Large-scale earthquakes have occurred frequently in South Korea in recent years with increasing magnitude and frequency. Vibration control technology is an effective seismic design strategy that mitigates earthquake-induced loads. However, conventional vibration control devices are often vulnerable to damage or failure owing to material yielding, making replacement inevitable and incurring additional maintenance costs. To address these limitations, this study developed and assessed a smart damper device fabricated from a superelastic shape memory alloy(SMA) and polyurethane, designed to maintain functionality even after exposure to strong seismic events. To investigate the resistance of the device to sustained loads such as seismic loads, structural experiments were conducted by applying repeated loads and evaluating the derived load–displacement curves. For comparison, structural experiments were also conducted on a conventional damper device made of structural steel. The smart damper device exhibited superior performance in terms of energy dissipation capacity and restoring force compared with the conventional damper device. The smart damper device was also flexibly responsive to cyclic loads such as seismic loads. Although conventional steel may be more economical considering the cost of superelastic SMAs, an optimized design method for the smart damper device can potentially serve as an efficient alternative for vibration control.
This study investigates the development and optimization of one-part geopolymers (OPGs) as sustainable, eco-friendly alternatives to Ordinary Portland Cement, utilizing Tunisian calcined clay from the El Hamma region. Using a Box-Behnken experimental design under Response Surface Methodology (RSM), the research was evaluated the interaction and influence of three critical molar ratios, SiO₂/Al₂O₃, Na₂O/Al₂O₃, and H₂O/Na₂O, on compressive strength. Statistical analysis identifies the Na₂O/Al₂O₃ ratio as the most significant factor governing mechanical performance. An optimal compressive strength of 3.99 MPa was achieved at ratios of 3.51 (SiO₂/Al₂O₃), 1.20 (Na₂O/Al₂O₃), and 14.00 (H₂O/Na₂O). Microstructural characterization via XRD, FTIR, and SEM/EDX confirmed the successful formation of an amorphous N-A-S-H gel network, validating the effectiveness of this optimized binder system.
Expansive soils, prone to volumetric changes in response to moisture content variations, pose challenges for infrastructure founded on them. In cold regions, these problems are further intensified by repeated freeze–thaw (F–T) cycles, which alter soil structure through ice lens formation, water migration, cracking, and pore redistribution. This paper critically reviews the effects of F–T cycles on the geotechnical properties of expansive soils, highlighting past research, current developments, and future directions for the design of durable infrastructure in colder regions. The reviewed studies observe that the greatest deterioration generally occurs during the initial F–T cycles, after which the rate of change often decreases toward a stable state. The severity of degradation depends on initial water content, degree of saturation, compaction state, freezing temperature, cycle duration, and clay mineralogy. Changes in pore structure and crack connectivity are closely associated with reductions in strength, resilient modulus, and ultrasonic pulse velocity, as well as increases in permeability, compressibility, and volumetric deformation. Although several traditional and emerging stabilizers have shown potential, their long-term performance under repeated F–T exposure remains insufficiently established. Important research gaps include open-system testing, coupled F–T and desiccation effects, suction and compressibility behavior, macro-micro correlations, and field-scale evaluation. Overall, this review provides directions for future research to advance understanding and improve stabilization strategies for durable infrastructure constructed on expansive soils in cold regions.
This study presents an integrated experimental investigation, design code assessment, and numerical analysis of the shear behavior of High-Strength Self-Compacting Concrete (HSSCC) beams without transverse reinforcement. Twenty-seven beam specimens were tested using three concrete strength grades (approximately 78, 87, and 94 MPa), three longitudinal reinforcement ratios (1
Red mud, a by-product of bauxite refining, presents environmental challenges due to its large volume and hazardous composition. Although reusing red mud in building materials is a viable solution for valorizing waste and preventing environmental pollution, it typically requires a thermomechanical pretreatment that consumes a significant amount of energy. This study optimizes red mud calcination to enhance its role as both an activator and precursor in sustainable one-part geopolymer production through the combination of Life Cycle Assessment, sensitivity analysis, machine learning techniques, and analysis of variance. This research evaluates sustainable alkali-activated mixtures using ground granulated blast furnace slag, waste ceramics, and a waste glass-based activator, with calcined red mud tested as a partial substitute. It optimizes red mud calcination parameters and evaluates environmental and economic feasibility through carbon footprint and cost analyses benchmarked against traditional cement mortar. The highest compressive strength (27.6 MPa) is obtained for the same mixture with lowest ecological impact (2.93 kg CO₂-eq/MPa) and lowest economic impact (1.49 /MPa). Upscaling results demonstrated that large-scale production yields significantly lower carbon emissions and costs compared to both lab-scale operations and conventional Portland cement.
