Experimental and Numerical Reassessment of Real-World Defects in Nuclear Components Through Additively Manufactured Physical Twins

The detection of complex, crack-like discontinuities in the reactor pressure vessels of Doel 3 and Tihange 2 highlighted the need for more advanced evaluation methods to assess the structural effects of internal defects. This study presents a combined numerical–experimental methodology based on additively manufactured physical twins to reassess such real-world defects in safety-critical components. Tensile specimens of the ferritic-bainitic reactor steel 22NiMoCr3-7 containing open (surface-visible) and closed (internal) cracks were produced using laser powder bed fusion to serve as physical twins. Four different crack configurations were selected, produced, and tested as open and closed crack field specimens. The additive manufacturing parameters and post-heat treatment were optimized to achieve microstructures and mechanical properties equivalent to those of the conventionally manufactured (forged) reference material used in nuclear power plant components. The incorporated crack fields are designed as replicas (geometry, configuration, spacing, inclination, etc.) of the internal crack fields found in Doel and Tihange. Digital image correlation (DIC) was used during mechanical testing to capture the evolution of full-field strain and crack propagation during tensile loading. DIC is crucial for the experimental testing because it allows events inside the material, such as crack interaction, to be recorded. The gained experimental data were then evaluated against finite-element simulations employing an extended non-local Rousselier damage model capable of describing ductile failure through void initiation, growth, and coalescence across varying triaxiality conditions. For a physical twin to serve as a valid experimental replica, its properties must be similar to those of the conventionally forged reference material. Otherwise, the findings cannot be directly applied to the real component. The initial material characterization demonstrated that the additively manufactured material closely matches the conventionally forged reference. In the next step, the failure behavior of the additively manufactured crack field specimens was examined and compared with that of the reference material and with the predictions of the numerical simulation. The experimental results demonstrated that the additively manufactured crack field specimens show a sufficiently similar failure behavior to reference specimens. Specimens with crack configurations that were expected to show crack interaction according to the interaction criterion proposed by Gauder did so in both the open and closed specimens. These results further validate his updated interaction criteria. Additionally, the results confirm that testing closed or open crack field specimens does not influence the occurrence of crack interaction, allowing for simplified experiments in the future. The crack initiation sequence in additively manufactured specimens was found to also match that of conventionally manufactured ones. Numerical simulations showed strong correlation with experimental force-displacement curves and strain field patterns, confirming the accuracy of the developed numerical damage mechanics model. These findings prove that physical twins can be used to realistically assess the structural impact of complex internal defect configurations. This study confirms the suitability of additive manufacturing for replicating defect geometries with high fidelity and demonstrates the capability of damage-based numerical models to predict crack interaction and failure behavior. This approach provides a powerful tool for structural integrity assessments in nuclear and other safety-critical applications.