paper. A simple reliability analysis example illustrates the use of these variables.