• Home
• Program
• Contact

10:45   Wall Bounded Turbulence 4 10:45 15 mins #224 TURBULENT BOUNDARY LAYER OF THE FLOW PAST AN INFINITE SWEPT-BACK WING Carlo Alessio Suardi, Alfredo Pinelli, Mohammad Omidyeganeh Abstract: Highly resolved incompressible large eddy simulations (LES) of the flow past an infinite 30 degrees swept-back wing are carried out and the results are compared to the ones obtained for an un-swept wing. The Reynolds number (based on the chord and the free stream velocity, normal to the leading edge) is set to 50000. The infinite wing is modelled by using spanwise periodic conditions over a wavelength set to 40% of the chord. Along the chord, a NACA-4412 cambered profile is chosen and the geometric incidence is set to 5 degrees. The far field, free stream boundary condition is assigned according to the potential flow solution obtained for the given profile at the mentioned incidence. Differently, on the portions of the outer boundary characterised by an outgoing mean velocity vector, a non reflective condition is imposed. The effect of the sweep angle is kept into account by assigning a mean spanwise velocity component to the free stream velocity vector. The boundary layer developing on the wing undergoes a laminar to turbulent transition forced by a set of perturbations superimposed to the incoming free stream. In particular, these finite size perturbations are produced by a twin DNS that simulates the flow behind a grid. By modulating the parameters of the grid generated turbulence, it is possible to choose the integral and the dissipative scales of the incoming turbulence and its intensity. Manipulating the character of the incoming flow allows to model conditions virtually similar to the landing or take off scenarios of commercial aircraft. Although the swept wing configuration is adopted to avoid the appearance of shock waves and to control the induced drag in transonic flight, the detailed structure of the resulting turbulent boundary layer has not yet been completely understood. In particular, our focus is on studying the modification of the wall cycle when the structures embedded in the buffer layer are submitted to both a pressure gradient and a mean cross wind. Recently, in the case of an un-swept wing, it has been found that the configuration of the coherent wall structures plays an important role in case of an incipient separation when the local pattern of the wall shear determines the appearance of localised nuclei of reversed flow [2]. Inspired by these recent findings, the objective of our investigation is to assess the effect of the sweep angle on the developing turbulent boundary layer experiencing an imposed pressure gradient distribution with particular emphasis on the characterisation of the buffer layer coherent motions and their role on the incipient separation. In this context we have already carried out simulations that consider both a swept and an un-swept wing. As an example, in figure we report the maps of the instantaneous friction coefficient on the wing suction side for both cases (isolines represent a zero wall shear). For the case of un-swept wing the flow experiences large separated areas, while in the swept case these regions appear as small spots embedded within street canyons formed by the streamwise velocity streaks that are bent by the action of the crossflow. Thus, the crossflow appears to have a behaviour promoting a more energetic boundary layer that delays separation by confining and squeezing the local spots of reversed flow. Concerning the numerical details, the simulations have been carried out by numerically integrating the incompressible LES equations, adopting the ILSA closure proposed by [1] while fully resolving the near-wall regions. The equations are space discretised with a second order finite volume method using a co-located formulation on a body fitted C-type structured mesh. FFTs are employed along the spanwise direction. The time evolution of the flow is carried out using a second order fractional step method. Apart from a detailed analysis of the structure of a turbulent boundary layer submitted to a concurrent pressure gradient and crossflow, we will report also the influence that the character of the incoming grid turbulent flow has on the transition and on the generation of localised separated spots in both swept and un-swept wings. 