To improve the seismic resilience of precast concrete moment frames, a replaceable fuse connection at the splice location representing the beam inflection point under gravity load was proposed previously by the authors. This connection incorporates shear tabs at the beam top acting as both the rotational center and shear transfer components, along with two buckling-restrained braces (BRBs) at the beam bottom serving as the fuse. The shear tabs are vertical steel plates embedded across the splice section and into the adjacent beam segments, with shear studs welded on the embedded parts to facilitate force transfer. While prior tests have validated the damage-control concept, design criteria for the shear tabs to ensure reliable force transfer between beam segments under combined axial, shear, and bending actions are still lacking. In this study, calibrated finite element models (FEMs) of the proposed connection were constructed to clarify the force transfer mechanism of the shear tabs. Informed by this mechanism, design methods for the shear tabs were presented, incorporating both strength criteria to ensure effective force transmission and damage-control criteria to avoid premature fracture under cyclic loading. These methods were validated using FEMs with variations in the cross-sectional dimensions and height-to-thickness ratios of the shear tabs, as well as in the arrangements and diameters of shear studs welded on the embedded parts.
Scalar damage indices for retrofitted reinforced concrete frames are calibrated against one intervention family and record how much damage has occurred rather than which mechanism produced it, so they neither transfer between retrofit types nor warn when a brittle mode is about to govern. The Mechanism Evolution Index (MEI) addresses this gap for non-ductile RC frames retrofitted with bonded carbon fiber-reinforced polymer (CFRP) jackets or steel buckling-restrained braces, within the scope of regular planar systems under single-direction motion. Six independently measurable response constituents combine through one frozen weight set, anchored once at the joint-panel cracking and sliding thresholds of a three-specimen calibration series, into a bounded scalar on the unit interval and a ternary coordinate that separates ductility, shear, and connection demand. The set is fixed once and applied without re-tuning across retrofit type and intensity. The framework is demonstrated on a one-third-scale specimen series, a bare frame with three-ply CFRP and tube-in-tube brace retrofits, reproduced by one calibrated OpenSees model. The bare frame becomes connection-dominated and approaches the sliding threshold at the design intensity, the CFRP jacket suppresses joint engagement then collapses through the same band at the ultimate level, and the brace stays on the ductility axis throughout. Transferability is examined on six external cases outside the calibration, four bare and two retrofitted; the source-reported mechanism sits on the leading ternary axis in every labelled case. The classical scalars track damage severity but cannot name the mechanism, whereas the connection component supplies an indicative retrofit-screening level.
This study investigates the cyclic performance of exterior reinforced concrete (RC) beam–column joints in which the normal concrete within the joint core is partially replaced by a fiber-strengthened epoxy composite (SFREC), consisting of an epoxy binder, silica sand, cement–microsilica filler, and hooked-end steel fibers. The experimental program was conducted in two stages. First, SFREC mixtures were characterized through compressive, direct tensile, and flexural tests, and the mixture F30-EB15-SF1.0 was selected based on a multi-criteria performance ranking. Second, three half-scale exterior RC beam–column joint specimens were tested under quasi-static reversed cyclic loading: BCJ-REF, a reference joint without transverse reinforcement in the joint core; BCJ-TR, a normal-concrete joint incorporating conventional joint-core transverse reinforcement; and BCJ-SFREC, a joint incorporating the selected SFREC mixture as a partial joint-core replacement. The verified results showed that BCJ-SFREC achieved the highest peak load, Pmax=6.23 kN, corresponding to increases of 55.4
This paper summarizes the findings of the recent project “Earthquake Risk Assessment of the City of Zagreb”, initiated as a national pilot study to serve as a model for future earthquake risk assessments across Croatia and as a foundation for changes to national regulations planned for 2026. Prompted by recent earthquakes, this initiative plays a key role in disaster risk reduction and complements ongoing post-earthquake retrofit efforts. The main activities of the project are presented, focusing on its most critical component: the collection of building data required to develop a GIS building inventory database. Attributes are primarily based on the latest Global Earthquake Model building taxonomy. The outcome is a comprehensive database of buildings and population, serving as the foundation for an exposure model. The earthquake risk model is presented through its three main components: hazard, exposure and vulnerability. The risk assessment was carried out in the OpenQuake Engine, using building-by-building data, and the main risk metrics are presented. The evaluation of the official city evacuation routes, considering potential road blockages caused by building debris, is briefly discussed. The limitations of the model are explained and some recommendations for future work are provided. This project marks a significant advancement following two earlier, more limited national earthquake risk assessments, both of which highlighted a critical earthquake scenario involving Zagreb. The new risk model at the building level was created, which can serve as a solid foundation for future risk mitigation activities and as an example for similar urban areas worldwide.
