Intensity Measures for Seismic Response Prediction and associated Ground Motion Selection and Modification

Based on the research of the first generation of Performance-Based Earthquake engineering methodology (PBEE), Pacific Earthquake Engineering Research Centre (PEER) has developed the second generation procedure aiming at a more robust methodology of PBEE where the process is broken into several logical elements that can be studied and resolved in a rigorous and consistent manner. Due to the inherent uncertainty properties of earthquake occurrence, e.g. earthquake intensity, ground motion features, nonlinear dynamic behaviour of structures and etc., it allows that the new generation of PBEE methodology should be formalized within a probabilistic basis. To apply this methodology it requires an interactive effort of multi-disciplinary experts, such as geology engineers, seismologist, structural engineers, loss experts and etc. For structural engineers the most interest can be relevant to the selection and estimation of two parameters in PBEE, i.e. Intensity Measures (IM) and Engineering Demand Parameters (EDP), which reflect ground motion hazard and structural response in terms of deformations, accelerations, or other response quantities of the building excited by input ground motions. The EDPs are strongly dependent on the Intensity Measure (IM) used to perform the selection of ground motions. The IM as an intermediate variable connecting seismic analysis and structural analysis plays a very important role for structural engineers. An ideal IM should generally be of efficiency and sufficiency. The efficiency means it yields low dispersion of values of engineering demand parameter (EDP), while the sufficiency implies that EDP predicted with the candidate IM should be only dependent on this IM, not be conditionally dependent on properties of ground motions, like magnitude, source to site distance, fault mechanism etc. Therefore it implies the need of comparison among different intensity measures (IMs), in particular the comparison of dispersion of the EDP in relation to each IM. To this purpose a set of IMs 27 IMs, including those commonly adopted and some modified IMs based on the existed ones, are investigated in order to find optimum IMs for predicting various EDPs. Not only is IMs for predicting the structural response of widely studied fixed base buildings are investigated, but also IMs for predicting structural response of the base-isolated buildings are initiatively studied. 80 ordinary and 59 pulse-like ground motion records are used to run nonlinear dynamic analyses on the 4-storey and 6-storey frame concrete buildings and these buildings equipped with base-isolation system on them. The EDPs considered in this study include the Maximum Inter-storey Drift Ratio (MIDR), the Maximum Roof Drift Ratio (MRDR) and the Maximum Base Displacement (MBD, only for base-isolated buildings). Base on the results from this study some energy-based intensity measures have been shown to be good predictors of both structural and non-structural damage for base-isolated structures. However, they are not usually employed in probabilistic seismic demand analyses because of the lack of reliable Ground Motion Prediction Equations (GMPEs). In order to define seismic hazard and thus to calculate demand hazard curves it is essential, in fact, to establish a GMPE for the earthquake intensity. In the light of this need, new GMPEs are proposed here for the energy-based intensity measure, in particular elastic input energy equivalent velocity spectra i.e. VEIa and VEIr. The new GMPE is developed by taking advantage of the more comprehensive NGA database with more completed meta-data compiled in recent years. This prediction equation has a wider magnitude and distance applicable range, considers the effect of soil site by VS30 and fault mechanism, and etc. However when the energy-based IMs are used in the selection and modification of ground motions for structural dynamic analyses, the uniform hazard spectrum derived from their GMPEs only gives the marginal distribution without information of joint occurrence of spectral values at different periods. In fact the uniform hazard spectrum of spectral acceleration is widely demonstrated to cause conservative results. Therefore the correlation of the elastic input energy spectral values at different periods is initiatively evaluated and the analytical predictive equation is also proposed to calculate the correlation of elastic input energy spectral values. Using the correlation their conditional mean spectrum recognized as a more appropriate target spectrum for ground motions selections can be developed. On the other hand this correlation also can be used to calculate the predicted mean value and the dispersion of some integral intensity measures (such as VEIaSI, VEIrSI, MVEIaSI, MVEIrSI), achieving the application of these IMs in Performance-Based Earthquake Engineering. Finally, we made a practical Matlab implementation for ground motion selection and modification. Here it is called RELACS (REaL ACcelerogram Selection). The total ground motion database used in RELACS, with more available ground motion records, is composed of three large ground motion database, i.e. NGA (Next Generation Attenuation) database, SISMA (Site of Italian Strong Motion Accelerograms) database, and ESGM (European Strong Ground Motion) Database. The RELACS brings to engineers and researchers more convenience to select ground motion accelerograms, using nowadays widely adopted GMSM methods in terms of not only some commonly used acceleration-based IMs and some other scalar intensity measures but also some energy-based IMs that have been approved good predictors for the response of base-isolated buildings. The RELACS contains two consecutive steps: selection according to the geophysical parameters; and selection according to the elastic response parameters (IMs). The user can easily obtain the acceleration time-history, and the acceleration spectrum, the velocity spectrum and the displacement spectrum of the ground motion record selected using the RELACS.