Aluminum pipes are widely employed in industrial applications due to their low density, corrosion resistance, and excellent mechanical properties, however, defects such as cracks and corrosion can significantly reduce their structural integrity. This study investigates the percentage variation in natural frequencies of damaged aluminum pipes repaired using composite patches made of glass, carbon, and basalt fiber-reinforced polymers (GFRP, CFRP, and BFRP). Experimental modal analysis was first conducted on intact, damaged, and repaired pipe specimens, revealing frequency reductions of up to 15-25% due to damage, depending on crack length and boundary conditions. Finite element (FE) modeling showed good agreement with experimental results, with discrepancies generally below 6% for the first two vibration modes. Parametric numerical simulations were then used to generate a comprehensive dataset incorporating crack length, patch type, number of layers, and fiber orientation. Based on these data, an artificial neural network (ANN) model was developed to predict frequency variation in repaired pipes, achieving a high prediction accuracy with correlation coefficients exceeding 0.95 and low validation errors. The results demonstrate that composite patch type and stacking sequence play a key role in stiffness recovery, with CFRP patches providing the highest frequency restoration. The proposed framework, combining experimental vibration measurements, numerical modeling, and ANN-based prediction, offers an effective and reliable tool for assessing and optimizing composite repair strategies for damaged aluminum pipes.

This study aims to evaluate the effect of an eco-friendly and cost-effective surface treatment method based on the use of commercial sodium bicarbonate on the physicochemical, morphological, and tensile properties of jute yarns. Specifically, raw jute yarns were treated with sodium bicarbonate (NaHCO3) solutions of different concentrations (5

This work presents a combined experimental and numerical investigation of 3D-printed long carbon fiber composites, with a focus on developing and validating an advanced finite element model as the primary contribution. Tensile and three-point bending specimens were fabricated using a continuous fiber additive manufacturing process with two different fiber volume fractions: 41

This study presents a novel approach to optimizing column mass in complex reinforced concrete structures by integrating multiobjective optimization with variants of the K-means Optimizer (KO) algorithm. A key technical contribution of the research lies in the development of a two-way computational system that establishes a direct connection between MATLAB and ETABS via the Open Application Programming Interface (OAPI). This integration enables MATLAB to function as an independent computational platform capable of automatically updating design parameters within the finite element model (FEM) in ETABS, eliminating the need for manual interaction with the software's graphical interface. Building upon this automation framework, objective functions are formulated with constraints related to structural drift and interstory displacement. The search for a robust and efficient optimization algorithm is conducted through performance evaluations of KO variants across 50 benchmark test functions. To validate the effectiveness and practical applicability of the proposed methodology, two case studies are carried out: (1) a 15-story planar reinforced concrete frame subjected to both gravity and lateral loads and (2) a realistic 10-story three-dimensional structural system under comprehensive load combinations. The results from both cases confirm that the proposed method satisfies all structural constraints while achieving near-optimal mass reduction. Furthermore, the study introduces an enhanced variant of the KO algorithm, demonstrating strong potential for application not only in structural optimization but also in solving other complex engineering optimization problems.

This study introduces an optimal polar wavelet transform for damage detection in laminated composite circular plates. The core concept behind the proposed method lies in strategically selecting the most informative polar detail signals to extract damage-related information from vibration mode shapes. To guide this selection process, a minimum energy ratio (MER) index is introduced as a benchmark for identifying the most effective polar wavelet function by evaluating the energy balance between the approximation signal and the various detail signals. To generate reliable vibration data for the proposed damage detection framework, the mathematical model for the circular plate has been developed based on the first-order shear deformation theory (FSDT), neglecting damping effects to simplify the analysis. By introducing central and non-central finite elements, the solution is calculated using the finite element method (FEM), and a convergence study is conducted to verify the accuracy of the numerical model. The effects of different parameters such as location of damage, level of damage, noise, number of layers, and lay-ups on detecting the damages of laminated composite circular plates are considered numerically and experimentally. Results show that the developed optimal wavelet transform can more accurately predict the location of the damage compared to the traditional polar wavelet transform.