Post-earthquake observations of failures in fine-grained soils in precedent strong ground motions point to loss of strength of such soils in cyclic loading. This study reports on the results of the cyclic tests showing both cyclic softening and liquefaction, and a third type of behavior of fine-grained soils that is intermediate between those two, termed as clay-like, sand-like and intermediate, respectively. The presented results urge to conduct cyclic tests on fine-grained soils to assess their cyclic failure susceptibility, while following the proposed classification to identify the relevant soil type. The attained cyclic resistance ratios (CRR) were mostly smaller than the estimated cyclic stress ratios (CSR) governed by the maximum considered earthquake. Even though the examined fine-grained soil is primarily classified as not susceptible to either liquefaction or cyclic softening, it was observed by conducting cyclic tests that at most of the studied depths the results showed opposite values in both investigated areas, exposing a hazardous tendency for cyclic softening. This has significant economic consequences. A point of novelty of this study is the observation that the correlations between the CRR and the number of cycles introduced in the literature for different soils should be used with particular caution, especially in tests with substantial amount of loading cycles. Another point is that when the CSR were calculated both from the estimations and from the site response analysis, they showed inconsistent results requiring engineering judgment to reach a reliable safety factor against liquefaction and cyclic softening.
This experiment investigates the ability of self-compacting concrete (SCC) corner beam-column joints to resist fire exposure, using the ASTM E119 curve. Two specimen series with concrete cover thicknesses of 15 mm and 25 mm were tested; each series contained a single unconditioned reference specimen and two specimens subjected to one-sided scaled fire conditions of 26.66min and 53.33 min, respectively. Monotonic loading tests were applied. Strain and crack were monitored through Digital Image Correlation (DIC). Performance measurements included load capacity, stiffness, ductility, and response to shear strain; microstructural damage was measured using scanning electron microscopy (SEM). The results showed that the joints retained a substantial proportion of their load-carrying capacity after fire exposure, although noticeable reductions in stiffness and deformation capacity were observed. The thinner-cover specimens developed higher peak loads but exhibited earlier crack localization and a more brittle post-peak response, whereas the thicker-cover specimens sustained larger shear strains and demonstrated a more stable failure process despite slightly lower strength. These findings are in line with SEM observations, international codes, and previous experimental research, indicating that one-sided scaled fire exposure is a reasonable representation of realistic fire effects on SCC corner joints. Cover thickness is highlighted as one of the key points in the balance between load resistance and ductility, contributing to a better understanding of asymmetric fire performance and creating new design recommendations in the future.
Accurate prediction of concrete compressive strength is essential for mix design optimization and structural reliability assessment. However, the extent to which explicit feature engineering improves prediction performance beyond the choice of machine learning algorithm remains unclear. This study systematically investigates this question using the widely used Concrete Compressive Strength dataset and three data representation scenarios: (i) a baseline scenario containing raw input variables (raw scenario), (ii) an intermediate scenario involving ratio-based transformations (basic ratios scenario), and (iii) an advanced scenario incorporating engineered features derived from established physical relationships in concrete technology, such as the water/binder ratio and the logarithmically transformed curing age (engineered ratios scenario). Prediction performance was compared across Random Forest, XGBoost, LightGBM, and Support Vector Regression (SVR) models using a 20 × 20 nested cross-validation scheme with grouped folds to prevent information leakage between specimens sharing the same mix design, and the statistical significance of performance differences was assessed using the Bonferroni-corrected Wilcoxon signed-rank test. The best-performing configuration (LightGBM under the engineered ratios scenario) achieved R^2 = 0.903 and Root Mean Square Error (RMSE) = 4.84 MPa. Ratio-based transformations alone did not yield a statistically significant improvement over raw inputs for any model. The engineered ratios scenario produced a statistically significant improvement only for Random Forest ( R^2 increased from 0.869 to 0.893, p = 0.019 ); the corresponding improvements for XGBoost, LightGBM, and SVR were not statistically significant, despite modest increases in mean R^2 . SHapley Additive exPlanations (SHAP)-based interpretability analyses showed that the engineered water/binder ratio and logarithmically transformed curing age were consistently among the most influential features across all models, with dependence patterns consistent with established relationships in concrete technology, such as Abrams’ Law. These findings indicate that the benefit of physically motivated feature engineering is model-dependent rather than universal, and underscore the importance of leakage-aware cross-validation design when evaluating feature engineering strategies for concrete strength prediction.
