Abstract
This study explores the application of machine learning (ML) models, including Random Forest (RF), Extreme Gradient Boosting (XGB), and Artificial Neural Network (ANN), for predicting the mechanical properties of viscoelastic dampers (VED), specifically the storage and loss modulus. VEDs play a crucial role in structural engineering for mitigating dynamic responses to seismic and wind forces. Despite their effectiveness, predicting the mechanical properties of VEDs remains a challenge due to their sensitivity to various factors such as loading amplitude, frequency, and temperature. Leveraging ML models and Explainable AI (XAI) techniques, this research aims to enhance understanding of VED behavior under cyclic loading and provide valuable insights for utilizing ML models for prediction of VED mechanical properties. The study conducts experiments within a temperature chamber, subjecting VEDs to cyclic loading in different conditions to discern the effects of these features on storage and loss modulus. The features such as loading amplitude, frequency, temperature, and loading cycle are then utilized to train ML models. XAI techniques are applied to provide insights into the predictive mechanisms of these models, ensuring the accuracy and reliability of predictions. The results indicate that all three ML models exhibit commendable prediction capabilities, with the ANN demonstrating superior performance compared to RF and XGB. According to SHAP (Shapley Additive Explanations) analysis reveals that loading amplitude exhibits the highest impact, followed by working temperature, loading cycle, and loading frequency.
Key Words
Kelvin-Voight model; machine learning; seismic retrofit; viscoelastic dampers
Address
Jonathan Dereje Assefa, Seungho Chun, Jinkoo Kim: Department of Global Smart City, Sungkyunkwan University, Suwon, Korea
Abstract
Geopolymer concrete (GPC) stands out as an environmentally friendly, sustainable structural material with superior mechanical and thermal properties, as an alternative to ordinary Portland cement (OPC) concrete. Due to this, investigating the dynamic properties of GPC materials is particularly crucial for structures subjected to vibration loads such as earthquakes. Therefore, eight heat-cured class F fly ash-based GPC and two OPC column specimens were subjected to modal tests using the Experimental Modal Analysis (EMA) method. Thus, the modal characteristics of columns produced from GPC were investigated and compared regarding reinforcement ratio, alkali activator ratio, and curing method. Furthermore, the dynamic properties of GPC and OPC columns were also compared. Additionally, the relationship between these columns' resonance frequencies, damping ratios, mode shapes, and mechanical properties was evaluated. The results show that although the resonance frequencies of OPC columns are 8.19% higher than those of GPC columns, their damping ratios are 13.01% lower. For GPC columns, the alkali activator ratio and curing method had a greater impact on the modal characteristics of the column samples than the reinforcement ratio. Besides, the relationship between strength and resonance frequency was closely related. In contrast, the damping ratio was closely related to the static modulus of elasticity and rigidity values. In particular, it was concluded that GPC columns had higher energy dissipation capacities, which was an important design parameter, under dynamic loads due to the high damping ratio.
Key Words
damping ratio; dynamic characteristic; empirical modal analysis; geopolymer concrete column; resonance frequency
Address
Hurmet Kucukgoncu: Civil Engineering Department, Engineering Faculty, Abdullah Gül University, 38080, Kayseri, Türkiye
Abstract
This study presents a unified computational framework for analyzing the influence of ply orientation on the in-plane force and bending moment resultants of three-layered functionally graded sandwich panels with a soft porous core. The face sheets comprise laminated SiC-Al plies with continuous axial gradation governed by a powerlaw distribution, with effective properties estimated through the Voigt rule of mixtures. The core elastic behaviour is characterized by a closed-cell foam porosity model governed by a non-dimensional porosity coefficient. Each ply is modelled as a linear elastic orthotropic continuum under plane stress, with laminate-level stiffness matrices assembled following Classical Laminate Plate Theory, while the full sandwich assembly is governed by Higher-Order Shear Deformation Theory to capture transverse shear deformation in the compliant core. Complete trigonometric expansions rigorously satisfy natural boundary conditions at all plate edges. Four representative stacking sequences-Distinct, quasi-isotropic, symmetric balanced, and antisymmetric-are systematically examined to establish laminate architecture as an independent structural design parameter. The framework enables quantitative identification of global stress resultant extrema and their spatial distributions, providing a theoretical basis for performance-based design of porous-core FGM sandwich panels in aerospace, automotive, and civil engineering applications.
Abstract
The ability to accurately estimate structural responses is essential for ensuring safety and enabling early damage detection in complex engineering systems. However, obtaining full-field structural state data is often hindered by the physical limitations of sensor installation in extreme operational environments and the scarcity of failure data required for data-driven approaches. To address these challenges, this paper proposes a physics-based virtual sensing technique that reconstructs the full-field strain distribution using a sparse array of strain sensors. The
proposed method utilizes the mode superposition principle, approximating the global structural response as a linear
combination of modal weights derived from limited sensor data. A key feature of this approach is the construction of
a hybrid basis set that integrates dominant low-order eigenmodes with quasi-static correction vectors, ensuring that both dynamic characteristics and static aeroelastic deformations are accurately captured with high computational
efficiency. The method is applied to a blended wing body (BWB) aircraft structure, and its performance is verified
through numerical simulations under various cruise conditions with elliptical lift distributions. The analysis results show that the proposed technique effectively estimates the strain field over the entire structure. Relative errors are mostly within 10% compared to the finite element analysis reference value. In addition, the error is less than 4% in the major deformation area, showing high precision. These findings confirm the potential of the proposed virtual sensing framework as a robust and efficient solution for real-time structural health monitoring in aerospace applications.