This paper introduces a novel deep learning framework for predicting the normalized and non-dimensional deflection of functionally graded composite plates subjected to sinusoidal loading. The proposed approach integrates a deep neural network (DNN) with a novel enhanced whale optimization algorithm (EWOA) to optimize deflection predictions considering mechanical parameters as input data, including stress, strain, plate geometry, and boundary conditions. The deflection outputs are expressed in both normalized and non-dimensional forms, demonstrating a robust and generalizable prediction model applicable to various structural configurations. During the training phase, the proposed EWOA significantly enhances convergence efficiency and prediction accuracy by introducing two key improvements: chaotic initialization and an adaptive leader mechanism. The EWOA-DNN model is trained using analytically derived deflection datasets, exhibiting strong adaptability to changes in material gradation and loading scenarios. Comparative studies confirm that the suggested hybrid framework outperforms conventional optimization-based models, creating an effective and reliable artificial intelligence (AI)-driven tool for structural design, computational mechanics, and the analysis of functionally graded composite materials.

PurposeThis study aims to facilitate the selection of an appropriate zinc-rich coating formulation, which remains challenging, particularly given the wide range of zinc-rich paints available. Therefore, the authors proposed a robust approach to produce an optimal zinc-rich epoxy (ZRE) coating. This strategy aims to use optimization techniques in conjunction with model reduction. This approach is based on the percentages of zinc, silica, epoxy and vinyl, and the immersion time in 3.5% NaCl, as decision variables to maximize anti-corrosion performance.Design/methodology/approachElectrochemical Impedance Spectroscopy (EIS), Open Circuit Potential (OCP) measurements and X-ray diffraction (XRD) patterns of the corrosion products were used to characterize the anticorrosion properties of each ZRE paint formulation, generating a comprehensive data set to support optimization procedures. Accordingly, the Proper Orthogonal Decomposition combined with the Radial Basis Function (POD-RBF) approach, in addition to Exponential Trigonometric Optimization (ETO) and Cuckoo Search (CS) algorithms, were used to perform the optimization.FindingsThe results confirm that all 11 formulations exhibit both sacrificial cathodic protection and a barrier effect, mainly governed by zinc oxidation. Furthermore, silica and vinyl resin reduce excessive zinc consumption and enhance barrier performance. ETO is an efficient and feasible methodology for predicting the best ZRE coating with the best anti-corrosion property, compared to CS.Originality/valueThis study introduces an original synthesis of the anticorrosion behavior of zinc-rich coatings by jointly varying pigment and resin compositions. Coupling anticorrosion analysis with optimization and inverse-problem approaches enables the direct prediction of coating formulations that meet targeted corrosion performance, while reducing the number of experiments required.

In this paper, a combined experimental–computational combined method for damage quantification and stiffness recovery evaluation of glass fiber-reinforced polymer (GFRP) and basalt fiber-reinforced polymer (BFRP) patch-strengthened aluminum plates is introduced. Central-hole damage was used to mimic localized loss in stiffness, and recovery potential of both patches was investigated through experimental modal analysis. The initial four natural frequencies obtained for undamaged, damaged, and patched configurations were used as input parameters for the training of a new hybrid artificial intelligence (AI) model SSA-SCA-XGBoost where the Salp Swarm Algorithm (SSA) offers effective global search and Sine Cosine Algorithm (SCA) optimizes local search for hyperparameter tuning of XGBoost. Numerical finite element modeling confirmed experimental trends and enabled composite–structure interaction interpretation. The findings demonstrate that patches of GFRP show enhanced recovery of stiffness and patches of BFRP show enhanced damping characteristics. The new hybrid SSA-SCA-XGBoost model also attained better predictive accuracy than the traditional metaheuristic–XGBoost hybrids with negligible prediction error in a large volume of damage range. The research verifies the dual role of composite patching as a high-strength structural retrofitting method and advanced hybrid AI modeling as an accurate, large-scale quantification tool for composite-strengthened systems.

