Accurate prediction of uniaxial compressive strength (UCS) in recycled concrete aggregate (RCA) is essential for advancing sustainable construction practices. This study presents a data-driven predictive framework based on Deep Neural Networks (DNN) to estimate UCS using a comprehensive dataset comprising 326 literature-derived records and 50 experimentally validated samples. The model was trained using optimized hyperparameters over 1000 epochs and evaluated through a combination of statistical metrics and cross-validation techniques. The proposed DNN model demonstrated superior predictive performance compared to benchmark regression models, including Multilayer Perceptron (MLP), Support Vector Machine (SVM), and Decision Tree (DT), achieving an accuracy of 0.925 with significantly reduced error values. The robustness of the model was further validated using independent experimental data, confirming its generalization capability in real-world conditions. The novelty of this study lies in integrating heterogeneous data sources with a rigorous validation framework, enhancing both reliability and applicability. The findings highlight the potential of deep learning techniques to support efficient material design and decision-making in recycled concrete applications, contributing to more sustainable and resource-efficient construction practices.
A visco-elasto-plastic fiber bundle-chain model for creep of concrete under constant loading is developed in this work. By introducing a Kelvin-Voigt element and a dashpot chain into the fiber bundle-plastic chain model, the proposed model effectively captures both the time-dependent creep and the instantaneous plastic damage of concrete. The Kelvin-Voigt element simulates the recoverable elastic creep, while the dashpot chain characterizes the irreversible plastic creep. Furthermore, by considering the randomness of micro mechanical properties of springs and dashpots, the model conveniently characterizes the mean and stochastic creep behaviors from a microstructural perspective, avoiding the complexity of multi-factor randomness in traditional approaches. It is found that the model predictions agree well with experimental results, demonstrating its reliability. Additionally, a comparative discussion with existing models in the literature confirms the advantages of the proposed approach in simulating both deterministic and stochastic creep behavior.
Full-depth precast deck panels can enhance constructability and productivity while reducing construction costs for highway bridges. This paper presents a finite element model (FEM) for direct shear tests and of precast concrete composite beams with discrete connections made with shear pockets and U-shaped steel connectors. Constitutive laws are proposed for these interfaces to accurately represent the resistance, interface slip, and strain in the U-shaped steel connectors observed during testing. Additionally, a parametric analysis of precast composite beams was conducted to investigate the influence of interface slip, shear interface stiffness, and the degree of shear connection on the flexural strength of composite beams with discrete connections. The results indicate that appropriate constitutive laws are essential for accurately representing the behavior of connections utilizing shear pockets and U-shaped steel connectors. Furthermore, parametric analysis shows that the bending resistance of precast concrete composite beams with discrete connections and low interface stiffness can be estimated using simplified analytical models, independent of the interface stiffness, with results within 15% of those obtained from finite element models.
Bridges are considered one of the most critical components of transportation infrastructure. Thus, an accurate response prediction of bridge structures is crucial for long-term monitoring and safety assessment. To accomplish this, the current study presents a deep learning-based approach for predicting strain responses of the I-35W Bridge at three different locations. For this purpose, the five-year measured data are adopted from this concrete box girder bridge located in Minneapolis, Minnesota. In this approach, the collected datasets are utilized as inputs to the multilevel deep neural networks, namely, deep long short-term memory (D-LSTM), deep gated recurrent unit (D-GRU), and modified generative adversarial networks (GANs). The performance of these networks is also assessed using the root mean square error (RMSE), the mean absolute error (MAE), and the coefficient of determination (R2) indices. The findings reveal that the deep learning-based approach is a computationally effective, promising, and reliable method for accurately predicting the future response of the bridge structure. The comparison of the final results also indicates that the proposed D-GRU model provides the best prediction accuracy and performance among the evaluated models.
