Research
Prediction of hydrodynamic pressure using deep learning in sloshing flow
(Main investigator: Ki Jong Kim)
Sloshing flow in fluid dynamics which refers to the movement of liquid back and forth inside a container due to external force. In terms of the prediction of slosh-induced pressure, deep learning (DL) is emerging as a powerful tool for predicting the pressure field using an accumulated database of the measured pressure and image recognition by flow visualization. We explore the prediction of slosh-induced pressure using image-based DL for various pressures such as regular, random, impulse, and peak pressure with a single algorithm. This study is valuable in that the dynamic characteristics using DL can be predicted from the wave image, which indicates the kinematic source.
Ocean wave pattern and wave power generation efficiency through vertical cylinder arrays
(Main investigator: Dong Hyup Youn)
Ocean structures, such as offshore wind turbines and bridges, experience phenomena in which waves interact with their supports, leading to effects like wave run-up or wave breaking. Additionally, the presence of these structures alters wave behavior through refraction, diffraction, scattering, reflection, and superposition, resulting in complex and unpredictable patterns, especially when multiple structures are present. To investigate these phenomena, we are conducting scaled-down experiments. We arrange cylindrical structures and observe the interactions between the cylinders and the surrounding waves to understand the effects of cylinder size, gap, and wavelength relationships on the amplification of wave heights through superposition. This amplification of wave heights through superposition can focus wave energy conversion efficiency, leading to increased effectiveness in wave power generation. Furthermore, we are also researching to verify the efficiency enhancement through scaled-down wave power generators.
Reinforcement learning-based optimization for the shape and valve position of a scroll compressor
(Main investigator: Janggon Yoo)
The scroll compressor, employed in large-capacity refrigeration cycles, functions by orbiting two scroll profiles to compress the refrigerant. Designing the scroll compressor is essential to enhance the energy efficiency, compression capacity, and discharge pressure. However, optimizing the compressor’s component positioning and shape is challenging because of complicated nonlinear boundary conditions. The optimization problem is solved using reinforcement learning utilizing an automatic scroll design algorithm and a low-order scroll compressor thermal analysis model. A novel technique involving multiple steps and degrees of freedom was proposed to optimize the position and number of discharge valves and the shape of the scroll compressor. The optimization method can be applied to the commercial scroll compressor design process.
Instability patterns of particles by the overpressure
(Main investigator: Jaehun Yoo)
Packed particles show certain instabilities when the pressure is induced and such instability forms specific patterns over time. We are conducting research to identify the factors that determine the formation of these patterns. This research can be applied in the field of fire safety and defense industry.
Liquid CO2 spray cooling for LNG tanks
(Main investigator: Siyoung Park, Hyeeun Shim)
“Cooling down” is an essential prelude to loading cryogenic LNG, necessitating the chilling of cargo tanks and conduits. The process involves the introduction of liquefied carbon dioxide (LCO2) via spray nozzles into each cargo tank. The transition from ambient temperature (post-gassing up) to a predetermined temperature constitutes the “initial cool down” phase, distinguishable from routine cooling during ballast voyages. Prior to the infusion of LNG into an LNG vessel’s cargo system, including the cargo tanks, requisite cooling to approximate LNG temperature is mandatory. Mitigating the cargo tank temperature curtails the heat reservoir available for transfer to and subsequent warming of the incoming LNG. This deliberate cooling curbs vapor generation within reasonable thresholds. To evaluate the efficacy of the LCO2 spray cool-down process and comprehend phase transition dynamics, computational fluid dynamics (CFD) simulations are employed. Subsequently, the study offers insights into optimal design parameters and operational conditions, culminating in recommendations for facilitating optimal states in the LNG cargo tank.
Potassium Heat Pipe for Small Modular Reactors (SMRs)
(Main investigator: Sohyeun Kang)
Heat pipes have garnered renewed attention in contemporary times, emerging as focal points of rigorous investigation within domains such as nuclear engineering and space exploration. Technological advancements, exemplified by the advent of Small Modular Reactors (SMRs), have garnered substantial interest due to their capacity to effectively address the escalating energy requisites. These reactors possess the capability to produce power in close proximity to densely populated regions, accommodating heightened energy consumption without necessitating extensive spatial occupancy or protracted construction periods. In this context, the deployment of SMRs mandates the integration of a space-efficient heat dissipation system, a criterion well-suited for realization through the utilization of heat pipes. This study endeavors to elucidate the meticulous design of heat pipes tailored for SMRs, employing a synergy of theoretical models and Computational Fluid Dynamics (CFD) simulations. This framework buttresses the subsequent research phase, wherein a prototypical implementation of the heat pipe within the SMR infrastructure is established, ensuring unwavering operational stability and requisite heat transfer efficiency.
Control of droplet scattering behavior in the process of spin coating/developing
(Main investigator: Dong Ju Kim)
Spin coating/developing constitutes a critical process in semiconductor manufacturing. Precisely spreading chemically sensitive liquids onto a rapidly rotating wafer is imperative for optimal results. Unfortunately, the presence of scattered droplets due to various factors poses a risk of wafer contamination and performance degradation. To address this challenge, our research involves high-speed imaging of droplets during this process and the development of reliable quantitative models to predict their behavior. By doing so, we aim to enhance the precision and efficiency of spin coating/developing, ensuring superior semiconductor production outcomes.
Design of sphere to suppress vortex-induced vibration
(Main investigator: Minseop Lee)
The sphere connected to the elastic cable (Drogue system) can be utilized in many areas of the industry. When the drogue system interacts with the uniform flow, the sphere oscillates from side to side under certain conditions (Vortex-induced vibration). Since such strong vibration is a negative factor from the point of view of industrial applications, it is necessary to design a stable drogue system. We present a design that adds a spiral-shaped pattern to the surface of a sphere that breaks the vortex structure, and the vibration of the drogue system is significantly reduced.
Flow-induced vibration of a hydroturbine
Flow-induced and acoustic-induced vibration of a pipe
Fluid-structure interaction of oil spill prevention technology
Oil separation of a compressor
Thin film flow on a rotating disc
Fluid mixing inside a reactor chamber