This paper studies the effect of different masonry infills on the seismic performance of masonry infilled reinforced concrete (MIRC) frames by experimentally investigating three small-scale multi-story multi-bay frames partially infilled with different masonry units, along with their strut-based macro-model evaluation. The studied masonry types include conventional burnt clay brick (BCB) and greener unburnt sandcrete (sand-cement) block (USB). Seismic performance was evaluated based on hysteretic behavior representing strength, stiffness, ductility and energy dissipation of the studied frames. Results show that the type of masonry infill has a significant influence on initial stiffness degradation, post-peak strength degradation, energy dissipation and ductility of multi-story multi-bay MIRC frames. Although BCB infills provided the highest lateral capacity, they exhibited sudden degradation of initial stiffness with brittle post-peak degradation. On the contrary, USB infills indicated improved ductility and the highest post-peak cumulative energy dissipation with slightly lower lateral capacity. Besides, the failure mechanism was discussed in detail, exploring extensive damage in the frame and infill panel near exterior beam-column joints.Test results were also numerically evaluated using two available macro-models of MIRC frames to assess the reliability of the models in predicting the seismic performance of MIRC frames. However, implementation of those macro-models was found somewhat difficult in predicting experimental behavior with acceptable accuracy due to either their dependency on substantial empirical calibration or overestimation of stiffness and energy dissipation.
Precast concrete (PC) structures provide significant advantages, including accelerated construction, superior quality control, and reduced on-site labor. Nevertheless, their seismic performance is often compromised by brittle beam-column connections that exhibit limited ductility and energy dissipation capacity, as observed in past earthquakes. This paper proposes a novel hybrid precast concrete beam-column connection incorporating a replaceable I-shaped steel ductile link acting as a seismic fuse. The link is bolted inside a steel-jacketed joint, where it effectively concentrates plastic deformations. This design ensures that yielding is confined exclusively to the replaceable fuse while protecting the integrity of the precast beams, columns, and joint regions. A parametric and numerical study, validated against experimental data, was carried out to investigate the parametric effects of the link to concrete beam strength ratio ( ), ρ factor effect, flange thickness ( t_f ), and depth ( h ) of I-shaped links. Results demonstrate stable hysteretic behavior, with all models achieving capacities exceeding Mp and minimal stiffness degradation. Results indicated that the most important variable on the system response is h rather than , ρ and t_f . Also, required equations were proposed for the design of the system.
The rapid and accurate generation of ground motion (GM) maps is crucial for effective seismic risk management in urban areas following an earthquake. However, conventional strong-motion networks are costly, sparsely distributed, and difficult to maintain, limiting their spatial resolution. This study proposes an alternative approach by leveraging structural response (SR) sensors installed in buildings, which offers a cost-efficient and scalable solution. The SR acceleration time histories are processed to extract multiple response parameters, which are then fused to estimate peak ground acceleration, peak ground velocity, and peak ground displacement. The study area is discretized into grid cells, where representative buildings are modeled as nonlinear MDOF shear structures. An optimized efficient channel-attention-based hybrid deep learning framework (ECA-Deep Net), integrating convolutional neural networks and long short-term memory networks, is trained for each representative building. The resulting GM maps for a simulated earthquake effectively reproduce the spatial variability of the reference ground-motion fields. The framework also exhibits stable performance under varying levels of sensor noise on a city scale. In addition, validation using field-recorded responses from an instrumented six-story building showed strong predictive performance on unseen earthquake records without architectural modification or hyper-parameter retuning, with correlation coefficients upto 0.97 and error values of 7.33
The region extending from Sylhet to Chittagong Division in northeastern to southeastern Bangladesh is one of the major tectonically active zones in South Asia, located at the convergence of the Indian, Eurasian, and Burmese plates. Several active faults, such as the Chittagong-Tripura Boundary Fault, the Tripura Fold Belt, and the Dauki Fault, have an impact on this region and frequently cause moderate-to-strong earthquakes. This study compares Probabilistic Seismic Hazard Analysis (PSHA) and the Analytic Hierarchy Process (AHP) based seismic susceptibility assessment to evaluate regional variations in seismic potential across the study area. The PSHA was conducted using a grid-based probabilistic seismic hazard assessment framework, incorporating the regional earthquake catalogue, Gutenberg-Richter recurrence analysis, and the Sharma et al. (2009) Ground Motion Prediction Equation (GMPE). For 10
