Abstract
Accurate prediction of the shear strength in open-web reinforced concrete deep beams (RCDBs) is essential
for ensuring structural safety and optimizing design efficiency. This study presents a supervised machine learning framework that integrates ensemble learning models with advanced data resampling and feature selection techniques to enhance predictive accuracy. Three powerful ensemble models, Random Forest (RF), Extra Trees (ET), and
Extreme Gradient Boosting (XGB) are utilized to capture the complex nonlinear relationships between structural parameters and shear strength. To address data imbalance and variability, two resampling strategies are applied: random
over-sampling and Gaussian noise-based augmentation. Feature selection is employed to reduce dimensionality and enhance model generalization, thereby contributing to more accurate and interpretable predictions of shear strength. The results show that combining supervised ensemble models with data resampling significantly outperforms conventional approaches. Among the tested models, XGB coupled with random oversampling achieved the highest accuracy and robustness with a correlation of determination (R2) of 0.96 and normalized root mean square error (RMSE) of 0.041in the testing stage. The key variables influencing the shear strength of RCDBs with web openings include the relative ratio of the opening's distance in the x-direction to the shear span, the yield strength of the transverse reinforcement, and the horizontal web reinforcement ratio, respectively. The proposed framework supports efficient
and dependable evaluation of shear performance in open-web RCDBs, even with limited data availability.
Key Words
deep beam; ensemble, feature selection; open-web; resampling; shear strength
Address
Mosbeh R. Kaloop: Department of Civil and Environmental Engineering, Incheon National University, Incheon 22012, Korea; Incheon Disaster Prevention Research Center, Incheon National University, Incheon 22012, Korea; Public Works Engineering Department, Mansoura University, Mansoura 35516, Egypt
Jong Wan Hu: Department of Civil and Environmental Engineering, Incheon National University, Incheon 22012, Korea; Incheon Disaster Prevention Research Center, Incheon National University, Incheon 22012, Korea
Basem S. Abdelwahed: Structural Engineering Department, Mansoura University, Mansoura 35516, Egypt
Abstract
The damping ratio is generally adopted as a fundamental parameter for the vibration analysis of cable structures. However, the damping ratio of cables is not constant; it varies with parameters such as length, mass, pretension
force, and friction between wires. To address this, a new theoretical model for out-of-plane free vibration is established, utilizing the ideal viscous damping coefficient as the primary parameter to predict vibration decay and
explain the variability of the damping ratio. Furthermore, a damping measurement method is proposed by analyzing the attenuation of vibration peaks. Vibration tests on CFRP cables were conducted to validate the model. Finally, the
influence of relevant parameters—including damping coefficient (c), mass per unit length (m), pre-tension (H),
length (l), and mode order (n)—on vibration attenuation is systematically analyzed. Results show that the proposed
model exhibits high accuracy. Under a fixed damping coefficient c, higher H and n, or lower m and l, lead to an
increase in the critical damping value, which consequently reduces the damping ratio (pn). The free vibration amplitude attenuates exponentially as wmax=e-(c/2m)t. Notably, the rate of time-history attenuation is primarily governed by c and m; a higher c or lower m accelerates decay. The H, l, n, and frequency do not exert a direct influence on vibration attenuation; instead, they may indirectly affect it by modulating the damping coefficient c.