The extensive use of Ordinary Portland Cement (OPC) in concrete pavements has become a major concern due to its significant contribution to greenhouse gas (GHG) emissions and environmental degradation. In recent years, the use of various materials to replace and reduce clinker consumption in the production of various types of pavement concrete has expanded. Limestone Calcined Clay Cement (LC3) has gained attention as a sustainable option for reducing environmental pollutants and improving mechanical properties, with a composition of 55% OPC, 30% calcined clay, and 15% limestone powder. In this study, mechanical properties including compressive strength and flexural strength, durability against freeze-thaw cycles, relative dynamic modulus of elasticity, and environmental analysis of pavement concretes made with LC3 and OPC at binder contents of 300, 400, and 500 kg/m3 were comprehensively evaluated and compared. The results of this research showed that LC3 concretes exhibited better compressive and flexural strength at 90 days compared to OPC concretes. Additionally, the durability of LC3 samples against damage caused by freeze-thaw cycles was significantly higher, and they exhibited less reduction in the dynamic modulus of elasticity. Environmental indicators were also examined, revealing that using LC3 in concrete pavements reduced CO2 emissions by up to 36% and significantly improved the Human health, Ecosystems, and Resources indicators. Overall, the results of this study indicate that LC3 has significant potential for constructing pavements with high mechanical properties and a lower environmental impact.
In this study, the optimization of concrete reinforced with iron chips, a waste product from the iron and steel industry was carried out. To this end, the Response Surface Method was used in the experimental design and empirical modelling phases. Genetic algorithm was also used to determine the optimum values. Different concrete samples with water/cement ratios of 0.40, 0.50 and 0.60 were prepared for the experimental studies. During the production of the concrete samples, waste iron chips with different shapes (flat, spiral and mixed) were added to the mix as reinforcement, based on the proportions of 0.25%, 0.50% and 0.75% of the total volume. Compressive strength and ultrasonic pulse velocity tests were performed on the concrete samples prepared in 12 series. Analysis of variance (ANOVA) was used to determine the statistical effects of the parameters on the experimental results. As a result of the study, optimum parameter levels were determined to achieve the desired properties in concrete samples reinforced with waste iron chips.
Although Fiber Reinforced Polymer (FRP) enhances the structural performance of concrete columns, existing predictive models for square and rectangular sections remain inadequate due to corner geometry variations and experimental uncertainties. This study aims to improve the reliability and efficiency of strength prediction for FRP-confined columns, particularly in engineering applications demanding both safety and cost-effectiveness. To address these challenges, the Group Method of Data Handling (GMDH) neural network was employed to develop a predictive model that minimizes reliance on expensive, time-consuming experiments. Several modeling approaches were examined, and the final GMDH-based neural network effectively captured the nonlinear relationships between input and output parameters, providing a robust predictive framework. Model evaluation using standard error metrics indicated strong performance, with coefficients of determination (R2) of 0.88 and 0.85 for training and testing datasets, respectively. Low error values, including Root Mean Square Error (RMSE) of 0.169 and 0.201 and Mean Absolute Error (MAE) of 0.128 and 0.157 for training and testing, confirmed the model's predictive reliability. The results demonstrate a close agreement between experimental and predicted strengths, validating the GMDH-NN as a practical, efficient alternative to extensive laboratory testing for estimating the compressive strength of FRP-confined square and rectangular columns. A significant contribution of this work is the integration of advanced neural network techniques with a comprehensive dataset of 171 specimens, offering improved insights into factors influencing strength and providing engineers with a data-driven, reliable tool for optimizing FRP-confined concrete column design.
Concrete-filled steel tube (CFST) is considered a promising solution for enhancing the structural performance of steel structures. This paper presents the development of empirical model for predicting the ultimate moment capacity and flexural stiffness of composite beams, emphasizing the interaction between the concrete core and steel sections. A comprehensive comparison was conducted between the experimental results collected from past studies and the predicted results from existing standards and analytical methods. The existing design models reveal significant inconsistencies and inaccuracies in predicting the results. To address this issue, Grey Correlation Analysis (GCA) was conducted to address the sensitivity of the key parameters influencing the moment capacity and flexural stiffness of CFST beams. The proposed formula considers the key parameters, including confinement ratio (xi), geometric properties (t/D), materials strength (fck and fy), and the elastic modulus of materials (Es and Ec), to improve the precision and practical applicability of the empirical formula. A reliable group of experimental and numerical database were collected from past research to predict the ultimate moment, initial stiffness, and serviceability-level stiffness. Validation was conducted based on the results collected from past research to ensure the applicability and reliability of the developed empirical formula. Furthermore, the predicted value from new methods were compared with the results obtained from the published literatures. The proposed models demonstrate good agreement and better consistency with the experimental results.