Damage from pounding between adjacent structures has often been observed after earthquakes. This study investigates the influence of ground motion duration on the seismic pounding response of adjacent buildings with varying height configurations. Three pounding scenarios were examined—4–5, 4–7, and 4–9 storeys—corresponding to period ratios of 1.25, 1.76, and 2.29, respectively. The buildings were subjected to 77 ensembles of long-duration (significant duration > 25 s) and short-duration (significant duration < 25 s) ground motions, and their responses were evaluated in terms of acceleration, displacement, base shear, pounding force, and fragility estimates. The results reveal that ground motion duration has a minor to moderate influence on response quantities for period ratios below 1.76. However, at the period ratio of 2.29 (4–9 pounding scenario), short-duration motions produce dramatically higher amplifications than long-duration motions, i.e. acceleration amplifications have increased by 103
This study evaluates the performance of feature importance and feature selection methods for rapidly and reliably estimating the seismic risk levels of reinforced concrete (RC) buildings prior to an earthquake. The aim is to identify the most influential parameters in machine learning models and to investigate whether accurate predictions can be achieved using a reduced set of input features. A simulation-based dataset representing an existing building stock was utilized. Analyses using three machine learning models demonstrate that combining feature importance and feature selection improves prediction performance. LightGBM with normalized feature importance and CORR-based feature selection achieved the highest performance, reaching a macro-F1 score of approximately 0.91 while using fewer input parameters and outperforming models trained with the full feature set. The findings demonstrate that reliable seismic risk predictions can be achieved using a limited number of critical parameters, thereby enabling rapid and efficient large-scale building stock assessments. In addition, the proposed framework provides an interpretable and data-driven decision-support approach for regional seismic risk evaluation and urban-scale vulnerability screening applications.
Evaluating structural seismic performance requires accurate prediction of hysteretic behavior and energy dissipation capacity. While data-driven surrogate models improve the computational efficiency of traditional physics-based constitutive models, they often suffer from cumulative error propagation and lack explicit physical constraints, limiting their reliability in long-duration seismic response prediction. This study proposes a Hysteretic Physics-guided Neural Network that integrates energy conservation constraints with a cumulative error correction mechanism for seismic response simulation. The framework utilizes a Transformer-LSTM (Long Short-Term Memory) architecture to exploit its robust temporal feature extraction capabilities, serving as the backbone for the surrogate model to directly predict restoring forces, enabling efficient seismic response prediction. To mitigate the cumulative error propagation inherent in multi-step-ahead predictions, a correction mechanism based on a hybrid true-pseudo sample training strategy is implemented. Furthermore, a consistency constraint based on hysteretic loop energy conservation within a defined time window is embedded into the loss function to further enhance the predictive accuracy and physical consistency of the model. Through comprehensive validation using bilinear and Bouc-Wen hysteretic models, along with systematic implementation in both single-degree-of-freedom and multiple-degree-of-freedom structural systems subjected to seismic ground motions, the proposed framework demonstrates superior prediction accuracy, enhanced extrapolation capability, and robust generalization performance. These results highlight its potential as a physics-informed data-driven approach for efficient seismic response prediction and provide a foundation for future investigations involving more complex hysteretic behaviors and engineering scenarios.
This study investigated the seismic performance of multi-story traditional hybrid pavilion-style timber structures through shaking table tests and numerical modelling. A 1/4.5 scaled model of the Guangyue Tower in Northern China was fabricated and subjected to shaking table tests. Three ground motion records with different intensity levels were selected as seismic excitations. The damage patterns, dynamic characteristics, and responses of the model were analyzed based on the test results. The results showed that significant joint loosening and cracking were observed in the model under strong seismic excitation, which led to a 22.8
The estimation of shear wave velocity (VS) is a key parameter for seismic site characterization; however, its direct determination is not always available in urban investigations. In this context, empirical correlations with the Standard Penetration Test (SPT) represent a practical alternative. This study develops and validates a local empirical correlation between VS and SPT blow count for fine-grained soils in the city of Cochabamba, Bolivia, based on the integration of 164 data points obtained from geophysical measurements using the 1D Multichannel Analysis of Surface Waves (MASW) method and geotechnical SPT records from three representative zones. Power-law models were evaluated considering both the field blow count (Nfield) and the energy-corrected value. The results indicate improved performance when using N60 yielding the correlation V_S=129.59· N_60^0.2836 , with satisfactory statistical indicators (R2 = 0.6713, r = 0.8209 y RMSE = 30.81 m/s). The proposed model provides a reliable tool for estimating VS and contributes to the local seismic characterization of the Cochabamba valley.
