In this article, we address the capabilities of physics-informed neural networks (PINNs) in assimilating the experimentally acquired mean flow of a turbulent separation bubble occurring in a diffuser test section. The training database contains discrete mean pressure and wall shear-stress fields measured on the diffuser surface as well as three-component velocity vectors obtained with particle image velocimetry throughout the volumetric flow domain. Imperfections arise from the measurement uncertainty and the inability to acquire velocity data in the near-wall region. We show that the PINN methodology is suited to handle both of these issues thanks to the incorporation of the underlying physics that, in the present study, are taken into account by minimizing residuals of the three-dimensional incompressible Reynolds-averaged Navier–Stokes equations. As a result, measurement errors are rectified and near-wall velocity profiles are predicted reliably. The latter benefits from the incorporation of wall shear-stress data into the PINN training, which has not been attempted so far to the best of our knowledge. In addition to demonstrating the influence of this novel loss term, we provide a three-dimensional, highly resolved, and differentiable model of a separating and reattaching flow that can be readily used in future studies.

Experimental methods are employed to investigate vortex rings generated by impulsively emitting pressurised air through a rectangular outlet of high aspect ratio into quiescent surroundings. The flow is characterised by a rapid transverse expansion as thick-cored, almost spherical vortex rings with a diameter more than 40 times the outlet width and entrainment rates of associated with Hill's spherical vortex, vorticity is continuously shed into a trailing jet, and the vortex rings evolve into unsteady thinner-core states. No separation between the vortex ring and the trailing jet (pinch-off) is observed. The present study provides new insights into the flow physics of non-parallel planar starting jets that significantly differ from parallel starting flows investigated previously. The potential for active flow control applications is discussed.

Spectral proper orthogonal decomposition (SPOD) is used to investigate the low-frequency behavior of a pressure-induced turbulent separation bubble (TSB) generated on a one-sided diffuser at a nominal velocity of 20 m/s. The input to the decomposition is the unsteady wall shear stress determined experimentally with an array of novel calorimetric shear-stress sensors. The SPOD spectrum indicates low-rank behavior for frequencies below 10 Hz, which correspond to Strouhal numbers [ ] when built with the average TSB length and the nominal incoming velocity. A low-order model constructed with only the first SPOD mode at each frequency up to 10 Hz reveals that the TSB is contracting and expanding in a manner reminiscent of the low-frequency breathing motion recently documented in other TSBs. A comparison between SPOD and space-only proper orthogonal decomposition (POD) indicates that both decompositions are able to capture the low-frequency breathing motion as long as space-only POD is applied on signals previously low-pass filtered at 10 Hz. However, the superiority of SPOD is revealed by its inherent capacity to express the frequency range amenable to low-rank behavior. In addition to offering new insights into the low-frequency dynamics of TSBs, the present results also demonstrate the usefulness of calorimetric shear-stress sensors for the study of turbulent separating and reattaching flows.

The interaction between starting jets and a steady cross-flow with a zero-pressure-gradient, turbulent boundary layer is studied experimentally. A device typically used in flow control applications is employed as jets of compressed air are injected into the cross-flow through a rectangular high-aspect-ratio outlet. Investigating different velocity ratios between starting jets and cross-flow within the interval recently determined for this type of non-parallel planar starting jets when emitted into quiescent surroundings (Steinfurth & Weiss, J. Fluid Mech., vol. 903, 2020, A16). The findings presented in the current article can promote the sophisticated selection of actuation parameters in active mixing and separation control.

Phase-resolved volumetric velocity measurements of a pulsed jet are conducted by means of three-dimensional particle tracking velocimetry (PTV). The resulting scattered and relatively sparse data are densely reconstructed by adopting physics-informed neural networks (PINNs), here regularized by the Navier–Stokes equations. It is shown that the assimilation remains robust even at low particle densities (ppp<10−3) where the mean particle distance is larger than 10% of the outlet diameter. This is achieved by enforcing compliance with the governing equations, thereby leveraging the spatiotemporal evolution of the measured flow field. Thus, the PINN reconstructs unambiguously velocity, vorticity, and pressure fields, enabling a robust identification of vortex structures with a level of detail not attainable with conventional methods (binning) or more advanced data assimilation techniques (vortex-in-cell). The results of this article suggest that the PINN methodology is inherently suited to the assimilation of PTV data, in particular under conditions of severe data sparsity encountered in experiments with limited control of the seeding concentration and/or distribution.

A novel method capable of assessing flow fields in a quick and relatively simple manner is introduced. In an extension to the classical qualitative flow visualization by means of cotton or polymeric tufts, digital data processing is used to extract the orientation of these tufts. This information can be related to physical quantities, in particular to time- and space-dependent velocity signals. The capability of this method is demonstrated in two test scenarios. First, it is applied to gain information on the unsteady near-wall flow along a turbulent separation bubble. Second, the two-component velocity field in the wake of a generic car model is measured, allowing for a quantification of the recirculation zone dimensions. Validation measurements with conventional techniques, e.g., particle image velocimetry, unsteady pressure measurements and hot wire anemometry, are conducted throughout the study. These generally suggest that the novel approach provides a quick and reasonably good quantitative overview of the flow configurations. However, the measurement error may be substantial in flow regions of low velocity or dominated by high-frequency oscillations. Graphic