Selection and modification of ground motion records using a weighted scaling method based on the Newmark-Hall target spectrum

For the problem of the selection and modification of ground motions in nonlinear time-history analysis, a weighted scaling method based on the Newmark-Hall target spectrum (WNM) was developed in this study. The average Newmark-Hall spectra of the three ground motion sets developed in the SAC Steel Project are defined as the target spectra, which represent 50 %, 10 %, and 2 % probabilities exceeded in 50 years, respectively. The matching errors (SSE) and scaling factors (SF) were calculated by the weighted least-squares method to match with the Newmark-Hall target spectrum. The 9-and 20-story steel-moment-resisting frame structures were used as examples. Their mean structural responses obtained by the WNM were compared with the results of the method using the Newmark-Hall target spectrum without weight in scaling records (NM). Both the WNM and WN can control the relative errors of the prediction of the mean structural response within +/- 20 % as compared with the benchmark results when 7 or 10 records are selected, if the SSE and SF used to scale and rank a record are both obtained by the same scaling method. Both the WNM and NM have better capacity for reducing the variability of structural responses than the acceleration target spectrum method. Especially for the 20-story structure with long fundamental period in regions with high earthquake intensity, this advantage of the WNM is more prominent than that of the WN. In addition, if the record has a large forward and back displacement peak, the steel-moment-resisting frame structures would produce a large offset at the low half of the building, and the panel zones are critical components for the structural deformation.