1-D site response analysis of horizontally layered soil profiles is commonly used to generate site-specific earthquake ground motions. While there are many existing methods, both equivalent-linear and nonlinear type, to simulate the response in small to medium strain range, application to large strain range is often considered questionable. The accurate representation of soil strength in the selected nonlinear model is the key to simulate response beyond the medium strain range. This paper presents a new analytical solution by combining a recently proposed double hyperbolic model with Mohr-Coulomb strength model. The paper describes the mathematical formulation of the new model and demonstrates a practical implementation using DYNES3D, an open-source site response analysis tool.

The application of the proposed strength formulation with double hyperbolic model is investigated by analyzing the response for an arbitrary soil profile and comparing with conventional hyperbolic model. Results indicate that using the proposed model could lead to smaller shear strains at shallow depths, potentially leading to larger surface accelerations and higher response acceleration in the shorter period range. A python package developed to interact with DYNES3D and analyze with the proposed method is also made available on GitHub.

The Philippines, being the archipelagic country that it is, has always understood the importance of inter-island transportation and, in a way, associated progress and development with it. Outside of the nation’s capital in Metro Manila, Davao Region is one of the strategic locations identified with the potential to thrive as a business center in Southern Philippines. In order to support and boost economic activity within the Davao Region, the Philippine government proposed the construction of a long-span extradosed bridge that connects two of the region’s major cities. This study presents the site characterization of the proposed bridge and the development of the site-specific ground motions. Situated on potentially liquefiable soil, this study implements an approach that couples porewater pressure generation with nonlinear site response analysis. The results show the expected period-dependent attenuation/amplification of the response spectra and reveal that, even at different hazard levels, the short- and intermediate-period spectral accelerations normalize to a certain intensity. The results are finally compared with the design spectra prescribed in the local design code as a way to demonstrate the effects of soil nonlinearity and liquefaction on the resulting ground motions. At the end of this paper, practical issues and key differences between this coupled approach and the conventional practice shall be highlighted.

Current building codes in the United States provide for five standard ground surface response spectra for seismic design purposes as a function of a site classification based on the weighted average shear wave velocity over the top 30 m of the profile, Vs30, and mapped spectral accelerations at period of 0.2 and 1.0 seconds. These mapped spectral accelerations were obtained using standard ground motion prediction equations and the average shear wave velocity and standard deviation for each site class. The current building codes allow for both the conduct of site-specific seismic hazard analyses and the conduct of site-specific response analyses, but the resulting ground surface response spectra must not fall below 80 percent of the standard response spectra. This paper describes several site conditions which are typical of those sites where it is advantageous to conduct nonlinear site response analyses and argues that there are conditions under which the lower limit is not applicable.

The design of buildings and infrastructures in seismic areas has to account for site effects, as they dramatically affect the expected ground shaking. The equivalent linear approach is commonly adopted for the numerical simulation of seismic site response. This scheme models the nonlinear soil behaviour as a linear viscoelastic medium, characterized by strain-compatible secant shear modulus and damping ratio extracted as a function of an equivalent uniform strain, usually termed effective shear strain. The effective strain is computed as the product between the maximum shear strain and the effective strain ratio, which is a scaling factor conventionally equal to 0.65 or linearly increasing with the magnitude. However, the proposed values are the result of recommendations without a rigorous demonstration and their reliability has been questioned. This study investigates the influence of this parameter on a collection of 1-D ground models, which are subjected to a set of acceleration time histories recorded from earthquakes with different intensities. Models are generated from a database of real soil profiles through a stochastic procedure and they are representative of a broad variety of soil deposits of engineering interest. The study addresses the sensitivity of the predicted ground motion amplification to variations in the effective strain ratio, considering the role of soil deformability, ground motion characteristics, and the investigated amplification parameter. This study contributes towards a more robust prediction of ground motion amplification of soil deposits, enhancing the reliability of design in seismic areas.

Ground classification refers to the process of categorizing the ground into different grades based on how soil properties impact ground motion. With the introduction of the National Standard of Seismic Design General (KDS 17 00 00) in South Korea, ground classification is now based on the average shear wave velocity of the soil layer (VSSO) and the depth to bedrock (BRDP). To consider the specific characteristics of the ground classification, this study develops a site amplification model based on a suite of nonlinear ground response analyses that utilize VSSO and BRDP as model variables. We collected borehole data from the National Geotechnical Information DB System and obtained V_S profiles through geophysical surveys. In cases where no survey record was available, the N-V_S relationship was used to estimate the profiles, or representative shear wave velocity profiles for each soil layer were obtained. The collected 93,486 profiles were then classified into 36 clusters based on VSSO and BRDP. 3,271 shear wave velocity profiles were randomly selected from each group and used for the site response analysis. This study utilized 34 observation records of horizontal components of ground motion from domestic and international earthquakes as input seismic waves. Then, each motion was scaled to 1, 1.5, and 2 times, resulting in 81 seismic waves with PGAs ranging from 0.1g to 1g. Nonlinear site response analyses were then performed using the NL_HH(Nonlinear Hybrid Hyperbolic) module using a Python-based site response analysis module, PySeismoSoil. This module requires shear wave velocity profiles and input seismic waves as input parameters. The NL_HH module uses the HH model, which employs an empirical dynamic curve (Darendeli, 2001)[1] to estimate the damping ratio. A nonlinear site response analysis of 81 input seismic waves was performed at 3,271 boreholes using the NL_HH module, and the response spectrum and site amplification ratio were determined. Based on these, this study develops a nonlinear site amplification model with the ground classification parameters, VSSO, BRDP, and input PGA level at the depth to bedrock as model variables.

