Reactor pressure vessel (RPV) steels are critical for maintaining the structural integrity and safety of nuclear reactors, designed to endure extreme conditions over prolonged operational lifetimes. Evaluating the mechanical properties of RPV steels frequently involves tests with sub-sized specimens, due to size constraints associated with irradiated materials. However, the reduced specimen dimensions introduce a size effect that alters material behavior and requires correlating the test results to full-sized specimens. Although numerous correlation methods have been previously proposed, they are typically applicable to specific test conditions. To address these challenges, this study introduces a public dataset of 4,961 Charpy impact test records for RPV steels. The dataset was compiled through a comprehensive literature review and incorporates data from 109 peer-reviewed publications. It provides detailed information on material composition, manufacturing treatments, specimen dimensions, testing conditions, and test results. The primary objective of the dataset is to advance the understanding of specimen size effect in Charpy impact testing, and support studies for validating existing methods and developing data-driven approaches for test results correlation.
Lightweight structural materials are required to increase the mobility of fission batteries. The materials must feature a robust combination of mechanical properties to demonstrate structural resilience. The primary objective of this project is to produce lightweight structural materials whose strength-to-weight ratios exceed those of the current widely used structural materials such as 316L stainless steels (316L SS). To achieve this, advanced modeling and simulation tools were employed to design lattice structures with different lattice parameters and different lattice types. A process was successfully developed for transforming lattice-structures models into Multiphysics Object Oriented Simulation Environment (MOOSE) inputs. Finite element modeling (FEM) was used to simulate the uniaxial tensile testing of the lattice-structured parts to investigate the stress distribution at a given displacement. The preliminary results showed that the lattice-structured sample displayed a lower Young’s modulus in comparison with the solid material and that the unit cell size of the lattice had a minimal effect. The novelty here is to apply up-front modeling to determine the best structure for the application before actually producing the sample. The approach of using modeling as a guiding tool for preliminary material design can significantly save time and cost for material development.
8 July 2026 in Frontiers in Manufacturing Technology
Introduction: Experimental tensile testing of sub-sized nuclear structural materials is often limited by the high cost, time requirements, and scarcity of available data. These constraints lead to sparse and imbalanced datasets that challenge the development of reliable predictive models. This study investigates the effectiveness of machine learning (ML) algorithms and data augmentation strategies for predicting tensile properties of sub-sized Stainless Steel 316 specimens. Methods: A dataset comprising 396 tensile testing records and 45 input features describing material composition, processing history, irradiation conditions, specimen geometry, and testing parameters was assembled from the literature. Nine ML algorithms were evaluated for predicting ultimate tensile strength, yield strength, total elongation, and uniform elongation. Two data augmentation methods, Generative Adversarial Networks (GAN) and Synthetic Minority Oversampling Technique for Regression with Gaussian Noise (SMOGN), were applied at multiple augmentation ratios. Model performance was assessed using five-fold cross-validation with Pearson correlation coefficient (r), coefficient of determination (R 2 ), and root-mean-square error (RMSE). Results: Tree-based ensemble methods achieved the highest predictive performance across most tensile properties. Random Forest and Extreme Gradient Boosting (XGBoost) attained Pearson correlation coefficients exceeding 0.98 for yield strength and uniform elongation. Data augmentation effects were property dependent: GAN augmentation generally improved predictions of strength-related properties, while SMOGN yielded greater benefits for elongation-related properties. The best augmentation configurations improved predictive accuracy and model robustness, particularly for sparse regions of the feature space. Physics-Informed Neural Networks incorporating empirical constraints underperformed conventional ML approaches, exhibiting approximately 10% lower correlation and higher prediction errors. Discussion: The results demonstrate that combining suitable ML algorithms with property-specific data augmentation strategies enables accurate prediction of tensile properties from limited experimental data. The differing effectiveness of GAN and SMOGN suggests that augmentation strategies should be selected according to the physical characteristics of the target property. The proposed framework provides a practical approach for addressing data scarcity in nuclear materials research and supports accelerated qualification and evaluation of nuclear structural materials.