This study aims to validate the spectral corrections given in ASCE Standard 4–98 as well as the coherency model developed by Abrahamson (2005) by comparing them with different coherency models widely used in the earthquake engineering field. Having implemented these coherency functions in the open-source software code_aster, in the first part of the study a rigid square foundation is analysed under spatially varying ground motion including the wave incoherency. Foundation response spectra are then determined for rock and soil site profiles using the compatible free-field response spectra. Based on the free-field response spectra and foundation response spectra, reduction factors are determined and compared with the spectral corrections given in EPRI (2005) and ASCE Standard 4–98. The analyses results reveal that the foundation response spectra can be very sensitive to the selected ground motion parameters of the assumed coherency models. In the second part of the paper, the effect of the spatial variability of ground motion on the reactor building of a nuclear power plant (NPP), including the soil-structure interaction (SSI), is investigated. The findings show that the effect of the incoherency is remarkable in the secondary systems which are sensitive to frequencies larger than 10 Hz.
To enable the quantitative evaluation of seismic collapse severity in historic masonry pagodas, this study proposes a dual-parameter damage index based on the Area Loss Ratio (ALR) and Residual Height Ratio (RHR). Seven representative pagodas are modeled using a three-dimensional discrete element method, in which key structural parameters, including internal infill condition, height-width ratio, and opening ratio, are systematically considered. Nonlinear dynamic analyses are conducted under multiple ground motions corresponding to frequent earthquake (PGA = 0.07 g) and rare earthquake (PGA = 0.40 g) to investigate the progressive collapse mechanisms. The results indicate that internal infill significantly enhances global stiffness and reduces seismic responses, while increases in height-width ratio and opening ratio lead to amplified displacement demand and higher collapse susceptibility. Under rare earthquakes, hollow-core and fully opening configurations exhibit pronounced block detachment and partial collapse, with ALR values reaching 10
Linear seismic design procedures prescribed by building codes typically establish a predefined peak inter-story drift ratio limit. However, the seismic damage and strength loss in Reinforced Concrete Beam-Column Connections (RCBCC), under identical peak drift ratios, vary significantly as a function of their structural and geometric attributes. Consequently, the peak inter-story drift ratio alone cannot fully characterize the seismic performance level (PL) of a structure. This research introduces a novel data-driven methodology for identifying the seismic performance level of Reinforced Concrete Moment-Resisting Frame (RCMRF), utilizing the projected peak drift ratio derived from code-prescribed linear design procedures. The methodology takes advantage of a robust experimental database containing force-deformation curves of 236 cyclic-loaded RCBCC subassemblies. The threshold values for peak drift ratios corresponding to ASCE41-23-compliant seismic Performance Levels, namely Immediate Occupancy (IO), Damage Control (DC), Life Safety (LS), and Collapse Prevention (CP), are then extracted. The detected broad range of peak drift ratios associated with every seismic performance level for samples with diverse structural characteristics highlights the crucial role of learning models in performance level recognition. The research employs nine shallow Machine Learning (ML) classification models, along with one deep learning model, achieving 88
Due to its excellent thermal insulation performance and low density and light weight, more and more ceramic aerated concrete blocks (CACB) are used as infill walls in frame structures. In order to study the influence of CACB infill wall on the seismic behavior of reinforced concrete(RC) frame structure, a bare RC frame and two CACB infill wall RC frames were constructed, and quasi-static loading tests were conducted. The results showed that CACB infill walls can significantly increase the horizontal bearing capacity of the structure compared with the bare frame, and changed the failure mode of the structure. This resulted in a more uniform distribution of cracks on the frame beams and columns, without forming obvious plastic hinges. The CACB infill wall increased the stiffness of the structure while reducing its ductility. Infill walls without windows reduced ductility even further. As the horizontal load increased, the stiffness degradation rate of the frame with infill walls was significantly higher than that of the bare frame. The energy dissipation capacity of CACB infill wall frames without windows was better than that of bare frame, while the capacity of the infill wall with window frames was slightly lower than that of bare frames. At the same time, ABAQUS finite element models were established for bare frame, CACB infill wall frame with window and without window. Simulation results showed that the hysteresis curve, skeleton curve, failure characteristics, and the entire force process were in good agreement with the experimental results. Furthermore, the influence of ethylene propylene diene monomer (EPDM) on the mechanical properties of CACB infill wall frames is analyzed. The structural deformation, energy dissipating capability and ductility is significantly increased. The EPDM is proved to be an effective flexible connection between the CACB infill wall and frame, which has the function to reduce earthquake damage.
