Controllability of welded parts in critical equipment - Virtual materials for efficient NDT inspection tools – ATALANTE

In the nuclear sector, ensuring the integrity of welded components subject to service constraints or technical hazards is a major challenge. The reliability of these components is therefore determined by non-destructive testing (NDT) methods, enabling to detect and size a wide range of defects and providing essential information on their occurrence, orientation and position. In this context, regular ultrasonic testing (UT) remains an essential method for the assessment of welds, despite the difficulty to perform such analysis on austenitic stainless steels. In these materials, the welds, made up of numerous passes, present a heterogeneous and anisotropic microstructure, formed of long grains developed by epitaxy. Deleterious phenomena are associated with this microstructure, such as deviation and divergence of the ultrasound beam, or its progressive attenuation. Predicting these phenomena using dedicated modeling tools is therefore essential for conducting and optimizing NDT methods. Among these tools, CIVA (CEA LIST) and A3D-CND (EDF R&D) answer this need by simulating the propagation of ultrasonic waves in welds for nuclear applications. However, software requires detailed knowledge and description of the weld microstructure and its variability induced by material properties, process parameters or welder practices. In this context, ATALANTE aims to develop virtual microstructures on thick-welded parts to simulate this ultrasonic propagation and improve NDT tools, by taking up the following challenges: 1) characterizing grain structure formation and variability, for thick industrial welds with a high number of passes, 2) advancing knowledge of the impact of microstructures on UT wave propagation, and 3) maximizing the efficiency of NDT inspection tools through the use of UT simulation tools. More broadly, ATALANTE aims to develop the industrial exploitation of virtual materials, particularly in NDT approaches.
The ATALANTE partnership is divided between experimentation, modeling and development, and focuses on the 316L stainless steel alloy due to its importance in the targeted nuclear applications. Experiments will be carried out to study the effect of gravity on bead shape and microstructural evolution for a high number of passes and under nuclear deposition conditions (ICB). Subsequently, instrumented tests covering various welding processes will be developed to analyze this deposition strategy. In parallel, the welding processes will be modeled using a multi-physics and multi-scale approach (CEMEF) based on a Cellular Automata - Finite Elements (CAFE) coupling method, and validated on the basis of the experiments previously conducted. Fine virtual microstructures, providing a precise description of welds, will be produced on thick parts. In the same time, Machine Learning (ML) tools will be implemented to exploit these data and improve microstructural prediction in welding. Simulations of ultrasonic propagation will be undertaken using CIVA and A3D-CND software by CEA LIST and EDF R&D, based on these virtual materials, analyzing the effect of microstructure, and its impact on the efficiency of NDT tools, in order to improve the latter. Dependence on deposition conditions will be studied, to define problematic welding conditions. The numerical developments done at CEMEF will be enriched and stabilized by TRANSVALOR and introduced into the TRANSWELD software, along with tools facilitating the use of virtual microstructures in NDT applications. This approach will promote the dissemination of the outputs of the project. ATALANTE will thus represent an undeniable technological advance in the inspection of welded parts and the reliability of NDT methods in response to the current needs of the nuclear industry.