Machine Learning-based Correlation of Charpy Impact Properties Between Sub-sized and Standard-sized Specimens for Nuclear Structural Materials

Jul 16, 2026

Upload by Levi

Reliable correlations between Charpy impact results from sub-sized and full-sized specimens are critical for assessing nuclear structural integrity when spatial constraints and limited material volume prevent testing standard sizes. Existing standards and analytical correlation methods offer limited accuracy and are often restricted by specific materials, treatments, and geometries. This study presents an ML framework that aligns sub-sized absorbed-energy data across the full ductile-to-brittle transition using a temperature shift and scaled residual projection, then extracts USE and DBTT via hyperbolic tangent fitting. Validation on 389 matched SA533B tests shows improved performance (R²=0.942 for USE, 0.892 for DBTT) without requiring full-sized data at inference.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages27

Reading Time1 h 8 min

Introduction

Charpy V-notch impact testing and key parameters (USE, DBTT)

Material surveillance needs in nuclear applications

Motivation for sub-sized specimens

Broader applications of small-size specimen testing

Limitations: geometry-dependent fracture behavior

Machine Learning-based Correlation of Charpy Impact Properties Between Sub-sized and Standardsized Specimens for Nuclear Structural Materials Yugandhar Kasala Sreenivasulu1, Isshu Lee2, John W. Merickel2, Fei Xu3,2, Yalei Tang2, Joshua E. Rittenhouse2, Aleksandar Vakanski1,\#, Rongjie Song2,\# 1 University of Idaho, Department of Computer Science, Moscow, ID 83843, USA 2 Idaho National Laboratory, Idaho Falls, ID 83415, USA 3 Department of Computer Science, University of Texas at El Paso, El Paso, TX, United States \#[Corresponding Authors: Aleksandar Vakanski: vakanski@uidaho.edu](Corresponding Authors: Aleksandar Vakanski: vakanski@uidaho.edu), Rongjie Song: [Rongjie.Song@inl.gov](Rongjie.Song@inl.gov) Abstract Reliable correlations of Charpy impact test results between sub-sized and full-sized specimens are essential for structural integrity assessments, particularly in nuclear applications, where spatial constraints and limited material volume restrict specimen size. Although standards such as ASTMA370 and BS 7910 provide guidance on conversion methodologies, and numerous analytical correlation methods have been proposed in prior studies, these approaches generally have limited accuracy and their applicability is often constrained to specific materials, treatment conditions, and specimen geometries. In this study, a Machine Learning (ML)-based framework is proposed for correlating Charpy impact properties across specimen sizes. The proposed approach maps absorbed energy values across the full ductile-to-brittle transition region by applying a temperature shift combined with scaled residual projection, to align sub-sized test data with full-sized response. From the resulting temperature-energy profiles, the correlated values for upper shelf energy (USE) and ductile-to-brittle transition temperature (DBTT) are extracted by fitting data with a hyperbolic tangent model. The framework is validated using a dataset comprising 389 matched sub-sized and full-sized Charpy impact tests on SA533B steel. This ML-based approach demonstrates an improved correlation performance relative to conventional analytical methods, achieving R² values of 0.942 for USE and 0.892 for DBTT. The trained ML models do not require access to full-sized Charpy data during inference, making this approach suitable for material surveillance programs, accelerated irradiation testing, and other applications involving small-size Charpy impact testing. Keywords: Charpy V-notch impact test; sub-sized specimens; machine learning; size effects; nuclear structural materials. 1. Introduction Charpy V-notch impact testing is a fundamental methodology for evaluating the fracture resistance and temperature-dependent impact toughness behavior of ferritic steels used in reactor pressure vessels (RPV) [1−3] . The primary parameters obtained from Charpy tests performed within a range of temperatures include the upper shelf energy (USE) and the ductile-to-brittle transition temperature (DBTT) . The USE is defined as the plateau value of absorbed impact energy measured at sufficiently high temperatures where fracture occurs predominantly by ductile mechanisms. The DBTT is defined as the temperature corresponding to the inflection point of the sigmoidal absorbed energy-temperature curve, representing the transition from brittle to ductile fracture behavior. These parameters serve as important indicators for quantifying a material’s capacity to withstand dynamic loading under unirradiated and neutron-irradiated service conditions. Within the material surveillance programs in nuclear power plants, long-term monitoring of USE and DBTT through periodic testing of archived samples fabricated from representative materials is indispensable for ensuring the structural integrity of reactor components over their licensed operational lifespan and extended operation [4,5] . However, challenges with the availability of archived irradiated material limit the extent and f

FAQs

Why are correlations between sub-sized and full-sized Charpy impact tests important in nuclear applications?

Nuclear structural integrity assessments require reliable USE and DBTT estimates, but spatial constraints and limited material volume often make full-sized testing impractical in surveillance and irradiation contexts. Correlating the two specimen sizes enables effective use of smaller samples.

What limitations affect conventional correlation methods and standards?

Existing guidance (e.g., ASTM A370 and BS 7910) and analytical correlation approaches generally show limited accuracy and reduced generality. Their applicability is commonly constrained by specific materials, heat-treatment conditions, and specimen geometries.

How does the proposed machine learning framework produce USE and DBTT correlations without full-sized data at inference?

The framework maps absorbed energy across the ductile-to-brittle transition by applying a temperature shift combined with a scaled residual projection to align sub-sized responses with full-sized behavior. It then fits temperature-energy profiles using a hyperbolic tangent model to extract USE and DBTT, while not requiring full-sized Charpy inputs during inference.

Jul 16, 2026

Upload by Levi