Conventional design methodology for the earthquake-resistant structures is based on the concept of ensuring 'no collapse' during the most severe earthquake event. This methodology does not envisage the possibility of continuous damage accumulation during several not-so-severe earthquake events, as may be the case in the areas of moderate to high seismicity, particularly when it is economically infeasible to carry out repairs after damaging events. As a result, the structure may collapse or may necessitate large scale repairs much before the design life of the structure is over. This study considers the use of design force ratio (DFR) spectrum for taking an informed decision on the extent to which yield strength levels should be raised to avoid such a scenario. DFR spectrum gives the ratios by which the yield strength levels of single-degree-of-freedom oscillators of different initial periods should be increased in order to limit the total damage caused by all earthquake events during the lifetime to a specified level. The DFR spectra are compared for three different seismicity models in case of elasto-plastic oscillators: one corresponding to the exponential distribution for return periods of large events and the other two corresponding to the lognormal and Weibull distributions. It is shown through numerical study for a hypothetical seismic region that the use of simple exponential model may be acceptable only for small values of the seismic gap length. For moderately large to large seismic gap lengths, it may be conservative to use the lognormal model, while the Weibull model may be assumed for very large seismic gap lengths.

Seismic hazard maps have been prepared for Northeast India based on the uniform hazard response spectra for absolute acceleration at stiff sites. An approach that is free from regionalizing the seismotectonic sources has been proposed for performing the hazard analysis. Also, a new attenuation model for pseudo-spectral velocity scaling has been developed by using 261 recorded accelerograms in Northeast India. In the present study, the entire area of Northeast India has been divided into 0.1° grid size, and the hazard level has been assessed for each node of this grid by considering the seismicity within a 300-km radius around the node. Using the past earthquake data, the seismicity for the area around each node has been evaluated by defining a and b values of the Gutenberg-Richter recurrence relationship, while accounting for the incompleteness of the earthquake catalogue. To consider the spatial distribution of seismicity around each node, a spatially smoothed probability distribution function of the observed epicentral distances has been used. Uniform hazard contours for pseudo-spectral acceleration as the hazard parameter have been obtained for an exposure time of 100 years and for 50% confidence level at different natural periods for both horizontal and vertical components of ground motion. The trends reflected by these contours are broadly consistent with the major seismotectonic features in the region.

SUMMARYIn damage-based seismic design it is desirable to account for the ability of aftershocks to cause further damage to an already damaged structure due to the main shock. Availability of recorded or simulated aftershock accelerograms is a critical component in the non-linear time-history analyses required for this purpose, and simulation of realistic accelerograms is therefore going to be the need of the profession for a long time to come. This paper attempts wavelet-based simulation of aftershock accelerograms for two scenarios. In the first scenario, recorded main shock and aftershock accelerograms are available along with the pseudo-spectral acceleration (PSA) spectrum of the anticipated main shock motion, and an accelerogram has been simulated for the anticipated aftershock motion such that it incorporates temporal features of the recorded aftershock accelerogram. In the second scenario, a recorded main shock accelerogram is available along with the PSA spectrum of the anticipated main shock motion and PSA spectrum and strong motion duration of the anticipated aftershock motion. Here, the accelerogram for the anticipated aftershock motion has been simulated assuming that temporal features of the main shock accelerogram are replicated in the aftershock accelerograms at the same site. The proposed algorithms have been illustrated with the help of the main shock and aftershock accelerograms recorded for the 1999 Chi-Chi earthquake. It has been shown that the proposed algorithm for the second scenario leads to useful results even when the main shock and aftershock accelerograms do not share the same temporal features, as long as strong motion duration of the anticipated aftershock motion is properly estimated.

No abstract

Summary Stochastic ground motion simulation techniques are becoming increasingly popular because of enhanced computation power enabling direct simulation of complex response quantities. Priestley process assumption is the most general approach for stochastic modeling of earthquake ground motion. However, a framework for multicomponent ground motion simulation using the general Priestley process assumption is not available. Multicomponent motions are useful especially when the correlation structure between them significantly influences the response. The present study proposes a framework for frequency‐dependent principal component analysis (PCA), which facilitates Priestley process–based simulation of multicomponent ground motions. The study focuses only on the frequency‐dependent PCA part, and the results show high dependency of the principal components/directions on the frequency bands of the signals. The present work also advocates that the frequency‐dependent PCA should be preferred to the conventional PCA as the former can address the issues related to the frequency‐independent uniform modulation associated with the latter.