11:00 15 mins #379 Experimental investigation of coherent structures in a flat plate turbulent boundary layer at ReΘ=10,000 Christina Voß, Reinhard Geisler, Andreas Schröder, Markus Rütten, Matteo Novara Abstract: The production of viscous drag within the turbulent boundary layer (TBL) leaves still unresolved questions in turbulence research. Coherent structures consisting of large meandering formations of positive and negative streamwise velocity fluctuations in the upper log-layer contribute to the near-wall stress-producing cycle as well [1]. The interplay between these large-scale structures and those of the near-wall, which can be clearly separated from each other for high Reynolds numbers (ReΘ>1,500) [2], is of considerable importance for getting a detailed understanding of turbulence length scales and spectra. For this reason, measurements were carried out on a zero pressure gradient flat plate that was installed in the DLR (German Aerospace Centre) low-speed wind tunnel SWG (Cross Wind Test facility Göttingen). The flat plate model was 7.67 m long and 2.39 m wide. In order to guarantee the development of a TBL without disturbances from the boundary layer of the wind tunnel floor, the flat plate was mounted 400 mm above and parallel to the bottom of the test section. Different measurement techniques were applied at a region of interest located 6.9 m (x=0) downstream of the elliptical leading edge of the flat plate model: On the one hand, flow investigations were made using three overlapping stereo-PIV-systems (Particle Image Velocimetry) with a measurement plane perpendicular to the flat plate (Fig 1-left). The complete TBL thickness was captured, whereas special attention was paid on the detection of large-scale motions in the upper log-region over a long distance. Additionally, for getting the full velocity information in the near-wall region of the TBL, the Multi-Pulse Shake-The-Box 3D Lagrangian Particle Tracking measurement technique (MP-STB) [3] was applied to a field of view of about 80 x 120 x 10 mm^3 in streamwise, spanwise and wall-normal direction, respectively. Velocity vectors measured at randomly distributed tracks within the measurement volume have been interpolated by the Navier-Stokes-regularized interpolation method FlowFit [4] for the purpose of determining the full velocity gradient tensor. On the right in Fig.1, iso-surfaces of the Q-criterion gained by FlowFit interpolation of an instantaneous MP-STB result (97,000 tracks) are shown. Even small scale vortical structures within the volume can be resolved. For both measurement setups, a statistical significant number of samples has been acquired which provides a huge database for a detailed structure analysis of the coherent structures (e.g. with two-point-correlations). Especially the interactions of these structures within the near-wall and log-region will be examined. [1] N. Hutchins, and I. Marusic. Evidence of very long meandering features in the logarithmic region of turbulent boundary layers. J. Fluid Mech. vol 579: 1–28, 2007. [2] P. Schlatter, Q. Li, R. Örlü, F. Hussain, and D. S. Henningson. On the near-wall vortical structures at moderate Reynolds numbers. European Journal of Mechanics B/Fluids vol 48: 75–93, 2014. [3] M. Novara, D. Schanz, N. Reuther, C. J. Kähler, and A. Schröder: Lagrangian 3D particle tracking for high-speed flows: Shake-The-Box for multi-pulse systems. Exp. In Fluids 57:128, 2016. [4] S. Gesemann, F. Huhn, D. Schanz, and A. Schröder: From noisy particle tracks to velocity, acceleration and pressure fields using B-splines and penalties. 18th Int Symp on Appl of Laser and Imaging Tech. to Fluid Mech, 2016. 11:15 15 mins #421 DNS study on Reynolds stress anisotropy in a turbulent boundary layer with separation and reattachment Hiroyuki Abe Abstract: As for the abstract, please see the attached pdf file. 11:30 15 mins #588 Embedded Large Eddy Simulation of Streamwise Vortices Within a Spatially Developing Turbulent Boundary Layer Andrew Mole, Alistair Revell Abstract: For the increasingly complex applications targeted by computational fluid dynamics, the use of Reynolds averaged Navier Stokes (RANS) modelling is either insufficiently inaccurate or inherently limited by the range of scales it returns, while the use of scale resolving simulations is too costly. As a compromise with wall-resolved Large Eddy Simulation (LES), hybrid RANS-LES methods endeavour to combine both approaches in a synergistic manner. One such approach, Embedded LES can be used to notable effect by minimising the required spatial regions of scale resolution and employing RANS in the remainder of the domain. Towards this aim, one must transition statistically-steady RANS quantities into a time-varying velocity field containing coherent eddy structures and a significant body of work has investigated the use of Synthetic Eddy Method (SEM) for this purpose [3]. As recently demonstrated by [5], with