The load-bearing performance of train bodies under complex service conditions requires accurate evaluation, necessitating substantial analysis of the damage and fracture behavior of thin-walled structures with heterogeneous materials under complex stress states. To address this requirement, this study focuses on MIG-welded thin-walled 6005A-T6 aluminum alloy and proposes a parameter identification method based on the bilevel parallel optimization principle. The welding regions were characterized through metallographic and microhardness tests, and specimens were designed with pre-crack tips located in various welding regions. This enabled the calibration of material parameters from the elastic to the fracture stages for each welding region. By smoothing the material properties at the boundaries of the welding regions based on surface interpolation principles, the complex fracture behaviors, such as slant fractures and V-shaped fractures, were successfully represented. The predicted load–displacement curves closely matched the experimental results, with a relative error in peak force prediction within 8%. The proposed damage characterization method effectively captures material deformation behavior and accurately predicts fracture performance, offering potential refinements to current standards for welding crack propagation tests.

Physics-based material models often require redundant testing and repeated calibration to characterize the mechanical behavior of materials exhibiting similar constitutive responses and deformation mechanisms, such as those produced by the same manufacturing process with varying parameters. To address this limitation, this study proposed an adaptive neural network (ANN) material modeling framework that bypasses explicit constitutive formulations. Training data with prior physical knowledge was generated through a physics-based model calibrated by both experimental and simulation data. Leveraging transfer learning, an evolutionary algorithm was introduced to extract and fine-tune neural network parameters by solving a non-convex optimization problem, thereby enabling heterogeneous model transfer with only limited experimental data. The framework was validated through tensile, punch, and bending tests on 2A14-O aluminum alloy plates with varying rolling thicknesses. Results demonstrated that the ANN model accurately captures material behavior under different processing conditions. At the source rolling thickness, its predictive accuracy was comparable to that of traditionally calibrated physics-based models, with relative L²-norm errors within 5 % for tensile specimens. At the target rolling thickness, the predicted force–displacement curves for validation tensile specimens yielded an average L²-norm error of 8.31 % compared to experimental data, reflecting a 14.07 % improvement in accuracy over models that neglect thickness-induced material variation. The proposed approach provides a generalizable and cost-efficient modeling framework for materials with similar mechanical behavior, substantially reducing experimental and calibration efforts while offering strong potential for engineering applications.

This study focuses on the utilization of high-frequency mechanical impact (HFMI) treatment for rehabilitating pre-fatigued steel bridge components. It incorporates time of flight diffraction (TOFD) for precise crack depth measurement, alongside strain range drop monitoring to enhance assessment accuracy. The experimental setup involves fillet weld specimens with cope hole geometry, using S355MC steel. The HFMI treatment process employs 3-mm diameter pins to achieve an HFMI indentation depth of 0.2 mm. The study demonstrated that HFMI treatment effectively extends the fatigue life of steel bridge components, showing significant improvements for cracks up to 1.2-mm deep. TOFD measurements, validated against manual optical measurements, consistently indicated crack depths within ± 0.1-mm accuracy. This precision is critical for assessing the HFMI treatment’s effectiveness in repairing pre-fatigued structures. The strain range drop method was used as a stop criterion to evaluate crack depth in real time, effectively reducing the number of TOFD measurements required during fatigue crack growth testing. The experimental results showed that HFMI treatment could improve fatigue life, moving specimens’ performance well above the IIW recommended FAT125 curve for treated steel details. In conclusion, this investigation confirms the significant potential of HFMI treatment for extending the life of pre-fatigued steel bridge components. The combined use of TOFD and strain range drop monitoring provides a robust framework for accurately assessing and optimizing HFMI treatment.

This study addresses the limitations of current parametric equations and artificial neural networks (ANNs) in accurately predicting the stress concentration factor (SCF) of fillet welded joints stemming from the simplification of their real weld profiles. To improve the accuracy, this study introduces bagged trees for estimating local stresses. The dataset used as the foundation for training the bagged trees is extracted from the actual weld geometry of T-shaped joints. It is created via a digitalization process involving the extraction of actual geometric parameters from the joints, which are transformed into finite element models (FEMs). These models are then employed to determine the ratio between the simulated sectional stress and the nominal stress (σsec/∆σnom) under an axial loading condition. A comprehensive comparison is carried out among existing parametric equations, ANNs, and the proposed bagged trees. The results emphasize the inadequacy of idealized geometry models in accurately determining local stresses for real weld profiles. In contrast, bagged trees are a promising method for accurately computing sectional weld stresses (σsec) within real weld geometry.

Aluminum 6061-T6 sheets with a Ti–6Al–4V titanium alloy interlayer were investigated using friction stir welding (FSW) to achieve strong bonding. The influence of welding parameters on welding quality and strength was assessed by varying rotational speeds and traverse speeds, the optimal welding conditions. Welded samples with a cross section of 28 mm and excellent surface smoothness were prepared and analyzed to measure the residual stress using X-ray diffraction (XRD) techniques. This study investigated the effect of tool geometry and type on residual stresses in welded specimens and highlighted the importance of choosing the appropriate tool geometry and type to minimize residual stresses. Furthermore, finite element simulation of the FSW process was conducted using a thermal modeling approach to calculate the heat generated and predict residual stresses using ABAQUS software. Comparison of the residual stress values obtained from numerical simulations with experimental measurements demonstrated the model’s ability to predict residual stresses in FSW adequately. The experimental and numerical results revealed that an increase in rotational speed and tool feeding led to higher stresses in the welded region due to an increased thermal gradient. Examination of the microstructure shows that during the welding process, the weld cross-section has become smaller than the base metal. The ultimate tensile strength and microhardness obtained in optimal conditions were 245 MPa and 108.2 HV, respectively. Examining the fracture surfaces from the tensile tests showed the soft fracture type, which is characterized by the presence of holes and depressions in the three-layer sheet.

