https://scholarworks.unist.ac.kr/handle/201301/72 2026-04-08T00:28:19Z 2026-04-08T00:28:19Z Jeong, Ugyu https://scholarworks.unist.ac.kr/handle/201301/91033 2026-03-26T13:14:54Z 2026-01-31T15:00:00Z

Title: SAFE-CLEAR: Integrated Safety Assessment Framework based on Advanced Numerical Modeling for Clearance of Decommissioning Radioactive Waste Author(s): Jeong, Ugyu Abstract: This dissertation develops and validates SAFE-CLEAR (Safety Assessment Framework for Exempt waste CLEARance), an integrated national safety assessment framework for clearance of nuclear facility decommissioning waste. The framework is designed to overcome methodological limitations and regulatory blind spots in the current RESRAD code family, including fragmented scenario-specific tools, outdated physical modelling, and the absence of linked scenarios across the waste management chain. SAFE-CLEAR comprises three advanced core modules that are consistently integrated into a integrated assessment platform. First, the external exposure module replaces simplified dose conversion factor corrections with a point-kernel integration scheme combined with probabilistic sampling. Benchmarking against MCNP and MicroShield® demonstrates that this module prevents the hazardous underestimation observed in existing models under complex three-dimensional source geometries and heterogeneous multi-layer shielding, while maintaining MCNP-level credibility with reasonable conservatism. Second, the groundwater transport module couples 1D and 2D advection–dispersion analysis to resolve physical inconsistencies in RESRAD-Onsite, such as neglect of dispersion and a single-path 1D flow assumption. Second, the groundwater transport module couples one-dimensional and two-dimensional advection–dispersion analysis in a way that retains the conservative assumptions embedded in RESRAD-Onsite but explicitly corrects the physical errors originating from its simplified formulations. Dispersion, multidimensional flow paths, and mass balance are treated in a physically consistent manner, removing non-physical error such as illogical overestimation of released activity and dose. In doing so, the module preserves the conservative frame of the original RESRAD approach, while replacing its oversimplified transport representation with a hydrologically and radiologically coherent description of plume evolution. Third, the multi-phase carbon-14 release module links pH- dependent geochemical reactions and isotopic exchange with gas–aqueous transport. This overcomes the empirical compartment approach in RESRAD that, for example, unrealistically blocks gaseous migration in cover layers. The module quantitatively captures dynamic phase partitioning and pathway shifts with pH, significantly improving the reliability of 14C safety assessments without unnecessary over-conservatism. Finally, Integrated scenario analyses using SAFE-CLEAR identify and quantify key missing links, including activity concentration in by-products (e.g., dust and slag), landfill behavior of residues, and pH-dependent 14C migration in alkaline disposal environments. Compared with RESRAD-Onsite, SAFE-CLEAR enables higher annual disposal capacities for landfill scenarios while satisfying regulatory dose criteria, and demonstrates that excluding the landfill pathway under a conditional clearance strategy can increase nuclide-specific clearance levels by orders of magnitude. Overall, SAFE-CLEAR functions not only as a dose calculation code but as a decision-support tool that enhances regulatory completeness and provides a technically robust basis for flexible, efficient, and safely conservative management of decommissioning waste. Major: Department of Nuclear Engineering

2026-01-31T15:00:00Z Jin, Ik Jae https://scholarworks.unist.ac.kr/handle/201301/91032 2026-03-26T13:14:54Z 2026-01-31T15:00:00Z

Title: Development of Heat Pipe-Cooled Microreactor Condition Monitoring Platform using Deep Learning for Digital Twin Implementation Author(s): Jin, Ik Jae Abstract: Heat pipe-cooled microreactors have been proposed to enhance the safety, reliability, and deployment of nuclear systems in remote and off-grid environments. The heat pipe-cooled microreactor utilizes passive heat transfer without the need for complex coolant circulation systems, compared to conventional NPPs that rely on active cooling mechanisms. Considering the purpose of microreactor development, the autonomous operation of these systems requires advanced condition monitoring techniques capable of real-time fault detection and predictive diagnostics. Conventional monitoring methods primarily depend on in-core instrumentation and threshold-based anomaly detection, which have limited adaptability, increased maintenance complexity, and the generation of additional radioactive waste. Although advanced condition monitoring technology has been proposed to address the limitation of the conventional method, the large-scale NPPs are usually considered. For regulatory approval, the development should be accompanied by research on advanced condition monitoring technology. In this study, a deep learning-based condition monitoring platform was developed to support the implementation of a digital twin for heat pipe-cooled microreactors. An experimental investigation was conducted to evaluate the thermal performance of heat pipes under various operating conditions, including steady-state operation, startup behavior, inclination changes, loss of cooling accidents, and shutdown sequences. The experimental results were analyzed to investigate heat transfer behavior on system performance. A data-driven monitoring approach was employed to infer fuel temperature and conditions using only indirect temperature measurements from adiabatic and condenser sections of the heat pipe, reducing reliance on in-core instrumentation. Deep learning models were trained using experimental datasets to enable real-time prediction of reactor conditions and anomaly detection. In addition, the discussion model was developed to help the decision making of the operator, using language model. The integration of explainable AI techniques and uncertainty quantification methods enhances the reliability and interpretability of model predictions, ensuring reliability in safety-critical nuclear systems. The proposed condition monitoring system demonstrated that it could detect deviations in heat pipe performance from transient thermal behavior, providing robust diagnostic capabilities for transient scenarios with outstanding performance. The proposed methodology enables early fault detection with a reduction of the dependency on manual inspections, enabling remote reactor operation. This study could contribute to the advancement of digital twin implementation for heat pipe-cooled microreactors by providing a framework for AI- assisted predictive maintenance and autonomous monitoring. Major: Department of Nuclear Engineering