The study aimed to forecast asphalt concrete's dynamic modulus (|E & lowast;|, |G & lowast;|) using hybrid machine learning, combining MLP (Multi-Layer Perceptron), SVM (Support Vector Machine), DT (Decision Tree), RF (Random Forest), LR (Logistic Regression), and AdaBoost classifiers to understand dataset complexities. With 2,238 data points from 2010 to 2023 and diverse asphalt concrete samples, a 70-30 train-testing (70% for training and 30% for testing) split ensured thorough analysis. This dataset's diversity was further enriched by incorporating asphalt concrete samples with varying geometries of 100 mmx200 mm and 100 mmx150 mm (diameterxheight), contributing to a more holistic understanding of the material's behavior. Evaluation metrics included confusion matrix, MAE, RMSE, and R2. Results highlighted predictive models' impact on |E & lowast;| and |G & lowast;| values, especially MLP's accuracy (0.964) and precision (0.955), making it reliable for engineers and researchers. MLP also excelled in the testing dataset (accuracy: 0.964, precision: 0.903). SVM followed with 0.880 accuracy and 0.852 precision. These outcomes reinforced the MLP model's reliability and underscored its potential as an asset in predicting |E & lowast;| and |G & lowast;| modulus values, affirming its practical applicability in geotechnical studies and research endeavors.
Structures in service may be subject to blast loading, which can result in significant damage or even failure of critical structural elements. Understanding and mitigating such effects is crucial for ensuring structural resilience. This study conducts a comprehensive numerical analysis to evaluate the efficiency of carbon fiber reinforced polymer (CFRP) as a reinforcing solution for reinforced concrete (RC) beams under explosive loads. Finite element (FE) models were developed using LS-DYNA to analyze the structural response, failure mechanisms of RC beams under blast conditions. To verify the reliability of the FE models, numerical results were systematically compared with experimental data from existing literature. CFRP reinforcement significantly enhances the load-carrying capacity and energy absorption of RC beams while also reducing mid-span deflection. Given excellent agreement with experimental data, the study further explores the impact of different CFRP reinforcement strategies through numerical analyses, considering key factors such as CFRP thickness, the number of reinforcement layers, and various strengthening configurations. Additionally, numerical simulations were conducted to generate Pressure-Impulse (P-I) diagrams for CFRP-strengthened RC beams subjected to blast loading. These diagrams help establish correlations between blast-induced damage, measured in terms of mid-span displacement, and applied explosive loads, offering valuable insights for structural design and retrofitting.
This study aimed to develop a methodology to predict the bending and shear P-I diagrams for any arbitrary RC columns based on a simple linear interpolation approach, which eliminates the need to conduct multiple blast analyses through rigorous nonlinear time-history approaches. The bending and shear P-I diagrams for the 15 RC column section are incorporated into the database, which covers the typical range of office buildings. The effectiveness of the proposed interpolation approach has been validated through comparisons between the interpolated P-I diagrams and those directly derived from trial-and-error-based blast analysis. The proposed methodology can be effectively used to assess whether to remove or retain the RC member that experienced the explosive loading, when the bending and shear P-I diagrams are not available, and to determine the required section dimensions and reinforcement ratios of RC columns during the explosion-proof design process.
The construction industry is seeking automation-based solutions to overcome labor shortages, improve construction quality, and reduce waste in production. Although 3D concrete printing (3DCP) has shown potential for automated on-site fabrication, the conventional layer-deposition approach faces limitations in reinforcement integration, pumping requirements, size constraints, and surface quality. This study, as an alternative, proposes a slipform-based vertical printing method specifically designed for fabricating precast concrete elements in a single process. A prototype vertical printing device was developed at the lab scale, and the printability of cementitious materials, produced with various water-to-binder ratios and high-range water-reducing admixture dosages, was evaluated through the channel flow tests and printing experiments. Consequently, yield stress criteria were established: An initial yield stress of 85 Pa or lower is required to ensure gravity-driven flow of material into the nozzle, whereas an increase in yield stress to 240 Pa after a 20 min waiting time is necessary to satisfy adequate shape stability for vertical precast printing. The printing materials satisfying both criteria were successfully printed at 30 mm/min. In addition to the printability, surface wrinkles and air voids observed during printing were effectively handled by applying localized vibration (4 V, 60 s), which improved filling ability of the material around the embedded rebar substitute. These results demonstrate the feasibility of the proposed non-layered vertical printing process and provide quantitative rheological criteria for material design in the proposed vertical precast printing method.
