Dynamic Mode Decomposition Analysis of Separated Boundary Layers Under Variable Reynolds Number and Free-Stream Turbulence

Author(s):  
M. Dellacasagrande ◽  
J. Verdoya ◽  
D. Barsi ◽  
D. Lengani ◽  
D. Simoni

Abstract A flat plate boundary layer has been surveyed by means of time-resolved particle image velocimetry (PIV) under variable Reynolds number (70000 < Re < 150000) and turbulence intensity level (1.5% < Tu < 2.5%). The PIV visualizations were completed in two measuring planes, that are oriented both normal and parallel to the wall. For the wall-parallel configuration, the measuring plane is located inside the boundary layer. The PIV data were post-processed by applying Dynamic Mode Decomposition (DMD), which provides frequency based modes and their corresponding growth rate. The effects of Re and Tu variation on the amplification of the dominant wavelength within the separated shear layer, which is responsible for transition, is the main subject of the present work. The DMD modes and related eigenvalues were computed with reference to the main streamwise coordinate. This allowed discussing the effects due to the main flow parameters on the amplification of the dominant streamwise wavelengths within the separated shear layer (Kelvin-Helmholtz modes). The growth of such streamwise modes ends with the formation of large scale vortices, whose breakup forces transition. In order to obtain the effective distribution of the maximum growth rate of fluctuations at different locations and times, the DMD domain was continuously extended in the streamwise direction, accounting for a specified number of periods characterizing the large scale K-H vortices. In order to reduce the time-space dependent results obtained by the DMD procedure, a probability density function of the most unstable wavelength and the corresponding growth rate has been computed. For the present data set, the spatial growth rate of fluctuations is found to increase at the higher Reynolds number, while it slightly reduces with increasing the Tu level. The procedure and findings discussed in this work shall be suitable for designing active control systems, such as harmonic blowing for separation control.

Energies ◽  
2020 ◽  
Vol 13 (18) ◽  
pp. 4886 ◽  
Author(s):  
Yang Yang ◽  
Xiao Liu ◽  
Zhihao Zhang

The current work is focused on investigating the potential of data-driven post-processing techniques, including proper orthogonal decomposition (POD) and dynamic mode decomposition (DMD) for flame dynamics. Large-eddy simulation (LES) of a V-gutter premixed flame was performed with two Reynolds numbers. The flame transfer function (FTF) was calculated. The POD and DMD were used for the analysis of the flame structures, wake shedding frequency, etc. The results acquired by different methods were also compared. The FTF results indicate that the flames have proportional, inertial, and delay components. The POD method could capture the shedding wake motion and shear layer motion. The excited DMD modes corresponded to the shear layer flames’ swing and convect motions in certain directions. Both POD and DMD could help to identify the wake shedding frequency. However, this large-scale flame oscillation is not presented in the FTF results. The negative growth rates of the decomposed mode confirm that the shear layer stabilized flame was more stable than the flame possessing a wake instability. The corresponding combustor design could be guided by the above results.


2014 ◽  
Vol 748 ◽  
pp. 848-878 ◽  
Author(s):  
Pramod K. Subbareddy ◽  
Matthew D. Bartkowicz ◽  
Graham V. Candler

AbstractWe study the transition of a Mach 6 laminar boundary layer due to an isolated cylindrical roughness element using large-scale direct numerical simulations (DNS). Three flow conditions, corresponding to experiments conducted at the Purdue Mach 6 quiet wind tunnel are simulated. Solutions are obtained using a high-order, low-dissipation scheme for the convection terms in the Navier–Stokes equations. The lowest Reynolds number ($Re$) case is steady, whereas the two higher $Re$ cases break down to a quasi-turbulent state. Statistics from the highest $Re$ case show the presence of a wedge of fully developed turbulent flow towards the end of the domain. The simulations do not employ forcing of any kind, apart from the roughness element itself, and the results suggest a self-sustaining mechanism that causes the flow to transition at a sufficiently large Reynolds number. Statistics, including spectra, are compared with available experimental data. Visualizations of the flow explore the dominant and dynamically significant flow structures: the upstream shock system, the horseshoe vortices formed in the upstream separated boundary layer and the shear layer that separates from the top and sides of the cylindrical roughness element. Streamwise and spanwise planes of data were used to perform a dynamic mode decomposition (DMD) (Rowley et al., J. Fluid Mech., vol. 641, 2009, pp. 115–127; Schmid, J. Fluid Mech., vol. 656, 2010, pp. 5–28).


Author(s):  
Susanne Horn ◽  
Peter J. Schmid ◽  
Jonathan M. Aurnou

Abstract The large-scale circulation (LSC) is the most fundamental turbulent coherent flow structure in Rayleigh-B\'enard convection. Further, LSCs provide the foundation upon which superstructures, the largest observable features in convective systems, are formed. In confined cylindrical geometries with diameter-to-height aspect ratios of Γ ≅ 1, LSC dynamics are known to be governed by a quasi-two-dimensional, coupled horizontal sloshing and torsional (ST) oscillatory mode. In contrast, in Γ ≥ √2 cylinders, a three-dimensional jump rope vortex (JRV) motion dominates the LSC dynamics. Here, we use dynamic mode decomposition (DMD) on direct numerical simulation data of liquid metal to show that both types of modes co-exist in Γ = 1 and Γ = 2 cylinders but with opposite dynamical importance. Furthermore, with this analysis, we demonstrate that ST oscillations originate from a tilted elliptical mean flow superposed with a symmetric higher order mode, which is connected to the four rolls in the plane perpendicular to the LSC in Γ = 1 tanks.


