Akademik Veri Yönetim Sistemi
Tez Türü: Doktora
Tezin Yürütüldüğü Kurum: İstanbul Teknik Üniversitesi, Lisansüstü Eğitim Enstitüsü, Kıyı Bilimleri ve Mühendisliği, Türkiye
Tez Danışmanı: Prof. Dr. Veysel Şadan Özgür Kırca
Tezin Onay Tarihi: 2025
Tezin Dili: Türkçe
Desteklendiği Program: Diğer
Özet:
Boundary layer turbulence has long been a significant issue in fluid mechanics, drawing considerable attention. In particular, the problem of laminar-to-turbulent transition in boundary layers remains partially unresolved despite being studied for over a century. While well-established Reynolds number-based criteria exist for the onset of transition in developing boundary layers, ongoing research continues to explore key uncertainties surrounding this phenomenon. These include identifying which coherent structures observed during transition enhance turbulence development or dissipate, determining the spatio-temporal variations in flow parameters such as velocity, acceleration, and pressure, and establishing cause-and-effect relationships between these parameters and the evolution of flow structures from the perspective of the transition mechanism. Therefore, understanding the laminar-to-turbulent transition is crucial for numerous areas of research, including flow-seabed or flow-riverbed interactions, flow-structure interactions, sediment transport, flow resistance, wave-seabed dynamics, and atmospheric boundary layer dispersion.
This thesis aims to quantitatively investigate the laminar-to-turbulent transition in oscillatory wave boundary layers. To this end, the coherent structures and turbulence characteristics in an oscillatory boundary-layer flow at transitional Reynolds number over a smooth bed are investigated using the Direct Numerical Simulation (DNS) method. The simulations are carried out using the open-source Nektar++ framework, which solves the governing equations for the oscillatory boundary layer, namely the Navier–Stokes equations and the continuity equations. The DNS model was validated against the results of a highly validated numerical study regarding bypass transition in Stokes boundary layers developing over a smooth bed. Following the validation, two techniques were used to trigger boundary layer transition: 1) A temporary roughness element, which was removed at the very moment (i.e., cutoff time tc) when a vortex tube appears on it; 2) A fluid jet, puffing from the center of the computational domain for a specified time interval. The model successfully reproduced the onset of turbulence, along with the development of coherent structures such as vortex tubes and turbulent spots, allowing for a detailed analysis of their spatio-temporal evolution. The bed shear stress was used as a reliable indicator for identifying and distinguishing the near-bed flow structures that appear in the flow region, as these coherent structures cause intense fluctuations in the bed shear stress signal.
The DNS results revealed that the current modeling approach can clearly detect both vortex tubes and turbulent spots. Both vortex tubes and turbulent spots were observed in the transitional oscillatory boundary layer, which is consistent with the findings reported in the literature. Among these, vortex tubes appeared as two-dimensional vortices, producing kinks and dips in the bed shear stress. These are relatively minor
fluctuations compared to the more abrupt and pronounced peaks (spikes) generated by
turbulent spots. The model accurately captured vortex tube generation, including the
phase of their emergence and the wavelength between successive vortex tubes, which
remains nearly constant, as demonstrated in the literature.
Turbulent spots emerge as small turbulent patches, developing into arrowhead-shaped isolated turbulent structures that produce single or multiple spikes in the bed shear stress. The DNS model successfully reproduced these spots, too, from their formation to dissipation. Data from various measurement points showed that turbulent spots generate peaks in the bed shear stress up to 4−5 times the maximum bed shear stress value measured in the far field (i.e., in the laminar regime). This result agrees well with the findings reported in the previous experimental and numerical studies. The fully developed turbulent boundary layer followed a quasi-cyclic process called the bursting cycle. The non-dimensional spacing between low-speed streaks near the wall was found in the 80 < λ + < 110 range, consistent with the commonly reported value of λ + = 100 for turbulent boundary layers developing over a smooth bed. These findings confirm previous observations that turbulent spots mimic the bursting process observed in fully developed turbulent boundary-layer flows.
The present study also showed that phase-resolved profiles of mean velocity u(y, ωt) and turbulence quantities revealed that the near-bed region responds to the adverse pressure gradient earlier than the free-stream flow. Following the removal of the roughness element, turbulence was weak or absent in most regions, and the flow remained predominantly laminar, maintaining a phase lead of approximately 45◦, characteristic of laminar oscillatory boundary layers. However, it was observed that the phase lead of bed shear stress to free-stream velocity decreased (i.e., the near-bed flow reversal delays) as turbulence developed in the following half-cycles, reflecting an increase in momentum exchange and more pronounced turbulence in the boundary layer flow. The phase lead reduced to values around 15◦ −20◦ (i.e., near-bed flow reversal takes place around ωt = 150◦ −165◦, consistent with enhanced near-bed momentum transfer from the upper flow layers.
These results further indicate that turbulence originates in the near-bed region, grows during the deceleration stage of the oscillatory motion due to the adverse pressure gradient, and diffuses away from the bed in the later phases of the oscillatory motion. It was also observed that turbulence was distributed more uniformly in the early phases, i.e., in the acceleration stage of the oscillatory flow, due to the residual background turbulence from the previous half-cycle. The flow remained predominantly laminar during acceleration when the Reynolds number was near the critical value, and transition to turbulence occurred during the deceleration part (approximately at ωt =105◦) in line with the increasing adverse pressure gradient.
Lastly, comparisons of the turbulence quantity (e.g., u′2) profiles with highly validated previous studies revealed a good agreement in the order of magnitude of near-wall turbulence. However, it was observed that the turbulence intensity in the early phases of the oscillatory cycle was lower in the present study than the values reported in the literature, which is expected given the relatively low Reynolds number (Rew = 1.8 ×105) used. This discrepancy can also be attributed to spatial averaging of turbulence quantities, which inherently includes flow regions where the boundary layer flow remains laminar during the early phases of the oscillatory motion. This result supports the findings that turbulence does not fully develop at these phases under such flow conditions.