This paper explores integrating artificial intelligence (AI) segmentation models, particularly the Segment Anything Model (SAM), into fluid mechanics experiments. SAM’s architecture, comprising an image encoder, prompt encoder, and mask decoder, is investigated for its application in detecting and segmenting objects and flow structures. Additionally, we explore the integration of natural language prompts, such as BERT, to enhance SAM’s performance in segmenting specific objects. Through case studies, we found that SAM is robust in object detection in fluid experiments. However, segmentations related to flow properties, such as scalar turbulence and bubbly flows, require fine-tuning. To facilitate the application, we have established a repository (https://github.com/AliRKhojasteh/Flow_segmentation) where models and usage examples can be accessed.

The dataset contains Eulerian velocity and pressure fields, and Lagrangian particle trajectories of the wake flow downstream of a smooth cylinder at a Reynolds number equal to 3900. An open source Direct Numerical Simulation (DNS) flow solver named Incompact3d was used to calculate the Eulerian field around the cylinder. The synthetic Lagrangian tracer particles were transported using a fourth-order Runge-Kutta scheme in time and trilinear interpolations in space. Trajectories of roughly 200,000 particles for two 3D sub-domains are available to the public. This dataset can be used as a test case for tracking algorithm assessment, exploring the Lagrangian physics, statistic analyses, machine learning, and data assimilation interests.

In the present study, we investigate the computation of the Lagrangian second-order structure-function to characterise the multiscale dynamics of turbulence from measured particle trajectories. We performed time-resolved three-dimensional particle tracking velocimetry (4D-PTV) to study the anisotropic and inhomogeneous flow field of the wake behind a cylinder at a Reynolds number equal to 3900. We performed Lagrangian statistical analysis on nearly 12000 trajectories for 4000 time steps.

Advances in time-resolved three-dimensional Particle Tracking Velocimetry (4D-PTV) techniques have consistently revealed more accurate Lagrangian particle motions. A novel track initialization technique as a complementary part of 4D-PTV, based on local temporal and spatial coherency of neighbor trajectories, is proposed. The proposed Lagrangian Coherent Track Initialization (LCTI) applies physics-based Finite Time Lyapunov Exponent (FTLE) to build four frame coherent tracks. We locally determine Lagrangian coherent structures among neighbor trajectories by using the FTLE boundaries (i.e., ridges) to distinguish the clusters of coherent motions. To evaluate the proposed technique, we created an open-access synthetic Lagrangian and Eulerian dataset of the wake downstream of a smooth cylinder at a Reynolds number equal to 3900 obtained from three-dimensional direct numerical simulation. Performance of the proposed method based on three characteristic parameters, temporal scale, particle concentration (i.e., density), and noise ratio, showed robust behavior in finding true tracks compared to the recent initialization algorithms. Sensitivity of LCTI to the number of untracked and wrong tracks is also discussed. We address the capability of using the proposed method as a function of a 4D-PTV scheme in the Lagrangian particle tracking challenge. We showed that LCTI prevents 4D-PTV divergence in flows with high particle concentrations. Finally, the LCTI behavior was demonstrated in a jet impingement experiment. LCTI was found to be a reliable tracking tool in complex flow motions, with a strength revealed for flows with high velocity and acceleration gradients.

In the present study, the film cooling effectiveness in the near field (x/D < 4) of a single inclined film-cooling jet was measured using fast-response pressure sensitive paint (fast-PSP) and a low-frame-rate CCD camera. Previous experimental data demonstrated considerable variation in this region, and good agreement was established beyond it (x/D > 4). The blowing ratios M = 0.5 and 1.0 were used. The coolant fluid was nitrogen and the air was the mainstream fluid, and both were kept at the same temperature. A fast-PSP measurement technique was used to determine the variations of the film cooling effectiveness with time. The contours of the time-averaged film cooling effectiveness demonstrated that the coolant spread on the surface throughout the near-hole region at M = 0.5, while at M = 1.0, the coolant jet detached from the surface immediately behind the hole and reattached downstream around 1.5 D behind the hole's trailing edge. The spatial distribution of the film cooling effectiveness fluctuations and its cross-correlation pattern convincingly reflected the substantial influence of the energetic unsteady flow structures in the jet and cross-flow interaction. Subsequently, the Proper Orthogonal Decomposition (POD) method was used to identify the coherent parts of the film cooling effectiveness, which are regarded as the signatures of the convective large-scale vortical structures above the wall. At M=0.5, the near-hole region was subjected to the dominant influence of the counter-rotating vortex pair (CRVP), characterized by the first two POD modes, which contained up to 40% of the fluctuation energy. Two signatures of large-scale symmetric structures were identified with similar energy levels. The second two POD modes corresponding to the horseshoe vortex near the leading edge of the hole were identified and contained around 10% of the fluctuation energy. Phase-dependent variations of the large-scale convective signatures in relation to the quasi-periodic CRVP and horseshoe vortex were separately detected. At M = 1.0, the signatures of the CRVP and the horseshoe vortex were also seen in the POD modes, though they were relatively difficult to distinguish.