A Fast Seismic Assessment Technique for Reinforced Concrete Buildings: Machine Learning-based Hassan Index
Assessing large inventories of reinforced concrete structures in urban areas with high seismicity is a daunting task that requires tools that can be applied quickly to produce reliable results. The first goal should be to identify the most vulnerable structures that require rapid intervention. Existing assessment standards are often too complex for this purpose. In the literature, the index-based methods, for example the Hassan Index, provide more efficient assessment options based on simple geometric parameters. The question addressed here is whether machine learning (ML) algorithms trained to use the same parameters can better match field observations. The developed algorithm has been trained and tested on survey data from 1320 low- to mid-rise buildings, the model achieved 74 % accuracy on a held-out test set with 5 % “risk” (false negatives) and 21 % “cost” (false positives), improving over the simple index-threshold baseline (61 % accuracy, 8 % risk, 31 % cost). On an external dataset from the 2024 Taiwan earthquake, performance remained comparable (73 % accuracy, 3 % risk, 24 % cost). The approach is intended to prioritize structures for detailed assessment and early intervention; its applicability is limited to buildings whose attributes fall within the training data domain in terms of statistical properties.
You can reach the full paper via the link below:
https://www.sciencedirect.com/science/article/pii/S2352012425022404
A Refined Experimental Dataset for Reinforced Concrete (RC) Columns [Article In Press]
This study provides a refined dataset of reinforced concrete columns compiled from
existing experimental studies available in the literature. Prior datasets have displayed
imperfections that may end up with deceiving outcomes of experimental observations
and misleading academic studies. A rigorous and structured methodology has been
applied in order to eliminate additional imperfections. Using a manual cross-validation
approach, each parameter in the ACI 369 dataset was directly compared and revised
with the corresponding values reported in the original experimental studies. A
representative nominal shear strength evaluation has been performed to show the
impact of using incorrect datasets for structural analysis and design. With its wide
range of input parameters, this newly refined dataset opens up exciting possibilities for
use in academic research and engineering practice. It can be used for validating
numerical models, deriving empirical equations, and developing data-driven models,
among other applications.
Methodological Framework for Assessing the Reliability Indices of Composite Bridge Superstructures Considering the Degradation of Shear Connectors: Aspects of Technical Implementation
The paper presents the methodological framework and technical aspects of reliability indices assessing of composite bridge superstructures considering the degradation of stud shear connectors. This study is a continuation of the previous ones, which established diagrams illustrating the time-dependent decline in the reliability index of a simple composite beam. A distinguishing feature of this reliability assessment task is the nonlinear approach of the calculation model, which allows for accounting for damage accumulation in stud shear connectors and the resultant reduction in their longitudinal stiffness. In the case of composite structures that are more intricate than simple beams (such as continuous beams and any spatial systems), the complexity of the calculation framework escalates considerably due to increased uncertainty surrounding the identification of critical point locations and load case pairs’ definitions for critical stress range values. Furthermore, considering the extensive list of input data within a probabilistic framework, it becomes imperative to select an efficient method for determining the reliability index to optimize computational resources. The paper proposes guidance to practicing engineers and researchers for developing a methodological framework for reliability analysis and presents illustrative examples that demonstrate the implementation of algorithms.
You can reach the full paper via the link below:
Unveiling the Implicitness: Kolmogorov-Arnold Networks for Structural Reliability Problems
The analysis and design process in structural engineering relies on the results obtained of the structural model from the black-box finite element analysis which causes implicit limit state function (i-LSF) in the structural reliability analysis (SRA). The current surrogate modeling techniques are based on evaluating the i-LSF to construct surrogates. However, even though their computational efficiencies and accuracies, the developed surrogates are mainly still implicit or yield highly complex i-LSFs. In this work, the Kolmogorov-Arnold Network (KAN) is used to discover an equivalent explicit LSF (ee-LSF) by generating a symbolic function for a given dataset. The discovered ee-LSF can be used in SRA since the expensive FEA is now able to be replaced by a simple explicit function. This paradigm allows us to unveil the implicitness of LSFs by discovering equivalent formulations through KANs which is novel to this work. Two examples are covered in this paper to present the ee-LSF approach. The ee-LSF approach demonstrates high accuracy, though its computational efficiency is currently lower compared to other surrogate modeling techniques. This limitation presents an opportunity for enhancement in future studies, particularly through integration with advanced sampling techniques.
