Asian Journal of Civil Engineering最新文献

The study investigates machine learning applications in civil engineering, which are biased towards construction management. The hybrid model was developed for better schedule deviation and budget overrun performance, based on Harris Hawks Optimization combined with Light GBM. Using HHO for feature selection, the model identified the most influencing factors like Project Size, Risk Score, and Change Orders. This optimized the prediction process. This hybrid approach outperformed the traditional machine learning models, including Random Forest and XGBoost, by an optimum RMSE of 15.32 days schedule deviations and $25,840 budget overruns, proving more accurate and efficient. Therefore, this underpins the potential AI-driven solutions for improving project planning, risk mitigation, and decision-making within construction management. Future work will need to refine models as artificial intelligence becomes integrated into practice within civil engineering. Additional predictive variables will be further investigated while extending the approach to other areas of construction management and civil engineering applications.

This study investigates the mechanical properties of lightweight reactive powder concrete (LWRPC) under normal curing conditions, with a focus on grades M70, M80, and M90. The research was conducted in two phases. In the first phase, conventional reactive powder concrete (RPC) was formulated using quartz sand and 0–30% supplementary cementitious materials (microsilica and alccofine), guided by the Elkem Material Mix Analyzer (EMMA) and the modified Andreassen model. In the second phase, lightweight expanded clay sand (LECS) was incorporated to develop LWRPC, and its mechanical properties were assessed. The study developed mix proportions for the specified grades and identified 10% microsilica and 20% alccofine as an effective blend for improving strength and workability, while LECS contributed to a more than 20% reduction in density. The developed LWRPC grades achieved 86–90% of its 28-day compressive strength within 7 days, with an average density of 1893 kg/m³, 22% lower than corresponding normal high-strength concrete (NHSC) grades, resulting in a 35% increase in structural efficiency. The modulus of elasticity of LWRPC was found to be 10% higher than high-strength lightweight concrete (HSLWC) in the literature. Additionally, flexural and splitting tensile strengths revealed improvements of 24% and 63%, respectively, compared to HSLWC, and 11% and 22% relative to NHSC grades. Although LWRPC has a higher cost ($239/m³) approximately three times that of NHSC, the results demonstrate that it offers superior structural performance, positioning it as a high-performance lightweight concrete.

The paper investigates the amalgamation of LightGBM and Enhanced Colliding Bodies Optimization (ECBO) to establish a resilient framework for sustainable interior design optimization in residential projects. The main goal is to harmonize aesthetic appeal, functionality, and energy efficiency by applying modern machine learning and metaheuristic optimization methods. LightGBM was utilized for predictive modeling of essential design outcomes, achieving good prediction accuracy, with R-squared values of 0.892 for energy savings, 0.839 for functional enhancements, and 0.782 for aesthetics. Critical elements, including sustainable materials, project budget, and energy efficiency ratings, surfaced as pivotal influences on design improvements. The ECBO further refined these design elements, yielding a 28.13% enhancement in aesthetic evaluations, a 22.86% gain in functionality, a 41.56% advancement in energy savings, and a 29.17% decrease in carbon footprint. Compared to conventional algorithms such as Particle Swarm Optimization and Genetic Algorithm, the ECBO exhibited enhanced convergence velocity and solution efficacy. This study presents a thorough, data-centric methodology for sustainable interior design, offering an efficient framework for attaining many design objectives in housing rehabilitation.

The use of waste and recycled materials in concrete is one potential solution to lessen the impact of environmental problems from the concrete industry. The purpose of this work is to use machine learning algorithms to forecast and create an empirical formula for the compressive strength (CS) of limestone powder (LP) incorporated concrete. Eight distinct machine learning models—XGBoost, Gradient Boosting, Support Vector Regression, Linear Regression, Decision Tree, K-Nearest Neighbors, Bagging, and Adaptive Boosting—were trained and tested using a dataset that included 339 experimental data of varying mix proportions. The most significant factors were used as input parameters in the creation of LP-based concrete models, and these included cement, aggregate, water, super plasticizer, cement, and additional cementitious material. Several statistical measures, such as mean absolute error (MAE), coefficient of determination (R²), mean square error (MSE), root man square error (RMSE) and mean absolute percentage error (MAPE), were used to evaluate the models. XGBoost model outperforms the other models with R² values of 0.99 (training) and 0.89 (testing), with RMSE values between 0.065 and 4.557. To ascertain how the input parameters affected the outcome, SHAP analysis was done. It was demonstrated that superplasticizer, cement, and SCM significantly affected the CS of limestone powder concrete (LPC) with high SHAP values. By eliminating experimental procedures, reducing the demand for labor and resources, increasing time efficiency and offering insightful information for enhancing LPC design, this research advances the development of sustainable building materials using machine learning.

