Asian Journal of Civil Engineering最新文献

Leakages in pipelines are still a significant challenge for fluid transportation systems, since they raise risks to efficiency, positive environmental impact and cost-effectiveness. Methods like eye inspection, pressure measurement and tracking flow rates do not usually catch leaks efficiently or accurately in big and busy pipeline installations. This study provides a way to use interpretable physical modeling and the predictive ability of machine learning to make the detection and classification of leaks more efficient. Second-degree polynomial regression and Random Forest regression models are both used in the study which are applied to synthetic data made using COMSOL Multiphysics. By analyzing pressure and velocity using regression, we can clearly understand the effects on leak size and position and by using Random Forest, we can attain much higher precision in predictions, with R² scores of 0.998 for leak size and 0.9999 for leak position. Looking at the importance of various features, it was clear that flow velocity has the most influence on leak dynamics and K-Means clustering organized the risks into helpful severity groups. All of these models together build a strong and flexible system designed for smart pipeline infrastructure use. It moves forward in predictive maintenance and helps unite our common sense with modern analytic methods used for pipeline condition monitoring.

The impact of the usage of sintered fly ash aggregates (SFA) as a partial substation of coarse natural aggregates (CNA) and metakaolin (MK) as partial replacement cement has been assessed in this study. To achieve this objective, a full factorial design has been adopted with choosing factors SFA(%) and MK(%) and compressive strength (CS), split tensile and flexural strength, water absorption (WA), volume of voids (VV) and density are selected as selected responses of this investigation. The levels of MK(%) are 0%, 10%, 15%, and 20%, and for the other factor SFA(%), 0%, 15%, 30%, and 45% are chosen as levels. The procedures of the general factorial approach have been followed for analysing the experimental outcomes. Analysis of variance (ANOVA) results and individual, contour, main effects and interaction plots have been used for annotation of the findings of the current research. The study depicts that the main effects of SFA(%) and MK(%) considerably influence the above-mentioned responses; however, the interaction of the said factors has the least significant impact on chosen concrete properties. Furthermore, the study reveals that a 26% reduction in CS, a 14% decline in STS, and a 20% decrease in FTS has been witnessed with the inclusion of SFA up to 45%, which can be compensated by the usage of 15% MK in SFA incorporated mix. Similarly, a reduction in density and rise in WA and VV of concrete is witnessed with the use of 45% SFA owing to inferior characteristics of SFA as compared to CNA. However, this degradation in concrete characteristics has been marginalised by using the beneficial effects of MK. The study recommends for adaptation of 30% SFA and 15% MK in making sustainable concrete for practical usage.

The growing emphasis on sustainable construction has spurred interest in utilizing recycled coarse aggregates (RCA) in structural concrete applications. However, the incorporation of RCA can significantly change the mechanical behavior of structural elements, particularly their punching shear resistance, a critical design consideration in flat slabs. Predicting the punching shear capacity (PSC) of reinforced concrete slabs is the goal of traditional analytical models and design guidelines. However, because material qualities are inherently variable and include intricate, nonlinear interactions, these models frequently fall short of producing accurate predictions. In response to this challenge, the present study proposes a robust data-driven framework for PSC prediction using four machine learning (ML) models: Gradient Boosting Machine (GBM), Extreme Learning Machine (ELM), Multiple Linear Regression (MLR), and Support Vector Regression (SVR). A curated dataset comprising 101 experimental observations was employed, encompassing eleven key input variables related to geometry, material properties, and reinforcement. The models were trained and validated using a 70:30 split and evaluated via multiple statistical indices, including R², RMSE, MAE, NSE, and WMAPE. GBM consistently outperformed the other models, achieving the highest prediction accuracy in both training and testing phases. To enhance model interpretability, advanced diagnostic tools such as Taylor diagrams, Regression Error Characteristic (REC) curves, and Cosine Amplitude Method (CAM)-based sensitivity analysis were employed. The results highlighted the dominant influence of concrete compressive strength, reinforcement properties, and cement content on PSC, providing critical insight into design priorities when using RCA.

