The Mechanics of Reliability Performance Evaluation

This paper discusses the issues with using the reliability index as a means for structural performance evaluation. The structural strength performance of components (members and connections) is discussed using reliability principles to manage the uncertainties of the design parameters. Three issues are discussed: (i) Reliability index as a performance parameter: This can only be established after the more basic parameters of actions and resistances are specified. The reliability index is sensitive to the assumptions made in establishing the action and resistance models. The specification of reliability indices must include the models and their assumptions to enable verification. (ii) Setting performance targets: Different targets are feasible depending on the objectives. Code calibration is a practical way for setting performance targets. Reliability differentiation should be derived from evidence-based analysis and should not restrict the freedom of the designers to innovate. A scheme is proposed where reliability targets are set as a function of the types of action and resistance only. (iii) Performance of design by prototype testing: A simple scheme is proposed in which the resistance is specified at its 5-percentile value, and the capacity factor ϕ is expressed as ϕ  = (1 − kV), where V is the coefficient of variation of the applicable population and k is a calibration factor between 0 and 1. The resulting reliability indices are compatible with current practice. This scheme is suitable for use with a limited number of tests.