Fiber filters are widely used as a core technology for removing impurities in industrial processes. Traditionally, CFD (Computational Fluid Dynamics)-based simulations have employed deterministic models that assume particles are always captured upon contacting the fiber. However, in reality, not all particles that come into contact with the fiber are captured, which necessitates a probabilistic approach to more accurately represent particle behavior.In this study, we developed a probabilistic particle capture model that determines the capture of each particle randomly based on a PRNG (Pseudo Random Number Generator) and calculates the capture probability in each cell using Single Fiber Filtration Efficiency (SFFE). The model was implemented in the Ansys Fluent environment using a User-Defined Function (UDF), and the filtration process for each particle is represented as a Bernoulli trial. This enables much more accurate prediction of particle distribution and capture locations compared to conventional models.

Associated members: Junho Jung

MP-PIC-PBE is an extension of MP-PIC to predict multi-scale fluid phenomena, namely, combining micro-scale particle formation phenomena with meso-scale fluid mechanics. The method employs the PBE in its homogeneous form, while maintaining equivalence with the full-dimensional population balance equation. Additionally, it enables the particulate phase to express particulate stresses using spatial gradients and adopts a Lagrangian description to predict particle properties such as mass, size, age, and velocity. As a result, this approach is robust and fast numerically in predicting particle size distributions with particulate fluid dynamics.

Applications

The first case addresses the simulation of antisolvent crystallization. This case is compared with the existing CFD-PDF-PBE method and the MP-PIC-PBE method. The test problem of crystallization demonstrates the advantages of MP-PIC-PBE in handling the particle size distribution. For variables that are computable by the two methods, the results are qualitatively similar but quantitatively different due to the differences in the models employed. Particularly, MP-PIC-PBE generates additional information not provided by CFD-PDF-PBE.

The second case addresses the simulation of suspension polymerization. The particle flow predicted by the Lagrangian frame clearly shows the suspension flow patterns induced by drag and lift forces. Moreover, the case study confirms the interaction between the blade angle of the impeller and the particle size of resin in the CSTR. To validate the suggested CFD model, the simulation results are compared with the reported experimental data in the literature.

The third case addresses the investigation of particle flow effects in slug flow crystallization. This research introduces a newly developed CFD numerical procedure, the MVP method, to investigate the slug crystallization phenomenon. The developed method can predict the three-phase fluid flow with particle size variation, which could not be shown by existing CFD models, and its ability is verified through comparison with experimental results. In the case studies, the MVP model predicts that the size and number of particles formed can change depending on the size of the slug. Moreover, as the slug size increases, it affects the internal mixing within the slug. These results indicate that the commonly assumed perfect mixing condition inside the slug may be inappropriate.

This study virtually analyzes the Passive Autocatalytic Recombiner (PAR) using Computational Fluid Dynamics (CFD) to remove radioactive substances and hydrogen generated during severe accidents in nuclear power plants. The PAR is designed as a combination of a catalytic section and an adsorption section. The key aspect of this technology is that it operates automatically when the hydrogen concentration in the containment building increases, without any external support.

To predict the correct operational pathway of such a passive system, this study conducts CFD-based modeling. First, a single-phase, multi-component model based on a porous medium is used to predict the gas flow through the adsorbent. The pressure drop caused by the packed adsorbent is expressed using Darcy’s law, with empirical coefficients derived from experimental data. The adsorption rate of CH₃I by the adsorbent is predicted based on the bi-Langmuir equation, which is implemented using a user-defined function (UDF). The catalytic reaction is modeled by directly incorporating plate-shaped elements into the computational domain, and the reaction mechanism is implemented using a wall surface reaction model.

The simulation results explain the self-operating principle of the PAR system and further demonstrate its potential application in optimizing device design and evaluating accident scenarios.

Associated members: Injung Jang

Flow-accelerated corrosion (FAC) is a major degradation mechanism that damages carbon steel piping in the secondary cooling system of nuclear power plants. This study analyzes the hydrodynamic parameters affecting FAC in the elbow section to identify particularly vulnerable locations and classify them according to their direct association with wall thinning.

 To achieve this, accelerated FAC experiments and Computational Fluid Dynamics (CFD) simulations have been conducted on elbow pipe specimens. The experiments have revealed, multiple sections prone to FAC, while CFD simulations have provided velocity and vorticity vectors. These vectors have been then decomposed into circumferential, axial, and radial components in a cylindrical coordinate system to evaluate the influence of each component on FAC.

 The results show that all velocity and vorticity components to contribute to wall thickness reduction due to FAC. Therefore, to accurately assess wall thinning due to FAC, It is essential to comprehensively consider both velocity and vorticity components.

Associated members: Hyunseung Jo

Hydrogen, which has recently attracted significant attention, faces substantial challenges in transportation and storage. Hydrogen production via ammonia decomposition has emerged as a promising solution to these challenges. However, the cost of ammonia decomposition varies greatly depending on the catalyst type and morphology.

To address this, the present study aims to design a reactor and catalyst system capable of efficiently performing ammonia decomposition under low-temperature conditions. A one-dimensional fixed-bed reactor model based on Temkin–Pyzhev reaction kinetics was developed, and a parameter estimation framework was established using experimental data. In addition, a gPROMS-based process simulator was developed to evaluate variations in thermal efficiency depending on catalyst composition. On the CFD side, user-defined functions (UDFs) were implemented to precisely analyze the thermal and flow characteristics within the reactor. Furthermore, Blender-based modeling of the catalyst packing structure was used to simulate the behavior of the catalyst bed.

Associated members: Wansik Yu, ByeongGyu Jeon, WooSol Jeon

  The population balance equation(PBE) is a mathematical model that calculates the particle size distribution in the spatial dimensions by considering both fluid flow and chemical changes. Studies conducted so far have discretized the particle distribution range and found a solution to solve the PBE. This general procedure, which must cover particle size changes from the nucleation of very small particles to the growth of larger particles, can sometimes lead to unnecessary computational domains or cause issues with a coarse grid. We develop a novel numerical procedure to solve the PBE using the moving boundary method. The moving boundary method is one of the techniques used in numerical analysis, applied to problems involving boundaries that changes in position over time. Uses a scaled dimensionless domain to represent the moving boundary system. If the boundary point is tracked using this dimensionless number and reflected in the equation, it can be though as if a fixed area is moving. The introduced method will enable efficient particle range setting and assist in finding faster and more accurate numerical solutions compared to conventional numerical methods.

Associated members: Junho Chung

 Data centers handle large volumes of data and produce a significant amount of heat. While air cooling is commonly used, it has limitations in thermal efficiency. Liquid cooling, which uses a coolant to dissipate heat, is a more effective alternative. Underwater data centers, designed to take advantage of surrounding ocean conditions such as wave activity, are still in the early stages of development but show promising potential for energy-efficient operation.

As part of an ongoing study, numerical simulations are being conducted to analyze fluid flow, heat transfer, and turbulence within an underwater cooling system. An open-channel wave boundary condition is applied to simulate the effects of waves, and the heat transfer coefficient is evaluated under various temperature differences. This research aims to enhance the understanding of how to improve cooling performance and energy efficiency in real-world applications.

Associated members: Nelson K. Muchabaiwa, Taegyun Han