The volumetric defocusing particle tracking velocimetry (DPTV) approach is applied to measure the flow in the sub-millimeter gap between the disks of a radially grooved open wet clutch. It is shown that DPTV is capable of determining the in-plane velocities with a spatial resolution of $$12\;\upmu \mathrm{m}$$ 12 μ m along the optical axis, which is sufficient to capture the complex and small flow structures in the miniature clutch grooves. A Couette-like velocity profile is identified at sufficient distance from the grooves. Moreover, the evaluation of the volumetric flow information in the rotor-fixed frame of reference uncovers a vortical structure inside the groove, which resembles a cavity roller. This vortex is found to extend well into the gap, such that the gap flow is displaced towards the smooth stator wall. Hence, the wall shear stress at the stator significantly increases in the groove region by up to $$15\%$$ 15 % as compared to the ideal linear velocity profile. Midway between the grooves, the wall shear stress is around $$4\%$$ 4 % lower than the linear reference. Furthermore, significant amounts of positive radial fluxes are identified inside the groove of the rotor; their counterpart are negative fluxes in the smooth part of the gap. The interaction of the roller in the groove and the resulting manipulation of the velocity profile has a strong impact on the wall shear stress and therefore on the drag torque production. In summary, this DPTV study demonstrates the applicability of such particle imaging approaches to achieve new insights into physical mechanisms of sub-millimeter gap flow scenarios in technical applications. These results help to bring the design- and performance-optimization processes of such devices to a new level. Graphic abstract

No abstract

The presented work addresses the problem of particle detection with neural networks in defocusing particle tracking velocimetry. A novel approach based on synthetic training data refinement is introduced, with the scope of revising the well documented performance gap of synthetically trained neural networks, applied to experimental recordings. In particular, synthetic particle image data is enriched with image features from the experimental recordings by means of deep learning through an unsupervised image-to-image translation. It is demonstrated that this refined synthetic training data enables the neural-network-based particle detection for a simultaneous increase in detection rate and reduction in the rate of false positives, beyond the capability of conventional detection algorithms. The potential for an increased accuracy in particle detection is revealed with neural networks that utilise small scale image features, which further underlines the importance of representative training data. In addition, it is demonstrated that neural networks are able to resolve overlapping particle images with a higher reliability and accuracy in comparison to conventional algorithms, suggesting the possibility of an increased seeding density in real experiments. A further finding is the robustness of neural networks to inhomogeneous background illumination and aberration of the images, which opens up DPTV for a wider range of possible applications. The successful application of synthetic training-data refinement advances the neural-network-based particle detection towards real world applicability and suggests the potential of a further performance gain from more suitable training data.

The present experimental study revolves around the applicability of a Bragg-shifted laser Doppler velocimetry profile sensor (LDV-PS) in use for open wet clutch flow scenarios, where sub-millimeter gap height and textured surfaces are present. It is shown that the LDV-PS is capable to determine angular-resolved 1D3C velocity information, with all complex flow structures, depicted properly that are present in a radial groove. For the flow measurements the sensor is tilted to $$\pm 30^{\circ }$$ ± 30 ∘ compared to the axial orientation to enable the opportunity to reconstruct angular-resolved 1D3C velocity fields from two consecutively conducted runs. This facilitates measurement results with high axial and angular resolution for the complete open clutch flow and proves for the first time, that a profile sensor is capable to extract 3C information with the mentioned method. The results show that all characteristic flow structures occurring in the investigated sub-millimeter rotor-stator gap flow can be recorded properly. This insight renders the LDV-PS a promising and straight-forward applicable means to support industry-relevant research so as to uncover formerly hidden flow features and thus contribute to advanced development approaches for the respectively considered applications. Graphical abstract

The present work aims at the improvement of particle detection in DPTV by means of a novel hybrid approach. Two deep learning approaches, namely Faster R-CNN and RetinaNet are compared to the performance of two benchmark conventional image processing algorithms for DPTV. For the development of a hybrid approach with improved performance, the different detection approaches are evaluated on synthetic and images from an actual DPTV experiment. First, the performance under the influence of noise, overlaps, seeding density and optical aberrations is discussed and consequently advantages of neural networks over conventional image processing algorithms for image processing in DPTV are derived. Furthermore, current limitations of the application of neural networks for DPTV are pointed out and their origin is elaborated. It shows that neural networks have a better detection capability but suffer from low positional accuracy when locating particles. Finally, a novel Hybrid Approach is proposed, which uses a neural network for particle detection and passes the prediction onto a conventional refinement algorithm for better position accuracy. A third step is implemented to additionally eliminate false predictions by the network based on a subsequent rejection criterion. The novel approach improves the powerful detection performance of neural networks while maintaining the high position accuracy of conventional algorithms, combining the advantages of both approaches.