In the field of fluid mechanics, there has been a significant shift towards the integration of machine and deep learning techniques to address challenges in reduced-order modeling, flow feature analysis, and control, especially within the realm of active flow control (AFC) for objectives such as lift optimization and drag reduction. Deep learning has taken a central role in advancing state-of-the-art AFC methods by creating data-driven models that mitigate the computational demands of conventional Computational Fluid Dynamics (CFD) simulations, enabling real-time fluid control. Despite the predominance of models trained offline and focused on simple scenarios like laminar flow around bluff bodies, the utility of sophisticated learning methods in AFC has remained largely unexplored.

This research introduces a novel benchmark in fluid dynamics—a soft robotic tentacle actuator—to evaluate the effectiveness of deep learning architectures in complex flow control situations. Through the comparison of online and offline learning frameworks for predicting system behavior, the study elucidates the strengths and limitations of deep learning networks in AFC. The findings underscore the constraints faced by deep learning architectures when dealing with aperiodic motions and demonstrate the significant benefits of adopting an online learning approach over offline training methods, thereby highlighting the advantages of adaptive learning strategies in complex AFC scenarios. The online learning framework displays more stability and increased quality of forecasts at larger time horizons.

Hydrodynamic drag significantly influences the fuel efficiency and operational performance of marine vessels. Gas Injection Drag Reduction (GIDR) techniques, such as Bubble Drag Reduction (BDR), Transitional Layer Drag Reduction (TLDR), and Air Layer Drag Reduction (ALDR), have shown promise in reducing skin friction drag by introducing gas into the boundary layer. However, the influence of hull protrusions, such as sacrificial anodes and fouling, on these techniques remains poorly understood. This research addresses the question: How do the placement and dimensions of upstream cylindrical obstacles affect the formation and stability of gas phases (bubbly, transitional, and air layer) in a turbulent boundary layer (TBL)? To investigate this, protrusions are represented as cylindrical objects placed upstream of the gas injector, exploring their effects on the gas injection regimes in a TBL.

The study examines three gas injection regimes: bubbly, transitional, and air layer. In the bubbly regime, the focus is on bubble morphology and velocity distributions. The transitional regime investigates non-wetted areas and air layer morphology. For the air layer regime, the study examines the length and stability of the air layer near the injector. Image processing techniques are used to distinguish the air layer from the surrounding fluid, providing accurate assessments of the properties of the air layer.

The results demonstrate that cylindrical obstacles have a significant effect on the gas phases. In the bubbly regime, they alter bubble morphology and velocity, producing localized velocity peaks. In the transitional regime, obstacles lead to the formation of non-wetted areas and funnel-shaped regions. For the air layer regime, although the length of the air layer remains mostly unaffected, its stability near the injector is disrupted by upstream obstacles due to wake-induced instabilities. These findings highlight the critical role of obstacle placement and dimensions in shaping gas injection regimes.

Future research is needed to enhance understanding in areas such as the integration of velocity measurements (e.g., Particle Image Velocimetry), widening the experimental field of view to capture more comprehensive flow dynamics, and exploring the impact of various obstacle sizes and geometries. These investigations will be crucial for refining gas injection drag reduction strategies and advancing practical marine applications.

With an annual production exceeding 390 million metric tons of plastic, the issue of plastic waste entering the environment and accumulating in rivers and other large bodies of water is escalating. Large-scale models used to estimate the transport and accumulation of said waste often neglect the effect of smaller turbulent scales.

This research focuses on studying the dynamics of floating particles in a turbulent free surface with a mean flow generated by a turbulent jet. Particles released in the surface are observed to primarily accumulate in clusters, with over 90% of all particles belonging to clusters as the seeding volume is increased. Clusters are divided into three groups based on size to highlight size-dependant clustering dynamics. The groups are defined as small, medium and large clusters. The dynamics of these clusters are studied based on the parameters size, shape and compactness, as well as the variation of those parameters over the tracked life of a cluster. A combination of these parameters is concluded to be crucial to fully characterize the cluster dynamics. Medium clusters are observed to experience the largest variations in both size and shape. The influence of the flow on the clusters is studied to conclude that turbulent flow regions lead to higher variations in both size and shape. Large-scale rotating flows are observed to vary the shape but not the size of the cluster. No flow dynamics are observed to affect the compactness.

Surface pressure measurements are a crucial part of aerodynamic/hydrodynamic analysis and are essential when studying complex flow phenomena. Although pressure-sensitive paint (PSP) has become established as a non-intrusive field measurement technique for high-speed applications in air, a comparable method does not yet exist for usage underwater....

The turbulent boundary layer development under the influence of an air cavity is studied experimentally using planar PIV, with the aim of gaining insight and building upon the flow physics typically encountered in the application of air layer drag reduction. A detection technique based on correlation values is implemented to obtain an approximate shape of the air cavity and the location of the air-water interface. The technique was successful in identifying the maximum cavity thickness with sufficient accuracy. The leading and trailing edges of the cavity however, were harder to identify, the former owing to a limitation of the developed technique and the latter due to the dynamic nature of the flow and a slightly limited FOV. The ratio of the initial boundary layer thickness to the maximum thickness of the air cavity is 6.7, and as a consequence the boundary layer did not separate at the leeward side of the air cavity. The turbulent boundary layer is observed to feel the presence of the air cavity up to 8.5-9.5 cm upstream due to an adverse pressure gradient. Alternating streamwise pressure gradients are generated due to the curvature of the air cavity: from an adverse to favourable and back to adverse. Compared to solid bump studies in literature, additional perturbations due to a free-slip boundary condition and the unstable nature of air cavity increase the complexity of the current flow. The mean velocity profile and stresses are able to capture the effects of alternating streamwise pressure gradients and air injection, with variations mostly restricted to the inner region. Effects of streamline curvature in the outer region are found to be minimal, while potential effects of the free-slip condition were much harder to identify separately and further research would be needed to appropriately assess them. The mean velocity profile is found to deviate from the classic logarithmic behaviour at the apex of the air cavity, although the flow does not seem to relaminarise. Quadrant analysis shows differences in Reynolds stress producing events compared to the baseline turbulent boundary layer case hinting at possible alteration to coherent structuring of the turbulent boundary layer developing below the air cavity.