In recent years, with the explosion of the 4.0 industrial revolution, terms such as machine learning (ML) have become familiar and are increasingly widely applied in the engineering field. This study focuses on proposing two new hybrid models named exponential-trigonometrie optimized extreme gradient boosting (ETO-XGBoost) and whale optimization algorithm extreme gradient boosting (WOA-XGBoost), which are developed based on extreme gradient boosting combined with exponential-trigonometric optimization and whale optimization algorithm. A data set has been built and analyzed using ABAQUS software, and combined with data from the empirical formula to construct a training data set for the proposed models. The two proposed hybrid models are compared with the available models including Kolmogorov-Arnold networks (KAN), artificial neural networks (ANN) and Eurocode 4 standard through important statistical indices such as mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE) and correlation coefficient R. The analysis results show that the ETO-XGBoost model and the WOA-XGBoost model are the most effective ML models when compared with other models. The correlation index R of both models reached very high values (0.9963 for ETO-XGBoost and 0.9956 for WOA-XGBoost, respectively). At the same time, the error indexes of the ETO-XGBoost model were the smallest among the compared models, with MAPE = 63.3738, MAE = 47.4643 and RMSE = 1.8221; meanwhile, the WOA-XGBoost model had corresponding indexes of MAPE = 67.8956, MAE = 49.1825 and RMSE = 1.9040. Besides, the predicted data from the ETO-XGBoost model and the WOA-XGBoost model showed the highest similarity with the actual data in predicting the axial strength of concrete-filled steel tubular (CFST) columns. Therefore, the ETO-XGBoost model and the WOA-XGBoost model can be considered as a powerful and accurate tool in predicting the compressive strength of CFST columns.

This paper introduces a robust damage detection strategy for π-shape laminated composite frames. A machine learning-based two-dimensional wavelet transform scheme processes one-dimensional signals by converting them into a two-dimensional format, enabling analysis of the signal and its details using a two-dimensional discrete wavelet transform in both horizontal and vertical directions. The proposed method transfers the damage information hidden in the signal from the one-dimensional graph domain to the pixel domain to accurately locate the damage position in the structural mode shape signals. A neural network is used to select the optimized vanishing moments in terms of wavelet coefficient properties. Results show that the proposed machine learning-based two-dimensional approach outperforms the traditional one-dimensional wavelet transform in detecting damage. The quantitative evaluation demonstrates high performance for the proposed GMDH model, achieving regression indices (R) of 0.8926 and 0.9091, along with Mean Squared Errors (MSE) of 0.1191 and 0.1056 for the training and testing phases, respectively, so that GMDH algorithm effectively predicts vanishing moments of the wavelet function with minimal error.

This work introduces two hybrid optimization frameworks for composite laminate blended problems involving discrete design variables and manufacturing constraints. In addition to reducing structural mass, the proposed frameworks aim to preserve layer continuity between adjacent multilayer regions by generating stacked sheet chains from a common guide layer, thereby improving manufacturability and reducing the complexity of layer dropping. The proposed approaches, referred to as ETO-SM and GA-SM, integrate metaheuristic optimization with surrogate modeling to improve computational efficiency. In these frameworks, the ETO and GA algorithms govern the generation and evolution of stacking sequence tables, whereas a pre-trained surrogate model is employed to rapidly estimate the corresponding thickness distributions. The performance of the frameworks is examined through three representative case studies, including the 18-panel horseshoe benchmark, a cantilever plate, and the main spar of a wind turbine blade. Numerical results indicate that both ETO-SM and GA-SM exhibit faster convergence, improved robustness, and lower optimal mass than original ETO and GA versions across all three case studies. In the wind turbine blade main spar optimization under multiple design constraints, the GASM framework yields the lowest structural mass of 156.45 kg with 1498 laminate layers, whereas the ETO-SM framework attains a comparable design with a mass of 162.88 kg and 1573 layers. These results indicate the effectiveness of the proposed approachs and support its applicability to a broad range of plate-type structural design problems.