Accurate non-destructive durability assessment of cover concrete using the Surface Water Absorption Test (SWAT) is significantly affected by high initial surface moisture conditions encountered in in-service concrete structures. This study develops and validates a Surface Moisture Correction Factor (SMCF) to account for the influence of high surface moisture content, measured using the Kett HI-100 surface moisture meter, on the Coefficient of Surface Water Absorption (CSWA) index of SWAT. Natural environmental exposure of six OPC concrete blocks over ten months confirmed that in-service structures consistently exhibit surface moisture contents spanning both the plateau region (120–210 HI-100 count values) and the wet region (above 210 count values), justifying the target scope of the SMCF. Statistical analysis of 150 paired CSWA – HI-100 count values revealed a statistically significant inverse linear relationship (β1 = − 0.04856, p = 1.19 × 10−17). Among three evaluated correction strategies, the additive bias correction approach achieved the most statistically stable result, reducing data variance by approximately 39
According to the Indian Standard, the target mean compressive strength of concrete is calculated using the formula fcm = fck + 1.65σ. This study investigates whether this fixed safety margin ensures consistent reliability when natural variations in concrete ingredients are evaluated using nonlinear strength models. Using 1,030 experimental concrete-strength records compiled by Yeh and archived in the UCI Machine Learning Repository, a standard Artificial Neural Network (ANN) was trained alongside two optimized hybrid models: GA-ANN and PSO-ANN. The dataset was strictly divided into training (761 records), validation (134), and test (135) sets. Several ensemble tree models were evaluated, with XGBoost achieving the best performance on the hidden test set with R2 = 0.968; RMSE = 3.04 MPa. For the neural networks, the test R2 values were 0.903 for the standard ANN, 0.915 for the GA-ANN, and 0.908 for the PSO-ANN. To evaluate practical safety, correlated Monte Carlo simulations were conducted with 80,000 trials for each concrete grade. These tests produced reliability indices (β) ranging from 0.861 to 2.575, meaning none reached the safety target of β = 3.5. Residual tests revealed significant heteroscedasticity, and applying the model to an independent slump dataset yielded a weak predictive transfer. Ultimately, these findings provide computational evidence of sensitivity to ingredient fluctuations, rather than an immediate basis for revising standards without further plant-calibrated experimental validation.
This paper develops an enhanced magneto-modal structural damage identification framework to find low levels of stiffness variation within noisy measurements, variable operating conditions and limited modal data. An MPPN uses MMPEN to encode magneto-modal perturbations; A TDEM maps structural evolution at various scales with a topology; A CSAO provides a causally constrained search algorithm; SUQRFE measures uncertainties using spectral methods and fuses residuals from each method; MSECV validates structural consistency through multi-scale energy methods. Compared with representative modal, swarm and recent learning-based SHM baselines, the framework is evaluated as a benchmark-level integrated enhancement rather than as a first-of-its-kind deployment claim. Evaluations based on the proposed framework were completed for beam, truss, frame and LANL benchmark tests with respective detection accuracies ranging from 96.8 to 97.5
Vehicle-bridge collision is usually represented in design by a nominal static force, although the response depends on the vehicle, contact, pier, and boundary conditions. This study examines a concrete-filled steel tube (CFST) pier struck by a deformable 16-ton NCAC truck at 45 and 80 km/h using archived LS-DYNA analyses. The baseline model had a fixed lower boundary and did not include deck mass, bearing elements, or explicit lateral or rotational restraint at the pier head. Peak contact forces were 3400 and 5870 kN, while the corresponding maximum pier-head displacements were 89 and 150 mm. Mapping these displacements to the available archived static-response points gave displacement-based equivalent static forces (ESFs) of approximately 2135 and 3500 kN. The two cases are reported separately and are not used to fit an ESF-velocity relation. The model developed local yielding, local buckling, and plastic-hinge formation without global instability during the simulated interval. Published RC-pier studies are used only for context. Quantitative soil-structure interaction results are excluded because the spring-dashpot input set cannot be reproduced. A previously published full-scale quasi-static test by the authors provides supporting evidence for the displacement-matching concept and stable lateral response, but not direct validation of the dynamic contact history. The results clarify the distinction between peak contact force and displacement-based ESF and identify the validation and parametric work needed before design calibration.