Accurate characterization and modeling of pulse characteristics from near-fault ground-motion velocities are fundamental to seismic hazard and risk assessment. To this end, two key aspects are addressed in this study. Firstly, we developed a novel procedure to extract pulses in near-fault records by integrating the parametric pulse model and the generalized continuous wavelet transform (GCWT) identification method. Compared with previous identification methods that usually extract a single type of pulse, the proposed procedure can capture various types of pulses in near-fault records, ensuring the diversity and accuracy of the extracted pulses. Using this procedure, 815 of 21,222 records in the PEER NGA-West2 database are identified as pulse-like ground motions. Results show that the extracted pulses in pulse-like ground motions can effectively match pulse characteristics in records. Secondly, empirical pulse models are developed to predict the pulse parameters from earthquake magnitude, source-to-site distance, and site conditions ( V_s,30 ). Model uncertainty is quantified through residual analysis, and the proposed models are validated against existing models and independent datasets. Finally, practical guidance for synthesizing near-fault broadband ground motions is introduced by combining the empirical pulse model and the stochastic residual component. The generated artificial ground motions are verified to effectively retain the pulse features and match the spectral velocity and spectral acceleration of the benchmark ground mtion. Therefore, the developed empirical pulse model not only links pulse characteristics with seismological properties but also provides an effective approach to generate artificial near-fault ground motion with considering magnitude, distance, and site conditions. This work would directly facilitate the seismic hazard and risk analysis in near-fault regions.
The present study focuses on the evaluation of the life-cycle cost of concrete tunnel-form system through an innovative approach. Such life-cycle cost analysis accounts for the roles of different building components in generating injury, fatalities, and downtime. Using such an approach, three component categories, namely structural, acceleration-sensitive non-structural, and displacement-sensitive non-structural elements, were considered. For the 5- and 10-story models examined, the results primarily indicate the high strength and stiffness of the tunnel-form system. Even when adopting moderate ductility provisions for the walls, the mean peak ground acceleration corresponding to slight damage state is approximately 0·6 g. The probability of reaching such damage state under the design-basis earthquake is below 10
Earthquake-induced surface fault rupture is a major permanent ground-deformation hazard for foundations and overlying structures in active fault zones. Existing setback-distance assessments commonly rely on mapped surface rupture traces and empirical distances, whereas physical-model evidence remains limited regarding how normal-fault surface rupture propagates through overburden soils and affects shallow foundations. In this study, six 1 g physical model tests were conducted to investigate earthquake-induced normal-fault surface rupture propagation and its interaction with rectangular foundations. The effects of overburden soil type, overburden thickness, fault dip, and imposed bedrock fault offset were examined. The tests simulate permanent ground deformation from quasi-static bedrock fault offset, excluding transient seismic-wave-induced shaking and associated inertial effects. A multi-indicator envelope procedure was established to convert rupture–deformation–foundation response characteristics into setback-distance estimates by integrating internal rupture development, ground deformation, soil-pressure variation, foundation tilt, bottom strain, and foundation-base contact pressure. The results show that rupture development progressed through microcrack damage, rupture propagation and coalescence, and stable rupture development. Fault dip, overburden thickness, and soil type controlled the final rupture–deformation domain and foundation influence range. Under the adopted similarity relationship, the total setback width was 30.0–43.5 m for the 30 m thick clayey overburden, 24.0 m for the 15 m thick clayey overburden, and 21.0–30.0 m for the sandy overburden. The hanging-wall-side setback distance was larger than the footwall-side distance in all cases. These results support earthquake engineering assessment of surface-fault-rupture hazards for rectangular foundations at sites underlain by active normal faults.
