Near-fault ground motions containing high energy and large amplitude velocity pulses may cause severe damage to structures. The most widely used intensity measure (IM) is the elastic spectral acceleration at the fundamental period of the structure (Sa(T 1 )); however, Sa(T 1 ) is not a sufficient IM with respect to the effects of the pulse-like ground motions on structural response. For nearfault ground motions, including pulse-like and non-pulse-like time histories, we propose a vector-valued IM consisting of a new IM called instantaneous power (IP(T 1 )) and the Sa(T 1 ). The IP(T 1 ) is defined as the maximum power of the bandpass-filtered velocity time series over a time interval of 0.5T 1 . The IP(T 1 ) is period-dependent because the velocity time series is filtered over a period range (0.2T 1 -3T 1 ). This allows the IP(T 1 ) to represent the power of the near-fault ground motions relevant to the response of the structure. Using two-dimensional models of the 2-and 9-story steel-frame buildings, we show that the proposed [Sa(T 1 ), IP(T 1 )] vector IM gives more accurate estimates of the maximum inter-story drift and collapse capacity responses from near-fault ground motions than using the vector IM consisting of the Sa(T 1 ), the presence of the velocity pulse, and the period of the velocity pulse. Moreover, for the structures considered, for a given Sa(T 1 ), the IP(T 1 ) is more strongly correlated with structural damage from near-fault ground motions than the combination of the velocity pulse and pulse period. K E Y W O R D Snear-fault ground motions, pulse, record selection, vector-valued intensity measure

The velocity pulse in near-fault ground motions has been used as a key characteristic of damaging ground motions. Characterization of the velocity pulse involves three parameters: presence of the pulse, period of the pulse, and amplitude of the pulse. The basic concept behind the velocity pulse is that a large amount of seismic energy is packed into a short time, leading to larger demands on the structure. An intensity measure for near-fault ground motions, which is a direct measure of the amount of energy arriving in short time, called instantaneous power (IP (T1)), is defined as the maximum power of the bandpass-filtered velocity time series measured over a time interval of 0.5T1, in which T1 is the fundamental period of the structure. The records are bandpass filtered in the period band (0.2T1−3T1) to remove the frequencies that are not expected to excite the structure. Zengin and Abrahamson (2020) showed that the drift is better correlated with the IP (T1) than with the velocity pulse parameters for records scaled to the same spectral acceleration at T1. A conditional ground-motion model (GMM) for the IP is developed based on the 5%-damped spectral acceleration at T1, the earthquake magnitude, and the rupture distance. This conditional GMM can be used for record selection for near-fault ground motions that captures the key features of velocity pulses and can lead to a better representation of the median and variability of the maximum interstory drift. The conditional GMM can also be used in a vector hazard analysis for spectral acceleration (T1) and IP (T1) that can be used for more accurate estimation of drift hazard and seismic risk.

The debate over the significance of ground-motion duration is long-standing and the literature on the influence of duration on structural response is extensive. Decoupling of the duration from other characteristics of the ground motion is crucial for accurate quantification of its effect on structural responses. This article presents a new methodology that isolates the duration from the amplitude, frequency content, and rate of energy build-up of the ground motion. This is achieved by selecting short- and long-duration record pairs that are equated on the basis of spectral shape and the slope of the Husid plot. The use of the initial rate of Arias Intensity as a control parameter is novel in the literature. The proposed approach enables the examination of the sole effect of the duration on structural responses of 2-story and 9-story steel frame buildings. We find that the maximum interstory-drift ratios are not generally sensitive to the duration differences between short- and long-duration record sets, whereas the cumulative damage parameters (i.e. dissipated hysteretic energy and Modified Park–Ang Damage Index) of the buildings considered in this study are affected by duration. Finally, we extend the study to collapse limit states and find that duration has a small effect on structural collapse capacity, after controlling three key ground-motion parameters.

Selection of ground motions for use in nonlinear dynamic analysis is one of the most critical steps for both code-based design and probabilistic seismic risk assessment of structures. In practice, time-domain spectrum-matching methods, which add wavelet functions to an initial acceleration time series, have been widely used to obtain a record whose response spectrum closely matches the desired target spectrum. Although the spectral shape is known to be a good predictor of structural response, it does not represent the critical aspects of the velocity pulses, such as pulse amplitude and pulse period for near-fault ground motions. The Instantaneous Power ( IP( T1)), defined as the maximum rate of change of energy of the bandpass-filtered velocity time series over a short time interval given by half of the structural period, has been shown to be an effective alternative parameter to capture effects of the presence of a velocity pulse and the pulse period in near-fault record selection. We introduce an approach to modify time series so as to simultaneously match a target response spectrum and IP spectrum over a specified period interval. We demonstrate that the records modified using the proposed approach produce results comparable to those obtained using unscaled records, and prevent potential bias in structural response, relative to results when matching is performed without consideration of IP.

No abstract