The development of high-rise buildings is associated with high complexity in design coordination, high capital at risk, high resource consumption, and high uncertainty in terms of cost, schedule, and embodied-carbon. Traditionally, cost overrun, schedule delay, and environmental impact are usually assessed independently even though they are highly correlated due to the geometry of the building, materials’ volumes, construction sequences, logistics, and resource usage. The purpose of this research is to propose a new framework for high-rise building risk assessment and management based on BIM integration and multi-objective optimization, using machine learning and explainable AI. 2,860 building designs were created, based on BIM data in IFC format regarding geometric parameters, structure design, envelope design, 4D scheduling, 5D costing, resources, and carbon footprint. Six machine learning models, including XGBoost, Random Forest, LightGBM, CatBoost, support vector regression, and artificial neural networks, were applied to estimate the three criteria of interest, i.e., cost overrun, schedule delay, and carbon emissions. XGBoost showed the best results with R² = 0.939 for cost, R² = 0.928 for schedule, and R² = 0.951 for carbon prediction. Concrete volume, planned cost, length of critical paths, number of changes in design, MEPS clashes, and transportation were the major risks according to SHAP analysis.
Limited studies have directly compared the fire performance of clay and concrete masonry units using integrated mechanical, chemical, and microstructural characterization under identical elevated-temperature conditions. This study addressed this gap by exposing both materials to temperatures ranging from 50 °C to 1000 °C and evaluating their response using compressive and flexural strength testing, FTIR, SEM, BET surface area analysis, ion migration, and moisture retention measurements. Concrete masonry units exhibited significant thermal degradation, with compressive strength decreasing from 25.4 to 16 MPa, flexural strength reducing from 1.42 to 0.03 kN, surface area declining from 14.7 to 1.2 m²/g, and ion migration reaching 2.751 ppm. ANOVA confirmed a statistically significant temperature effect (F = 20.3, p = 8.18 × 10⁻⁷). In contrast, clay masonry units maintained stable compressive strength (18.8–21.07 MPa), showed minimal flexural variation (0.36–0.67 kN), lower ion migration (1.161 ppm), and negligible shrinkage (0.0–0.05
Soybean crude urease-calcite precipitation (SCU-CP) is an environmentally sustainable bio-grouting technique that employs soybean extract as a urease biosource to initiate calcium carbonate precipitation. However, conventional extraction methods frequently result in elevated organic matter content, which can hinder the efficacy and distribution of calcite within the soil. Furthermore, the seed coat of soybeans contains lignocellulosic compounds that contribute to the mass of the organic precipitates. This study evaluated the impact of various soybean seed pretreatments and extraction methods on precipitation parameters and their role in enhancing the strength of sandy soil. The treatments involved the use of soybeans without seed coats, preparation of soybeans under dormancy break conditions, use of two distinct extraction methods (a No. 400 sieve and pressurized filtration), and incorporation of ethanol alongside distilled water. The findings indicated that the removal of the soybean seed coat effectively enhanced the purity of the SCU-CP solution, with an average reduction in organic content of 3.0
Crack detection plays a critical role in the structural health monitoring of civil engineering infrastructure to ensure safety, durability, and maintenance planning. Traditional manual inspection methods are time-consuming, labor-intensive, and prone to human error. Recently, automated crack detection systems have been significantly enhanced using image processing, computer vision, and deep learning techniques. This review article provides a comprehensive analysis of image processing and deep learning-based crack detection techniques for civil engineering structures. It provides a critical review of various methods, including edge detection, thresholding, morphological operations, Convolutional Neural Networks (CNNs), Fully Convolutional Networks (FCNs), U-shaped convolutional networks (U-Net), You Only Look Once (YOLO), and Digital Image Correlation (DIC). The performances of the different algorithms were compared in terms of accuracy, precision, computational efficiency, and crack characterization capability. Moreover, publicly available datasets, preprocessing methods, segmentation techniques, and evaluation metrics are discussed. This study also highlights key challenges, including variations in illumination, noisy backgrounds, limited datasets, and computational complexity. Finally, future research directions involving hybrid AI models, transfer learning, UAV-based inspection, and real-time crack monitoring are discussed. This review provides useful information for researchers and practitioners working on structural health monitoring and intelligent infrastructure systems.
