Asian Journal of Civil Engineering最新文献

Non-destructive testing (NDT) techniques such as the rebound hammer (yielding rebound number, RN) and ultrasonic pulse velocity (UPV) are widely used to infer concrete strength without damaging specimens, yet their standalone accuracy remains limited. In this study, we propose a weighted-ensemble machine learning framework that simultaneously predicts RN and UPV based on mix design parameters (cement, aggregates, water–cement ratio, admixtures) and curing age. Six traditional regressors–ElasticNet, SVR, KNN, Random Forest, XGBoost, and LightGBM–were each tuned via Optuna hyperparameter optimization. Ensemble weights were derived from inverse-RMSE scores on out-of-fold validation. On a hold-out test set of 30 specimens, the ensemble achieved RMSE = 0.83 and (R^2) = 0.94 for RN, and similarly strong performance for UPV, representing a 40–50% improvement over individual models. We further quantify prediction uncertainty with 95% bootstrap intervals (empirical coverage (> 90%)) and interpret model behavior via SHAP and Sobol sensitivity analyses. Feature attributions reveal that nonlinear interactions–particularly second-order terms of aggregate contents and the water–cement ratio–dominate predictions, while sensitivity indices confirm cement dosage and (w/c) ratio as primary drivers. This integrated approach offers highly accurate, well-calibrated predictions and actionable insights for NDT-based quality control in concrete construction.

This study presents a comprehensive investigation of the performance assessment of reinforced concrete (RC) multistory buildings with varying shear wall configurations under seismic and wind loading conditions, utilizing ETABS software for structural analysis. Six building models were designed, each with distinct arrangements of shear walls, including their placement in the X and Y directions. These configurations were evaluated on the basis of key structural performance parameters such as lateral displacement, interstory drift, and shear force, following the guidelines of IS:875-2002. Among the six configurations, Model 4, which incorporates shear walls in both the X and Y directions, demonstrated the most effective performance by minimizing displacements and enhancing the overall lateral resistance. To further support efficient structural design and reduce computational demand, this study integrated machine learning (ML) techniques for predictive analysis. A dataset comprising 153 samples was generated from the ETABS analytical results, capturing critical parameters such as model type, story height, displacement, drift, and shear. Three supervised regression models, a decision tree regressor, a random forest regressor, and a multilayer perceptron (MLP), were trained and tested to predict structural responses. Among these, the random forest regressor achieved the best balance between accuracy and generalization, demonstrating its robustness in modeling complex structural behavior. This dual-phase methodology, which combines detailed finite element analysis with data-driven ML modeling, provides a novel and practical framework for evaluating the seismic and wind performance of multistory RC buildings. The results offer valuable insights for structural engineers, enabling quicker decision-making and optimized design strategies in performance-based design and urban construction planning.

The use of nano graphene oxide (GO) in cement composites has shown tremendous potential for improving strength and performance characteristics. The impact of the addition of GO in concrete remains uncertain due to the interaction of the mix ingredients with the graphene oxide. To examine the influence of multiple coupling parameters on forecasting the mechanical properties using traditional experimental methods are cumbersome. In this study, Machine Learning (ML) approaches are used to investigate the intricate relationship between the multiple influencing parameters on the mechanical properties of GO reinforced cement composites. A comprehensive collection of 260 datasets related to GO, with 10 input parameters, was collected to train and test the machine learning models. Different Machine Learning techniques were applied to predict the multi-output parameters simultaneously. The SHapley Additive exPlanations approach identified the most influential parameters of the composite strength characteristics. The results revealed that the XGBoost model delivered highly accurate predictions, with lower RMSE, MSE, and MAE values, and a higher R² value of 0.9 compared to other ML models. Multi-Output Machine Learning Techniques have proven to be a quick and cost-effective solution, an alternative to time-consuming traditional tests.

