표제지
요약
목차
기호설명 9
I. 서론 11
II. 역으로 반전된 연성 와류생성기를 이용한 열전달 시스템 수치해석 14
1. 문제 정의 14
2. 수치 계산 방법 및 검증 14
3. 역방향 깃발의 동적 수치 계산 결과 18
4. 채널 내 와류 거동과 열전달 분석 결과 25
5. 차원분석 (Buckingham의 파이 이론) 29
III. 기계 학습 기반의 LES 모델 개발 33
1. 문제 정의 33
2. 수치 계산 방법 및 검증 33
3. 기계 학습 알고리즘을 통한 채널 및 실린더 주위 난류데이터 예측 37
IV. 결론 41
1. 역으로 반전된 연성 와류생성기를 이용한 열전달 시스템 수치해석 41
2. 기계 학습 기반의 LES 모델 개발 42
참고문헌 43
Abstract 47
Table 1. The mean drag coefficient and minimum velocity along the centerline in the wake of a circular cylinder at Re=3900. 37
Table 2. Computational parameter of DNS at turbulence channel flow. superscript + denotes the wall unit. △T is sampling time interval of the instantanous DNS fields... 38
Table 3. Computational parameter of DNS at flow past a circular cylinder 39
FIG. 1. Schematic diagram of the heat transfer system including an inclined inverted flag and computational domain with boundary conditions. 14
FIG. 2. Time history of the y-position of the leading edge for different grid resolutions of 1/32 (red dashed line), 1/64 (black solid line, present), 1/96 (blue... 17
FIG. 3. (a)-(c) The superimposed inverted flags and (d)-(f) tip position of the flexible flag in the y-direction at γ=0.4 for three different inclination angles; θi=0°,...[이미지참조] 19
FIG. 4. (a)-(c) The superimposed inverted flags and (d)-(f) tip position of the flexible flag in the y-direction at γ=0.7 for three different inclination angles; θi=0°,...[이미지참조] 20
FIG. 5. (a)-(c) The superimposed inverted flags and (d)-(f) tip position of the flexible flag in the y-direction at γ=1.4 for three different inclination angles; θi=0°,...[이미지참조] 20
FIG. 6. (a) The flapping amplitude and (b) the average of y-position for different inclination angles. The symbols indicate LAF(●), SAF(■), D(▼), and S(▲) modes,... 22
FIG. 7. Phase diagram of the dynamic modes as a function of the bending rigidity and the inclination angle. 22
FIG. 8. The mean drag force as a function of the bending rigidity and the inclination angle. The symbols indicate LAF(●), SAF(■), D(▼), and S(▲) modes, respectively. 23
FIG. 9. (a) The mean heat flux, (b) the mechanical energy loss, (c) the heat transfer efficiency for different bending rigidities and inclination angles. The symbols indicate... 25
FIG. 10. Instantaneous vorticity, pressure, and temperature fields at the instants of A-E in Fig. 4(d). The inclination angle θi is 0°, and the bending rigidity γ is 0.7,...[이미지참조] 27
FIG. 11. Nusselt number of the upper and lower walls along the x-direction at the instants of A-E in Fig. 4(d). The inclination angle θi is 0°, and the bending rigidity γ...[이미지참조] 27
FIG. 12. Instantaneous vorticity, pressure, and temperature fields at the instants of A-E in Fig. 4(e). The inclination angle θi is 12°, and the bending rigidity γ is 0.7,...[이미지참조] 28
FIG. 13. Nusselt number of the upper and lower walls along the x-direction at the instants of A-E in Fig. 4(e). The inclination angle θi is 12°, and the bending rigidity...[이미지참조] 28
FIG. 14. Phase diagram of (a) mean heat flux, (b) mean mechanical energy loss, and (c) heat transfer efficiency as a function of the inclination angle θi and bending...[이미지참조] 30
FIG. 15. The correlations between the mean drag force and (a) heat flux, (b) mechanical energy loss, and (c) heat transfer efficiency. The red and blue circles... 30
FIG. 16. The correlations between the flapping kinematics and (a) mean drag force, (b) heat flux, (c) mechanical energy loss, and (d) heat transfer efficiency. The red... 32
FIG. 17. (a) The mean velocity, (b) turbulence intensity of u velocity. (Lx, Ly, Lz)=(8π, 1, 2π), (N₁, N₂, N₃)=(512, 65, 128). The friction velocity Reτ=590[이미지참조] 35
FIG. 18. The Schematic of the computational domain and coordinate system at Re=3900. 36
FIG. 19. (a) The mean streamwise velocity along the centerline in the wake of a circular cylinder, and (b) the transverse velocity profile of the mean flow at... 36
FIG. 20. The schematic diagram of the machine learning algorithm (ANN, RF). 38
FIG. 21. Comparing the SGS stress tensor predictive result with (a) ANN one-point strain rate tensor, (b) ANN nine-point strain rate tensor, and (c) RF nine-point... 39
FIG. 22. Comparing the SGS stress tensor predictive result with ANN nine-point strain rate tensor with (a) ANN, 2 hidden layer and (b) ANN, 6 hidden layer (c) RF... 40
FIG. 23. Comparing the SGS stress tensor predictive result with (a) ANN nine-point strain rate tensor, and (b) RF nine-point strain rate tensor in the case of flow... 40