• 대한기계학회논문집B
• /
• 제31권11호
• /
• pp.895-903
• /
• 2007

CMBT(Curved Moving Boundary Treatment) is a newly developed scheme for the treatment of a no slip condition on the curved solid wall of moving obstacle in a flow field. In our research CMBT was used to perform LBM simulation of a flow over a moving circular cylinder to determine the flow feature and aerodynamics characteristic of the cylinder. To ascertain the applicability of CMBT on the complex shape of the obstacle, it was first simulated for the case of the flow over a fixed circular cylinder in a channel and the results were compared against the solution of Navier-Stokes equation with deforming mesh technique. The simulations were performed in a moderate range of reynolds number at each moving cylinder to identify the flow feature and aerodynamic characteristics of circular cylinder in a channel. The drag coefficients of the cylinder were calculated from the simulation results. We have numerically confirmed that the critical reynolds number for vortex shedding is ar Re=250 and the result is the same as the case of fixed cylinder. As the cylinder approaching to one wall, the 2nd vortex is developed by interacting with the wall boundary-layer vorticity. As the velocity ratio increase the third vortex are generated by interacting with the 2nd vortexes developed on the upper and lower wall boundary layer. The resultant $C_d$ decrease as reynolds number increasing and the Cd approached to a value when Re>1000.

• 한국공작기계학회논문집
• /
• 제16권6호
• /
• pp.146-152
• /
• 2007

An active micro-mixer, which was composed of an oscillating micro-stirrer in the microchannel to provide rapid, effective mixing at high flow, rates was analyzed. The effects of molecular diffusion and disturbance by the stirrer were considered with regard to two types of mixer models: the simple straight microchannel and microchannel with an oscillating stirrer. Two types of mixer models were studied by analyzing mixing behaviors such as their interaction after the stirrer. The mixing was calculated by Lattice Boltzmann methods using the D2Q9 model. In this study, the time-averaged mixing index formula was used to estimate the mixing performance of time-dependent flow. The mixing indices of the two models compared. From the results, it was found that the mixer with an oscillating stirrer was much more enhanced and stabilized. Therefore, an optimum design for a dynamic micro-mixer with an oscillating stirrer was performed using Taguchi method in order to obtain a robust solution. The design parameters were established as the frequency, the length and the angle of the stirrer and the optimal values were determined to be 2, 0.8D and ${\pm}75^{\circ}$, respectively. It was found that the mixing index of the optimal design increased 80.72% compared with that of the original design.

• Composites Research
• /
• 제37권1호
• /
• pp.40-45
• /
• 2024

본 연구에서는 고분자 전해질막 연료전지용 가스확산층의 투과도를 예측하기 위해 삼차원 합성곱 신경망 모델을 사용하는 방법론을 소개한다. 먼저, 기계학습 모델을 학습시키기 위해 X-선 단층 촬영을 통해 얻은 실제 가스확산층 이미지에서 형태학적 특성을 추출해 가스확산층의 대표 체적 요소로 이루어진 인공 데이터셋을 생성한다. 이러한 형태학적 특성은 다공성, 섬유 배향, 직경의 통계적 분포가 포함된다. 구축한 인공 데이터셋 대표 체적 요소들의 투과도를 평가하기 위해 격자 볼츠만 방법이 사용되었으며 각각의 대표 체적 요소들의 투과도를 도출하였다. 이러한 인공 데이터셋을 통해 삼차원 합성곱 신경망 모델을 학습시켰으며 인공 데이터셋을 학습한 삼차원 합성곱 신경망 모델이 실제 가스확산층의 대표 체적 요소 투과도 또한 잘 예측하는 것을 확인하였다.