Key Words
cable; damping; out-of-plane vibration; vibration attenuation
Address
Yujian Zhou: Research Institute of Urbanization and Urban Safety, School of Future Cities, University of Science and Technology Beijing, Beijing, China; Yangtze River Delta Carbon Fiber & Composite Innovation Center, Changzhou, China
Maoqiang Wang: Highway Bridges National Engineering Research Centre, Beijing, China
Qingrui Yue: Research Institute of Urbanization and Urban Safety, School of Future Cities, University of Science and Technology Beijing, Beijing, China
Xiaogang Liu: Research Institute of Urbanization and Urban Safety, School of Future Cities, University of Science and Technology Beijing, Beijing, China
Abstract
Seismic intensity measurements constitute are the core scale for evaluating the intensity of regional
seismic ground motion, which directly affects the accuracy of assessing the seismic fragility of reinforced concrete
(RC) bridges. This study proposes an innovative intensity measure that considers both continuous and discrete
seismic intensity estimation. A time history and frequency spectrum curve considering the real ground motion
parameters of the August 2024 Kyushu earthquake in Japan and the May 2008 Wenchuan earthquake in China were
generated via a nonlinear dynamic time history and frequency spectrum theory framework. Using Chinese
earthquake intensity tables and lifeline engineering damage assessment standards, a vulnerability level (damage state)
assessment was conducted on 303 RC bridges surveyed during the Wenchuan earthquake. A regression prediction
model for evaluating the seismic vulnerability of RC bridges in medium- to low-intensity zones has been proposed
using nonlinear prediction frameworks and strategies. To further improve the efficiency of assessing the seismic
vulnerability of RC bridges, artificial intelligence and machine learning algorithms are being considered. Three
innovative RC bridge seismic vulnerability prediction models considering typical intelligent algorithms (decision tree
(DT), random forest (RF), and multilayer perceptron (MLP)) have been developed. The intelligent learning and
regression models were validated using bridge damage data obtained from actual onsite investigations. The results
indicate that the proposed RC bridge seismic vulnerability prediction model is consistent with the onsite
investigation. The established vulnerability framework can significantly improve the prediction and assessment of the
seismic vulnerability of RC bridges in medium- to low-intensity zones.
Key Words
intelligent and learning models; medium- and low-intensity zones; reinforced concrete bridges; seismic intensity measures; seismic vulnerability
Address
Si-Qi Li: National Science Center for Earthquake Engineering, Tianjin University, Tianjin, 300350, China; School of Civil Engineering, Heilongjiang University, No. 74, Xuefu Road, Harbin City, China
Lin-Lin Zheng: School of Civil Engineering, Heilongjiang University, No. 74, Xuefu Road, Harbin City, China
Yong-Sheng Chen: Institute of Engineering Mechanics, China Earthquake Administration, No.29, Xuefu Road, Harbin City, China
Zhi-Peng Li: National Science Center for Earthquake Engineering, Tianjin University, Tianjin, 300350, China
Abstract
Machine learning (ML) emerges as an advanced tool that is promising to improve the smart design level of ultra-high performance concrete (UHPC) structures. However, the class imbalance in datasets seriously affects the prediction accuracy of ML models. This study establishes ML classification models to identify the UHPC beam's
shear failure modes. To address the class-imbalanced datasets issue, the synthetic minority oversampling technique and the adaptive synthetic sampling approach for imbalanced learning are employed, thereby enhancing prediction accuracy for minority classes. Specifically, six machine learning models are investigated and compared. The shapley additive explanations (SHAP) is employed to explain the contributing factors for each failure mode. The results
indicate that the CatBoost model achieving the best performance. After balancing the training data using the
oversampling method, the F1 scores for compression-shear and flexure-shear failures improved by 263.6% and 27.5%, respectively, and the overall accuracy of the CatBoost model increased by 6.8%. SHAP analysis revealed that the shear span-to-depth ratio (y), stirrup reinforcement ratio (psv), and tensile strength of UHPC (fcu) are the key factors
influencing the failure modes, with dominant influencing factors varying across different failure types.