You can reach the full paper via the link below:
Seismic performance assessment of structural systems in the aftermath of the 2023 Kahramanmaraş earthquakes: Observations and fragility analyses
This article evaluates how different reinforced concrete (RC) building systems in Türkiye behaved during the extreme 2023 Kahramanmaras earthquakes. The analysis relies on a comprehensive field survey covering 242 RC buildings across various heavily affected locations. Most surveyed buildings were low- and mid-rise RC moment frames and frame-wall (hybrid) systems, with RC wall construction being less commonly observed. Both RC frame and hybrid buildings exhibited several common deficiencies, resulting in significant structural and non-structural damage due to high drift demands. The performance of RC wall construction varied, with some buildings sustaining severe damage while others remained largely unaffected. An analysis of structural plans revealed that RC wall buildings with adequate wall amounts demonstrated exceptional performance, while those with inadequate amounts of walls experienced severe damage. In addition, fragility analyses using simplified models based on surveyed buildings reinforced these findings. The analyses suggested that RC frame and hybrid systems were insufficient in ensuring life safety during the earthquakes. Conversely, properly designed RC wall buildings are expected to perform well. This alignment between field observations and fragility analyses underscores the reliability of the study’s findings and emphasizes the effectiveness of RC wall construction in mitigating seismic risks and protecting life and property.
You can reach the full paper via the link below:
Investigation of the Effect of Artificial Neural Network Performance Parameters and Training Dataset on the Probability Estimate Capacity in Structural Reliability Problems
This study highlights two of the important details of the implementation of artificial neural networks to the structural reliability problems by pointing out the effect of training dataset, and the relationship between the performance parameters (coefficient of determination of train, validation, and test sets) of a network and its probability estimation capacity when it is used as a surrogate model in structural reliability problems. Four numerical examples are covered regarding these key aspects including one that is derived from a real-life reinforced concrete structure. Results have shown that the dataset can affect the probability estimation capacity for complex problems. Furthermore, it is also observed that having a neural network with good performance parameters does not mean that the network always has good probability estimation capacity. However, in order to have a network that can be used for probability estimate purposes, its performance parameters must be at a satisfactory level.
You can reach the full paper via the link below:
https://link.springer.com/chapter/10.1007/978-3-031-60271-9_37
Correlation Between Intensity Measures and Damage Caused by the 2023 Türkiye Earthquakes
This article presents comparisons between different ground-motion intensity measures and damage frequencies observed following the Pazarcık (Mw7.8) and Elbistan (Mw7.5) earthquakes that struck Türkiye in February 2023. Various intensity measures were examined including peak ground acceleration (PGA), peak ground velocity (PGV), and spectral ordinates (e.g., Sa0.3, Sa1, Sd1). Two data sources were used to quantify damage frequency: a survey conducted by the team deployed by ACI 133 Reconnaissance Committee involving more than 200 reinforced concrete buildings, and surveys published by the Ministry of Environment, Urbanization, and Climate Change of Türkiye. The differences and commonalities between the two surveys are discussed, and plausible correlations between damage and intensity measures are examined. It is concluded that, relative to other intensity measures, peak ground velocity (PGV) and spectral displacement at a fundamental period of 1.0 second (Sd1) had better correlations with the damage observed from Antakya to Malatya. This observation is consistent with two previously published ideas: a) in the absence of structural damage, drift demand can be expressed as a linear function of PGV, and b) seismic design ought to be focused on drift (and PGV) instead of force (and PGA).
You can reach the full paper via the link below:
https://proceedings-wcee.org/view.html?id=25188&conference=18WCEE
Lessons from the 2023 Southeast Türkiye Earthquakes: A Study on Damaged RC Buildings Considering the Hassan Index
A survey was conducted across 10 cities in Southeast Türkiye to classify damage in 242 reinforced concrete (RC) buildings constructed in the last 15 years, ranging from 2 to 16 stories. The ‘robustness’ of these buildings was quantified using ratios of cross-sectional areas of vertical elements (walls and columns) to floorplan areas. The results are compared with similar measures obtained for buildings in Erzincan and Duzce (Türkiye) and buildings in Chile and Japan as well. These comparisons suggest that excessive drift was one of the primary causes of the widespread damage in RC buildings across the cities surveyed, from Antakya to Malatya. Drift a) exposed a myriad of defects in structural layouts and reinforcing detailing, b) caused nearly destruction of partitions and other non-structural building components (leading to disruptions of functionality even in the absence of structural damage), and c) induced instability even in structures with better detailing. In contrast, stiff (albeit uncommon) structures with abundant and well-distributed structural walls had lower drifts and performed well. Except for sporadic failures in details placed at critical locations, those structures are still in use and should serve as models for reconstruction.