Accurate prediction of fire-damaged RC structural behavior is essential for safety assessment; however, traditional numerical analysis becomes computationally intensive due to complex thermo-mechanical responses including non-mechanical strains at elevated temperatures. This study develops ML surrogate models (NN, XGB, LGBM) to predict RC member fire endurance time using a 4.37 million-point dataset generated from P-M diagrams obtained via high-fidelity numerical analyses. Tree-based ensembles outperformed NN in large-data regimes (test error <1%, geometric fitness >96%), providing superior interpretability through feature importance and monotonic constraints. Specifically, by defining the input conditions through a flexible 7-variable framework (B, H, BN, HN, M, P, R), this study enables the direct generation of P-M interaction diagrams under any arbitrary loading scenarios. Frame-level validation comparing predictive model and numerical analysis results on a 1-bay, 1-story RC frame confirmed applicability of proposed model with consistent column failure patterns, enabling very rapid predictions showing similar trends.
Functionally graded materials (FGMs) constitute a category of advanced composites characterized by a continuous variation of mechanical and thermal properties, resulting from gradual changes in composition or microstructure. This gradation enables the design of components with tailored performance for specific engineering applications. In this study, an elasticity-based analytical solution is developed for the static analysis of a functionally graded cantilever beam under a cubic load distribution. In contrast to most existing studies that primarily address uniformly distributed or simple load cases, the present work introduces a new analytical framework capable of capturing the response of graded beams under higher-order distributed loading. The formulation uses the Airy stress function to predict the internal stress fields within the beam, ensuring that both equilibrium and compatibility conditions are rigorously satisfied. By applying the boundary conditions consistent with the cantilever configuration, closed-form expressions for the stress components and deflection are derived. The obtained results show the capability of the proposed approach to accurately capture the stress distribution and deformation response of functionally graded beams under cubic loading. This highlights not only the efficiency of the method but also its novelty in extending elasticity-based solutions beyond conventional load assumptions, providing new insights for analyzing graded structures under higher-order load distributions.
This paper investigates the interfacial slip stresses in continuous steel-concrete composite beams reinforced with externally bonded composite plates. The study employs mathematical formulations to capture the nonlinear behavior of continuous composite beams, accounting for both equilibrium and deformation compatibility throughout all components of the reinforced structure. An analytical nonlinear modeling approach is developed alongside a finite element model to analyze interfacial slip in continuous composite beams. The results demonstrate that adhesive bonding between the concrete slab and steel beam can significantly reduce cracking in the negative moment regions of continuous beams. The theoretical and numerical predictions are validated through comparison with existing experimental and analytical results, enhancing the understanding of the mechanical behavior of continuous composite beams and informing the design of hybrid steel-concrete structures.
This study investigates the nonlinear low-velocity impact response of functionally graded (FG) fluid-conveying pipes (FCPs), addressing critical gaps in existing research. Unlike prior studies on graphene-reinforced beams (Chen and She 2024), our work firstly incorporates fluid-structure interaction effects with graded material properties, enabling more accurate predictions for industrial pipeline systems. The key contributions include: (1) A novel analytical framework combining Euler beam theory, geometric nonlinearity, and fluid velocity coupling; (2) Systematic validation of damping effects on impact resilience, which previous studies have overlooked; (3) Demonstration that initial geometric imperfections exhibit a stronger influence on dynamic response than in homogeneous structures. Through Galerkin discretization and Runge-Kutta integration, we reveal parameter-sensitive regimes that could guide protective designs for high-speed fluid pipelines. These advancements distinguish our work by providing comprehensive solutions for FG materials under coupled mechanical and fluid dynamic loads.