Key Words
class imbalance; machine learning; oversampling; SHAP; shear failure mode; UHPC beams
Address
Shiqiang Qin: School of Civil Engineering and Architecture, Wuhan University of Technology, Wuhan, 430070, China; Sanya Science and Education Innovation Park of Wuhan University of Technology, Sanya 572000, Hainan, China
Jun Li, Le Zhong, Qiuping Wang: School of Civil Engineering and Architecture, Wuhan University of Technology, Wuhan, 430070, China
Yun-Lai Zhou: State Key Laboratory for Strength and Vibration of Mechanical Structures, School of Aerospace Engineering, Xi'an Jiaotong University, China
Abstract
This study presents a systematic investigation into the coupled effects of curing regime, incorporation of ground granulated blast furnace slag (GBFS), and polyamide fiber reinforcement on the performance of fly ashbased geopolymer concrete. A comprehensive experimental program was designed to evaluate varying GBFS contents (0-20%) and polyamide fiber ratios (0-3%) cured under both ambient and oven conditions. The mechanical behavior was evaluated through compressive, splitting tensile strength tests, while microstructural characteristics were examined using SEM, EDX, and TGA analyses. The results reveal that the curing regime plays a dominant role in strength development, with oven curing significantly enhancing geopolymerization and resulting in superior mechanical performance compared to ambient curing. GBFS incorporation contributed to strength enhancement by promoting the formation of calcium-rich reaction products, leading to a denser and more homogeneous matrix. Polyamide fibers improved mechanical performance up to an optimal content of 2%, beyond which strength degradation was observed due to fiber clustering and reduced matrix continuity. A strong interaction between GBFS
content and fiber dosage was identified, indicating that optimal performance is achieved through a balanced mixture
design rather than individual parameter optimization. The best-performing mixture was obtained with 2% polyamide
fiber and 20% GBFS under oven curing, demonstrating the combined effect of chemical and physical modifications.
Microstructural observations confirmed the formation of a compact matrix with improved fiber-matrix bonding in optimized mixtures. Overall, the study provides a comprehensive understanding of the multi-parameter interaction governing geopolymer concrete performance and offers practical guidance for optimizing sustainable fiberreinforced geopolymer systems.
Key Words
curing regime; GBFS; geopolymer concrete; polyamide fiber; strength development
Address
Ali İhsan Çelik, Ufuk Tunç, Atahan Güven: Department of Construction, Tomarza Mustafa Akincioglu Vocational School, Kayseri University, Kayseri, 38940, Turkey
Memduh Karalar: Department of Civil Engineering, Zonguldak Bulent Ecevit University, Zonguldak, Turkey
Yasin Onuralp Özkiliç: Department of Civil Engineering, Faculty of Engineering, Necmettin Erbakan University, Konya 42000, Turkey; Department of Technical Sciences, Western Caspian University, Baku, 1001, Azerbaijan; Department of Unique Buildings and Constructions Engineering, Don State Technical University, Gagarin Sq. 1, 344003 Rostov-on-Don, Russia
Abstract
The increasing demand for sustainable construction material has led researchers to explore alternatives to traditional cement. This study investigates the use of bagasse ash as a partial replacement of cement and natural fibers to improve concrete properties. Bagasse ash is an agro-industrial byproduct from the sugar industry rich in silica and pozzolanic properties and is adequate to replace a cement content level. This study replaces the cement partially with
bagasse ash in different percentages (5%, 10%, 15% and 20%) and introduces natural fiber (coir fiber) to improve toughness and crack resisting. A series of test were conducted to determine the compressive strength, flexural strength, split tensile strength and durability characteristics of the modified concrete mixes. Based on the findings, using 10% bagasse ash and 0.75% of coir fiber yield the superior mechanical properties, water permeability and propagation resistance. There in an increase of 12.98%, 8.21% compressive strength at 28 and 56 days respectively. The optimum percentage of coir fibre is 0.75% with addition of 10% of bagasse ash with an increase in compressive strength of 16.8% and 20.1% at 28 days and 56 days. These enhancements were attributed to the combined effect of pozzolanic reaction from bagasse ash and the fiber bridging mechanism.
