JUST THINK
Laboratory Research Topics
Thermal-hydraulic Characteristics Analysis of Helical Cruciform Fuel
– A CFD Approach and Experimental Study Using MRI Equipment
– A CFD Approach and Experimental Study Using MRI Equipment
This study analyzes the thermal-hydraulic characteristics of a HALEU (~20%)-based helical cruciform metallic fuel assembly designed for high-density power outputs in next-generation nuclear reactors. A flow visualization experiment is conducted using Magnetic Resonance Velocimetry (MRV), a non-invasive method capable of measuring internal flow velocities in an opaque experimental model. This experiment enables the detailed analysis of localized and intrinsic flow characteristics within the flow channel, thereby providing academically significant data. Additionally, CFD analyses are performed to validate the data from both approaches.
Development of a Non-Toxic Refractive Index Matching Fluid
for Flow Visualization using 3-D Printing
for Flow Visualization using 3-D Printing
This study develops a new refractive index matching (RIM) technique designed for flow visualization using 3D-printed transparent models.A custom-formulated fluid—non-toxic, colorless, and free of strong smell—is used to match the refractive index of the printed models, allowing internal structures to appear optically seamless. The system enables clear and accurate observation of flow fields using Particle Image Velocimetry (PIV).
Development of a Simplified Lattice CFD Model for Metal Foams
Applied to Industrial Thermal Energy Storage Systems
Applied to Industrial Thermal Energy Storage Systems
This study focuses on developing a simplified CFD modeling approach for metal foams by constructing lattice-based representations of their complex porous structures. The goal is to enable efficient thermal-fluid simulations applicable to industrial systems such as Thermal Energy Storage (TES), while significantly reducing computational cost. The proposed model aims to retain the essential heat transfer and flow characteristics of actual metal foams, ensuring both accuracy and practicality for engineering applications.
CFD Analysis of Helical Coil Flow Characteristics for
Thermal-Hydraulic Optimization of SMR Heat Exchangers
Thermal-Hydraulic Optimization of SMR Heat Exchangers
This study investigates the thermal-hydraulic performance of heat exchangers used in Small Modular Reactors (SMRs) by conducting CFD analysis of internal flows in helical coils. Key focuses include heat transfer characteristics, pressure drop behavior, and flow structure analysis, aiming to provide insight into the optimization and safe design of compact, high-efficiency heat exchangers suitable for nuclear applications.
Reduced-Order Model (ROM) for High-Capacity, Multidimensional
3-D Flow Data for Deep Learning Training Applications
3-D Flow Data for Deep Learning Training Applications
This research explores a methodology that employs dimensionality reduction to convert high-capacity, multidimensional data into training datasets for deep learning models. In developing a deep learning-based virtual nuclear reactor, high-precision 3D data for each reactor component must be produced and provided as training data. However, the transient 3D CFD results are excessively large to be directly processed by the learning model. For this reason, data compression via dimensionality reduction is essential during the preprocessing stage. The reduced-order model (ROM) must capture the essential patterns and features of the 3D flow while delivering adequate compression performance. Our goal is to develop a deep learning-specific ROM by integrating existing linear and nonlinear methodologies with innovative ideas.
Physics-Informed Neural Networks with Adaptive Dynamic Adjustment for Neutron Transport in Single Geometry
The Neutron Transport Equation (NTE) is essential for predicting neutron behavior in reactor cores but is analytically intractable due to its high dimensionality. Traditional numerical methods—both deterministic and stochastic—face limitations such as high computational cost and sensitivity to geometry. To address these challenges, this study introducesAdaptive Dynamic Adjustment PINN (ADA-PINN), a deep learning approach based on Physics-Informed Neural Networks. ADA-PINN automatically adjusts loss function weights using curvature-based metrics, improving training stability and eliminating manual tuning. It also incorporates domain decomposition, multi-group solving, and boundary-focused data placement, making it a promising alternative to conventional NTE solvers.
Development of Probabilistic Safety Assessment (PSA) Model
for Floating Molten Salt Reactor
for Floating Molten Salt Reactor
This research aims to develop a probabilistic safety(PSA) model for a floating molten salt reactor (MSR) deployed at sea. Unlike conventional reactors, the float MSR operates on the ocean and uses molten salt instead of water, introducing unique risk factors and accident scenarios. By identifying maritime-specific hazards and developing accident scenarios tailored to the molten salt reactor systems, an independent probabilistic safety assessment (PSA) model is constructed. The resulting PSA model serves as a critical reference for regulatory approval and plays a significant role in shaping international stakeholder engagement.
Development of Thermal-Hydraulic Analysis Code
for Steam Generators of Nuclear System
for Steam Generators of Nuclear System
This study aims to develop a thermal-hydraulic analysis code specifically designed for steam generators in the nuclear system. This code overcomes the limitations of existing FORTRAN-based tools, which struggle to perform detailed local flow and interaction analyses within analysis components, by supporting the entire analysis spectrum from simple correlation-based methods to advanced CFD numerical simulations. Developed in Python, the code offers numerous possibilities, including the automation of repetitive analysis tasks, optimization, and even deep learning applications. Furthermore, by featuring a user-friendly graphical user interface (GUI).