To better approximate the actual behavior of reinforced concrete structures under static and monotonic loading, we consider the effect of shear, ductility, and the contribution of concrete in tension between two cracks, as well as the location and distribution of stresses and strains with their directions. Based on the model established by Smahi and Bouafia for concrete and its combination with the damage variable for steel (derived from the behavior law for strain-hardened steel), a homogenization law for composite structures has been proposed. The proposed model is essentially based on the theory of continuum mechanics (generalized Hooke’s law), the damage theory of irreversible processes applied to homogeneous and isotropic materials, and the analytical model established by Vecchio and Collins. The latter is applied to reinforced concrete structures in a plane stress state and is extended in this study to a tridirectional stress state. Taking into account the geometric percentage of steel, two independent damage variables (deviatoric and volumetric) have been used to influence the properties of the composite material in the nonlinear domain, and then a law of variation of Poisson’s ratio is proposed. A numerical finite element program has been developed and applied to slabs and beams “with and without stirrups” in three-point and four-point bending tests. The latter, based on secant stiffness, was compared with other existing software, allowing us to verify its performance in the simulation of reinforced concrete elements and to monitor the actual behavior of these structures, both theoretically and graphically, until failure.

One of the main challenges in welding structural sections is selecting the optimal welding sequences to minimize residual stresses and distortions. In welded structural sections, such as built-up steel box sections, residual stresses and distortions can lead to failures due to the non-uniform expansion and contraction of the weld and surrounding materials. This study investigates the impacts of two different multi-pass welding sequences on residual stress and distortion in fillet welded built-up steel box sections, aiming to identify the most effective solution for minimizing residual stresses and distortions in these structural sections. To achieve this, both thermal and mechanical analyses were conducted using the finite element method, implemented in Abaqus software and programmed in Fortran programming language. The numerical study was validated against existing experimental tests documented in the literature, demonstrating good agreement. The analysis revealed that the sequance of welding operations can affect peak residual stresses; with some sequences resulting in lower stresses and distortions. Consequently, an optimal multi-pass welding sequence is proposed to minimize distortion and residual stresses in fillet welded built-up steel box sections.

Trade-off problem requires a balance between the project objectives taken as time and cost, known as the NP-hard optimization problem. Due to this, any metaheuristic algorithm like the arithmetic optimization algorithm (AOA) gaining popularity for its simplicity and fast convergence might suffer from finding the optimal solution(s) when the construction project scale is increasing. To improve the overall optimization ability and overcome the drawbacks of the plain AOA in solving the time–cost trade-off optimization problems, in this study, the generation jumping phase of the opposition-based learning strategy is proposed and integrated with AOA. This enhancement realizes complementary advantages of the opposition jumping rate to avoid falling into the local optimum and premature convergence. Construction engineering projects involving 63, 81, and 146 activities are applied to verify the effectiveness and feasibility of the enhanced AOA. The experimental results reveal that the proposed model is more effective than the plain AOA and other emerging algorithms for simultaneously optimizing the trade-off problems in construction management.