Optimizing concrete mix design is essential for advancing sustainable construction practices. Conventional methods for evaluating the compressive strength (CS) of concrete, a critical mechanical property, are often time-intensive and resource-demanding. This study investigates the application of machine learning (ML) models to predict CS of concrete more efficiently, utilizing the Python interface on Google Colab. Multiple regression models have been assessed using performance metrics including mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and the coefficient of determination (R²). A stacked regression (SR) model has been developed by integrating 14 base models, with CatBoost (CB) employed as the meta-learner. The models have been trained and tested on a dataset comprising 1315 samples collected from the Concrete Laboratory at Veer Surendra Sai University of Technology (VSSUT), Burla, using an 80/20 train-test split. To enhance model performance, hyper-parameter tuning via Grid Search and validation through K-Fold cross-validation have been employed. The optimized SR-CB model has achieved superior predictive accuracy, recording an RMSE of 1.95 and an R² of 0.93. Furthermore, SHAP and LIME analyses have been conducted to interpret the feature importance and model behaviour. The model’s generalizability has been validated by predicting the CS of 21 new concrete mixes from literature, resulting in prediction errors ranging from 0.5% to 9.9% and a R² of 0.93. The findings demonstrate that the proposed stacked regression approach significantly improves prediction accuracy and robustness compared to individual models, thereby facilitating more efficient and sustainable concrete mix design with reduced dependence on conventional experimental methods.

There has been an increased push in the research literature pushing the need for sustainable, high-performing concrete. Studies on partial cement replacement with industrial by-products, such as fly ash (FA) and biomedical waste ash (BMWA), have gained traction. However, the models traditionally used to estimate the compressive strength and durability of such concrete systems lag behind in the ability to capture intricate multi-scale interactions when associated with hybrid binder systems. Most of these predictive models are empirical or purely data-driven by neglecting or failing to rely on physical principles and generalization with varying curing conditions and material synergies. This would be addressed by this study with a fully derived ensemble proposing a new framework that incorporates the physics informed and data-driven technique for the prediction and optimization for M25-grade concrete mixes containing FA and BMWA in compressive strength (at 7, 14, 28 and 90 days) and durability prediction. The main psychic module called by this predictive framework is PINN-CSM (Physics Informed Neural Network for Concrete Strength Modeling); it puts physical laws such as Abram’s water-cement relationship and pozzolanic bounds into the loss function of the neural network, improving extrapolative performance and interpretability, achieving an R² ~ 0.96–0.98. Optimization is managed by MODE-STR (Multi-Objective Design Engine for Strength-Tradeoffs and Replacement) using NSGA II for optimal combinations of binders that balance strength, cost, and carbon impact sets. To further strengthen constituent interaction modeling, GNN-CS (Graph Neural Network for Constituent Synergy) leverages a graph-based message passing architecture to capture cross-material synergies while SHAP-FS (Shapley Aware Feature Synthesis) builds interpretable hybrid features with high joint influence on strength sets. It does so by aligning pre-trained models with experimental domains through transfer learning and hydration kinetics ontologies, TLO-CS (Transfer Learning with Ontological Fine-Tuning). This paper also used Random Forest (RF) and Gene Expression Programming (GEP) for evaluating the results under different scenarios. Thus, this integrated framework provides better accuracy and empowers the informed mix design towards low-carbon durable concrete optimized over multiple performance metrics.

Pre-engineered steel buildings (PESBs) are increasingly adopted for industrial applications due to their cost efficiency and rapid deployment. However, ensuring sustainability in PESBs requires a balanced evaluation of economic, environmental, and certification-related goals. This study develops a hybrid multi-objective optimization framework that combines the non-dominated sorting genetic algorithm III (NSGA-III) with multi-objective particle swarm optimization (MOPSO) to simultaneously optimize life cycle cost, embodied carbon emissions, green framework compliance scores, and construction time. A case study of an industrial warehouse in Hyderabad, India, is used to demonstrate the framework, incorporating green building standards such as LEED, IGBC, and GRIHA. The optimization explores alternative design configurations involving material selection, insulation thickness, sheeting type, and bracing systems. The resulting Pareto-optimal solutions highlight trade-offs among key performance metrics, enabling informed decision-making for stakeholders. Sensitivity analysis under varied stakeholder preferences further supports targeted design strategies. Comparative evaluation with other optimization techniques confirms the superiority of the proposed hybrid approach in convergence quality and solution diversity. This study offers a practical decision-support tool for sustainable PESB design, aligning industry practices with climate goals and certification requirements.

