Approximate Analysis and Economics of Structures

The influence of approximate methods of analysis on the cost and reliability of structures is examined. It is first assumed that the design code is optimum, and curves of total expected cost, including the cost of failure, are determined as functions of a common code parameter (proportional to the nominal safety factors), using second-moment reliability concepts. The approximation is represented by a random variable factor on the nominal strength in the design inequalities. Design for constant reliability and optimum design are studied as functions of the mean and coefficient of variation of this factor. These two statistics are next estimated for a few common methods of analysis. It is concluded that constant reliability design gives results that are close but not identical to optimum design as the coefficient of variation of the analysis method is varied. Further, if this coefficient is less than approx 0.12, the method is effectively exact in the sense that the total expected cost is not in excess of 3% above this cost when using exact analysis.