Accurately predicting the Depth of Penetration (DOP) is essential for understanding the impact behavior of Functionally Graded Composite Materials (FGCMs) under high-velocity conditions. However, the nonlinear dynamic behavior governing FGCMs, coupled with the high cost and time-consuming nature of experimental tests, make precise DOP estimation a challenging task. This study proposes an efficient predictive framework that combines finite element (FE) simulations with artificial neural networks (ANNs) to overcome the constraints of conventional numerical and experimental approaches. This work focuses on applying the Whale Optimization Algorithm (WOA) and the Arithmetic Optimization Algorithm (AOA) as independent optimization strategies to improve the control parameters during the initiation phase, avoid convergence to local optima, and accelerate the learning process. Finite Element (FE) simulations were used to generate input-output datasets, which were used to train three ANN models: conventional back-propagation (BP), WOA-ANN, and AOA-ANN. The optimized ANNs accurately captured the nonlinear mapping between impact velocity ( V (imp )) and DOP. Results confirm that WOA-ANN model outperformed both AOA-ANN and BP models, achieving lower mean square error (MSE) and higher correlation coefficients. Moreover, the optimized ANNs provide reliable predictions with a limited number of FE simulations, considerably reducing computational effort while maintaining accuracy.

This experimental study evaluates the influence of recycled metallic fibers (MF) from steel machining waste and polypropylene grids (PPG) on the flexural behavior of concrete slabs. The metallic fibers are randomly dispersed throughout the concrete, while the PPG grids (with fine and large meshes) are positioned through the slab thickness. Three-point bending tests were conducted on slabs measuring 250 × 500 × 70 mm. Results show that metallic fibers at a fiber content of 0.8 _f = 0.8

When using the continuous wavelet transform (CWT), wavelet coefficients in higher scales cannot detect damages and singularities in the structural signal; this is a weakness of CWT. This paper introduces a greedy wavelet transform (GWT), an enhanced version of the one-dimensional continuous wavelet transform, for improved damage localization. In this method, first, the CWT wavelet transform is applied to the damaged structural signal. Then the standard deviation of each scale is calculated from the wavelet coefficients to obtain the standard deviation vector of the wavelet coefficients. A Greedy algorithm finds the scale corresponding to the optimal wavelet coefficients with the lowest standard deviation, so that in the next step, these optimal wavelet coefficients are re-entered into the CWT. This process is iterated until the desired solution is obtained. Experimental validation on a polyurethane sandwich beam under various damage scenarios and mode shapes, using modal analysis for signal acquisition, demonstrates the GWT’s effectiveness in damage detection. The GWT consistently outperforms the conventional wavelet transform across all tested mode shapes and damage scenarios.

This study presents a numerical solution for the thermal bending analysis of laminated bio-inspired helicoidal composite plates using isogeometric analysis integrated with refined zigzag plate theory (IGA-RZT). The sinusoidal thermal variation across the plate thickness is considered as an external load. By employing refined zigzag theory to simulate the displacement field of laminated plate, the present IGA-RZT solution effectively captures the layerwise mechanical behavior by accounting for the discontinuities in the first derivatives of in-plane displacements across interfaces layer along the thickness direction, in accordance with specified C0z-requirement. Four distinct helicoidal architectures including helicoidal recursive (HR), helicoidal exponential (HE), helicoidal semicircular (HS), and helicoidal linear (HL) are examined to assess their structural behavior under thermal loading. The influence of sequential parameter, number layers, geometry aspect ratio, thermal expansion ratio and length-to-thickness ratio on thermal bending responses is thoroughly analyzed. Numerical results confirm the accuracy and reliability of the IGA-RZT method and reveal the mechanical response of bio-inspired spiral designs under thermal environments. The new findings of this study provide a meaningful insights that can guide future research on the laminated bio-inspired structures.

In this paper, a comprehensive modal analysis is conducted, utilizing experimental and numerical approaches to investigate and predict the influence of the damage size on structural behavior. A finite element (FE) model is created from different loading conditions of the glass fiber reinforced polymer (GFRP) composite pipe with a fixed damage size. The model is subjected to three pressure levels (10, 20, and 30 bar). Next, a dynamic frequency load is then added to the FE model to visualize the different frequency mode shapes and their values under different loading conditions on a pressurized and unpressurized composite pipe. The results were collected for the proposed training models. A Kolmogorov Arnold model is employed to develop a robust predictive model that accurately estimates these geometrical and mechanical parameters by analyzing vibration data from various scenarios. The results show a significant correlation between the actual and predicted results across a wide range of test settings and scenarios, and the precision of the predictions is increased by combining data-driven methodologies with classical modal analysis.

