SMIP 2023 Seminar

This paper presents a methodology to obtain the time history of the structural response using the Temporal Convolutional Network, a deep learning method. The presented methodology, in conjunction with sensor data from instrumented buildings, facilitates the prediction of the response in future earthquakes without the need for a structural analysis model. In this way, a computationally effective complement, or even alternative, to standard nonlinear time history analysis is possible. The applications of the developed method for different cases, including available number of records, buildings with higher mode effects, and linear and nonlinear response, are explored using accelerometer data from buildings instrumented by the California Strong Motion Instrumentation Program. Fundamental concepts of structural response and structural dynamics are used to guide the development of the training datasets and to explain the predictions. Furthermore, interpretation of the results is presented using earthquake engineering concepts. 

An approach to interrogate measured response on the behavior of inherent damping during nonlinear excursions is presented. The scheme computes a signal that approximates the base shear (to within a scalar) and decides on the inherent damping during nonlinear excursions on the premise that the derivative of this signal, with respect to time, is small within these segments. Preliminary results suggest that the inherent damping model should include a reduction in effectiveness when hysteretic dissipation is activated. 

This paper disseminates the ongoing research conducted for the assessment of the Alternative Design Provisions for Diaphragms in buildings per ASCE/SEI 7-22 Section 12.10.3 by utilizing recorded strong-motion acceleration data. Details of the workflow developed for the assessment of the design provisions are presented. 

The potential dam failure modes related to earthquakes are often the driving design criteria for new dams and the primary concern when evaluating the safety of existing dams. Ground motion recordings on dam sites at different locations were studied to estimate dams' fundamental and second mode vibration. Earthquake-based horizontal to vertical ratio (HVSR) are evaluated. Fourier amplification ratios using crest records over the abutment or downstream records are also assessed. The vibration characteristics are evaluated based on different methods and the resulting estimate of the fundamental frequency and second mode of vibration for Briones and Terminus dams are presented. 

We have estimated the thickness distribution of a geotechnical layer in the SCEC CVM-S4.26M01 (updated with recent ambient noise results) that generates the least-biased fit between 3D 0-1 Hz physics-based ground motion simulations and strong motion data in the greater Los Angeles area, CA, for 7 Mw4.4-5.4 earthquakes. Outside the basins, the optimal GTL thickness distribution shows strong spatial variation, generally increasing from near 0 m at the edges of the basins to values of 1,000 m or larger at distances of about 10-50 km, in particular toward the northeast, east and southeast. 

The Community Seismic Network (CSN) is a low-cost, MEMS-sensor seismic network with smaller average station-to-station spacing than stations for other networks. We have downloaded and processed CSN data for 29 earthquakes with M > 4 from 2012 to 2023 using NGA procedures. Visual checks of data useability were applied to distinguish rejected records form records with clear seismic signals. We compare recordings from proximate (within 3 km) CSN and non-CSN (generally SCSN or CSMIP) stations with usable signals. Results show no systematic differences for peak acceleration and similar spectra when the CSN motions have large usable bandwidths. 

The Learning From Earthquakes (LFE) program of the Earthquake Engineering Research Institute (EERI) deployed several teams to conduct reconnaissance activities in Turkey following the February 6, 2023 earthquakes. The findings from the teams, along with the observations from the GEER Association field teams are summarized in the report available here: https://learningfromearthquakes.org/2023-02-06-nurdagi-turkey/images/2023_02_06_nurdagi_turkey/GEER_2023_Turkey_Earthquake_FullReport_ReducedSize.pdf. 

This presentation brings forward highlights from the field observations with a focus on typical damage to reinforced concrete frame structures, and the impact of this damage to property and to the community functions. The presentation also reports on the post-earthquake functionality of two hospitals serving the same community, along with data collected from additional healthcare facilities in the region. Although the code requirements pertaining to structural engineering in the last two decades are on par with the requirements in the U.S., the observations from the field indicate that lack of code enforcement during the design and construction phase may have an impact on the performance of the buildings. 

The 2023 Türkiye earthquake sequence includes the February 6 M7.8 mainshock followed approximately nine hours later by a M7.7 aftershock, and many smaller aftershocks including M6.8 and M6.3 events on February 6 and 20, respectively. These events occurred in a region near the plate boundary on the East Anatolian Fault, in the proximity of which numerous strong motion recording stations had been installed north of the Türkiye-Syria border. Within hours of these significant events, strong motion data became available through the Earthquake Data Center System of Türkiye (TDVMS) and the Incorporated Research Institutions for Seismology (IRIS) from over 700 seismic stations operating in the region. Early releases of some data through TDVMS were found to contain baseline-correction and instrument metadata errors, which have been corrected over time. Additionally, records at a few number of stations which had be identified to terminate prematurely (i.e., the record ends while the site is experiencing significant shaking), have been properly windowed in subsequent releases. 

Raw time-series have been screened manually to remove noise-dominated and spurious records, and the remaining records were processed using standard procedures developed during Next Generation Attenuation (NGA) projects. A total of 310, 351, 291, and 229 usable three-component recordings were processed from the M7.8, M7.7, M6.8, and M6.3 events, respectively, with maximum peak ground accelerations at several stations exceeding 1.0 g. Source, path, and site metadata were compiled according to uniform protocols, which benefited greatly from extensive site characterization performed at seismic stations by the Disaster and Emergency Management Authority (AFAD). The strong motion data and associated metadata are publicly available at https://doi.org/10.17603/ds2-t115-bk16, and will be incorporated into the NGA-West3 database. 

Comparisons to ground motion models (GMMs) for active tectonic regions demonstrate the existence of complex path effects that result in relatively poor fits between the GMMs and observed data at large distances (R_JB > 200 km). Residual maps produced from these analyses demonstrate that ground motions generally over-predicted on the Anatolian block and under-predicted on the Arabian block. The repercussions of these events will have a lasting effect on the region, and their scientific and engineering impacts will prove to be influential for future ground motion related studies and efforts. 

GEER (Geotechnical Extreme Events Reconnaissance) mobilized several joint US and Turkish teams following the Feb 6th earthquakes in southwest Turkiye to collect perishable data. A collaborative EERI and GEER report was published disseminating the reconnaissance observations: https://www.geerassociation.org/index.php/component/geer_reports/?view=geerreports&layout=build&id=109 

This presentation provides a brief summary of the surface fault rupture, landslides, rock fall, liquefaction, lateral spreading, and other ground damage and how these impacted the civil infrastructure and built environment. The damage zone from these earthquakes was on the order of 200 km wide by 350 km long with widespread left-lateral surface fault rupture and liquefaction/lateral spreading affecting many cities and towns in the region. Several earth dams experienced some form of seismic induced deformations and one experienced surface fault rupture. Ports and harbors were damaged primarily due to liquefaction, and landsliding and rock falls were documented throughout the damage zone. The goal of these efforts is to learn from this event so that these hazards can be mitigated in future seismic events.