This study investigates the flexural behavior of reinforced concrete beams containing recycled concrete aggregate (RCA) under inclined loading conditions. Four concrete mixtures with RCA replacement ratios of 0
Porous geopolymers are being developed as multifunctional components that can combine hydraulic transport, contaminant retention, chemical resistance, and construction-scale formability. This critical review evaluates their use in water-containment systems and in tertiary treatment, defined here as post-secondary polishing for removal of residual suspended matter, nutrients, dissolved metals, trace organics, and microorganisms before discharge or reuse. The synthesis links precursor chemistry and batch variability to gel assemblage, pore connectivity, tortuosity, anisotropy, surface charge, permeability, and mechanical integrity. Chemical foaming, sacrificial templating, freeze casting, pressing, retardation-induced porosity, and additive manufacturing provide different degrees of pore control, but no universal strength–porosity relationship applies across low-calcium N-A-S-H and calcium-rich C-(A)-S-H/N-A-S-H systems. Adsorption and ion exchange remain the most mature mechanisms for dissolved metals and dyes, whereas photocatalytic and antimicrobial functions require stable modifiers and explicit leaching assessment. The review further shows that total porosity and short-term removal efficiency are insufficient design criteria; construction deployment requires permeability retention, anisotropic transport characterization, resistance to hydraulic cleaning, chemical and ultraviolet stability, regeneration performance, modular manufacturability, and service-life evidence under real wastewater chemistry. A tiered reporting framework and provisional engineering screening envelope are proposed. The principal need is a chemistry-aware, performance-based route that connects pore engineering to structural reliability, treated-water quality, life-cycle impact, and scalable manufacture.
This study presents a performance-based multi-objective optimization framework for evaluating hybrid steel-basalt fiber-reinforced polymer (BFRP) reinforcement strategies in recycled aggregate concrete (RC) beams subjected to flexure. A three-dimensional nonlinear finite element model was developed in ABAQUS using the Concrete Damaged Plasticity (CDP) formulation and validated against experimental results for beams reinforced with conventional steel and BFRP bars. The validated numerical model was subsequently used as a computational framework to investigate a range of hybrid reinforcement layouts combining steel and BFRP tensile bars. Structural performance was quantified using normalized indicators including ultimate load capacity, energy absorption, energy-based ductility, and post-cracking tensile stiffening. These indicators were integrated into a global mechanical deviation index relative to the steel reference beam, while the reduction in tensile steel area was adopted as a durability-oriented proxy. A Pareto-based multi-objective formulation was then employed to explicitly quantify the trade-off between mechanical performance and steel reduction, and acceptable hybrid configurations were identified through an ε-constraint selection procedure. The results indicate a nonlinear relationship between steel reduction and structural response. Across all hybrid configurations, ultimate load capacity and post-cracking stiffness were maintained or exceeded relative to the steel reference beam, whereas the fully BFRP-reinforced beam exhibited a 2.9
The extensive use of natural resources in construction and development projects across the globe poses serious environmental challenges and ecological imbalance and has led to the search for alternative construction materials. In view of this, the present study investigates performance of novel M40 grade concrete blend containing Portland Slag Cement, Alccofine as constant partial replacement (15