Composite beams consist of profiled steel decking topped with in-situ reinforced concrete. This research investigates the load-slip behaviour of steel–concrete composite beams with different configurations of shear connectors and steel decking. Composite beams combine steel and concrete to create structures with superior strength, durability, and load-bearing capabilities, frequently employed in high-rise buildings and bridges. The study focuses on the role of shear connectors and steel profiles, which are essential for effective load transfer and slip resistance between the materials. By examining various shear connector types (e.g., channel, stud, and bolted connectors) and steel decking shapes (trapezoidal, rectangular, and re-entrant), the research aims to determine their impact on the load-slip performance and ultimate strength of composite beams. The project methodology includes experimental testing and numerical analysis to assess bond strength, slip behaviour, and structural performance under load. Advanced modelling techniques, including finite element analysis and machine learning algorithms, are employed to predict structural characteristics like load capacity and slip resistance. The study contributes to optimizing composite beam design, providing insights for construction applications that demand high strength, durability, and efficient material usage.

The study present an interpretable deep-learning framework, optimized using a hybrid Genetic Algorithm-Particle Swarm Optimization (GA-PSO), to predict and enhance the compressive strength of nano-modified geopolymer concrete (GPC). The framework integrates attention-augmented neural networks with SHAP-based explainability, Monte Carlo dropout uncertainty quantification, and surrogate-assisted multi-objective optimisation to simultaneously maximise strength while minimising cost and embodied CO₂ emissions. A curated dataset comprising 234 experimental GPC mixes–incorporating variables such as precursor type, nano-silica dosage, activator content, and curing conditions—was subjected to advanced preprocessing and polynomial feature engineering. A Binary Grey Wolf Optimiser (BGWO) was used for feature selection. The proposed DeepGA-PSO model outperformed conventional regressors (e.g., SVR, Random Forest, XGBoost) with an (R^2) of 0.994 and RMSE of 3.86 MPa. Explainability analyses identified curing regime, sodium hydroxide, and nano-silica content as key predictors. Optimisation via NSGA-II yielded Pareto-optimal mix designs suitable for cost-effective and low-carbon construction. A MATLAB-based GUI was developed to facilitate real-time mix design and prediction. This study offers a robust, scalable, and interpretable pipeline for data-driven GPC optimisation and provides a methodological foundation for intelligent infrastructure materials engineering.

Generation of waste tire increases with end-of its life. This scenario made the researchers to explore the feasibility of reusing and recycling the waste tire as an alternative material. Recent literatures mainly focus on the engineering properties of used tires alone or their behaviour when mixed with soil. To understand the research towards reuse/recycle of waste tire, a bibliometric study has been carried out to report a comprehensive and detailed bibliometric network mapping and evaluation of research progress connected to the utilization of used tires in geotechnical application. For the last two decades, it has been systematically documented through the Dimensions database. To understand the influence of publications, affiliations, journals, countries, authors, and keywords etc.in this field of research, the statistical analysis has been carried out. By using a bibliometric mapping tool, the evolving pattern of authors’ research themes and collaboration structures were examined. This bibliometric study findings revealed that there have been a significant number of publications and influence of authors to this studied topic in the recent two decades, as well as an increase in authors’ collaboration. Moreover, the objective is extended to explore the use of waste tire as geo-material to its use in geo-engineering practices.

The proliferation of Artificial Intelligence (AI) in Structural Health Monitoring (SHM) has catalyzed a paradigm shift from traditional, feature-based damage detection to end-to-end, data-driven methodologies. While Deep Learning (DL) models, particularly Convolutional Neural Networks (CNNs), have demonstrated remarkable efficacy, the advent of Transformer architectures presents a new frontier with unparalleled capabilities for sequence modeling. However, a direct comparative analysis of these architectures on a standardized experimental benchmark, coupled with a deep investigation into their decision-making processes, remains a critical research gap. This study addresses this void by conducting a comprehensive investigation using a publicly available experimental dataset from a six-storey laboratory shear building. We develop, train, and evaluate two distinct DL models: a lightweight one-dimensional CNN (Fast CNN) and a state-of-the-art Transformer-based model (Fast Transformer). Both models are tasked with directly classifying the structural state (undamaged vs. damaged) from raw accelerometer time-series data. Performance evaluation based on standard metrics reveals that both models achieve exceptional accuracy, with the Fast CNN reaching 99.44% and the Fast Transformer reaching 98.87% on validation datasets. This work’s core contribution lies in applying Explainable AI (XAI) techniques, including Integrated Gradients and saliency mapping, to deconstruct these models’ “black box” nature. Our analysis reveals a non-intuitive yet consistent finding: both the CNN and the Transformer primarily focus on the vibration signature of the base sensor (Sensor 1) to detect damage located at the fourth storey. This suggests the models have learned to identify damage through their influence on the structure’s global dynamic response as reflected at their boundary conditions. Furthermore, XAI reveals distinct operational strategies: the CNN acts as a highly localized feature detector, whereas the Transformer leverages its self-attention mechanism to weigh a broader spatiotemporal context. This paper provides a rigorous benchmark for modern DL architectures in vibration-based SHM and tells a technical story of how interpretable AI can uncover novel, physically meaningful damage detection strategies, enhancing trust and guiding future development of intelligent monitoring systems.