Inspection information plays an important role in time-dependent reliability assessment of deteriorating structures, and various types of inspection information can be incorporated into reliability updating. This study proposes a unified reliability-updating framework that systematically integrates both inequality-type inspection events and equality-type inspection data into the reliability analysis of deteriorating structures. Two illustrative examples are presented to demonstrate the applicability of the proposed framework. The first example illustrates an reinforced concrete (RC) bridge subjected to a coupled corrosion-fatigue deterioration process, in which inspection outcomes for no damage detection, corrosion pit detection, and fatigue crack measurement are investigated. The second example focuses on a steel bridge dominated by fatigue crack deterioration. The effects of multiple inspection events under various inspection strategies, including inspection timing, inspection intervals, inspection order, inspection quality, and correlation among inspection outcomes, are discussed. The results show that detailed inspection data, such as measured corrosion pit depth or fatigue crack size, significantly change the updated probability of failure, whereas qualitative inspection events, such as crack detection without size measurement, provide less constraining information compared to precise quantitative measurements, although they still contribute useful information for reliability updating. The proposed framework offers practical guidance for inspection planning and life-cycle reliability management of aging infrastructures.
Pervious concrete pavements offer stormwater management benefits but are limited by their low mechanical capacity. To address this issue, this paper introduces a hybrid paver composed of a 3D-printed paste frame and post-cast pervious concrete core. A comprehensive experimental program was conducted to investigate the effects of frame geometry (lattice, circular, hexagonal, and diamond), bottom-layer thickness (10-30 mm), aggregate gradation, and post-casting time on density, porosity, infiltration, and flexural performance. The results indicated that the frame geometry and bottom-layer thickness were the dominant factors governing the mechanical-hydraulic balance. Increasing the printed-layer thickness significantly enhanced the flexural strength by increasing the section stiffness and confinement, although it reduced infiltration owing to the formation of deeper constricted flow channels. Circular and diamond geometries provided superior confinement and strength, whereas the lattice and hexagonal patterns maintained higher drainage continuity. The hybrid system developed in this study demonstrated flexural strengths up to 4 MPa, exhibited performance consistent with the functional drainage standards and offering a designable platform to tailor pavement performance.
The durability of the reinforced cement concrete (RCC) structures exposed to open areas, i.e., bridges and marine structures, is affected by the chloride ions. The fly ash, silica fume, and ground granulated blast slag materials are used to reduce the chloride ions penetration. Still, the laboratory procedure for determining the chloride permeability is time-consuming and lengthy. This investigation introduces an optimal performance model to assess the chloride permeability of fly ash concrete. For that purpose, a database of chloride permeability results of 288 concrete specimens has been compiled from the literature. This research employs genetic and particle swarm-optimized relevance vector machine (RVM) models. Moreover, these RVM models have been configured by single and dual kernels. In addition, the extreme gradient-boosting (XGBoost) model has been developed and compared with RVM models. The performance comparison reveals that the RVM1 model has predicted chloride permeability with a performance index of 1.95, root mean square error of 286.8311C, a correlation coefficient of 0.9923, and the variance accounted for of 98.42 in the testing phase, close to the ideal values, followed by XGBoost model. The variance inflation factor (VIF) revealed that the binder and water-to-binder ratio features have considerable multicollinearity. This research also demonstrates that the database and structural multicollinearity highly influence the prediction capabilities of the RVM4 models. The score analysis, regression error characteristics curve, accuracy matrix, computational cost, and reliability analysis confirm that the RVM1 model is an optimal performance model in predicting the chloride permeability of fly ash concrete. This investigation will help concrete designers and engineers to determine the chloride permeability without performing laboratory procedures in mega construction projects.
This research investigates the experimental, optimization and prediction of the mechanical properties of normal and self-compacting concrete by examining varying ratios of gravel, basalt, and dolomite aggregates. Each mix consists of one or two different types of aggregates combined in varying proportions to assess their influence on concrete performance. The study aims to evaluate how these different aggregate compositions influence key mechanical properties, including compressive strength, tensile strength and flexural strength. Experimental tests were conducted on concrete samples with varying proportions of the aggregates, followed by a comprehensive analysis of their performance. The results indicate that self-compacting concrete generally outperforms normal concrete in mechanical properties, with dolomite contributing most to compressive strength, basalt excelling in flexural and tensile strengths, and gravel showing the weakest performance across all cases. Genetic Algorithm (GA), were employed to develop predictive models for the mechanical properties of both normal and self-compacting concrete based on the aggregate ratios. The formula developed through GA demonstrates practical applicability in assessing the mechanical properties of both normal and self-compacting concrete, while accounting for varying ratios of gravel, basalt, and dolomite aggregates. This research provides valuable insights for engineers and researchers in the field of construction materials, contributing to the development of more efficient and sustainable concrete solutions.