Plate thickness influences the fatigue performance of welded components. In fatigue design standards and recommendations, the thickness effect and fatigue strength reduction have been considered by modifying the S–N curve for plates thicker than a reference thickness. However, increasing fatigue strength due to the thinness effect is often disregarded. The current study focuses on the size effect in fatigue of butt welded and non-load carrying cruciform welded joints under constant amplitude tension load. Literature data is evaluated using the effective notch stress method with a reference radius of 1 mm, which is used for all finite element models to ensure that FAT-value corresponding to a 1 mm notch radius remains constant across all models. A probabilistic assessment of the results using the weakest-link theory is applied to improve the prediction accuracy of thinner members outside the recommended thickness range of the used radius. The method reduces the S–N data scatter in comparison to the variation of test data and shows applicability also for thinner members. A comparison of the size effect for the current method with extrapolated values from standards and recommendations shows a difference in the size effect for thinner members.

The objective of this study is to investigate defect tolerance of HFMI-treated welds in bridge application for life extension based on numerical FE simulations. Following the authors’ previous study on 3D HFMI simulation to the rat-hole welds, 3D crack propagation analysis (CPA) based on linear elastic fracture mechanics (LEFM) was performed, using the HFMI-treated rat-hole welds including different defect depths in the HFMI-treated welds. At first, compressive residual stress introduced by the 3D HFMI simulation was considered over these crack faces in the HFMI-treated welds. Crack opening and closing stress at the crack tip in the HFMI-treated weld were investigated considering the crack opening-closing behavior. The results indicate that cracks in the HFMI-treated welds would propagate after the HFMI-treated crack surface opened. Then, 3D CPA based on LEFM was carried out under fatigue load. The results show that it is difficult to represent crack propagation behavior in a deeper region where the given residual stress distributions transition from compressive to tensile residual stress. The calculated fatigue life results demonstrated that the life extension may be obtained effectively when HFMI treatment is performed to pre-fatigued welds including fatigue cracks that are shallower than 1.5 mm in depth.

A probabilistic framework is developed utilizing a two-scale modeling approach to efficiently consider the material variability that is typical for composite materials. The modeling integrates a macro-scale model with effective elastic properties extracted from micro-mechanical simulations. Using a weakest link modeling approach for fatigue assessment the combined effects of defects on fatigue strength in a Carbon Fiber Reinforced Polymer (CFRP) material can be investigated. A full fatigue test program is presented and is used to calibrate the probabilistic fatigue model. By including material variability in the numerical model, the calibrated probabilistic model improves the fatigue life prediction.

The study presents a two-scale modeling approach allowing for an efficient fatigue strength evaluation on a macro scale considering a micro-mechanical defect characterization of a Carbon Fiber Reinforced Polymer (CFRP) material. The modeling approach integrates a macro model with the effective elastic properties from micro-mechanical simulations considering voids. This enables the analysis of defects’ influence on material fatigue strength using a probabilistic weakest link approach. A CFRP laminate with a cross-ply layup was investigated. Two simulation case studies demonstrate the effect of void content and size on the characteristic fatigue strength. An experimental investigation was conducted testing the laminates in tension–tension fatigue verifying the predicted numerical behavior. The numerical models identify a difference in the characteristic fatigue strength consistent with the fatigue test results. It is numerically concluded that the investigated CFRP material's fatigue strength is affected by the presence of voids and even with only a slight difference in the global void volume fraction a scatter in fatigue strength is identified.

Aerospace welds are non-destructively evaluated (NDE) during manufacturing to identify defective parts that may pose structural risks, often using digital radiography. The analysis of these digital radiographs is time consuming and costly. Attempts to automate the analysis using conventional computer vision methods or shallow machine learning have not, thus far, provided performance equivalent to human inspectors due to the high reliability requirements and low contrast to noise ratio of the defects. Modern approaches based on deep learning have made considerable progress towards reliable automated analysis. However, limited data sets render current machine learning solutions insufficient for industrial use. Moreover, industrial acceptance would require performance demonstration using standard metrics in non-destructive evaluation, such as probability of detection (POD), which are not commonly used in previous studies. In this study, data augmentation with virtual flaws was used to overcome data scarcity, and compared with conventional data augmentation. A semantic segmentation network was trained to find defects from computed radiography data of aerospace welds. Standard evaluation metrics in nondestructive testing were adopted for the comparison. Finally, the network was deployed as an inspector’s aid in a realistic environment to predict flaws from production radiographs. The network achieved high detection reliability and defect sizing performance, and an acceptable false call rate. Virtual flaw augmentation was found to significantly improve performance, especially for limited data set sizes, and for underrepresented flaw types even at large data sets. The deployed prototype was found to be easy to use indicating readiness for industry adoption.