Polymer-modified concrete (PMC) is an advanced building material with more excellent durability, tensile strength, adhesion, and lesser susceptibility to chemical degradation. Recent developments in machine learning (ML) have shown that prediction of compressive strength (CS) of PMC key input factors needed to obtain an optimized mix design are among the areas of applicability of ML. This study used eight machine learning models, which are Decision Tree, Support Vector Machine, K-Nearest Neighbors, Bagging Regression, XG-Boost, Ada-Boost, Linear Regression, Gradient Boosting to predict compressive strength and perform SHAP (Shapley additive explanation) analysis. These hybrid predictive PMC models were developed using a wide-ranging dataset of 382 experimental data points compiled from the literature. A SHAP interaction plot was also used to show how each feature affected predictions on the model outputs. As highlighted in the results, hybrid models had significantly higher performance than conventional models, and the XG-Boost and decision tree model had the highest accuracy. In particular, the XG-Boost and decision tree model reached R² scores of 0.987 for training and 0.577 for testing, proving its remarkable prediction ability for PMC compressive strength. The SHAP analysis confirmed that coarse aggregate, cement, and SCMs had the most significant influence on CS, with all other variables contributing lower values. The Partial Dependence Plots (PDP) analysis allowed a relatively simple interpretation of the contribution of individual inputs to the CS predictions. These results are useful for construction purposes and provide engineers and builders with first-hand knowledge and insight into the importance of individual components on PMC development and performance.

Detecting and quantifying multiple damages in structures remains a significant challenge in structural health monitoring (SHM), particularly in complex civil engineering systems. This study presents an experimental approach for the detection of multiple damages and their severity using the Electromechanical Impedance (EMI) technique. The EMI method, which utilizes piezoelectric transducers, offers a sensitive and reliable means to monitor structural integrity by measuring the coupled mechanical and electrical response of structures under various damage conditions. In this research, multiple damage scenarios were simulated in concrete specimens, and the corresponding conductance signatures were recorded. Particularly shifts in conductance values were analyzed to identify and localize damages. Conventional statistical metrics such as root-mean square deviation, correlation coefficient, mean absolute percentage deviation are employed to quantify the changes in conductance signature. Additionally, a methodology for localizing the damage is presented. Additionally, a severity index based on impedance variations was developed to quantify the extent of damage. The experimental results demonstrate the effectiveness of the EMI technique in accurately detecting, locating, and assessing the severity of multiple damages in complex structural systems. Further machine learning approach viz. artificial neural network model was applied to predict the damages. The data trained an artificial neural network model, which found suitable for predicting multiple damages levels. This approach contributes to enhanced safety and reliability in structural health monitoring (SHM) and sustainable building practices by offering a scalable and sustainable approach for real-time durability assessment, performance of concrete structures, contributing to a more sustainable development.

The coupled Eulerian–Lagrangian (CEL) method has shown good capability to simulate large deformation behavior in the blast response of complex structural and multi-physics systems. However, the published literature has not addressed blast wave characteristics in free open-air space or directly compared the results of such studies with experiments. In this study, the authors performed three-dimensional (3-D) non-linear finite element (FE) analysis of CEL model utilizing ABAQUS/Explicit finite element software to estimate blast wave parameters, peak overpressure (P_so), time of arrival (t_a), positive phase duration (({t}_{o}^{+})), and blast shock wave front velocity (U) in comparison to empirical Kingery–Bulmash free air-blast predictions and recently published small-scale blast field test results. The height of spherical and cubical TNT charges (HOE) is 5.0-m inside a Eulerain domain (ED). The air and trinitrotoluene (TNT) charge are modeled using (C3D8R) continuum solid elements and the Eulerian domain is modeled as a volume element using (EC3D8R). The CEL model results show good agreement with Kingery–Bulmash predictions and experimental data for incident peak-overpressure, time of arrival, and blast shock wave velocity of considered scaled distances. However, the CEL model outcomes of positive phase duration showed a difference from Kingery–Bulmash model as high as 55% due to secondary shock waves moving inward and reflected toward the source of burst. Despite the extensive validation of the Kingery–Bulmash empirical model, direct measurements in open-space indicate that incorporating blast wave propagation phenomena is critical in different explosion scenarios, especially when reflection phenomena are probable. As a practical model of the CEL model, the blast response and damage evolution of a X70 steel pipe subjected to contact pipe bomb charge is investigated. This grade of steel pipe is a reliable material and used in oil and gas transmission pipelines. The post-damage simulation showed wall thickness has significant contribution to improve the response of the pipe and blast-post damage evolution. This study aims to highlight the efficiency of coupled Eulerian–Lagrangian (CEL) technique to simulate blast for a better understanding of wave propagation in free space and wave-structure interaction phenomena when blast waves interact with structures.