Accelerogram-scaling procedures for near-fault motions

Technical Abstract

This study tackles the problem that the procedures of the New Zealand standard NZS1170.5:2004 for earthquake actions on structures lead to different seismic demands on structures from near-fault earthquake motions depending on whether the model response spectrum (MRS) or the numerical-integration time-history (NITH) methods of analysis are used. The requirements appear to penalize the NITH method for which the estimated responses are generally greater. This conflicts with the desire to encourage designers to use the NITH method, because it has features that are likely to provide more realistic representations of structural response than other analysis methods.

Two modifications to the NZS1170 procedures have been identified to overcome these problems. These modifications relate to how to scale the records of earthquake motions (“accelerograms”) they are used as input to NITH analyses, and how to combine the results from the analyses for individual records.

The first medication is to increase the number of records required from three to seven or more. These are selected and scaled according to the current NZS1170 procedures, but the maximum inter-story drifts are averaged across the total number of records rather than taken as the largest of the individual drifts. The second modification, to be used in conjunction with the first, is to increase the upper period limit used for determining the record scale factors to the effective period Teff associated with the maximum response, which could be up to about double the current limit. This modification has only slight effect o the average maximum response, but generally changes the more extreme responses considerably, leading to reduced scatter across the records.

The study has been successful in identifying simple modifications to the NZS1170 procedures for scaling earthquake records and combining NITH analysis results from different records that in general produce similar or reduced seismic demand estimates for NITH analyses compared to MRS methods.