Foreword to the Special Issue for the RINTC (The Implicit Seismic Risk of Code-Conforming Structures) Project

The need for a code in which the design objectives are expressed in terms of probability of failure of the structure has been around for about 50 years [e.g., Cornell, 1969]. Such a need has been accepted by the community, but the way these concepts have been implemented in modern codes is a compromise between a probabilistic approach and a deterministic format compatible with the technical background of practitioners; that is, the so-called load-resistance-factor-design or LRFD [Ellingwood, 2000]. In the LRFD approach, conservative percentiles of load and resistance distributions of structural elements are employed in verifications. The assumption is that, in such a way, a level of structural reliability is indirectly warranted. In seismic design, the design action results from probabilistic evaluations consistent with the probabilistic seismic hazard analysis [e.g., Cornell, 1968; McGuire, 2004]. The design ground motion (usually in the form of elastic response-spectrum ordinates) has a (rare) return period of exceedance at the construction site, which corresponds to some structural performance threshold that, if surpassed, determines failure; that is, the attainment of a limit-state [OES, 1995]. It is well known that this approach, although has represented a giant leap in earthquakeresistant design when introduced, does not allow to explicitly control the reliability of the designed structure, that is ultimately unknown. In fact, if design is such that the structure does not fail for intensities lower than the design intensity and fails with certainty for larger intensities, then the failure probability in any time interval coincides with the exceedance probability of the design intensity in the same time interval. However, because groundmotion intensity has limited explanatory power with respect to structural response, there is always a non-negligible probability that the structure fails for intensities below that used in design. Moreover, the design procedures are such that, typically, there is a level of structural resistance surplus even at intensities larger than design. This renders the relationship between the exceedance return period of the design groundmotion and the structural reliability less direct. Recently, some approaches have been proposed to improve design towards more controlled failure probabilities. These ongoing attempts are collectively known as risk-targeted ground motion [i.e., Luco et al., 2007]. The latter approaches are promising (although requiring some careful calibration) and possibly will find their way into the next generation of design codes. However, given the present situation, insights about the seismic structural reliability implied by design according to current standards are the starting point to plan improvements, especially with respect to what should be the targeted reliability for code-conforming structures. This was