Trustworthiness of Digital-Twin-based Automation Technology in Nuclear Power Plant Operation

The project seeks to establish a technical basis for the development and demonstration of a trustworthiness assessment framework for automation enabled by digital twin (DT) technology for nuclear reactor operation and maintenance (O&M) procedures. DTs can be used to automate the assessment of plant state and make O&M decisions, but such automation requires that the output of DTs be trustworthy and explainable. These elements are derived in the proposed project within a modern reasoning-based system, making use of evidence from the DT's development and assessment process (DAP), knowledge-based coverage assessment, contextual and environmental factors, O&M procedures, and expert judgement. The proposed framework uses predictive capability maturity model, assurance case, and logic reasoning to evaluate and establish (1) the trustworthiness stemming from process assurance during DT development, and (2) the trustworthiness stemming from functionality assurance during DT deployment. Theoretically, the task boils down to a complex assurance case for automation technology. The complexity stems from having to deal with a large amount of heterogeneous and potentially conflicting evidence, multi-physics reactor transients, reactor safety, and high-consequence outcomes. Notably, the framework targets the growing uses of machine-learning algorithms in automation technology. A two-tier assessment process, supported by the state-of-the-art uncertainty quantification techniques, will be used to collect and characterize evidence about the inherent trustworthiness of automation technology. A comprehensive evidence pool for the transparent and consistent management of uncertainty will improve the explainability of automation technology in reactor O&M. A reasoning-based system will then be used to manage and analyze the massive evidence, compiling information that can be used to support final decision-making on trustworthiness. The use of intelligent decision support system can vastly improve the robustness and effectiveness of trustworthiness assessment. To address the complexity, the project will exercise a multi-disciplinary approach and leverage on the team's experience to formulate and demonstrate the capability of the framework. This will be achieved through a case study designed with an advanced reactor simulator in support of the application and trustworthiness assessment for an automated O&M procedure. The framework will guide the development and validation of DTs in automation technology. Notably, methods and lessons learned from this project will be instrumental for the independent trustworthiness assessment of different automation systems using DTs, the identification of toplevel automation goals based on the safety and economic requirements of nuclear reactors, the identification of sub-level DTs development plan, and the effective collection of evidence for the target trustworthiness levels. Ultimately, the framework creates the potential for the development and deployment of DT technology, advanced modeling approaches, intelligent automation systems, and unattended reactor systems to improve the reliability and economic viability of existing and next-generation commercial reactor designs. The project's co-investigators bring the necessary complementary expertise and capabilities, with track records of their productive collaboration in previous joint efforts.