Author(s):  
Shubham Agarwal ◽  
Laurent Gicquel ◽  
Florent Duchaine ◽  
Nicolas Odier ◽  
Jérôme Dombard

Abstract Film cooling is a common technique to manage turbine vane and blade thermal environment. Optimizing its cooling efficiency is furthermore an active research topic which goes in hand with a strong knowledge of the flow associated with a cooling hole. The following paper aims at developing deeper understanding of the flow physics associated with a standard cooling hole and helping guide future cooling optimization strategies. For this purpose, Large Eddy Simulations (LES) of the 7-7-7 fan-shaped cooling hole [1] is performed and the flow inside the cooling hole is studied and discussed. Use of mathematical techniques such as the Fast Fourier Transforms (FFT) and Dynamic Mode Decomposition (DMD) is done to quantitatively access the flow modal structure inside the hole based on the LES unsteady predictions. Using these techniques, distinct vortex features inside the cooling hole are captured. These features mainly coincide with the roll-up of the internal shear layer formed at the interface of the separation region at the hole inlet. The topology of these vortex features is discussed in detail and it is also shown how the expansion of the cross-section in case of shaped holes aids in breaking down these vortices. Indeed upon escaping, these large scale features are known to not be always beneficial to film cooling effectiveness.


2009 ◽  
Vol 131 (2) ◽  
Author(s):  
Brian R. McAuliffe ◽  
Metin I. Yaras

Through experiments using two-dimensional particle-image velocimetry (PIV), this paper examines the nature of transition in a separation bubble and manipulations of the resultant breakdown to turbulence through passive means of control. An airfoil was used that provides minimal variation in the separation location over a wide operating range, with various two-dimensional modifications made to the surface for the purpose of manipulating the transition process. The study was conducted under low-freestream-turbulence conditions over a flow Reynolds number range of 28,000–101,000 based on airfoil chord. The spatial nature of the measurements has allowed identification of the dominant flow structures associated with transition in the separated shear layer and the manipulations introduced by the surface modifications. The Kelvin–Helmholtz (K-H) instability is identified as the dominant transition mechanism in the separated shear layer, leading to the roll-up of spanwise vorticity and subsequent breakdown into small-scale turbulence. Similarities with planar free-shear layers are noted, including the frequency of maximum amplification rate for the K-H instability and the vortex-pairing phenomenon initiated by a subharmonic instability. In some cases, secondary pairing events are observed and result in a laminar intervortex region consisting of freestream fluid entrained toward the surface due to the strong circulation of the large-scale vortices. Results of the surface-modification study show that different physical mechanisms can be manipulated to affect the separation, transition, and reattachment processes over the airfoil. These manipulations are also shown to affect the boundary-layer losses observed downstream of reattachment, with all surface-indentation configurations providing decreased losses at the three lowest Reynolds numbers and three of the five configurations providing decreased losses at the highest Reynolds number. The primary mechanisms that provide these manipulations include: suppression of the vortex-pairing phenomenon, which reduces both the shear-layer thickness and the levels of small-scale turbulence; the promotion of smaller-scale turbulence, resulting from the disturbances generated upstream of separation, which provides quicker transition and shorter separation bubbles; the elimination of the separation bubble with transition occurring in an attached boundary layer; and physical disturbance, downstream of separation, of the growing instability waves to manipulate the vortical structures and cause quicker reattachment.


Author(s):  
B. R. McAuliffe ◽  
M. I. Yaras

Through numerical simulations, this paper examines the nature of instability mechanisms leading to transition in separation bubbles. The results of two direct numerical simulations are presented in which separation of a laminar boundary layer occurs over a flat surface in the presence of an adverse pressure gradient. The primary difference in the flow conditions between the two simulations is the level of freestream turbulence with intensities of 0.1% and 1.45% at separation. In the first part of the paper, transition under a low-disturbance environment is examined, and the development of the Kelvin-Helmholtz instability in the separated shear layer is compared to the well-established instability characteristics of free shear layers. The study examines the role of the velocity-profile shape on the instability characteristics and the nature of the large-scale vortical structures shed downstream of the bubble. The second part of the paper examines transition in a high-disturbance environment, where the above-mentioned mechanism is bypassed as a result of elevated freestream turbulence. Filtering of the freestream turbulence into the laminar boundary layer results in streamwise streaks which provide conditions under which turbulent spots are produced in the separated shear layer, grow, and then merge to form a turbulent boundary layer. The results allow identification of the structure of the instability mechanism and the characteristic structure of the resultant turbulent spots. Recovery of the reattached turbulent boundary layer is then examined for both cases. The large-scale flow structures associated with transition are noted to remain coherent far downstream of reattachment, delaying recovery of the turbulent boundary layer to an equilibrium state.


Sign in / Sign up

Export Citation Format

Share Document