11:40   PIV/PTV methods and applications II Volumetric PTV on the flow behind a rolling F1 wheel Isabella Fumarola, Parv Khurana, Alexandra I. Liosi, Harris Warburton, Fahez Siddiqi, Spencer J. Sherwin, Jonathan F. Morrison Abstract: Abstract For open-wheel racing cars, such as Formula 1 (F1), the wheels contribute 30-40% of the total aerodynamic drag [1]. This is due to the complex flow structures generated behind each wheel, which effectively is a rotating bluff body in proximity of the ground. A limited number of studies on rolling wheels are available in the literature, either numerical or experimental. Recent developments in Particle Tracking Velocimetry (PTV) for measuring flow within large volumes make the technique particularly suitable for analysing such complex flows. In this work, results from an experiment using PTV to measure the flow in the wake of a rotating wheel on a rolling road in the wind tunnel are presented. The dataset is also used to validate numerical simulations using Large Eddy Simulations (LES) based on high-order spectral h/p element methods [2]. The results are unique and represent a valuable contribution to the F1 community. Introduction The wheels on a F1 car are required by regulation to be uncovered. This means that they contribute significantly to the total drag of the car, not only because they effectively are bluff bodies rotating on the ground, but also because their wake interferes with the other components of the car, reducing the overall efficiency of the aerodynamically optimised vehicle. The contribution of the tyres to the total drag has been estimated to be between 30-40% for an F1 car [1]. In recent years, FIA regulations have limited the use of front wing elements to control air flow around the tyres. This makes the understanding of the flow behind the rotating tyre essential to enhance aerodynamic performance and stability of the car. The wake behind a rolling wheel in isolation exhibits a complex 3D flow, with separation, a recirculation regions and several vortical structures. Literature is relatively limited; a list of papers can be found in [3]. In addition, the wake topology is highly sensitive to factors such as geometry details (tyre grooves or deformation) and the contact patch. The latter represents the most challenging problem for both numerical simulation and experiments. In fact, for the former it is hard to numerically model a reliable deformation of the tyre, for the latter it is difficult to experimentally control or measure the shape of the contact. Available experiments have not always used a moving ground plane, despite is importance in accurately estimating forces. Most experimental studies have measured the forces acting on the wheel or the pressure either in the wake or on the wheel. Some have analysed the flow field, but mostly used point wise measurement techniques, laser Doppler or hot-wire anemometry, and fewer two-dimensional PIV. Numerical studies on wheel aerodynamics have not always considered rotation, and most of them assumed a non-deformable tyre. Croner et al. [4] carried out unsteady RANS simulations, which showed a good agreement validated by an experiment, showing a good agreement for the average flow topology as well as the unsteady behaviour. In this work, volumetric PTV is used for the first time to investigate the flow behind a rolling F1 wheel in a 10x5 wind tunnel at Imperial College London. The recent development of this technique, by combining high-power LEDs with Helium-Filled Soap Bubbles (HFSB), enables experiments to measure the velocity field over a large volume using a relatively simple setup [5]. The motivation for this work is to create a unique database for validating numerical simulations of this geometry. In particular, the recent development of Large Eddy Simulations (LES) using high-order spectral h/p element methods [2] within the Nektar++ framework [6] makes it a valuable tool for investigating flows around complex geometries at high Reynolds (Re) numbers [7].The experimental data used to validate novel numerical simulations are on the isolated rolling wheel using Nektar++ for the first time. The simulation assumes a rigid rotating wheel equipped with the same strut used in the experiment to secure the model to the tunnel floor. Experimental setup The experiment was carried out in the lower test section of the 10x5 wind tunnel at Imperial College London. The facility, as shown in Fig. 1, features a test section measuring 3 m x 1.5 m, 20 m long, equipped with a moving ground plane (rolling road) that simulates the effect of ground proximity. Two suction systems, located ahead of the rolling road, are used to remove the boundary layer. The tyre model used in this experiment is a 50% scale model of the Michelin tyre used on the front left of the McLaren MP4-17D, which competed in the 2002 and 2003 F1 World Championship. To simplify the flow physics, the wheel hub was sealed on both sides with a wooden panel. This prevents any flow through the hub, which would add unnecessary complexity. The tyre is pneumatic, made of rubber, and has grooves. The pressure was held constant at 7 psi for every run during the experiment. The tyre was held by a strut from the outboard side and bold to the tunnel floor, figure 1 b). Two LaVision LED-Flashlight 300 were mounted side by side on the tunnel roof. Helium-Filled Soap Bubbles (HFSB) were inserted upstream of the model using seven Linear Nozzle Array (LNA), each of which produce about 280,000 bubbles/second. Due to the presence of the suction system used to remove the boundary layer generated on the floor of the wind tunnel, the LNA could not be mounted in the contraction of the wind tunnel, but they were mounted just after the primary suction system. Four high-speed 4 Mpx Phantom v641 cameras were mounted in line on the side of the wind