Key Words
blended wing body; finite element analysis; full-field strain reconstruction; mode superposition method; virtual sensing
Address
Mincheol Shin: Department of Aerospace and Mechanical Engineering, Korea Aerospace University, 76 Hanggongdaehak-ro, Deogyang-gu, Goyang, Gyeonggi 10540, Republic of Korea
Sungbo Lee: Gen-IV Reactor Technology Development Division, Korea Atomic Energy Research Institute, 111 Daedeok-daero, 989beon-gil, Yuseong-gu, Daejeon 34057, Republic of Korea
Seongpil Cho: Department of Aerospace and Mechanical Engineering, Korea Aerospace University, 76 Hanggongdaehak-ro, Deogyang-gu, Goyang, Gyeonggi 10540, Republic of Korea
Abstract
As the state-of-the-art in seismic resilience evolves from basic life-safety toward damage mitigation and continuous functionality, piloti-type reinforced concrete (RC) buildings remain a critical vulnerability due to their inherent vertical irregularities. While extensive literature addresses general soft-story retrofits, few studies detail the specific plastic hinge evolution and directional isolator–column interactions required to optimize isolation strategies. To bridge this gap, this study evaluates a representative piloti-type RC prototype (Ministry of Land, Infrastructure and Transport, R.O.K.), explicitly selected because its mid-rise height, asymmetric wall layout, and column dimensions accurately represent the broader stock of vulnerable piloti structures. To ensure strict methodological reproducibility, including ASCE-41 plastic-hinge definitions, material nonlinearity parameters, and effective section properties, all modeling choices are comprehensively detailed in SAP2000. Also, distinct from dynamic earthquake simulations, this study employs displacement-controlled quasi-static analyses to systematically map capacity and collapse progression without ground motion variability. Comparative analyses in both principal directions for non-isolated and base-isolated (lead-rubber bearing) configurations reveal that the non-isolated frame develops collapse-level softstory hinges at low displacements. On the other hand, the base-isolated model completes the prescribed displacement history without collapse by dissipating input energy through isolator hysteresis, dramatically reducing superstructure hinge demand and standardizing the inter-story drift profile. Differentiating this work from prior research, the results highlight that directional stiffness disparities and column sizing dictate energy absorption pathways as larger column sections and higher-stiffness axes significantly enhance isolator efficiency. The findings in this study provide novel, reproducible insights into typical piloti-type RC structural interactions, offering practical guidance for performancebased design in high-density urban seismic regions.
Address
Mo Shi: School of Economics and Management, Ankang University, 92 Yucai Road, Hanbin, Ankang 725000, People's Republic of China
Yeol Choi: School of Architecture, Kyungpook National University, 80 Daehak-ro, Buk-gu, Daegu 41566, Republic of Korea
Abstract
Designing fiber-reinforced concrete (FRC) panels to resist projectile impact is a challenging task, as it requires balancing the panel's failure resistance with construction feasibility, such as minimizing its weight. These objectives often conflict since lighter FRC panels tend to be more vulnerable to damage than heavier ones. Additionally, the panels must be designed to achieve predefined failure modes. To address these challenges, this study develops a multi-objective optimization process to minimize both the penetration ratio and the weight of FRC panels under missile impact while incorporating failure mode as a constraint. The optimization process is implemented using the nondominated sorting genetic algorithm-II (NSGA-II). Machine learning (ML) models are employed to predict penetration depth and classify failure modes using experimental datasets. However, due to the dataset's limitations, including class imbalance and insufficient samples, the k-means SMOTE technique is applied to generate additional data for the minor classes. Moreover, the Giant Trevally Optimizer (GTO) is utilized to adjust the hyperparameters of the ML models, aiming to achieve optimal performance. The results demonstrate that kmeans-SMOTE and GTO algorithms significantly improve the predictive accuracy of the models. Furthermore, the optimization algorithm effectively identifies multiple optimal solutions, exhibiting a clear trade-off between the objectives. The strong convergence toward boundary values and the even distribution of Pareto-front points confirms the algorithm's efficiency in exploring the solution space. Finally, a cloud-based platform is developed to employ the application of the proposed model in real-world design processes.
Address
Thai-Hoan Pham, Dai-Nhan Le, Thanh-Tung Pham: Department of Concrete Structures, Hanoi University of Civil Engineering, 55 Giai Phong, Hanoi, Vietnam
Ngoc-Phuong Nguyen: Hanoi Architectural University, Km10 Nguyen Trai Street, Ha Dong District, Hanoi, Viet Nam
Duc-Kien Thai: Department of Civil and Environmental Engineering, Sejong University, 98 Gunja-dong, Gwangjin-gu, Seoul, 143-747, Korea
open access