2026-01-31T15:00:00Z Seo, Joo Hyung https://scholarworks.unist.ac.kr/handle/201301/91031 2026-03-26T13:14:53Z 2026-01-31T15:00:00Z

Title: Effect of Gas Entrainment on Liquid Maldistribution in the Freeze Valve Assembly of a Molten Salt Reactor Drain System Author(s): Seo, Joo Hyung Abstract: The drain system functions to transfer fuel salt by gravity during shutdown of molten salt reactors (MSRs). This system has been recognized as a passive approach for enhancing reactor safety, and its performance and importance as a key safety component were confirmed through the Molten Salt Reactor Experiment (MSRE Despite its importance as a major safety feature, previous studies have primarily relied on simplified numerical models, and experimental investigations to evaluate and verify system performance have remained limited. In particular, potential phenomena such as gas entrainment observed in experiments can induce two-phase flow within the drain system, leading to siphon break and degradation of drain performance. These phenomena therefore constitute important considerations in drain system design. This study focused on analyzing the hydraulic behavior of the drain system and evaluating the effects of gas entrainment on liquid maldistribution within the freeze valve assembly as well as on overall drain system performance. A scaled-down mock-up experimental facility was constructed to simulate the hydraulic drain behavior of the MSR, and combined experimental and numerical results were used to characterize the drain behavior under gas entrainment conditions. A drain-time delay of approximately 10% was observed when vortex-induced gas entrainment occurred. When gas entrainment was initiated while the liquid level exceeded 13.2cm of the initial fuel-salt inventory, liquid maldistribution developed and resulted in an insufficient amount of salt remaining within the freeze valve assembly. Liquid maldistribution was quantitatively identified as being caused by siphon break induced by early gas entrainment. The identified hydraulic mechanism was subsequently extended to molten-salt conditions and evaluated through numerical analysis. An advanced freeze valve design was developed to effectively suppress liquid maldistribution. A cross-vane fin structure extending from the flat region of the freeze valve to the tank bottom was experimentally confirmed to be effective in suppressing vortex-induced gas entrainment. An 8 cm-high fin structure restricted the maximum vortex critical height to 8.8 cm and prevented liquid maldistribution under all experimental conditions. In addition, numerical analyses demonstrated that the increased heat-transfer area introduced by the fin structures reduced the valve opening time and consequently decreased the total drain time by more than 60 s, despite the additional pressure loss caused by the fins. This study highlights the significance of vortex-induced gas entrainment in drain systems and demonstrates the effectiveness of its suppression through an improved freeze valve design. The results provide fundamental insight and quantitative data to support the design of drain systems and freeze valves and ultimately contribute to enhanced safety and reliability of future MSR technologies. Major: Department of Nuclear Engineering

2026-01-31T15:00:00Z Moon, Taeuk https://scholarworks.unist.ac.kr/handle/201301/91030 2026-03-26T13:14:52Z 2026-01-31T15:00:00Z

Title: ADVANCED CAD-TO-SIMULATION INTERFACE FOR UNSTRUCTURED MESH SIMULATIONS IN VERSATILE FUSION DEVICES Author(s): Moon, Taeuk Abstract: The exponential escalation in global energy demand, catalyzed by the rapid industrialization of artificial intelligence, has precipitated an urgent requirement for commercial-scale fusion energy. As magnetic confinement devices transition from experimental facilities to pilot plants, a critical disparity has emerged between idealized physics modeling and the intricate engineering reality of tokamak assemblies. Traditional simplified models frequently fail to capture the complex interactions between plasma behavior and "as-built" reactor components—such as discrete tiles, cooling channels, and diagnostic ports—leading to uncertainties in heat load predictions and diagnostic validity. This dissertation addresses this challenge by developing a modular and unified CAD-to-simulation geometric analysis pipeline. Unlike legacy workflows that rely on geometry simplification or intermediate file conversions, this framework functions as a stand-alone library that integrates high-fidelity computer-aided design (CAD) data directly into unstructured mesh simulations. The pipeline introduces optimized algorithms for rapid particle collision detection and realistic physics mapping, allowing simulations to account for engineering installation tolerances and misalignments. The efficacy of this framework is validated through comprehensive application to KSTAR and ITER geometries. Key results include the precise characterization of non-axisymmetric physics phenomena, specifically neutral beam injection (NBI) induced fast ion losses and the modeling of amplified resonant magnetic perturbations (RMP) on realistic plasma-facing components . The analysis successfully identified localized heat peaks and tile-gap heat flux splitting patterns often missed by 2D assumptions. Furthermore, the framework facilitated the optimization of Lyman-alpha diagnostic line-of-sights (LOSs) through synthetic diagnostic generation. By successfully mapping multi-physics results onto authentic engineering structures, this work establishes a standardized, error-resistant foundation for digital twin development. It bridges the gap between scientific analysis and engineering design, enabling proactive risk assessment, uncertainty quantification, and the future integration of spatial intelligence into fusion power plant operation. Major: Department of Nuclear Engineering

2026-01-31T15:00:00Z