tunnel at a working distance from the measurement volume of about 1 m, each of them mounting a Nikon 50 mm f/1.4D lens. The experiment was carried out at 9 m/s in single frame mode at image rate of 1.4 kHz and at 12.5 m/s in double pulse mode at 250 Hz. Results Fig. 2 shows the 3D isosurfaces using the λ2 criterion obtained from the experimental results. Large coherent 3D structures shedding from the rolling wheel are clearly observed. The system of six vortices already described in the literature [4] appears evident in the plot. The evolution of the two large upper vortices (A and B in figure 3) shedding from the top surface of the wheel follows a downwards trajectory that carries the vortex location aligned with the centre of the wheel axis at x/D < 0.6 after the bluff body. In the mid-plane, two small vortices (C and D) are visible. These are generated by the wheel hubs, causing the flow to become rotational around the wheel axis. Finally, close to the rolling road, a pair of strong jetting vortices (E and F) is evident. They are produced due to the interaction of the rotating front face of the tyre with the rolling ground plane, acting like a viscous pump and causing these two jets on either side of the contact patch. Figure 3 shows the average streamwise velocity (U/U∞) and streamwise average vorticity ωxD/U∞, normalised using the freestram velocity (U∞) and diameter of the wheel (D), in the wall-normal spanwise planes at three streamwise locations (X/D = 0.7, X/D = 1 and X/D = 1.2). The wake height is approximately 1D at X/D = 0.7 and it is almost constant up to X/D = 1.2. The width of the wake from the mid-plane upwards is roughly equal to the width of the wheel, 165 mm. The wake narrows then widens significantly at the floor due to the strong jetting vortices being pushed outwards by the rotating contact patch, widening the effective width of the tyre at the floor. Two regions of reversed flow are observed, one close to the mid-plane vortices and another close to the floor. The mid-plane is caused by the recirculation region behind the bluff body. The region close to the floor corresponds to a negative pressure peak, which has been observed in other experimental studies of rolling wheels. The influence of the structure holding the wheel appears evident in the mean velocity profile, breaking the geometrical symmetry of the wheel. The mean vorticity shows the system of six vortices as observed in the λ2 isosurfaces and in agreement with previous studies. The two upper vortices, A and B, are the strongest and of similar in size, while the mid-plane vortices quickly interact with each other and are the first to dissipate. Further analysis of the data will include spectral analysis to identify the frequencies associated to these unsteady structures and Proper Orthogonal decomposition (POD) analysis. The results will then be compared to numerical simulations. Conclusion The work explored the flow behind a rotating F1 wheel using for the volumetric PTV. The results are in agreement with previous studies and they will be further compared to novel numerical simulation using LES based on high- order spectral h/p element. The offers a unique dataset for the advance of understanding the flow behind a rotating tyre and offers a perfect example of the applicability of PTV to complex flows. Volumetric Flow Analysis Around an Accelerating Hand Using AI-Based Object Segmentation Ali R Khojasteh, Edwin Overmars, Gertjan Mulder, Alex Nila, Thomas Rockstroh, Jerry Westerweel Abstract: Advancements in three-dimensional flow velocimetry techniques, such as Shake-The-Box (STB), have significantly improved volumetric reconstruction of flow fields when an object is outside the field-of-view. However, accurately reconstructing flow around moving objects remains challenging due to occlusions and the potential introduction of ghost particles within the object domain, particularly when the object geometry is not a canonical case. In this study, we propose an artificial intelligence (AI)-based approach to reconstruct 3D objects in volumetric PIV/PTV experiments that relies on original calibration and camera image segmentations. A Hybrid Clustering-Based Post-Processing Method for Analyzing PIV Flow Field Measurements in Scramjet Combustor Guannan Men, Guang Chang, Hongjie Zhong Abstract: In scramjet combustor, there exists intense supersonic turbulence coupled with complex physical and chemical phenomena such as the intense light emissions caused by flame combustion. The velocity field measurement within a scramjet combustor faces significant challenges due to factors such as strong background light interference and particle injection inhomogeneity caused by high pressure gradients. This study proposes a postprocessing algorithm based on hybird clustering to effectively identify and eliminate erroneous vectors from transient velocity fields. The technique inputs the original velocity vectors, corresponding coordinates, and vorticity into a Gaussian Mixture Model (GMM) for binary classification, which yields probability values representing the likelihood of being classified as correct or erroneous. These probability values and velocity vectors (threedimensional vectors) are then clustered using the DBSCAN algorithm to assign labels that indicate whether the vectors are correct or erroneous. The time-averaged flow field information is subsequently obtained from the filtered velocity fields. The velocity field measurements were performed under the scramjet combustor cramjet operating mode, and the results demonstrate that the proposed post-processing algorithm effectively captures the velocity field patterns under different operating conditions. It also exhibits the capabilities of solving low-quality particle images caused by complex background interference and intense turbulence environments.

end %-->