You can reach the full paper via the link below:
https://proceedings-wcee.org/view.html?id=25297&conference=18WCEE
Quantitative evaluation of the damage to RC buildings caused by the 2023 southeast Turkey earthquake sequence
Data from 15 earthquakes that occurred in 12 different countries are presented showing that, without better drift control, structures built with building codes allowing large seismic drifts are likely to keep leaving a wide wake of damage ranging from cracked partitions to building overturning. Following the earthquake sequence affecting southeast Turkey in 2023, a team led by Committee 133 of the American Concrete Institute surveyed nearly 250 reinforced concrete buildings in the area extending from Antakya to Malatya. Buildings ranging from 2 to 16 stories were surveyed to assess their damage and evaluate the robustness of their structures in relation to overall stiffness, as measured by the relative cross-sectional areas of structural walls and columns. The majority of the buildings were estimated to have been built in the past 10 years. Yet, the structures surveyed were observed to have amounts of structural walls and columns comparable with amounts reported after the Erzincan (1992), Duzce (1999), and Bingol (2003) Earthquakes in Turkey. These amounts are, on average, much smaller than the wall and column amounts used in Chile and Japan. Because of that lack of robustness and given the intensities of the motions reported from Antakya to Malatya (with 10 stations with peak ground velocity (PGV) of 100 cm/s or more), it is concluded that structures in this region experienced large drifts. Excessive drift (1) exposed a myriad of construction and detailing problems leading to severe structural damage and collapse, (2) induced overturning caused by p-delta for some buildings, and (3) caused widespread damage to brittle masonry partitions. The main lesson is simple: ductility is necessary but not sufficient. It is urgent that seismic drift limits are tightened in high-seismicity regions worldwide.
You can reach the full paper via the link below:
https://journals.sagepub.com/doi/full/10.1177/87552930231211208
Evaluation of the Structural Damage Caused by the 2023 Türkiye Earthquakes in Light of the Design-Basis and Measured Ground Motion Intensities
The 2023 Kahramanmaraş earthquakes (Mw7.7 Pazarcık and Mw7.6 Elbistan) struck eleven cities in Türkiye within nine hours. Over 200,000 buildings collapsed or suffered severe damage, while an additional 500,000 had moderate or light damage. Following the earthquakes, the Ministry of Environment, Urbanization, and Climate Change of Türkiye conducted a comprehensive damage assessment survey. The earthquakes were also recorded by a dense network of seismic stations. This study aims to investigate the relationship between the ground motion intensity measures and observed structural damage, utilizing the extensive damage assessment and ground motion data available for various locations in the affected region. Several intensity measures were employed, including effective peak ground acceleration (EPGA) and velocity (PGV), spectral accelerations and displacements for various periods (e.g. Sa0.3, Sa1, Sd0.3, Sd1). These measures were also compared with the design-basis ground motion values provided by the seismic design codes of Türkiye. Notably, the design-basis acceleration and displacement spectra, and code-predicted PGV values were mostly exceeded during the earthquakes, indicating their significance in assessing the damage. These two intensity measures also demonstrated stronger correlations with the observed structural damage compared to other measures. The comparisons with the design-basis values showed that the design spectra should be carefully scrutinized. The findings highlight the importance of a drift-based design approach, which can be based on the expected PGV, instead of a force-based design approach relying on the expected EPGA. Furthermore, this study emphasizes the importance of updating the predictions for PGV considering the consequences of the recent earthquakes.
You can reach the full paper via the link below:
https://link.springer.com/chapter/10.1007/978-3-031-57659-1_8
Analyzing Structural Performance of Buildings in the Kahramanmaraş Earthquakes: The Role of Structural Systems
The devastating earthquakes that struck Kahramanmaraş on February 6, 2023 impacted 11 cities and millions of people in Türkiye, leading to immediate demolition of over 270,000 buildings due to extensive damage or collapse. The aftermath left more than 2 million people facing accommodation issues, underscoring the urgent need to enhance residential building structural performance beyond the standard “life safety performance level”. This study, focusing on reinforced concrete structures, explores the correlation between structural performance and the structural systems of buildings. In the earthquake-affected region, around 90% of buildings utilize reinforced concrete (RC) structural systems. RC moment frames with or without shear walls dominate building construction in Turkey. Besides regular moment frames, wide-beam RC frames with thin slabs and one-way joists are also common. The current seismic design code allows these systems if adequate shear walls are provided, but it is observed that some recently constructed buildings violate this requirement. Both types of structural systems experienced damage during the earthquakes. A less frequent building type was shear-wall buildings, which rely solely on shear walls for lateral-load resistance. It was observed that the buildings with evenly distributed shear walls in both plan directions exhibited robust seismic performance if these wall amounts were sufficient compared to their height. Otherwise, these buildings suffered heavy structural damage. The findings of this study stress the need for a comprehensive assessment of structural systems to enhance earthquake performance and ensure the safety of lives and property in seismic regions.
You can reach the full paper via the link below:
https://link.springer.com/chapter/10.1007/978-3-031-57659-1_9