improved normalisation of the fluctuations on the inlet plane, the method can accomplish a short recovery of the turbulent statistics. However, it remains sensitive to the prescribed variation of length scale, as shown in Figure 1. In the present work, the SEM is applied to a spatially developing boundary layer case in the fully turbulent regime. This flow presents a different length scale variation which is first investigated to minimise development length, according to trends identified in earlier work for internal flows [2]. The method will also be adapted to implement a region of Zonal Detached Eddy Simulations in place of the LES in order to further reduce the computational cost according to [1]. Turbulent boundary layers with embedded large scale streamwise vortices represent an important class of three- dimensional wall-bounded flows due to their common occurrence; either intentional as with vortex generators or otherwise. Mean streamline curvature effects are often poorly modelled by standard unmodified RANS models the complex interplay between vortex breakdown and turbulent boundary layer physics remain beyond even more advanced single point turbulent closures. Many studies have examined at the interaction between a spatially developing turbulent boundary layer with embedded streamwise vortices, particularly using scale-resolving methods as shown in Figure 3 [4], but cost and complexity of domain geometry are limiting in these cases. We will look at replicating these results using embedded eddy simulations with the synthetic eddy method providing the turbulent structures in order to develop the method for more complex inflow conditions and industrial engineering applications. 11:45 15 mins #299 A comparison of surface roughness effect in channel and Taylor-Couette flows Pourya Forooghi, Pieter Berghout, Richard Stevens, Philipp Schlatter, Detlef Lohse, Daniel Chung, Roberto Verzicco, Bettina Frohnapfel Abstract: Surface roughness is relevant in a wide range of applications such as atmospheric boundary layers over urban structures, turbomachinery and biological flows. The majority of available knowledge on rough wall-bounded turbulence, however, stems from canonical flow configurations such as plane channels and straight pipes. This knowledge includes, among others, the predictive correlations for the estimation of equivalent sand roughness, which is the main parameter sought by engineers . An open question is the extent to which the derived correlations and scaling rules are flow-independent. We investigate this question by comparing channel flow (CH) with the fundamentally different Taylor-Couette flow (TC). The TC is studied for both outer- and inner-cylinder rotation (TC-O and TC-I). The former case features a stable shear flow while in the latter case centrifugal instabilities give rise to Taylor vortices. The roughness geometry is identical in all flow configurations and generated by a random distribution of similar roughness elements. The governing Navier-Stokes equations are directly solved using a pseudo-spectral solver for CH and a finite difference solver for TC . The roughness is implemented using Immersed Boundary Method. The simulations are run at two friction Reynolds numbers $Re_\tau=400$ and 550$ and the roughness element height $k^+$ varies between 50 and 100, allowing the observation of the fully rough regime. In TC, the inner-outer cylinder ratio is fixed at 0.714. Figure 1 presents a sample time-averaged TC-I solution, where the presence of Taylor vortices is indicated by non-parallel wakes behind the roughness elements. The values of the roughness function $\Delta U^+$ -- the roughness-induced downward shift in the mean inner-scaled streamwise velocity profile -- are calculated by comparing the rough case with the corresponding smooth one. The behaviour of $\Delta U^+$ with $k^+$ for CH and TC is compared in the fully rough regime to determine any flow-dependence of the equivalent sand roughness. Moreover, the extent of the roughness sublayer -- the near-wall region where turbulence is directly affected by roughness -- for different flow types is compared. We also provide a comparison of higher-order turbulence statistics. 