In this work, a newly numerical approach for the global and local analysis of laminated composite plates, incorporating size-dependent effects through refined zigzag kinematics and nonlocal strain gradient theory, is developed. By applying the principle of virtual displacement, we derive a weak form of the equilibrium equations based on the nonlocal strain gradient theory, resulting in a local continuum model. This model inherently accounts for deviations in stiffness due to material inhomogeneity and interatomic forces, particularly at small scales. The refined zigzag theory is used to model the displacement field of the plates, which is numerically approximated using the NURBS-based isogeometric analysis technique. This approach allows us to account for the size-dependent characteristics of the laminate at micro/nanoscale through higher-order nonlocal parameters and nonlocal gradient length coefficients. We validate the proposed model for static and free vibration analyses by comparing our numerical results with those reported in the literature. As an original contribution, this research extensively explores the size-dependent effects on the layerwise local responses of laminated composite plates under static loading conditions with various stacking sequences and thickness ratios being investigated.

In the present study, mechanical behavior of Ceresit 20 mortar is investigated through non-destructive experimental modal testing on a mortar beam. The objective was to determine the Young's modulus (E) in three principal directions longitudinal, transverse, and vertical in terms of bending frequencies. Torsion mode data was utilized in order to calculate the shear modulus (G) and to calculate Poisson's ratio (ν). A finite element model was used for validation, with very good correlation with experiment. Besides, a new Machine Learning (ML) model, Self-Attention Interpretable Neural Transformer optimized by Honey Badger Algorithm (SAINT-HBA), was introduced to predict crack depth of mortar beams. Utilizing bending and torsion frequencies of failed specimens, the model predicted single and multiple crack scenarios very well. SAINT-HBA outperformed hybrid SAINT models that used Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Artificial Ecosystem-based Optimization (AEO) and was found to be useful for structural health monitoring applications.

This paper presents a novel Multi-Objective Flow Direction Algorithm (MOFDA) for complex engineering optimization problems. The key innovation is a hybrid leader selection mechanism, which replaces the conventional roulette wheel selection and significantly enhances convergence and diversity in identifying Pareto-optimal solutions. The proposed MOFDA is rigorously evaluated on 31 standard benchmark problems and 11 constrained engineering design cases-including truss optimization, welded beam design, and a large-scale steel frame structure-to assess its accuracy, stability, and solution diversity comprehensively. Comparative studies with state-of-the-art multi-objective algorithms such as MOMVO, MOMSA, MSSA, and MOGNDO further highlight the strong performance of MOFDA. In addition, MOFDA is integrated into a MATLAB-SAP2000 framework and applied to the real-world structural optimization of the Dong Bai ferry terminal steel frame in Vietnam. The results show that MOFDA consistently achieves competitive or superior outcomes on benchmark functions, delivers substantial weight reduction, and improves structural efficiency in engineering applications. These findings demonstrate both the proposed approach's technical novelty and practical effectiveness. Source codes of MOFDA is publicly available at .

In recent years, with the explosion of the industrial revolution 4.0, terms such as artificial intelligence (AI) have become familiar and increasingly widely applied in the engineering field. This study focuses on the study and evaluation of AI models to predict the axial strength of concrete-filled steel tube columns (CFST). In particular, this study introduces and highlights a new AI model, Kolmogorov-Arnold Networks (KAN), and compares its performance with the previously existing AI model, support vector regression (SVR), along with Eurocode 4. A large dataset consisting of two types of CFST columns (hollow circular and hollow square CFST columns) with different concrete strengths was created using ABAQUS software. The AI models were evaluated based on important statistical indices such as MAPE, MAE, RMSE, and correlation coefficient R. The analysis results showed that the KAN model was the most effective AI model when compared with other models. The R indices were always greater than 0.9 and the MAPE, MAE, RMSE indices were the lowest among the compared models. At the same time, the predicted data from the KAN model showed the highest similarity with the actual data in predicting the axial strength of four types of CFST columns. Therefore, the KAN model can be considered as a powerful and accurate tool in predicting the compressive strength of CFST columns.