This study investigates the effects of incorporating fly ash (FA) and silica fume (SF) into concrete and evaluates the predictive accuracy of machine learning models such as Backpropagation Neural Network (BPNN), Random Forest Regressor (RFR), and Gradient Boosting Regressor (GBR), on compressive strength. Optimal performance was achieved with 50–60% FA and 8–10% SF, reaching strengths above 76 MPa at 90 days, while 100% FA with 10% SF reached 71.13 MPa at 90 days versus 27.6 MPa at 14 days. Among all models, GBR showed the best accuracy (R² = 0.996, MSE = 0.578, MAPE = 0.941%), with SHAP and Partial Dependence analyses confirming curing time as the most influential factor, followed by %SF and %FA. Perturbation analysis confirmed GBR’s robustness to input variation, and monotonicity analysis revealed a strong positive trend between curing time and strength (Spearman correlation = 0.9245), confirming GBR’s suitability for strength prediction and mix optimization.

While several studies have previously explored the prediction of compressive strength in geopolymer concrete, many suffer from limitations in feature selection, model generalizability, and prediction accuracy. This invention aims to enhance the prediction process by employing advanced machine learning algorithms capable of capturing complex, non-linear relationships between mix design parameters and compressive strength outcomes. To realize this objective, a dataset consisting of 276 geopolymer concrete mixes and their corresponding 28-day compressive strength values was compiled. Input features were selected based on two key criteria: their proven relevance in prior literature and their statistical significance in model performance. Multiple regression models—including Linear Regression, Decision Tree, Random Forest, Gradient Boosting, XGBoost, and NGBoost—were implemented and evaluated. Through trial-and-error, optimal hyperparameters such as the number of training epochs and k-fold values for cross-validation were determined. Model performance was assessed using standard evaluation metrics (R, RMSE, MAE, MSE), and further validated via score-based analysis. The model’s adaptability was tested using an independent secondary dataset. The results confirm that the NGBoost model achieved the most accurate predictions among all tested models, outperforming traditional approaches in both accuracy and consistency. This invention offers a scalable and reliable solution for predicting compressive strength, significantly reducing the need for physical trial mixes and enabling efficient, data-driven mix design in geopolymer concrete applications.

The incorporation of deep learning (DL) methodologies such as Neural Networks, Convolutional Neural Networks (CNNs), and CNNs-based hybrid AI systems, has tremendously shifted the paradigm in the field of structural retrofitting. This review analyses the architectural frameworks, practical implementations, and the structural safety measures undertaken using DL models aimed at improving the performance and cost efficiency in retrofitting techniques. Additional focus areas include damage identification, performance assessment of treated structures, and retrofitting design optimisation. The review critically assesses the data sufficiency, model training steps, and validation processes within the scope of civil engineering to deploy DL driven models. Clearly, further work is warranted with respect to sparsity of data, the ‘black box’ nature of the models, high computational costs, and absence of uniform benchmark criteria. Interdisciplinary approaches—combining civil engineering, data science, and legal policy—are essential to mitigate these challenges and fully exploit AI-enhanced capabilities for retrofitting. This paper will serve as a single point of reference for anyone intending to research or practically implement intelligent, adaptable, and safety-oriented retrofitting strategies.