12:00 15 mins #448 Network-based characterization of passive-scalar plume dynamics in a turbulent boundary layer Stefania Scarsoglio, Giovanni Iacobello, Luca Ridolfi, Massimo Marro, Pietro Salizzoni Abstract: The investigation of the interplay between the passive-scalar dynamics and the velocity field in a turbulent boundary layer (TBL) is essential to understand and model the transport of contaminants in atmospheric TBLs. Specifically, the characterization of concentration peaks is crucial to identify the possibility to exceed concentration limits [1]. In this work, we take advantage of the recent developments in network science to study the vertical transport of a passive-scalar in a TBL. Experimental measurements of velocity and passive-scalar are performed in a wind tunnel which generates a TBL with a free-stream velocity u∞=4.94 m/s and a thickness δ=314 mm [5]. A mixture of air-ethane was released from an L-shaped source of diameter D, located at a distance h_s/δ ≈ 0.24 from the wall (see Fig.1a). Time-series of vertical passive-scalar flux, w'c', were obtained for two diameter values, D={3, 6} mm, at different streamwise and wall-normal positions, (x,z). Here, w' and c' are the fluctuating components of wall-normal velocity and concentration, respectively. Data were measured over a time interval of 180 s, with sampling frequency equal to 1000 Hz, thus obtaining NT=1.8·10^5 time-samples. To investigate the time-series of w'c', the visibility algorithm was exploited [4]. The visibility-graph technique is one of the simplest and most employed approaches to map a time-series into a complex network [6]. According to this method, each point of the time-series corresponds to a node of the network, which therefore has NT nodes. A link between two nodes exists if the straight line connecting the two data points lies above all the other in-between data. To highlight the key mechanisms affecting the plume dynamics -- i.e., the meandering of the plume centre of mass and its relative dispersion [2] -- we investigated two network metrics: the average peak occurrence, Φ, and the transitivity, Tr. These metrics are related to the occurrence of extreme events and irregularities in the time-series, respectively [3]. In Fig.1b the behaviour of Φ and Tr is shown as a function of z/δ at two streamwise locations: in the source proximity and far from the source. Relative dispersion along z is well captured by Φ, which significantly decreases for both D cases towards the far field (see Fig. 1b, left panels). The Tr magnitude does not substantially vary from near to far field. However, the Φ and Tr differences between the two diameters along the source axis denote an important variation of the meandering effect in the near field (see Fig.1b, top panels). The network-based analysis helps characterizing the impact of the source diameter on the passive-scalar plume dynamics in a TBL. Based on the present findings, the near-field meandering is influenced by the source size much more than relative dispersion throughout the domain is. References [1] J.E. Fackrell and A.G. Robins. The effects of source size on concentration fluctuations in plumes. Bound.-Lay. Meteorol., 22(3):335–350, 1982. [2] F. Gifford Jr. Statistical properties of a fluctuating plume dispersion model. In Adv. Geophys., 6, pages 117–137. Elsevier, 1959. [3] G. Iacobello, S. Scarsoglio, and L. Ridolfi. Visibility graph analysis of wall turbulence time-series. Phys. Lett. A, 382(1):1–11, 2018. [4] L. Lacasa, B. Luque, F. Ballesteros, J. Luque, and J. C. Nuño. From time series to complex networks: The visibility graph. Proc. Natl. Acad. Sci., 105(13):4972–4975, 2008. [5] C. Nironi, P. Salizzoni, M. Marro, P. Mejean, N. Grosjean, and L. Soulhac. Dispersion of a passive scalar fluctuating plume in a turbulent boundary layer. part i: Velocity and concentration measurements. Bound.-Lay. Meteorol., 156(3):415–446, 2015. [6] Y. Zou, R. V. Donner, N. Marwan, J. F. Donges, and J. Kurths. Complex network approaches to nonlinear time series analysis. Phys. Rep., 2018. 12:15 15 mins #178 Experimental and numerical investigation of attached flow structures in tubulent boundary layers Ilkay Solak, Sricharan Srinath, Jean-Philippe Laval, John Christos Vassilicos, Christophe Cuvier, Jean-Marc Foucaut, Michel Stanislas Abstract: Srinath et al. (PRE 2018) have shown from experimental PIV data that the Towsend's attached eddy range of the streamwise energy spectra can be fitted with a -1-p scaling with p varying smoothly with distance to the wall from negative values in the buffer layer to positive ones in the inertial layer. The authors related this scaling with a simple on-off model that relies on a −2 slope of the length distribution of the fluctuating streamwise velocity large-scale structures. Recently, Solak et al (PRE, 2018) investigated the streamwise large scale structures from a DNS of turbulent boundary layer at moderate Reynolds number. Using a skeletonization technique, the large scale structures have been investigated in details and the hypotheses of the Srinath's model was confirmed.