Virtual X-ray Tomography as a Digital Twin for NDT and Synchrotron Imaging

X-ray computed tomography (XCT) is increasingly used in non-destructive testing (NDT) to inspect complex material combinations and various components. However, the growing complexity of XCT systems and inspection tasks makes experimental planning, parameter optimisation, and result interpretation both time-consuming and resource-intensive. Virtual X-ray simulation offers a powerful complement and addition to physical experimentation by enabling realistic, physics-based representations of XCT systems. This contribution presents how realistic X-ray simulation, based on an extended Beer– Lambert law that accounts for the whole image acquisition system and implemented on high performance graphics processing units (GPUs), can be used to create reliable and trustworthy digital representations of laboratory and industrial XCT scanners. Building on recent developments of the open source gVirtualXray (gVXR) framework, the work demonstrates how virtual XCT can support experiment planning, dosage and experimental setup, develop scripts for more complex experiments involving several motors/detectors, feasibility studies, inspection performance evaluation, develop and evaluate in situ sample environments and subsequent the generation of synthetic data for machine-learning-based NDT. All this information mentioned above can furthermore be used to create even more realistic simulation models to move to a simulation driven and predictive approach. Examples include virtual scan configuration, optimisation of acquisition parameters, assessment of image quality and artefacts, and realistic simulation of detector response and noise. The ability to closely match simulated and experimental data enables systematic investigation of inspection scenarios that would be impractical, dangerous/impossible or costly to reproduce physically. By placing these capabilities in the context of 30 years of NDT evolution, the contribution highlights the transition from purely experimental inspection toward simulation-driven and data-assisted decision support in XCT-based NDT.