Address
S. Sivakumar: Department of Civil Engineering, PSNA College of Engineering and Technology, Dindigul, 624622, India
S. Srividhya: Department of Civil Engineering, Kangeyam Institute of Technology, Tirupur 638108, India
C. Subramanian, R. Prakash: Department of Civil Engineering, Alagappa Chettiar Government College of Engineering and Technology, Karaikudi 630003, India
Abstract
Reinforced concrete columns play a vital role in structural stability against lateral forces, carrying and transferring loads from other structural elements to the foundation. Columns can be formed to meet various requirements, such as height and cross-sectional shape, and their shapes may vary by location. This study numerically investigated the effect of the sectional shape of a long concrete column on its structural performance under a concentric axial load using Abaqus. It assessed seven cross-sectional shapes of concrete columns while maintaining equal cross-sectional areas, reinforcement ratios, tie-bar spacing, and diameters. All columns featured fixed free ends with equal concrete compressive strength. The results indicated that the square section could endure the highest load, while the L-shaped section exhibited the lowest load capacity. Nevertheless, square, plus-shaped, rectangular, octagonal, T-shaped, and circular columns sustained higher loading than L-shaped columns by 90.9%, 57.1%, 56.8%, 25.3%, 24%, and 22.9%, respectively. The square and plus-shaped columns failed under axial loading due to the yielding of longitudinal bars without buckling. In contrast, the rectangular columns failed from buckling and yielding of bars on the convex buckling side, while the other shapes failed in buckling without yielding the longitudinal bars. The cracking load and longitudinal displacement were approximately similar in the square, plus-shaped, and rectangular columns, while circular columns experienced the lowest cracking load and
displacement. Furthermore, estimating the column bearing under axial loads using the ACI equations reveals that the
ACI equation cannot be applied to all column section shapes. Therefore, a theoretical study was conducted to introduce a column-section shape factor into the ACI equation.
Key Words
Abaqus software; column's sectional shape; cracking load; load capacity; long column; longitudinal displacement
Address
Ahmed Sh. J. Al-Zuheriy: College of Civil Engineering, University of Technology-Iraq, Baghdad, Iraq
Adil M. Jabbar: Civil Engineering Department, College of Engineering, Wasit University, Iraq
Qais A. Hasan: College of Civil Engineering, University of Technology-Iraq, Baghdad, Iraq
Abstract
The increasing height and slenderness of modern wind turbines make tower vibration a critical structural challenge. Previously proposed tuned mass dampers (TMD) cannot adequately mitigate this without imposing excessive mass or impractical spatial installations. Pendulum-based solutions, in particular, require long pendulum lengths to be installed inside the nacelle or the tower. This study introduces a novel approach employing a compact stand-alone TMD mounted externally to the nacelle. Thus, avoiding structural interference with the wind turbine and enhancing maintainability. This approach is employed by a sprung inverted pendulum tuned mass damper (SIP-TMD). The proposed inverted pendulum, along with the linear spring enables substantial size reduction without compromising performance. A dynamic model of the tower SIP-TMD system is developed. The model is used to analyze the system in both frequency and time domains to determine optimal tuning parameters. Under equivalent suppression performance, the SIP-TMD requires a pendulum length of 55% shorter than conventional pendulum TMDs and 51% shorter than prestressed pendulum TMDs. This geometric reduction makes nacelle-level installation both structurally and practically feasible. Furthermore, the SIP-TMD eliminates the need for extensive prestressed wiring associated with prestressed pendulum designs. The proposed framework is adaptable to a broad range of wind turbine configurations.
Key Words
inverted pendulum; sprung inverted pendulum; tuned mass damper; vibration absorber; wind turbines
Address
Islam Mohamed Heikal: Design and Production Engineering Department, Ain Shams University, 11517, Cairo, Egypt
Adel Elsabbagh: Design and Production Engineering Department, Ain Shams University, 11517, Cairo, Egypt
Bahi Bakeer: Design and Production Engineering Department, Ain Shams University, 11517, Cairo, Egypt; Department of Mechanical Engineering, Galala University, 43713 Suez, Egypt