Seismic-hazard analysis (SHA) is typically performed using ergodic ground-motion models (GMMs), wherein the site response component is derived from global data and conditioned on the time-averaged shear-wave velocity in the upper 30 meters (V_S30) and a “basin depth” term (e.g., Z_1.0 or Z_2.5). In the ergodic GMMs, for a given V_S30, there is an implicit shear-wave velocity (V_S) profile associated with the site response prediction that has smooth increases of velocity with depth. When a site-specific V_S profile is characterized by abrupt velocity contrasts, for example at the rock-soil interface, the site response is likely to differ significantly from ergodic model predictions. This limitation of the ergodic models can be overcome by incorporating non-ergodic site response in the SHA. This approach involves customizing the site response for site-specific conditions, which has the effect of decreasing overall model uncertainty.

In this paper, we describe results from ergodic SHA and SHA that incorporates non-ergodic site response at two sites in the San Francisco Bay Area. Both sites are characterized by a strong impedance contrast at the top of competent bedrock. Depth to bedrock at these sites varies, ranging from 75 meters to more than 400 meters. At each of the sites, nearby ground-motion records indicate that the ergodic GMMs tend to underestimate spectral accelerations at oscillator periods that are close to the fundamental site period. Conversely, there are typically broad period ranges where the ergodic GMMs overestimate spectral acceleration. Since the non-ergodic site response considers these local ground-motion data, these differences are reflected in the non-ergodic results. The findings from these two sites underscore the importance of estimating the fundamental site period, the limitations of ergodic models at sites with strong impedance contrasts, and the benefits of implementing non-ergodic site response into SHA.

The 2018 Northern Osaka earthquake (Mw5.6) caused residential damages in the Ibaraki and Takatsuki cities in Osaka. However, the distribution of damage was not uniform and there were locally damaged areas. To clarify the cause of this damage distribution, seismic records are pulled back to the engineering basement by using the velocity models evaluated from the microtremor surveys, and the ground motion distribution in Ibaraki city is calculated. The velocity models are evaluated by performing microtremor array observations in the vicinity of the stations where the seismic records are available. Phase velocity dispersion curves are estimated from the microtremor array observation records, and an initial velocity model is established to explain the phase velocity as well as the HVSR and nearby borehole data. A genetic algorithm is, then, used to optimize the model to fit the phase velocity more precisely. Using this model, the input motions on the engineering basement are estimated from the records at each station. The peak velocities at the engineering basement in the locally damaged areas are larger than those in the other areas. Since the effect of site amplification near the surface is excluded, it may represent source-induced differences such as forward directivity. A S-wave velocity model derived from borehole data is used to map the distribution of peak ground velocities in Ibaraki city. The results show a good agreement with the distribution of the residential damages. The amplification factors from the engineering basement to the ground surface are also greater in the locally damaged areas than in the other areas. This implies that the cause of the damage in the local areas was due to the combined effects of the forward directivity and the site amplifications.

The Atlantic and Gulf Coastal Plains (CPs) in the eastern and southern United States consist of thick and soft sediments that overlie a stiff bedrock that increase the amplification and duration of earthquake ground motions. Understanding both the depth and the dynamic properties of these sediments is critical yet challenging at a regional scale. Sediment thickness and geology have been shown to be good indicators for regional characterization of shear wave velocity (V_S). However, existing site amplification models within the CPs have limited characterization and modeling of these parameters. Therefore, we have developed a new site response and hazard model that uses a geology-based V_S model and geotechnical model to characterize the CP sediments so that amplification and hazard levels can be estimated. A sensitivity study using one-dimensional linear site response analyses is conducted to determine the influence of sediment thickness and reference condition assumptions within the model. Sediment thickness values of 100, 200, and 1000 m are tested. Additionally, reference condition velocities of 2500, 3000, and 3500 m/s at the base are tested. Response spectra- and Fourier amplitude spectra-based amplification ratios are calculated. The change in sediment thickness values is found to have the largest impact on longer-period (lower-frequency) amplification ratios as the fundamental period shifts. The change in reference condition also impacts longer-period (lower-frequency) amplification ratios more significantly. Overall, the amplification ratios estimated by the model are found to be more sensitive to the change in sediment thickness than to the change in reference condition. The results of the sensitivity study indicate that caution should be taken when using the model for shallow sediment thickness values (100 m or less), as this may lead to unrealistically large amplification ratios depending on the V_S value assumed for the reference condition.