SMIP 2015 Seminar

Ground-motion prediction equations (GMPEs) typically give amplitudes of ground motion as a function of distance from earthquakes of a particular magnitude. They are the foundations on which the seismic hazard maps used in building codes are built, they provide motions for the design of critical structures, and they and the databases used in their derivation conveniently summarize a large amount of information about the seismic waves radiated from earthquakes. The development of GMPEs requires knowledge of many aspects of seismology, including data acquisition, data processing, source physics, the determination of crustal structure, the effects of that structure on the propagation of seismic waves, the measurement and characterization of the geotechnical properties near the Earth’s surface, and the nonlinear response of soils to strong shaking. Generally, GMPEs are developed for three regions: active crustal regions (ACR), stable continental regions (SCR), and subduction zones (SZ). Most GMPEs in ACRs and SZs are based on empirical analysis of observed ground motion, while those in data-poor areas such as SCRs rely primarily on simulations of ground shaking. As data sets increase and theoretical simulations improve, previous GMPEs are revised and new ones are proposed. As a result, many hundreds of GMPEs have been published, and more are on their way. As an example of the current state-of-practice for GMPEs in ACRs, I will discuss a recent multi-year project undertaken by the Pacific Earthquake Engineering Research Center (PEER). The future is bound to bring more data, but most of these data will be for magnitudes and distances where present GMPEs are well constrained by existing data, at least in ACRs. Significant gaps will continue to exist in our knowledge of ground shaking in certain distance and magnitude ranges for ACRs and for SCRs in general. For this reason, combinations of simulated and observed motions will be used to create future GMPEs.

We investigate the ability of 1-D ground response simulations to match observed levels of site amplification from California vertical arrays. Using 10 vertical arrays, we find simulations to best match data using a VS-based damping model from the literature. We find a higher percentage of California sites, as compared to KiK-net sites from Japan, to have a reasonable match of empirical and theoretical transfer function shapes. The empirical transfer functions also have a greater degree of event-to-event consistency than has been found previously in Japan. Cases with poor matches highlight that 1-D simulations can fail to accurately model site response.

This paper summarizes our findings from a previous study on the effects of topography on ground motions. We analyzed the NGA-West2 dataset and proposed a model to predict topographic effects at a site. The model proposes modification factors for the expected amplifications or de-amplifications at a site, as a function of the relative elevation value at the site. As a part of this study, we also computed 2D topographic amplification at ground motion stations from simplistic numerical analyses and found that the logarithm of amplifications at stations, averaged over multiple orientations, were highly correlated with relative elevation value at the stations.

The data recorded from seismically instrumented buildings over the past approximately 40 years is used to indirectly evaluate the ASCE/SEI 7 direction of loading provisions. Direction of loading provisions require combining the maximum response in one direction, with a percentage of the maximum response in the orthogonal direction. In ASCE/SEI 7, a value of 30% is used for response in the orthogonal direction. This research shows that, for a wide range of conditions and assumptions, building response exceeds combinations with maximum response in one direction and only 30% of the maximum in the other direction. Alternative combination values are provided that better bound the recorded data.

When the design response spectra for the two horizontal components of motion are of equal intensity the expectation of the peak seismic demand is given by the SRSS combination of the uni-directional responses. The variance of this response depends, however, on the probability densities of the cross-correlations and thus the upper bounds on the ratios of response to design level in elements that have important contributions from both loading directions are larger than in those dominated by uni-directional motion. This paper operates on the premise that it is desirable to equalize these bounds and attempts to do so by specifying the cross-correlations at a probability of exceedance that attains the objective. Derivation of the pdf of the cross correlation is required and is found that for closely spaced frequencies it is well approximated by the pdf of the unlagged coherency times the standard correlation coefficient. The pdf of the unlagged coherency, in turn, is shown to be well approximated by a shifted and scaled beta distribution with parameters . It is shown that when the results obtained are translated into the 100%+X% combinations rule the consistent value of X is sixty.

A new procedure for rapid post-earthquake safety evaluation of bridges is being developed, using existing strong motion records, PGA data immediately available following an earthquake, and fragility databases, to assist responsible parties in making timely, informed decisions regarding post-earthquake bridge closures. The New Carquinez Bridge was selected to demonstrate the procedure. This paper provides a procedure overview and its application to safety evaluation of a bridge following an earthquake event, and the development to date of this process, including earthquake scenario selection and generation of ground motions for nonlinear time history analyses of the bridge to establish component fragility data.

Sixty-four buildings, with a total of 693 distinct seismic event and building direction records, are selected from the CSMIP database to identify modal quantities (i.e., natural periods and equivalent viscous damping ratios). The selected buildings include steel and reinforced concrete moment resisting frames (i.e., SMRF, and RCMRF), and reinforced concrete walls (RCW). Variation of modal quantities to structural system types, building height, amplitude of excitation, and system identification technique is studied. Results, tentatively, show median values for modal damping ratio are %2.7, %3.1, and %3.6 for RCW, RCMRF, and SMRF structures, with COVs in the order of 50%.

Substructure method is generally used in engineering practice to take Soil-Structure Interaction (SSI) effects into account in seismic design. In this method, soil is modeled using discrete spring elements—Impedance Functions (IF)—that are attached to the superstructure; and the Foundation Input Motions (FIMs) are applied at the remote ends of these springs. While the application of the substructure method is simple and it is computational costs is low, the determination of FIMs and the IFs are generally quite challenging. In the present study, we present a new approach to identify IFs and FIMs from response signals recorded during earthquakes. To do so, we use a flexible-based Timoshenko beam model is to represent the structure and its soil-foundation system and updated the parameters of this model such that its responses match real-life data. The impedance functions of a large set of instrumented buildings are identified using this novel method and compared against various analytical solutions. Additionally, a computer program named CSMIP-CIT is developed that automatically extracts data for selected buildings in the CSMIP database and applies the method developed in this study.