Evaluation of structural performance in the immediate aftermath of an earthquake: a case study of the 2011 Christchurch earthquake. (1st January 2012)

• Record Type: Journal Article
• Authors: Stafford, Peter J.
• Is Part Of: International journal of forensic engineering. Volume 1:Number 1(2012)
• Page Start: 58
• Page End: 77
• Publication Date: 2012-01-01
• Subjects: structural performance -- ground motion -- aftershock -- spatial correlation -- earthquake loss estimation -- Christchurch earthquake; Forensic engineering -- Periodicals; 624.176
• Languages: English
• ISSNs: 1744-9944

Abstract

Immediately following an earthquake event, there is pressure for experts to provide information regarding what has just happened and why. What is reported in the media can influence the confidence within the society regarding the engineering profession, and may also influence policy that is established as a result of the event. Using the earthquake that occurred in Christchurch, New Zealand, in early 2011, this paper looks to investigate what sort of information we really have in the immediate aftermath of an earthquake and what inferences can be made by experts engaged in dialogue with the media and other interested parties. The paper demonstrates that very little in the way of explanation can be given for the observed performance of buildings because of the large uncertainties associated with trying to reconstruct the input ground-motion field responsible for the observed damage. To demonstrate this point, established methods for discretising random fields are presented within a new framework for re-creating spatial fields of ground-motions for use in earthquake loss estimation. The results show that even when high-resolution analyses are conducted, the variance of the discretised field remains large owing to the overwhelming contribution of the aleatory variability of ground motions. Any assessment of structural performance is necessarily bound by the uncertainties in the input motions, and this study demonstrates that, in the best-case scenario, the input motions will only be known to within 20% of their expected values 70% of the time.