This study examines the seismic response of plan-irregular reinforced concrete (RC) buildings with C-, I-, L-, and T-shaped configurations, integrating statistical and machine learning (ML) techniques to quantify the influence of irregularity on structural performance. Focusing on five plan irregularity ratios (L/B, L/B₁, B/L₂, B₂/B, L₁/L), The paper evaluated their impact on key seismic responses—maximum storey drift, roof displacement, roof acceleration, and base shear—under far-fault (FF) and near-fault (NF-PL, NF-NP) ground motions across 16 building models. Pearson correlation analysis revealed moderate to strong linear relationships, with L/B significantly influencing base shear (r = 0.71) and roof displacement, though it lacked depth in capturing shape-specific effects. One-Way ANOVA showed that building shape alone was not consistently significant (p > 0.05, e.g., Max Drift-Y: p = 0.27), but ground motion effects were shape-specific (e.g., T-shape Roof Displacement in Y: p ≈ 0.041). Two-way ANOVA confirmed significant shape-motion interactions, notably for roof displacement (p = 0.012). XGBoost with SHAP analysis identified L/B as the dominant feature (SHAP of 1033.1 for base shear in I-shaped, Y-direction), with B₂/B affecting drift in T-shaped buildings. PCA and K-Means clustering revealed four latent groups, with Cluster 2 exhibiting extreme responses (base shear of 19,789 kN) driven by a high L/B₁ ratio (5.5). These findings advocate for parameter-driven design over shape-based classifications, offering a data-driven foundation for fragility modelling and code refinement in seismic engineering.

This study presents a comparative analysis of reinforced concrete (RC) buildings and composite (steel–concrete) structures subjected to blast loads. Structures were modelled and analysed using extended 3D Analysis of Building Systems (ETABS) 20.0.3 under uniform blast load intensities across varying building heights. Key performance indicators such as maximum storey displacement and overturning moment were evaluated. The results show that composite structures exhibit superior performance, with significantly lower maximum lateral displacements (6300 mm) compared to RC structures (9800 mm), indicating greater lateral stiffness. Storey drift was also notably lower in composite systems 0.09 versus 0.145 particularly in the mid-storey region, while overturning moments were slightly higher in composites due to their increased stiffness. These results demonstrate better deformation control and overall stability of composite structures under dynamic loading. Machine learning algorithms, including AdaBoost, CatBoost, and Random Forest, were applied to predict structural responses. All models performed well, achieving R2 values greater than 0.91.

Structural health monitoring (SHM) of buildings is critically important as it directly affects human safety and economic activities. This paper proposes a real-time system design for SHM to monitor the building’s status through WSNs, make intelligent inferences, and predict the risk. The system architecture consists of multiple stages. The first stage is the transmitting side attached directly to the building’s structure, which comprises several sensors, including ADXL345, SW-420, LVDT, and strain gauge. LoRa wireless communication technology is established to transfer the sensors’ data from the transmitter side to an on-site central node. The central node processes and transmits the data to the cloud via Wi-Fi. The artificial neural networks (ANN) algorithm is employed to classify healthy and abnormal data to determine the damage severity value of the building’s status based on the peak ground acceleration (PGA), which ensures high accuracy in determining the damage value exposed to the building. The system utilizes the ThingSpeak IoT platform, which integrates the trained neural network and central node for storing sensors’ data and damage severity value to enable real-time monitoring. The system was validated using a shake table experiment by applying three PGA values, 0.05 g, 0.15 g, and 0.32 g, to the building model. The results demonstrate that the system is reliable and more effective for damage prediction, achieving a mean absolute error (MAE) of 0.0126 and 0.014 for neural network training and testing, respectively. Moreover, the ANN performed a correlation coefficient (R²) of 0.95892 and 0.95961 for training and testing. The main achievement of this research involves developing an advanced integrated system that combines sensors with an IoT platform and neural networks to track building damage severity in real-time.

Structural health monitoring (SHM) is essential for ensuring the safety and functionality of civil infrastructure. This study presents a deep learning-based approach to SHM in the ASCE benchmark building. To achieve this, the ASCE benchmark building is modelled in the ANSYS environment to simulate its response under various structural conditions, including both undamaged and multiple damaged states. The acceleration data obtained from these simulations is converted into scalogram images using the continuous wavelet transform. These images are employed to train two deep learning algorithms for structural state classification: the Convolutional Neural Network (CNN) and the Alex Net algorithms. Compared to Alex Net, the CNN algorithm excelled at detecting subtle damage patterns. Additionally, MobileNetV2 is employed to evaluate performance under limited data conditions, achieving better classification accuracy. This approach offers a valuable and automated tool for real-time damage identification and decision-making in SHM applications.