Amplitude scaling is a common approach to modify recorded ground motions to achieve a desired intensity level. The possible bias introduced by scaling the amplitude of ground motions when using the first-mode spectral ordinate as the intensity measure is evaluated using intensity-based analyses. This study evaluates whether upward scaling introduces bias in lateral displacement demands, but more importantly, in the probability of collapse. The latter, which is of utmost importance, has received little attention in previous studies. Analyses were conducted using degrading single-degree-of-freedom and multiple-degree-offreedom systems with different fundamental periods of vibration and normalized strengths subjected to different sets of recorded accelerograms requiring different scale factors to reach a target intensity. The results demonstrate that this type of amplitude scaling introduces a bias in which lateral displacement demands and collapse estimates are increasingly overestimated with an increasing scale factor and that the bias is strongly dependent on the period and lateral strength of the system. Furthermore, the bias is considerably larger in collapse risk estimates.

Summary Although for many years it was thought that amplitude scaling of acceleration time series to reach a target intensity did not introduce any bias in the results of nonlinear response history analyses, recent studies have showed that scaling can lead to an overestimation of deformation demands with increasing scale factors. Some studies have suggested that the bias can be explained by differences in spectral shape between the response spectra of unscaled and scaled records. On the basis of these studies, some record selection procedures assume that if records are selected using spectral‐shape‐matching procedures, amplitude scaling does not induce any bias on the structural response. This study evaluates if bias is introduced on lateral displacement demands and seismic collapse risk estimates even when spectral shape is carefully taken into consideration when selecting ground motions. Several single‐degree‐of‐freedom and multiple‐degree‐of‐freedom systems are analyzed when subjected to unscaled and scaled ground motions selected to approximately match the mean and the variance of the conditional spectrum at the target level of intensity. Results show that an explicit consideration of spectral shape is not enough to avoid a systematic overestimation of lateral displacement demands and collapse probabilities as the scale factor increases. Moreover, the bias is observed in practically all cases for systems with strength degradation and it increases with decreasing period and decreasing lateral strength relative to the strength required to remain elastic. Key reasons behind the bias are presented by evaluating input energy, causal parameters, and damaging pulse distributions in unscaled and scaled ground motion sets.

Summary This study evaluates the performance of a new intensity measure, referred to as filtered incremental velocity FIV3, which is computed using time‐domain features extracted from an acceleration time series and is aimed at the evaluation of structural collapse. This novel approach focuses on the area under a small number of acceleration pulses in the ground motion instead of focusing on the peak response of linear elastic oscillators as in many recently proposed measures of ground motion intensity. FIV3 is developed based on previous research that has highlighted the close relation between the incremental velocity of a ground motion and its potential to induce large inelastic incursions on structures. However, unlike the original definition of incremental velocity which provides a single level of intensity for a ground motion, this new intensity measure is period‐dependent and computed as the sum of the three largest incremental velocities obtained from a low‐pass filtered ground acceleration time series. Efficiency and sufficiency with respect to several ground motion parameters such as magnitude, source‐to‐site‐distance, spectral shape, scale factor, and duration are carefully evaluated and compared against those computed with some traditional and recently proposed intensity measures using collapse results from a four‐story reinforced concrete frame. Results indicate that FIV3 leads to lower variability of collapse estimates and therefore higher efficiency as well as high sufficiency compared with those of other ground motion intensity parameters indicating that this new intensity measure is a promising parameter for structural collapse risk assessment.

An intraslab normal-faulting earthquake struck the central region of Mexico on 19 September 2017, leading to the collapse of 44 buildings in Mexico City. After the earthquake, the authors collected information in situ and through social media about the collapsed buildings, which was statistically processed to identify the causes of their collapse. This article presents the main collapse statistics, which revealed that 64% of the collapsed buildings had between 1 and 5 stories, 61% had a seismic-force-resisting system based on reinforced concrete columns with flat slabs, 57% experienced a soft-story mechanism, 91% were built before 1985, 43% were located at the corner blocks, and 10% exhibited pounding with neighboring buildings. The spatial distribution of the collapsed buildings and the recorded ground motion features suggest that short- and medium-period buildings having well-known vulnerabilities were particularly prone to collapse under amplified high-frequency seismic waves typical of intraslab normal-faulting earthquakes, such as the 2017 Puebla–Morelos earthquake.

This paper presents a ground motion prediction model (GMPM) for estimating medians and standard deviations of the random horizontal component of the peak inelastic displacement of 5% damped single-degree-of-freedom (SDOF) systems, with bilinear hysteretic behavior and 3% postelastic stiffness ratio, directly as a function of the earthquake magnitude and the distance to the source. The equations were developed using a mixed effects model, with 1,662 recorded ground motions from 63 seismic events. In the proposed model, the median is computed as a function of the vibration period and the normalized strength of the system, as well as the event magnitude and the Joyner-Boore distance to the source. The standard deviation of the model is computed as a function of the vibration period and the normalized strength of the system. The proposed model has the advantage of not requiring an auxiliary elastic GMPM to predict the median and dispersion of peak inelastic displacement.